Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Assessing and monitoring the stability of stream detrital dynamics under forest disturbances Yeung, Alex Chee Yu 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2019_may_yeung_alexcheeyu.pdf [ 3.58MB ]
Metadata
JSON: 24-1.0376538.json
JSON-LD: 24-1.0376538-ld.json
RDF/XML (Pretty): 24-1.0376538-rdf.xml
RDF/JSON: 24-1.0376538-rdf.json
Turtle: 24-1.0376538-turtle.txt
N-Triples: 24-1.0376538-rdf-ntriples.txt
Original Record: 24-1.0376538-source.json
Full Text
24-1.0376538-fulltext.txt
Citation
24-1.0376538.ris

Full Text

ASSESSING AND MANAGING THE STABILITY OF STREAM DETRITAL DYNAMICS UNDER FOREST DISTURBANCES  by  Alex Chee Yu Yeung  B.Sc., The University of Hong Kong, 2010 M.Phil., The University of Hong Kong, 2013  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  February 2019  © Alex Chee Yu Yeung, 2019 ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: Assessing and managing the stability of stream detrital dynamics under forest disturbances  submitted by Alex Chee Yu Yeung in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Forestry  Examining Committee: Dr. John Richardson Supervisor  Dr. David Kreutzweiser Supervisory Committee Member  Dr. Cindy Prescott Supervisory Committee Member Dr. Marwan Hassan University Examiner Dr. Mary O’Connor University Examiner  Additional Supervisory Committee Members: Dr. Dan Moore Supervisory Committee Member  iii  Abstract  Disturbances affect ecosystem stability and the delivery of ecosystem services. Stability responses to disturbances involve multiple components, including changes in the levels (indicating resistance) and recovery time (resilience) of state variables. The stability of stream ecosystems is an active research area, however, empirical studies commonly measure single components of stability, and the basis of its spatio-temporal variability is often unclear. Bivariate and multi-dimensional frameworks were proposed to facilitate stability comparisons within and across ecosystems. I recommended adjusting two bivariate frameworks to better address diverse disturbance-response trajectories, shifting baselines, and broaden their applications for streams and other ecosystems. In my dissertation, I quantified multiple stability components of terrestrial-derived particulate organic matter (POM) availability and breakdown in small, temperate streams under forest harvesting. I focussed on stream POM dynamics, as they underpin the stability of consumer productivity (e.g., fish) and nutrient cycling, which are important aquatic ecosystem services. First, I investigated whether the natural variability of leaf litter breakdown – critical for detecting disturbance impacts – was affected by natural, weather-driven discharge variations across years, and decomposer preference of local versus exotic litter species. I showed that inter-annual hydrologic variations poorly explained litter breakdown, whose natural range of variation exceeded the benchmarks set by a popular bioassessment framework. Accordingly, I recalibrated reference conditions of litter breakdown to allow more robust bioassessments. Decomposer preference did not differ significantly between high-quality native and exotic litter, supporting their standardised use for disturbance studies across geographic regions. Secondly, I demonstrated that the resilience of litter breakdown to forest harvesting was likely greater in streams affected by thinning (with 50% basal area removal of riparian trees; 2-9 years post-harvest) than those affected by clear-cutting with or without riparian buffers (8-15 years). Thirdly, I modelled the resistance of stream POM quantity under variable, realistic harvesting impacts. Logging-induced changes in litterfall were more influential than peak discharge and stream temperature alterations in regulating POM quantity. Management strategies minimising riparian forest disturbances would more likely sustain detrital resource iv  availability and productivity in small streams. My results indicated that riparian vegetation, through litterfall and shading, importantly controlled the stability of stream POM dynamics. v  Lay Summary  The reliable delivery of ecosystem services depends on ecosystem stability. In small streams, the availability and breakdown of organic matter (OM) originating from streamside vegetation control productivity and nutrient cycling. Forest harvesting affects the stability of stream ecosystems via OM dynamics, but its effects vary across space and time. To better understand the causes of such variation, I first evaluated potential mechanisms that could affect the natural variation of stream OM breakdown. I found that natural, weather-driven discharge differences across years, and preference of decomposers, had small effects on the breakdown of high-quality leaf litter. I then demonstrated smaller changes in stream OM quantity, and faster recovery of OM breakdown, when streamside vegetation was minimally affected by forest harvesting practices. These practices include establishing no-harvest riparian buffers, and selective harvesting with a low basal area of trees removed. The stability of stream ecosystems is therefore critically controlled by surrounding forests. vi  Preface  Chapter 2: A version of this chapter is published in two peer-reviewed journals with the following citations. 1. [Yeung, A.C.Y.], and Richardson, J.S. 2016. Some conceptual and operational considerations when measuring “resilience”: a response to Hodgson et al. Trends Ecol. Evol. 31(1): 2–3; 2. [Yeung, A.C.Y.], and Richardson, J.S. 2018. Expanding resilience comparisons to address management needs: a response to Ingrisch and Bahn. Trends Ecol. Evol. 33(9): 647–649. I conceptualised the study, and wrote the manuscripts with input from John Richardson (hereafter JSR).  Chapter 3: A version of this chapter is published in a peer-reviewed journal with the following citation. [Yeung, A.C.Y.], Musetta-Lambert, J., Kreutzweiser, D.P., Sibley, P.K., and Richardson, J.S. 2018. Relations of interannual differences in stream litter breakdown with discharge: bioassessment implications. Ecosphere 9(9): e02423. I conceived and designed the study with David Kreutzweiser (hereafter DK) and JSR. I performed the experiment, processed samples, and analysed the data with Jordan Musetta-Lambert. I wrote the manuscript with input from all co-authors.  Chapter 4: A version of this chapter is in preparation for a peer-reviewed journal. I conceived and designed the study with help from JSR. I performed the experiment, processed samples, and analysed the data. I wrote the paper with input from DK and JSR.  Chapter 5: A version of this chapter is in preparation for a peer-reviewed journal. I conceived the study with JSR. O. Magnus Karlsson and JSR developed the model, and Karolina Stenroth (hereafter KS) made improvements to the model. I performed model simulations, and analysed the data. I wrote the paper with input from KS and JSR.  Chapter 6: A version of this chapter is published in a peer-reviewed journal with the following citation. [Yeung, A.C.Y.], Lecerf, A., and Richardson, J.S. 2017. Assessing the long-term ecological effects of riparian management practices on headwater streams in a coastal temperate vii  rainforest. For. Ecol. Manage. 384: 100–109. I conceived and designed the study with help from JSR. I performed the experiment, processed samples, and analysed the data. I wrote the manuscript with input from all co-authors. viii  Table of Contents  Abstract ......................................................................................................................................... iii Lay Summary .................................................................................................................................v Preface ........................................................................................................................................... vi Table of Contents ....................................................................................................................... viii List of Tables ............................................................................................................................... xii List of Figures ...............................................................................................................................xv Acknowledgements .................................................................................................................... xxi Dedication ................................................................................................................................. xxiii Chapter 1: Introduction ................................................................................................................1 1.1 Ecological stability in relation to disturbances ................................................................... 1 1.2 Forested headwater ecosystems: importance and threats.................................................... 4 1.3 Organic matter dynamics as indicators of ecosystem services provided by small streams 5 1.4 Forestry impacts on headwater streams: considerations of environmental drivers, multiple scales and casual mechanisms .................................................................................................... 6 1.5 Thesis structure and objectives ........................................................................................... 7 Chapter 2: Towards standardising comparisons of ecological stability across ecosystems: conceptual and operational considerations ...............................................................................13 2.1 Introduction ....................................................................................................................... 13 2.2 General considerations ...................................................................................................... 14 2.3 Framework-specific considerations .................................................................................. 15 2.4 Conclusion ........................................................................................................................ 18 Chapter 3: Relations of interannual differences in stream litter breakdown with discharge: bioassessment implications ..........................................................................................................20 3.1 Introduction ....................................................................................................................... 20 3.2 Methods............................................................................................................................. 23 ix  3.2.1 Study region and site characteristics ...................................................................... 23 3.2.2 Field sampling ........................................................................................................ 24 3.2.2.1 Litter breakdown ....................................................................................... 24 3.2.2.2 Shredders and fungal biomass .................................................................. 25 3.2.2.3 Water chemistry and temperatures ............................................................ 27 3.2.3 Data analyses .......................................................................................................... 28 3.2.3.1 Litter breakdown rates .............................................................................. 28 3.2.3.2 Hydrologic indices .................................................................................... 29 3.2.3.3 Statistical analyses .................................................................................... 30 3.3 Results ............................................................................................................................... 32 3.3.1 Litter breakdown rates and their spatio-temporal variability ................................. 32 3.3.2 Selection of variables ............................................................................................. 33 3.3.3 Variance partitioning .............................................................................................. 33 3.3.4 Relationships between hydrology and litter breakdown rates ................................ 34 3.3.5 Spatiotemporal variability of shredder assemblage structure ................................. 34 3.4 Discussion ......................................................................................................................... 35 3.4.1 Sources of variation in litter breakdown rates ........................................................ 35 3.4.2 Spatio-temporal variability of litter breakdown rates ............................................. 37 3.4.3 Bioassessment implications .................................................................................... 38 3.4.4 Conclusions ............................................................................................................ 40 Chapter 4: Litter quality and decomposer community attributes mediate the effects of litter-decomposer interactions on stream litter breakdown ....................................................52 4.1 Introduction ....................................................................................................................... 52 4.2 Methods............................................................................................................................. 55 4.2.1 Study sites and litter species selected ..................................................................... 55 4.2.2 Field sampling ........................................................................................................ 56 4.2.2.1 Litter breakdown experiment .................................................................... 56 4.2.2.2 Shredders, fungal and algal biomass ......................................................... 58 4.2.2.3 Water characteristics and microhabitat variables ..................................... 59 4.2.2.4 Litter chemistry ......................................................................................... 59 x  4.2.3 Data analyses .......................................................................................................... 60 4.3 Results ............................................................................................................................... 63 4.3.1 Litter quality analysis ............................................................................................. 63 4.3.2 HFA: magnitude, variability, and drivers ............................................................... 63 4.3.3 Shredder assemblage dissimilarity ......................................................................... 65 4.4 Discussion ......................................................................................................................... 65 4.4.1 Litter-microbe interactions ..................................................................................... 66 4.4.2 Litter-shredder interactions .................................................................................... 68 4.4.3 Variability of HFA on litter breakdown ................................................................. 69 4.4.4 Implications for comparing breakdown between native and exotic litter .............. 70 4.4.5 Conclusions ............................................................................................................ 71 Chapter 5: Modelling biophysical controls on stream organic matter standing stocks under variable forest harvesting impacts .............................................................................................77 5.1 Introduction ....................................................................................................................... 77 5.2 Methods............................................................................................................................. 79 5.2.1 Literature analysis of post-harvest responses of model drivers ............................. 79 5.2.2 Stream CPOM model ............................................................................................. 81 5.2.3 Sensitivity analysis and heuristic modelling .......................................................... 82 5.2.4 Data analyses .......................................................................................................... 83 5.3 Results ............................................................................................................................... 84 5.3.1 Published post-harvest responses of model drivers ................................................ 84 5.3.2 Sensitivity analysis and heuristic modelling of CPOM .......................................... 85 5.3.3 Interactive effects of model drivers on CPOM ...................................................... 86 5.4 Discussion ......................................................................................................................... 87 5.4.1 Post-harvest response ranges of model drivers ...................................................... 88 5.4.2 Effects of individual drivers on CPOM .................................................................. 88 5.4.3 Effects of multiple drivers on CPOM ..................................................................... 91 5.4.4 Model limitations ................................................................................................... 92 5.4.5 Implications for forest and watershed management ............................................... 93 xi  Chapter 6: Assessing the long-term ecological effects of riparian management practices on headwater streams in a coastal temperate rainforest .............................................................101 6.1 Introduction ..................................................................................................................... 101 6.2 Methods........................................................................................................................... 103 6.2.1 Study sites ............................................................................................................. 103 6.2.2 Water temperature and chemistry ......................................................................... 104 6.2.3 Leaf bags .............................................................................................................. 105 6.2.4 Shredders .............................................................................................................. 105 6.2.5 Statistical analyses ................................................................................................ 106 6.3 Results ............................................................................................................................. 108 6.3.1 Canopy openness and water characteristics ......................................................... 108 6.3.2 Litter decomposition ............................................................................................. 108 6.3.3 Shredders .............................................................................................................. 109 6.4 Discussion ....................................................................................................................... 110 6.4.1 Changes in shredder density, rarefied richness and assemblage structure ........... 111 6.4.2 Changes in litter decomposition ........................................................................... 113 6.4.3 Implications for forest management ..................................................................... 114 Chapter 7: Conclusion: synthesis and implications ................................................................122 7.1 Standardising stability comparisons within and across ecosystems ............................... 123 7.2 Evaluating mechanisms underpinning the stability of stream particulate OM dynamics123 7.3 Re-evaluating the sensitivity of litter breakdown as a bioassessment tool ..................... 126 7.4 Study caveats and limitations.......................................................................................... 127 7.5 Management implications ............................................................................................... 128 Bibliography ...............................................................................................................................130 Appendices ..................................................................................................................................176 Appendix A Supporting information for Chapter 3 ............................................................... 176 Appendix B Supporting information for Chapter 4 ................................................................ 186 Appendix C Supporting information for Chapter 5 ................................................................ 195 Appendix D Supporting information for Chapter 6 ............................................................... 212  xii  List of Tables  Table 3.1 Catchment- and reach-scale characteristics for sites in the study regions in British Columbia and Ontario. ................................................................................................. 42 Table 3.2 Initial sets of candidate hydrologic indices, shredder-related, fungal biomass, and water chemistry variables that were expected to influence stream litter breakdown in the study regions in British Columbia and Ontario. ..................................................... 44 Table 3.3 Extent of spatio-temporal variability of in-stream total breakdown rate (kc), fragmentation rate (λF), and dissolution and microbial decomposition rate (λm) in the study regions in British Columbia and Ontario............................................................ 45 Table 3.4 Results of forward selection of hydrologic indices, shredder-related, fungal biomass, and water chemistry variables on in-stream total breakdown rate (kc), fragmentation rate (λF), and dissolution and microbial decomposition rate (λm) in the study regions in British Columbia and Ontario. Variables selected for use in the variance partitioning analysis, and the directional effect of each variable (+, −) on litter breakdown rates are given (see Table 3.2 for notations of variables). .......................................................... 46 Table 3.5 Results of linear mixed-effects models explaining the relationships between in-stream total breakdown rate (kc) and dissolution and microbial decomposition rate (λm), and the composite hydrologic index, for each study region in British Columbia and Ontario. Site was treated as a random effect for each region in regression analyses. Model fit was assessed by the marginal and conditional determination coefficients (R2) using the rsquared.GLMM function in the MuMIn package (Nakagawa and Schielzeth 2013). The significance of regression relationships was estimated using an approximate F-test based on the Kenward-Roger approach. ....................................... 47 Table 3.6 Recommended baseline conditions for assessing the functional integrity of small streams using total litter breakdown rate (kc) in the study regions in British Columbia and Ontario. .................................................................................................................. 48 Table 4.1 Results of regression analyses on the proportional effects of litter origin on stream litter breakdown rates (λF and kf), as influenced by water characteristics, and either xiii  average litter quality and litterbag-associated community attributes of home-derived and foreign litter species, or proportional effects of litter origin on these attributes. The optimal model for each breakdown rate is shown (* P < 0.05; ** P < 0.01). Data from G stream in Malcolm Knapp Research Forest were omitted from the analysis due to incomplete records of microhabitat variables. ............................................................. 71 Table 5.1 Changes in litterfall, peak flows, and stream temperature assigned to each level of severity of forest harvesting disturbance for the heuristic modelling of CPOM standing stocks in East Creek, British Columbia........................................................................ 95 Table 5.2 Difference in average daily CPOM standing stocks (g AFDM m-2) between the values simulated using the heuristic model of forest harvesting disturbance and the unharvested (control) scenario in East Creek, British Columbia. For each pair of model drivers, individual effects of drivers (litterfall, L; peak flows, P; stream temperature, T) and the cumulative effects of drivers were modelled. Positive (negative) values indicate that daily CPOM standing stocks in the given scenario are higher (lower) than in the unharvested scenario. For each heuristic scenario with cumulative effects, the difference in daily CPOM standing stocks between values expected from the additive null model and the unharvested scenario are given in parentheses. Also shown are the directional interaction types of two-model driver interactions (+A: positive antagonistic, -A: negative antagonistic; sensu Piggott et al. (2015)), and the results of Wilcoxon’s signed-rank tests indicating whether these directional interaction types were significantly non-additive. P values < 0.05 are in bold font. Refer to Table 5.1 for changes in model drivers associated with each category of disturbance severity................................................................................... 96 Table 5.3 Results of three-way ANOVA testing the modelled main effects and interactions of litterfall (L), peak flows (P), and stream temperature (T) affected by forest harvesting on ln(hydrograph-specific response ratio of CPOM) in East Creek, British Columbia. P-values < 0.05 are in bold typeface. ........................................................................... 97 Table 6.1 Reach-scale characteristics for the sixteen stream sites in the study. Water temperature during the study period (early November 2013 to late January 2014) is given as daily mean values. Other water characteristics obtained in the middle of the study period are xiv  shown. Soluble reactive phosphorous (SRP) concentration was below detection limit (i.e., 10 µg P L−1) in all sites, and is therefore not shown in the table. ...................... 116 Table 6.2 Summary of ANOVA and planned comparisons of the effects of forest treatments on biological parameters. Shredder densities were ln-transformed to normalise the data and to improve the homogeneity of variance. Note that three sites were omitted in the analyses of rarefied shredder richness. P values are shown in parentheses, and are in bold typeface when significant at 0.05. ...................................................................... 117 Table 6.3 Changes in the effects of forest treatments on functional stream integrity between studies conducted in 2006 (i.e., data from Lecerf and Richardson (2010)) and in 2013 (i.e., present study). The assessment of functional stream integrity was based on comparing daily, non-temperature-corrected decomposition rate coefficients at treatment (kt) versus reference sites (kr). Scores (in parentheses and bolded) represent the assessed conditions of the treatment sites (0: severely impaired; 1: mildly impaired; 2: no evidence of impairment). For details of assessment criteria, refer to Gessner and Chauvet (2002). ..................................................................................... 118  xv  List of Figures  Figure 1.1 Major types of disturbance-response patterns of ecosystems. The onset and end of disturbance are indicated by vertical, grey dotted lines. Three common types of relationships between disturbance (dotted lines) and response variables (solid lines) are illustrated: (a) a pulse disturbance may lead to a pulse response; (b) a pulse disturbance may lead to a press response; (c) a press disturbance may lead to a press response (modified from Lake 2000). The response variable may return to pre-disturbance levels (i.e., asymmetric patterns of trajectories indicating hysteresis; dash-dot line 1), or may remain at persistently lower (or higher) levels, if the system reaches and remains in an alternative stable state, and/or if baseline conditions shift (dash-dot line 2). Ramp disturbance and response that change steadily with time without clear endpoints are not illustrated here. Note that the response variable can also follow a contrasting disturbance-induced trajectory; that is, it increases during disturbance, and decreases afterwards to reach predisturbance state (see Chapter 2.3). .................................................................................................................................... 10 Figure 1.2 Schematic diagram illustrating engineering and ecological resilience. Valleys represent stability domains, balls represent the system and arrows represent different levels of disturbances. Engineering resilience is a measure of the time taken for the system to return to pre-disturbance levels (t) within (or back to) the original stability domain (A), with the length of the upward arrow (solid line) indicating resistance to the disturbance. Ecological resilience is a large-scale topographical measure of a specified stability domain. The amount of disturbance (dotted line) required for the transition of a system to an alternative stable state (from A to B) depends on how close the system is to the threshold (i.e., precariousness). Note this is a simplified graphical representation (modified from Gunderson 2000), as transitions are not always reversible, and the topography of the valleys may not be static over time. ... 11 Figure 1.3 Hypothetical trajectory of the response variable y before and after disturbance, and quantifiable components of ecological stability associated with the disturbance-xvi  induced change in y. The dotted line indicates the end of the disturbance. The difference between yt0 (at baseline conditions, and before disturbance) at t0 and yti (at the end of disturbance) at ti is the magnitude of change, and the inverse of which represents resistance (as defined in section 1.1). The response variable recovers to ytr within its range of natural variability at tr. tr – ti indicates the recovery time, and the inverse of which represents resilience. The response variable is considered to have recovered to its pre-disturbance levels when it reaches yt2 at t2, but not yt1 at t1. The focal components of ecological stability investigated in the thesis chapters are indicated. .................................................................................................................... 12 Figure 2.1 Disturbance responses and resilience components of an ecosystem in two common management scenarios. Hypothetical disturbance-induced change in an ecosystem relative to the predisturbance state at time t0 is baseline-normalised (Sbase). Baseline-normalised impact (∆Sti(base)) is expressed as the percentage of baseline state at the end of disturbance (time ti), as shown in Figure IB in Ingrisch and Bahn (2018). The orange boxes indicate the occurrence of disturbance. (a) The baseline state of the reference (control) system decreases to Rtx (= Rti) for trajectories 1 and 2 at time ti, which remains unchanged from time ti to tx for trajectory 1, and further decreases to R’tx for trajectory 2. The disturbed ecosystem reaches a steady state at time tx and does not return to the predisturbance state, due to its shifting baseline. (b) The disturbed ecosystem reaches Sti (above management target ST, a minimum acceptable level of an environmental condition) for trajectory 1 and S’ti (below ST) for trajectory 2 at time ti. Therefore, the proximity to management target at ti is positive for trajectory 1 (P; 20%), and negative for trajectory 2 (P’; -40%). The baseline-normalised impact for trajectory 2 is greater than that of trajectory 1, and both reach the predisturbance state at time tr with identical recovery time in this example. ....... 19 Figure 3.1 Map showing the locations of study regions in British Columbia and Ontario, Canada. Abbreviations of study regions are as in Table 3.1. ..................................... 49 Figure 3.2 Results of variance partitioning for (a) total breakdown rate (kc), (b) fragmentation rate (λF), and (c) dissolution and microbial decomposition rate (λm) across all study regions in British Columbia and Ontario, and in each of these regions, using forward-xvii  selected hydrologic (H), shredder-related (S), and/or water chemistry (W) variables as predictors (see Table 3.2). The global RDA model involving fungal biomass is non-significant for all breakdown metrics, and hence it is excluded from the variance partitioning analysis. Values displayed are adjusted R2 as portion of variance explained, including the residual, unexplained variation, and negative values are not shown. The sum of variance explained by the explanatory matrices and residual variance may exceed 1 due to negative explained variances. Significance levels of the unique effects of hydrology, shredders and water chemistry are indicated with asterisks (* P < 0.05, ** P < 0.01, *** P < 0.001). Abbreviations of study regions are as in Table 3.1. ........................................................................................................... 50 Figure 3.3 Relationships between (a) total breakdown rate (kc), (b) dissolution and microbial decomposition rate (λm) of leaf litter (shown on a log10 scale), and the corresponding fitted site scores (i.e., linear combination of selected hydrologic indices) on the first RDA axis, generated from the global RDA across all study regions in British Columbia and Ontario. The study years are colour-coded. Note that RDA axis 1 scores are not comparable between (a) and (b). Abbreviations of study regions are as in Table 3.1. ................................................................................................................ 51 Figure 4.1 Principal component analysis (PCA) biplot of 8 litter quality traits (arrows) measured for the 6 plant litter species (points) in the study regions in British Columbia (closed circles) and Ontario (open circles), Canada. RA = red alder (Alnus rubra); SA = speckled alder (Alnus incana ssp. rugosa); VM = vine maple (Acer circinatum); SM = sugar maple (Acer saccharum); WRC = western red cedar (Thuja plicata); EWC = eastern white cedar (Thuja occidentalis). ................................................................... 73 Figure 4.2 Observed and estimated (from resampling) home-field advantage index (HFAI) for in-stream (a) fragmentation rate, λF, (b) litter breakdown rate in fine-mesh bags, kf, (c) algal biomass on leaf discs in coarse-mesh bags, (d) fungal biomass on leaf discs in fine-mesh bags, and (e) shredder density and (f) shredder taxonomic richness of three litter pairs incubated in the study regions. For estimated HFAI, median values and error ranges representing the 95th percentile confidence intervals (CI) are shown. xviii  Asterisks indicate that values of estimated HFAI are significant (i.e., 95% CI not overlapping zero; * P < 0.05). .................................................................................... 74 Figure 4.3 Non-metric multidimensional scaling ordination of shredder assemblages on two litter pairs of alder (red alder: RA; speckled alder: SA), and maple (vine maple: VM; sugar maple: SM) in (a) British Columbia and (b) Ontario, Canada. Ellipses represent 95% confidence intervals for the individual species across sites. Note that NMDS axis scores are not comparable between study regions. ............................................. 75 Figure 4.4 Relationships between (a) differences in litter C:N between native and exotic species, and the ln-response ratio of kf (in day-1) of native species relative to the exotic species (RRnative:exotic of kf), and between (b) litter C:N and kf of native and exotic species. The source of data in (a) and (b) were identical, and from studies selected by the meta-analysis by Kennedy and El-Sabaawi (2017) and this study (only including TLW data). Only studies using litterbags with mesh size <0.5 mm to measure kf were included. When there were multiple native (exotic) species involved in a given study selected by Kennedy & El-Sabaawi (2017), kf values were averaged to give a single value for each native (exotic) species group, prior to calculating RRnative:exotic of kf. Values from multiple sites in each study were averaged. For my study, comparisons between values of native and exotic litter were made for specified litter pairs for plotting (a). The black line in (a) indicates ordinary least-squares regression (y = -0.0125x – 0.407; R2 = 0.74) through all plotted points, whose slope is significantly negative. Results of power regression analysis in (b) show weak and insignificant power relationships between litter C:N and kf of native litter (y = 0.0352x-0.409; F1,12 = 1.19; p = 0.30; R2 = 0.09) and exotic litter (y = 0.0637x-0.465; F1,8 = 2.69; p = 0.14; R2 = 0.25). ....................................................................................................................... 76 Figure 5.1 Simulated daily CPOM standing stocks (g ash-free dry mass [AFDM] m-2) in East Creek, British Columbia, from model runs under unharvested (control) conditions and with logging-associated changes of high severity in (a) single model drivers, including litterfall (L), peak flows (P), and stream temperature (T), and (b) multiple model drivers (L+P, P+T, L+T, L+P+T). Model outputs of 730 time steps (i.e., across two years) averaged across an ensemble of ten simulations are shown, and results xix  from the first 90 days of each model run (spin-up period) are excluded to allow for the model to stabilise. Refer to Table 5.1 for changes in individual model drivers for each scenario. ............................................................................................................. 98 Figure 5.2 Effect size of logging-associated changes in litterfall, peak flows (P), and stream temperature (T) on average daily CPOM standing stocks (ln[response ratio of CPOM]) in East Creek, British Columbia, simulated by full-factorial heuristic modelling of forest harvesting impacts. Effect size for each heuristic scenario averaged across ten model simulations, and its associated 95% confidence interval (represented by error ranges which are in most cases very small), are shown. Effect sizes with corresponding error ranges overlapping zero (the solid line) are not significant. Panels are arranged by increasing severity of logging-associated changes in litterfall: (a) no changes (i.e., control), (b) low, (c) moderate, (d) high, and (e) very high severity. Note that the scale of y-axis in (e) is different from other panels. ...... 99 Figure 5.3 Relationships between the temperature-dependent function for leaf litter consumption by shredders and stream temperature. These relationships are part of the CPOM model by Stenroth et al. (2014), and were established according to consumption model 2 for warm-water species (Kitchell et al. 1977) in the Fish Bioenergetics 4.0 model (Deslauriers et al. 2017). Shaded area denotes the range of smoothed daily stream temperature from June 1st to September 30th at East Creek, British Columbia. The solid line denotes the average daily stream temperature (11.3°C) of this period, which is below the optimum temperature (15°C) for shredder consumption used to parameterise the stream CPOM model. .................................................................... 100 Figure 6.1 Comparison of the effects of forest treatments on (a) daily and (b) temperature-corrected decomposition of alder leaves, on (c) shredder density and (d) rarefied taxonomic richness between studies conducted in 2006 (i.e., data from Lecerf and Richardson (2010)) and in 2013 (i.e., present study). Letters indicate homogeneous groups derived from ANOVA and planned contrasts uniquely for each study (see Table 6.2). Error bars denote standard errors. Note that scales of y-axis differ across studies in (c). ............................................................................................................ 120 xx  Figure 6.2 Non-metric multidimensional scaling ordination of shredder assemblages across sixteen streams sampled in (a) 2006 and (b) 2013. Dashed ellipses represent 95% confidence intervals for the centroids of the groups of forest treatments. ............... 121  xxi  Acknowledgements I feel like a cased caddisfly larva drifting in this academic journey along streams. High-flow events were common, so were joyful (re)discoveries of patches of highly nutritious leaf litter in no insect’s land. Luckily, I didn’t get consumed prematurely. I now see what I processed has become useful for others locally and distantly. Let me emphasise that it has been a well-directed, humbling, and enjoyable drift. I am tremendously thankful for the constant guidance and patience of my supervisor, Dr. John Richardson, and importantly, from taking me on in the first place. I am also grateful to members of my supervisory committee, Dr. David Kreutzweiser, Dr. Dan Moore, Dr. Cindy Prescott, who challenged my thinking, and offered valuable advice throughout my study. The widely read and forthright advice for graduate students by Stearns (1987) and Huey (1987) helped me (re)gain focus throughout this research journey. I must thank NSERC Canadian Network for Aquatic Ecosystem Services, and the University of British Columbia (UBC) for funding my research. I am very fortunate to be given the freedom to undertake a research-intensive journey with additional support from a Natural Sciences and Engineering Research Council of Canada CGS-D Fellowship, a UBC Killam Doctoral Scholarship, and Society for Freshwater Science Simpson Fund. Members of the Stream and Riparian Research lab have been a continuous source of intellectual and emotional empowerment. I appreciate the good company of brilliant folks including Claire Cathcart, Ana Chara, Danielle Courcelles, Felipe Rossetti de Paula, Gillian Fuss, Liliana García, Roseanna Gamlen-Greene, Brian Kielstra, Lenka Kuglerová, Kasey Moran, Sean Naman, Tonya Ramey, David Tavernini, and many others. My field and laboratory assistants provided enormous support from the ground up; in particular, without their unwavering dedication to picking macroinvertebrates from organic matter chunks under microscope, I wouldn’t have gone this far. I am indebted to the research staff at Canadian Forest Service in Sault Ste. Marie (especially Scott Capell, Kevin Good, and Derek Chartrand) and staff at Malcolm Knapp Research Forest (especially Ionut Aron) for their advice, logistical and/or laboratory assistance. Jason Leach and Junting Guo were extremely helpful for generating discharge data for my research. I would also like to extend my thanks to Diane Srivastava (UBC) and Irena Creed (Western University) for offering collaborative opportunities to do impactful, xxii  extracurricular research related to environmental change, which have broadened my horizons and extended my networks. My immersive intellectual drift from the very beginning has benefitted from a highly stable, resourceful, and motivating environment provided by my family. I cannot ask for more from them. Finally, a great thanks to Beverly Po for her care and emotional support, particularly throughout the later writing stages, and the joy she has sparked in me. xxiii  Dedication  To my parents Victor Yeung and Wendy Lu, and brother Paul Yeung for their care and enduring support for my pursuit of knowledge. Without them I wouldn’t have found smoother pebbles and fresher leaves than ordinary along the streamside.1  Chapter 1: Introduction  1.1 Ecological stability in relation to disturbances The era of the Anthropocene has seen pervasive, ever-increasing human transformations of ecosystems worldwide (Vitousek et al. 1997; Zalasiewicz et al. 2011). Researchers, managers, and policy-makers are confronted with a greater uncertainty to forecast how ecosystems respond to intensifying and novel human disturbances, especially with respect to the functions and services they provide for humanity (Folke et al. 2004; Bulling et al. 2010). Ecosystems naturally experience disturbance events, which affect their biotic and abiotic components subsequent to the removal of organisms, and opening of niches and habitat space (Resh et al. 1988). The responses of these biotic and abiotic components can follow some generalised patterns in terms of their ability to withstand and recover from disturbances in question (Fig. 1.1; Lake 2000). Previous literature has given a plethora of terminologies and definitions describing similar or different aspects of the ‘stability’ of ecosystems (Grimm and Wissel 1997), which is regarded as a complex and multifaceted concept (Donohue et al. 2013). Therefore, an undefined and uncritical use of terminologies may result in confusion when using ‘stability’ concepts to understand and predict ecosystem responses to disturbances, and misinterpretation of results from different studies. According to Grimm and Wissel (1997), two of the identifiable properties that provide quantitative measures of ‘stability’ include constancy (referring to the variable staying essentially unchanged) and resilience (returning to the pre-disturbance state after the cessation of disturbance). To address how an anthropogenic disturbance modifies or interacts with natural disturbances to influence ‘stability’, resistance (in lieu of constancy) and resilience are the two major quantifiable measures of stability considered in this thesis, and henceforth defined as follows:  Resistance: Inverse of the magnitude of change in an ecological state variable in response to the disturbance. More resistant systems show a lesser extent of change.  2  Resilience: Inverse of the time taken for an ecological state variable to return to its pre-disturbance level. More resilient systems require less time for recovery. In the absence of pre-disturbance data, recoveries to within the range of natural variability of reference sites or systems are instead adopted for space-for-time substitution (chronosequence) studies which are considered to approximate pre-disturbed conditions of impacted sites.  Resistance defined here reflects the ability of the state variable to withstand changes attributed to the disturbance (similar to “inertia” in Orians 1975). Together with resilience, they are close to the Westman-Webster definition (Westman 1978; Webster et al. 1983; Lake 2013), and the modified definition of ‘engineering resilience’ – ‘the intrinsic ability of a system to adjust its functioning prior to, during, and following changes and disturbances, so that it can sustain operations under both expected and unexpected conditions’ (sensu Hollnagel 2010; see Fig. 1.2). While acknowledging some emerging definitions that have also become mainstream, such as ‘ecological resilience’ (sensu Peterson et al. 1998), its quantification is generally considered challenging (Baho et al. 2017), as comprehensive data collections and data-intensive modelling are required to establish and assess the amount of disturbance required for the systems to cross thresholds, and enter into alternative stable states (e.g., Costanza 1992; Carpenter et al. 2001; Lake 2013). Studies of disturbance ecology commonly characterise disturbances with regard to their intensity, duration, frequency, and response dynamics of the ecological states of communities and ecosystems in terms of their spatial extent and temporal duration (Fig. 1.1; e.g., Romme et al. 1998; White and Jentsch 2001; Elmqvist et al. 2003). Sometimes disturbance may propagate to yield large-scale, long-lasting influences throughout the impacted landscapes (Niemi et al. 1990; Turner et al. 2003). The responses of ecosystem functions following disturbance in particular tend to be more complex, given their dependence on the interactions between abiotic and biotic drivers which may operate across dissimilar spatial and temporal scales (e.g., Peterson et al. 1998; Seidl et al. 2011; Graça et al. 2015). Other challenges facing studies evaluating resilience include the requirement of long-term (e.g., multi-annual up to centennial) continuous sampling for long recoveries (Turner et al. 2003; Lake 2013) and potentially delayed effects occurring decades after disturbances (Perry and Jones 2016; Jackson et al. 2018), fragmented 3  data availability, and shifting baseline conditions owing to ongoing environmental changes (Steen and Jachowski 2013). In order to robustly assess how disturbances affect ecological stability, empirical and modelling studies should establish a well-defined focus and relevant spatial and temporal scales for inquiry. The range of natural variability of the ecological state variable should also be quantified (Fig. 1.3). These approaches provide the basis for distinguishing the effects of anthropogenic disturbance from natural variations and disturbances, disentangling drivers, and forecasting trends (see Landres et al. 1999; White and Jentsch 2001). Developing a causal understanding of disturbance-response dynamics is key to identifying ecological variables that are more (or less) resistant and resilient to disturbances. Such an understanding can inform the adaptive management of ecosystems, which seeks to iteratively implement, evaluate, and adjust management practices to maintain and restore the resilience of ecosystems (Groffman et al. 2006; Folke 2006). Appropriate characterization of disturbances and response variables is also important for analysing and explaining the variability in measures of ecological stability across space and time. Numerous bivariate frameworks were recently formulated to use two distinct components of change after disturbance (i.e., resistance and recovery) to simplify the characterization and facilitate coherent comparisons of ecological stability (Hodgson et al. 2015; Nimmo et al. 2015; Ingrisch and Bahn 2018). The choice of these components differs across frameworks, and a good understanding of disturbance-response dynamics can assist with the selection of an appropriate framework for comparisons across ecosystems. In addition, the application of the concept of ecological stability critically depends on such understanding to address variable management needs of ecosystems. For instance, when the current state of ecosystems is already in close proximity to a threshold, they have high precariousness and may undergo shifts to an undesired alternative (stable) state under disturbances. Disturbances may disrupt internal controls by biotic interactions (e.g., removal of top predators or keystone species), which lead to state shifts of ecosystems (see Hilderbrand and Utz 2015). These ecosystems may be less resilient when recovery is difficult and takes a long time (Folke et al. 2004; Walker et al. 2004). Management practices may be targeted to minimise and/or mitigate disturbances that can trigger state shifts, so 4  as to sustain the capacity of ecosystems to provide functions and services (Folke et al. 2004). In contrast, an evolving framework of adaptive management that emulates natural disturbance regimes has been increasingly applied in forest and stream-riparian ecosystems (Long 2009; Sibley et al. 2012). Intentional disturbances (e.g., forest harvesting) are applied to mimic historical (or acceptable) ranges of natural variation of the ecosystems, in order to maintain habitat complexity and confer ecological resilience (Drever et al. 2006). However, more empirical evidence is still needed to assess the effectiveness of this type of disturbance-based management in maintaining multiple essential ecosystem functions and services (Sibley et al. 2012; Lindenmayer and Laurance 2012).  1.2 Forested headwater ecosystems: importance and threats Riverine ecosystems provide vital resources for humans, and have been under ever-increasing stress globally due to human uses (Vörösmarty et al. 2010; Strayer and Dudgeon 2010). Anthropogenic disturbances have massively transformed riparian zones and watersheds by altering the inputs of water, nutrients, organic matter and sediments, in many cases imperilling biodiversity and compromising the functions and services these ecosystems provide (e.g., Williamson et al. 2008; Strayer and Dudgeon 2010; Dodds et al. 2013). A large body of research has sought to characterise structure-function-service relationships in riverine systems, contributing to identifying suitable indicators of ecosystem integrity and services (e.g., Sweeney et al. 2004; Young et al. 2008; Feld et al. 2009; Cardinale et al. 2012). These established relationships have guided us toward making quantitative predictions of how riverine ecosystem functions and services change when affected by human-induced disturbances. Notably, we now better understand 1) how disturbances can independently or interactively influence environmental factors which regulate the structural and functional metrics of these systems; 2) the inherent differences in these regulatory relationships across spatial and temporal scales; and 3) linearity (or non-linearity) within or outside the range of natural variation (see Stanley et al. 2010; Dodds et al. 2010; McCluney et al. 2014) Small, low-order streams in forested catchments were chosen to be the main study system for this thesis to investigate the effects of disturbance on riverine ecosystem structure, functions and services. Headwater systems are considered to be confined to ≤1 km2, typically taking up 5  around 70-80% of the total catchment area. The downstream limits of headwater systems are variable and depend on the strength of processes linking hillslopes to streams, but normally include first-order stream channels (see Gomi et al. 2002), where most of study sites in this thesis are located. Accordingly, first-order streams are referred to as ‘headwaters’, and ‘small streams’ include first- to third-order stream channels.  Headwaters can profoundly influence downstream receiving waters owing to a tight and dynamic coupling of hydrological and biogeochemical processes between them (e.g., Gomi et al. 2002; MacDonald and Coe 2007; Alexander et al. 2007). The tight linkage between stream and terrestrial ecosystems is manifested in various controls of riparian vegetation on the physical environment and biological processes in streams. Examples include the input of terrestrial organic matter and invertebrates, provision of shade to control stream temperature and primary production, modification of nutrient and sediment runoff (see review by Naiman and Décamps 1997). Collectively, headwater streams and the adjacent riparian habitats contribute significantly to maintaining the diversity and functional integrity of the entire river network (Freeman et al. 2007). They support rich and distinct biological communities (Meyer et al. 2007; Clarke et al. 2008), export substantial amounts of subsidies (aquatic and terrestrial insects, and organic matter) to support downstream foodwebs (Wipfli and Gregovich 2002; Freeman et al. 2007), and play a similar or disproportionately more important role (e.g., per unit time) in retaining and transforming nutrients when compared to larger rivers downstream (Peterson et al. 2001; Ensign and Doyle 2006).  1.3 Organic matter dynamics as indicators of ecosystem services provided by small streams The amount and processing of in-stream organic matter (OM) can reflect several types of aquatic ecosystem services, such as provisioning (e.g., fish production) and supporting services (e.g., nutrient cycling) associated with water quality. The availability of terrestrial-derived OM (such as leaf litter) provides critical energy and resource subsidies to consumers in nutrient-limited headwater streams, including detritivorous invertebrates (shredders), and consumers (such as fish) dependent on them (e.g., Wallace et al. 1999; Hall et al. 2000; Richardson and Sato 2015). OM processing (commonly measured as the rate of leaf litter breakdown) integrates a number of 6  biotic and abiotic processes, and is widely measured as a key functional metric of stream integrity (Gessner and Chauvet 2002; Young et al. 2008). Particulate OM provides substrates for microbial activity and indirectly mediates stream metabolism and nutrient retention (Crenshaw et al. 2002; Aldridge et al. 2009). Dissolved organic carbon derived from terrestrial OM can influence productivity in streams and lakes further downstream (e.g. Carpenter et al. 2005; Battin et al. 2008; Tanentzap et al. 2014). Some structural measures of stream ecosystems, such as invertebrate shredder abundance and diversity, are indirectly related to provisioning and supporting services. Macroinvertebrates within the headwaters and drifting downstream can constitute an important source of nutrients and energy to predators, particularly fish (e.g., Covich et al. 1999; Vanni 2002; Wipfli and Gregovich 2002). Shredders and collector-gatherers facilitate the breakdown of terrestrial plant materials and the transport of coarse and fine particulate OM (CPOM and FPOM) for downstream consumers (see reviews by Wallace and Webster 1996; Graça 2001). Higher diversity of detritivores tends to increase decomposition rates, which may be explained by facilitation and functional dissimilarly among species (Dangles and Malmqvist 2004; Srivastava et al. 2009). It is thus vital to also examine how these structural attributes can mediate observed changes in functional processes in headwaters following forest disturbances, such as OM processing and transport.   1.4 Forestry impacts on headwater streams: considerations of environmental drivers, multiple scales and casual mechanisms Forest harvesting is a dominant form of disturbance in headwater ecosystems. The removal of riparian vegetation can incur differential physical, chemical, and biological changes in streams, influencing water quality and quantity, and biota (see reviews by Campbell and Doeg 1989; Mellina and Hinch 2009; Richardson and Béraud 2014). Therefore, the responses of the structural and functional attributes of headwater systems can be quite complex, which may be associated with a highly variable strength of stream-terrestrial linkage in these systems (≤ 3 m bankfull width; Sakamaki and Richardson 2013). Furthermore, differences in forest harvesting practices, and their effects on the underlying environmental drivers of the response variables, can 7  add to a strong scale-dependency of response patterns (e.g., Kreutzweiser et al. 2010; Sweeney and Newbold 2014; Richardson and Béraud 2014).  Potential interactions among changes in these drivers can hamper efforts to generalise patterns of ecological stability across headwater streams. For instance, temperature, nutrient and sediment concentrations are demonstrated empirically to interact to produce non-additive responses of macroinvertebrate abundances (e.g., Wagenhoff et al. 2011; Piggott et al. 2015a; Chará-Serna and Richardson 2017) and OM processing (Greig et al. 2011; Piggott et al. 2012). The extent of shifts from reference to disturbed conditions represent another dimension of complexity, as non-linear, drastic responses to disturbance would occur when ecological thresholds are crossed; for instance, the loss in macroinvertebrate richness can be exacerbated when nutrients and turbidity levels exceed some thresholds (Evans-White et al. 2009). It is thus imperative to adopt a process-oriented approach to elucidate mechanisms underpinning the spatio-temporal differences in forest harvesting impacts on ecosystem structure and functions. Field-based correlative and manipulative studies, and process-oriented modelling studies, are suitable approaches to achieve this goal.  1.5 Thesis structure and objectives In this thesis, the effects of natural and forest harvesting disturbances on the stability of terrestrial-derived particulate OM dynamics (in terms of its quantity and processing rate) in small streams are investigated (see Fig. 1.3). My thesis focus is on mechanistically linking the spatial (i.e., across watersheds and geographic regions) and temporal (i.e., across years) variability of disturbance-induced responses of leaf-litter OM dynamics to their underlying biotic and abiotic drivers. I use a combination of field studies and process-based modelling to quantitatively evaluate how resistant and resilient OM dynamics are to watershed disturbances.  I first give introductory remarks on the concept of ecological stability, and discuss its relevance to disturbance-induced ecological responses in small streams (this Chapter). In Chapter 2, I review several conceptual frameworks that are formulated to bivariately map and compare quantifiable components of ecological stability across ecosystems and studies. I make recommendations to further operationalise these frameworks to accommodate more diverse disturbance-response dynamics, which can be applied to describe the stability of a wide range of 8  ecological variables, and not just limited to stream OM dynamics. In this chapter, I highlight that establishing the natural variability, or baseline conditions, of ecological variables is fundamental to accurately determining components of stability (particularly those related to recovery). Therefore, in the subsequent data chapters, I first describe studies that address the range and causes of natural variability of stream OM dynamics. I then report studies that examine the resistance- and resilience-related components of OM dynamics in streams where data on baseline conditions are available or can be modelled. In Chapter 3, I empirically establish the range of natural variability of the breakdown rates of riparian litter in unharvested perennial streams in temperate regions. In particular, I assess the contributions of interannual, weather-driven hydrologic variation to that of total and microbe-mediated litter breakdown. I hypothesise that physical abrasion – a component of total breakdown – increases with the frequency and duration of high-flow events, whereas microbial decomposition of litter tends to be unrelated to discharge variability. In Chapter 4, I examine how the choice of litter species could influence the results of litter breakdown responses to disturbances shown by harmonised broad-scale investigations. Specifically, I determine how differences in litter origin (i.e., native vs. exotic) and quality can modulate the spatial variability of litter breakdown rates, using reciprocal transplant incubations in streams in non-harvested catchments across geographic regions. I hypothesise that the adaptation of shredders to processing locally-derived litter resources is weaker than that of microbes. This is because shredders are known to prefer to consume high-quality litter resources, regardless of whether the litter is derived locally or from elsewhere. In Chapter 5, I make projections of the resistance of stream coarse particulate OM (CPOM) standing stocks to variable forest harvesting impacts, using a published process-oriented CPOM model developed for small streams. I examine the independent and interactive effects of major biophysical controls on CPOM quantity, along a realistic severity gradient of forest harvesting disturbance. In Chapter 6, I assess the signs of recovery of litter breakdown and shredders to various contemporary riparian forest management practices, by following up on a previous study which still detected harvesting impacts on these ecological variables. I hypothesise that the ecological variables examined are more resilient to the effects of riparian thinning (removing 9  50% of basal area of trees), than those of clearcutting with or without fixed-width (10 m and 30 m) riparian buffers. The empirical and process-based modelling studies are based in Malcolm Knapp Research Forest (MKRF), coastal British Columbia, Canada. Past timber harvests in the rainforests in MKRF with diverse riparian management practices allow the comparison of OM responses along a gradient of forest harvesting disturbance and histories. Streams in the mixed hardwood forests in Turkey Lakes Watershed and in the boreal mixedwoods of the White River Watershed, central Ontario, Canada, are also included in the field studies to widen the scope of investigations of broad-scale differences in OM responses to disturbances. Finally, in Chapter 7, I conclude by discussing the relevance of my work to the understanding of key environmental drivers that mediate the availability and fate of terrestrial-derived OM subsidies in the recipient environment, and more broadly, the resistance and resilience of small streams to disturbances. I also remark on the utility of study findings (especially on litter breakdown rate) in guiding forest and watershed management to assess and mitigate the effects of land-use change and disturbances on small stream ecosystems.10   Figure 1.1 Major types of disturbance-response patterns of ecosystems. The onset and end of disturbance are indicated by vertical, grey dotted lines. Three common types of relationships between disturbance (dotted lines) and response variables (solid lines) are illustrated: (a) a pulse disturbance may lead to a pulse response; (b) a pulse disturbance may lead to a press response; (c) a press disturbance may lead to a press response (modified from Lake 2000). The response variable may return to pre-disturbance levels (i.e., asymmetric patterns of trajectories indicating hysteresis; dash-dot line 1), or may remain at persistently lower (or higher) levels, if the system reaches and remains in an alternative stable state, and/or if baseline conditions shift (dash-dot line 2). Ramp disturbance and response that change steadily with time without clear endpoints are not illustrated here. Note that the response variable can also follow a contrasting disturbance-induced trajectory; that is, it increases during disturbance, and decreases afterwards to reach predisturbance state (see Chapter 2.3). 11   Figure 1.2 Schematic diagram illustrating engineering and ecological resilience. Valleys represent stability domains, balls represent the system and arrows represent different levels of disturbances. Engineering resilience is a measure of the time taken for the system to return to pre-disturbance levels (t) within (or back to) the original stability domain (A), with the length of the upward arrow (solid line) indicating resistance to the disturbance. Ecological resilience is a large-scale topographical measure of a specified stability domain. The amount of disturbance (dotted line) required for the transition of a system to an alternative stable state (from A to B) depends on how close the system is to the threshold (i.e., precariousness). Note this is a simplified graphical representation (modified from Gunderson 2000), as transitions are not always reversible, and the topography of the valleys may not be static over time. 12   Figure 1.3 Hypothetical trajectory of the response variable y before and after disturbance, and quantifiable components of ecological stability associated with the disturbance-induced change in y. The dotted line indicates the end of the disturbance. The difference between yt0 (at baseline conditions, and before disturbance) at t0 and yti (at the end of disturbance) at ti is the magnitude of change, and the inverse of which represents resistance (as defined in section 1.1). The response variable recovers to ytr within its range of natural variability at tr. tr – ti indicates the recovery time, and the inverse of which represents resilience. The response variable is considered to have recovered to its pre-disturbance levels when it reaches yt2 at t2, but not yt1 at t1. The focal components of ecological stability investigated in the thesis chapters are indicated.  13  Chapter 2: Towards standardising comparisons of ecological stability across ecosystems: conceptual and operational considerations  2.1 Introduction The concept of ecological stability is increasingly applied in managing complex socio-ecological systems for the goods and services they provide. However, ecological stability is a dynamic multifaceted concept with many quantifiable dimensions (Grimm and Wissel 1997; Donohue et al. 2013). While measurements of multiple stability properties allow a more holistic understanding of ecological dynamics in response to disturbances, in practice a reductionist approach of quantifying single or a few properties is usually adopted given that their measurements can be difficult and/or prohibitively costly in empirical conditions (Egli et al. 2018). One major challenge to operationalise conceptual stability properties is therefore centred on integrating holistic and reductionist approaches to characterising stability to widen its applications in environmental management. Numerous bivariate frameworks have been proposed to facilitate comparisons of ecological stability across ecosystems and studies, in which two major stability properties of ecological state variables are simultaneously mapped and quantified (Hodgson et al. 2015; Nimmo et al. 2015; Ingrisch and Bahn 2018). These three noted frameworks both address resistance (i.e., magnitude of change of the state variable) and recovery (magnitude or rate of recovery) after the disturbance ends, but with differences in the methods of quantifying these stability properties. These frameworks also differ in how concepts of ‘engineering resilience’ (Lake 2013) and ‘ecological resilience’ (Walker et al. 2004) are integrated into their stability comparisons, in terms of how changes of state variable within and across stability domains (or alternative stable states) are quantitatively described. For instance, Ingrisch and Bahn (2018) suggested jointly considering percentage reduction in state (= disturbance impact) and recovery rate (both normalised to the predisturbed or undisturbed state of the ecosystem) under disturbances within a stability domain. The stability properties bivariately represented in Hodgson et al. (2015) are absolute change in state and return time after disturbance, considered 14  in the context of multiple alternative stable states. Ecosystems shifted to new alternative stable states and not showing recovery to the predisturbance state can be suitably included for comparisons using this framework. Here, I present two key conceptual and operational considerations for establishing appropriate stability comparisons when using the bivariate frameworks. These considerations pertain to developing comprehensive understanding of the nature of disturbance(s) acting on the ecosystem in question, and characterization of baseline conditions of the ecosystem. In addition, I recommend adjustments to specific components of the frameworks of Hodgson et al. (2015) and Ingrisch and Bahn (2018). These adjustments are intended to demonstrate how such frameworks can be devised to address a wider range of ecological responses and management needs to broaden their applications in environmental monitoring and restoration.  2.2 General considerations When taking measurements of stability properties, I suggest to characterise the temporal pattern of the given exogenous disturbance on the ecosystem (Lake 2000). For use in the proposed bivariate frameworks, these measurements are applicable when targeting pulse disturbances, but not press and ramp disturbances that remain persistent (e.g. spread of exotic species, dam construction, or river channelization). This is because the disturbance in question has to cease or become substantially alleviated for the measurements of recovery (or return) time, rate, or the capacity to recover from recovery (Lake 2013; Nimmo et al. 2015). It should be made clear whether the disturbance refers to changes in a particular abiotic and/or biotic variable, or a natural (e.g. storms, wildfires, or hurricanes) and human-related event (e.g. forest harvesting, pesticide application, or bottom trawling) that can encompass multiple variables. For the latter, attention should be paid to the variables affected, and their possible non-additive interactions (Darling and Côté 2008; Piggott et al. 2015b). This is because the change in state will not necessarily increase (or decrease) monotonically with the magnitude of disturbance, as assumed in Hodgson et al. (2015) (see also Zhang et al. 2015). Accurate measurements of the predisturbance state of the ecosystem (or range of variability of state irrespective of disturbance), is prerequisite to determining the change of state for stability comparisons. However, these measurements can be hampered by shifting baselines 15  that pervade environmental management globally, due to ongoing climate change, and intensifying anthropogenic disturbances (e.g., Pauly 1995; Steen and Jachowski 2013; Donohue et al. 2016). Some disturbed ecosystems will unlikely recover to the predisturbance state, when their baselines are shifting concomitantly under additional disturbances, and/or when these ecosystems undergo (irreversible) shifts to alternative stable states after disturbance. When the extent of baseline shifts is known for these ecosystems (or reference systems to be compared against) during and/or after disturbances which still remain in the same stability domain, disturbance impact and recovery rate can be normalised relative to the new baseline (e.g., using the framework of Ingrisch and Bahn (2018); Fig. 2.1). Further, many ecosystems are affected by successive disturbances (e.g., multiple pulse, or ramp disturbances) with longer recovery time than disturbance frequencies (Nimmo et al. 2015). Therefore, the state before the most recent disturbance may not be the ‘true’ baseline, which needs to be modelled or hindcasted to more accurately determine recovery rate. Data deficiency on baselines may arise when current reference systems fail to reflect the undisturbed state, or no longer exist. In such cases, stability properties need to be normalised relative to the state estimated or desired by researchers and managers as the ‘working baseline’. Alternatively, to include ecosystems exhibiting only partial recovery after disturbances for stability comparisons, the measures of return time (Hodgson et al. 2015) and baseline-normalised recovery rate (Ingrisch and Bahn 2018) can be substituted by another component of change, such as the capacity to recover from disturbance (i.e., completeness of recovery; Nimmo et al. 2015). This representation does not assume full recovery, and would be more suitable for tracking the long-term stability of systems affected by frequent, multiple disturbances (see Fig. 2 in Nimmo et al. (2015)).  2.3 Framework-specific considerations In the framework of Hodgson et al. (2015), when comparing changes in state across ecosystems, the predisturbance state of ecosystems can differ in its position within the stability domain. Thus, the same magnitude of disturbance may or may not trigger shifts to the alternative stable state. A given position of the post-disturbance state on the bivariate stability space does not give a good indication of how close it is to the nearest threshold (i.e. precariousness) relative to other ecosystems, particularly if the ecosystems considered vary widely in their positions of 16  thresholds. Therefore, to avoid potential ambiguity in comparing the stability of ecosystems with well-supported existence of alternative stable states, such as coral reefs (Knowlton 1992), shallow lakes (Capon et al. 2015), and savannah-forest ecosystems (Hirota et al. 2011), I suggest a mutual consideration of precariousness and the two stability properties. An ecosystem with high precariousness should deserve management attention, because it may be tipped into undesired alternative stable states with a small amount of disturbance. Also, an ecosystem close to a tipping point of state shifts is commonly predicted to require longer time to recover from small perturbations of the state – a phenomenon called ‘critical slowing down’ (CSD; also shown in Fig. 2C in Hodgson et al. (2015)). Indeed, CSD-based indicators have been demonstrated to detect state shifts in advance in some ecosystems, which illustrates the practicality of this concept (Dakos et al. 2015; Scheffer et al. 2015). In the framework of Ingrisch and Bahn (2018), to normalise stability properties relative to the new (or estimated) baseline, modifications of the equations for calculating the stability properties are necessary, which are illustrated in Fig. 2.1a. In a hypothetical example of disturbance-induced ecosystem change, the predisturbance baseline of the ecosystem (St0) has a 1:1 relationship with that of the reference system (Rt0). Following reference system trajectory 1, when the disturbed ecosystem reaches Sti at the end of disturbance, its baseline changes to Rti, which is also the new baseline at time tx (= Rtx). Hence, baseline-normalised impact (∆Sti(base)) =  𝑆𝑡0−𝑆𝑡𝑖𝑅𝑡𝑥=  100−20𝑅𝑡𝑥=  80𝑅𝑡𝑥, and the rate of recovery normalised to the new baseline (Rbase) =𝑅𝑅𝑡𝑥 where 𝑅 =𝑆𝑡𝑥−𝑆𝑡𝑖𝑡𝑥−𝑡𝑖. Following reference system trajectory 2, the baseline of the disturbed ecosystem decreases further to R’tx at time tx, after the end of disturbance. Rtx can be substituted by the average of R’tx values measured in the reference system from ti to tx to generate an approximate Rbase. The time taken for partial recovery is given by tx – ti (Fig. 2.1a). The framework of Ingrisch and Bahn (2018) can also be adjusted to address a contrasting, but common, type of disturbance-induced trajectory to those depicted in the framework. Ecosystem state variables can show increases (rather than decreases) during the period of disturbance (see Steen and Jachowski 2013), followed by decreases post-disturbance to eventually reach the predisturbance disturbed state (i.e., a ‘concave downward trajectory’). For instance, during a warming-induced coral bleaching event in the Great Barrier Reef, a seaweed 17  (Lobophora variegata) bloom was observed and declined afterwards as corals re-established (Diaz-Pulido et al. 2009). The flood response of nitrate uptake length of a desert stream followed a similar trajectory (Martí et al. 1997). For this type of trajectory, both baseline-normalised impact (i.e., 𝑆𝑡0−𝑆𝑡𝑖𝑆𝑡0; see Fig. 2.1a) and recovery rate (𝑆𝑡𝑥−𝑆𝑡𝑖(𝑡𝑥−𝑡𝑖)𝑆𝑡0) are negative. This is because the difference between the ecosystem state before and after disturbance (St0 – Sti), and between a particular time during the recovery period and the end of disturbance (Stx – Sti), is negative, respectively (see supplementary material in Ingrisch and Bahn (2018)). Hence, these stability properties cannot be properly mapped onto Ingrisch and Bahn’s framework with positive axis values, and inferences about recovery time and perturbation (or ‘recovery debt’, sensu Moreno-Mateos et al. (2017)) cannot be made. A ‘concave downward’ trajectory can be transformed to become ‘concave upward’ by reflecting it along y = (St0 + Sti)/2. The transformed state at time tx (Stn) equals St0 + Sti – Stx, and the original equations in Ingrisch and Bahn (2018) remain valid for computing and mapping baseline-normalised impact and recovery rate for this transformed trajectory. Moreover, for ecosystems showing partial recovery or overcompensation (i.e., exceeding the baseline; Hillebrand et al. (2018)), or ‘net change’ (sensu Nimmo et al. (2015)) relative to the new baseline of such systems (i.e., Stx/Rtx or Stx/R’tx in Fig. 2.1a), these features can be represented by varying the appearance of data points (e.g., shape, size, colour) mapped on Fig. 2 in Ingrisch and Bahn (2018). In ecosystems with well-defined management and/or conservation targets, it is also informative to determine the proximity of the disturbed ecosystem state to such targets (e.g., management target ST in Fig. 2.1b). These targets are measurable thresholds, and management measures can be triggered below minimum (or above maximum) acceptable levels of specific environmental conditions (e.g., species richness, population size, index of ecological integrity; Samhouri et al. 2012). Such management measures are needed to shorten the duration the system exists in an undesirable state (i.e, tx – ti in Fig. 2.1b), even if the baseline-normalised impact is small. Magnitude and sign of proximity to management targets can be mutually considered alongside important stability properties.  18  2.4 Conclusion Bivariate frameworks, such as those proposed by Hodgson et al. (2015) and Ingrisch and Bahn (2018), provide easily understandable representations of the stability of ecosystems. I highlighted major limitations in real-life applications of the various approaches undertaken for reducing the representation to fewer dimensions of ecological stability (see also Scheffer et al. 2015). Adjustments of some components of these frameworks are necessary to enable them to address a wider range of ecosystem response dynamics and management concerns in stability comparisons. It is important to recognise that no single bivariate framework can serve to represent and compare stability in a way to address diverse management needs. Given the multidimensional nature of ecological stability (Donohue et al. 2013), other properties (e.g., ‘net change’, degree of temporal fluctuations, proximity to management target), not considered in the three noted bivariate frameworks can also be of relevance to stability-based management. However, these components may not be well correlated with the stability properties chosen by the noted frameworks, such as disturbance impact and recovery rate (see Hillebrand et al. 2018; Radchuk et al. 2019). Therefore, to comprehensively understand and practically manage for ecological stability, I recommend a flexible adoption of a bivariate framework, and quantify additional, relevant measures of stability properties.19    Figure 2.1 Disturbance responses and resilience components of an ecosystem in two common management scenarios. Hypothetical disturbance-induced change in an ecosystem relative to the predisturbance state at time t0 is baseline-normalised (Sbase). Baseline-normalised impact (∆Sti(base)) is expressed as the percentage of baseline state at the end of disturbance (time ti), as shown in Figure IB in Ingrisch and Bahn (2018). The orange boxes indicate the occurrence of disturbance. (a) The baseline state of the reference (control) system decreases to Rtx (= Rti) for trajectories 1 and 2 at time ti, which remains unchanged from time ti to tx for trajectory 1, and further decreases to R’tx for trajectory 2. The disturbed ecosystem reaches a steady state at time tx and does not return to the predisturbance state, due to its shifting baseline. (b) The disturbed ecosystem reaches Sti (above management target ST, a minimum acceptable level of an environmental condition) for trajectory 1 and S’ti (below ST) for trajectory 2 at time ti. Therefore, the proximity to management target at ti is positive for trajectory 1 (P; 20%), and negative for trajectory 2 (P’; -40%). The baseline-normalised impact for trajectory 2 is greater than that of trajectory 1, and both reach the predisturbance state at time tr with identical recovery time in this example. 20  Chapter 3: Relations of interannual differences in stream litter breakdown with discharge: bioassessment implications  3.1 Introduction The inclusion of key ecosystem-level processes that reflect ecosystem functional integrity is recommended to complement the use of structural biological attributes in bioassessment schemes, particularly in streams (Gessner and Chauvet 2002; Young et al. 2008; Clapcott et al. 2012). Litter breakdown plays a pivotal role in regulating the availability and transport of allochthonous organic matter that supports stream foodwebs and energetics (Gessner and Chauvet 2002). As litter breakdown is sensitive to environmental stressors and relatively easy to measure, it is one of the most widely used functional indicators to assess the impacts of anthropogenic disturbances in watersheds (von Schiller et al. 2017), such as forestry activities (see references in Chauvet et al. 2016), nutrient and pesticide contamination (Woodward et al. 2012; Brosed et al. 2016), flow regulation (Mollá et al. 2017), and urbanization (Chadwick et al. 2006). However, litter breakdown rates do not respond equivalently to the same stressor type in terms of direction and/or magnitude (Ferreira et al. 2015, 2016b; Yeung et al. 2017). Concomitant changes in naturally varying abiotic factors (e.g., temperature, water chemistry, discharge) can induce high variability of litter breakdown even in undisturbed streams, potentially masking the effects associated with stressors in impacted streams. The current criteria for linking litter breakdown rates and stream integrity are partly derived from limited information on the former’s natural variability across space and time. For instance, Gessner and Chauvet (2002) tentatively suggested a range of breakdown rates (i.e., 75-133% around the mean of local reference streams; 50-200% around the mean at the regional scale), and a ratio of 1.2-1.5 for breakdown rates in coarse- relative to fine-mesh bags, to indicate ‘no impacts’ on ecosystem functioning. Nevertheless, these recommended criteria were established based on data from a few sites unaffected by human disturbances, and also not explicitly linked to any potential confounding factors. Moreover, the setting of these benchmarks has not incorporated the extent 21  of interannual or regional variability of litter breakdown rates (Pozo et al. 2011). Indeed, litter breakdown studies are not commonly repeated on an annual basis (Chauvet et al. 2016). Importantly, the same assessment benchmarks may not be robust to variation between years, which could be problematic for routine monitoring and impact assessments.  Litter breakdown rates in temperate, forested headwater streams are known to differ markedly across years (e.g., Jonsson et al. 2001; Dangles et al. 2004; Kreutzweiser et al. 2010; Yeung et al. 2017). Weather-driven, year-to-year hydrologic variations can be an important top-down driver of stream litter breakdown (Graça et al. 2015), through altering decomposer communities, organic matter availability and distribution, and water chemistry (e.g., Negishi and Richardson 2006, Feller 2010, Stenroth et al. 2014). However, there is a lack of quantitative understanding about the effects of interannual variability of hydrologic regime on litter breakdown, which depend on the differential responses of key mechanistic pathways of litter breakdown. These pathways include physical abrasion (by hydraulic forces and transport of sediments), feeding of detritivorous invertebrates (shredders), microbial decomposition, and dissolution. The duration of hydrologic influences are probably not only limited to within the period of leaf bag incubation, because hydrologic patterns preceding incubation (i.e., in the preconditioning phase) can also set the conditions for the development of benthic communities and resource availability, thereby leaving legacy effects on litter breakdown (see Peckarsky et al. 2015; Arroita et al. 2018). Hydrologic effects on litter breakdown rates are expected to be mainly controlled by fragmentation (through mechanical abrasion and shredder feeding), and less by microbial decomposition and dissolution. This is because (1) shredder communities are known to vary markedly with hydrologic variability (Kreutzweiser et al. 2010; Imberger et al. 2016); (2) shredders tend to exert much greater control on breakdown than microbes (Hieber and Gessner 2002; Kreutzweiser et al. 2010; Kominoski et al. 2011; Lecerf 2017); and (3) the contribution of microbial decomposition to breakdown was shown to vary little with current velocity (Ferreira et al. 2006). Extreme hydrologic events (e.g., spates and low-flow periods) may have varying consequences on litter breakdown rates, despite the strong association of these events with the shredder communities (Negishi and Richardson 2006; Booker et al. 2015; Patrick and Yuan 2017). For instance, spates could result in differential extent of reductions in shredder 22  abundances and/or organic matter standing stocks, depending on the characteristics of spates (e.g., magnitude, frequency, duration) and antecedent flows (Snyder and Johnson 2006). Extended low flows could lead to altered abundance, activity and feeding patterns of shredders, and litter availability (e.g., Leberfinger et al. 2010; Jeanette and Michael 2011; Northington and Webster 2017). These responses could also be mediated by other environmental conditions, such as the timing and magnitude of litter inputs, and availability of hydraulic refuge patches (Negishi and Richardson 2006). The interplay of these processes thus regulates the responses of litter breakdown rates to hydrologic variations (Tiegs et al. 2008). The primary goal of this study was to examine how litter breakdown rates in perennial, small forest streams varied with interannual differences in hydrologic conditions, both during and before leaf bag incubations. Multi-year measurements of litter breakdown rates were undertaken in the same sites across distinct geographical regions of similar latitude. The study was therefore a natural experiment that relied on regional disparities in weather patterns to induce differences in the extent of temporal hydrologic variation, while minimising the potential influences of latitudinal differences between regions on the relative contributions of shredders and microbes to litter breakdown (Boyero et al. 2011). The relative roles of fragmentation, and microbial decomposition and dissolution were elucidated by measuring litter breakdown rates in coarse- (kc) and fine-mesh (kf) leaf bags. I hypothesised that, at the site level (1) fragmentation rate (λF) would tend to scale positively with the magnitude of hydraulic forces (i.e., frequency and duration of high-flow pulses), when the direct hydraulic effects on enhancing abrasion exceeds the indirect effects on reducing shredder populations; and (2) dissolution and microbial decomposition rate (λm) would be generally invariant along the gradient of hydraulic forces. The importance of hydrologic characteristics relative to variables associated with decomposers and water chemistry in driving the variability of litter breakdown rates was evaluated by variance partitioning. Furthermore, the spatio-temporal variability of litter breakdown rates, as well as the ratio of breakdown rates (i.e., kc/kf, λF/λm), within study regions was quantified. Results would help establish regional and site-specific ranges of the natural variability of litter breakdown rates, in order to refine bioassessment benchmarks.    23  3.2 Methods 3.2.1 Study region and site characteristics This study was conducted in three geographically separate regions in the temperate zone of Canada. They include the University of British Columbia’s Malcolm Knapp Research Forest (MKRF) in British Columbia, and the Turkey Lakes Watershed (TLW) and White River Forest Management Area (WR) in Ontario (Fig. 3.1). MKRF is located in the Pacific Maritimes ecozone approximately 60 km east of Vancouver, and has a moderate oceanic climate (Köppen-Geiger climate classification Cfb), bordering on a warm-summer Mediterranean climate (Csb) (Peel et al. 2007). TLW lies on the Boreal Shield ecozone about 60 km north of Sault Ste. Marie. WR is also located on the Boreal Shield, about 150 km away from TLW, and 75 km inland from the northeastern shore of Lake Superior. Both TLW and WR have a warm-summer humid continental climate (Dfb). These study regions typically have wet early-fall and winter months. The summer months (July and August) of MKRF are generally dry, often resulting in low-flow conditions, whereas the precipitation at TLW and WR is fairly evenly distributed across seasons. The precipitation and discharge regimes of the study catchments are known to vary substantially across years (Kiffney et al. 2002; Foster et al. 2005; Kreutzweiser et al. 2009a). The age of most forest stands in the study catchments, at the time of this study, ranged from around 85 years in MKRF (Kiffney and Richardson 2010), ~20 years or 140-200 years (Creed and Band 1998) in TLW, and ~40-50 years in WR (Musetta-Lambert et al. 2017). For additional descriptions of the study regions in terms of climate, geology, vegetation, etc., refer to Kiffney and Richardson (2010) for MKRF, Foster et al. (2005) for TLW, and Kreutzweiser et al. (2009a) for WR. Within each region, sampling was undertaken in four to five small (1st- to 3rd-order), forested streams (Appendix A.1: Fig. A.1a-c) over three or four consecutive (2014-2017), or near-consecutive (2009-2010, 2014) years. During the study period (early autumn) in all regions, I observed that interannual hydrologic variability did not cause complete surficial streambed drying, nor catastrophic debris flows that could remove riparian vegetation from large portions of the stream networks. Most of the study sites were minimally affected by fires and forest harvesting in catchments and in the adjacent riparian areas in the past 20 years (see Table 3.1). G-4 stream in MKRF was an exception, where riparian clear-cutting occurred about 10 m 24  downstream of the site in 2015 and substantial surficial drying occurred in late September, 2017; thus at G-4, only data from 2014 and 2016 were used. One TLW site (TLW34) affected by selection-based logging in its catchment 17 years prior to the first study year included partial-harvest (at least 60% retention) riparian buffers along the study reach. Stream temperature during the incubation of leaf bags was comparable across years (MKRF: 9.2-11.4°C; TLW: 7.6-10.6°C; WR: 7.3-8.0°C). Within individual streams, the interannual difference in average daily temperature was within the range of 1.9-2.5°C in MKRF, 2.2-3.7°C in TLW, and 0.6-1.1°C in WR.  3.2.2 Field sampling 3.2.2.1 Litter breakdown Coarse- (10 mm mesh size; 5 mm for WR) and fine-mesh (0.5 mm) leaf bags were incubated along 30 to 40-m reaches of all study streams to determine litter breakdown rates. Prior to deployment, leaves of red alder (Alnus rubra Bong.) and speckled alder (Alnus incana ssp. rugosa (Du Roi) Clausen) had been collected at senescence, air-dried to constant mass, and stored at room temperature. Speckled alder leaves were additionally pre-leached in water for 24 h, oven-dried at 30°C for 24 h before storage. Leaves of red alder and speckled alder were used in MKRF and WR, respectively, as these deciduous tree species are commonly present in the riparian vegetation of the study streams. In TLW, leaves of foreign red alder − highly palatable to decomposers − were used in lieu of native speckled alder. Therefore, the spatio-temporal variability of breakdown rates in TLW would be influenced by fluctuations in decomposers’ abundance and activities as in the other regions, and not a reduced preference for red alder (see also Boyero et al. 2011). The mesh size of leaf bags, and the procedures of constructing and handling them were identical in MKRF and TLW, whereas some of them differed from those in WR (see following paragraphs). Three coarse- and three fine-mesh bags, each containing 4 ± 0.01 g of red alder leaves, were deployed in each study stream. In WR, seven coarse-mesh bags, each enclosing 10 ± 0.01 g of speckled alder leaves (4 g in 2014), and seven fine-mesh bags, each enclosing ten pre-weighed 23-mm-diameter speckled alder leaf disks, were used in each stream. In WR, no fine-mesh bags were deployed in the third year of study (i.e., 2014). Placement of leaf bags in streams at all sites 25  occurred in early autumn (mid-September to early October), which coincided with the period of natural litterfall, and were retrieved 4-6 weeks after incubation. This length of incubation period allowed for sufficient breakdown (>20% mass loss), which enabled the testing of possible differences in breakdown rates between sites and years. Each coarse-mesh bag was placed alongside a fine-mesh bag, the latter excluding the access to litter by large-bodied invertebrate shredders. Some meiobenthic decomposers (e.g., nematodes) could be present in the fine-mesh bags, but previous work has shown that meiofauna density is unrelated to litter breakdown inside these bags (Majdi et al. 2015). Upon retrieval, each leaf bag was sealed in a plastic bag containing some stream water, stored on ice (whole leaf bag preserved in 5% formalin in WR) and transported to the laboratory. The leaf bags in MKRF and TLW were frozen at -20°C until later processing. Leaf materials from each bag were thawed, and gently rinsed with tap water onto a 250-µm sieve to remove invertebrates and sediments from leaf surfaces. In WR, leaves removed from coarse-mesh bags were separated from invertebrates and sediments by elutriation using a 250-µm sieve (Kreutzweiser et al. 2008a). Invertebrates retained on the sieves were stored in ethanol for later enumeration. Two sets of five 9.5-mm leaf disks were randomly cut from leaves, using a cork borer, from each coarse-mesh leaf bag in MKRF and TLW. Central veins were avoided when removing leaf disks. One set was frozen at -20°C until analysis of ergosterol – a surrogate for fungal biomass in decomposing leaves (Gessner 2005). The other set of disks, as well as leaf materials from leaf bags in all regions, were oven-dried at 60°C to a constant mass and weighed to obtain dry mass (DM). Leaf materials in MKRF and TLW were further ashed at 550°C for 4 h in a muffle furnace, and reweighed to determine ash-free dry mass (AFDM), by subtracting ash mass from DM. Final AFDM (DM in WR) was corrected by the mass of leaf materials from handling losses (and leaf disks removed for fungal and AFDM determinations), prior to the calculations of kc and kf.  3.2.2.2 Shredders and fungal biomass Macroinvertebrates associated with the coarse-mesh leaf bags were counted and identified to the lowest practical taxonomic resolution (at family or genus level), except for Chironomidae in which subgroups corresponding to subfamily or tribe were identified (e.g., Orthocladiinae, 26  Chironomini, Tanytarsini, Tanypodinae). The assignment of macroinvertebrates to shredders was based on Merritt et al. (2008). Shredder density at the site level was obtained by dividing the average shredder abundance by the average final leaf AFDM across leaf bags. The density of Lepidostoma and Micrasema caddisflies (as numerically dominant shredder taxa), and Tipula craneflies and Pteronarcys stoneflies (as large-bodied taxa with relatively high contribution to total shredder biomass) were also calculated, whose feeding possibly accounted for an appreciable proportion of shredder effects on litter breakdown (e.g., Ruesink and Srivastava 2001; Lecerf and Richardson 2011; Andrushchenko et al. 2017; Tonin et al. 2018a). The procedures of determining the ergosterol content for each set of frozen leaf disks followed Gessner (2005). Briefly, lipids were extracted from leaf disks by heating (30 min, 80 °C) in 0.8 % KOH-methanol, and the extract was purified by solid-phase extraction (using Sep-Pak C18 Vac RC cartridges, 500 mg, 3 cc; Waters®, Milford, MA, USA) and eluted in isopropanol. Ergosterol extracts were then quantified by high-performance liquid chromatography. Fungal biomass was expressed as ergosterol mass per gram of leaf AFDM. Stream discharge.— In MKRF, daily discharge (m3/s) of the study streams from 2014-2017 was simulated using a process-based, coupled hydrology and temperature model, which was previously developed for adjacent forested catchments within the same study region (Leach and Moore 2015, 2017). This model was calibrated and evaluated against field data, and generally captured the timing of high-flow events and low-flow conditions but might underestimate peakflow magnitude (see model fit and comparisons between observed and simulated discharge in Upper East in Leach and Moore 2017). An adjustment of several model parameters was necessary to account for differences in catchment characteristics among study streams. These parameters included catchment area, mean elevation and mean hillslope for the catchment determined from either 1-m or 20-m digital elevation models. Thirty time series (= iterations) of discharge were generated for each study stream based on a generalised likelihood uncertainty estimate approach. Mean discharge specific to each date of the year was obtained by averaging across all the 30 iterations for each stream. In TLW, discharge in study streams was modelled based on the Regional Hydro-Ecological Simulations System (RHESSys). This hydrologic model has been well calibrated based on the discharge data in a subset of gauged streams in the Batchawana River Watershed in 27  central Ontario, and the study streams were within this Watershed (Sanford et al. 2007). The RHESSys model was evaluated to represent the timing and magnitude of observed discharge reasonably well, and monthly to seasonal flow metrics (e.g., duration of low/high-flow pulses) appeared to be better estimated by the model than metrics associated with daily discharge (see model fit between observed versus simulated discharge in the study and nearby streams in Sanford et al. 2007). Air temperature and precipitation data from 2014-2017 were obtained from nearby meteorological stations to run the simulations to generate daily discharge. In WR, water levels were recorded every two hours from April to November during 2009-2010 and in 2014, using level-loggers (Levelogger Edge model 3001; Solinst®, Georgetown, ON, Canada). Measurements of stream discharge were made at multiple (≥ 3) stage levels throughout each year at the approximate location of the level-loggers. Discharge (Q) was estimated using the area-velocity equation Q = AV, where A is the cross-sectional area of the stream, calculated by multiplying stream width by average water depth, at 6-10 points across the width of the stream, and V is average velocity, estimated with a flow meter at each point where depth was measured at 60% maximum depth (Gordon et al. 2004). Water level measurements were converted to discharge using linear rating curves established from stream discharge measurements, and discharge data were averaged for each day to give daily discharge.  3.2.2.3 Water chemistry and temperatures Water samples were collected during the retrieval of leaf bags (except for 2009 in WR), and analysed by the Canadian Forest Service water chemistry laboratory at Sault Ste. Marie, Ontario, following standardised methods (see procedures in Nicolson 1988; Hazlett et al. 2008). Water chemistry variables measured included pH, conductivity, total dissolved nitrogen (TDN), and total dissolved phosphorus (TDP). TDN and TDP were analysed by cadmium reduction and ascorbic acid reduction of phosphomolybdic acid, respectively, following autoclave digestion. Water temperature was recorded every one or two hours during the period of leaf pack incubation, using temperature loggers (MKRF and TLW: TidbiT v2; Onset®, Bourne, MA, USA; WR: Levelogger Edge model 3001; Solinst®, Georgetown, ON, Canada). Data were averaged to yield average daily temperature for each stream during the incubation period.  28  3.2.3 Data analyses 3.2.3.1 Litter breakdown rates Additional samples of unleached red alder litter were leached, oven-dried and ashed as for speckled alder litter (see section 3.2.2.1) to determine the percent loss in AFDM due to the initial release of water-soluble compounds. This percent of leaching correction (i.e., 20%) was applied to adjust the initial leaf litter AFDM of red alder (Pozo et al. 2011), in order to standardise the treatment of leaf litter across study regions. Additional release of solutes and dissolved organic matter from alder litter (or dissolution) could continue for a few more days after leaf bag incubation (see Gessner and Konstanz 1989; McArthur and Richardson 2002). Temperature-corrected litter breakdown rate in coarse-mesh (kc; total breakdown rate) and fine-mesh (kf) bags was calculated as decay rate coefficients, and averaged across leaf bags per site, using a first-order exponential decay model: mt = e-kt, where mt is the proportion of initial leaf litter AFDM, and t is the cumulative degree-days (in degrees Celcius) during the incubation period. The mean rates of litter breakdown through fragmentation (denoted as λF), and the combination of microbial decomposition and dissolution (λm), were computed following Lecerf (2017): 𝜆𝐹 =  𝑘𝑐 −𝑘𝑓 − 𝑘𝑐ln(𝑘𝑓) − ln(𝑘𝑐) 𝑘𝑐 =  𝜆𝐹 + 𝜆𝑚 All metrics of breakdown rate (kc, kf, λF, and λm) have the same unit (degree-day-1), and are thus comparable with each other. The approach by Lecerf (2017) assumes that litter breakdown is not constant through time, and that these pathways of litter breakdown are independent (e.g., the conditioning effects of microbial decomposers on fragmentation are not considered). This approach overcomes several shortcomings of previously used breakdown metrics in approximating and comparing pathway-specific litter breakdown rates. In WR, limited variability of site-specific kc/kf between 2009 and 2010 was observed (range of coefficient of variation: 2-33%; Appendix A.3: Table A.3.2). Therefore, kf at each site in this region in 2014 was calculated based on the average of kc/kf from these two years, allowing λF and λm to be estimated.  29  3.2.3.2 Hydrologic indices To explore the potential aspects of flow regimes both during and before leaf bag incubation, which were hypothesised to influence the spatio-temporal variability of litter breakdown rates, a group of 14 hydrologic indices was selected, and calculated for each site in each year (Table 3.2). These hydrologic indices were initially chosen to broadly reflect magnitude, frequency and duration of high- and low-flow events, and flashiness (Baker et al. 2004), which are representative and ecologically important components of flow regimes in perennial, runoff-driven streams (Clausen and Biggs 2000; Olden and Poff 2003; Kakouei et al. 2017; Poff 2018). Some of these indices are used as indicators of hydrological alteration (Richter et al. 1996). The beginning of the pre-incubation period was defined to be 30 days prior to incubation to capture the effects of recent flow conditions on decomposer communities.  Median daily discharge from May 1st to November 30th (July 1st/18th to November 30th for WR) averaged across the three study years (excluding the fourth year in MKRF and TLW) was obtained. It was considered to be the best available long-term summer-to-fall discharge estimate normal for the study regions, as the basis for characterising high- and low-flow events using the high.spells and low.spells functions, respectively, in the hydrostats package in R (Bond 2016). Daily discharge exceeding 3 times, and below 25% of, the 3-year mean of median daily discharge constituted high-flow and low-flow periods, respectively. Flow events of the same type were considered to be separate when they were at least 5 days apart. The frequency and duration of these flow events were first converted to percentage data by dividing each index by the number of days of incubation to standardise across sites or years for differences in study duration, and by 30 for indices for the pre-incubation period. These percentage data were then multiplied by [(D-x+1)/D], where x is the time difference between the last day of flow event(s) and the first day of incubation/pre-incubation period (in days; averaged in the case of multiple events), and D is the duration of incubation/pre-incubation period. This scaling factor incorporates the possible effects of the timing of extreme flow events on litter breakdown rates, as it controls the time available for decomposers to re-establish, or ‘time since community reset’ (Campbell et al. 2015). Adjusted by this scaling factor, more recent extreme flow events (i.e., smaller x), given the same frequency and duration, would be expected to more strongly influence decomposer (particularly shredder) communities than earlier flow events, as the former allow 30  less time for community reset. Base flow index was computed using the IHA package in R (Law 2013). Hydrologic indices for each of the 3 streams at TLW (i.e., TLW34, TLW96, TLW97) – generated based on RHESSys-modelled discharge – were highly similar across the first three years of study (coefficient of variation usually <10%). AY1-1 and AY4-1 – without modelled discharge – were in proximity to these 3 streams (~ 15 km away), with similar physical and biological habitat characteristics. Therefore, each hydrologic index at AY1-1 and AY4-1 for a given study year was assumed to be the average of the corresponding values at the other 3 study streams at TLW.  3.2.3.3 Statistical analyses Linear mixed-effects models (LMMs) were used to compare between kc and kf, and between λF and λm, averaged at the site level across study regions and years. Litter breakdown rate type and study region were the main factors, while site was a random factor nested within study region in this analysis. The magnitude of temporal variability of selected litter breakdown rates (i.e., kc, λF, and λm) at each site was expressed as (1) the inter-year coefficient of variation (CV), and (2) the relative range limits around the inter-year mean (i.e., enclosed by the minimum and maximum ratios of breakdown rates for a given year to the inter-year mean). For each year, the spatial variability of litter breakdown rates at each region was expressed as (1) inter-site CV, and (2) the relative range limits around the inter-site mean (i.e., enclosed by the minimum and maximum ratios of breakdown rates at a given site to the inter-site mean). Mean values of inter-year CV (herein referred to as temporal CV) were compared among litter breakdown rates and study regions (as main factors) using a two-way analysis of variance (ANOVA). For the analysis of inter-site CV (herein spatial CV), LMMs were fitted and included breakdown rate type and region as main factors, and year as a random factor. LMMs were undertaken using the lmer function in the lme4 package (Bates et al. 2017). The significance of main factors was estimated using an approximate F-test based on the Kenward-Roger approach implemented using the anova function in the lmerTest package (Luke 2017). For LMMs, P values were adjusted for multiple comparisons using a Holm’s correction (Holm 1979). Significant main effects (P < 0.05) were analysed by Tukey’s HSD post-hoc tests.  31  To further quantify the combined spatial and temporal variability of litter breakdown rates attributable to the influences by hydrology, shredders (not for λm) and fungal decomposers, and water chemistry, redundancy analysis (RDA) was undertaken (Borcard et al. 2011) with the varpart function in vegan (Oksanen et al. 2017). Prior to RDA, forward selection was undertaken to select significant variables for each group of variables (see Table 3.2) that could explain litter breakdown rate, using 2 stopping rules with the packfor package based on Blanchet et al. (2008). Relationships between hydrology and litter breakdown rates were further explored by a global RDA, using only forward-selected hydrologic indices. The fitted site scores on the first RDA axis (i.e., ‘site constraints’; hereinafter referred to as ‘composite hydrologic index’) were considered as a composite measure of important hydrologic indices that were strongly associated with litter breakdown. The regression coefficients from forward-selected hydrologic indices to composite hydrologic index were extracted, which were then used for calculating the composite hydrologic indices of the sites at MKRF and TLW for the fourth study year. For each region, LMMs were constructed to test for inter-annual differences in composite hydrologic index, with year as the main factor, and site as a random factor. The significance of regression relationships between litter breakdown rate and composite hydrologic index was also tested for each region, while controlling for the random effects of site. Details about procedures for RDAs are provided in Appendix A.2. For each region, community dissimilarity of shredders at study streams between years was visualised using non-metric multidimensional scaling (NMDS) graphs. One-way permutational multivariate analysis of variance (PERMANOVA) was performed to assess the significance of differences in shredder assemblages between years. Site-level shredder density data were used to generate the Bray-Curtis similarity matrix as the basis for PERMANOVA (999 permutations), using the adonis function in the vegan package (Oksanen et al. 2017). Within PERMANOVA, year was the main factor, and sites were used as strata to ensure that randomizations were only made within each site, given the differences in shredder assemblages among sites. Inter-annual differences in shredder assemblages were further assessed by pairwise PERMANOVA, and a sequential Bonferroni correction was used to adjust P values for post hoc pairwise comparisons.  32  Prior to all analyses, data were ln(x+1)- or arcsine-transformed as appropriate to meet assumptions of normality. All data analyses were carried out using R 3.4.1 (R Development Core Team 2018).  3.3 Results 3.3.1 Litter breakdown rates and their spatio-temporal variability Litter breakdown rates in coarse-mesh bags were uniformly greater than in fine-mesh bags (F1,83: 116.3, P < 0.001; Appendix A.3: Table A.3.1). This was also indicated by the lower end of the range of kc/kf exceeding 1 for red alder (range: 1.06-8.59) and speckled alder (1.32-3.09; Appendix A.3: Table A.3.2). Fragmentation rate of litter was almost always lower than dissolution and microbial decomposition rate for both litter species (F1,83: 109.6, P < 0.001; Appendix A.3: Table A.3.1). λF/λm ranged from 0.03-1.43 for red alder, and 0.14-0.67 for speckled alder, and λF was about 3-59% of kc (Appendix A.3: Table A.3.2). Temporal CV of litter breakdown rates was similar among breakdown rate types (F2,37: 2.14, P = 0.13) and regions (F2,37: 0.20, P = 0.82). Across regions, the maximum value of temporal CV was comparable between kc (50.1%) and λm (47.9%), and was considerably larger for λF (82.4%; Table 3.3). Temporal CV of kc was positively and more strongly associated with that of the corresponding λm (Pearson’s correlation: r = 0.88; P < 0.001) than that of λF (Pearson’s correlation: r = 0.60; P < 0.05) at each site. Spatial CV differed significantly across breakdown rate types (F2,25: 70.57, P < 0.001) and regions (F2,25.5: 10.57, P < 0.001). Spatial CV of λF (75.0%) was the largest, followed by that of kc (36.8%) and λm (24.4%; Tukey’s HSD for all pairwise comparisons: P < 0.001). On average, spatial CV of litter breakdown rates was the largest in MKRF (58.3%; Tukey’s HSD: P < 0.01), whereas that in TLW (40.1%) and WR (37.6%) was not significantly different from each other (Table 3.3). In certain years, maximum λF in a study region could be more than double that of the inter-site mean. Temporal CV of kc tended to be of a smaller magnitude than spatial CV, especially in MKRF (range: 17.9-28.6% vs. 27.5-63.1%) and TLW (11.8-28.3% vs. 27.5-41.6%).  33  3.3.2 Selection of variables Global RDAs indicated that both kc and λm were strongly associated with hydrologic indices, but not λF (Table 3.4 and Appendix A.4: Table A.4.1). Water chemistry was significantly related to all litter breakdown rates, and shredder-related variables related to kc and λF, whereas fungal biomass was unimportant for predicting the variation of all breakdown rates. The forward-selected variable(s) from each variable set associated with litter breakdown rates were similar (Table 3.4). In particular, both CV of daily discharge during the incubation period (CVD), and rate of change of discharge during the pre-incubation period (P.ROC), were the selected hydrologic indices for kc and λm. CVD positively affected, and accounted for more variation of kc and λm than did P.ROC, which negatively affected these breakdown rates (adjusted R2 for kc: 0.32 (CVD) vs. 0.17 (P.ROC); λm: 0.43 vs. 0.19; Appendix A.4: Table A.4.1). Total shredder density and density of Tipula craneflies and Pteronarcys stoneflies were the shredder-related variables selected for, and positively related to, both kc and λF. The only selected water chemistry variable was total dissolved phosphorus, which reduced all breakdown rates at high concentrations.  3.3.3 Variance partitioning Total variance explained by the forward-selected pRDA models was the highest for kc (adjusted R2: 0.80), followed by λm (0.64) and λF (0.53; Fig. 3.2), when data pooled across all regions were analysed. Hydrologic indices appeared to explain a considerable proportion of variability of kc and λm across all regions. The proportion of total explained variance of kc attributed uniquely to hydrologic indices was 0.13, smaller than that to shredder-related variables (0.24). Hydrologic indices uniquely accounted for a much higher proportion of variance of λm (0.40) than water chemistry variables did (0.02). Shredder-related variables uniquely explained more variance of λF (0.30) than did water chemistry variables (0.15). However, within individual regions, variance partitioning of litter breakdown rates revealed considerable differences in the order of relative importance of variable sets, and in the significance of their unique effects. For instance, the proportion of variance explained by the unique effects of hydrologic indices (shredder-related variables) ranged from 0 to 0.69 (0-0.76) for kc, and 0.04-0.20 for λm, and their significance differed across regions.  34  3.3.4 Relationships between hydrology and litter breakdown rates Variation in both kc and λm explained by forward-selected hydrologic indices did not differ significantly between linear and the polynomial form of RDAs (kc: P = 0.96; λm: P = 0.89). Hence, results of linear RDAs are presented herein, and were used to generate the composite hydrologic index. Inter-year differences of the composite hydrologic index for both kc and λm were significant in MKRF and TLW, but not in WR (Fig. 3.3, Appendix A.4: Table A.4.2). The inter-year variation of the composite hydrologic index was not consistent among sites in WR. When pooled across regions, an overall positive relationship between the composite hydrologic index and kc (and λm) was observed (Fig. 3.3). However, in MKRF and TLW, significant interannual differences in the composite hydrologic index did not result in consistent shifts of kc within sites across years (Table 3.5). A considerably smaller proportion of variance of kc in these regions was explained by the composite hydrologic index alone (marginal R2: 0.02 and 0.009, respectively) than the random effects of sites (conditional R2: 0.81 and 0.62; variance explained by both factors), as observed for λm (Table 3.5). The composite hydrologic index in WR tended to be lower than that in the other two regions, due to more sites experiencing a greater rate of change of flows, and was also unrelated to kc across years.  3.3.5 Spatiotemporal variability of shredder assemblage structure The NMDS ordination plots well represented data on shredder assemblage structure in all regions, given the fair stress values (stress = 0.10-0.11; Appendix A.5: Fig. A.5.1a-c). Inter-year differences significantly affected assemblage structures in MKRF (PERMANOVA F2,8: 2.26, R2: 0.34, P = 0.005, 431 permutations) and WR (PERMANOVA F2,14: 1.38, R2: 0.19, P = 0.003, 999 permutations), but not in TLW (PERMANOVA F2,14: 0.56, R2: 0.08, P = 0.28, 999 permutations). In WR, assemblage structure differed significantly between the 2nd and 3rd year (or 2010 and 2014; adjusted P: 0.033); however, no pairwise comparisons were significant in MKRF after a sequential Bonferroni correction of p values.  35  3.4 Discussion In this study, I reported a wide range of inter-annual variability of temperature-standardised litter breakdown rates in small forested streams, through multi-year repeated measurements in three geographically separate but climatically similar regions in temperate Canada. Using the approach of approximating pathway-specific contributions to litter breakdown rates in coarse-mesh bags proposed by Lecerf (2017), fragmentation was found to be consistently a less important breakdown agent than dissolution and microbial decomposition in my study regions. Contrary to my hypotheses, across regions, differences in site-level hydrologic conditions were unrelated to the interannual variability of λF. Furthermore, the relationships between hydrologic conditions (mainly in terms of flow flashiness) and the interannual variability of λm and kc were not consistent among study sites. Overall, the temporal CV of λF at individual sites did not exceed that of λm, except for several TLW sites. While weather-driven differences in hydrologic conditions could account for some spatial variability of litter breakdown rates across study regions, their effects on breakdown rates across years within sites were inconsistent. The effects of site-level hydrologic fluctuations might be modified by unmeasured reach- and patch-scale attributes (e.g., benthic litter quality and quantity), and/or in-channel physical features (e.g., channel morphology, availability of hydraulic refugia), thereby inducing varied responses of hydraulic conditions (Turner and Stewardson 2014) and hence breakdown rates (Colas et al. 2017).  3.4.1 Sources of variation in litter breakdown rates Across the study regions, λm represented a considerably greater proportion of kc than λF. The range of percent contribution of λF to kc found in this study overlapped with that in 100 temperate European streams (range: <1 to 70%; reanalysed by Lecerf 2017), suggesting that the dominating role of dissolution and microbial decomposition in litter breakdown occurs over a broad geographic scale. Therefore, the observed pattern of variability of kc was more sensitive to that of λm. Indeed, the forward-selected hydrologic indices and water chemistry variables associated with kc and λm were identical, whereas no hydrologic indices were significantly associated with λF. The significantly negative effects of total dissolved phosphorus on both kc, λF and λm should 36  be interpreted with caution in the forested, nutrient-limited study sites, as phosphorus enrichment in these systems typically enhances litter breakdown rates (Rosemond et al. 2015; Ferreira et al. 2015). Considering that the direct effects of temperature on litter breakdown rates were normalised and accounted for by degree days (see Boyero et al. 2011), the rather small interannual differences in water temperature within study regions (i.e., 0.6-3.7°C) likely had minor contributions to the variation in litter breakdown in my analyses. Global RDAs and subsequent variance partitioning suggested that the variability of λF across sites/regions and years was negligibly related to differences in hydrologic conditions, but rather driven by shredder-related and water chemistry variables. Changes in forward-selected shredder density, density of large-bodied shredders, and total dissolved phosphorus affected fragmentation through shredder feeding (and not mechanical abrasion). The overall effects on fragmentation by changing shredder densities in leaf bags across years did not appear to be associated with spate occurrences in the study regions. Some shredder taxa might be able to seek hydraulic refugia (e.g., leaf accumulations, hyporheic zone) for shelter during spates, and/or they could quickly colonise leaf bags post-disturbance (Whiles et al. 1993; Negishi and Richardson 2006) to maintain shredder feeding rates during the incubation period. It could be inferred that the availability of benthic refugia and litter quantity, which could drive the variations in shredder densities and assemblages (Tiegs et al. 2008) and hence litter breakdown, were not strongly influenced by site-level changes in hydrologic conditions.  Hydrologic indices reflecting flashiness during and before leaf bag incubation significantly influenced λm. Fungal biomass did not contribute significantly to the variability of λm in MKRF and TLW (unmeasured in WR), although previous studies showed strongly positive relationships between fungal biomass and litter breakdown rates (Gessner and Chauvet 1994; Lecerf and Richardson 2010). In my study, microbial decomposition and dissolution was influenced more by the degree of flow variability than by water chemistry. Similarly, in low-order streams in France, human-induced changes in flow quantity, and the intensity and frequency of extreme flow events (and geomorphological parameters) were shown to affect microbial decomposition (kf) more strongly than water chemistry did (Colas et al. 2017). Hydrologic changes could give rise to the variability of λm through numerous potential pathways (unmeasured in this study), for instance, via altered microbial diversity and assemblage structure 37  on litter (Dang et al. 2005; Judd et al. 2006; Zeglin 2015), and primary producers whose priming effects through the release of labile carbon exudates could stimulate bacterial activity (Danger et al. 2013). Several lines of evidence suggest that variable ecological responses to flood events across watersheds could be modulated by ecological and hydromorphological parameters operating at the patch, site, and watershed scale. These variables can include current velocity, substrate composition, vegetation characteristics, channel morphology, etc. (e.g., Tiegs et al. 2009; McMullen and Lytle 2012; Stenroth et al. 2014; Robertson et al. 2015; Colas et al. 2017). These factors might account for much of the unexplained variance of litter breakdown rates in the perennial streams of this study. For instance, site-level hydrologic indices that I examined might not be good surrogates of hydraulic conditions (or shear stress) within and in the vicinity of leaf bags, as a similar magnitude of hydrologic variation could produce diverse, complex hydraulic responses due to the influences of channel morphology (Turner and Stewardson 2014). Hydraulic stress-discharge relationships are potentially nonlinear and spatially heterogeneous, and are operationally difficult to establish (Kakouei et al. 2017), which could limit the success in generalising the responses of litter breakdown rates to natural hydrologic variations in perennial streams (Turner and Stewardson 2014). In contrast, hydrologic variability has clearer impacts on litter breakdown rates in drought-prone streams experiencing precipitation-induced flow intermittency (e.g., Datry et al. 2011, Dieter et al. 2011, Martínez et al. 2015), and small, perennial streams experiencing experimental flow reduction (Northington and Webster 2017). These studies demonstrated that drastic flow decreases generally slowed litter breakdown due to reduced colonization of shredders; and in the absence of surface flow, dissolution and photodegradation became the dominant mechanisms of breakdown.  3.4.2 Spatio-temporal variability of litter breakdown rates The temporal variability of kc was more strongly associated with that of λm at individual sites than λF. This was probably due to the greater contribution by λm to kc, and similar temporal CV among λF and λm. In WR, the apparently greater temporal CV of λF might be due to the interannual differences in shredder assemblage structure and hence feeding rates. There are limited multi-year studies from which the interannual variability of either kc or kf at individual 38  sites could be inferred, and their range of variability resembled that found in the present study. For example, the range of temporal CV of kc (~26-54%) and kf (~37-46%) in three forested reference streams in WR during 2002-2007 (Kreutzweiser et al. 2010) was comparable to that of the present study sites within the same region (i.e., ~10-50%; Table 3.3). Temporal CV of kc in Satellite Branch, Coweeta Hydrologic Laboratory, USA was 27% for rhododendron, Rhododendron maximum L. (6 years; 1985-1990) and 28% for red maple, Acer rubrum L. (7 years; 1985-1990, 1992) (Webster et al. 1999). However, these previous studies did not explicitly quantify the sources of interannual variability of kc, unlike the present study.  In all regions, considerable interannual differences in the across-watershed variability of litter breakdown rates were observed. In addition to hydrological characteristics, stream temperature regime, water chemistry, and benthic litter quality and quantity are important watershed-level controls of litter breakdown via their influences on shredder feeding and microbial activity (Royer and Minshall 2003; Graça et al. 2015). Such pattern of variability could be attributed to temporally-varying, climatic factors inducing watershed-specific responses of these controls (other than turbulence), which differed across years (Kreutzweiser et al. 2010).  3.4.3 Bioassessment implications The selected study sites covered a range of landscape attributes, riparian vegetation characteristics and hydrologic regimes. They were largely unaffected by recent watershed disturbances such as forest harvesting and fires, which are common in the study regions. Therefore, these sites – with efforts made to control for stream size – could be regarded as a group of objectively and realistically chosen forest streams to best approximate minimally disturbed conditions (Stoddard et al. 2006), which provided baseline (reference) conditions for evaluating the robustness of existing bioassessment benchmarks. The range of natural variability of kc was mostly smaller than that of λF, particularly in terms of spatial CV across years. Therefore, kc is a more preferred breakdown metric to use in bioassessment, given its smaller temporal background variation and interference with disturbance effects. The variability of kc at the regional scale indicative of ‘no impacts’ suggested by Gessner and Chauvet (2002) was inferred from a study conducted in streams from 1st- to 7th-order in Sweden, situated in the northern temperate zone (Jonsson et al. 2001). When including 39  only 1st- to 3rd-order streams, the range became 73-127% of the mean. Comparatively, the upper end of the corresponding range determined in the present study was higher (and also year-specific), as shown by the upper end of the inter-site range of kc (i.e., MKRF: up to 170%; TLW: 163%; WR: 154%; see Table 3.3). A similar range of natural variability of kc were also reported in many European streams (CV: ≈ 50%; see Boyero et al. 2015). At the reach scale, the range of natural variability of kc (i.e., 75-133% of the mean) proposed by Gessner and Chauvet (2002) encompasses that of most sites in the present study, with some exceptions in WR (range: 46-151%). The extent of weather-driven variability of kc obtained in this study could be used to develop site- and region-specific baseline conditions to refine the bioassessment framework based on Gessner and Chauvet (2002), depending on assessment approaches and data availability (Table 3.6). When pre-disturbance data are available at the site level, using the temporal CV of kc as baseline conditions will likely have higher sensitivity to detect disturbance impacts, compared to using the spatial CV of other reference sites. This is because the temporal CV of kc tended to be smaller than the spatial CV. Applying the baseline conditions established in the present study, previous findings of the effects of riparian/upland forest harvesting on stream litter breakdown appear to have a stronger ‘ecological significance’ (in addition to ‘statistical significance’) in the same study regions. For example, the conclusions of the absence of harvesting effects on litter breakdown rates in impacted sites in MKRF (Yeung et al. 2017) and WR (Kreutzweiser et al. 2010; Musetta-Lambert et al. 2017) are in accordance with these rates falling within the range of region-specific baseline conditions (i.e., spatial CV). Conversely, the presence of such effects in MKRF (Lecerf and Richardson 2010; in a different time frame from Yeung et al. 2017) and WR (Kreutzweiser et al. 2008a) corresponds to these rates falling outside the range of baseline conditions. Nevertheless, it should be noted that the upper/lower end of natural variability of kc in some sites (particularly in WR) approximates the mean effect sizes of other common agents of watershed disturbance, such as nutrient enrichment (Ferreira et al. 2015), and replacement of native forests by plantations and invasive plant species (Ferreira et al. 2016a). Hence, relying on litter breakdown rates as a single indicator of disturbances affecting streams might be problematic, as the possibility of interannual (hydrologic) variation obscuring putative disturbance effects in certain years cannot be excluded.  40  In cases where considerable hydrologic changes occurred in conjunction with particular disturbances of interest (e.g., fertiliser application), kc (or λm) in impacted sites may be ‘hydrologically-adjusted’ to yield the expected value under the undisturbed hydrologic condition. This is achievable only when robust relationships between kc and the composite hydrologic index have been established for these sites prior to disturbances. Furthermore, this adjustment will not be applicable when the effects of hydrologic condition on kc as altered by disturbances are themselves of interest in bioassessment (Chauvet et al. 2016). The values of kc/kf (and λF/λm) observed in this study varied less within sites (across years) than among sites (Appendix A.3: Table A.3.2), and its range in reference sites (MKRF: 1.2-8.6; TLW: 1.1-4.4; WR: 1.3-3.1) would have been considered to reflect ‘no impact’ (i.e., kc/kf : 1.2-1.5) to ‘severely compromised’ (>2.0) stream functioning according to the criterion suggested by Gessner and Chauvet (2002). This shows that the relative contribution of fragmentation, and microbial decomposition and dissolution could be highly inconsistent at least across sites in the study regions. Indeed, variable extent of consistency of kc/kf in reference sites across watersheds/regions has been reported in temperate (Tiegs et al. 2009; Hladyz et al. 2010) as well as tropical streams (Boyero et al. 2015). Thus, the classification of stream functional integrity solely based on the absolute values of kc/kf, without prior data of the study regions, could yield inaccurate results. It would be a suitable metric only when the values are known to differ significantly among impacted and reference sites (see Hladyz et al. 2010).  3.4.4 Conclusions My study findings provide a basis for incorporating interannual variability and regional specificity to better operationalise the bioassessment framework using litter breakdown assays for temperate small streams. The apparently weak and inconsistent effects of the temporal variability of hydrologic conditions on litter breakdown rates are likely well accounted for in the recommended range of natural variability, within the gradient of hydrologic conditions encompassed in the study. However, given the relatively short span of study years and increasing occurrences of hydrologically extreme events (e.g., droughts) and changing litterfall timing anticipated under climate change (e.g., Creed et al. 2015; Coulthard et al. 2016; Imberger et al. 2016), the recommended baseline conditions for stream bioassessment should not be regarded as 41  stationary, and this baseline may shift over time and necessitate future revisions. This also strengthens the notion of using litter breakdown rates in combination with other structural and functional measures to comprehensively assess stream ecological integrity under anthropogenic disturbances (e.g., Gessner and Chauvet 2002; Young et al. 2008; Chauvet et al. 2016).42  Table 3.1 Catchment- and reach-scale characteristics for sites in the study regions in British Columbia and Ontario.  Region† and site Location Watershed area (ha) Elevation (m) Stream order Years since last major harvest Extent (%) and year of harvest Reach gradient (%) Wetted width (m) ‡ Mid-channel depth (m) ‡ Canopy openness (%) § Latitude (N) Longitude (W) MKRF            G-4 49°17'44" 122°35'48" 28 257 1 ~85 –  30.6 0.97 (0.22) 0.11 (0.02) 17.0 Mike 49°16'40" 122°32'46" 30 314 1 ~85 –  5.2 1.43 (0.12) 0.14 (0.03) 11.0 ¶ Spring 49°17'41" 122°34'2" 38 340 3 ~85 –  8.8 2.40 (0.37) 0.15 (0.02) 8.7 ¶ Upper East 49°17'3" 122°33'43" 36 306 2 ~85 –  57.7 1.92 (0.20) 0.15 (0.03) 7.7 ¶ TLW            TLW34 47°3'27" 84°24'59" 68 391 2 17 61 (1997) 8.8 2.33 (0.26) 0.19 (0.04) 25.4 TLW96 47°4'39" 84°24'39" 71 362 2 140-200 –  3.5 1.97 (0.35) 0.14 (0.03) 26.0 TLW97 47°4'34" 84°24'59" 37 363 2 140-200 –  5.2 1.81 (0.31) 0.14 (0.02) 34.7 AY1-1 46°58'17" 84°17'59" 987 290 3 21 65.7 (1990); 14.1 (1993) 1.8 4.60 (0.31) 0.19 (0.03) 27.1 AY4-1 47°0'28" 84°18'41" 219 248 3 24 35.2 (1990); 0.22 (2013) 10.5 2.25 (0.33) 0.19 (0.04) 28.3 WR            EWR4 48°45’4” 85°10’21” 655 389 2 ~40-50 –  10.7 2.83 (0.31)  0.10 (0.01) 26.2 EWR5 48°55’11” 85°14’33” 344 355 2 ~40-50 –  4.2 1.13 (0.15) 0.10 (0.02) 14.5 EWR6 48°44’56” 85°9’58” 787 391 2  ~40-50 –  3.3 1.42 (0.07) 0.08 (0.01) 18.0 EWR8 48°14’3” 85°25’11” 299 450 2 ~40-50 –  1.8 2.57 (0.20) 0.17 (0.02) 28.1 EWR9 48°15’11” 85°23’48” 45 465 1 ~40-50 –  2.4 1.45 (0.09) 0.09 (0.01) 41.6 Notes: MKRF, Malcolm Knapp Research Forest; TLW, Turkey Lakes Watershed; WR: White River. Watershed area, stream order, and extent of forest harvesting were determined in ArcGIS using watershed delineations in Whitebox Geospatial Analysis Tools (version 3.4). 43  † Dominant forest vegetation in each region: (MKRF) western hemlock (Tsuga heterophylla (Raf.) Sarg.), western red cedar (Thuja plicata Donn ex D. Don), Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco); (TLW) sugar maple (Acer saccharum Marsh.), with scattered stands of yellow birch (Betula alleghaniensis Britton); (WR) black and white spruce (Picea mariana (Miller) BSP and Picea glauca (Moench) Voss), balsam fir (Abies balsamea (L.) Miller), jack pine (Pinus banksiana Lamb.). Riparian vegetation in each region: (MKRF) red alder (Alnus rubra Bong.), vine maple (Acer circinatum Pursh), salmonberry (Rubus spectabilis Pursh); (TLW) Species similar to those in forest vegetation; (WR) speckled alder (Alnus incana ssp. rugosa (Du Roi) Clausen), high bush cranberry (Viburnum trilobum Marshall), beaked hazel (Corylus cornuta Marshall), red osier dogwood (Cornus sericea L.). ‡ Data are presented as mean values (± SE) measured in September or October across 2014-2017 for MKRF (2014-2016 for G-4) and TLW sites, and in 2010 for WR sites. § Measurements were obtained using either spherical densiometer, or digital hemispherical pictures taken by fisheye lens (denoted by ¶; see Yeung et al. (2017) for details), in 2014 for MKRF and TLW, and 2010 for WR.  44  Table 3.2 Initial sets of candidate hydrologic indices, shredder-related, fungal biomass, and water chemistry variables that were expected to influence stream litter breakdown in the study regions in British Columbia and Ontario.  Notation Variable Hydrologic indices (n = 14) Magnitude BFI for incubation period, P.BFI for pre-incubation period Base flow index: Ratio of 7-day minimum flow to total discharge Frequency of flow events † LF, P.LF Low-flow period: No. of times daily discharge is below 25% of 3-year mean of median daily discharge from summer to fall HF, P.HF High-flow pulse: No. of times daily discharge is above 3 times the 3-year mean of median daily discharge from summer to fall Duration of flow events †  LFD, P.LFD Low-flow duration: Total no. of days of low-flow period(s) HFD, P.HFD High-flow duration: Total no. of days of high-flow pulse(s) Flashiness  CV, P.CV Variability: Coefficient of variation of daily discharge ROC, P.ROC Rate of change: Ratio of the sum of the absolute values of day-to-day changes in daily discharge to the sum of daily discharge (Richards-Baker Flashiness Index; see Baker et al. 2004) Shredder-related variables (n = 4) SD Shredder density  LMD Density of Lepidostoma and Micrasema (only in MKRF) caddisflies TPD Density of Tipula craneflies and Pteronarcys (only in MKRF) stoneflies STR Shredder taxonomic richness Fungal-related variable (n = 1) FB Fungal biomass Water chemistry variables (n = 4) PH pH CO Conductivity TN Total dissolved nitrogen TP Total dissolved phosphorus Note: MKRF, Malcolm Knapp Research Forest. † Variables were adjusted by the time difference between the last day of these flow event(s) and the first day of incubation/pre-incubation period (see main text for details of calculations). 45  Table 3.3 Extent of spatio-temporal variability of in-stream total breakdown rate (kc), fragmentation rate (λF), and dissolution and microbial decomposition rate (λm) in the study regions in British Columbia and Ontario.  Region and site kc   λF   λm TCV Inter-year range   TCV Inter-year range   TCV Inter-year range MKRF         G-4 27.17 0.81-1.19 13.73 0.90-1.10 32.85 0.77-1.23 Mike 28.58 0.64-1.34  18.38 0.87-1.27  34.84 0.59-1.44 Spring 20.89 0.79-1.25  59.55 0.58-1.89  17.39 0.89-1.26 Upper East 17.87 0.84-1.19  26.63 0.79-1.37  24.18 0.84-1.36 SCV (inter-site range) 2014 47.68 (0.68-1.70)  115.48 (0.25-2.70)  24.62 (0.82-1.36) 2015 42.99 (0.64-1.48)  85.75 (0.29-1.95)  21.89 (0.82-1.24) 2016 46.78 (0.65-1.68)  101.25 (0.34-2.50)  22.70 (0.79-1.29) 2017 63.09 (0.34-1.60) †  84.93 (0.16-1.86)  41.96 (0.53-1.33) TLW         TLW34 23.89 0.78-1.34 53.60 0.52-1.75 10.68 0.91-1.12 TLW96 11.83 0.84-1.09  30.56 0.62-1.30  12.79 0.89-1.18 TLW97 14.31 0.80-1.11  20.68 0.84-1.30  19.06 0.77-1.23 AY1-1 22.71 0.68-1.21  58.77 0.61-1.87  26.02 0.64-1.24 AY4-1 28.27 0.83-1.42  82.38 0.15-2.10  20.52 0.80-1.29 SCV (inter-site range) 2014 34.04 (0.59-1.49)  84.45 (0.07-2.31)  22.30 (0.75-1.23) 2015 27.52 (0.81-1.49) 54.36 (0.56-1.94) 19.20 (0.84-1.33) 2016 41.58 (0.60-1.63)  68.13 (0.33-2.07)  25.98 (0.75-1.37) 2017 31.16 (0.58-1.42)  55.16 (0.19-1.70)  17.81 (0.82-1.25) WR         EWR4 16.50 0.82-1.14 21.04 0.81-1.23 15.74 0.82-1.10 EWR5 9.70 0.89-1.08  7.59 0.94-1.08  10.17 0.89-1.08 EWR6 34.78 0.69-1.37  55.57 0.66-1.64  29.26 0.70-1.28 EWR8 47.23 0.46-1.34  46.63 0.46-1.27  47.92 0.46-1.38 EWR9 50.10 0.51-1.51  76.42 0.48-1.88  43.68 0.52-1.37 SCV (inter-site range) 2009 29.77 (0.66-1.32)  60.65 (0.30-1.73)  17.01 (0.82-1.21) 2010 19.29 (0.80-1.27) 51.35 (0.36-1.55) 11.93 (0.84-1.16) 2014 41.31 (0.57-1.54)   63.47 (0.65-2.13)   43.34 (0.51-1.56) Notes: Tcv, coefficient of variation (CV; in %) of breakdown rates across years; Scv, CV (%) of breakdown rates across sites; MKRF, Malcolm Knapp Research Forest; TLW, Turkey Lakes Watershed; WR, White River. Inter-year(site) range refers to the relative range limit around the inter-year(site) mean for a given litter breakdown rate (see main text for details of calculation). † TCV (inter-year range) becomes 28.63% (0.80-1.20) when data from Mike are excluded. 46  Table 3.4 Results of forward selection of hydrologic indices, shredder-related, fungal biomass, and water chemistry variables on in-stream total breakdown rate (kc), fragmentation rate (λF), and dissolution and microbial decomposition rate (λm) in the study regions in British Columbia and Ontario. Variables selected for use in the variance partitioning analysis, and the directional effect of each variable (+, −) on litter breakdown rates are given (see Table 3.2 for notations of variables).  Variable set Degrees of freedom F P Variable(s) selected kc Hydrology 14, 26 4.31 < 0.001 (+) CVD; (−) P.ROC Shredder-related 4, 36 11.83 < 0.001 (+) SD; (−) STR; (+) TPD; (−) LMD Fungal biomass 1, 24 0.45 0.52 None selected Water chemistry 4, 31 9.24 < 0.001 (−) TP λF Hydrology 14, 26 0.74 0.74 None selected Shredder-related 4, 36 9.38 < 0.001 (+) SD; (+) TPD Fungal biomass 1, 24 0.21 0.65 None selected Water chemistry 4, 31 9.24 < 0.001 (−) TP λm Hydrology 14, 26 8.34 < 0.001 (+) CVD; (−) P.ROC Fungal biomass 1, 24 2.76 0.11 None selected Water chemistry 4, 31 9.70 < 0.001 (−) TP 47  Table 3.5 Results of linear mixed-effects models explaining the relationships between in-stream total breakdown rate (kc) and dissolution and microbial decomposition rate (λm), and the composite hydrologic index, for each study region in British Columbia and Ontario. Site was treated as a random effect for each region in regression analyses. Model fit was assessed by the marginal and conditional determination coefficients (R2) using the rsquared.GLMM function in the MuMIn package (Nakagawa and Schielzeth 2013). The significance of regression relationships was estimated using an approximate F-test based on the Kenward-Roger approach.  Variable Marginal R2 Conditional R2 Estimate Standard error df F P kc        MKRF 0.02 0.81 0.00058 0.00054 1, 9.10 1.16 0.31 TLW 0.009 0.62 0.00017 0.00026 1, 14.00 0.43 0.52 WR 0.17 0.17 0.00056 0.00034 1, 12.41 2.23 0.16 λm        MKRF 0.02 0.37 -0.00025 0.00038 1, 9.40 0.43 0.53 TLW 5.4E-5 0.30 -6.7 × 10-6 0.00017 1, 14.00 0.001 0.97 WR 0.24 0.24 0.00038 0.00018 1, 12.04 3.67 0.08 Notes: MKRF, Malcolm Knapp Research Forest; TLW, Turkey Lakes Watershed; WR, White River. 48  Table 3.6 Recommended baseline conditions for assessing the functional integrity of small streams using total litter breakdown rate (kc) in the study regions in British Columbia and Ontario.  Assessment approach Description Recommended range of natural variability (% of the mean of reference site(s)) 1. Before-After (BA) Sampling before and after disturbance at the same site (also applicable for sampling upstream and downstream of the impacted reach after disturbance) Best to be site-specific; if long-term dataset is unavailable, MKRF: 60-135%; TLW: 65-145%; WR: 50-155% (corresponding to temporal CV in this study) 2. Control-Impact (CI, i.e., space-for-time substitution) Sampling (once) after disturbance in (more than one) impacted and reference site(s) MKRF: 60-170%; TLW: 55-165%; WR: 55-155% (corresponding to spatial CV in this study) † 3. Before-After-Control-Impact (BACI), paired BACI (BACIPS), multiple BACI (MBACI), and beyond BACI (sensu Downes et al. (2002)) BACI: Sampling once before and once after disturbance in a reference and an impacted site  BACIPS: Multiple paired samplings before and after disturbance in a reference and an impacted site  MBACI: Multiple samplings before and after disturbance in multiple reference and impacted sites  Beyond BACI: Multiple samplings before and after disturbance in multiple reference sites and one impacted site Not necessary, as detection of impacts based on the significance of BA×CI interaction (for Beyond BACI, the interaction between the times of sampling and contrast of impacted and reference sites) already takes into account the range of natural variability (see Underwood 1992, 1994) Notes: MKRF, Malcolm Knapp Research Forest; TLW, Turkey Lakes Watershed; WR, White River. † Data from one site (i.e., Mike) in 2017 were excluded, as the inclusion of this site would have lowered the minimum of the range of variability at MKRF to 0.30 of the mean of reference sites. 49   Figure 3.1 Map showing the locations of study regions in British Columbia and Ontario, Canada. Abbreviations of study regions are as in Table 3.1.50   Figure 3.2 Results of variance partitioning for (a) total breakdown rate (kc), (b) fragmentation rate (λF), and (c) dissolution and microbial decomposition rate (λm) across all study regions in British Columbia and Ontario, and in each of these regions, using forward-selected hydrologic (H), shredder-related (S), and/or water chemistry (W) variables as predictors (see Table 3.2). The global RDA model involving fungal biomass is non-significant for all breakdown metrics, and hence it is excluded from the variance partitioning analysis. Values displayed are adjusted R2 as portion of variance explained, including the residual, unexplained variation, and negative values are not shown. The sum of variance explained by the explanatory matrices and residual variance may exceed 1 due to negative explained variances. Significance levels of the unique effects of hydrology, shredders and water chemistry are indicated with asterisks (* P < 0.05, ** P < 0.01, *** P < 0.001). Abbreviations of study regions are as in Table 3.1.51   Figure 3.3 Relationships between (a) total breakdown rate (kc), (b) dissolution and microbial decomposition rate (λm) of leaf litter (shown on a log10 scale), and the corresponding fitted site scores (i.e., linear combination of selected hydrologic indices) on the first RDA axis, generated from the global RDA across all study regions in British Columbia and Ontario. The study years are colour-coded. Note that RDA axis 1 scores are not comparable between (a) and (b). Abbreviations of study regions are as in Table 3.1.  52  Chapter 4: Litter quality and decomposer community attributes mediate the effects of litter-decomposer interactions on stream litter breakdown  4.1 Introduction The preference and dependence of consumers on allochthonous resource subsidies are known to vary across populations and communities, due to spatial differences in resource availability, consumer traits, and environmental context (e.g., Ayres et al. 2009; Bassar et al. 2010; Rodriguez-Cabal et al. 2016). The availability and flows of resource subsidies in the recipient ecosystem are therefore modulated by the geographic mosaic of resource-consumer interactions, which can arise from ecological and/or evolutionary processes. For instance, some decomposer communities can undergo phenotypic and/or genotypic shifts to more efficiently process allochthonous plant litter in the vicinity of the plant (i.e., home-derived litter) to which they have higher evolutionary and/or contemporary exposure, compared to litter derived from further away (e.g., Ayres et al. 2009; Jackrel and Wootton 2014; Leroy et al. 2017). Taxon sorting may also occur among decomposers whereby species (or genotypes and/or phenotypes within species) more adapted to process home-derived litter outcompete their less-adapted counterparts (Jackrel and Wootton 2014; Jackrel et al. 2016), thereby giving rise to community-level resource specialisation (Devictor et al. 2010). These resource-consumer interactions can accelerate the consumption of resource subsidies (i.e., litter breakdown) originating from their home relative to those from away – a phenomenon known as ‘home-field advantage’ (HFA; Gholz et al. 2000). In terrestrial ecosystems, HFA is strongly driven by the degree of specialisation of the decomposer community, particularly microbes, to the chemical composition (i.e., resource quality) of home-derived litter relative to foreign litter (Veen et al. 2014). Such specialisation can be mediated by the past and present environment of litter incubation (e.g., conditions of soil and litter layer, presence of herbivores; Freschet et al. 2012; Kagata and Ohgushi 2013; Palozzi and Lindo 2018). For example, litter breakdown rate tends to be reduced when the incubated litter is of different quality to that found in the litter layer (Freschet et al. 2012; Veen et al. 2014). These 53  drivers underlie marked variations in both the direction and magnitude of terrestrial HFA outcomes (see Palozzi and Lindo 2018, and references therein). Forested headwater streams are typically nutrient-limited, and their food webs rely on litter inputs from the riparian vegetation as resource subsidies to consumers, particularly detritus-feeding macroinvertebrates (shredders; Wallace et al. 1997; Richardson et al. 2010). Previous research has shown either faster or similar in-stream breakdown rate of litter derived from adjacent riparian habitats, compared with foreign litter (of the same species) originating from upstream locations and/or more distant riparian zones (e.g., Kominoski et al. 2011; Jackrel and Wootton 2014; Fenoy et al. 2016; Stoker et al. 2017). Relatively few studies have explicitly considered how differences in quality between home-derived and foreign litter, and incubation conditions among streams combine to induce such spatial variation in HFA effects. Several lines of evidence demonstrate more positive HFA effects in streams when the incubated litter is more prevalent in the riparian vegetation, and hence in the benthic litter layer (Kominoski et al. 2011; Jackrel and Wootton 2015; Lidman et al. 2017). It is suggested that stream decomposers can ‘perceive’ intraspecific (including within-river) differences in locally abundant and high-quality alder litter in terms of nutritional quality and defensive compounds (Lecerf and Chauvet 2008; Jackrel et al. 2016), thereby inducing positive HFA on its breakdown in streams (Jackrel and Wootton 2014). However, for streams draining mid- to late-successional riparian forests that receive litter inputs of diverse quality, it is unclear how HFA effects may differ across litter species which are present in the riparian vegetation but are of varying quality. In these streams, decomposers’ past exposure to litter resources and their degree of specialisation on particular litter species could be important drivers of HFA effects (see Milcu and Manning 2011; Palozzi and Lindo 2018). It is important to recognise that the quality and quantity of the litter layer, and decomposer communities in small streams can be spatially and temporally dynamic, in contrast with those in the relatively stable terrestrial environment (but see Šnajdr et al. 2008). Such dynamic changes in the incubation environment may modify HFA effects. For example, spatial heterogeneity of riparian vegetation and canopy cover can create a mosaic of litter inputs, regulating stream decomposer distributions and consumption patterns (e.g., Fazi et al. 2005; Gjerløv and Richardson 2010; Tonin et al. 2018b). Stream litter availability and decomposer 54  community attributes can show high temporal fluctuations especially during peak riparian litterfall periods with concurrently high precipitation, when riparian litter inputs, litter transport from upstream, and litter residence time vary greatly (e.g., O’Keefe and Naiman 2006; Stenroth et al. 2014; Mora-Gómez et al. 2015). The magnitude of these fluctuations differs between streams, which is dependent on the hydraulic conditions and local geomorphic features that affect the availability of refugium patches and retention structures (e.g., Findlay et al. 2002; Negishi and Richardson 2006; Hoover et al. 2006). Another possible source of variation in HFA effects among streams could be differences in the selectivity and plasticity in litter processing between shredder and microbial decomposer communities (Kominoski et al. 2011; Stoker et al. 2017). Hence, the independent contributions of these decomposer groups to potential HFA effects need to be discerned. In this study, I performed spatially-replicated, reciprocal multi-species litter transplant incubations in small streams across two distant temperate regions in Canada. Two pairs of congeneric riparian broadleaf species (alder and maple) and a congeneric pair of conifer species (cedar) spanning a gradient of chemical quality were reciprocally transplanted, and species in each pair are non-overlapping between study regions. I measured litter breakdown rates in coarse- and fine-mesh bags to derive shredder feeding rate (not for conifer litter) and microbial decomposition rate. Through multiple congeneric litter comparisons, I quantified the strength and inter-stream variability of HFA effects on temperature-normalised litter breakdown, and related them to important biophysical drivers of litter breakdown. With this study design, I addressed several questions pertaining to HFA effects, including: (1) Is the accelerated breakdown of home-derived litter relative to transplanted foreign litter widely detectable across streams, with high natural variability of biophysical drivers operating at the reach- and patch-scale?; (2) If detectable, how are these effects associated with the chemical quality of litter?; and (3) Does the efficiency of processing home-derived litter vary among shredder and microbial decomposer communities? I hypothesised that stream microbes processed home-derived litter more efficiently than expected relative to foreign litter. In contrast, HFA effects on shredder feeding were expected to be considerably weaker, as the preference for colonising and consuming litter by shredders tends to be more strongly driven by litter quality than by litter origin (e.g., Kominoski et al. 2011; 55  Casas et al. 2013; Kuglerová et al. 2017a). HFA effects on microbial decomposition rate would be more positive for more recalcitrant conifer litter than broadleaf litter, when the processing of conifer litter required greater specialization by microbial decomposer communities, possibly through the production of specific enzymes to degrade recalcitrant compounds (see Wallenstein et al. 2013).   4.2 Methods 4.2.1 Study sites and litter species selected The reciprocal transplant experiment was conducted in University of British Columbia’s Malcolm Knapp Research Forest (MKRF) in southwestern British Columbia, and the Turkey Lakes Watershed (TLW) in central Ontario – both in the temperate zone of Canada. MKRF is located in the Pacific Maritimes ecozone approximately 60 km east of Vancouver, and TLW lies on the Boreal Shield ecozone about 60 km north of Sault Ste. Marie. Sampling was conducted in five or six small forest streams (1st- to 3rd-order) within each region (Appendix B: Fig. B.1) in 2014 (for alder litter) and 2015 (for maple and cedar litter; see litter descriptions below). The study streams were draining mid- to late-successional forests, and were in catchments with minimal disturbances by fires and forest harvesting in the past 20 years (for catchment- and reach-scale characteristics, see Appendix B: Table B.1). Nonetheless, these streams differed considerably in terms of riparian vegetation and canopy cover, and geomorphic and local habitat features, which were probable biophysical drivers of the spatial variability of litter-decomposer interactions. Owing to riparian or upland forest harvesting in some catchments in 2015, sampling occurred in other undisturbed sites near the affected streams in that year, in order to maintain the sample size. Detailed descriptions of climate, geology and vegetation of the study regions are available in Kiffney & Richardson (2010) for MKRF, and Foster et al. (2005) for TLW. I selected three congeneric litter pairs originating from the two study regions: two pairs of broadleaf leaves (alder and maple), and one pair of coniferous leaves (cedar). Species within each litter pair had similar leaf morphology (in terms of shape, edge, venation, etc) and colonisable surface area for decomposers. My quantification of HFA effects through comparisons of congeneric litter species reduced the potential confounding effects of leaf 56  morphology and area on decomposer (especially shredder) consumption (Davis and Winterbourn 1977; Moline and Poff 2008). The selected litter pairs span a gradient of mean values in initial chemical quality traits that control litter breakdown (see Fig. 4.1). Red alder (Alnus rubra Bong.), vine maple (Acer circinatum Pursh), and western red cedar (Thuja plicata Don.) originated from British Columbia, whereas speckled alder (Alnus incana ssp. rugosa (Du Roi) Clausen), sugar maple (Acer saccharum Marsh.), and eastern white cedar (Thuja occidentalis L.) originated from Ontario. They are common riparian tree and shrub species that provide important sources of litter inputs to streams (e.g., Kreutzweiser et al. 2004; Kiffney and Richardson 2010; Hoover et al. 2011). I did not determine benthic litter composition during and before the reciprocal transplant experiment, which could characterise the quality of litter resources that decomposers have been exposed to. I observed that the benthic litter layers commonly consisted of lower-quality litter of western red cedar and western hemlock (Tsuga heterophylla (Raf.) Sarg.) in the study streams in MKRF, and mid-quality sugar maple and yellow birch (Betula alleghaniensis Britt.) in TLW, typically in benthic samples of macroinvertebrates collected as part of other studies.  4.2.2 Field sampling 4.2.2.1 Litter breakdown experiment Litter materials of species originating from British Columbia and Ontario were collected from MKRF and TLW, respectively. The only exception was speckled alder, whose litter was collected near Sault Ste. Marie (46°33'11"N, 84°18'1"W), about 60 km away from TLW. The collection localities of other species were of variable distance (>200 m to 8 km) from, and not located in the riparian areas adjacent to the study sites in each region. Hence, the collected litter likely differed in nutritional and defensive chemistry to some degree from the conspecific litter from riparian vegetation adjacent to (or upstream of) each study site, the latter of which stream decomposers could have been most adapted to (see Jackrel and Wootton 2014; Jackrel et al. 2016). My study focus was to determine HFA effects between litter species (rather than within species), evaluating decomposer adaptation to litter resources in their native range, compared to those from distant regions. I therefore used a standardised, regional (rather than local) litter source for each species (see also Kominoski et al. 2011; Fenoy et al. 2016). The choice of litter 57  source could indeed contribute to the variations of HFA effects evaluated through my comparisons of litter derived from home and foreign regions. Alder, maple, and cedar leaves ready for abscission were collected by shaking the trees. Some near-senescent cedar leaves (turning yellow/brown) were picked from standing trees. Leaves of each species were thoroughly mixed, and air-dried to constant mass prior to use.  In the reciprocal incubation experiment, three coarse-mesh (10 mm mesh size) and three fine-mesh (0.5 mm) litterbags of each species (fine-mesh bags only for cedar) were deployed in riffle-run habitats, along a 30 to 40-m reach in each stream. The 0.5-mm mesh netting precluded access by most stream macrofauna, and the presence of meiobenthic decomposers (e.g., nematodes) was shown to be poorly related to litter breakdown inside fine-mesh litterbags (Majdi et al. 2015). Therefore, microbial decomposition was considered the dominant pathway of litter breakdown in the 0.5-mm mesh litterbags. Each litterbag enclosed pre-weighed maple leaves (dry mass: 2 ± 0.01 g), and alder and cedar leaves (4 ± 0.01 g), and the litterbags were of similar volume across species. Litterbag deployment occurred in early autumn (mid-September to early October), coinciding with the period of natural litterfall, and litterbags were retrieved after 4-5 weeks. During retrieval, litterbags were placed inside plastic bags, and transported to the laboratory on ice. They were stored frozen at -20°C and in darkness if later processing was necessary. Leaves were (thawed and) gently rinsed with tap water onto a 250-µm sieve to remove invertebrates and sediments. Invertebrates retained on the sieves were preserved in ethanol for later identification and enumeration. For each coarse-mesh litterbag, three sets of five leaf disks (9.5-mm diameter) were randomly cut from the leaves using a cork borer, avoiding central veins. Two sets were used for the analyses of ergosterol – a surrogate for fungal biomass in decomposing leaves (Gessner 2005), and chlorophyll a – a surrogate for algal biomass (APHA 2017). The third set of disks, as well as leaves in litterbags, were oven-dried at 60°C to a constant mass, weighed (to the nearest 0.1 mg), and ashed at 550°C for 4 h in a muffle furnace to obtain ash-free dry mass (AFDM). Final leaf AFDM was corrected for handling losses and the leaf disks removed. Samples of each litter species not used in the experiment were leached for 24 hours in DI water, oven-dried, weighed, and ashed as above, in order to determine the percent loss in AFDM due to the initial release of water-soluble compounds (see Pozo et al. 2011). A 58  species-specific leaching correction factor was applied to adjust the initial litter AFDM to minimise the effects of potential leaching differences across species on HFA effects.  4.2.2.2 Shredders, fungal and algal biomass The determination of shredder densities, taxonomic richness, and community evenness in litterbags, and fungal biomass on leaf disks, was described in detail in Yeung, Lecerf & Richardson (2017) and Yeung et al. (2018). Briefly, macroinvertebrates associated with coarse-mesh litterbags were counted and identified to the lowest practical taxonomic unit (at family or genus level), except for Chironomidae in which subgroups corresponding to subfamily or tribe were identified (e.g., Orthocladiinae, Chironomini, Tanytarsini, Tanypodinae). The assignment of macroinvertebrates to shredders followed Merritt et al. (2008). Shredder density and taxonomic richness in each retrieved litterbag are equal to shredder abundance divided by remaining leaf AFDM, and the number of shredder taxa, respectively. The evenness of shredder community was represented by the Smith and Wilson evenness index Evar (Smith and Wilson 1996), which ranges from 0 (minimum evenness) to 1 (maximum evenness). To determine the ergosterol content on leaf disks, lipids were extracted from each set of disks by heating (30 min, 80 °C) them in 0.8% KOH-methanol. The extract was purified by solid-phase extraction (using Sep-Pak C18 Vac RC cartridges, 500 mg, 3 cc; Waters®, Milford, MA, USA) and eluted in isopropanol. Ergosterol extracts were then quantified by high-performance liquid chromatography (see experimental procedures in Gessner 2005). To convert ergosterol concentration on leaf disks to fungal biomass (per gram of leaf AFDM), I assumed an ergosterol concentration of 5.5 mg g-1 of mycelial dry mass (Gessner and Chauvet 1993). To estimate chlorophyll a concentrations on leaf disks, each set of disks was placed into a polypropylene conical tube with 5 ml ethanol. Chlorophyll a was extracted using the hot ethanol technique described in Francoeur, Rier & Whorley (2013). Briefly, the tubes were heated in a water bath (80°C) for 5 min, and steeped in ethanol at 4°C in the dark overnight. Prior to spectrophotometric analysis, the tubes were centrifuged at 4000 rpm for 2 min. Chlorophyll a concentrations (normalised for surface area of leaf disks) were calculated from the optical densities of the algal extract measured with a Beckman Coulter DU® 800 UV/Vis spectrophotometer (Fullerton, CA, USA). Optical densities were measured before and after the 59  addition of 0.3 M HCl (125 μl for each sample) to correct for phaeophytin a when determining chlorophyll a concentrations (Francoeur et al. 2013).  4.2.2.3 Water characteristics and microhabitat variables Water samples were collected during litterbag retrieval. Water characteristics (pH, conductivity, nitrate (NO3-) and soluble reactive phosphorus (SRP) concentrations) were analysed by the water chemistry laboratory at Canadian Forest Service, Great Lakes Forestry Centre, in Sault Ste. Marie, Ontario, using standardised methods (Nicolson 1988; see procedures in Hazlett et al. 2008). Water temperature was monitored every one or two hours during litterbag incubation by temperature loggers (TidbiT v2; Onset®, Bourne, MA, USA). Data were averaged to generate average daily temperature during the litterbag incubation period. Along the study reach in each stream, 10 measurements of wetted width, mid-channel depth, and current velocity at mid channel (Swoffer Model 2100, Swoffer Instruments Inc., Seattle, WA, USA) were made during litterbag placement, when the sites were at or close to baseflow levels. High-flow events (i.e., daily discharge exceeding 3 times the 3-year mean of median daily discharge from summer to fall) occurred once during the second half of the incubation period in each study year in MKRF, and once in the first study year in TLW (see Yeung et al. 2018).   4.2.2.4 Litter chemistry Foliar concentrations of macronutrients (C, N, P), micronutrients (K, Ca, Mg), and a recalcitrant compound (lignin) of the litter species were determined before incubation. Litter was ground to fine powder using a coffee grinder prior to analysis of elemental concentrations. To determine litter C and N content, samples were transferred to and pressed in 8 × 5 mm tin capsules, and then analysed by an Elementar vario EL cube elemental analyser (Elementar Analysensysteme GmbH, Germany) at the Stable Isotope Facility of the University of British Columbia (Vancouver, BC, Canada). For litter P, K, Ca, and Mg, samples underwent microwave digestion, followed by analysis by inductively coupled plasma mass spectrometry (ICP-MS) in the laboratory at Great Lakes Forestry Centre, Sault Ste. Marie, Ontario. Litter lignin content was 60  analysed using the acid detergent method by the Technical Services Section (Laboratory) of the British Columbia Ministry of Environment (Victoria, BC, Canada).  4.2.3 Data analyses A principal component analysis (PCA) was performed to summarise the differences in major chemical traits across litter species. The correlation of each chemical trait with the principal components was evaluated. The distances between the points on the PCA biplot represent the differences in measured chemical traits between litter species, and it was computed for species within each litter pair. The average PC axis 1 and 2 scores (PC1 and PC2) for each litter pair were also calculated. Temperature-normalised litter breakdown rates in coarse-mesh (kc; i.e., total breakdown rate) and fine-mesh (kf) bags were calculated as decay rate coefficients of the first-order exponential decay model: mt = e-kt, where mt is the proportion of initial leaf litter AFDM at litterbag retrieval, and t is the cumulative degree-days (in degrees Celcius) during the incubation period. Mean fragmentation rate of litter (denoted as λF; combining shredder feeding and physical abrasion) in coarse-mesh bags was determined, after accounting for microbial decomposition and dissolution rate (λm). Following the equations in Lecerf (2017), 𝜆𝐹 =  𝑘𝑐 −𝑘𝑓−𝑘𝑐ln(𝑘𝑓)−ln(𝑘𝑐) (Equation 1) 𝑘𝑐 =  𝜆𝐹 + 𝜆𝑚 (Equation 2) kf and λF are the focal breakdown rates, which are both standardised to the same unit (degree-day-1). The ratio of λF to kc approximates the proportional contribution of fragmentation rate to total litter breakdown. Lecerf’s (2017) approach of estimating pathway-specific breakdown rates takes into consideration the differences in these rates through the breakdown process (i.e., not constant through time), and assumes that the effects of these pathways are independent (i.e., the influences of microbial conditioning on fragmentation are not considered). However, this approach can better approximate the contributions of various agents of litter breakdown, and overcomes several shortcomings of previously used breakdown metrics. Home-field advantage index (HFAI) for each litter pair was calculated for kf, λF, shredder densities, taxonomic richness and community evenness, algal and fungal biomass on leaf disks, 61  using the equation of Ayres et al. (2009). HFAI indicates the percentage difference of each variable for the litter between its ‘home’ region and ‘away’ region, for each litter pair. HFAI (%) = [(𝐴𝑣𝑎+𝐵𝑣𝑏2/𝐴𝑣𝑏+𝐵𝑣𝑎2)] × 100 − 100 (Equation 3) where 𝐴𝑣𝑎 represents the value of variable v of species A averaged across sites in region a, and region a and b are ‘home’ regions to species A and B, respectively. For instance, Aa of kf equals kf of species A in its ‘home’ region a, and Ab of kf equals kf of species A in its ‘away’ region b. This index can control for inherent differences in breakdown rates between regions (e.g., due to stream nutrient concentrations) and litter species (Veen et al. 2014). To estimate the confidence intervals associated with the within-region variability of HFAIs, a resampling procedure was performed. For each variable in a litter pair, values of two species from 3 sites in each region were randomly resampled without replacement for 1000 iterations. This procedure allowed the assessment of whether the choice of sites and hence inherent inter-site variability might, by chance, mask HFA effects. The mean values of two species in their ‘home’ and ‘away’ regions for each iteration were recorded and used to compute the corresponding HFAI. The 95% confidence interval of the distribution of 1000 iterated HFAI values was equivalent to the range between 2.5% and 97.5% quantiles of the distribution. The significance of estimated HFAI from resampling iterations was assessed based on whether this confidence interval included zero. Values of each variable for home-derived and foreign litter were separately averaged within each site. Then, the site-averaged values for home-derived and foreign litter were paired in each region. Standardised influences of litter origin on each variable at the regional level were measured as Hedges’ d for these paired site data (or dunbiased; using equations 4 and 14 in Nakagawa & Cuthill, 2007). Note that Hedges’ d of a variable could still be negative (i.e., value for foreign litter greater than that for home-derived litter) in one region, when overall HFAI is positive. This is because the latter quantifies the additional extent of the variable when the incubation site of litter matches with its origin, and is averaged across two species. The proportional influences of litter origin on the variable at the site level were computed as ln-response ratio (RR), which is the natural log-transformed ratio of the mean of home-derived litter over the mean of foreign litter at a given site. The response ratio for litter quality PC1 and PC2 was not transformed as negative values were present. These effect size indices were hereafter 62  referred to represent the standardised (Hedges’ d) and proportional (RR) effects of litter origin. The 95% confidence intervals associated with Hedges’ d were calculated using the t distribution with appropriate df corresponding to the number of site pairs, so as to adjust for small sample size (instead of 1.96; using equations 15 and 18 in Nakagawa and Cuthill 2007). General linear models (GLMs) were used to test, across litter pairs and study regions, how the proportional influences of litter origin on kf and λF at each site were associated with initial litter quality (represented by PC1 and PC2 of litter chemistry analysis), litterbag-associated community attributes (shredder densities, taxonomic richness and community evenness, algal and fungal biomass on leaf disks; fungal biomass only for kf) as a proxy of decomposer preference and consumption, water characteristics (pH, conductivity, NO3-), and reach-scale habitat variables (wetted width, mid-channel depth, current velocity). SRP values in many streams were below detection limits, and were not included in the regression analysis. These predictor variables were normalised (using z-scores) prior to GLMs. For each of the litter quality and litterbag-associated community attributes, its mean value of each litter pair, and the proportional influences of litter origin on it, were separately used to construct GLMs. Linear mixed-effects models were not used in my study as the random effects of sites were assessed to be insignificant by log-likelihood ratio tests. Collinearity among predictor variables was assessed using the variance inflation factor (VIF) for each variable. Variable with the highest VIF was dropped sequentially and the VIF recalculated through backwards step-wise selection, until all remaining predicator variables had a VIF <2 (i.e., non-collinear variables; Zuur et al. 2010). The GLMs were fitted with an initial set of all non-collinear predictor variables, and their two-way interactions. Following Zuur et al. (2009), a backward stepwise selection of predictors was performed to obtain the simplest (optimal) model, which was validated with regard to assumptions of normality and heterogeneity of variance. For each region, the community dissimilarity of shredders on litter pairs was visualised by a non-metric multidimensional scaling (NMDS) graph. The differences in shredder community structure between litter pairs was assessed by one-way permutational multivariate analysis of variance (PERMANOVA). Site-level shredder density data were first used to generate the Bray-Curtis similarity matrix. PERMANOVA was performed using the adonis function in the vegan package (Oksanen et al. 2017). Within PERMANOVA, species was the 63  fixed factor, and the strata option was used to ensure that randomizations were only made within each site, in order to account for differences in shredder communities among sites. Post hoc, pairwise comparisons were performed to assess the significance of the dissimilarity in shredder community structure between home-derived and foreign litter (separately for alder and maple litter pairs), with p values adjusted by Holm’s correction (Holm 1979) using the pairwise.adonis function (Arbizu 2017) in R. Prior to all analyses, data were ln(x+1)-transformed as appropriate to meet assumptions of normality. All data analyses were carried out using R 3.4.1 (R Development Core Team 2018).  4.3 Results 4.3.1 Litter quality analysis The PCA of litter quality generated the first two principal components with eigenvalues >1, which cumulatively accounted for ~75% of the variation of chemical traits of the 6 study species (Appendix B: Table B.2). Therefore, the overall correlation between these traits between species was considered to be well summarised by the PCA biplot (Fig. 4.1). The first PCA axis (PC1) represented a strong gradient in litter N content, which was significantly positively correlated with % N, and negatively correlated with C:N and lignin:N. The second axis (PC2) was significantly positively correlated with % lignin. None of the micronutrients analysed were significantly related to PC1 and PC2. Average litter quality varied substantially between pairs (Fig. 4.1). The value of % N was highest in alder, intermediate in maple, and lowest in cedar, whereas C:N was lowest in alder, intermediate in maple, and highest in cedar. Within-litter-pair difference in litter quality (indicated by the distance between points of a pair on the PCA biplot) was greatest for alder (4.75), followed by maple (2.01) and cedar (0.85). Such differences were mainly contributed by % lignin for alder, and % N, C:N, and lignin:N for maple and cedar.   4.3.2 HFA: magnitude, variability, and drivers Percent mass loss of litter (in AFDM) in coarse-mesh bags tended to be higher for red alder (geometric mean across sites in ON: 40%; BC: 54%), followed by vine maple (ON: 29%; BC: 64  40%), speckled alder (ON: 28%; BC: 31%), and sugar maple (ON: 13%; BC: 25%). In fine-mesh bags, red alder also had greater percent mass loss (ON: 27%; BC: 42%), in comparison with those of vine maple (ON: 22%; BC: 22%), eastern white cedar (ON: 19%; BC: 10%), speckled alder (ON: 11%; BC: 14%), western red cedar (ON: 11%; BC: 13%), and sugar maple (ON: 11%; BC: 9%). Litter fragmentation rate (λF) was considered zero (rather than negative) when kf exceeded kc. This phenomenon only occurred for maple litter at some sites in ON, and the regional average of λF for all litter pairs were positive (Appendix B: Table B.3). Maple λF at Upper East in MKRF was excluded from HFA-related calculations, as the longitudinal displacement of some coarse-mesh bags during high flows appeared to cause additional dislodgement of leaf materials. Nevertheless, remaining leaf materials in these bags were sufficient for the determination of litterbag-associated community attributes at this site. Percent contribution of fragmentation to litter breakdown (λF/kc) ranged from 16-38% in BC and 19-34% in ON, which was comparable between home-derived and foreign litter. HFAI was observed to be positive for kf, and negative for λF, across all litter pairs (Fig. 4.2; Appendix B: Table B.4). In particular, HFAI for kf was most positive for cedar (45.2%), followed by alder (14.2%) and maple (7.9%). An average of 45.2% HFAI for kf of cedar indicated that, without HFA, the absolute mass loss rate of home-derived cedar litter in fine-mesh bags would have been roughly 4-5% smaller than that of foreign litter during the incubation period (Appendix B: Table B.4). In MKRF, western red cedar litter had an overall higher kf in comparison with eastern white cedar, as shown by a significantly positive Hedges’ d (Appendix B: Table B.5), although the former was of lower quality (higher C:N) and more recalcitrant (higher lignin:N). For litterbag-associated community attributes, observed HFA effects on them tended to show opposite directions between alder and maple litter pairs (Appendix B: Table B.4). The exceptions were shredder community evenness on which HFA effects were positive for both litter pairs, and fungal biomass in fine-mesh litterbags on which HFA effects were negative. Observed HFAI of all biotic variables examined was numerically similar to the medians of corresponding estimated HFAI from resampling iterations (Fig. 4.2; Appendix B: Table B.4). All values of observed HFAI were bounded by the 95th confidence intervals of estimated HFAI. Estimated HFAI was significant only for cedar kf (positive) and fungal biomass in fine-mesh bags 65  (negative), and nearly significantly positive for alder and maple kf. The variability of estimated HFAI (represented by its 95% confidence interval) for kf (alder: 41.9%; maple: 41.5%) was much smaller than that for λF (alder: 152.8%; maple: 68.8%) (Fig. 4.2; Appendix B: Table B.4). The optimal models of the proportional effects of litter origin (RR) on breakdown rates tended to be better explained by the proportional effects on explanatory variables (i.e., litter quality and litterbag-associated community attributes) than by the mean values of these variables (Table 4.1). All the retained variables were also shown to significantly influence RR on breakdown rates in the initial model with all first-order predictor variables. RR on λF was positively associated with RR on shredder taxonomic richness, but was unrelated to the mean value of any predictor variable. RR on kf was positively associated with RR on fungal biomass, and negatively associated with the mean value of mid-channel depth and RR on litter quality PC2 (i.e., % lignin), respectively. Proportional effects on fungal biomass and litter quality combined to explain more variation of RR on kf (36%) than did the mean values of mid-channel depth (18%). However, when analysing cedar data only (rather than all litter pairs), none of the proportional effects on explanatory variables were significantly associated with RR on cedar kf.  4.3.3 Shredder assemblage dissimilarity The NMDS ordination plots well represented data on shredder community structure in litterbags across regions, as indicated by the fair stress values (0.12 and 0.20; Fig. 4.3). When accounting for site-level variability within regions, overall shredder community dissimilarity between litter pairs was significant in ON (PERMANOVA F3,18: 1.83, R2: 0.23, p = 0.004), and marginally significant in BC (PERMANOVA F3,18: 1.69, R2: 0.22, p = 0.059). However, in both regions, no pairwise comparisons of shredder community structure were significantly different between home-derived and foreign litter within litter pairs.  4.4 Discussion Our results indicate that the efficiency of stream microbial decomposers to process litter resources in the home region was considerably greater for conifer litter than for broadleaf litter. This was reflected by more positive and significant HFA effects on temperature-normalised 66  microbial decomposition (kf) of lower-quality conifer litter. In addition, consistent with my hypothesis, HFA effects on fragmentation rate (λF) were less positive than those on kf, and more variable among streams. My findings therefore suggest that the community-level efficiency of litter processing by stream microbial decomposers, and not shredders, likely depended on their past exposure to litter in the home region (see Kominoski et al. 2011; Jackrel and Wootton 2015; Lidman et al. 2017), which could indirectly control the relative difference in kf between home-derived and foreign litter. Litter origin did not seem to strongly influence the preference of shredders for colonising and consuming broadleaf litter.  4.4.1 Litter-microbe interactions Litter kf generally increased when litter had higher fungal biomass and lower initial lignin content. This finding generally corroborated those of previous studies (Gessner and Chauvet 1994; Lecerf and Richardson 2010; but see Richardson et al., 2004), as lignin content in litter regulates the availability of carbon to aquatic fungi and in turn controls kf. In most of my study streams, kf contributed more to kc than λF for broadleaf litter, as has been previously observed (Yeung et al. 2018; Elosegi et al. 2018). Litter-microbe interactions apparently played a more essential role than litter-shredder interactions in mediating in-stream breakdown and transport, at least of broadleaf litter. Significantly positive HFA effects on kf of for cedar litter, but not of broadleaf litter, illustrated the importance of both litter origin and quality in modifying the outcomes of litter-microbe interactions. One plausible explanation of positive HFA effects is that stream microbes are physiologically primed to degrade home-derived litter within particular ranges of chemical traits, such as lignin and/or other phenolic compounds (Freschet et al. 2012; Mora-Gómez et al. 2016). This past exposure to benthic litter could also have induced compositional shifts of microbes, whereby taxa more efficiently processing home-derived litter outcompeted and replaced less efficient ones (Jackrel and Wootton 2014). More positive HFA effects on kf of conifer litter suggest that microbes specialised on degrading home-derived conifer litter were probably primed to process litter within narrower ranges of nutrient concentrations (or stoichiometric ratios) and/or recalcitrance, compared to microbes that degrade broadleaf litter. This can explain why microbial decomposers were much less efficient in processing foreign 67  conifer litter than the home-derived one, despite their relatively small differences in chemical quality as observed in this study. I am uncertain why higher kf was associated with lower fungal biomass on home-derived cedar litter. It is possible that fungal biomass might peak prior to, or after the end of litterbag incubation (Ferreira and Chauvet 2011; Artigas et al. 2011), which could affect fungal influences on home-derived and foreign litter observed in my study. Also, I did not have data on bacterial and fungal dynamics on litter, in terms of their biomass, community composition, and extracellular enzyme activities, throughout the incubation period, which would help elucidate kf differences between home-derived and foreign litter (e.g., Hieber and Gessner 2002; Duarte et al. 2010; Mora-Gómez et al. 2016). For instance, positive HFA effects on kf occurred for black alder (Alnus glutinosa (L.) Gaertner) and common reed (Phragmites australis (Cav.) Trin. ex Steud.) in mountain and lowland streams in southern Spain (when regional means of kf were used to calculate HFAI; Fenoy et al. 2016). Home-derived litter in these streams tended to support higher than expected fungal biomass and activity of β-glucosidase – a key enzyme involved in cellulose decomposition (see Artigas et al. 2011). Thus, I suggest further studies that link HFA effects on kf to the activity and accrual dynamics of stream microbial communities across multiple time points and over longer periods of incubation (i.e., >4-5 weeks; particularly for slowly decomposing conifer species). These studies can reveal the relative contributions of fungi and bacteria to HFA (see Lin et al. 2018), and potential changes in HFA effects along the course of litter decay (see Wallenstein et al. 2013; Chávez-Vergara et al. 2018). It is suggested that the algal biofilm on litter can provide labile dissolved organic carbon to stimulate the growth of heterotrophic microbes, promoting litter breakdown through priming effects (Danger et al. 2013; Kuehn et al. 2014). In fine-mesh bags, I found on average higher algal biomass on home-derived maple (but not alder) litter. Priming effects appeared to only partly explain positive HFA effects on kf during fall, and could be overridden by other drivers of litter breakdown (see also Elosegi et al. 2018).  In this study, I did not collect and incubate litter from the riparian zone adjacent to (or upstream of) each site to determine HFA effects, thus I did not strictly evaluate the adaptation of decomposers to litter resources at the local stream scale. It is difficult to predict how HFA effects on kf would compare to those observed in this study if local litter sources were used instead, 68  without knowing the differences in quality between local litter and the standardised litter stock from the home region, and the benthic litter composition at each site (Freschet et al. 2012; Jackrel and Wootton 2015; Lidman et al. 2017). Future work is needed to assess whether the dissimilarity in the quality of incubated litter and the average quality of the benthic litter layer (driven by dominant species) can importantly explain the proportional effects of litter origin on kf in individual streams (see Freschet et al. 2012; Frossard et al. 2013; Veen et al. 2014).  4.4.2 Litter-shredder interactions Shredders did not appear to colonise and consume home-derived litter more than expected at the end of the incubation period, as inferred from negligible, or slightly negative HFA effects on λF. In particular, in TLW, foreign red alder and vine maple had lower litter lignin content and/or C:N, and supported higher shredder densities than home-derived speckled alder and sugar maple, respectively. Numerous studies also demonstrated greater influences of litter quality on shredder aggregation and feeding on litter than those of litter origin (Richardson et al. 2004; Hladyz et al. 2009; Hisabae et al. 2011; Kuglerová et al. 2017a), as shredders tended to preferentially feed on higher-quality litter to meet their energetic and elemental requirements (see Hladyz et al. 2009). I detected clear colonization preference of higher-quality litter by the numerically dominant shredder taxa, including the stonefly Zapada spp., caddisfly Lepidostoma spp., and orthoclad midge Brillia sp. in MKRF (see also Richardson et al. 2004; Kominoski et al. 2011; García et al. 2012), and Lepidostoma spp. in TLW. However, this preference was absent among many of the less abundant taxa, which might lead to an overall similar shredder taxonomic richness, and community evenness and structure between home-derived and foreign litter within the same litter pair (see also Muto et al. 2011). Therefore, more abundant shredder taxa under greater degree of intraspecific competition appeared to have higher resource-use plasticity, and they could opportunistically colonise and consume higher-quality litter. In this study, I did not compare λF and shredder colonization between cedar species. Future studies will be required to examine if differences in taxon-specific selectivity of litter resources among shredders extend to low-quality conifer litter for consumption and/or case-building by caddisflies (see Richardson et al. 2005). When provided litter choices with more subtle differences in quality (e.g., defensive compounds), such as the same litter species (red alder) from different populations, Jackrel et al. 69  (2016) found that shredders consumed litter derived from the ‘home’ river more quickly than from the ‘away’ river. Hence, the outcome of litter-shredder interactions could vary with the choice of litter pairs offered to shredders, in terms of their differences in multiple aspects of litter quality. Importantly, HFA effects on litter breakdown could be affected by the study design through the designation of ‘home-derived’ and ‘foreign’ litter by researchers (see also Palozzi and Lindo 2018). Shredder taxonomic richness was the only community attribute measured whose differences between home-derived and foreign litter contributed positively to those of shredder feeding. This finding is in line with previous observations of faster litter breakdown rates in the presence of higher shredder richness in my study streams (e.g., Ruesink and Srivastava 2001; Kreutzweiser et al. 2005a; Lecerf and Richardson 2010), which might be explained by the facilitation of smaller species by larger ones, and niche partitioning of litter resources (Tonin et al., 2018).  4.4.3 Variability of HFA on litter breakdown While I highlighted the importance of considering litter-decomposer interactions in explaining local patterns of litter breakdown, inter-stream variability in biophysical variables, such as channel depth, could be large enough to mask the positive HFA effects on kf, and negative HFA effects on λF, of broadleaf litter. In the study regions, inter-stream differences in riparian forest composition and physical habitat attributes could cause variations in benthic litter quality and quantity (Kreutzweiser et al. 2004; Negishi and Richardson 2006; Kiffney and Richardson 2010; Hoover et al. 2011), and the activity and community of decomposers (McArthur and Richardson 2002; Kreutzweiser et al. 2005b; Kominoski et al. 2011). These variations likely induced the observed inter-stream variability of HFA effects on litter breakdown. Quantitatively linking the responses of microbial decomposer communities to the biophysical variables described above and kf is beyond the scope of this study. However, as a first step, I established that the significance and magnitude of HFA effects on kf (and λF) across streams co-varied with the identity and number of study sites with inherent differences in local decomposer communities. HFA effects on total litter breakdown (kc) further depend on the relative contributions of its components – shredder feeding and microbial decomposition – which 70  could differ considerably across streams as shown in this study and others (Tiegs et al. 2009; Hladyz et al. 2010; Yeung et al. 2018). Therefore, this study established that the significance and magnitude of HFA effects on litter breakdown across streams co-varied with the identity and number of study sites with inherent differences in local decomposer communities.  4.4.4 Implications for comparing breakdown between native and exotic litter Many studies have sought to investigate how invasions by exotic plant species (derived from outside their natural range) and replacement of native forests by plantations affected stream litter breakdown (see meta-analyses by Ferreira et al. 2016a; Kennedy and El-Sabaawi 2017). Local adaptation of decomposers to process litter is not typically considered as an ecological mechanism driving differences (or the lack thereof) between the breakdown rate of native and exotic litter in streams. I found that benthic litter quality and quantity could be site-specific environmental factors that interacted with microbial decomposers to confer faster breakdown (kf) of home-derived litter than expected. Characterization of the quality traits of the benthic litter layer, and their differences from those of exotic litter species (indicating relative processing efficiency of microbial decomposers), may allow more mechanistic, a priori predictions of exotic litter kf. If the designated native species is locally abundant in the benthic litter layer, the quality of the home-derived species may be directly used to approximate the average quality of the litter layer (see Veen et al. 2014). Using the data from studies comparing the breakdown between native and exotic litter included in the meta-analysis by Kennedy and El-Sabaawi (2017) and this study, I showed that the difference in quality (represented by C:N) was fairly well associated with that in kf between native and exotic litter (F1,8 = 22.6; p < 0.01; R2 = 0.74; Fig. 4.4a). In comparison, uniquely considering the absolute values of litter C:N (and not litter-microbe interactions) within the same dataset led to poor predictions of the kf of native and exotic litter, and the dissimilarity in kf between them (Fig. 4.4b). Results of this simple regression analysis were derived from a few studies and only considered C:N, and not other litter traits (e.g., lignin and other phenolic compounds) that microbial decomposers might also be specialised to. Also, it did not consider potentially greater HFA effects on kf if more recalcitrant litter species were involved in kf comparisons between native and exotic litter. Despite these caveats, it shows a strong relevance of incorporating some 71  aspect of local litter-microbe interactions into predicting exotic litter kf, in addition to intrinsic litter quality and extrinsic environmental factors (e.g., pH, conductivity; Kennedy and El-Sabaawi 2017). Future studies are recommended to quantify these factors and move beyond categorically examining the effects of litter origin, in order to better account for the differences in kf between native and exotic litter.  4.4.5 Conclusions I conclude that consumer specialisation by microbes on home-derived litter can drive the spatial variability of the processing of litter subsidies in stream ecosystems. Stream microbial decomposers probably had lower resource-use plasticity than shredders at the community level, and/or underwent compositional shifts favouring taxa which more efficiently processed locally abundant litter. Additional investigations of the activity and community composition of microbes across litter species (particularly conifer), and benthic litter composition, will allow us to better understand the basis of the local adaptation of microbes in resource processing. It is worth testing further whether the specialisation of freshwater microbes on locally derived litter is widespread (see Franzitta et al. 2015; Leroy et al. 2017; Xie et al. 2017), and temporally variable. If so, litter-decomposer interactions would likely be more dynamic than have been recognised, under the intensifying impacts of global environmental change (including range shifts of species and genotypes/populations) and human activities on riparian vegetation and freshwater decomposer communities (e.g., Kominoski and Rosemond 2011; Kominoski et al. 2013; Ferreira et al. 2016a).72  Table 4.1 Results of regression analyses on the proportional effects of litter origin on stream litter breakdown rates (λF and kf), as influenced by water characteristics, and either average litter quality and litterbag-associated community attributes of home-derived and foreign litter species, or proportional effects of litter origin on these attributes. The optimal model for each breakdown rate is shown (* P < 0.05; ** P < 0.01). Data from G stream in Malcolm Knapp Research Forest were omitted from the analysis due to incomplete records of microhabitat variables.  Model fit Variable Parameter estimate SE t Degrees of freedom P (A) Influences of water characteristics, and average litter quality and litterbag-associated community attributes, and   1. λF (no explanatory variables were significant in the full model) 2. kf        Adjusted R2: 0.176 Mid-channel depth -0.197 0.072 -2.721 29 0.011 *        (B) Influences of water characteristics, and the proportional effects of litter origin on litter quality and litterbag-associated community attributes  1. λF       Adjusted R2: 0.199 Shredder taxonomic richness 0.314 0.134 2.338 17 0.032 * 2. kf       Adjusted R2: 0.355 Fungal biomass 0.232 0.064 3.625 28 0.001 ** Litter quality PC2 -0.145 0.063 -2.277  0.031 *         73    Figure 4.1 Principal component analysis (PCA) biplot of 8 litter quality traits (arrows) measured for the 6 plant litter species (points) in the study regions in British Columbia (closed circles) and Ontario (open circles), Canada. RA = red alder (Alnus rubra); SA = speckled alder (Alnus incana ssp. rugosa); VM = vine maple (Acer circinatum); SM = sugar maple (Acer saccharum); WRC = western red cedar (Thuja plicata); EWC = eastern white cedar (Thuja occidentalis). 74    Figure 4.2 Observed and estimated (from resampling) home-field advantage index (HFAI) for in-stream (a) fragmentation rate, λF, (b) litter breakdown rate in fine-mesh bags, kf, (c) algal biomass on leaf discs in coarse-mesh bags, (d) fungal biomass on leaf discs in fine-mesh bags, and (e) shredder density and (f) shredder taxonomic richness of two or three litter pairs incubated in the study regions. For estimated HFAI, median values and error ranges representing the 95th confidence intervals (CI) are shown. Asterisks indicate that values of estimated HFAI are significant (i.e., 95% CI not overlapping zero; * P < 0.05).75    Figure 4.3 Non-metric multidimensional scaling ordination of shredder assemblages on two litter pairs of alder (red alder: RA; speckled alder: SA), and maple (vine maple: VM; sugar maple: SM) in (a) British Columbia and (b) Ontario, Canada. Ellipses represent 95% confidence intervals for the individual species across sites. Note that NMDS axis scores are not comparable between study regions.76    Figure 4.4 Relationships between (a) differences in litter C:N between native and exotic species, and the ln-response ratio of kf (in day-1) of native species relative to the exotic species (RRnative:exotic of kf), and between (b) litter C:N and kf of native and exotic species. The source of data in (a) and (b) were identical, and from studies selected by the meta-analysis by Kennedy and El-Sabaawi (2017) and this study (only including TLW data). Only studies using litterbags with mesh size <0.5 mm to measure kf were included. When there were multiple native (exotic) species involved in a given study selected by Kennedy & El-Sabaawi (2017), kf values were averaged to give a single value for each native (exotic) species group, prior to calculating RRnative:exotic of kf. Values from multiple sites in each study were averaged. For my study, comparisons between values of native and exotic litter were made for specified litter pairs for plotting (a). The black line in (a) indicates ordinary least-squares regression (y = -0.0125x – 0.407; R2 = 0.74) through all plotted points, whose slope is significantly negative. Results of power regression analysis in (b) show weak and insignificant power relationships between litter C:N and kf of native litter (y = 0.0352x-0.409; F1,12 = 1.19; p = 0.30; R2 = 0.09) and exotic litter (y = 0.0637x-0.465; F1,8 = 2.69; p = 0.14; R2 = 0.25).77  Chapter 5: Modelling biophysical controls on stream organic matter standing stocks under variable forest harvesting impacts  5.1 Introduction Land-use change and disturbances in watersheds can differentially alter physical, chemical, and biological processes in streams (e.g., Mellina and Hinch 2009; Richardson and Béraud 2014; Ferreira et al. 2016a; Rogger et al. 2017). Understanding the relative importance of the controlling processes and their interactions is paramount to better assessing and forecasting disturbance effects on ecological variables (Evans et al. 2012). In addition, such process-based understanding can be applied to formulate more targeted management and/or restoration measures to mitigate anthropogenic disturbance impacts on stream ecosystems, and facilitate their recovery (e.g., Quinn et al. 2007; Beechie et al. 2010; Cuddington et al. 2013).  In forested headwater streams, terrestrial-derived particulate (e.g., leaf litter, fruits, seeds) and dissolved organic matter provides critical energy and resource subsidies to consumers, and controls many ecosystem processes in the nutrient-limited environment (Richardson and Sato 2015). In particular, the quantity and quality of coarse particulate organic matter (CPOM, >1 mm in diameter) – the focus of this study – can strongly influence stream food-web productivity through detritivorous invertebrates (shredders), and the predators depending on them (Richardson 1991; Wallace et al. 1999; Kominoski et al. 2012). The breakdown of CPOM (through shredder feeding, microbial decomposition, and physical abrasion; Graça et al. 2015) and subsequent production of fine-particulate organic matter (~0.45 µm – 1 mm in diameter) can subsidise invertebrate collector-gatherers in local reaches and further downstream (Richardson and Neill 1991; Bundschuh and McKie 2016). CPOM can also provide substrates for microbial activity and indirectly mediates stream metabolism and nutrient retention (Crenshaw et al. 2002; Aldridge et al. 2009). Given the ecological importance of CPOM in shaping stream food-web dynamics, a number of empirical studies have sought to investigate the ecological impacts of anthropogenic watershed disturbances, predominantly forest harvesting, through assessing the responses of CPOM standing stocks. Among studies of forest harvesting impacts, both post-78  harvest increases (Garman and Moring 1991), decreases (Golladay et al. 1989; Garman and Moring 1991; Göthe et al. 2009; Santiago et al. 2011), or insignificant changes (Smolders et al. 2018) in CPOM standing stocks have been reported. Such diverse responses could be attributed to differences in wood inputs after harvesting (Garman and Moring 1991), harvesting practices (e.g., variation in the configurations of riparian buffers), and/or environmental context across study sites (Richardson and Béraud 2014). In order to better explain the spatial variability in stream CPOM responses to forest harvesting (and other disturbances), it is necessary to quantitatively relate them to the key biophysical processes controlling CPOM standing stocks, including, for instance, riparian litter inputs, discharge, and temperature (Webster 1983; Webster et al. 2001; Stenroth et al. 2014). It is logistically prohibitive to deploy long-term (e.g., multi-year) manipulations of these reach- and catchment-scale processes in combination, using a replicated, factorial design to understand their relative importance in mediating CPOM standing stocks (Bennett and Adams 2004). I therefore adopted process-based modelling as a means towards attaining this goal through varying rates of these processes to simulate post-harvest CPOM responses. In this study, I used the process-based CPOM model formulated by Stenroth et al. (2014), which is a mass-balance, ecosystem model that can simulate temporal dynamics of leaf litter-derived CPOM (excluding wood) on the streambed. This model was modified from its original versions developed by Richardson et al. (2009) and Karlsson et al. (2005), and was specifically developed for small streams (<10 m bankfull width). The external biophysical parameters of this model (or model drivers) are known to exhibit variable responses to forest harvesting (i.e., aerial litter inputs, Kiffney and Richardson (2010); discharge, Moore and Wondzell (2005); stream temperature, Moore et al. (2005a)). Therefore, I advanced the application of this model by investigating how differential changes in the model drivers, reflecting a wide range of possible forest harvesting scenarios, could influence CPOM standing stocks in similar stream-riparian systems. My emphasis was on the short-term responses (<5 years after forest harvesting) of model drivers, which are typically greatest and vary little within this timeframe (see also Grant et al. 2008; Sweeney and Newbold 2014). My first study objective was to perform a sensitivity analysis as an initial comparison of the relative importance of the model drivers in altering CPOM standing stocks in a focal small, 79  forested stream in coastal British Columbia, Canada. I considered a broad, realistic range of changes in model drivers likely to be induced by forest harvesting in this sensitivity analysis, in contrast with only two levels for each model driver in the analysis by Stenroth et al. (2014), in which actual field data for a forested and an open stream were interchanged for the model drivers. To establish a realistic range of disturbance severity, I conducted a literature summary and analysis of the post-harvest responses of aerial litter inputs, discharge, and temperature in small temperate streams in North America. The second study objective was to heuristically vary model drivers in different combinations to examine specific forest harvesting scenarios under which CPOM standing stocks in the focal stream would be more (or less) likely to be reduced relative to unharvested conditions. This exploratory modelling approach also allowed us to detect complex, previously unknown interactions between multiple model drivers (e.g., synergism, antagonism), and to describe the nature of two-model driver interactions based on the classification of Piggott et al. (2015b). My study findings can reveal which model driver(s) should be the management focus to more effectively and efficiently minimise the impacts of anthropogenic disturbances in similar forested watersheds on stream CPOM standing stocks.  5.2 Methods 5.2.1 Literature analysis of post-harvest responses of model drivers To evaluate the relative influence of post-harvest changes in discharge and stream temperature on CPOM standing stocks, I focussed on their major components, which were peak flows, and late spring to Northern summer (June to September) stream temperature, respectively. This is because post-harvest increases in peak flows could more substantially enhance the advection input of CPOM and its re-entrainment than changes in low-flow conditions (e.g., Gurtz et al. 1988; Kiffney et al. 2000; Richardson et al. 2005). Also, post-harvest increases in stream temperature – affecting many biological processes related to CPOM – tended be greater in summer than in winter, due to lower energy inputs and greater discharge in winter in similar rain-dominated streams (Moore et al. 2005a; Leach et al. 2012). Furthermore, this approach limited the number of drivers in the model to ensure an informative sensitivity analysis (Cuddington et al. 2013). 80  I reviewed published responses of riparian litterfall, peak flows, and stream summer temperature (from around June to September) within about 5 years since forest harvesting in small temperate streams in North America. Data selection was restricted to low-elevation (≤1000 m above sea level), rain-dominated streams, as their physical habitats are most comparable to those of my focal study stream, East Creek in Malcolm Knapp Research Forest (MKRF; 49°16’N, 122°34’W), in coastal British Columbia. When responses were analysed across time periods (e.g., 0-8 years) elapsed since harvesting (rather than individual time points), the mid-point of the specified period (i.e., 4 years) was noted to check whether the response qualified for inclusion in the review (i.e., within a 5-year post-harvest window). For each measured response, the characteristics of watershed disturbances associated with forest harvesting, including the type of harvesting practices (e.g., clearcut, partial cut), percentage of the watershed area harvested, the presence and the width of riparian buffers, were also recorded. The first phase of literature search was conducted by cross-checking references in relevant review studies of logging-associated changes in peak flows (Guillemette et al. 2005; Moore and Wondzell 2005) and stream summer temperature (Moore et al. 2005a; Sweeney and Newbold 2014). The reported values of changes were verified by checking the original papers cited in these studies. To identify additional literature, the second phase of bibliographic searches using online databases (including Web of ScienceTM and Google Scholar®) were performed in September 2018. Keywords used in the searches were stream*, harvest*, logging*, clearcut*, thinning*, and retention*, in combination with litterfall*, litter input*, with peak flow*, peak discharge*, and with temperature*. Very few studies on riparian litterfall responses to forest harvesting within 5 years of harvests were returned in the initial search. For this reason, the search extended to all literature of post-harvest responses of riparian litterfall in temperate North America, regardless of the timing of measurements taken. The literature range of responses of model drivers was considered to realistically encompass the range of uncertainty surrounding these drivers, or the severity of forest harvesting disturbance affecting them, that could occur in East Creek. The 25th percentile, 50th percentile (median), 75th percentile, and the maximum of the data distribution of published responses of stream temperature and peak flows were computed. Peak flows responses varied greatly with the extent of watershed area harvested, therefore these percentile values were calculated separately 81  for response distributions specific to the groups of watershed area logged (i.e., 0-40%; 41-80%, 81-100%). These percentage harvest group ranges were established by Grant et al. (2008) in consideration of the conceptual scaling of changes in hydrologic processes affecting peak flows with the spatial extent of harvesting, and also the data spread of peak flow responses to harvesting. Post-harvest stream temperature responses tended to be much attenuated when riparian buffers were present, hence these temperature responses were excluded from percentile calculations. This was to ensure that these percentiles would not be skewed towards lower values by a considerable portion of studies of post-harvest thermal responses that addressed the mitigating effects of riparian protection. The percentile values of the response distributions of peak flows and stream temperature, and values chosen along the range of litterfall responses based on their data spread, provided a realistic basis of assigning various levels along the gradient of disturbance severity (i.e., low, moderate, high, and very high severity) to the model drivers under forest harvesting.  5.2.2 Stream CPOM model The stream CPOM model employed in this study incorporates important pathways of CPOM inputs and outputs (e.g., riparian litter inputs, re-entrainment and downstream transport, consumption by shredders, microbial respiration; Stenroth et al. 2014). This model estimates the probability of litter re-entrainment as a function of the relative change in stream discharge (to account for antecedent hydrologic conditions), rather than discharge as in the previous model of Richardson et al. (2009). The contributions of in-stream wood and its fragments to CPOM standing stocks are not considered in this model. The parameterization of the stream CPOM model was identical to that in Stenroth et al. (2014) for the sensitivity analysis and heuristic modelling, using the same set of empirical data from representative small forest streams (including East Creek) in coastal British Columbia. These streams typically have wetted width <2.5 m with steep slopes (gradient: ~15-30%), and riparian vegetation dominated by red alder (Alnus rubra Bong.) and coniferous tree species (for descriptions of climate and other habitat characteristics, see Karlsson et al. 2005; Kiffney and Richardson 2010).  82  5.2.3 Sensitivity analysis and heuristic modelling I performed a simple sensitivity analysis to quantify the effects of individual model drivers relative to others on CPOM standing stocks in East Creek, along a severity gradient of forest harvesting disturbance. One driver was adjusted by a categorical level of disturbance severity, while all other drivers were kept unchanged. Note that in this sensitivity analysis, systematic adjustment by the same degree of disturbance was applied to model drivers, but not by the same ratio. This is because stream temperature is an interval scale, and hence could not be altered by a ratio of change as for the other two drivers. In the heuristic modelling, I considered scenarios under the effects of all three model drivers across all severity levels of forest harvesting disturbance. Model drivers were manipulated across five levels of disturbance severity (i.e., undisturbed, low, medium, high and very high severity) in a full-factorial design. Among a total of 125 possible combinations of hypothetical disturbance scenarios generated, 12 of these scenarios (with changes in single drivers only) were the same as those considered in my sensitivity analysis. The procedures of modelling CPOM standing stocks followed Stenroth et al. (2014), and was run in Stella version 10.0.4 (ISEE Systems, Lebanon, NH, USA). For each scenario, the model was set to generate outputs with 1500 time steps (i.e., between 4 to 5 years), with each time step set to one day. CPOM standing stocks were expressed as ash-free dry mass per unit area of streambed (g AFDM m-2). Daily input values of leaf litter were empirically determined and averaged across three streams close to East Creek, which added up to an annual input of 310 g AFDM m-2 yr-1 (Richardson 1992). Daily mean stream temperatures of an average year were derived from temperature records for East Creek (empirically measured from May 1997 to August 2002). The value for each day of the annual temperature curve was smoothed using the temperature of that day and seven subsequent days (see Fig. 2a in Stenroth et al. 2014). Hydrographs used in the model were generated from discharge data from East Creek (1989-2002). A 1500-day hydrograph of average daily discharge (L s-1) was used for each simulation, and the first 90 days of modelling results – as the spin-up period – were discarded to accommodate for the initial stabilization for the model components. Daily CPOM values of the rest of the period were then averaged. An ensemble of ten simulations, each associated with a hydrograph starting on 1st January of a different year (i.e., 1989-1993, 1990-1994,…, 1998-83  2002), were performed. Therefore, ten hydrograph-specific values of average daily CPOM standing stocks were simulated for each heuristic scenario. In the heuristic modelling, I assumed constancy of disturbance severity with no recovery of the model drivers throughout the duration of model runs. Annual (and daily) litterfall values were perturbed to the same extent for each level of disturbance severity, without altering the time-specificity of litterfall (i.e., unchanged percentage contribution by individual days to annual litterfall). Peak flow events driven by storm runoff were delimited from the hydrograph, prior to perturbations. Two adjacent local maxima of discharge were considered storm-induced, when the smaller of these peaks was more than double the intervening local minimum. Exploratory runs of this arbitrary splitting procedure identified distinct discharge pulses, whose timing closely tracked that of rainfall episodes recorded at a climate station in MKRF (see example in Appendix C: Fig. C.1). This procedure effectively avoided generating false-positive ‘peak flows’ due to minor, natural fluctuations of flows. In each simulation, all peak flows identified (~30-40 events in each annual hydrograph) were perturbed to the same extent for each level of disturbance severity, while all the other components of the hydrograph were kept unchanged. Daily stream temperature during late spring to summer (i.e., June to September) of the smoothed annual temperature curve was also perturbed to the same extent for each level of disturbance severity.  5.2.4 Data analyses For each heuristic scenario, the natural log of the ratio of hydrograph-specific daily CPOM standing stocks in East Creek to that of the undisturbed (control) scenario (i.e., ln[hydrograph-specific response ratio of CPOM]) was calculated. Values of this response ratio were then averaged across hydrographs (i.e., simulations) to give a mean response ratio (i.e., ln[response ratio of CPOM]), which represented the average effect size of each heuristic disturbance on CPOM standing stocks. A three-way ANOVA was used to test the main effects of the three model drivers across levels of disturbance severity, and their two-way and three-way interaction terms, on ln(hydrograph-specific response ratio of CPOM). This response ratio was analysed as the dependent variable in ANOVA, rather than daily CPOM standing stocks, to isolate the confounding effects of interannual variations in discharge. 84  The effects of two-way interactions of model drivers on CPOM standing stocks at each level of disturbance severity was classified into various directional interaction types (i.e., additive, positive/negative synergistic, positive/negative antagonistic), according to the conceptual framework by Piggott et al. (2015b). Directional interaction types were determined based on the magnitude and response direction of daily CPOM standing stocks, and the interaction effect (i.e., effect deviation from the additive sum of effects of model drivers acting in isolation). For each heuristic scenario, hydrograph-specific daily CPOM standing stocks predicted by a simple additive null model were calculated according to Schäfer and Piggott (2018). The additive null model assumes that the effects of two-model driver interactions are absent, and therefore predicts the cumulative effects of two model drivers based on the effects of individual drivers (Schäfer and Piggott 2018). The difference in hydrograph-specific daily CPOM standing stocks between the values simulated under a given heuristic scenario and the unharvested scenario was compared with the difference between the values predicted using the additive null model and the unharvested scenario, using a Wilcoxon signed-rank test across ten simulations. At a given level of disturbance severity, statistically significant results of this test indicate that the directional interaction type classified for each two-model driver interaction is significantly non-additive. Data were ln(x+1)-transformed as appropriate to meet assumptions of normality. All data analyses were carried out using R 3.5.1 (R Development Core Team 2018).  5.3 Results 5.3.1 Published post-harvest responses of model drivers Most compiled data of logging-associated changes in peak flows and stream summer temperature in low-elevation watersheds in North America were collected from the Pacific Northwest region, particularly MKRF in British Columbia, Canada, and H.J. Andrews Experimental Forest, Alsea and Trask Watersheds in Oregon, USA (Appendix C: Tables C.1-2). Post-harvest litterfall changes were measured in a relatively small number of studies, which reported greater litterfall reductions in streams without riparian buffers than with buffers (Appendix C: Table C.3). Post-harvest changes in peak flows ranged from -20% to 300%, and the magnitude of peak flow increases tended to be greater for sites with 41-80% of watershed area harvested than for sites 85  with less than 40% of watershed area harvested (Appendix C: Table C.4). There were variations in the designation of summer months by studies of post-harvest changes in stream temperature, but typically included the period from June to August and/or September, thus matching the period of temperature perturbation in the heuristic modelling. The range of responses of stream summer temperature to forest clear-cutting without riparian buffers was from -1.6 °C to 6.0°C (Appendix C: Table C.4). To arbitrarily assign the levels of disturbance severity of forest harvesting at East Creek (from low to very high), I referred to the quartiles and maximum values of the response distributions of stream temperature (1-6°C increases), and peak flows (20-300% increases; considered across groups of percentage watershed area harvested). Representative values of litterfall reductions (10-90% reductions) reported across studies were selected to form the gradient of disturbance severity (Table 5.1).  5.3.2 Sensitivity analysis and heuristic modelling of CPOM Modelled average daily CPOM standing stocks at East Creek decreased under the independent effects of reduced litterfall and elevated peak flows, and increased with higher stream temperature (Table 5.2). The magnitude of these effects increased significantly with the severity of forest harvesting disturbance (P < 0.001; Table 5.3). Along this severity gradient, the effects of litterfall reductions on depleting CPOM standing stocks were at least an order of magnitude greater than those of elevated peak flows. For example, litterfall reductions at low severity led to greater CPOM decreases on average (-14%) than did peak flow increases at very high severity of disturbance (-5%; Table 5.2). The magnitude of CPOM changes induced by litterfall reductions was consistently greater than stream temperature increases, but their differences in magnitude became smaller at higher levels of disturbance severity (Table 5.2). The associated alterations in the temporal dynamics of modelled CPOM standing stocks depended on the identity of model driver(s) affected, and the level of disturbance severity. With litterfall reductions and peak flow increases, CPOM standing stocks were depleted faster in winter (from November to January) to reach near-zero levels than in unharvested conditions (Fig. 5.1). CPOM accumulation was slower in late summer (especially from August to September) with reduced litterfall in general, and faster with increased temperatures only at high or very high 86  levels of disturbance severity. With simultaneous increases in peak flows and temperature, CPOM standing stocks tended to decline more slowly during winter, in comparison with unharvested conditions (Fig. 5.1). For all other heuristic scenarios involving changes in multiple drivers, the accumulation of CPOM standing stocks in late summer was more delayed, and their depletion in winter was faster than in unharvested conditions (Fig. 5.1). For heuristic scenarios involving perturbations of multiple model drivers, the effect size of disturbance was significantly negative (indicating significantly lower CPOM standing stocks than in unharvested conditions) whenever litterfall reductions reached 50% or above (i.e., high severity; Fig. 5.2). When litterfall reductions were 30% or below, the effect size of disturbance varied with the relative changes in peak flows and stream temperature (Fig. 5.2). For instance, a 4 or 6°C-increase in stream temperature could reverse decreases in CPOM standing stocks caused by litterfall reductions and peak flow increases (Fig. 5.2).  5.3.3 Interactive effects of model drivers on CPOM Only the effects of litterfall-temperature interactions on CPOM standing stocks were significant (F4,1116 = 4.79; P < 0.001), whereas the effects of all other two-way and three-way interactions were insignificant overall (Table 5.3). The cumulative effects of paired model drivers were additive at low disturbance severity (Table 5.2). From moderate to very high levels of disturbance severity, two-model driver interactions were significantly non-additive, and either positive or negative antagonistic (Wilcoxon signed-rank tests: P < 0.05), except for the effects of litterfall-temperature interactions which were additive (P = 0.56; Table 5.2). The considerations involved in determining the directional interaction types for these interactions are given in the following two representative heuristic scenarios under high severity of disturbance (shown in Table 5.2). Firstly, for litterfall-peak flows interactions, the individual effects of changing litterfall and peak flows on CPOM standing stocks were both negative, and their sum of effects expected from the additive null model would lead to an average reduction of 4.61 g AFDM m-2 relative to the unharvested conditions. Their cumulative effects, on average, resulted in a reduction of 4.46 g AFDM m-2 relative to the unharvested conditions (Table 5.2), which were less negative than predicted from the additive null model, and thus their interactions were classified as negative 87  antagonistic. Secondly, for litterfall-temperature interactions, the individual effects of changing litterfall and stream temperature had opposing directions (which was also the case for peak flows-temperature interactions). The cumulative effects of litterfall and temperature (-3.18 g AFDM m-2 relative to the unharvested conditions) were more negative than the expected sum of effects (-2.14 g AFDM m-2), and thus determined as positive antagonistic. For their cumulative effects to be classified as synergistic, they would have to increase CPOM standing stocks by more than 2.27 g AFDM m-2 on average (i.e., exceeding the individual effects of temperature; classified as positive synergistic), or reduce them by more than 4.41 g AFDM m-2 (i.e., exceeding the individual effects of litterfall; classified as negative synergistic).  5.4 Discussion My model-based analysis revealed that litterfall rate was more influential than stream temperature (during late spring to summer) and peak flows as biophysical controls on the standing stocks of leaf litter-derived CPOM in a small forested stream, across a realistic range of severity of forest harvesting disturbances. Post-harvest litterfall reductions were simulated to cause much greater decreases in CPOM standing stocks than did elevated peak flows. Stream warming enhanced CPOM standing stocks, and could counteract the negative effects of litterfall and peak flows at lower severity levels of forest harvesting disturbance. In particular, when stream summer temperature at East Creek was increased by 4°C or more, it approached or exceeded the optimum temperature for leaf litter consumption by shredders (Topt). This resulted in a greater extent of CPOM gains through reduced shredder biomass and consumption than losses through increased microbial processing (see below under section 5.4.2). In the heuristic modelling of forest harvesting scenarios, the interplay of model drivers yielded diverse CPOM responses, except at high or very high severity of litterfall reductions under which CPOM reductions were more probable. My modelling results showed that non-additive, antagonistic interactions between paired model drivers could emerge at higher levels of disturbance severity, suggesting the need for forest management activities to limit changes in drivers of concern to minimise harvesting impacts on CPOM standing stocks.  88  5.4.1 Post-harvest response ranges of model drivers The published post-harvest response ranges of model drivers in North American temperate streams reflected the effects of diverse forest harvesting practices and their spatial extent in individual watersheds, as well as the environmental context of study sites (e.g., elevation, bankfull width, aspect). The ranges of post-harvest changes in litterfall and stream summer temperature generally encompass those reported for small streams in the region of MKRF (litterfall: -2 to -91%, temperature: 0.4 to 3.9°C; Appendix C: Table C.4). Responses of peak flows in a MKRF stream to forest clear-cutting were the lowest of the published range and negative (-22%; Cheng et al. 1975), although variable extent of post-harvest increases (up to 233%) in peak flows have been observed in the Pacific Northwest. It is possible that small, forested streams similar to East Creek would not experience peak flow increases corresponding to the higher levels of disturbance severity that I specified, even when the entire watershed is clear-cut, as mediated by soil water drainage, evapotranspiration rates, and antecedent flow conditions (see Storck et al. 1998; Lewis et al. 2001; Abdelnour et al. 2011). It follows that some heuristic scenarios involving perturbations of peak flows at higher levels of disturbance severity might not occur in these streams. Nevertheless, I expected to have captured the realistic ranges of post-harvest responses of model drivers for the purpose of the sensitivity analysis and heuristic modelling, and the published responses I compiled could be a useful reference for additional modelling studies that simulate CPOM (or OM) responses to disturbances in other small streams.  5.4.2 Effects of individual drivers on CPOM Results of the sensitivity analysis indicated that litterfall was a more influential model driver of CPOM standing stocks at East Creek than stream temperature and peak flows, under variable impacts of forest harvesting. The extent of modelled reduction of riparian litterfall – as the source of leaf litter-derived CPOM – approximated the extent of depletion of CPOM standing stocks. This occurred along the severity gradient of forest harvesting disturbance, not only with near-complete exclusion of litter inputs (see Eggert et al. 2012). It is probable that when litterfall reductions exceeded 50% (i.e., similar to the “open-canopy stream” conditions in Stenroth et al. 2014), shredders were subject to intense resource limitation which further supressed their biomass and CPOM consumption (e.g., Baer et al. 2001; Melody and Richardson 2004; Zhang 89  and Richardson 2011). Litterfall reductions during early spring could also advance the onset of the period of extremely low availability of CPOM, potentially limiting the growth and productivity of many summer-growing shredder taxa (Richardson 2001). Logging-associated increases in peak flows could enhance the re-entrainment of litter, particularly during winter months with higher precipitation and discharge (see Appendix C: Fig. C.1; Kiffney and Richardson 2010). Modelled effects of increasing peak flows on CPOM standing stocks in East Creek were measurable but weak, and were comparable to the effects of the interannual variations in discharge regime. The discharge threshold beyond which bed sediments are mobilised (Qcritical) was considered to approximate 250 L s-1, based on empirical measurements in an experimental stream channel near East Creek (T.M. Hoover, unpublished data). Depending on antecedent flow conditions, when Qcritical is crossed, the probability of litter re-entrainment would go up from ~5-26% (T.M. Hoover, unpublished data) to 95% (Stenroth et al. 2014). Under unharvested conditions, Qcritical is crossed 1-4 times each year, normally from November to January, and there were numerous high-flow episodes with daily discharge reaching 160-250 L s-1 during this period (Appendix C: Fig. C.1). Additional events of Qcritical exceedances due to peak flow increases tended to occur at moderate to very high disturbance severity, which led to further CPOM reductions compared with unharvested conditions. Such further CPOM reductions usually lasted for several days to about one month during winter, compared to the effects of reduced litterfall, which occurred all year round. It is important to recognise that the effects of elevated peak flows on CPOM standing stocks could be greater in certain years (or streams) than modelled in this study, if discharge levels are already more frequently near Qcritical in unharvested conditions (e.g., Papangelakis and Hassan 2016).  Logging-associated summer warming effects on CPOM standing stocks were more apparent at higher levels of disturbance severity, and depended on the thermal regime of the study stream, and trade-offs between changes in shredder feeding and microbial processing of CPOM at higher temperatures. Under unharvested conditions, the average smoothed stream temperature from June to September at East Creek was about 11°C, and below Topt (15°C). At higher levels of disturbance severity (e.g., 4-6°C temperature increases), Topt were exceeded for longer periods of time (Fig. 5.3). This caused greater declines in shredder consumption and biomass, which lasted till fall (Appendix C: Fig. C.2; see also Stenroth et al. 2014). Modelled 90  CPOM gains due to lower shredder consumption exceeded CPOM losses through accelerated microbial processing under warming conditions. This is due to a much lower modelled contribution of microbial processing (~1%) to CPOM losses than that of shredder consumption (~20-40%) in this stream. However, such positive responses of CPOM may not occur in streams where microbial processing contributes to CPOM losses similarly to or more than shredder consumption (see Buzby and Perry 2000; Yeung et al. 2018). CPOM responses to temperature increases are evidently also contingent on Topt used for parameterising the model. Laboratory feeding trials established Topt to approximate 15°C for many temperate shredder taxa, particularly caddisflies (e.g., González and Graça 2003; Rumbos et al. 2010; Batista et al. 2012; Correa-Araneda et al. 2015) which could vary with taxonomic identity and the quality of litter consumed. In cool streams in Nova Scotia, Canada, Leuctra stoneflies – a numerically dominant shredder (other leuctrid species are also common in East Creek) – were shown to attain the highest density (Taylor and Andrushchenko 2014) and probably highest total consumption at 14°C. Indeed, in the study region, the temperature-dependence for litter consumption is still unknown for the numerically dominant and large-bodied taxa, such as Lepidostoma caddisflies and Pteronarcys stoneflies (see Ruesink and Srivastava 2001; Lecerf and Richardson 2011; Yeung et al. 2018). It is possible that Topt is taxon-specific, and below or above 15°C. In the case of the latter, and in higher-elevation streams, stream temperature under forest harvesting disturbances may still be much below Topt. This would lead to a lesser extent of net gains of CPOM than what I modelled, and would render temperature a less influential model driver in forest harvesting scenarios. There are other sources of uncertainty associated with warming effects on shredder consumption, for instance, through compositional shifts in litter-associated shredder and microbial communities (Fernandes et al. 2012; Moghadam and Zimmer 2014; Domingos et al. 2015), which were not incorporated into the model. Despite this uncertainty, my modelling results suggested a potential mechanism by which stream warming could enhance CPOM standing stocks via suppression of shredder consumption, when summer stream temperature is already close to, or above Topt. There is a clear need to empirically evaluate the validity of this mechanism (e.g., using mesocosms) for individual shredder taxa and entire communities across a temperature gradient. 91  My modelled effects of litterfall and peak flow changes are generally consistent with those shown by modelling studies in other temperate streams. CPOM decreases occurred with simulated litterfall reductions (by 10%) in central Appalachian streams in Virginia, USA (Buzby and Perry 2000), and during periods of lower litter inputs in a northern Spain stream (Pozo et al. 1997). Peak flow increases (by 10%) had limited effects on CPOM standing stocks (Buzby and Perry 2000). Simulated warming (by 2°C) led to reduced CPOM, as the consequence of increased shredder consumption and microbial processing (Buzby and Perry 2000), which differed from the suggested mechanism under warming effects for East Creek. This difference is probably due to the modelling assumption of linear increases in shredder consumption rates with temperature, without considering Topt, in Buzby and Perry (2000). I therefore suggest to empirically determine the temperature-dependence of shredder consumption, in order to understand the influences of stream temperature relative to other model drivers on CPOM standing stocks in any given stream.   5.4.3 Effects of multiple drivers on CPOM The response pattern of stream CPOM to multiple model drivers highlights the complexity of biophysical controls on CPOM standing stocks, which gives rise to context-dependent effects of forest harvesting disturbance. Modifying three model drivers already yielded countervailing effects (i.e., increase, decrease, or no change) on and tremendous variability in CPOM standing stocks. In addition, interaction effects between pairs of model drivers could vary between additive or antagonistic, depending on the severity of forest harvesting disturbance. This illustrated the difficulty of predicting the cumulative effects of multiple drivers from knowledge of single driver effects. I showed that antagonistic interactions were the dominant type of non-additive interactions examined, which are also frequently observed in multiple stressor studies in freshwater ecosystems (Piggott et al. 2015b; Jackson et al. 2015). It is unclear how common and variable these interactions are in streams affected by forest harvesting, due to the lack of field-based correlative and manipulative studies of multiple drivers. These interactions arose due to the emergent properties of the biophysical processes modelled, and not relevant factors excluded from the CPOM model, such as litter quality (Fernandes et al. 2012; Foucreau et al. 2016), the 92  activity and community composition of shredders, and their interspecific interactions (Tiegs et al. 2008; Ferreira et al. 2014). Taking the scenario with high severity of litterfall reductions and peak flow increases (especially in fall) as an example, a much-reduced quantity of CPOM by one driver would allow a smaller magnitude (in absolute amount) of further decrease by another driver, since reductions were made on a proportional basis. Hence, cumulative CPOM reduction by the two drivers combined would be less than the additive sum of the negative effects of these drivers acting in isolation, thus giving rise to antagonistic interactions (see also interactions between warming and storm events on CPOM in Buzby and Perry 2000).  5.4.4 Model limitations My modelling results are more applicable to stream reaches similar to East Creek where in-stream wood is spatially patchy, which has minor contributions to leaf litter retention. In forested streams where large wood is a major type of CPOM retentive structure, their presence would likely further reduce the modelled effects of elevated peak flows on CPOM standing stocks (Ehrman and Lamberti 1992; Small et al. 2008). Nevertheless, I expect the differences between water temperature and Topt, and number of Qcritical exceedances to be commonly important drivers of shredder and discharge influences on CPOM standing stocks, respectively, across many streams systems. Therefore, site-specific testing and recalibrations (if necessary) of assumed model parameters such as Topt and Qcritical will ensure more accurate CPOM estimates. I assumed constant alterations in the values of model drivers throughout the duration of model runs. Forest harvesting is known to also alter the temporal variability and/or seasonality of litterfall (Kreutzweiser et al. 2004; Kiffney and Richardson 2010), discharge (Moore and Wondzell 2005), and stream temperature (Moore et al. 2005a; Gomi et al. 2006) within the first few years since harvest, which is also highly variable across sites. In particular, I did not model the effects of post-harvest increases in the frequency of large floods on CPOM reduction, which would be considerably greater than those of peak flow increases (see Buzby and Perry 2000). Incorporating the effects of known changes in these components of model drivers into model simulations is important for improving the accuracy of predicting site-specific responses (and recovery) of CPOM standing stocks to forest harvesting, but it is beyond of the scope of the present study. 93  I also did not quantify the uncertainty of model outcomes related to the influences of leaf litter types (broad leaves vs. conifer needles; Hoover et al. 2010; Marcarelli et al. 2011), and other co-occurring CPOM-related processes, without making the model too analytically complex. For instance, post-harvest stream narrowing, streambed reduction, and loss of large wood can lead to reduced channel roughness (Sweeney et al. 2004), which in turn enhances the probability of CPOM re-entrainment during high-flow events. Forest harvesting-induced changes in stream nutrient concentrations, sediment loads (Feller and Kimmins 1984; Sweeney and Newbold 2014), and light availability (Kiffney et al. 2004; Kaylor et al. 2016) can influence shredder consumption and microbial processing (e.g., Gulis and Suberkropp 2003; Cross et al. 2005; Lagrue et al. 2011; Danger et al. 2012; Connolly and Pearson 2013). These processes can interact with the model drivers (e.g., temperature) to affect CPOM responses (e.g., Ferreira and Chauvet 2011; Piggott et al. 2012; Fernandes et al. 2014).  5.4.5 Implications for forest and watershed management The main purpose of this study was to identify relatively important biophysical controls on CPOM flows and transport in small, forested streams in coastal British Columbia, Canada. My modelling results are relevant not only to forest and watershed management associated with forestry activities, but also with other anthropogenic impacts (e.g., agriculture, urbanization) and climate change in this region that can alter organic matter inputs, discharge, and stream temperature. Establishing riparian buffer zones can most likely mitigate post-harvest changes in CPOM standing stocks, through maintaining litter inputs and stream summer temperature (see also Karlsson et al. 2005; Kiffney and Richardson 2010). In doing so multiple attributes related to water quality, stream-riparian habitats, and biota therein can also be protected (Dosskey et al. 2010; Sweeney and Newbold 2014). Under selective harvesting, limiting basal area harvested (e.g., to below 50%), and retaining larger trees within a few meters of the stream channel may also minimise impacts on litter inputs in some streams (Kreutzweiser et al. 2004; Muto et al. 2009). In contrast, reach- and catchment-scale management measures that reduce the extent of peak flow increases are expected to have minimal effects on enhancing stream CPOM retention on an annual basis. Nevertheless, limiting peak flow increases (and debris flows) may be effective for achieving other management objectives, such as minimising post-harvest changes in 94  stream geomorphology, sediment loads, and biotic integrity (Moore and Wondzell 2005; Reid et al. 2010; Hawley et al. 2016). For instance, this can be attained by reducing proportion of watershed area harvested to below 40%, and positioning harvested areas and logging roads further away from streams (Rogger et al. 2017). My exploratory modelling reveals antagonistic interactions between reach- to catchment-scale biophysical processes on an ecological variable at higher disturbance severity, which adds to the limited work showing non-additive cumulative impacts of environmental stressors in small streams (see Buzby and Perry 2000; Harvey and Railsback 2011). These results reiterate the need for management actions to limit changes in multiple stressors, particularly litter inputs and stream temperature (e.g., setting up riparian buffers), to minimise disturbance effects on stream food webs (Jackson et al. 2015). It is normally challenging to quantify the impacts of biophysical processes operating at reach to catchment scales (e.g., discharge, substrate composition) on the structural and functional attributes of streams, in the presence of multiple controlling processes and interactions between them. As illustrated in this study and others, process-based model simulations and scenario analysis can provide a mechanistic approach to better understand the impacts and drivers of these large-scale processes (e.g., Futter et al. 2007; Katsuyama et al. 2009; Harvey and Railsback 2011; Bussi et al. 2018; Groom et al. 2018). I recommend wider adoption of this approach to forecast stream ecological responses to forest harvesting and other watershed disturbances, which can help prioritise management strategies to meet multiple ecological targets under limited budgets.95  Table 5.1 Changes in litterfall, peak flows, and stream temperature assigned to each level of severity of forest harvesting disturbance for the heuristic modelling of CPOM standing stocks in East Creek, British Columbia.  Severity Litterfall Peak flows Stream temperature (June to September) Low -10% 20% 1°C Moderate -30% 40% 2°C High -50% 100% 4°C Very high -90% 300% 6°C 96  Table 5.2 Difference in average daily CPOM standing stocks (g AFDM m-2) between the values simulated using the heuristic model of forest harvesting disturbance and the unharvested (control) scenario in East Creek, British Columbia. For each pair of model drivers, individual effects of drivers (litterfall, L; peak flows, P; stream temperature, T) and the cumulative effects of drivers were modelled. Positive (negative) values indicate that daily CPOM standing stocks in the given scenario are higher (lower) than in the unharvested scenario. For each heuristic scenario with cumulative effects, the difference in daily CPOM standing stocks between values expected from the additive null model and the unharvested scenario are given in parentheses. Also shown are the directional interaction types of two-model driver interactions (+A: positive antagonistic, -A: negative antagonistic; sensu Piggott et al. (2015)), and the results of Wilcoxon’s signed-rank tests indicating whether these directional interaction types were significantly non-additive. P values < 0.05 are in bold font. Refer to Table 5.1 for changes in model drivers associated with each category of disturbance severity.  Model drivers Severity of disturbance Low Moderate High Very high Individual effects                 L -1.06 -2.92 -4.41 -7.06 P -0.05 -0.10 -0.20 -0.37 T 0.09 0.46 2.27 5.64          Cumulative effects  Interaction type  Interaction type  Interaction type  Interaction type L+P -1.11 (-1.11) -A; P = 0.08 -2.97 (-3.02) -A; P < 0.01 -4.46 (-4.61) -A; P < 0.01 -7.09 (-7.43) -A; P < 0.01 L+T -0.96 (-0.97) -A; P = 0.43 -2.49 (-2.46) +A; p = 0.56 -3.18 (-2.14) +A; P < 0.01 -6.27 (-1.42) +A; P < 0.01 P+T 0.03 (0.03) +A; P = 0.06 0.35 (0.36) +A; p = 0.02 2.01 (2.07) +A; P = 0.02 1.84 (5.27) +A; P < 0.01 Note: Simulated average daily CPOM standing stocks under unharvested conditions are 7.82 g AFDM m-2. 97  Table 5.3 Results of three-way ANOVA testing the modelled main effects and interactions of litterfall (L), peak flows (P), and stream temperature (T) affected by forest harvesting on ln(hydrograph-specific response ratio of CPOM) in East Creek, British Columbia. P-values < 0.05 are in bold typeface.   Degrees of freedom Mean square F P L 4 220.19 31450.61 < 0.001 P 4 0.09 12.62 < 0.001 T 4 13.82 1974.07 < 0.001 L × P 16 0.00 0.09 1.000 L × T 16 0.03 4.79 < 0.001 P × T 16 0.00 0.03 1.000 L × P × T 63 0.00 0.002 1.000 Error  1116 0.01     98    Figure 5.1 Simulated daily CPOM standing stocks (g ash-free dry mass [AFDM] m-2) in East Creek, British Columbia, from model runs under unharvested (control) conditions and with logging-associated changes of high severity in (a) single model drivers, including litterfall (L), peak flows (P), and stream temperature (T), and (b) multiple model drivers (L+P, P+T, L+T, L+P+T). Model outputs of 730 time steps (i.e., across two years) averaged across an ensemble of ten simulations are shown, and results from the first 90 days of each model run (spin-up period) are excluded to allow for the model to stabilise. Refer to Table 5.1 for changes in individual model drivers for each scenario.99    Figure 5.2 Effect size of logging-associated changes in litterfall, peak flows (P), and stream temperature (T) on average daily CPOM standing stocks (ln[response ratio of CPOM]) in East Creek, British Columbia, simulated by full-factorial heuristic modelling of forest harvesting impacts. Effect size for each heuristic scenario averaged across ten model simulations, and its associated 95% confidence interval (represented by error ranges which are in most cases very small), are shown. Effect sizes with corresponding error ranges overlapping zero (the solid line) are not significant. Panels are arranged by increasing severity of logging-associated changes in litterfall: (a) no changes (i.e., control), (b) low, (c) moderate, (d) high, and (e) very high severity. Note that the scale of y-axis in (e) is different from other panels.100    Figure 5.3 Relationships between the temperature-dependent function for leaf litter consumption by shredders and stream temperature. These relationships are part of the CPOM model by Stenroth et al. (2014), and were established according to consumption model 2 for warm-water species (Kitchell et al. 1977) in the Fish Bioenergetics 4.0 model (Deslauriers et al. 2017). Shaded area denotes the range of smoothed daily stream temperature from June 1st to September 30th at East Creek, British Columbia. The solid line denotes the average daily stream temperature (11.3°C) of this period, which is below the optimum temperature (15°C) for shredder consumption used to parameterise the stream CPOM model. 101  Chapter 6: Assessing the long-term ecological effects of riparian management practices on headwater streams in a coastal temperate rainforest  6.1 Introduction A range of management practices have been designed to address and mitigate the impacts of forest harvesting on watersheds, which operate on various spatial scales. These practices typically include the establishment of riparian reserve strips of fixed or variable widths and configurations (reach scale); and harvesting of various extent and species (e.g., partial harvesting, progressive harvesting, selective and shelterwood logging), re-routing of stream crossings and road networks, etc. (catchment scale). Many of these practices are now formulated as forestry best management practices (BMPs), and incorporated in forestry guidelines and regulations worldwide (e.g., Lee et al. 2004; McDermott et al. 2010; Cristan et al. 2016), as well as forest certification programs (Ice et al. 2010). Contemporary forestry BMPs and other riparian protection measures are known to differ in the effectiveness of maintaining water quality, aquatic and riparian biodiversity, ecosystem functions and services (Cristan et al. 2016). Such effectiveness tends to increase with the width of riparian reserves, and fixed-width reserves of 30 m have been typically set to balance stream-riparian protection and timber production. Nevertheless, forest harvesting even ≥ 30 m beyond streamsides is sometimes shown to cause apparent abiotic (e.g., temperature and nitrate removal efficiency; see Sweeney and Newbold 2014) and biotic changes in stream and riparian systems (e.g., Pearson and Manuwal 2001; Kreutzweiser et al. 2008a; Lecerf and Richardson 2010; Marczak et al. 2010). Partial harvesting of riparian vegetation (or thinning; usually affecting ≤ 50% of the basal area) could elevate stream temperature (Macdonald et al. 2003; Guenther et al. 2014) and nutrient levels shortly after logging (Wang et al. 2006). The effects of thinning upland and riparian vegetation, or thinning upland forests while retaining riparian buffers, on stream habitats and communities may be evident (e.g., Martel et al. 2007; Moseley et al. 2008; Kara et al. 2014; Burton et al. 2016) or undetectable (e.g., Wilkerson et al. 2006; Olson and Rugger 102  2007; Kreutzweiser et al. 2009b; Chizinski et al. 2010), depending on the extent and methods of harvesting. The effectiveness of riparian protection measures, such as fixed-width buffers and partial harvesting, has typically been evaluated based on short-term, post-harvest water quality data (e.g., within three years after harvesting; see Cristan et al. 2016). Post-harvest ecological recovery of streams, including biotic communities and their functions, has been much less studied, which can require longer time frames (but see Jackson et al. 2007; Gravelle et al. 2009). Many studies of ecological recovery adopted various sampling designs to minimise site-specific differences in ecological processes post-harvest, by studying streams affected by a gradient of logging history (Herlihy et al. 2005; Martel et al. 2007), multiple paired watersheds and three or more sites in treatment/control groups in replicated designs (e.g., Frady et al. 2007; Hemstad et al. 2008; Medhurst et al. 2010). Some studies detected signs of recovery for macroinvertebrate or shredder communities >10 years after harvesting (Davies et al. 2005; Frady et al. 2007; Medhurst et al. 2010), but did not address long-term changes in ecosystem processes, such as litter decomposition. Measures of stream ecosystem processes are important to assess ecological integrity (Gessner and Chauvet 2002). As forested headwater streams receive most of their energy inputs from allochthonous detritus, decomposition rates of leaf litter have been used to indicate forestry impacts on important ecosystem processes, with post-harvest increases (Griffith and Perry 1991; Benfield et al. 2001; McKie and Malmqvist 2009) and decreases (Kreutzweiser et al. 2008a; Lecerf and Richardson 2010) being documented. The substantial variability of the post-harvest responses of litter decomposition rate (in terms of direction and magnitude) could be attributed to logging-induced alterations in stream habitats (Webster and Waide 1982), and compositional shifts of riparian forest vegetation (Kominoski et al. 2011) (see also Richardson and Béraud 2014). Several lines of evidence indicate that these changes affected the abundance and diversity of invertebrate shredders, as a key agent of organic matter breakdown in aquatic ecosystems (e.g., Griffith and Perry 1991; McKie and Malmqvist 2009; Lecerf and Richardson 2010). As part of the riparian management study at the University of British Columbia’s Malcolm Knapp Research Forest (MKRF), near Vancouver, Canada, experimental harvests were executed across watersheds, with replicate streams for each harvesting practice (see Methods for 103  detailed descriptions). The recovery of stream temperature and water chemistry in the logged sites to reference levels was detected within 5 years post-harvest (Gomi et al. 2006; Feller 2010a). In a study conducted in 2006, Lecerf and Richardson (2010) showed lower litter decomposition rates in streams 8 years after clear-cut logging with or without riparian reserves, and 2 years after thinning (i.e., logging 50% of the basal area of riparian trees) than in reference ones. The reduced decomposition rates in clear-cut reaches with or without riparian reserves were associated with lower densities of shredder invertebrates. I conducted this study in 2013 (i.e., 15 years after logging and 9 years after thinning) by re-measuring litter decomposition rate, shredder density and richness, and water quality characteristics in the same study sites. This study thus investigated the post-logging recovery trajectories of headwater streams in the presence of experimental riparian management practices. After an additional 7-year period of forest regeneration, I predicted that the measured biological parameters in some of the logged reaches, especially those with wider riparian reserves, would become close to, or no longer be significantly different from, those in reference reaches. Such convergence would indicate signs of ecological recovery from forest harvests, and would likely be associated with the altered physical environment of streams approximating reference conditions. I expected that several hydro-climatic characteristics, including temperature, precipitation, and stream discharge, could potentially influence the responses of the measured biological parameters, in addition to the legacy effects of forest harvesting. For instance, stream discharge and thermal regimes in the study period and the preceding summer could affect shredder recruitment and survival, and hence litter decomposition, which need to be considered when comparing them across forest treatments.  6.2 Methods 6.2.1 Study sites The set of study sites were the same as used in Lecerf and Richardson (2010). The study area is located within the Coast Mountains of the Pacific Northwest and about 45 km east of Vancouver, British Columbia. A total of 16 stream reaches (up to 5 m wide) in MKRF (49°18’40’’N, 122°32’40’’W) were chosen and randomly assigned (with one exception) to various riparian management practices, including (1) clear-cutting where no reserves were left; (2) upland clear-104  cut logging with 30-m continuous riparian reserves on both sides of the study reach; (3) upland logging with 10-m riparian reserves on both sides; (4) upland and riparian thinning (50% basal area removal). There were 3 or 4 replicate stream reaches in each riparian management practice (for details, see Kiffney et al., 2003 and Lecerf and Richardson, 2010). Experimental streamside logging with manipulations of fixed-width riparian reserves (0, 10-m and 30-m wide) took place in 1998, whereas thinning occurred in 2004. Thinning involved uniform tree removals spanning the stream channel within a harvest unit, with the goal of removing a similar basal area of trees on a catchment scale as in earlier upland logging. Prior to the experimental harvests, these reaches and the unharvested, reference reaches were located in second-growth forests which had not been logged since 1931. Early successional riparian forests in the clear-cut sites was primarily dominated by rapidly-growing, nitrogen-fixing red alder, Alnus rubra Bong. (Betulaceae) in 2006 (Lecerf and Richardson 2010). At the time of this study, detrital inputs of red alder (and other deciduous broadleaves) likely continued to dominate riparian litterfall to streams (Bilby and Bisson 1992; Hoover et al. 2011), and enhance the productivity of lower trophic levels in clear-cut sites (Kominoski et al. 2012). Some red alder trees were also present in the riparian vegetation in sites with riparian reserves and thinned sites, and much fewer in coniferous-dominated references sites. Canopy openness in each site was determined from digital hemispherical pictures taken vertically upward about 1 metre from the streambed in autumn 2013 (Nikon Coolpix 4500 with fisheye lens, Nikon USA Inc., Melville, NY, USA). Pictures were processed using the software package Gap Light Analyzer version 2.0 (Simon Fraser University, Burnaby, BC, Canada) to determine percent gap area within a 0-30° zenith range (Lecerf et al. 2016).  6.2.2 Water temperature and chemistry Water temperature was recorded every 15 min throughout the deployment period of leaf bags in streams using temperature loggers (HOBO TidbiT v2 water temperature data loggers, Onset Computer Application). Temperature data were averaged for each day, in order to generate average daily water temperature for each stream during the deployment period. Measurements of pH and conductivity were conducted in situ at bag introduction and retrieval dates in all study sites. Water-quality parameters, including total dissolved nitrogen (TDN) and soluble reactive 105  phosphorous (SRP), were analysed in the laboratory at the Great Lakes Forestry Centre at Sault Ste. Marie, Ontario, Canada, following standard methods (see procedures in Nicolson 1988; Hazlett et al. 2008).  6.2.3 Leaf bags Senescent leaves of red alder were collected in autumn 2013, and then air-dried for several weeks. Nine leaf bags (made of 10-mm mesh size plastic netting), each containing 5 g of alder leaf litter (± 0.025 g), were deployed for each study site. Three sets of 3 bags were tied to iron bars, which were anchored into the stream bed in riffles of each of the study sites in early November 2013. A total of five additional leaf bags were submerged, and then immediately retrieved and transported back to the laboratory. The handling losses of leaf materials were estimated (mean: 7%; SE: 0.4%). One set of 3 leaf bags were retrieved at each of 3 sampling occasions (mid-November, mid-December 2013, and mid-January 2014) from each stream, which were of similar timing as in Lecerf and Richardson (2010). In the field they were stored on ice, and returned to the laboratory for determination of remaining leaf mass. Retrieved litter was carefully removed from bags, and rinsed over sieves (500-μm mesh size) to remove invertebrates and sediments from leaf surfaces. Leaf materials were oven-dried at 60°C for 72 h and weighed to the nearest 0.01 g to obtain final leaf dry mass. Leaf tissue was then ashed at 550°C for 4 h in a muffle furnace to determine ash mass, and hence ash-free dry mass (AFDM), for subsequent calculations of litter decomposition rates.  6.2.4 Shredders Benthic macroinvertebrate shredders were sampled in early November 2013, prior to the introduction of leaf bags in streams. Five 900-cm2 Surber (mesh size: 500 μm) samples were taken from riffle habitats, immediately stored in 70% ethanol within plastic containers and transported to the laboratory. The mesh size of the Surber sampler was larger than the one used in the previous study (i.e., 250 μm; Lecerf and Richardson 2010), and a small fraction of smaller shredders thus might not have been captured in the sampling process. Benthic samples from each site were pooled together as in Lecerf and Richardson (2010). Shredders were counted and 106  identified to the lowest practical taxonomic resolution to determine density and taxonomic richness. Shredder richness needs to undergo individual-based rarefaction to properly correct for the number of shredders collected (see Gotelli and Colwell 2001). Since the total number of shredders varied between sites and scaled positively with the number of taxa in both studies, taxonomic richness was uniformly rarefied to 11 individuals at each site. Sites with total abundance lower than 11 were excluded from the rarefaction (2 sites in Lecerf and Richardson (2010) and 3 sites in this study). Rarefied richness determined was more than 50% of total, unrarefied richness for all sites in both studies, hence inter-site differences in rarefied richness were unlikely to be underestimated owing to the convergence of rarefaction curves that might occur at low abundances (Gotelli and Colwell 2001).  6.2.5 Statistical analyses Final leaf AFDM was adjusted by subtracting by the weight of leaf materials from handling losses. Daily and temperature-corrected litter decomposition rates were determined by obtaining the slope of the regression of the proportion of remaining leaf AFDM against time (in days) and degree days (sum of positive daily mean temperature), respectively. The ratio of decomposition rate coefficients (daily, non-temperature-corrected k values) in streams affected by each riparian management practice versus reference ones was calculated. This measure could reflect changes in stream functional integrity within the assessment framework proposed by Gessner and Chauvet (2002). Differences in water characteristics (temperature, pH, and conductivity) across forest treatments were assessed by one-way multivariate analysis of variance (MANOVA), and those for biological parameters (litter decomposition rate, shredder density, and rarefied shredder richness) by separate ANOVAs. Two-way ANOVA was performed to detect differences in canopy openness across studies (i.e., years) and forest treatments. Planned comparisons further tested the hypothesis that logging impacts on biological parameters continued to be inversely related to riparian reserve width. These comparisons, using Helmert contrasts [reference vs mean (30-m reserves + 10-m reserves + no reserve), 30-m reserves vs mean (10-m reserves + no 107  reserve), and 10-m reserves vs no reserve], identified groups which did not differ significantly (P > 0.05). These groups were then tested against thinned sites in the second set of contrasts. Community dissimilarity of shredders between forest treatments for both studies was visualised using non-metric multidimensional scaling (NMDS) graphs, which showed centroids and 95% confidence ellipses associated with forest treatments. Shredders identified to be in the early instars stage (except the family Capniidae) in Lecerf and Richardson (2010) were excluded from the ordination analyses and richness calculations. Note these excluded taxa were still included in the calculations of shredder density. One-way permutational multivariate analysis of variance (PERMANOVA) was performed to assess whether the effects of forest treatments on shredder assemblages were different, using a Bray-Curtis similarity matrix. PERMANOVA is preferred as it is more robust to heterogeneity of variances (i.e., greater dispersion among groups) and unbalanced design than other widely used resemblance-based permutation methods such as analysis of similarity (ANOSIM; Anderson and Walsh 2013), which apply to both studies. Complementarily, indicator species analysis (IndVal) was conducted to determine, for each taxon, whether it was a significant indicator of a particular forest treatment for both studies. The ‘IndVal.g’ value given is a statistic that measures the degree of association between a taxon and a forest treatment, which ranges from 0 (no association) to 1 (complete association; Cáceres and Legendre 2009). Ln(x+1) transformations down-weighted the effects of abundant taxa on PERMANOVA and IndVal. The variability in assemblage structure (i.e., dispersion, or beta diversity) within each treatment group was measured by the average distance of site assemblages to the group centroid in multivariate space. Differences in dispersion among forest treatments for both studies were tested through the permutational analysis of multivariate dispersions (PERMDISP; Anderson 2006) using the ANOVA F statistic. Post-hoc pairwise comparisons of group mean dispersions were made using a permutation-based test. All MANOVA and ANOVAs were conducted in R (R Development Core Team 2018). All ordination analyses, including the construction of 95% confidence ellipses (ordiellipse function), PERMANOVA (adonis) and PERMDISP (betadisper), were carried out using the vegan package in R (Oksanen et al. 2017). Indicator species analysis was performed using the ‘multipatt’ function in the indicspecies package in R (Cáceres and Legendre 2009).  108  6.3 Results 6.3.1 Canopy openness and water characteristics Percent canopy openness varied significantly among forest treatments (ANOVA: F4,21 = 43.1; P < 0.001) in this study. Canopy openness was generally higher in reaches without riparian reserves than with 10-m and 30-m reserves and thinned reaches, and it was the lowest in reference reaches. After a 7-year period of forest regrowth subsequent to the study by Lecerf and Richardson (2010), canopy openness declined overall (ANOVA: F1,11 = 57.6; P < 0.001) and was below 20% for all sites. On average, in clear-cut sites it had been reduced by half to 17%, and in 10-m reserves, by one-third to 13% (Table 6.1). For other treatments, the extent of openness decline over the 7-year period was much less remarkable, which was thus associated with a significant interaction effect between sampling years (i.e., time since logging) and forest treatments (ANOVA: F4,11 = 15.0; P < 0.001). Variability of water characteristics among sites within a forest treatment was similar to, or greater than, their differences among treatments. These parameters were not significantly different overall across forest treatments (MANOVA: F12,33 = 1.9; P = 0.08), except for pH when analysed individually (ANOVA: F4,11 = 3.5; P = 0.045). Notably, pH in clear-cut sites (mean: 5.7) was lower than that in 30-m reserves (mean: 6.4; Tukey’s HSD: P = 0.04); however, all other pairwise comparisons showed no significant differences.  6.3.2 Litter decomposition Decomposition rates for alder leaves were less variable across study sites both on a daily basis (0.0034-0.011 day-1) and temperature-corrected basis (0.0006-0.0018 degree day-1; Table 6.1) than previously measured (0.0019-0.0118 day-1; 0.0003-0.0032 degree day-1). Decomposition rates among logged reaches were similar regardless of reserve width, and did not differ significantly from thinned and reference streams (Table 6.2). The reduction of differences across forest treatments appeared to be attributable to considerable increases in decomposition in logged and thinned reaches (daily: 76-173%; temperature-corrected: 61%-189%), and decreases in reference ones (25%; 46%), relative to those in the previous study (Fig. 6.1a and b). Stream integrity – as indicated by decomposition rate coefficients (Gessner and Chauvet 2002) – 109  improved across all forest treatments, which was no longer severely impaired and became either mildly or not impaired at all (Table 6.3).  6.3.3 Shredders In this study, mean benthic shredder density was about 36-90% lower (range: 40.7-200 individuals m-2), and with smaller variability across forest treatments than those in the previous investigation (Fig. 6.1c). Densities of shredders at reference and thinned reaches were no longer significantly higher than those at logged ones (P = 0.44; Table 6.2), although the mean value at thinned sites tended to be more than double that of the other treatments (200 vs 40.7-82.2 individuals m-2; Fig. 6.1C, see also Appendix D: Table D.2).  When reanalysing data on shredder richness from Lecerf and Richardson (2010), rarefied richness actually did not vary significantly across forest treatments (ANOVA: F4,9 = 1.18; P = 0.38), and was incongruent with the results of unrarefied richness presented. Thus, only rarefied richness was analysed and compared among studies (see Fig. 6.1d and Appendix D: Fig. D.1). In this study, rarefied richness also tended to be lower than in the previous study, with mean rarefied richness ranging from 2.1-3.4 taxa across treatments (Fig. 6.1d). Reference reaches in 2013 had significantly higher richness on average than logged and thinned sites, as indicated by the contrast analysis (P = 0.01; Table 6.2). Among logged sites, 30-m reserves supported the lowest richness compared to sites with 10-m and no reserves (P = 0.01). As for rarefied richness, the absence of Yoraperla stoneflies in most 30-m reserves and thinned sites (Appendix D: Table D.1), but not reference ones, could partly explain the lower richness in these logging-impacted sites in 2013. The NMDS ordination plots represented data on shredder assemblage structure well, given the fair stress values (stress = 0.06 and 0.12; Fig. 6.2a and b). Differences in forest treatments did not significantly affect assemblage structures in the study by Lecerf and Richardson (2010) (PERMANOVA F4,11: 1.04, R2 = 0.28, P = 0.39, 999 permutations) nor the present study (PERMANOVA F4,11: 1.14, R2 = 0.29, P = 0.32, 999 permutations). This was also revealed by the broad overlap of the 95% confidence ellipses for the centroids of forest treatments (Fig. 6.2a and b). In the present study, more shredder assemblages in sites affected by logging, both with and without riparian reserves, fell within the 95% confidence ellipse for the 110  centroids of assemblages in reference reaches (Fig. 6.2a and b). Therefore, shredder composition in some logging-impacted sites began to be more similar to that in reference ones. Indicator species analysis showed, for the shredder assemblage in 2006, that Moselia (IndVal.g = 0.91, P < 0.01) and Tipula (IndVal.g = 0.86, P = 0.03) were significantly associated with reference and thinned sites. Yoraperla was representative of both 30-m reserves, reference and thinned sites (IndVal. g = 0.92, P = 0.01). However, no indicator taxa were found for any forest treatment in 2013, suggesting that taxa present became more commonly found across forest treatments (see also Appendix D: Table D.1). Dispersion of assemblage structures differed significantly among treatments when reanalysing the data from Lecerf and Richardson (2010) (PERMDISP; ANOVA F4,11: 5.38, P = 0.01). Post-hoc comparisons revealed greater dispersion among clear-cut reaches than reference and thinned ones (P < 0.01; also see Fig. 6.2a). However, in the present study, significant differences in dispersion among treatments did not exist anymore (ANOVA F4,11: 0.39, P = 0.81), as assemblages within clear-cut sites became more homogeneous, and had comparable within-treatment variability with that in other treatments (P > 0.52; Fig.6.2b). This appeared to be partly associated with the occurrence of some taxa in more clear-cut reaches, such as Despaxia and Lepidostoma (see Appendix D: Table D.1), which led to the increase in similarity in shredder composition among sites. The dispersion of assemblages within sites influenced by thinning and in reference conditions was consistently low relative to other treatments across studies.   6.4 Discussion I revealed that important biological parameters in streams affected by different forest treatments, such as litter decomposition and density of macroinvertebrate shredders, began to converge with reference conditions during the additional 7-year period of forest regrowth since the last investigation (Lecerf and Richardson 2010). Successional riparian vegetation in logging-impacted streams experienced increases in canopy cover, although it was still less dense than that in reference sites. The denser streamside canopy mainly consisted of regenerating red alder stands and Rubus spectabilis Pursh (Rosaceae), a common shrub, particularly in no-reserve sites (Kiffney and Richardson 2010). Winter stream temperature and water chemistry remained 111  largely similar across forest treatments, except for TDN differences between thinned sites and the rest noted in Lecerf and Richardson (2010). This indicates that the recovery of abiotic variables, or water quality parameters, to reference levels was probably occurring in the previous study (i.e., within 8 years after logging and 2 years after thinning; see also Feller 2010a), which could precede that of litter decomposition rate and shredder density. This is also consistent with study findings elsewhere describing faster water quality recovery than ecological responses (e.g., Danehy et al. 2007; Kobayashi et al. 2010; Ishikawa et al. 2016). Plant cover and the biological uptake of essential elements normally increases rapidly in early forest succession following logging, thus revegetation would limit the export of sediments and nutrients (e.g., nitrates), which cease to affect water quality usually within several years after harvests (Webster et al. 2015).  6.4.1 Changes in shredder density, rarefied richness and assemblage structure I found that shredder densities and rarefied shredder richness were uniformly, and often substantially, lower across all streams in the present study, when compared with Lecerf and Richardson (2010). While the effects of forest treatments on shredder densities appeared to have diminished, differences in rarefied richness across treatments were detected in this study. The emerging differences in rarefied richness between forest treatments was more likely attributed to changes in shredder diversity than in the evenness of relative abundance distributions of shredders (the latter could also affect the shape of rarefaction curves; Gotelli and Graves 1996). This is because the evenness of the shredder community (represented by the Smith and Wilson evenness index Evar; Smith and Wilson 1996) was largely similar among treatments (range for 30-m reserves: 0.23-0.92; thinned sites: 0.28-0.90; clear-cut sites: 0.32-0.59; reference sites: 0.40-0.75; 10-m reserves: 0.91-0.92). The apparent reduction in shredder densities and rarefied richness across the two studies might be due to inter-annual climatic variability and/or the legacies of forest treatment that influenced hydrologic disturbances (e.g., droughts, floods) in the summer or earlier months (Negishi and Richardson 2006; Richardson 2008; Perry and Jones 2016). Differences in some catchment- and reach-level conditions (e.g., riparian litter inputs) might also affect the recruitment, survival, and distribution of shredder communities between the study years (Hemstad et al. 2008; Kreutzweiser et al. 2010; Kiffney and Richardson 2010). The 112  larger mesh size used in the present study (500 μm) might partly lead to fewer shredders being captured. However, the extent of this effect on density reduction was likely to be small (< 10-30%), even assuming that the early instars of leuctrid and nemourid stoneflies were all excluded (Lecerf and Richardson, unpublished data). The return time of shredder densities to reference conditions was indistinguishable between clear-cut logging with and without reserves given my sampling regime, which could range from 8-15 years after harvests. The initial proportion of watershed harvested, along with other landscape features, could have long-lasting effects on the relative differences in litterfall, and benthic organic matter quality and quantity between clear-cut and reference streams (Frady et al. 2007; Ely and Wallace 2010; Kiffney and Richardson 2010). Importantly, these differences could influence consumer-resource interactions, and result in the apparent variability of shredder recovery across forest types. For instance, the effects of riparian and upland clear-cutting on shredders have been shown to diminish faster (< 5 years post-harvest) in the sub-boreal interior region of British Columbia, Canada (Fuchs et al. 2003), or a near-decadal time-frame in coastal western Oregon (generally < 20 years; Herlihy et al. 2005) and also dry forests in central Washington, USA (mean: 16 years; Medhurst et al. 2010). The ranges of percentage of watershed area harvested in my study area (range for logged sites: 20-53%) were comparable with those in interior British Columbia (30-52%; Fuchs et al. 2003) and Oregon (0-42%; Herlihy et al. 2005). When large portions or whole watersheds were clear-cut, the impacts of forest harvesting on shredders were typically shown to be longer-lasting in other studies. Shredder density in clear-cut streams was still substantially (~ 200-400%) higher than in reference sites 16 years after whole-watershed clear-cutting of the forest in the Coweeta Hydrologic Laboratory in North Carolina, USA (Stone and Wallace 1998), but not 26 years post-harvest (Ely and Wallace 2010). About 80% higher shredder density was recorded in streams draining young-growth forests even 40 years after clear-cutting than unharvested ones in the H.J. Andrews Experimental Forest in Oregon (Frady et al. 2007). In temperate north-eastern Tasmania, Australia, taxon-specific responses of the abundance of facultative/obligate shredders 15 years after clear-cutting (50-90% of watershed area) were observed, with higher abundances of notonemourid stoneflies and leptophlebiid mayflies in clear-cut sites, but not for tipulid craneflies (Davies et al. 2005). 113  The legacy effects of riparian thinning on shredders or even macroinvertebrates appeared to be more short-lived than catchment-scale clear-cutting, as their numerical recovery in sites affected by thinning treatments (even in the presence of upland clear-cutting) could occur within 3-4 years post-harvest (Kreutzweiser et al. 2010), or sooner when only a single side of the stream was thinned (Chizinski et al. 2010). Thinning caused a relatively small reduction in canopy cover in these studies (< 10%; see also Kreutzweiser et al., 2009), as was similarly found by Lecerf and Richardson (2010). Therefore, thinning appeared to cause lower overall disturbance to riparian forests and hence the shredder communities than that of clear-cutting. Despite the statistical similarity of shredder assemblage structure across forest treatments, the assemblages in some logging-impacted streams were considerably different from those in reference sites in the previous study (see Fig. 3B in Lecerf and Richardson 2010). After a further 7 years of forest regrowth, shredder taxa present could be found in most forest treatments. Also, assemblage dispersion among clear-cut sites, relative to that of reference ones, was reduced. This was associated with the re-appearance of certain shredder taxa (e.g., the stonefly Despaxia augusta and caddisfly Lepidostoma spp.) in clear-cut sites that were also present in reference ones. The mutual consideration of assemblage dispersion and overlap demonstrated subtle site-level variability in the delay of shredder responses to clear-cutting.  6.4.2 Changes in litter decomposition In this study, I showed that some biological and physical parameters controlling decomposition rates in logged sites likely returned to within the range of variation of reference sites, although my sampling regime is not temporally intensive enough to confirm whether this had occurred earlier in reaches with wider riparian reserves. In thinned sites, faster decomposition was associated with lower TDN, relative to values in Lecerf and Richardson (2010), and both became comparable with reference conditions by 2013. The observed changes in decomposition rate did not appear to be related to any other water characteristics, because they continued to be similar among thinned and reference sites. The improvement of stream integrity across forest treatments observed in this study appeared to occur faster than in streams affected by catchment-scale clear-cutting in eastern United States, where decade-long post-harvest data are available. For instance, stream integrity 114  was reported to remain moderately (kt/kr = 1.42; Griffith and Perry 1991) to severely compromised (kt/kr > 2; Webster et al. 2014) even ≥20 years after harvests. The differences in recovery duration and trajectories of decomposition rates across forest types may arise from markedly variable responses of the underlying control processes of decomposition to riparian management histories. For instance, long-term increases in decomposition rate at clear-cut sites were associated with reduced benthic litter quantity and higher shredder density per unit area of litter (Griffith and Perry 1991; Webster et al. 2014a). Increased dominance of fast-processing leaves in riparian litterfall and/or reductions in total litterfall could lead to reduced availability of benthic organic matter in late fall and early winter. During the period of exacerbated resource limitation for shredders in detrital-based streams, litter in leaf packs could become an attractive food source and more rapidly consumed (e.g., Rowe and Richardson 2001; Benfield et al. 2001; Tiegs et al. 2008). In my study, litterfall in some of the logged sites was approximating, or even greater than, that in reference ones within 8 years after harvests (Kiffney and Richardson 2010), and I did not find higher shredder densities in logged and thinned reaches relative to reference ones, which unlikely intensified resource limitation. Elevated nutrient concentrations were identified as a cause of persistently higher decomposition rate in a stream 29 years after clear-cutting, within a montane deciduous forest in the Coweeta Hydrologic Laboratory (Webster et al. 2014a); however, nutrient concentrations did not differ significantly among my study streams. Taken together, the return of some underlying drivers of litter decomposition to reference conditions may explain the absence of significant differences in decomposition rate across forest treatments.  6.4.3 Implications for forest management The effectiveness of current forestry BMPs and other riparian protection measures has been predominantly evaluated on the basis of whether certain water quality standards are met in post-logging monitoring (see review by Cristan et al. 2016). The recovery of ecosystem structure (e.g., macroinvertebrate and fish communities) and functioning of streams may not necessarily be assumed even when there are no apparent signs of water quality impairment. Logging-induced changes in litter decomposition rates, the quantity and quality of riparian litter inputs, and discharge regime, could have long-term consequences on the local retention and downstream 115  transport of particulate and dissolved organic matter (Kreutzweiser et al. 2008b). Importantly, these changes have the potential to affect the delivery of aquatic ecosystem services throughout the catchment, such as fish and habitat productivity (Bilby and Bisson 1992; Wipfli et al. 2007). Therefore, I recommend the evaluation of harvest operations and forestry BMPs to look beyond water quality (see Karr 1993), and consider the temporal scales of recovery for hydrologic, physiochemical, and biological parameters from logging (see also reviews by Moore and Wondzell 2005; Moore et al. 2005a). Caution should be used when interpreting the recovery of shredder abundance and richness based on a few sampling times, in cases where the inter-annual variability of shredder recruitment and survival can mediate the legacy effects of logging. Commitment to long-term monitoring along the course of riparian and upland forest regeneration would be necessary to link forest regeneration to in-stream responses, and to verify the completeness of recovery to reference conditions, especially for catchment-scale clear-cutting. My study indicates that thinning of approximately 20% of the watershed area, including the riparian area, appeared to promote the rate of ecological recovery of streams, when compared to clear-cutting with and without fixed-width reserves in a coastal temperate rainforest in the Pacific Northwest. The sampling of two time points only characterised part of the post-logging recovery trajectory, and additional years of sampling are needed to confirm the recovery of the measured biological parameters from these forestry treatments. Thinning, or partial harvesting, near water bodies is often recommended in the emerging paradigms of emulating natural disturbance regimes (END; e.g., wildfires and windthrow) and enhancing future climatic resilience in forest management, by maintaining forest complexity and functioning (e.g., Sibley et al. 2012; O’Hara and Ramage 2013; Hood et al. 2016). Indeed, this practice has been used on a commercial scale (usually affecting ~ 50% of the watershed area) to accelerate forest regeneration and sustain timber yields in the temperate and boreal regions (Curtis and Carey 1996; Mielikäinen and Hynynen 2003). Future efforts should therefore address how closely thinning at such scale could mimic natural disturbances in terms of the ecological legacies on headwater streams, as have been done for terrestrial biota (e.g., Schieck and Song 2006; Zwolak 2009; O’Donnell et al. 2015).116  Table 6.1 Reach-scale characteristics for the sixteen stream sites in the study. Water temperature during the study period (early November 2013 to late January 2014) is given as daily mean values. Other water characteristics obtained in the middle of the study period are shown. Soluble reactive phosphorous (SRP) concentration was below detection limit (i.e., 10 µg P L−1) in all sites, and is therefore not shown in the table.  Site Forest treatment Canopy openness (% open to sky) Water temperature (°C) pH Conductivity (µS cm−1) TDN (µg N L−1) k (day-1) k (degree day-1) 2006 2013 East Reference 8.0 7.7 5.3 6.4 17 165.0 0.0086 0.0016 Mike Reference 8.3 11.0 6.1 5.9 14 649.0 0.0034 0.0006 Spring Reference 9.0 8.7 5.0 6.5 16 129.0 0.0079 0.0016 D 30m reserves 12.5 13.3 6.5 5.6 13 66.0 0.0056 0.0011 H 30m reserves 14.2 9.8 5.1 5.7 11 201.0 0.0052 0.0014 South 30m reserves 13.5 12.0 6.2 5.8 13 79.0 0.0062 0.0009 C 10m reserves 20.3 11.0 5.2 5.7 11 63.0 0.0086 0.0016 F 10m reserves 21.8 15.5 6.9 6.3 15 66.0 0.0065 0.0013 G 10m reserves 20.3 13.5 4.8 5.7 10 120.0 0.0060 0.0013 A No reserve 33.2 16.9 4.2 6.5 14 238.0 0.0048 0.0008 B No reserve 28.4 14.0 4.9 6.4 9 224.0 0.0054 0.0009 E No reserve - 17.4 6.1 6.2 11 < 50 0.0058 0.0008 I No reserve 36.8 18.9 3.3 6.5 10 217.0 0.0071 0.0018 J Thinning 18.5 13.2 5.7 5.8 11 299.0 0.0107 0.0015 K Thinning 19.9 16.0 4.7 6.5 16 127.0 0.0105 0.0017 L Thinning 14.2 12.8 4.4 5.9 10 73 0.0056 0.0010 117  Table 6.2 Summary of ANOVA and planned comparisons of the effects of forest treatments on biological parameters. Shredder densities were ln-transformed to normalise the data and to improve the homogeneity of variance. Note that three sites were omitted in the analyses of rarefied shredder richness. P values are shown in parentheses, and are in bold typeface when significant at 0.05.  Variable All treatments (P) Reference vs (30 m+10 m+no) (P) 30 m vs (10 m+no) (P) 10 m vs no (P) Thin vs reference (P) Thin vs (30 m+10 m+no) (P) k (day-1) F4,11 = 1.5 (0.27) t12 = -0.5 (0.66) t12 = 0.7 (0.52) t12 = 1.1 (0.32) - - k (degree day-1) F4,11 = 0.4 (0.78) – – – – – Ln(shredder density) F4,11 = 1.0 (0.44) t12 = -0.8 (0.44) t12 = -0.2 (0.86) t12 = -0.4 (0.70) – – Rarefied shredder richness F4,8 = 7.0 (0.01) t9 = -3.6 (0.01) t9 = 3.5 (0.01) t9 = < 0.1 (0.99) t11 = 3.2 (0.01) t11 = -0.1 (0.96)   118  Table 6.3 Changes in the effects of forest treatments on functional stream integrity between studies conducted in 2006 (i.e., data from Lecerf and Richardson (2010)) and in 2013 (i.e., present study). The assessment of functional stream integrity was based on comparing daily, non-temperature-corrected decomposition rate coefficients at treatment (kt) versus reference sites (kr). Scores (in parentheses and bolded) represent the assessed conditions of the treatment sites (0: severely impaired; 1: mildly impaired; 2: no evidence of impairment). For details of assessment criteria, refer to Gessner and Chauvet (2002).  Forest treatment kt/kr 2006 2013 No reserve 0.42 (0) 0.87 (2) 10-m reserves 0.45 (0) 1.06 (2) 30-m reserves 0.23 (0) 0.85 (2) Thinning 0.43 (0) 1.34 (1) 119    120  Figure 6.1 Comparison of the effects of forest treatments on (a) daily and (b) temperature-corrected decomposition of alder leaves, on (c) shredder density and (d) rarefied taxonomic richness between studies conducted in 2006 (i.e., data from Lecerf and Richardson (2010)) and in 2013 (i.e., present study). Letters indicate homogeneous groups derived from ANOVA and planned contrasts uniquely for each study (see Table 6.2). Error bars denote standard errors. Note that scales of y-axis differ across studies in (c).121    Figure 6.2 Non-metric multidimensional scaling ordination of shredder assemblages across sixteen streams sampled in (a) 2006 and (b) 2013. Dashed ellipses represent 95% confidence intervals for the centroids of the groups of forest treatments.             122  Chapter 7: Conclusion: synthesis and implications  Ecological stability broadly describes changes and variability of an ecological state variable. With intensifying and novel disturbances in the Anthropocene, there have been recent calls for operationalising this complex, multifaceted concept in environmental management (e.g., Hodgson et al. 2015; Donohue et al. 2016; Egli et al. 2018), which is foundational to my thesis. The main emphasis of my thesis was on developing a better understanding of the causal mechanisms underlying spatial and temporal differences of stability properties of stream particulate organic matter (OM) quantity and breakdown. The stability of particulate OM-related processes underpins the stability of consumer populations dependent on this basal resource, and hence trophic productivity in waterbodies throughout the watershed (Huxel and McCann 1998; Worm and Duffy 2003; Richardson et al. 2010). Responses of stream particulate OM dynamics to disturbances have key implications on the rates of delivery of important aquatic ecosystem services (e.g., fish production, nutrient cycling) to human populations.  In my thesis, I demonstrated a cohesive approach to devise empirical studies of the components of ecological stability, with an aim to more holistically represent the overall stability of an ecosystem (Chapter 2). Recognising forest harvesting as pulse disturbances that affect adjacent stream ecosystems, my thesis chapters were developed to address several stability properties of stream OM dynamics (depicted in Fig. 1.3 that shows a response trajectory to a pulse disturbance). The range and drivers of natural variability of stream litter breakdown that I determined (Chapters 3 and 4) helped identify the signs of post-harvest recovery of litter breakdown (Chapter 6). I also simultaneously modelled the natural variability and resistance of stream coarse particulate OM quantity to different forest harvesting scenarios (within ~4 years post-harvest; Chapter 5). Given the limited availability of pre-harvest data, I was only able to address single stability components in individual chapters, which is indeed typical in the majority of studies of ecological stability (Donohue et al. 2016). However, these stability components are not intended to be interpreted only individually, but collectively. Future efforts can build on my initial work to quantify additional stability components of stream OM dynamics, or adopt a similar approach to determine multiple stability components of ecological variables of interest (e.g., through process-based modelling; see Radchuk et al. 2019). These additional studies can 123  provide the basis of investigating the correlations among these components (i.e., dimensionality of stability; sensu Donohue et al. 2013), and mapping them bivariately (or trivariately; based on my recommendations in Chapter 2), in order to better characterise the overall stability of stream ecosystems. Here, I synthesise how the contributions of my thesis work advanced the field of disturbance ecology, and to better managing disturbance effects in stream-riparian and other ecosystems.  7.1 Standardising stability comparisons within and across ecosystems More generalised use of stability properties within and across ecosystems can be facilitated by adopting common frameworks that harmonise quantitative comparisons. I provided conceptual and practical recommendations (in Chapter 2) to adjust two existing bivariate frameworks (Hodgson et al. 2015; Ingrisch and Bahn 2018) that illustrate and compare stability properties to better address diverse disturbance-response trajectories, and shifting baselines. Empirical applications of bivariate or multidimensional frameworks of stability are rare (Donohue et al. 2016; Hillebrand et al. 2018). To the best of my knowledge, they have yet to be made for assessing the relations between stability properties of stream ecosystems. There is great potential for future efforts to use such frameworks to establish the relations between resistance and resilience (and other stability properties) of stream particulate OM dynamics along an intensity gradient of human disturbances in watersheds (e.g., percentages of tree cover, and impervious surface in watersheds; Young and Collier 2009; Voß et al. 2015). Data on the resistance and resilience of stream biotic communities to hydrological perturbations are also available for wider applications of these frameworks (e.g., Grimm and Fisher 1989; Peterson and Stevenson 1992; McMullen and Lytle 2012).  7.2 Evaluating mechanisms underpinning the stability of stream particulate OM dynamics The responses to forest harvesting of terrestrial-derived subsidy inputs and processing in streams have been studied to provide insights into the stability of functional attributes (e.g., stream 124  productivity), and mechanisms conferring stability (Webster et al. 1983; Golladay et al. 1992). The stability of streams is connected with that of riparian and upland forests, and upstream reaches, as mediated by the flows of energy, water, materials (e.g., OM, nutrients), and biota (e.g., Richardson 1992; Naiman and Décamps 1997; Wipfli and Gregovich 2002). In local reaches, the stability of stream particulate OM dynamics is known to depend on habitat stability that control OM retention and macroinvertebrate consumers (e.g., Death and Winterbourn 1995; Death 2002; Negishi and Richardson 2003). The responses of macroinvertebrate consumers to a given disturbance are also influenced by the predictability of disturbances these consumers have experienced over ecological and evolutionary time scales (Poff and Ward 1990; Poff 1992). Differences in post-harvest responses of particulate OM dynamics among streams may also reflect the variability of past discharge regimes among streams. Hence, stream particulate OM response to forest harvesting are often highly variable, given the interplay of controlling processes that operate across multiple spatial and temporal scales. Our understanding of resilience of streams particulate OM to forest harvesting is hindered by the lack of data, particularly for catchment-scale clearcutting which induces large-scale disruptions of terrestrial-aquatic and upstream-downstream linkages. Full recoveries of litter inputs and breakdown rates are difficult to detect, unless the duration of continuous, post-harvest measurements is commensurate with and even longer than the regrowth of riparian vegetation to reach pre-harvest levels, which can take multiple years to decades (Kiffney and Richardson 2010; Webster et al. 2014a). Conversely, litter inputs and breakdown are often highly resistant and resilient to a smaller extent and lower intensity of forest disturbances, where riparian buffers are established and/or selective harvesting practices are carefully implemented (Kreutzweiser et al. 2004; Lecerf et al. 2012; Smolders et al. 2018; but see Lecerf and Richardson 2010). In my thesis, I quantified multiple stability properties of leaf litter-derived coarse particulate OM quantity and breakdown in coastal rainforest streams in British Columbia, Canada. A major thesis focus is on evaluating the importance of mechanisms mediated by riparian vegetation (e.g., through litterfall, shading, and discharge regulation) that underpin these stability properties. The recovery of litter breakdown rate in logging-affected streams to within the range of streams in non-harvested catchments appeared to be connected with that of shredder densities, and lagged behind that of water quality characteristics (Chapter 6). Enhanced microbial activity 125  on litter – though not measured in this thesis – might also be another process contributing to litter breakdown recovery. In combination with findings from other empirical studies, the temporal scales of post-harvest recovery are now known to range from a few years to close to a century among physical (temperature: Gomi et al. 2006; large wood availability: Zhang et al. 2009; light levels: Hoover et al. 2011), chemical (Feller 2010a), and biological parameters in streams (OM inputs: Kiffney and Richardson 2010; fish density: Young et al. 2011). My work contributed additional evidence that show disparate responses of a stability property (i.e., resilience) to disturbances among state variables in the same ecosystem.  Elucidating and quantitatively comparing the causal mechanisms (and their interactions) of the responses of stability properties can be more definitively achieved by replicated manipulative and process-oriented (or agent-based) modelling studies (see Costello et al. 2018). One main challenge to empirically manipulating individual controlling processes of many stream ecological responses pertains to the large spatial scales (reach to catchment) at which these processes operate (Bennett and Adams 2004), since it is logistically prohibitive especially when assessing the long-term recovery of OM dynamics to forest harvesting. Through modelling biophysical controls on stream particulate OM standing stocks across a realistic range of forest harvesting impacts, I showed that short-term logging-associated changes in riparian litter inputs generally had greater impacts than those of peak flows and stream temperature on the resistance of consumption and retention of stream particulate OM to forest harvesting (Chapter 5). My modelling work demonstrated that resistance of ecological variables could decrease nonlinearly with increasing severity of disturbance, as driven by the exponential relationships linking these variables and their controlling processes (see Karlsson et al. 2005; Stenroth et al. 2014). Complex, non-additive interactions between pairs of controlling processes emerged in the modelling process, and the strength of interactions also depend on disturbance severity. Other model-based non-toxicological analyses of multiple stressor effects in streams have also revealed variable interaction types, contingent on the magnitude of disturbance and environmental context (e.g., Harvey and Railsback 2011; Crossman et al. 2013; Belarde and Railsback 2016). Despite the inherent model uncertainty and limitations, heuristic modelling can be a valuable analytic approach to exploring the complexity of multiple environmental changes and forecasting ecological responses at focal spatial and temporal scales. 126   7.3 Re-evaluating the sensitivity of litter breakdown as a bioassessment tool Measuring and detecting the ecological impacts of disturbances require explicit descriptions of reference conditions. For in-stream total litter breakdown rates, Gessner and Chauvet (2002) benchmarked their range of variability under unimpacted conditions in a bioassessment framework, which has been widely used in empirical studies assessing disturbance impacts on stream functional integrity (see Chauvet et al. 2016). I evaluated whether the range of reference conditions established by this framework could encompass natural, weather-driven, interannual variations of litter breakdown rates (both total and microbially mediated) in temperate perennial streams, which were expected to be induced by the variability of discharge as a major controlling factor (Chapter 3). On the other hand, to quantify the resistance and resilience of litter breakdown to disturbances in question on a large geographic scale, studies commonly standardised the use of leaf material by deploying high-quality alder litter across streams (Woodward et al. 2012; Chauvet et al. 2016). I evaluated the effects of inter-stream differences in decomposer adaptation to locally-derived litter on litter breakdown relative to the range of natural variability of litter breakdown (Chapter 4). Through these two cross-regional studies, I provided region- and site-specific calibration of reference conditions of total litter breakdown within the study regions, which are of practical importance for assessing the impacts of forestry activities as the major agent of human-induced watershed disturbances. When site-specific reference conditions are not determined, total litter breakdown would have to be reduced and/or increased by about 50% of the mean of reference sites to reflect probable impacts of anthropogenic disturbances. Hence, the sensitivity of total litter breakdown in temperate perennial streams – at least within the study regions – appeared to be lower than previously recognised. My results highlights the need for regional- and site-specific adjustments to include the temporal variability of reference conditions for more accurately assessing stream ecological stability (see also Mazor et al. 2009; Hawkins et al. 2010; Pozo et al. 2011). I consider that the use of alder litter from the same source across geographic regions was suitable for investigating the broad-scale stability of stream litter breakdown to disturbances in mid- to late-successional forests, when its measurements are taken in three or more streams in each study region. The average effects of local adaptation to alder litter by 127  microbial decomposers appeared to be minor, which likely resulted in about 20% difference in microbial decomposition rate of litter from the expected values in the absence of local adaptation (see also Jackrel and Wootton 2014; Fenoy et al. 2016). Such an effect size was not distinguishable from the range of natural variation of litter breakdown in most study streams. My two cross-regional studies revealed a generally more dominant control of microbial decomposition on stream litter breakdown than shredder consumption and mechanical abrasion combined during fall, except in some riffle habitats with high current velocity. Previous studies in Malcolm Knapp Research Forest have shown substantial variability in the relative contributions of shredder consumption and microbial decomposition to litter breakdown (Kominoski et al. 2011; Elosegi et al. 2018), although they also differed in study timing (summer vs. fall) and litter species used (red alder vs. black cottonwood (Populus trichocarpa Torr. & Gray ex. Hook)). The biomass (other than fungal biomass), composition, and activity of litter-associated microbes were not measured in my studies. Therefore, future research is needed to better link these attributes to the spatial and temporal differences of microbial decomposition (e.g., Hieber and Gessner 2002; Duarte et al. 2010; Mora-Gómez et al. 2016), and to understand the basis of differences in the relative rates of shredder consumption and microbial decomposition of leaf litter.  7.4 Study caveats and limitations I primarily addressed inter-site and regional variability of stability properties, and their mechanistic drivers in this thesis. Given my relatively limited sampling efforts within individual stream reaches, I might not have fully captured the range of reach-scale variability of litter breakdown rates. In stream reaches with high heterogeneity of riparian canopy cover (and hence light availability), natural variation in litter breakdown can be high, as induced by variability of fungal decomposition, and their facilitation of shredder consumption (Lagrue et al. 2011; Tonin et al. 2018b). Sampling efforts may need to be scaled with the heterogeneity of canopy cover to better characterise the within-reach variability of OM dynamics and their responses to disturbances. I examined the stability of stream particulate OM dynamics focussing on those of alder leaf litter (other litter types were also considered in Chapters 5 and 6). This is because of the greater 128  trophic support by higher-quality alder litter to stream foodwebs in the study regions, given its generally higher colonization and consumption rates compared to other detrital substrates (Hofer and Richardson 2007; Muto et al. 2011; Kominoski et al. 2011). My results with alder litter may not be extrapolated to other litter species, which are avenues for future research. Importantly, disturbance-induced changes in lower-quality conifer litter availability and breakdown could have important trophic and/or non-trophic consequences on stream foodwebs, as shredders and collector-gatherers are known to consume conifer litter, and caddisflies utilise conifer needles to build cases (e.g., Richardson et al. 2004, 2005; Sakai et al. 2016).  7.5 Management implications To limit forest harvesting impacts on the stability of particulate OM quantity and breakdown in small streams, my study findings (in Chapters 5 and 6) support that management practices should prioritise minimising the disturbances of riparian vegetation. Establishing riparian buffers (e.g., ≥30 m wide), and selective harvesting with low basal area of riparian trees removed (or thinning) are likely to maintain particulate OM quantity and terrestrial-derived subsidies to consumers in small streams. Beyond these benefits, these riparian management strategies can limit harvesting impacts on the delivery of ecosystem services related to the provisioning of clean water, carbon sequestration, and the support of ecological communities in stream-riparian systems (e.g., Naiman et al. 1993; Sweeney et al. 2004; Richardson and Béraud 2014; Dybala et al. 2019). In some regions, riparian management paradigms have undergone shifts to allow intentional harvesting to emulate natural disturbances and renewal processes in riparian stands (e.g., insect outbreaks, windthrow) to maintain long-term ecological stability (Kreutzweiser et al. 2012; Sibley et al. 2012). Site-specific intentional harvesting practices should ensure that ecological impacts are within acceptable ranges, for instance, by minimising riparian soil erosion and compaction by avoiding wet soils for skidding and locating roads, and reducing landing areas for logs (see Kreutzweiser et al. 2012). Additionally, hydrologically and ecologically sensitive groundwater-discharge areas should receive sufficient protection during harvesting (Kuglerová et al. 2014, 2017b). Given the high natural, interannual variability of litter breakdown rates in perennial forested streams (shown in Chapter 3), caution should be exercised by managers when using litter assays 129  to assess responses of functional integrity to disturbances. Without well-established reference conditions for individual sites (e.g., with before-after-control-impact study designs), a multi-metric approach that considers the responses of litter breakdown and other structural and functional measures (e.g., ecosystem metabolism) of stream ecological status would ensure comprehensive and reliable bioassessments (e.g., Young et al. 2008; Young and Collier 2009; Tank et al. 2010). Forecasting the consequences of watershed disturbances on aquatic ecosystem services (particularly their resistance) under different land-use change and management scenarios can be achieved by performing simulations using process-based models. There is great potential for collaborations between managers and researchers to develop and apply modelling approaches to meet desired management targets (Katsuyama et al. 2009; Harvey and Railsback 2011; Bussi et al. 2018), and to manage for complex, interactive effects between environmental stressors in watersheds. 130  Bibliography  Abdelnour, A., Stieglitz, M., Pan, F., and McKane, R. 2011. Catchment hydrological responses to forest harvest amount and spatial pattern. Water Resour. Res. 47(9). doi:10.1029/2010WR010165. Aldridge, K.T., Brookes, J.D., and Ganf, G.G. 2009. Rehabilitation of stream ecosystem functions through the reintroduction of coarse particulate organic matter. Restor. Ecol. 17(1): 97–106. doi:10.1111/j.1526-100X.2007.00338.x. Alexander, R.B., Boyer, E.W., Smith, R.A., Schwarz, G.E., and Moore, R.B. 2007. The role of headwater streams in downstream water quality. JAWRA J. Am. Water Resour. Assoc. 43(1): 41–59. doi:10.1111/j.1752-1688.2007.00005.x. Anderson, M.J. 2006. Distance-based tests of homogeneity of multivariate dispersions. Biometrics 62(1): 245–253. doi:10.1111/j.1541-0420.2005.00440.x. Anderson, M.J., and Walsh, D.C.I. 2013. PERMANOVA, ANOSIM, and the Mantel test in the face of heterogeneous dispersions: What null hypothesis are you testing? Ecol. Monogr. 83(4): 557–574. doi:10.1890/12-2010.1. Andrushchenko, I. V, Taylor, B.R., Toxopeus, J., and Wilson, E. 2017. Congregations of the leaf-shredding insect Lepidostoma togatum mediate exceptionally rapid mass loss from leaf litter in Nova Scotia rivers. Hydrobiologia 788(1): 245–265. doi:10.1007/s10750-016-3001-6. APHA. 2017. Standard methods for the examination of water and wastewater. In 23rd edition. Edited ByE.W. Rice, R.B. Baird, and A.D. Eaton. American Public Health Association, American Water Works Association, Water Environment Federation, Washington, DC, USA. Arbizu, P.M. 2017. pairwiseAdonis: Pairwise multilevel comparison using adonis. R package version 0.0.1. Available from https://github.com/pmartinezarbizu/pairwiseAdonis [accessed 13 August 2018]. Arroita, M., Flores, L., Larrañaga, A., Chauvet, E., and Elosegi, A. 2018. Hydrological contingency: drying history affects aquatic microbial decomposition. Aquat. Sci. 80(3): 31. doi:10.1007/s00027-018-0582-3. 131  Artigas, J., Gaudes, A., Munoz, I., Romani, A.M., and Sabater, S. 2011. Fungal and bacterial colonization of submerged leaf litter in a Mediterranean stream. Int. Rev. Hydrobiol. 96(3): 221–234. doi:10.1002/iroh.201111355. Ayres, E., Steltzer, H., Simmons, B.L., Simpson, R.T., Steinweg, J.M., Wallenstein, M.D., et al. 2009. Home-field advantage accelerates leaf litter decomposition in forests. Soil Biol. Biochem. 41(3): 606–610. doi:10.1016/j.soilbio.2008.12.022. Baer, S.G., Siler, E.R., Eggert, S.L., and Wallace, J.B. 2001. Colonization and production of macroinvertebrates on artificial substrata: upstream-downstream responses to a leaf litter exclusion manipulation. Freshw. Biol. 46(3): 347–365. doi:10.1046/j.1365-2427.2001.00677.x. Baho, D.L., Allen, C.R., Garmestani, A., Fried-Petersen, H., Renes, S.E., Gunderson, L., et al. 2017. A quantitative framework for assessing ecological resilience. Ecol. Soc. 22(3): 17. doi:10.5751/ES-09427-220317. Baker, D.B., Richards, R.P., Loftus, T.T., and Kramer, J.W. 2004. A new flashiness index: characteristics and applications to Midwestern rivers and streams. J. Am. Water Resour. Assoc. 40(2): 503–522. doi:10.1111/j.1752-1688.2004.tb01046.x. Bassar, R.D., Marshall, M.C., López-Sepulcre, A., Zandonà, E., Auer, S.K., Travis, J., et al. 2010. Local adaptation in Trinidadian guppies alters ecosystem processes. Proc. Natl. Acad. Sci. 107(8): 3616–3621. doi:10.1073/pnas.0908023107. Bates, D., Maechler, M., Bolker, B., and Walker, S. 2017. lme4: Linear mixed-effects models using Eigen and S4. R package version 1.1.13. http://cran.r-project.org/web/packages/lme4. Batista, D., Pascoal, C., and Cássio, F. 2012. Impacts of warming on aquatic decomposers along a gradient of cadmium stress. Environ. Pollut. 169: 35–41. doi:10.1016/j.envpol.2012.05.021. Battin, T.J., Kaplan, L.A., Findlay, S., Hopkinson, C.S., Marti, E., Packman, A.I., et al. 2008. Biophysical controls on organic carbon fluxes in fluvial networks. Nat. Geosci. 1(2): 95–100. doi:10.1038/ngeo101. Beechie, T.J., Sear, D.A., Olden, J.D., Pess, G.R., Buffington, J.M., Moir, H., et al. 2010. Process-based principles for restoring river ecosystems. Bioscience 60(3): 209–222. doi:10.1525/bio.2010.60.3.7. 132  Belarde, T.A., and Railsback, S.F. 2016. New predictions from old theory: emergent effects of multiple stressors in a model of piscivorous fish. Ecol. Modell. 326: 54–62. doi:10.1016/j.ecolmodel.2015.07.012. Benfield, E.F., Webster, J.R., Tank, J.L., and Hutchens, J.J. 2001. Long-term patterns in leaf breakdown in streams in response to watershed logging. Int. Rev. Hydrobiol. 86(4–5): 467–474. doi:10.1002/1522-2632(200107)86:4/5<467::AID-IROH467>3.0.CO;2-1. Bennett, L.T., and Adams, M.A. 2004. Assessment of ecological effects due to forest harvesting: approaches and statistical issues. J. Appl. Ecol. 41(4): 585–598. doi:10.1111/j.0021-8901.2004.00924.x. Bilby, R.E., and Bisson, P.A. 1992. Allochthonous versus autochthonous organic matter contributions to the trophic support of fish populations in clear-cut and old-growth forested streams. Can. J. Fish. Aquat. Sci. 49(3): 540–551. doi:10.1139/f92-064. Blackburn, W.H., Wood, J.C., and DeHaven, M.G. 1986. Storm flow and sediment losses from site-prepared forestland in east Texas. Water Resour. Res. 22(5): 776–784. doi:10.1029/WR022i005p00776. Bladon, K.D., Cook, N.A., Light, J.T., and Segura, C. 2016. A catchment-scale assessment of stream temperature response to contemporary forest harvesting in the Oregon Coast Range. For. Ecol. Manage. 379: 153–164. doi:10.1016/j.foreco.2016.08.021. Bladon, K.D., Segura, C., Cook, N.A., Bywater-Reyes, S., and Reiter, M. 2017. A multicatchment analysis of headwater and downstream temperature effects from contemporary forest harvesting. Hydrol. Process. 32(2): 293–304. doi:10.1002/hyp.11415. Blanchet, F.G., Legendre, P., and Borcard, D. 2008. Forward selection of explanatory variables. Ecology 89(9): 2623–2632. doi:10.1890/07-0986.1. Bond, N. 2016. hydrostats: Hydrologic Indices for Daily Time Series Data. R package version 0.2.5. http://cran.r-project.org/web/packages/hydrostats. Booker, D.J., Snelder, T.H., Greenwood, M.J., and Crow, S.K. 2015. Relationships between invertebrate communities and both hydrological regime and other environmental factors across New Zealand’s rivers. Ecohydrology 8(1): 13–32. doi:10.1002/eco.1481. Borcard, D., Gillet, F., and Legendre, P. 2011. Canonical ordination. In Numerical ecology with R. Edited by D. Borcard, F. Gillet, and P. Legendre. Springer, New York, USA. pp. 153–133  225. Bourque, C.-A., and Pomeroy, J.H. 2001. Effects of forest harvesting on summer stream temperatures in New Brunswick, Canada: an inter-catchment, multiple-year comparison. Hydrol. Earth Syst. Sci. 5(4): 599–614. doi:10.5194/hess-5-599-2001. Boyero, L., Pearson, R.G., Gessner, M.O., Barmuta, L.A., Ferreira, V., Graça, M.A.S., et al. 2011. A global experiment suggests climate warming will not accelerate litter decomposition in streams but might reduce carbon sequestration. Ecol. Lett. 14(3): 289–294. doi:10.1111/j.1461-0248.2010.01578.x. Boyero, L., Pearson, R.G., Gessner, M.O., Dudgeon, D., Ramírez, A., Yule, C.M., et al. 2015. Leaf-litter breakdown in tropical streams: is variability the norm? Freshw. Sci. 34(2): 759–769. doi:10.1086/681093. Brosed, M., Lamothe, S., and Chauvet, E. 2016. Litter breakdown for ecosystem integrity assessment also applies to streams affected by pesticides. Hydrobiologia 773(1): 87–102. doi:10.1007/s10750-016-2681-2. Bulling, M.T., Hicks, N., Murray, L., Paterson, D.M., Raffaelli, D., White, P.C.L., et al. 2010. Marine biodiversity–ecosystem functions under uncertain environmental futures. Philos. Trans. R. Soc. London B Biol. Sci. 365(1549): 2107–2116. doi:10.1098/rstb.2010.0022. Bundschuh, M., and McKie, B.G. 2016. An ecological and ecotoxicological perspective on fine particulate organic matter in streams. Freshw. Biol. 61(12): 2063–2074. doi:10.1111/fwb.12608. Burton, J.I., Olson, D.H., and Puettmann, K.J. 2016. Effects of riparian buffer width on wood loading in headwater streams after repeated forest thinning. For. Ecol. Manage. 372: 247–257. doi:10.1016/j.foreco.2016.03.053. Bussi, G., Whitehead, P.G., Gutiérrez-Cánovas, C., Ledesma, J.L.J., Ormerod, S.J., and Couture, R.-M. 2018. Modelling the effects of climate and land-use change on the hydrochemistry and ecology of the River Wye (Wales). Sci. Total Environ. 627: 733–743. doi:10.1016/j.scitotenv.2018.01.295. Buzby, K.M., and Perry, S.A. 2000. Modeling the potential effects of climate change on leaf pack processing in central Appalachian streams. Can. J. Fish. Aquat. Sci. 57(9): 1773–1783. doi:10.1139/f00-113. 134  Cáceres, M. De, and Legendre, P. 2009. Associations between species and groups of sites: indices and statistical inference. Ecology 90(12): 3566–3574. doi:10.1890/08-1823.1. Campbell, I.C., and Doeg, T.J. 1989. Impact of timber harvesting and production on streams: a review. Mar. Freshw. Res. 40(5): 519–539. doi:10.1071/MF9890519. Campbell, R.E., Winterbourn, M.J., Cochrane, T.A., and McIntosh, A.R. 2015. Flow-related disturbance creates a gradient of metacommunity types within stream networks. Landsc. Ecol. 30(4): 667–680. doi:10.1007/s10980-015-0164-x. Capon, S.J., Lynch, A.J.J., Bond, N., Chessman, B.C., Davis, J., Davidson, N., et al. 2015. Regime shifts, thresholds and multiple stable states in freshwater ecosystems; a critical appraisal of the evidence. Sci. Total Environ. 534: 122–130. doi:10.1016/j.scitotenv.2015.02.045. Cardinale, B.J., Duffy, J.E., Gonzalez, A., Hooper, D.U., Perrings, C., Venail, P., et al. 2012. Biodiversity loss and its impact on humanity. Nature 486: 59–67. Carpenter, S., Walker, B., Anderies, J.M., and Abel, N. 2001. From metaphor to measurement: resilience of what to what? Ecosystems 4(8): 765–781. doi:10.1007/s10021-001-0045-9. Carpenter, S.R., Cole, J.J., Pace, M.L., Van de Bogert, M., Bade, D.L., Bastviken, D., et al. 2005. Ecosystem subsidies: terrestrial support of aquatic food webs from 13C addition to contrasting lakes. Ecology 86(10): 2737–2750. doi:10.1890/04-1282. Casas, J.J., Larrañaga, A., Menéndez, M., Pozo, J., Basaguren, A., Martínez, A., et al. 2013. Leaf litter decomposition of native and introduced tree species of contrasting quality in headwater streams: How does the regional setting matter? Sci. Total Environ. 458–460: 197–208. doi:10.1016/j.scitotenv.2013.04.004. Chadwick, M.A., Dobberfuhl, D.R., Benke, A.C., Huryn, A.D., Suberkropp, K., and Thiele, J.E. 2006. Urbanization affects stream ecosystem function by altering hydrology, chemistry, and biotic richness. Ecol. Appl. 16(5): 1796–1807. doi:10.1890/1051-0761(2006)016[1796:UASEFB]2.0.CO;2. Chará-Serna, A.M., and Richardson, J.S. 2017. Chlorpyrifos interacts with other agricultural stressors to alter stream communities in laboratory microcosms. Ecol. Appl. 28(1): 162–176. doi:10.1002/eap.1637. Chauvet, E., Ferreira, V., Giller, P.S., McKie, B.G., Tiegs, S.D., Woodward, G., et al. 2016. 135  Litter decomposition as an indicator of stream ecosystem functioning at local-to-continental scales: insights from the European RivFunction Project. Adv. Ecol. Res. 55: 99–182. doi:10.1016/bs.aecr.2016.08.006. Chávez-Vergara, B., Merino, A., González-Rodríguez, A., Oyama, K., and García-Oliva, F. 2018. Direct and legacy effects of plant-traits control litter decomposition in a deciduous oak forest in Mexico. PeerJ 6: e5095. doi:10.7717/peerj.5095. Cheng, J.D., Black, T.A., de Vries, J., Willington, R.P., and Goodell, B.C. 1975. The evaluation of initial changes in peak streamflow following logging of a watershed on the west coast of Canada. In International Association of Hydrological Sciences Symposium of Tokyo. Publication No 117. pp. 475–486. Chizinski, C.J., Vondracek, B., Blinn, C.R., Newman, R.M., Atuke, D.M., Fredricks, K., et al. 2010. The influence of partial timber harvesting in riparian buffers on macroinvertebrate and fish communities in small streams in Minnesota, USA. For. Ecol. Manage. 259(10): 1946–1958. doi:10.1016/j.foreco.2010.02.006. Clapcott, J.E., Collier, K.J., Death, R.G., Goodwin, E.O., Harding, J.S., Kelly, D., et al. 2012. Quantifying relationships between land-use gradients and structural and functional indicators of stream ecological integrity. Freshw. Biol. 57(1): 74–90. doi:10.1111/j.1365-2427.2011.02696.x. Clarke, A., Mac Nally, R., Bond, N., and Lake, P.S. 2008. Macroinvertebrate diversity in headwater streams: a review. Freshw. Biol. 53(9): 1707–1721. doi:10.1111/j.1365-2427.2008.02041.x. Clausen, B., and Biggs, B.J.F. 2000. Flow variables for ecological studies in temperate streams: groupings based on covariance. J. Hydrol. 237(3–4): 184–197. doi:10.1016/S0022-1694(00)00306-1. Clinton, B.D., Vose, J.M., and Fowler, D.L. 2010. Flat Branch monitoring project: stream water temperature and sediment responses to forest cutting in the riparian zone. US Department of Agriculture Forest Service, Southern Research Station, Asheville, NC, USA. Colas, F., Baudoin, J.-M., Gob, F., Tamisier, V., Valette, L., Kreutzenberger, K., et al. 2017. Scale dependency in the hydromorphological control of a stream ecosystem functioning. Water Res. 115: 60–73. doi:10.1016/j.watres.2017.01.061. 136  Cole, E., and Newton, M. 2013. Influence of streamside buffers on stream temperature response following clear-cut harvesting in western Oregon. Can. J. For. Res. 43(11): 993–1005. doi:10.1139/cjfr-2013-0138. Connolly, N.M., and Pearson, R.G. 2013. Nutrient enrichment of a heterotrophic stream alters leaf litter nutritional quality and shredder physiological condition via the microbial pathway. Hydrobiologia 718(1): 85–92. doi:10.1007/s10750-013-1605-7. Correa-Araneda, F., Boyero, L., Figueroa, R., Sánchez, C., Abdala, R., Ruiz-García, A., et al. 2015. Joint effects of climate warming and exotic litter (Eucalyptus globulus Labill.) on stream detritivore fitness and litter breakdown. Aquat. Sci. 77(2): 197–205. doi:10.1007/s00027-014-0379-y. Costanza, R. 1992. Towards an operational definition of ecosystem health. In Ecosystem health: New goals for environmental management. Edited by R. Costanza, B.G. Norton, and B.D. Haskell. Island Press, Washington, D.C., USA, D.C., USA. pp. 237–256. Costello, D.M., Kulacki, K.J., McCarthy, M.E., Tiegs, S.D., and Cardinale, B.J. 2018. Ranking stressor impacts on periphyton structure and function with mesocosm experiments and environmental-change forecasts. PLoS One 13(9): e0204510. doi:10.1371/journal.pone.0204510. Coulthard, B., Smith, D.J., and Meko, D.M. 2016. Is worst-case scenario streamflow drought underestimated in British Columbia? A multi-century perspective for the south coast, derived from tree-rings. J. Hydrol. 534(Supplement C): 205–218. doi:10.1016/j.jhydrol.2015.12.030. Covich, A.P., Palmer, M.A., and Crowl, T.A. 1999. The role of benthic invertebrate species in freshwater ecosystems: zoobenthic species influence energy flows and nutrient cycling. Bioscience 49(2): 119–127. doi:10.2307/1313537. Creed, I.F., and Band, L.E. 1998. Exploring functional similarity in the export of nitrate-N from forested catchments: a mechanistic modeling approach. Water Resour. Res. 34(11): 3079–3093. doi:10.1029/98WR02102. Creed, I.F., Hwang, T., Lutz, B., and Way, D. 2015. Climate warming causes intensification of the hydrological cycle, resulting in changes to the vernal and autumnal windows in a northern temperate forest. Hydrol. Process. 29(16): 3519–3534. doi:10.1002/hyp.10450. 137  Crenshaw, C.L., Valett, H.M., and Webster, J.R. 2002. Effects of augmentation of coarse particulate organic matter on metabolism and nutrient retention in hyporheic sediments. Freshw. Biol. 47(10): 1820–1831. doi:10.1046/j.1365-2427.2002.00928.x. Cristan, R., Aust, W.M., Bolding, M.C., Barrett, S.M., Munsell, J.F., and Schilling, E. 2016. Effectiveness of forestry best management practices in the United States: literature review. For. Ecol. Manage. 360: 133–151. doi:10.1016/j.foreco.2015.10.025. Cross, W.F., Johnson, B.R., Wallace, J.B., and Rosemond, A.D. 2005. Contrasting response of stream detritivores to long-term nutrient enrichment. Limnol. Oceanogr. 50(6): 1730–1739. doi:10.4319/lo.2005.50.6.1730. Crossman, J., Whitehead, P.G., Futter, M.N., Jin, L., Shahgedanova, M., Castellazzi, M., et al. 2013. The interactive responses of water quality and hydrology to changes in multiple stressors, and implications for the long-term effective management of phosphorus. Sci. Total Environ. 454–455: 230–244. doi:10.1016/j.scitotenv.2013.02.033. Cuddington, K., Fortin, M.-J., Gerber, L.R., Hastings, A., Liebhold, A., O’Connor, M., et al. 2013. Process-based models are required to manage ecological systems in a changing world. Ecosphere 4(2): art20. doi:10.1890/ES12-00178.1. Curtis, R.O., and Carey, A.B. 1996. Timber supply in the Pacific Northwest: managing for economic and ecological values in Douglas-fir forests. J. For. 94(9): 4–37. doi:10.1093/jof/94.9.4. Dakos, V., Carpenter, S.R., van Nes, E.H., and Scheffer, M. 2015. Resilience indicators: prospects and limitations for early warnings of regime shifts. Philos. Trans. R. Soc. London B Biol. Sci. 370(1659): 20130263. doi:10.1098/rstb.2013.0263. Danehy, R.J., Chan, S.S., Lester, G.T., Langshaw, R.B., and Turner, T.R. 2007. Periphyton and macroinvertebrate assemblage structure in headwaters bordered by mature, thinned, and clearcut Douglas-fir stands. For. Sci. 53(2): 294–307. doi:10.1093/forestscience/53.2.294. Dang, C.K., Chauvet, E., and Gessner, M.O. 2005. Magnitude and variability of process rates in fungal diversity-litter decomposition relationships. Ecol. Lett. 8(11): 1129–1137. doi:10.1111/j.1461-0248.2005.00815.x. Danger, M., Cornut, J., Elger, A., and Chauvet, E. 2012. Effects of burial on leaf litter quality, microbial conditioning and palatability to three shredder taxa. Freshw. Biol. 57(5): 1017–138  1030. doi:10.1111/j.1365-2427.2012.02762.x. Danger, M., Cornut, J., Chauvet, E., Chavez, P., Elger, A., and Lecerf, A. 2013. Benthic algae stimulate leaf litter decomposition in detritus-based headwater streams: a case of aquatic priming effect? Ecology 94(7): 1604–1613. doi:10.1890/12-0606.1. Dangles, O., and Malmqvist, B. 2004. Species richness–decomposition relationships depend on species dominance. Ecol. Lett. 7(5): 395–402. doi:10.1111/j.1461-0248.2004.00591.x. Dangles, O., Gessner, M.O., Guerold, F., and Chauvet, E. 2004. Impacts of stream acidification on litter breakdown: implications for assessing ecosystem functioning. J. Appl. Ecol. 41(2): 365–378. doi:10.1111/j.0021-8901.2004.00888.x. Darling, E.S., and Côté, I.M. 2008. Quantifying the evidence for ecological synergies. Ecol. Lett. 11(12): 1278–1286. doi:10.1111/j.1461-0248.2008.01243.x. Datry, T., Corti, R., Claret, C., and Philippe, M. 2011. Flow intermittence controls leaf litter breakdown in a French temporary alluvial river: the “drying memory.” Aquat. Sci. 73(4): 471–483. doi:10.1007/s00027-011-0193-8. Davies, P.E., Cook, L.S.J., McIntosh, P.D., and Munks, S.A. 2005. Changes in stream biota along a gradient of logging disturbance, 15 years after logging at Ben Nevis, Tasmania. For. Ecol. Manage. 219(2–3): 132–148. doi:10.1016/j.foreco.2005.09.006. Davis, S.F., and Winterbourn, M.J. 1977. Breakdown and colonization of Nothofagus leaves in a New Zealand stream. Oikos 28(2/3): 250–255. doi:10.2307/3543978. Death, R.G. 2002. Predicting invertebrate diversity from disturbance regimes in forest streams. Oikos 97(1): 18–30. doi:10.1034/j.1600-0706.2002.970102.x. Death, R.G., and Winterbourn, M.J. 1995. Diversity patterns in stream benthic invertebrate communities: the influence of habitat stability. Ecology 76(5): 1446–1460. doi:10.2307/1938147. Deslauriers, D., Chipps, S.R., Breck, J.E., Rice, J.A., and Madenjian, C.P. 2017. Fish Bioenergetics 4.0: an R-based modeling application. Fisheries 42(11): 586–596. doi:10.1080/03632415.2017.1377558. Devictor, V., Clavel, J., Julliard, R., Lavergne, S., Mouillot, D., Thuiller, W., et al. 2010. Defining and measuring ecological specialization. J. Appl. Ecol. 47(1): 15–25. doi:10.1111/j.1365-2664.2009.01744.x. 139  Diaz-Pulido, G., McCook, L.J., Dove, S., Berkelmans, R., Roff, G., Kline, D.I., et al. 2009. Doom and boom on a resilient reef: climate change, algal overgrowth and coral recovery. PLoS One 4(4): e5239. doi:10.1371/journal.pone.0005239. Dieter, D., von Schiller, D., García-Roger, E.M., Sánchez-Montoya, M.M., Gómez, R., Mora-Gómez, J., et al. 2011. Preconditioning effects of intermittent stream flow on leaf litter decomposition. Aquat. Sci. 73(4): 599–609. doi:10.1007/s00027-011-0231-6. Dodds, W.K., Clements, W.H., Gido, K., Hilderbrand, R.H., and King, R.S. 2010. Thresholds, breakpoints, and nonlinearity in freshwaters as related to management. J. North Am. Benthol. Soc. 29(3): 988–997. doi:10.1899/09-148.1. Dodds, W.K., Perkin, J.S., and Gerken, J.E. 2013. Human impact on freshwater ecosystem services: a global perspective. Environ. Sci. Technol. 47(16): 9061–9068. doi:10.1021/es4021052. Domingos, C., Ferreira, V., Canhoto, C., and Swan, C. 2015. Warming, and the presence of a dominant shredder, drive variation in decomposer communities in a mountain stream. Aquat. Sci. 77(1): 129–140. doi:10.1007/s00027-014-0378-z. Donohue, I., Petchey, O.L., Montoya, J.M., Jackson, A.L., McNally, L., Viana, M., et al. 2013. On the dimensionality of ecological stability. Ecol. Lett. 16(4): 421–429. doi:10.1111/ele.12086. Donohue, I., Hillebrand, H., Montoya, J.M., Petchey, O.L., Pimm, S.L., Fowler, M.S., et al. 2016. Navigating the complexity of ecological stability. Ecol. Lett. 19(9): 1172–1185. doi:10.1111/ele.12648. Dosskey, M.G., Vidon, P., Gurwick, N.P., Allan, C.J., Duval, T.P., and Lowrance, R. 2010. The role of riparian vegetation in protecting and improving chemical water quality in streams. JAWRA J. Am. Water Resour. Assoc. 46(2): 261–277. doi:10.1111/j.1752-1688.2010.00419.x. Downes, B.J., Barmuta, L.A., Fairweather, P.G., Faith, D.P., Keough, M.J., Lake, P.S., et al. 2002. Monitoring ecological impacts: concepts and practice in flowing waters. Cambridge University Press, Cambridge, UK. Drever, C.R., Peterson, G., Messier, C., Bergeron, Y., and Flannigan, M. 2006. Can forest management based on natural disturbances maintain ecological resilience? Can. J. For. Res. 140  36(9): 2285–2299. doi:10.1139/x06-132. Duarte, S., Pascoal, C., Alves, A., Correia, A., and Cássio, F. 2010. Assessing the dynamic of microbial communities during leaf decomposition in a low-order stream by microscopic and molecular techniques. Microbiol. Res. 165(5): 351–362. doi:10.1016/j.micres.2009.06.002. Dybala, K.E., Matzek, V., Gardali, T., and Seavy, N.E. 2019. Carbon sequestration in riparian forests: a global synthesis and meta-analysis. Glob. Chang. Biol. 25(1): 57–67. doi:10.1111/gcb.14475. Eggert, S.L., Wallace, J.B., Meyer, J.L., and Webster, J.R. 2012. Storage and export of organic matter in a headwater stream: responses to long-term detrital manipulations. Ecosphere 3(9): art75. doi:10.1890/ES12-00061.1. Egli, L., Weise, H., Radchuk, V., Seppelt, R., and Grimm, V. 2018. Exploring resilience with agent-based models: state of the art, knowledge gaps and recommendations for coping with multidimensionality. Ecol. Complex. doi:10.1016/j.ecocom.2018.06.008. Ehrman, T.P., and Lamberti, G.A. 1992. Hydraulic and particulate matter retention in a 3rd-order Indiana stream. J. North Am. Benthol. Soc. 11(4): 341–349. doi:10.2307/1467556. Elmqvist, T., Folke, C., Nyström, M., Peterson, G., Bengtsson, J., Walker, B., et al. 2003. Response diversity, ecosystem change, and resilience. Front. Ecol. Environ. 1(9): 488–494. doi:10.1890/1540-9295(2003)001[0488:RDECAR]2.0.CO;2. Elosegi, A., Nicolás, A., and Richardson, J.S. 2018. Priming of leaf litter decomposition by algae seems of minor importance in natural streams during autumn. PLoS One 13(9): e0200180. doi:10.1371/journal.pone.0200180. Ely, D.T., and Wallace, J.B. 2010. Long-term functional group recovery of lotic macroinvertebrates from logging disturbance. Can. J. Fish. Aquat. Sci. 67(7): 1126–1134. doi:10.1139/F10-045. Ensign, S.H., and Doyle, M.W. 2006. Nutrient spiraling in streams and river networks. J. Geophys. Res. Biogeosciences 111(G4): G04009. doi:10.1029/2005JG000114. Evans-White, M.A., Dodds, W.K., Huggins, D.G., and Baker, D.S. 2009. Thresholds in macroinvertebrate biodiversity and stoichiometry across water-quality gradients in Central Plains (USA) streams. J. North Am. Benthol. Soc. 28(4): 855–868. doi:10.1899/08-113.1. Evans, M.R., Norris, K.J., and Benton, T.G. 2012. Predictive ecology: systems approaches. 141  Philos. Trans. R. Soc. B Biol. Sci. 367(1586): 163–169. doi:10.1098/rstb.2011.0191. Fazi, S., Amalfitano, S., Pernthaler, J., and Puddu, A. 2005. Bacterial communities associated with benthic organic matter in headwater stream microhabitats. Environ. Microbiol. 7(10): 1633–1640. doi:10.1111/j.1462-2920.2005.00857.x. Feld, C.K., Martins da Silva, P., Paulo Sousa, J., De Bello, F., Bugter, R., Grandin, U., et al. 2009. Indicators of biodiversity and ecosystem services: a synthesis across ecosystems and spatial scales. Oikos 118(12): 1862–1871. doi:10.1111/j.1600-0706.2009.17860.x. Feller, M.C. 2010a. Forest harvesting and stream chemistry. Streamline Watershed Manag. Bull. 13(2): 6–8. Feller, M.C. 2010b. Trends in precipitation and streamwater chemistry in East Creek watershed in southwestern British Columbia, 1971–2008. J. Limnol. 69: 77–91. doi:10.4081/jlimnol.2010.s1.77. Feller, M.C., and Kimmins, J.P. 1984. Effects of clearcutting and slash burning on streamwater chemistry and watershed nutrient budgets in southwestern British Columbia. Water Resour. Res. 20(1): 29–40. doi:10.1029/WR020i001p00029. Fenoy, E., Casas, J.J., Díaz-López, M., Rubio, J., Guil-Guerrero, J.L., and Moyano-López, F.J. 2016. Temperature and substrate chemistry as major drivers of interregional variability of leaf microbial decomposition and cellulolytic activity in headwater streams. FEMS Microbiol. Ecol. 92(11): fiw169. doi:10.1093/femsec/fiw169. Fernandes, I., Pascoal, C., Guimaraes, H., Pinto, R., Sousa, I., and Cassio, F. 2012. Higher temperature reduces the effects of litter quality on decomposition by aquatic fungi. Freshw. Biol. 57(11): 2306–2317. doi:10.1111/fwb.12004. Fernandes, I., Seena, S., Pascoal, C., and Cássio, F. 2014. Elevated temperature may intensify the positive effects of nutrients on microbial decomposition in streams. Freshw. Biol. 59(11): 2390–2399. doi:10.1111/fwb.12445. Ferreira, V., and Chauvet, E. 2011. Synergistic effects of water temperature and dissolved nutrients on litter decomposition and associated fungi. Glob. Chang. Biol. 17(1): 551–564. doi:10.1111/j.1365-2486.2010.02185.x. Ferreira, V., Graça, M.A.S., de Lima, J.L.M.P., and Gomes, R. 2006. Role of physical fragmentation and invertebrate activity in the breakdown rate of leaves. Arch. für 142  Hydrobiol. 165(4): 493–513. doi:10.1127/0003-9136/2006/0165-0493. Ferreira, V., Chauvet, E., and Canhoto, C. 2014. Effects of experimental warming, litter species, and presence of macroinvertebrates on litter decomposition and associated decomposers in a temperate mountain stream. Can. J. Fish. Aquat. Sci. 72(2): 206–216. doi:10.1139/cjfas-2014-0119. Ferreira, V., Castagneyrol, B., Koricheva, J., Gulis, V., Chauvet, E., and Graça, M.A.S. 2015. A meta-analysis of the effects of nutrient enrichment on litter decomposition in streams. Biol. Rev. 90(3): 669–688. doi:10.1111/brv.12125. Ferreira, V., Koricheva, J., Pozo, J., and Graça, M.A.S. 2016a. A meta-analysis on the effects of changes in the composition of native forests on litter decomposition in streams. For. Ecol. Manage. 364: 27–38. doi:10.1016/j.foreco.2016.01.002. Ferreira, V., Koricheva, J., Duarte, S., Niyogi, D.K., and Guérold, F. 2016b. Effects of anthropogenic heavy metal contamination on litter decomposition in streams – a meta-analysis. Environ. Pollut. 210(Supplement C): 261–270. doi:10.1016/j.envpol.2015.12.060. Findlay, S., Tank, J., Dye, S., Valett, H.M., Mulholland, P.J., McDowell, W.H., et al. 2002. A cross-system comparison of bacterial and fungal biomass in detritus pools of headwater streams. Microb. Ecol. 43(1): 55–66. doi:10.1007/s00248-001-1020-x. Folke, C. 2006. Resilience: The emergence of a perspective for social–ecological systems analyses. Glob. Environ. Chang. 16(3): 253–267. doi:10.1016/j.gloenvcha.2006.04.002. Folke, C., Carpenter, S., Walker, B., Scheffer, M., Elmqvist, T., Gunderson, L., et al. 2004. Regime shifts, resilience, and biodiversity in ecosystem management. Annu. Rev. Ecol. Evol. Syst. 35: 557–581. doi:10.2307/30034127. Foster, N.W., Beall, F.D., and Kreutzweiser, D.P. 2005. The role of forests in regulating water: The Turkey Lakes Watershed case study. For. Chron. 81(1): 142–148. doi:10.5558/tfc81142-1. Foucreau, N., Piscart, C., Puijalon, S., and Hervant, F. 2016. Effects of rising temperature on a functional process: consumption and digestion of leaf litter by a freshwater shredder. Fundam. Appl. Limnol. 187(4): 295–306. doi:10.1127/fal/2016/0841. Frady, C., Johnson, S., and Li, J. 2007. Stream macroinvertebrate community responses as legacies of forest harvest at the H.J. Andrews Experimental Forest, Oregon. For. Sci. 53(2): 143  281–293. doi:10.1093/forestscience/53.2.281. Francoeur, S.N., Rier, S.T., and Whorley, S.B. 2013. Methods for sampling and analyzing wetland algae. In Wetland Techniques SE - 1. Edited by J.T. Anderson and C.A. Davis. Springer Netherlands. pp. 1–58. Franzitta, G., Hanley, M.E., Airoldi, L., Baggini, C., Bilton, D.T., Rundle, S.D., et al. 2015. Home advantage? Decomposition across the freshwater-estuarine transition zone varies with litter origin and local salinity. Mar. Environ. Res. 110: 1–7. doi:10.1016/j.marenvres.2015.07.012. Freeman, M.C., Pringle, C.M., and Jackson, C.R. 2007. Hydrologic connectivity and the contribution of stream headwaters to ecological integrity at regional scales. JAWRA J. Am. Water Resour. Assoc. 43(1): 5–14. doi:10.1111/j.1752-1688.2007.00002.x. Freschet, G.T., Aerts, R., and Cornelissen, J.H.C. 2012. Multiple mechanisms for trait effects on litter decomposition: moving beyond home-field advantage with a new hypothesis. J. Ecol. 100(3): 619–630. doi:10.1111/j.1365-2745.2011.01943.x. Frossard, A., Gerull, L., Mutz, M., and Gessner, M.O. 2013. Litter supply as driver of microbial activity and community structure on decomposing leaves: a test in experimental streams. Appl. Environ. Microbiol.: AEM.00747-13. doi:10.1128/AEM.00747-13. Fuchs, S.A., Hinch, S.G., and Mellina, E. 2003. Effects of streamside logging on stream macroinvertebrate communities and habitat in the sub-boreal forests of British Columbia, Canada. Can. J. For. Res. 33(8): 1408–1415. doi:10.1139/x03-070. Futter, M.N., Butterfield, D., Cosby, B.J., Dillon, P.J., Wade, A.J., and Whitehead, P.G. 2007. Modeling the mechanisms that control in-stream dissolved organic carbon dynamics in upland and forested catchments. Water Resour. Res. 43(2). doi:10.1029/2006WR004960. García, L., Richardson, J.S., and Pardo, I. 2012. Leaf quality influences invertebrate colonization and drift in a temperate rainforest stream. Can. J. Fish. Aquat. Sci. 69(10): 1663–1673. doi:10.1139/f2012-090. Garman, G.C., and Moring, J.R. 1991. Initial effects of deforestation on physical characteristics of a boreal river. Hydrobiologia 209(1): 29–37. doi:10.1007/BF00006715. Gessner, M.O. 2005. Ergosterol as a measure of fungal biomass. In Methods to Study Litter Decomposition: A Practical Guide. Edited by M.S. Graça, F. Bärlocher, and M. Gessner. 144  Springer Netherlands. pp. 189–196. Gessner, M.O., and Chauvet, E. 1993. Ergosterol-to-biomass conversion factors for aquatic hyphomycetes. Appl. Environ. Microbiol. 59(2): 502–507. Available from http://aem.asm.org/content/59/2/502.abstract. Gessner, M.O., and Chauvet, E. 1994. Importance of stream microfungi in controlling breakdown rates of leaf litter. Ecology 75(6): 1807–1817. doi:10.2307/1939639. Gessner, M.O., and Chauvet, E. 2002. A case for using litter breakdown to assess functional stream integrity. Ecol. Appl. 12(2): 498–510. doi:10.1890/1051-0761(2002)012[0498:ACFULB]2.0.CO;2. Gessner, M.O., and Konstanz, J.S. 1989. Leaching kinetics of fresh leaf-litter with implications for the current concept of leaf-processing in streams. Arch. für Hydrobiol. 115(1): 81–90. Gholz, H.L., Wedin, D.A., Smitherman, S.M., Harmon, M.E., and Parton, W.J. 2000. Long-term dynamics of pine and hardwood litter in contrasting environments: toward a global model of decomposition. Glob. Chang. Biol. 6(7): 751–765. doi:10.1046/j.1365-2486.2000.00349.x. Gjerløv, C., and Richardson, J.S. 2010. Experimental increases and reductions of light to streams: effects on periphyton and macroinvertebrate assemblages in a coniferous forest landscape. Hydrobiologia 652(1): 195–206. doi:10.1007/s10750-010-0331-7. Golding, D.L. 1987. Changes in streamflow peaks following timber harvest of a coastal British Columbia watershed. In International Association of Hydrological Sciences. Publication No 167. pp. 509–517. Golladay, S.W., Webster, J.R., and Benfield, E.F. 1989. Changes in stream benthic organic matter following watershed disturbance. Holarct. Ecol. 12(2): 96–105. doi:10.2307/3682193. Golladay, S.W., Webster, J.R., Benfield, E.F., and Swank, W.T. 1992. Changes in stream stability following forest clearing as indicated by storm nutrient budgets. Arch. für Hydrobiol. - Suppl. 90(1): 1–33. Gomi, T., Sidle, R.C., and Richardson, J.S. 2002. Understanding processes and downstream linkages of headwater systems. Bioscience 52(10): 905–916. doi:10.1641/0006-3568(2002)052[0905:UPADLO]2.0.CO;2. Gomi, T., Moore, R.D., and Dhakal, A.S. 2006. Headwater stream temperature response to clear-145  cut harvesting with different riparian treatments, coastal British Columbia, Canada. Water Resour. Res. 42(8): W08437. doi:10.1029/2005WR004162. González, J.M., and Graça, M.A.S. 2003. Conversion of leaf litter to secondary production by a shredding caddis-fly. Freshw. Biol. 48(9): 1578–1592. doi:10.1046/j.1365-2427.2003.01110.x. Gordon, N.D., McMahon, T.A., Finlayson, B.L., Gippel, C.J., and Nathan, R.J. 2004. Stream hydrology: an introduction for ecologists. In 2nd edition. John Wiley & Sons Ltd, West Sussex, England. Gotelli, N.J., and Colwell, R.K. 2001. Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecol. Lett. 4(4): 379–391. doi:10.1046/j.1461-0248.2001.00230.x. Gotelli, N.J., and Graves, G.R. 1996. Species diversity. In Null Models in Ecology. Smithsonian Institution Press, Washington, D.C., USA. pp. 21–46. Göthe, E., Lepori, F., and Malmqvist, B. 2009. Forestry affects food webs in northern Swedish coastal streams. Fundam. Appl. Limnol. für Hydrobiol. 175(4): 281–294. doi:10.1127/1863-9135/2009/0175-0281. Graça, M.A.S. 2001. The role of invertebrates on leaf litter decomposition in stream – a review. Int. Rev. Hydrobiol. 86(4–5): 383–393. doi:10.1002/1522-2632(200107)86:4/5<383::AID-IROH383>3.0.CO;2-D. Graça, M.A.S., Ferreira, V., Canhoto, C., Encalada, A.C., Guerrero-Bolaño, F., Wantzen, K.M., et al. 2015. A conceptual model of litter breakdown in low order streams. Int. Rev. Hydrobiol. 100(1): 1–12. doi:10.1002/iroh.201401757. Grant, G.E., Lewis, S.L., Swanson, F.J., Cissel, J.H., and McDonnell, J.J. 2008. Effects of forest practices on peak flows and consequent channel response: a state-of-science report for western Oregon and Washington. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, Portland, Oregon. Gravelle, J.A., Link, T.E., Broglio, J.R., and Braatne, J.H. 2009. Effects of timber harvest on aquatic macroinvertebrate community composition in a northern Idaho watershed. For. Sci. 55(4): 352–366. doi:10.1093/forestscience/55.4.352. Greig, H.S., Kratina, P., Thompson, P.L., Palen, W.J., Richardson, J.S., and Shurin, J.B. 2011. 146  Warming, eutrophication, and predator loss amplify subsidies between aquatic and terrestrial ecosystems. Glob. Chang. Biol. 18(2): 504–514. doi:10.1111/j.1365-2486.2011.02540.x. Griffith, M.B., and Perry, S.A. 1991. Leaf pack processing in two Appalachian mountain streams draining catchments with different management histories. Hydrobiologia 220(3): 247–254. doi:10.1007/BF00006580. Grimm, N.B., and Fisher, S.G. 1989. Stability of periphyton and macroinvertebrates to disturbance by flash floods in a desert stream. J. North Am. Benthol. Soc. 8(4): 293–307. doi:10.2307/1467493. Grimm, V., and Wissel, C. 1997. Babel, or the ecological stability discussions: an inventory and analysis of terminology and a guide for avoiding confusion. Oecologia 109(3): 323–334. doi:10.2307/4221528. Groffman, P.M., Baron, J.S., Blett, T., Gold, A., Goodman, I., Gunderson, L.H., et al. 2006. Ecological thresholds: the key to successful environmental management or an important concept with no practical application? Ecosystems 9(1): 1–13. doi:10.1007/s10021-003-0142-z. Groom, J.D., Madsen, L.J., Jones, J.E., and Giovanini, J.N. 2018. Informing changes to riparian forestry rules with a Bayesian hierarchical model. For. Ecol. Manage. 419–420: 17–30. doi:10.1016/j.foreco.2018.03.014. Guenther, S.M., Gomi, T., and Moore, R.D. 2014. Stream and bed temperature variability in a coastal headwater catchment: influences of surface-subsurface interactions and partial-retention forest harvesting. Hydrol. Process. 28(3): 1238–1249. doi:10.1002/hyp.9673. Guillemette, F., Plamondon, A.P., Prévost, M., and Lévesque, D. 2005. Rainfall generated stormflow response to clearcutting a boreal forest: peak flow comparison with 50 world-wide basin studies. J. Hydrol. 302(1–4): 137–153. doi:10.1016/j.jhydrol.2004.06.043. Gulis, V., and Suberkropp, K. 2003. Leaf litter decomposition and microbial activity in nutrient-enriched and unaltered reaches of a headwater stream. Freshw. Biol. 48(1): 123–134. doi:10.1046/j.1365-2427.2003.00985.x. Gunderson, L.H. 2000. Ecological resilience – in theory and application. Annu. Rev. Ecol. Syst. 31(1): 425–439. doi:10.1146/annurev.ecolsys.31.1.425. 147  Gurtz, M.E., Marzolf, G.R., Killingbeck, K.T., Smith, D.L., and McArthur, J.V. 1988. Hydrologic and riparian influences on the import and storage of coarse particulate organic matter in a Prairie stream. Can. J. Fish. Aquat. Sci. 45(4): 655–665. doi:10.1139/f88-079. Hall, R.O., Wallace, J.B., and Eggert, S.L. 2000. Organic matter flow in stream food webs with reduced detrital resource base. Ecology 81(12): 3445–3463. doi:10.1890/0012-9658(2000)081[3445:OMFISF]2.0.CO;2. Harr, R.D., Harper, W.C., Krygier, J.T., and Hsieh, F.S. 1975. Changes in storm hydrographs after road building and clear-cutting in the Oregon Coast Range. Water Resour. Res. 11(3): 436–444. doi:10.1029/WR011i003p00436. Harris, D.D. 1977. Hydrologic changes after logging in two small Oregon coastal watersheds. Geological Survey Water-Supply Paper 2037. U.S. Geological Survey, Washington, D.C. Harvey, B.C., and Railsback, S.F. 2011. Estimating multi-factor cumulative watershed effects on fish populations with an individual-based model. Fisheries 32(6): 292–298. doi:10.1577/1548-8446(2007)32[292:EFCWEO]2.0.CO;2. Hawkins, C.P., Olson, J.R., and Hill, R.A. 2010. The reference condition: predicting benchmarks for ecological and water-quality assessments. J. North Am. Benthol. Soc. 29(1): 312–343. doi:10.1899/09-092.1. Hawley, R.J., Wooten, M.S., MacMannis, K.R., and Fet, E. V. 2016. When do macroinvertebrate communities of reference streams resemble urban streams? The biological relevance of Qcritical. Freshw. Sci. 35(3): 778–794. doi:10.1086/687808. Hazlett, P., Broad, K., Gordon, A., Sibley, P., Buttle, J., and Larmer, D. 2008. The importance of catchment slope to soil water N and C concentrations in riparian zones: implications for riparian buffer width. Can. J. For. Res. 38(1): 16–30. doi:10.1139/X07-146. Hemstad, N.A., Merten, E.C., and Newman, R.M. 2008. Effects of riparian forest thinning by two types of mechanical harvest on stream fish and habitat in northern Minnesota. Can. J. For. Res. 38(2): 247–256. doi:10.1139/X07-157. Henry, N. 1998. Overview of the Caspar Creek watershed study. In Proceedings of the Conference on Coastal Watersheds: The Caspar Creek Story. Edited by R.R. Ziemer. U.S. Department of Agriculture, Forest Service, Pacific Southwest Research Station, Albany, CA, USA. pp. 1–9. 148  Herlihy, A.T., Gerth, W.J., Li, J., and Banks, J.L. 2005. Macroinvertebrate community response to natural and forest harvest gradients in western Oregon headwater streams. Freshw. Biol. 50(5): 905–919. doi:10.1111/j.1365-2427.2005.01363.x. Hetherington, E.D. 1982. A first look at logging effects on the hydrologic regime of Carnation Creek experimental watershed. In Proceedings of the Carnation Creek Workshop: A 10-Year Review. Edited by G. Hartman. Pacific Biological Station, Nanaimo, British Columbia, Canada. pp. 45–62. Hieber, M., and Gessner, M.O. 2002. Contribution of stream detritivores, fungi, and bacteria to leaf breakdown based on biomass estimates. Ecology 83(4): 1026–1038. doi:10.1890/0012-9658(2002)083[1026:COSDFA]2.0.CO;2. Hilderbrand, R.H., and Utz, R.M. 2015. Ecological thresholds and resilience in streams. In Rivers – Physical, Fluvial and Environmental Processes. Edited by P. Rowiński and A. Radecki-Pawlik. Springer International Publishing, Cham. pp. 461–478. Hillebrand, H., Langenheder, S., Lebret, K., Lindström, E., Östman, Ö., and Striebel, M. 2018. Decomposing multiple dimensions of stability in global change experiments. Ecol. Lett. 21(1): 21–30. doi:10.1111/ele.12867. Hirota, M., Holmgren, M., Van Nes, E.H., and Scheffer, M. 2011. Global resilience of tropical forest and savanna to critical transitions. Science 334(6053): 232–235. doi:10.1126/science.1210657. Hisabae, M., Sone, S., and Inoue, M. 2011. Breakdown and macroinvertebrate colonization of needle and leaf litter in conifer plantation streams in Shikoku, southwestern Japan. J. For. Res. 16(2): 108–115. doi:10.1007/s10310-010-0210-0. Hladyz, S., Gessner, M.O., Giller, P.S., Pozo, J., and Woodward, G. 2009. Resource quality and stoichiometric constraints on stream ecosystem functioning. Freshw. Biol. 54(5): 957–970. doi:10.1111/j.1365-2427.2008.02138.x. Hladyz, S., Tiegs, S.D., Gessner, M.O., Giller, P.S., Rîşnoveanu, G., Preda, E., et al. 2010. Leaf-litter breakdown in pasture and deciduous woodland streams: a comparison among three European regions. Freshw. Biol. 55(9): 1916–1929. doi:10.1111/j.1365-2427.2010.02426.x. Hodgson, D., McDonald, J.L., and Hosken, D.J. 2015. What do you mean, ‘resilient’? Trends Ecol. Evol. 30(9): 503–506. doi:10.1016/j.tree.2015.06.010. 149  Hofer, N., and Richardson, J.S. 2007. Comparisons of the colonisation by invertebrates of three species of wood, alder leaves, and plastic “leaves” in a temperate stream. Int. Rev. Hydrobiol. 92(6): 647–655. doi:10.1002/iroh.200610979. Hollnagel, E. 2010. Prologue: the scope of resilience engineering. In Resilience Engineering in Practice. A Guidebook. Edited by E. Hollnagel, J. Pariès, D. Woods, and J. Wreathall. Ashgate, Farnham, Surrey, UK. pp. 29–35. Holm, S. 1979. A simple sequentially rejective multiple test procedure. Scand. J. Stat. 6(2): 65–70. doi:10.2307/4615733. Hood, S.M., Baker, S., and Sala, A. 2016. Fortifying the forest: thinning and burning increase resistance to a bark beetle outbreak and promote forest resilience. Ecol. Appl. 26(7): 1984–2000. doi:10.1002/eap.1363. Hoover, T.M., Richardson, J.S., and Yonemitsu, N. 2006. Flow-substrate interactions create and mediate leaf litter resource patches in streams. Freshw. Biol. 51(3): 435–447. doi:10.1111/j.1365-2427.2005.01499.x. Hoover, T.M., Marczak, L.B., Richardson, J.S., and Yonemitsu, N. 2010. Transport and settlement of organic matter in small streams. Freshw. Biol. 55(2): 436–449. doi:10.1111/j.1365-2427.2009.02292.x. Hoover, T.M., Pinto, X., and Richardson, J.S. 2011. Riparian canopy type, management history, and successional stage control fluxes of plant litter to streams. Can. J. For. Res. 41(7): 1394–1404. doi:10.1139/x11-067. Hudson, R.O. 2001. Roberts Creek Study Forest: preliminary effects of partial harvesting on peak streamflow in two S6 Creeks. For. Res. Ext. Note EN-007: 1–9. Huey, R.B. 1987. Reply to Stearns: some acynical advice for graduate students. Bull. Ecol. Soc. Am. 68(2): 150–153. doi:10.2307/20166581. Huxel, G.R., and McCann, K. 1998. Food web stability: the influence of trophic flows across habitats. Am. Nat. 152(3): 460–469. doi:10.1086/286182. Ice, G.G., Schilling, E., and Vowell, J. 2010. Trends for forestry best management practices implementation. J. For. 108(6): 267–273. doi:10.1093/jof/108.6.267. Imberger, S.J., Walsh, C.J., Tsyrlin, E., Kerr, D.G., and Tewman, M. 2016. Variability in the response of amphipods and macroinvertebrate assemblage structure to prolonged drought in 150  forested upland streams. Biodivers. Conserv. 25(8): 1465–1480. doi:10.1007/s10531-016-1119-5. Ingrisch, J., and Bahn, M. 2018. Towards a comparable quantification of resilience. Trends Ecol. Evol. 33(4): 251–259. doi:10.1016/j.tree.2018.01.013. Ishikawa, N.F., Togashi, H., Kato, Y., Yoshimura, M., Kohmatsu, Y., Yoshimizu, C., et al. 2016. Terrestrial–aquatic linkage in stream food webs along a forest chronosequence: multi-isotopic evidence. Ecology 97(5): 1146–1158. doi:10.1890/15-1133.1. Jackrel, S.L., and Wootton, J.T. 2014. Local adaptation of stream communities to intraspecific variation in a terrestrial ecosystem subsidy. Ecology 95(1): 37–43. doi:10.1890/13-0804.1. Jackrel, S.L., and Wootton, J.T. 2015. Diversity of riparian plants among and within species shapes river communities. PLoS One 10(11): e0142362. doi:10.1371/journal.pone.0142362. Jackrel, S.L., Morton, T.C., and Wootton, J.T. 2016. Intraspecific leaf chemistry drives locally accelerated ecosystem function in aquatic and terrestrial communities. Ecology 97(8): 2125–2135. doi:10.1890/15-1763.1. Jackson, C.R., Batzer, D.P., Cross, S.S., Haggerty, S.M., and Sturm, C.A. 2007. Headwater streams and timber harvest: channel, macroinvertebrate, and amphibian response and recovery. pp. 356–370. Jackson, C.R., Webster, J.R., Knoepp, J.D., Elliott, K.J., Emanuel, R.E., Caldwell, P. V, et al. 2018. Unexpected ecological advances made possible by long-term data: a Coweeta example. Wiley Interdiscip. Rev. Water 5(2): e1273. doi:10.1002/wat2.1273. Jackson, M.C., Loewen, C.J.G., Vinebrooke, R.D., and Chimimba, C.T. 2015. Net effects of multiple stressors in freshwater ecosystems: a meta-analysis. Glob. Chang. Biol. 22(1): 180–189. doi:10.1111/gcb.13028. Jeanette, S., and Michael, M. 2011. Leaf decay processes during and after a supra-seasonal hydrological drought in a temperate lowland stream. Int. Rev. Hydrobiol. 96(6): 633–655. doi:10.1002/iroh.201111322. Johnson, S.L., and Jones, J.A. 2000. Stream temperature responses to forest harvest and debris flows in western Cascades, Oregon. Can. J. Fish. Aquat. Sci. 57: 30–39. Jones, J.A. 2000. Hydrologic processes and peak discharge response to forest removal, regrowth, and roads in 10 small experimental basins, Western Cascades, Oregon. Water Resour. Res. 151  36(9): 2621–2642. doi:10.1029/2000WR900105. Jonsson, M., Malmqvist, B., and Hoffsten, P.-O. 2001. Leaf litter breakdown rates in boreal streams: does shredder species richness matter? Freshw. Biol. 46(2): 161–171. doi:10.1046/j.1365-2427.2001.00655.x. Judd, K.E., Crump, B.C., and Kling, G.W. 2006. Variation in dissolved organic matter controls bacterial production and community composition. Ecology 87(8): 2068–2079. doi:10.1890/0012-9658(2006)87[2068:VIDOMC]2.0.CO;2. Kagata, H., and Ohgushi, T. 2013. Home-field advantage in decomposition of leaf litter and insect frass. Popul. Ecol. 55(1): 69–76. doi:10.1007/s10144-012-0342-5. Kakouei, K., Kiesel, J., Kail, J., Pusch, M., and Jähnig, S.C. 2017. Quantitative hydrological preferences of benthic stream invertebrates in Germany. Ecol. Indic. 79(Supplement C): 163–172. doi:10.1016/j.ecolind.2017.04.029. Kara, F., Kalin, L., and Loewenstein, E.F. 2014. Sediment dynamics following active management of a stream buffer zone: a case study. J. Soil Water Conserv. 69(2): 87–94. doi:10.2489/jswc.69.2.87. Karlsson, O.M., Richardson, J.S., and Kiffney, P.M. 2005. Modelling organic matter dynamics in headwater streams of south-western British Columbia, Canada. Ecol. Modell. 183(4): 463–476. doi:10.1016/j.ecolmodel.2004.08.022. Karr, J.R. 1993. Defining and assessing ecological integrity: beyond water quality. Environ. Toxicol. Chem. 12(9): 1521–1531. doi:10.1002/etc.5620120902. Katsuyama, M., Shibata, H., Yoshioka, T., Yoshida, T., Ogawa, A., and Ohte, N. 2009. Applications of a hydro-biogeochemical model and long-term simulations of the effects of logging in forested watersheds. Sustain. Sci. 4(2): 179. doi:10.1007/s11625-009-0079-z. Kaylor, M.J., Warren, D.R., and Kiffney, P.M. 2016. Long-term effects of riparian forest harvest on light in Pacific Northwest (USA) streams. Freshw. Sci. 36(1): 1–13. doi:10.1086/690624. Kennedy, K.T.M., and El-Sabaawi, R.W. 2017. A global meta-analysis of exotic versus native leaf decay in stream ecosystems. Freshw. Biol. 62(6): 977–989. doi:10.1111/fwb.12918. Kibler, K.M., Skaugset, A., Ganio, L.M., and Huso, M.M. 2013. Effect of contemporary forest harvesting practices on headwater stream temperatures: initial response of the Hinkle Creek 152  catchment, Pacific Northwest, USA. For. Ecol. Manage. 310: 680–691. doi:10.1016/j.foreco.2013.09.009. Kiffney, P.M., and Richardson, J.S. 2010. Organic matter inputs into headwater streams of southwestern British Columbia as a function of riparian reserves and time since harvesting. For. Ecol. Manage. 260(11): 1931–1942. doi:10.1016/j.foreco.2010.08.016. Kiffney, P.M., Richardson, J.S., and Feller, M.C. 2000. Fluvial and epilithic organic matter dynamics in headwater streams of southwestern British Columbia, Canada. Arch. fur Hydrobiol. 149(1): 109–129. doi:10.1127/archiv-hydrobiol/149/2000/109. Kiffney, P.M., Bull, J.P., and Feller, M.C. 2002. Climatic and hydrologic variability in a coastal watershed of southwestern British Columbia. J. Am. Water Resour. Assoc. 38(5): 1437–1451. doi:10.1111/j.1752-1688.2002.tb04357.x. Kiffney, P.M., Richardson, J.S., and Bull, J.P. 2004. Establishing light as a causal mechanism structuring stream communities in response to experimental manipulation of riparian buffer width. J. North Am. Benthol. Soc. 23(3): 542–555. doi:10.1899/0887-3593(2004)023<0542:ELAACM>2.0.CO;2. Kitchell, J.F., Stewart, D.J., and Weininger, D. 1977. Applications of a bioenergetics model to yellow perch (Perca flavescens) and walleye (Stizostedion vitreum vitreum). J. Fish. Res. Board Canada 34(10): 1922–1935. doi:10.1139/f77-258. Knowlton, N. 1992. Thresholds and multiple stable states in coral reef community dynamics. Am. Zool. 32(6): 674–682. doi:10.1093/icb/32.6.674. Kobayashi, S., Gomi, T., Sidle, R.C., and Takemon, Y. 2010. Disturbances structuring macroinvertebrate communities in steep headwater streams: relative importance of forest clearcutting and debris flow occurrence. Can. J. Fish. Aquat. Sci. 67(2): 427–444. doi:10.1139/F09-186. Kominoski, J.S., and Rosemond, A.D. 2011. Conservation from the bottom up: forecasting effects of global change on dynamics of organic matter and management needs for river networks. Freshw. Sci. 31(1): 51–68. doi:10.1899/10-160.1. Kominoski, J.S., Marczak, L.B., and Richardson, J.S. 2011. Riparian forest composition affects stream litter decomposition despite similar microbial and invertebrate communities. Ecology 92(1): 151–159. doi:10.1890/10-0028.1. 153  Kominoski, J.S., Larrañaga, S., and Richardson, J.S. 2012. Invertebrate feeding and emergence timing vary among streams along a gradient of riparian forest composition. Freshw. Biol. 57(4): 759–772. doi:10.1111/j.1365-2427.2012.02740.x. Kominoski, J.S., Shah, J.J.F., Canhoto, C., Fischer, D.G., Giling, D.P., González, E., et al. 2013. Forecasting functional implications of global changes in riparian plant communities. Front. Ecol. Environ. 11(8): 423–432. doi:10.1890/120056. Kreutzweiser, D., Capell, S., Good, K., and Holmes, S. 2009a. Sediment deposition in streams adjacent to upland clearcuts and partially harvested riparian buffers in boreal forest catchments. For. Ecol. Manage. 258(7): 1578–1585. doi:10.1016/j.foreco.2009.07.005. Kreutzweiser, D., Muto, E., Holmes, S., and Gunn, J. 2010. Effects of upland clearcutting and riparian partial harvesting on leaf pack breakdown and aquatic invertebrates in boreal forest streams. Freshw. Biol. 55(11): 2238–2252. doi:10.1111/j.1365-2427.2010.02410.x. Kreutzweiser, D.P., Capell, S.S., and Beall, F.D. 2004. Effects of selective forest harvesting on organic matter inputs and accumulation in headwater streams. North. J. Appl. For. 21(1): 19–30. doi:10.1093/njaf/21.1.19. Kreutzweiser, D.P., Capell, S.S., and Good, K.P. 2005a. Effects of fine sediment inputs from a logging road on stream insect communities: a large-scale experimental approach in a Canadian headwater stream. Aquat. Ecol. 39(1): 55–66. doi:10.1007/s10452-004-5066-y. Kreutzweiser, D.P., Capell, S.S., and Good, K.P. 2005b. Macroinvertebrate community responses to selection logging in riparian and upland areas of headwater catchments in a northern hardwood forest. J. North Am. Benthol. Soc. 24(1): 208–222. doi:10.1899/0887-3593(2005)024<0208:MCRTSL>2.0.CO;2. Kreutzweiser, D.P., Good, K.P., Capell, S.S., and Holmes, S.B. 2008a. Leaf-litter decomposition and macroinvertebrate communities in boreal forest streams linked to upland logging disturbance. J. North Am. Benthol. Soc. 27(1): 1–15. doi:10.1899/07-034R.1. Kreutzweiser, D.P., Hazlett, P.W., and Gunn, J.M. 2008b. Logging impacts on the biogeochemistry of boreal forest soils and nutrient export to aquatic systems: a review. Environ. Rev. 16: 157–179. doi:10.1139/A08-006. Kreutzweiser, D.P., Capell, S.S., and Holmes, S.B. 2009b. Stream temperature responses to partial-harvest logging in riparian buffers of boreal mixedwood forest watersheds. Can. J. 154  For. Res. 39(3): 497–506. doi:10.1139/X08-191. Kreutzweiser, D.P., Sibley, P.K., Richardson, J.S., and Gordon, A.M. 2012. Introduction and a theoretical basis for using disturbance by forest management activities to sustain aquatic ecosystems. Freshw. Sci. 31(1): 224–231. doi:10.1899/11-114.1. Kuehn, K.A., Francoeur, S.N., Findlay, R.H., and Neely, R.K. 2014. Priming in the microbial landscape: periphytic algal stimulation of litter-associated microbial decomposers. Ecology 95(3): 749–762. doi:10.1890/13-0430.1. Kuglerová, L., Ågren, A., Jansson, R., and Laudon, H. 2014. Towards optimizing riparian buffer zones: ecological and biogeochemical implications for forest management. For. Ecol. Manage. 334: 74–84. doi:10.1016/j.foreco.2014.08.033. Kuglerová, L., García, L., Pardo, I., Mottiar, Y., and Richardson, J.S. 2017a. Does leaf litter from invasive plants contribute the same support of a stream ecosystem function as native vegetation? Ecosphere 8(4): e01779. doi:10.1002/ecs2.1779. Kuglerová, L., Hasselquist, E.M., Richardson, J.S., Sponseller, R.A., Kreutzweiser, D.P., and Laudon, H. 2017b. Management perspectives on Aqua incognita: connectivity and cumulative effects of small natural and artificial streams in boreal forests. Hydrol. Process. 31(23): 4238–4244. doi:10.1002/hyp.11281. Lagrue, C., Kominoski, J.S., Danger, M., Baudoin, J., Lamothe, S., Lambrigot, D., et al. 2011. Experimental shading alters leaf litter breakdown in streams of contrasting riparian canopy cover. Freshw. Biol. 56(10): 2059–2069. doi:10.1111/j.1365-2427.2011.02637.x. Lake, P.S. 2000. Disturbance, patchiness, and diversity in streams. J. North Am. Benthol. Soc. 19(4): 573–592. doi:10.2307/1468118. Lake, P.S. 2013. Resistance, resilience and restoration. Ecol. Manag. Restor. 14(1): 20–24. doi:10.1111/emr.12016. Landres, P.B., Morgan, P., and Swanson, F.J. 1999. Overview of the use of natural variability concepts in managing ecological systems. Ecol. Appl. 9(4): 1179–1188. doi:10.1890/1051-0761(1999)009[1179:OOTUON]2.0.CO;2. Law, J. 2013. IHA: Indicators of Hydrologic Alteration. R package version 0.2.41. http://rdrr.io/rforge/IHA/. Leach, J.A., and Moore, D. 2017. Insights on stream temperature processes through development 155  of a coupled hydrologic and stream temperature model for forested coastal headwater catchments. Hydrol. Process. 31(18): 3160–3177. doi:10.1002/hyp.11190. Leach, J.A., and Moore, R.D. 2015. Observations and modeling of hillslope throughflow temperatures in a coastal forested catchment. Water Resour. Res. 51(5): 3770–3795. doi:10.1002/2014WR016763. Leach, J.A., Moore, R.D., Hinch, S.G., and Gomi, T. 2012. Estimation of forest harvesting-induced stream temperature changes and bioenergetic consequences for cutthroat trout in a coastal stream in British Columbia, Canada. Aquat. Sci. 74(3): 427–441. doi:10.1007/s00027-011-0238-z. Leberfinger, K., Bohman, I., and Herrmann, J. 2010. Drought impact on stream detritivores: experimental effects on leaf litter breakdown and life cycles. Hydrobiologia 652(1): 247–254. doi:10.1007/s10750-010-0337-1. Lecerf, A. 2017. Methods for estimating the effect of litterbag mesh size on decomposition. Ecol. Modell. 362: 65–68. doi:10.1016/j.ecolmodel.2017.08.011. Lecerf, A., and Chauvet, E. 2008. Intraspecific variability in leaf traits strongly affects alder leaf decomposition in a stream. Basic Appl. Ecol. 9(5): 598–605. doi:10.1016/j.baae.2007.11.003. Lecerf, A., and Richardson, J.S. 2010. Litter decomposition can detect effects of high and moderate levels of forest disturbance on stream condition. For. Ecol. Manage. 259(12): 2433–2443. doi:10.1016/j.foreco.2010.03.022. Lecerf, A., and Richardson, J.S. 2011. Assessing the functional importance of large-bodied invertebrates in experimental headwater streams. Oikos 120(6): 950–960. doi:10.1111/j.1600-0706.2010.18942.x. Lecerf, A., Baudoin, J.-M., Besson, A.A., Lamothe, S., and Lagrue, C. 2012. Is smaller necessarily better? Effects of small-scale forest harvesting on stream ecosystems. Ann. Limnol. - Int. J. Lim. 48(4): 401–409. doi:10.1051/limn/2012028. Lecerf, A., Evangelista, C., Cucherousset, J., and Boiché, A. 2016. Riparian overstory–understory interactions and their potential implications for forest-stream linkages. For. Ecol. Manage. 367: 112–119. doi:10.1016/j.foreco.2016.02.031. Lee, P., Smyth, C., and Boutin, S. 2004. Quantitative review of riparian buffer width guidelines 156  from Canada and the United States. J. Environ. Manage. 70(2): 165–180. doi:10.1016/j.jenvman.2003.11.009. Leroy, C., Corbara, B., Dézerald, O., Trzcinski, M.K., Carrias, J.-F., Dejean, A., et al. 2017. What drives detrital decomposition in neotropical tank bromeliads? Hydrobiologia 802(1): 85–95. doi:10.1007/s10750-017-3242-z. Lewis, J., Mori, S.R., Keppeler, E.T., and Ziemer, R.R. 2001. Impacts of logging on storm peak flows, flow volumes and suspended sediment loads in Caspar Creek, California. In Land Use and Watersheds: Human Influence on Hydrology and Geomorphology in Urban and Forest Areas Volume 2. Edited by M.S. Wigmosta and S.J. Burges. American Geophysical Union. pp. 85–125. Lidman, J., Jonsson, M., Burrows, R.M., Bundschuh, M., and Sponseller, R.A. 2017. Composition of riparian litter input regulates organic matter decomposition: implications for headwater stream functioning in a managed forest landscape. Ecol. Evol. 7(4): 1068–1077. doi:10.1002/ece3.2726. Lin, D., Pang, M., Fanin, N., Wang, H., Qian, S., Zhao, L., et al. 2018. Fungi participate in driving home-field advantage of litter decomposition in a subtropical forest. Plant Soil. doi:10.1007/s11104-018-3865-5. Lindenmayer, D.B., and Laurance, W.F. 2012. A history of hubris – cautionary lessons in ecologically sustainable forest management. Biol. Conserv. 151(1): 11–16. doi:10.1016/j.biocon.2011.10.032. Long, J.N. 2009. Emulating natural disturbance regimes as a basis for forest management: a North American view. For. Ecol. Manage. 257(9): 1868–1873. doi:10.1016/j.foreco.2008.12.019. Luke, S.G. 2017. Evaluating significance in linear mixed-effects models in R. Behav. Res. Methods 49(4): 1494–1502. doi:10.3758/s13428-016-0809-y. Macdonald, J.S., MacIsaac, E.A., and Herunter, H.E. 2003. The effect of variable-retention riparian buffer zones on water temperatures in small headwater streams in sub-boreal forest ecosystems of British Columbia. Can. J. For. Res. 33(8): 1371–1382. doi:10.1139/x03-015. MacDonald, L.H., and Coe, D. 2007. Influence of headwater streams on downstream reaches in forested areas. For. Sci. 53(2): 148–168. doi:10.1093/forestscience/53.2.148. 157  Majdi, N., Traunspurger, W., Richardson, J.S., and Lecerf, A. 2015. Small stonefly predators affect microbenthic and meiobenthic communities in stream leaf packs. Freshw. Biol. 60(9): 1930–1943. doi:10.1111/fwb.12622. Makarenkov, V., and Legendre, P. 2002. Nonlinear redundancy analysis and canonical correspondence analysis based on polynomial regression. Ecology 83(4): 1146–1161. doi:10.1890/0012-9658(2002)083[1146:NRAACC]2.0.CO;2. Manga, M. 1996. Hydrology of spring-dominated streams in the Oregon Cascades. Water Resour. Res. 32(8): 2435–2439. doi:10.1029/96WR01238. Marcarelli, A.M., Baxter, C. V, Mineau, M.M., and Hall, R.O. 2011. Quantity and quality: unifying food web and ecosystem perspectives on the role of resource subsidies in freshwaters. Ecology 92(6): 1215–1225. doi:10.1890/10-2240.1. Marczak, L.B., Sakamaki, T., Turvey, S.L., Deguise, I., Wood, S.L.R., and Richardson, J.S. 2010. Are forested buffers an effective conservation strategy for riparian fauna? An assessment using meta-analysis. Ecol. Appl. 20(1): 126–134. doi:10.1890/08-2064.1. Martel, N., Rodríguez, M.A., and Bérubé, P. 2007. Multi-scale analysis of responses of stream macrobenthos to forestry activities and environmental context. Freshw. Biol. 52(1): 85–97. doi:10.1111/j.1365-2427.2006.01674.x. Martí, E., Grimm, N.B., and Fisher, S.G. 1997. Pre- and post-flood retention efficiency of nitrogen in a Sonoran desert stream. J. North Am. Benthol. Soc. 16(4): 805–819. doi:10.2307/1468173. Martínez, A., Pérez, J., Molinero, J., Sagarduy, M., and Pozo, J. 2015. Effects of flow scarcity on leaf-litter processing under oceanic climate conditions in calcareous streams. Sci. Total Environ. 503(Supplement C): 251–257. doi:10.1016/j.scitotenv.2014.06.018. Mazor, R.D., Purcell, A.H., and Resh, V.H. 2009. Long-term variability in bioassessments: a twenty-year study from two northern California streams. Environ. Manage. 43(6): 1269–1286. doi:10.1007/s00267-009-9294-8. McArthur, M.D., and Richardson, J.S. 2002. Microbial utilization of dissolved organic carbon leached from riparian litterfall. Can. J. Fish. Aquat. Sci. 59(10): 1668–1676. doi:10.1139/f02-135. McBroom, M.W., Beasley, R.S., Chang, M., Gowin, B., and Ice, G.G. 2003. Runoff and 158  sediment losses from annual and unusual storm events from the Alto Experimental Watersheds, Texas: 23 years after silvicultural treatments. In First Interagency Conference on Research in the Watersheds, October 27-30, 2003. Edited by K.G. Renard, S.A. McElroy, W.J. Gburek, H.E. Canfield, and R.L. Scott. U.S. Department of Agriculture, Agricultural Research Service. pp. 607–613. McCluney, K.E., Poff, N.L., Palmer, M.A., Thorp, J.H., Poole, G.C., Williams, B.S., et al. 2014. Riverine macrosystems ecology: sensitivity, resistance, and resilience of whole river basins with human alterations. Front. Ecol. Environ. 12(1): 48–58. doi:10.1890/120367. McDermott, C., Cashore, B.W., and Kanowski, P. 2010. Global environmental forest policies : an international comparison. Earthscan, Washington, DC. McKie, B.G., and Malmqvist, B. 2009. Assessing ecosystem functioning in streams affected by forest management: increased leaf decomposition occurs without changes to the composition of benthic assemblages. Freshw. Biol. 54(10): 2086–2100. doi:10.1111/j.1365-2427.2008.02150.x. McMullen, L.E., and Lytle, D.A. 2012. Quantifying invertebrate resistance to floods: a global-scale meta-analysis. Ecol. Appl. 22(8): 2164–2175. doi:10.1890/11-1650.1. Medhurst, R.B., Wipfli, M.S., Binckley, C., Polivka, K., Hessburg, P.F., and Salter, R.B. 2010. Headwater streams and forest management: does ecoregional context influence logging effects on benthic communities? Hydrobiologia 641(1): 71–83. doi:10.1007/s10750-009-0058-5. Mellina, E., and Hinch, S.G. 2009. Influences of riparian logging and in-stream large wood removal on pool habitat and salmonid density and biomass: a meta-analysis. Can. J. For. Res. 39(7): 1280–1301. doi:10.1139/X09-037. Melody, K.J., and Richardson, J.S. 2004. Responses of invertebrates and algae of a boreal coniferous forest stream to experimental manipulation of leaf litter inputs and shading. Hydrobiologia 519(1): 197–206. doi:10.1023/B:HYDR.0000026506.16929.e1. Merritt, R.W., Cummins, K.W., and Berg, M.B. 2008. An Introduction to the Aquatic Insects of North America. Kendall/Hunt Publishing Company, Dubuque, Iowa, USA. Meyer, J.L., Strayer, D.L., Wallace, J.B., Eggert, S.L., Helfman, G.S., and Leonard, N.E. 2007. The contribution of headwater streams to biodiversity in river networks. JAWRA J. Am. 159  Water Resour. Assoc. 43(1): 86–103. doi:10.1111/j.1752-1688.2007.00008.x. Mielikäinen, K., and Hynynen, J. 2003. Silvicultural management in maintaining biodiversity and resistance of forests in Europe–boreal zone: case Finland. J. Environ. Manage. 67(1): 47–54. doi:10.1016/S0301-4797(02)00187-1. Milcu, A., and Manning, P. 2011. All size classes of soil fauna and litter quality control the acceleration of litter decay in its home environment. Oikos 120(9): 1366–1370. doi:10.1111/j.1600-0706.2010.19418.x. Moghadam, F.S., and Zimmer, M. 2014. Effects of warming and nutrient enrichment on how grazing pressure affects leaf litter-colonizing bacteria. J. Environ. Qual. 43: 851–858. doi:10.2134/jeq2013.12.0503. Moline, A.B., and Poff, N.L. 2008. Growth of an invertebrate shredder on native (Populus) and non-native (Tamarix, Elaeagnus) leaf litter. Freshw. Biol. 53(5): 1012–1020. doi:10.1111/j.1365-2427.2008.01960.x. Mollá, S., Casas, J.J., Menéndez, M., Basaguren, A., Casado, C., Descals, E., et al. 2017. Leaf-litter breakdown as an indicator of the impacts by flow regulation in headwater streams: responses across climatic regions. Ecol. Indic. 73(Supplement C): 11–22. doi:10.1016/j.ecolind.2016.09.005. Moore, D.R., Spittlehouse, D.L., and Story, A. 2005a. Riparian microclimate and stream temperature response to forest harvesting: a review. J. Am. Water Resour. Assoc. 41(4): 813–834. doi:10.1111/j.1752-1688.2005.tb03772.x. Moore, R.D., and Wondzell, S.M. 2005. Physical hydrology and the effects of forest harvesting in the Pacific Northwest: a review. JAWRA J. Am. Water Resour. Assoc. 41(4): 763–784. doi:10.1111/j.1752-1688.2005.tb03770.x. Moore, R.D., Sutherland, P., Gomi, T., and Dhakal, A. 2005b. Thermal regime of a headwater stream within a clear-cut, coastal British Columbia, Canada. Hydrol. Process. 19(13): 2591–2608. doi:10.1002/hyp.5733. Mora-Gómez, J., Elosegi, A., Mas-Martí, E., and Romaní, A.M. 2015. Factors controlling seasonality in leaf-litter breakdown in a Mediterranean stream. Freshw. Sci. 34(4): 1245–1258. doi:10.1086/683120. Mora-Gómez, J., Elosegi, A., Duarte, S., Cássio, F., Pascoal, C., and Romaní, A.M. 2016. 160  Differences in the sensitivity of fungi and bacteria to season and invertebrates affect leaf litter decomposition in a Mediterranean stream. FEMS Microbiol. Ecol. 92(8): fiw121. doi:10.1093/femsec/fiw121. Moreno-Mateos, D., Barbier, E.B., Jones, P.C., Jones, H.P., Aronson, J., López-López, J.A., et al. 2017. Anthropogenic ecosystem disturbance and the recovery debt. Nat. Commun. 8: 14163. doi:10.1038/ncomms14163. Moseley, K.R., Mark Ford, W., Edwards, J.W., and Schuler, T.M. 2008. Long-term partial cutting impacts on Desmognathus salamander abundance in West Virginia headwater streams. For. Ecol. Manage. 254(2): 300–307. doi:10.1016/j.foreco.2007.03.073. Musetta-Lambert, J., Muto, E., Kreutzweiser, D., and Sibley, P. 2017. Wildfire in boreal forest catchments influences leaf litter subsidies and consumer communities in streams: implications for riparian management strategies. For. Ecol. Manage. 391: 29–41. doi:10.1016/j.foreco.2017.01.028. Muto, E.A., Kreutzweiser, D.P., and Sibley, P.K. 2009. The influence of riparian vegetation on leaf litter inputs to Boreal Shield streams: implications for partial-harvest logging in riparian reserves. Can. J. For. Res. 39(5): 917–927. doi:10.1139/X09-017. Muto, E.A., Kreutzweiser, D.P., and Sibley, P.K. 2011. Over-winter decomposition and associated macroinvertebrate communities of three deciduous leaf species in forest streams on the Canadian Boreal Shield. Hydrobiologia 658(1): 111–126. doi:10.1007/s10750-010-0455-9. Naiman, R.J., and Décamps, H. 1997. The ecology of interfaces: riparian zones. Annu. Rev. Ecol. Syst. 28(1): 621–658. doi:10.1146/annurev.ecolsys.28.1.621. Naiman, R.J., Decamps, H., and Pollock, M. 1993. The role of riparian corridors in maintaining regional biodiversity. Ecol. Appl. 3(2): 209–212. doi:10.2307/1941822. Nakagawa, S., and Cuthill, I.C. 2007. Effect size, confidence interval and statistical significance: a practical guide for biologists. Biol. Rev. 82(4): 591–605. doi:10.1111/j.1469-185X.2007.00027.x. Nakagawa, S., and Schielzeth, H. 2013. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol. Evol. 4(2): 133–142. doi:10.1111/j.2041-210x.2012.00261.x. 161  Negishi, J.N., and Richardson, J.S. 2003. Responses of organic matter and macroinvertebrates to placements of boulder clusters in a small stream of southwestern British Columbia, Canada. Can. J. Fish. Aquat. Sci. 60(3): 247–258. doi:10.1139/f03-013. Negishi, J.N., and Richardson, J.S. 2006. An experimental test of the effects of food resources and hydraulic refuge on patch colonization by stream macroinvertebrates during spates. J. Anim. Ecol. 75(1): 118–129. doi:10.1111/j.1365-2656.2005.01027.x. Nicolson, J.A. 1988. Water and chemical budgets for terrestrial basins at the Turkey Lakes Watershed. Can. J. Fish. Aquat. Sci. 45(S1): s88–s95. doi:10.1139/f88-271. Niemi, G.J., DeVore, P., Detenbeck, N., Taylor, D., Lima, A., Pastor, J., et al. 1990. Overview of case studies on recovery of aquatic systems from disturbance. Environ. Manage. 14(5): 571–587. doi:10.1007/BF02394710. Nimmo, D.G., Mac Nally, R., Cunningham, S.C., Haslem, A., and Bennett, A.F. 2015. Vive la résistance: reviving resistance for 21st century conservation. Trends Ecol. Evol. 30(9): 516–523. doi:10.1016/j.tree.2015.07.008. Northington, R.M., and Webster, J.R. 2017. Experimental reductions in stream flow alter litter processing and consumer subsidies in headwater streams. Freshw. Biol. 62(4): 737–750. doi:10.1111/fwb.12898. O’Donnell, K.M., Thompson, F.R., and Semlitsch, R.D. 2015. Prescribed fire and timber harvest effects on terrestrial salamander abundance, detectability, and microhabitat use. J. Wildl. Manage. 79(5): 766–775. doi:10.1002/jwmg.884. O’Hara, K.L., and Ramage, B.S. 2013. Silviculture in an uncertain world: utilizing multi-aged management systems to integrate disturbance. Forestry 86(4): 401–410. doi:10.1093/forestry/cpt012. O’Keefe, T.C., and Naiman, R.J. 2006. The influence of forest structure on riparian litterfall in a Pacific Coastal rain forest. Can. J. For. Res. 36(11): 2852–2863. doi:10.1139/x06-180. Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, R.P., O’Hara, R.B., et al. 2017. vegan: Community Ecology Package. R package version 2.4.4. http://cran.r-project.org/web/packages/vegan/. Olden, J.D., and Poff, N.L. 2003. Redundancy and the choice of hydrologic indices for characterizing streamflow regimes. River Res. Appl. 19(2): 101–121. doi:10.1002/rra.700. 162  Olson, D.H., and Rugger, C. 2007. Preliminary study of the effects of headwater riparian reserves with upslope thinning on stream habitats and amphibians in western Oregon. pp. 331–342. Orians, G.H. 1975. Diversity, stability and maturity in natural ecosystems. In Unifying Concepts in Ecology SE - 11. Edited by W.H. van Dobben and R.H. Lowe-McConnell. Springer Netherlands. pp. 139–150. Palozzi, J.E., and Lindo, Z. 2018. Are leaf litter and microbes team players? Interpreting home-field advantage decomposition dynamics. Soil Biol. Biochem. 124: 189–198. doi:10.1016/j.soilbio.2018.06.018. Papangelakis, E., and Hassan, M.A. 2016. The role of channel morphology on the mobility and dispersion of bed sediment in a small gravel-bed stream. Earth Surf. Process. Landforms 41(15): 2191–2206. doi:10.1002/esp.3980. Patrick, C.J., and Yuan, L.L. 2017. Modeled hydrologic metrics show links between hydrology and the functional composition of stream assemblages. Ecol. Appl. 27(5): 1605–1617. doi:10.1002/eap.1554. Pauly, D. 1995. Anecdotes and the shifting baseline syndrome of fisheries. Trends Ecol. Evol. 10(10): 430. doi:10.1016/S0169-5347(00)89171-5. Pearson, S.F., and Manuwal, D.A. 2001. Breeding bird response to riparian buffer width in managed Pacific Northwest Douglas-fir forests. Ecol. Appl. 11(3): 840–853. doi:10.1890/1051-0761(2001)011[0840:BBRTRB]2.0.CO;2. Peckarsky, B.L., McIntosh, A.R., Àlvarez, M., and Moslemi, J.M. 2015. Disturbance legacies and nutrient limitation influence interactions between grazers and algae in high elevation streams. Ecosphere 6(11): art241. doi:10.1890/ES15-00236.1. Peel, M.C., Finlayson, B.L., and McMahon, T.A. 2007. Updated world map of the Köppen-Geiger climate classification. Hydrol. Earth Syst. Sci. 11(5): 1633–1644. doi:10.5194/hess-11-1633-2007. Peres-Neto, P.R., Legendre, P., Dray, S., and Borcard, D. 2006. Variation partitioning of species data matrices: estimation and comparison of fractions. Ecology 87(10): 2614–2625. doi:10.1890/0012-9658(2006)87[2614:VPOSDM]2.0.CO;2. Perry, T.D., and Jones, J.A. 2016. Summer streamflow deficits from regenerating Douglas-fir 163  forest in the Pacific Northwest, USA. Ecohydrology 10(2): e1790. doi:10.1002/eco.1790. Peterson, B.J., Wollheim, W.M., Mulholland, P.J., Webster, J.R., Meyer, J.L., Tank, J.L., et al. 2001. Control of nitrogen export from watersheds by headwater streams. Science 292(5514): 86–90. doi:10.1126/science.1056874. Peterson, C.G., and Stevenson, R.J. 1992. Resistance and resilience of lotic algal communities: importance of disturbance timing and current. Ecology 73(4): 1445–1461. doi:10.2307/1940689. Peterson, G., Allen, C.R., and Holling, C.S. 1998. Ecological resilience, biodiversity, and scale. Ecosystems 1(1): 6–18. doi:10.1007/s100219900002. Piggott, J.J., Lange, K., Townsend, C.R., and Matthaei, C.D. 2012. Multiple stressors in agricultural streams: a mesocosm study of interactions among raised water temperature, sediment addition and nutrient enrichment. PLoS One 7(11): e49873. doi:10.1371/journal.pone.0049873. Piggott, J.J., Townsend, C.R., and Matthaei, C.D. 2015a. Climate warming and agricultural stressors interact to determine stream macroinvertebrate community dynamics. Glob. Chang. Biol. 21: 1887–1906. doi:10.1111/gcb.12861. Piggott, J.J., Townsend, C.R., and Matthaei, C.D. 2015b. Reconceptualizing synergism and antagonism among multiple stressors. Ecol. Evol. 5(7): 1538–1547. doi:10.1002/ece3.1465. Plamondon, A.P., Lévesque, D., Ma, Y., and Prévost, M. 1998. Long-term effects of forest mosaic management on storm and snowmelt flow, Quebec. In Hydrology in a Changing Environment Volume I. Edited by H. Wheater and C. Kirby. Proceedings of the British Hydrological Society International Conference, Exeter, UK. pp. 503–515. Poff, N.L. 1992. Why disturbances can be predictable: a perspective on the definition of disturbance in streams. J. North Am. Benthol. Soc. 11(1): 86–92. doi:10.2307/1467885. Poff, N.L. 2018. Beyond the natural flow regime? Broadening the hydro-ecological foundation to meet environmental flows challenges in a non-stationary world. Freshw. Biol. 63(8): 1011–1021. doi:10.1111/fwb.13038. Poff, N.L., and Ward, J. V. 1990. Physical habitat template of lotic systems: recovery in the context of historical pattern of spatiotemporal heterogeneity. Environ. Manage. 14(5): 629. doi:10.1007/BF02394714. 164  Pozo, J., González, E., Díez, J., and Elosegi, A. 1997. Leaf-litter budgets in two contrasting forested streams. Limnetica 13(2): 77–84. Pozo, J., Casas, J., Menéndez, M., Mollá, S., Arostegui, I., Basaguren, A., et al. 2011. Leaf-litter decomposition in headwater streams: a comparison of the process among four climatic regions. J. North Am. Benthol. Soc. 30(4): 935–950. doi:10.1899/10-153.1. Quinn, J.M., Phillips, N.R., and Parkyn, S.M. 2007. Factors influencing retention of coarse particulate organic matter in streams. Earth Surf. Process. Landforms 32(8): 1186–1203. doi:10.1002/esp.1547. R Development Core Team. 2018. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, http://www.r-project.org/. Radchuk, V., Laender, F. De, Cabral, J.S., Boulangeat, I., Crawford, M., Bohn, F., et al. 2019. The dimensionality of stability depends on disturbance type. Ecol. Lett. doi:10.1111/ele.13226. Reid, D.J., Quinn, J.M., and Wright-Stow, A.E. 2010. Responses of stream macroinvertebrate communities to progressive forest harvesting: influences of harvest intensity, stream size and riparian buffers. For. Ecol. Manage. 260(10): 1804–1815. doi:10.1016/j.foreco.2010.08.025. Reinhart, K.G., Eschner, A.R., and Trimble Jr., G.R. 1963. Effects on streamflow of four forest practices in the mountains of West Virginia. U.S. Department of Agriculture, Forest Service, Northeastern Forest Experimental Station. Resh, V.H., Brown, A. V, Covich, A.P., Gurtz, M.E., Li, H.W., Minshall, G.W., et al. 1988. The role of disturbance in stream ecology. J. North Am. Benthol. Soc. 7(4): 433–455. doi:10.2307/1467300. Richardson, J.S. 1991. Seasonal food limitation of detritivores in a montane stream: an experimental test. Ecology 72(3): 873–887. doi:10.2307/1940589. Richardson, J.S. 1992. Coarse particulate detritus dynamics in small, montane streams of southwestern British Columbia. Can. J. Fish. Aquat. Sci. 49(2): 337–346. doi:10.1139/f92-038. Richardson, J.S. 2001. Life cycle phenology of common detritivores from a temperate rainforest stream. Hydrobiologia 455(1–3): 87–95. doi:10.1023/A:1011943532162. 165  Richardson, J.S. 2008. Aquatic arthropods and forestry: effects of large-scale land use on aquatic systems in Nearctic temperate regions. Can. Entomol. 140(04): 495–509. Richardson, J.S., and Béraud, S. 2014. Effects of riparian forest harvest on streams: a meta-analysis. J. Appl. Ecol. 51(6): 1712–1721. doi:10.1111/1365-2664.12332. Richardson, J.S., and Neill, W.E. 1991. Indirect effects of detritus manipulations in a montane stream. Can. J. Fish. Aquat. Sci. 48(5): 776–783. doi:10.1139/f91-093. Richardson, J.S., and Sato, T. 2015. Resource subsidy flows across freshwater–terrestrial boundaries and influence on processes linking adjacent ecosystems. Ecohydrology 8(3): 406–415. doi:10.1002/eco.1488. Richardson, J.S., Shaughnessy, C.R., and Harrison, P.G. 2004. Litter breakdown and invertebrate association with three types of leaves in a temperate rainforest stream. Arch. für Hydrobiol. 159(3): 309–325. doi:10.1127/0003-9136/2004/0159-0309. Richardson, J.S., Bilby, R.E., and Bondar, C.A. 2005. Organic matter dynamics in small streams of the Pacific Northwest. J. Am. Water Resour. Assoc. 41(4): 921–934. doi:10.1111/j.1752-1688.2005.tb03777.x. Richardson, J.S., Hoover, T.M., and Lecerf, A. 2009. Coarse particulate organic matter dynamics in small streams: towards linking function to physical structure. Freshw. Biol. 54(10): 2116–2126. doi:10.1111/j.1365-2427.2009.02279.x. Richardson, J.S., Zhang, Y., and Marczak, L.B. 2010. Resource subsidies across the land–freshwater interface and responses in recipient communities. River Res. Appl. 26(1): 55–66. doi:10.1002/rra.1283. Richter, B.D., Baumgartner, J. V, Powell, J., and Braun, D.P. 1996. A method for assessing hydrologic alteration within ecosystems. Conserv. Biol. 10(4): 1163–1174. doi:10.1046/j.1523-1739.1996.10041163.x. Robertson, A.L., Brown, L.E., Klaar, M.J., and Milner, A.M. 2015. Stream ecosystem responses to an extreme rainfall event across multiple catchments in southeast Alaska. Freshw. Biol. 60(12): 2523–2534. doi:10.1111/fwb.12638. Rodriguez-Cabal, M.A., Barrios-Garcia, M.N., Rudman, S.M., McKown, A.D., Sato, T., and Crutsinger, G.M. 2016. It is about time: genetic variation in the timing of leaf-litter inputs influences aquatic ecosystems. Freshw. Biol. 62(2): 356–365. doi:10.1111/fwb.12872. 166  Rogger, M., Agnoletti, M., Alaoui, A., Bathurst, J.C., Bodner, G., Borga, M., et al. 2017. Land use change impacts on floods at the catchment scale: challenges and opportunities for future research. Water Resour. Res. 53(7): 5209–5219. doi:10.1002/2017WR020723. Romme, W.H., Everham, E.H., Frelich, L.E., Moritz, M.A., and Sparks, R.E. 1998. Are large, infrequent disturbances qualitatively different from small, frequent disturbances? Ecosystems 1(6): 524–534. doi:10.1007/s100219900048. Rosemond, A.D., Benstead, J.P., Bumpers, P.M., Gulis, V., Kominoski, J.S., Manning, D.W.P., et al. 2015. Experimental nutrient additions accelerate terrestrial carbon loss from stream ecosystems. Science 347(6226): 1142–1145. doi:10.1126/science.aaa1958. Rowe, L., and Richardson, J.S. 2001. Community responses to experimental food depletion: resource tracking by stream invertebrates. Oecologia 129(3): 473–480. doi:10.1007/s004420100748. Royer, T. V, and Minshall, G.W. 2003. Controls on leaf processing in streams from spatial-scaling and hierarchical perspectives. J. North Am. Benthol. Soc. 22(3): 352–358. doi:10.2307/1468266. Ruesink, J.L., and Srivastava, D.S. 2001. Numerical and per capita responses to species loss: mechanisms maintaining ecosystem function in a community of stream insect detritivores. Oikos 93(2): 221–234. doi:10.1034/j.1600-0706.2001.930206.x. Rumbos, C.I., Stamopoulos, D., Georgoulas, G., and Nikolopoulou, E. 2010. Factors affecting leaf litter decomposition by Micropterna sequax (Trichoptera: Limnephilidae). Int. Rev. Hydrobiol. 95(4‐5): 383–394. doi:10.1002/iroh.201011236. Sakai, M., Fukushima, K., Naito, R.S., Natuhara, Y., and Kato, M. 2016. Coniferous needle litter acts as a stable food resource for stream detritivores. Hydrobiologia 779(1): 161–171. doi:10.1007/s10750-016-2813-8. Sakamaki, T., and Richardson, J.S. 2013. Nonlinear variation of stream–forest linkage along a stream-size gradient: an assessment using biogeochemical proxies of in-stream fine particulate organic matter. J. Appl. Ecol. 50(4): 1019–1027. doi:10.1111/1365-2664.12099. Samhouri, J.F., Lester, S.E., Selig, E.R., Halpern, B.S., Fogarty, M.J., Longo, C., et al. 2012. Sea sick? Setting targets to assess ocean health and ecosystem services. Ecosphere 3(5): art41. doi:10.1890/ES11-00366.1. 167  Sanford, S.E., Creed, I.F., Tague, C.L., Beall, F.D., and Buttle, J.M. 2007. Scale-dependence of natural variability of flow regimes in a forested landscape. Water Resour. Res. 43(8): W08414. doi:10.1029/2006WR005299. Santiago, J., Molinero, J., and Pozo, J. 2011. Impact of timber harvesting on litterfall inputs and benthic coarse particulate organic matter (CPOM) storage in a small stream draining a eucalyptus plantation. For. Ecol. Manage. 262(6): 1146–1156. doi:10.1016/j.foreco.2011.06.018. Schäfer, R.B., and Piggott, J.J. 2018. Advancing understanding and prediction in multiple stressor research through a mechanistic basis for null models. Glob. Chang. Biol. 24(5): 1817–1826. doi:10.1111/gcb.14073. Scheffer, M., Carpenter, S.R., Dakos, V., and van Nes, E. 2015. Generic indicators of ecological resilience: inferring the chance of a critical transition. Annu. Rev. Ecol. Evol. Syst. 46: 145–167. doi:10.1146/annurev-ecolsys-112414-054242. Schieck, J., and Song, S.J. 2006. Changes in bird communities throughout succession following fire and harvest in boreal forests of western North America: literature review and meta-analyses. Can. J. For. Res. 36(5): 1299–1318. doi:10.1139/x06-017. von Schiller, D., Acuña, V., Aristi, I., Arroita, M., Basaguren, A., Bellin, A., et al. 2017. River ecosystem processes: a synthesis of approaches, criteria of use and sensitivity to environmental stressors. Sci. Total Environ. 596(Supplement C): 465–480. doi:10.1016/j.scitotenv.2017.04.081. Seidl, R., Fernandes, P.M., Fonseca, T.F., Gillet, F., Jönsson, A.M., Merganičová, K., et al. 2011. Modelling natural disturbances in forest ecosystems: a review. Ecol. Modell. 222(4): 903–924. doi:10.1016/j.ecolmodel.2010.09.040. Sibley, P.K., Kreutzweiser, D.P., Naylor, B.J., Richardson, J.S., and Gordon, A.M. 2012. Emulation of natural disturbance (END) for riparian forest management: synthesis and recommendations. Freshw. Sci. 31(1): 258–264. doi:10.1899/11-094.1. Small, M.J., Doyle, M.W., Fuller, R.L., and Manners, R.B. 2008. Hydrologic versus geomorphic limitation on CPOM storage in stream ecosystems. Freshw. Biol. 53(8): 1618–1631. doi:10.1111/j.1365-2427.2008.01999.x. Smith, B., and Wilson, J.B. 1996. A consumer’s guide to evenness indices. Oikos 76(1): 70–82. 168  doi:10.2307/3545749. Smolders, K.E., Rolls, R.J., Boulton, A.J., Webb, A.A., and Sheldon, F. 2018. Effects of selective forest harvesting best management practices on organic matter and invertebrate detritivores in streams draining subtropical eucalypt forest. Ecol. Eng. 122: 271–285. doi:10.1016/j.ecoleng.2018.08.010. Šnajdr, J., Valášková, V., Merhautová, V., Herinková, J., Cajthaml, T., and Baldrian, P. 2008. Spatial variability of enzyme activities and microbial biomass in the upper layers of Quercus petraea forest soil. Soil Biol. Biochem. 40(9): 2068–2075. doi:10.1016/j.soilbio.2008.01.015. Snyder, C.D., and Johnson, Z.B. 2006. Macroinvertebrate assemblage recovery following a catastrophic flood and debris flows in an Appalachian mountain stream. J. North Am. Benthol. Soc. 25(4): 825–840. doi:10.1899/0887-3593(2006)025[0825:MARFAC]2.0.CO;2. Sopper, W.E., and Lynch, J.A. 1970. Changes in water yield following partial forest cover removal on an experimental watershed. Int. Assoc. Hydrol. Sci. Publ. No 96: 369–389. Srivastava, D.S., Cardinale, B.J., Downing, A.L., Duffy, J.E., Jouseau, C., Sankaran, M., et al. 2009. Diversity has stronger top-down than bottom-up effects on decomposition. Ecology 90(4): 1073–1083. doi:10.1890/08-0439.1. Stanley, E.H., Powers, S.M., and Lottig, N.R. 2010. The evolving legacy of disturbance in stream ecology: concepts, contributions, and coming challenges. J. North Am. Benthol. Soc. 29(1): 67–83. doi:10.1899/08-027.1. Stearns, S.C. 1987. Some modest advice for graduate students. Bull. Ecol. Soc. Am. 68(2): 145–150. doi:10.2307/20166580. Steen, D.A., and Jachowski, D.S. 2013. Expanding shifting baseline syndrome to accommodate increasing abundances. Restor. Ecol. 21(5): 527–529. doi:10.1111/rec.12040. Stenroth, K., Hoover, T.M., Herrmann, J., Bohman, I., and Richardson, J.S. 2014. A model-based comparison of organic matter dynamics between riparian-forested and open-canopy streams. Riparian Ecol. Conserv. 2(1): 1–13. doi:10.2478/remc-2014-0001. Stoddard, J.L., Larsen, D.P., Hawkins, C.P., Johnson, R.K., and Norris, R.H. 2006. Setting expectations for the ecological condition of streams: the concept of reference condition. 169  Ecol. Appl. 16(4): 1267–1276. doi:10.1890/1051-0761(2006)016[1267:SEFTEC]2.0.CO;2. Stoker, D., Falkner, A.J., Murray, K.M., Lang, A.K., Barnum, T.R., Hepinstall-Cymerman, J., et al. 2017. Decomposition of terrestrial resource subsidies in headwater streams: does consumer diversity matter? Ecosphere 8(6): e01868. doi:10.1002/ecs2.1868. Stone, M.K., and Wallace, J.B. 1998. Long-term recovery of a mountain stream from clear-cut logging: the effects of forest succession on benthic invertebrate community structure. Freshw. Biol. 39(1): 151–169. doi:10.1046/j.1365-2427.1998.00272.x. Storck, P., Bowling, L., Wetherbee, P., and Lettenmaier, D. 1998. Application of a GIS-based distributed hydrology model for prediction of forest harvest effects on peak stream flow in the Pacific Northwest. Hydrol. Process. 12(6): 889–904. doi:10.1002/(SICI)1099-1085(199805)12:6<889::AID-HYP661>3.0.CO;2-P. Strayer, D.L., and Dudgeon, D. 2010. Freshwater biodiversity conservation: recent progress and future challenges. J. North Am. Benthol. Soc. 29(1): 344–358. doi:10.1899/08-171.1. Sun, G., Riedel, M., Jackson, R., Kolka, R., Amatya, D., and Shepard, J. 2004. Influences of management of Southern forests on water quantity and quality. U.S. Department of Agriculture, Forest Service, Southern Research Station, Asheville, NC, USA. Sweeney, B.W., and Newbold, J.D. 2014. Streamside forest buffer width needed to protect stream water quality, habitat, and organisms: a literature review. JAWRA J. Am. Water Resour. Assoc. 50(3): 560–584. doi:10.1111/jawr.12203. Sweeney, B.W., Bott, T.L., Jackson, J.K., Kaplan, L.A., Newbold, J.D., Standley, L.J., et al. 2004. Riparian deforestation, stream narrowing, and loss of stream ecosystem services. Proc. Natl. Acad. Sci. U. S. A. 101(39): 14132–14137. doi:10.1073/pnas.0405895101. Tanentzap, A.J., Szkokan-Emilson, E.J., Kielstra, B.W., Arts, M.T., Yan, N.D., and Gunn, J.M. 2014. Forests fuel fish growth in freshwater deltas. Nat. Commun. 5. doi:10.1038/ncomms5077. Tank, J.L., Rosi-Marshall, E.J., Griffiths, N.A., Entrekin, S.A., and Stephen, M.L. 2010. A review of allochthonous organic matter dynamics and metabolism in streams. J. North Am. Benthol. Soc. 29(1): 118–146. doi:10.1899/08-170.1. Taylor, B.R., and Andrushchenko, I. V. 2014. Interaction of water temperature and shredders on leaf litter breakdown: a comparison of streams in Canada and Norway. Hydrobiologia 170  721(1): 77–88. doi:10.1007/s10750-013-1650-2. Tiegs, S.D., Peter, F.D., Robinson, C.T., Uehlinger, U., and Gessner, M.O. 2008. Leaf decomposition and invertebrate colonization responses to manipulated litter quantity in streams. J. North Am. Benthol. Soc. 27(2): 321–331. doi:10.1899/07-054.1. Tiegs, S.D., Akinwole, P.O., and Gessner, M.O. 2009. Litter decomposition across multiple spatial scales in stream networks. Oecologia 161(2): 343–351. doi:10.1007/s00442-009-1386-x. Tonin, A.M., Pozo, J., Monroy, S., Basaguren, A., Pérez, J., Gonçalves Jr., J.F., et al. 2018a. Interactions between large and small detritivores influence how biodiversity impacts litter decomposition. J. Anim. Ecol. 87(5): 1465–1474. doi:10.1111/1365-2656.12876. Tonin, A.M., Hepp, L.U., and Gonçalves, J.F.J. 2018b. Spatial variability of plant litter decomposition in stream networks: from litter bags to watersheds. Ecosystems 21(3): 567–581. doi:10.1007/s10021-017-0169-1. Tremblay, Y., Rousseau, A.N., Plamondon, A.P., Lévesque, D., and Jutras, S. 2008. Rainfall peak flow response to clearcutting 50% of three small watersheds in a boreal forest, Montmorency Forest, Québec. J. Hydrol. 352(1–2): 67–76. doi:10.1016/j.jhydrol.2007.12.028. Turner, M., and Stewardson, M. 2014. Hydrologic indicators of hydraulic conditions that drive flow–biota relationships. Hydrol. Sci. J. 59(3–4): 659–672. doi:10.1080/02626667.2014.896997. Turner, M.G., Collins, S.L., Lugo, A.L., Magnuson, J.J., Rupp, T.S., and Swanson, F.J. 2003. Disturbance dynamics and ecological response: the contribution of long-term ecological research. Bioscience 53(1): 46–56. doi:10.1641/0006-3568(2003)053[0046:DDAERT]2.0.CO;2. Underwood, A.J. 1992. Beyond BACI: the detection of environmental impacts on populations in the real, but variable, world. J. Exp. Mar. Bio. Ecol. 161(2): 145–178. doi:10.1016/0022-0981(92)90094-Q. Underwood, A.J. 1994. On beyond BACI: sampling designs that might reliably detect environmental disturbances. Ecol. Appl. 4(1): 3–15. doi:10.2307/1942110. Ursic, S.J. 1991. Hydrologic effects of two methods of harvesting mature southern pine. JAWRA 171  J. Am. Water Resour. Assoc. 27(2): 303–315. doi:10.1111/j.1752-1688.1991.tb03135.x. Vanni, M.J. 2002. Nutrient cycling by animals in freshwater ecosystems. Annu. Rev. Ecol. Evol. Syst. 33: 341–370. doi:10.2307/3069266. Veen, G.F. (Ciska), Freschet, G.T., Ordonez, A., and Wardle, D.A. 2014. Litter quality and environmental controls of home-field advantage effects on litter decomposition. Oikos 124(2): 187–195. doi:10.1111/oik.01374. Verry, E.S., Lewis, J.R., and Brooks, K.N. 1983. Aspen clearcutting increases snowmelt and storm flow peaks in north central Minnesota. JAWRA J. Am. Water Resour. Assoc. 19(1): 59–67. doi:10.1111/j.1752-1688.1983.tb04557.x. Vitousek, P.M., Mooney, H.A., Lubchenco, J., and Melillo, J.M. 1997. Human domination of earth’s ecosystems. Science 277(5325): 494–499. doi:10.1126/science.277.5325.494. Vörösmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D., Prusevich, A., Green, P., et al. 2010. Global threats to human water security and river biodiversity. Nature 467(7315): 555–561. doi:10.1038/nature09440. Voß, K., Fernández, D., and Schäfer, R.B. 2015. Organic matter breakdown in streams in a region of contrasting anthropogenic land use. Sci. Total Environ. 527–528: 179–184. doi:10.1016/j.scitotenv.2015.04.071. Wagenhoff, A., Townsend, C.R., Phillips, N., and Matthaei, C.D. 2011. Subsidy-stress and multiple-stressor effects along gradients of deposited fine sediment and dissolved nutrients in a regional set of streams and rivers. Freshw. Biol. 56(9): 1916–1936. doi:10.1111/j.1365-2427.2011.02619.x. Walker, B., Holling, C.S., Carpenter, S.R., and Kinzig, A. 2004. Resilience, adaptability and transformability in social–ecological systems. Ecol. Soc. 9(2): 5. Available from http://www.ecologyandsociety.org/vol9/iss2/art5/. Wallace, J.B., and Webster, J.R. 1996. The role of macroinvertebrates in stream ecosystem function. Annu. Rev. Entomol. 41(1): 115–139. doi:10.1146/annurev.en.41.010196.000555. Wallace, J.B., Eggert, S.L., Meyer, J.L., and Webster, J.R. 1997. Multiple trophic levels of a forest stream linked to terrestrial litter inputs. Science 277(5322): 102–104. doi:10.1126/science.277.5322.102. Wallace, J.B., Eggert, S.L., Meyer, J.L., and Webster, J.R. 1999. Effects of resource limitation 172  on a detrital-based ecosystem. Ecol. Monogr. 69(4): 409–442. doi:10.2307/2657224. Wallenstein, M.D., Haddix, M.L., Ayres, E., Steltzer, H., Magrini-Bair, K.A., and Paul, E.A. 2013. Litter chemistry changes more rapidly when decomposed at home but converges during decomposition–transformation. Soil Biol. Biochem. 57: 311–319. doi:10.1016/j.soilbio.2012.09.027. Wang, X., Burns, D.A., Yanai, R.D., Briggs, R.D., and Germain, R.H. 2006. Changes in stream chemistry and nutrient export following a partial harvest in the Catskill Mountains, New York, USA. For. Ecol. Manage. 223(1–3): 103–112. doi:10.1016/j.foreco.2005.10.060. Webster, J.R. 1983. The role of benthic macroinvertebrates in detritus dynamics of streams: a computer simulation. Ecol. Monogr. 53(4): 383–404. doi:10.2307/1942645. Webster, J.R., and Waide, J.B. 1982. Effects of forest clearcutting on leaf breakdown in a southern Appalachian stream. Freshw. Biol. 12(4): 331–344. doi:10.1111/j.1365-2427.1982.tb00627.x. Webster, J.R., Gurtz, M.E., Hains, J.J., Meyer, J.L., Swank, W.T., Waide, J.B., et al. 1983. Stability of stream ecosystems. In Stream Ecology SE - 14. Edited by J. Barnes and G.W. Minshall. Springer US. pp. 355–395. Webster, J.R., Benfield, E.F., Golladay, S.W., Kazmierczak Jr., R.F., Perry, W.B., and Peters, G.T. 1988. Effects of watershed disturbance on stream seston characteristics. In Forest Hydrology and Ecology at Coweeta SE - 21. Edited by W. Swank and D.A. Crossley Jr. Springer, New York, NY, USA. pp. 279–294. Webster, J.R., Golladay, S.W., Benfield, E.F., D’Angelo, D.J., and Peters, G.T. 1990. Effects of forest disturbance on particulate organic matter budgets of small streams. J. North Am. Benthol. Soc. 9(2): 120–140. doi:10.2307/1467446. Webster, J.R., Benfield, E.F., Ehrman, T.P., Schaeffer, M.A., Tank, J.L., Hutchens, J.J., et al. 1999. What happens to allochthonous material that falls into streams? A synthesis of new and published information from Coweeta. Freshw. Biol. 41(4): 687–705. doi:10.1046/j.1365-2427.1999.00409.x. Webster, J.R., Benfield, E.F., Hutchens, J.J., Tank, J.L., Golladay, S.W., and Adams, J.C. 2001. Do leaf breakdown rates actually measure leaf disappearance from streams? Int. Rev. Hydrobiol. 86(4‐5): 417–427. doi:10.1002/1522-2632(200107)86:4/5<417::AID-173  IROH417>3.0.CO;2-L. Webster, J.R., Benfield, E.J., and Golladay, Stephen W. McTammany, M.E. 2014a. Recovery of particulate organic matter dynamics in a stream draining a logged watershed. In Long-term response of a forest watershed ecosystem: clearcutting in the Southern Appalachians. Edited by W.T. Swank and J.R. Webster. Oxford University Press. pp. 156–176. Webster, J.R., Benfield, E.F., Golladay, S.W., and McTammany, M.E. 2014b. Stream POC dynamics. In Long-term response of a forest watershed ecosystem: clearcutting in the Southern Appalachians. Edited by W.T. Swank and J.R. Webster. Oxford University Press, New York, NY, USA. pp. 156–176. Webster, K.L., Beall, F.D., Creed, I.F., and Kreutzweiser, D.P. 2015. Impacts and prognosis of natural resource development on water and wetlands in Canada’s boreal zone. Environ. Rev. 23(1): 78–131. doi:10.1139/er-2014-0063. Westman, W.E. 1978. Measuring the inertia and resilience of ecosystems. Bioscience 28(11): 705–710. doi:10.2307/1307321. Whiles, M.R., Wallace, J.B., and Chung, K. 1993. The influence of Lepidostoma (Trichoptera: Lepidostomatidae) on recovery of leaf-litter processing in disturbed headwater streams. Am. Midl. Nat. 130(2): 356–363. doi:10.2307/2426133. White, P.S., and Jentsch, A. 2001. The search for generality in studies of disturbance and ecosystem dynamics. In Progress in Botany SE - 17. Edited by K. Esser, U. Lüttge, J.W. Kadereit, and W. Beyschlag. Springer Berlin Heidelberg. pp. 399–450. Wilkerson, E., Hagan, J.M., Siegel, D., and Whitman, A.A. 2006. The effectiveness of different buffer widths for protecting headwater stream temperature in Maine. For. Sci. 52(3): 221–231. doi:10.1093/forestscience/52.3.221. Williamson, C.E., Dodds, W., Kratz, T.K., and Palmer, M.A. 2008. Lakes and streams as sentinels of environmental change in terrestrial and atmospheric processes. Front. Ecol. Environ. 6(5): 247–254. doi:10.1890/070140. Wipfli, M.S., and Gregovich, D.P. 2002. Export of invertebrates and detritus from fishless headwater streams in southeastern Alaska: implications for downstream salmonid production. Freshw. Biol. 47(5): 957–969. doi:10.1046/j.1365-2427.2002.00826.x. Wipfli, M.S., Richardson, J.S., and Naiman, R.J. 2007. Ecological linkages between headwaters 174  and downstream ecosystems: transport of organic matter, invertebrates, and wood down headwater channels. J. Am. Water Resour. Assoc. 43(1): 72–85. doi:10.1111/j.1752-1688.2007.00007.x. Woodward, G., Gessner, M.O., Giller, P.S., Gulis, V., Hladyz, S., Lecerf, A., et al. 2012. Continental-scale effects of nutrient pollution on stream ecosystem functioning. Science 336(6087): 1438–1440. doi:10.1126/science.1219534. Worm, B., and Duffy, J.E. 2003. Biodiversity, productivity and stability in real food webs. Trends Ecol. Evol. 18(12): 628–632. doi:10.1016/j.tree.2003.09.003. Xie, Y., Xie, Y., Xiao, H., Chen, X., and Li, F. 2017. Controls on litter decomposition of emergent macrophyte in Dongting Lake wetlands. Ecosystems 20(7): 1383–1389. doi:10.1007/s10021-017-0119-y. Yeung, A.C.Y., Lecerf, A., and Richardson, J.S. 2017. Assessing the long-term ecological effects of riparian management practices on headwater streams in a coastal temperate rainforest. For. Ecol. Manage. 384: 100–109. doi:10.1016/j.foreco.2016.10.044. Yeung, A.C.Y., Musetta-Lambert, J., Kreutzweiser, D.P., Sibley, P.K., and Richardson, J.S. 2018. Relations of interannual differences in stream litter breakdown with discharge: bioassessment implications. Ecosphere 9(9): e02423. doi:10.1002/ecs2.2423. Young, K.A., Hinch, S.G., and Northcote, T.G. 2011. Status of resident coastal cutthroat trout and their habitat twenty-five years after riparian logging. North Am. J. Fish. Manag. 19(4): 901–911. doi:10.1577/1548-8675(1999)019<0901:SORCCT>2.0.CO;2. Young, R.G., and Collier, K.J. 2009. Contrasting responses to catchment modification among a range of functional and structural indicators of river ecosystem health. Freshw. Biol. 54(10): 2155–2170. doi:10.1111/j.1365-2427.2009.02239.x. Young, R.G., Matthaei, C.D., and Townsend, C.R. 2008. Organic matter breakdown and ecosystem metabolism: functional indicators for assessing river ecosystem health. J. North Am. Benthol. Soc. 27(3): 605–625. doi:10.1899/07-121.1. Zalasiewicz, J., Williams, M., Haywood, A., and Ellis, M. 2011. The Anthropocene: a new epoch of geological time? Philos. Trans. R. Soc. London A Math. Phys. Eng. Sci. 369(1938): 835–841. doi:10.1098/rsta.2010.0339. Zeglin, L.H. 2015. Stream microbial diversity in response to environmental changes: review and 175  synthesis of existing research. Front. Microbiol. 6: 454. doi:10.3389/fmicb.2015.00454. Zhang, Y., and Richardson, J.S. 2011. Contrasting effects of cross-ecosystem subsidies and predation on benthic invertebrates in two Pacific coastal streams. Aquat. Sci. 73(1): 53–62. doi:10.1007/s00027-010-0159-2. Zhang, Y., Richardson, J.S., and Pinto, X. 2009. Catchment-scale effects of forestry practices on benthic invertebrate communities in Pacific coastal streams. J. Appl. Ecol. 46(6): 1292–1303. doi:10.1111/j.1365-2664.2009.01718.x. Zhang, Z., Yan, C., Krebs, C.J., and Stenseth, N.C. 2015. Ecological non-monotonicity and its effects on complexity and stability of populations, communities and ecosystems. Ecol. Modell. 312: 374–384. doi:10.1016/j.ecolmodel.2015.06.004. Ziemer, R.R. 1998. Flooding and stormflows. In Proceedings of the Conference on Coastal Watersheds: The Caspar Creek Story. Edited by R.R. Ziemer. U.S. Department of Agriculture, Forest Service, Pacific Southwest Research Station, Albany, CA, USA. pp. 15–24. Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., and Smith, G.M. 2009. Mixed effects models and extensions in ecology with R. Springer, Berlin, Germany. Zuur, A.F., Ieno, E.N., and Elphick, C.S. 2010. A protocol for data exploration to avoid common statistical problems. Methods Ecol. Evol. 1(1): 3–14. doi:10.1111/j.2041-210X.2009.00001.x. Zwolak, R. 2009. A meta-analysis of the effects of wildfire, clearcutting, and partial harvest on the abundance of North American small mammals. For. Ecol. Manage. 258(5): 539–545. doi:10.1016/j.foreco.2009.05.033.   176  Appendices  Appendix A  Supporting information for Chapter 3   A.1 Geographic locations of study sites in British Columbia and Ontario, Canada    177   Figure A.1 Maps showing the locations of study sites within study regions in British Columbia and Ontario, including (a) Malcolm Knapp Research Forest (MKRF), (b) Turkey Lakes Watershed (TLW), and (c) White River (WR).  178  A.2 Variance partitioning of stream litter breakdown rates explained by their important drivers, and redundancy analysis (RDA) to uniquely quantify the relationships between hydrology and litter breakdown rates  Partial RDA (pRDA) was used to partition the variation of litter breakdown rates explained by the unique effects of each variable (e.g., fungal biomass), or group of variables (i.e., hydrologic, shredder-related, water chemistry; see Table 3.2), shared variance explained by them, and residual variance not explained by these explanatory variables. The variance partitioning was based on partial linear regression, since univariate response variables were tested. Data collected from the fourth year of study (i.e., 2017) in MKRF and TLW were excluded in this particular analysis. Prior to variance partitioning, separate global RDA models were run, in which litter breakdown rate was constrained to be a linear combination of predictor variables (i.e., variable sets). A global RDA model involving fungal biomass was only run for MKRF and TLW, as fungal biomass was not measured in WR. If the global model involving a particular variable set was not significant, its contribution to the variation of litter breakdown rate was assumed to be zero. If the global model was significant, forward selection using 2 stopping rules was then performed with the packfor package based on Blanchet et al. (2008), in order to determine which subset of variables was the most important in explaining litter breakdown rate. Specifically, each variable retained should be associated with a P value < 0.05 after Holm’s correction, and the adjusted R2 of the final RDA model should not be greater than that of the respective global model. Then, pRDA was performed using the forward-selected variables to compare the degree of variance of litter breakdown rate explained by each set of explanatory variable(s). Variance fraction was measured by adjusted R2 (Peres-Neto et al. 2006). Relationships between hydrology and litter breakdown rates were further explored by global RDA, using only forward-selected hydrologic indices with data from all regions and study years. A polynomial form of RDA was also performed with the program Polynomial RDACCA (Makarenkov and Legendre 2002), which allows potential nonlinearity between evaluated relationships to be modelled, by including the second-order polynomials of hydrologic indices in the linear combination of explanatory variables. A permutation procedure (999 permutations) 179  provided in the program was employed to test the significance of the linear and polynomial RDA models, and to assess which provided a better model fit. For the RDA model with better fit, the fitted site scores on the first RDA axis (i.e., ‘site constraints’) were considered as a composite measure of important hydrologic indices that were strongly associated with litter breakdown.180  A.3 Parameters of stream litter breakdown rates measured in study regions in British Columbia and Ontario Table A.3.1 Temperature-corrected litter breakdown rate in coarse-mesh (kc) and fine-mesh bags (kf), fragmentation rate (λF), and dissolution and microbial decomposition rate (λm) across all study regions in British Columbia and Ontario, and study years.  Region and site kc   kf   λF   λm  1st 2nd 3rd 4th   1st 2nd 3rd 4th   1st 2nd 3rd 4th   1st 2nd 3rd 4th MKRF                    G-4 0.00223 - 0.00151 -  0.00183 - 0.00104 -  0.00021 - 0.00025 -  0.00202 - 0.00126 - Mike 0.00232 0.00175 0.00173 0.00111  0.00185 0.00126 0.00107 0.00063  0.00024 0.00026 0.00036 0.00026  0.00207 0.00149 0.00137 0.00085 Spring 0.00304 0.00240 0.00220 0.00345  0.00185 0.00117 0.00128 0.00082  0.00064 0.00068 0.00050 0.00162  0.00240 0.00171 0.00170 0.00183 Upper East 0.00558 0.00402 0.00392 0.00520  0.00178 0.00111 0.00092 0.00061  0.00226 0.00176 0.00186 0.00306  0.00333 0.00226 0.00207 0.00214                     TLW                    TLW34 0.00239 0.00197 0.00238 0.00339  0.00124 0.00113 0.00087 0.00087  0.00064 0.00046 0.00088 0.00154  0.00175 0.00151 0.00150 0.00185 TLW96 0.00171 0.00223 0.00222 0.00201  0.00098 0.00119 0.00079 0.00071  0.00040 0.00057 0.00084 0.00076  0.00131 0.00166 0.00139 0.00125 TLW97 0.00325 0.00361 0.00356 0.00260  0.00120 0.00149 0.00082 0.00077  0.00119 0.00122 0.00169 0.00110  0.00206 0.00239 0.00187 0.00150 AY1-1 0.00228 0.00214 0.00143 0.00254  0.00170 0.00148 0.00069 0.00093  0.00031 0.00035 0.00041 0.00094  0.00197 0.00179 0.00102 0.00160 AY4-1 0.00129 0.00219 0.00130 0.00139  0.00121 0.00122 0.00081 0.00106  0.00004 0.00053 0.00027 0.00017  0.00125 0.00166 0.00104 0.00122                     WR                    EWR4 0.00136 0.00125 0.00097 –  0.00044 0.00051 0.00036 –  0.00054 0.00042 0.00036 –  0.00081 0.00082 0.00061 – EWR5 0.00068 0.00079 0.00083 –  0.00050 0.00060 0.00062 –  0.00009 0.00010 0.00011 –  0.00059 0.00069 0.00072 – 181  Region and site kc   kf   λF   λm  1st 2nd 3rd 4th   1st 2nd 3rd 4th   1st 2nd 3rd 4th   1st 2nd 3rd 4th EWR6 0.00076 0.00111 0.00055 –  0.00048 0.00050 0.00030 –  0.00015 0.00034 0.00014 –  0.00061 0.00076 0.00041 – EWR8 0.00105 0.00094 0.00036 –  0.00045 0.00035 0.00014 –  0.00035 0.00035 0.00013 –  0.00071 0.00060 0.00024 – EWR9 0.00130 0.00085 0.00044 –   0.00054 0.00056 0.00022 –   0.00044 0.00015 0.00011 –   0.00087 0.00070 0.00033 – Notes: MKRF, Malcolm Knapp Research Forest; TLW, Turkey Lakes Watershed; WR, White River. Note kf, λF, and λm at WR in the 3rd year (i.e., 2014) were estimated based on site-specific kc/kf. The sum of λF and λm may differ from the corresponding kc at some sites due to rounding differences. 182  Table A.3.2 Ratio of temperature-corrected litter breakdown rate in coarse-mesh (kc) and fine-mesh bags (kf), ratio of fragmentation rate (λF), and dissolution and microbial decomposition rate (λm), and λF/kc across all study regions in British Columbia and Ontario, and study years.  Region and site kc/kf   λF/λm   λF/kc 1st 2nd 3rd 4th   1st 2nd 3rd 4th   1st 2nd 3rd 4th MKRF               G-4 1.22 – 1.46 –  0.10 – 0.20 –  0.09 – 0.17 – Mike 1.25 1.39 1.62 1.76  0.12 0.17 0.26 0.31  0.10 0.15 0.21 0.24 Spring 1.64 2.04 1.72 4.23  0.27 0.40 0.29 0.89  0.21 0.29 0.23 0.47 Upper East 3.14 3.61 4.29 8.59  0.68 0.78 0.90 1.43  0.40 0.44 0.47 0.59                TLW               TLW34 1.93 1.75 2.73 3.89  0.36 0.31 0.58 0.83  0.27 0.23 0.37 0.45 TLW96 1.75 1.87 2.81 2.85  0.31 0.35 0.60 0.61  0.23 0.26 0.38 0.38 TLW97 2.70 2.43 4.35 3.38  0.58 0.51 0.91 0.73  0.37 0.34 0.48 0.42 AY1-1 1.34 1.44 2.06 2.73  0.15 0.20 0.40 0.59  0.13 0.16 0.29 0.37 AY4-1 1.06 1.79 1.62 1.32  0.03 0.32 0.26 0.14  0.03 0.24 0.21 0.13                WR               EWR4 3.09 2.45 2.77 –  0.67 0.51 0.59 –  0.40 0.34 0.37 – EWR5 1.36 1.32 1.34 –  0.16 0.14 0.15 –  0.14 0.13 0.13 – EWR6 1.57 2.23 1.90 –  0.24 0.45 0.34 –  0.19 0.31 0.25 – EWR8 2.36 2.70 2.53 –  0.49 0.58 0.53 –  0.33 0.37 0.35 – EWR9 2.42 1.50 1.96 –   0.51 0.22 0.34 –   0.34 0.18 0.25 – Notes: MKRF, Malcolm Knapp Research Forest; TLW, Turkey Lakes Watershed; WR, White River. Site-specific kc/kf in WR for the 3rd year is estimated as the average of the corresponding values for the 1st and 2nd year. 183  A.4 Forward selection of variables influencing stream litter breakdown rates Table A.4.1 Results of forward selection of hydrologic, shredder-related, and water chemistry variables which strongly influenced in-stream total breakdown rate (kc), fragmentation rate (λF), and dissolution and microbial decomposition rate (λm) across all study regions in British Columbia and Ontario, using global RDA models.  Variable set Variable(s) selected Cumulative R2 Cumulative adjusted R2 F P (adjusted) kc Hydrology      1 CVD 0.338 0.321 19.91 0.014 2 P.ROC 0.521 0.496 14.54 0.014 Shredder-related      1 SD 0.229 0.209 11.60 0.016 2 STR 0.368 0.334 8.31 0.021 3 TPD 0.484 0.442 8.34 0.021 4 LMD 0.568 0.520 7.00 0.021 Water chemistry      1 TP 0.446 0.430 27.36 0.004 λF Shredder-related      1 SD 0.260 0.241 13.73 0.004 2 TPD 0.404 0.373 9.19 0.012 Water chemistry      1 TP 0.446 0.430 27.36 0.004 λm Hydrology      1 CVD 0.451 0.437 32.04 0.014 2 P.ROC 0.643 0.624 20.35 0.014 Water chemistry      1 TP 0.498 0.483 33.68 0.004 184  Table A.4.2 Results of linear mixed-effects models testing the inter-year differences in composite hydrologic index for in-stream total breakdown rate, and dissolution and microbial decomposition rate (λm) for each study region in British Columbia and Ontario.  Variable Degrees of freedom F P Comparison kc     MKRF 3,7.58 223.88 < 0.001 4th > 1st > 2nd > 3rd TLW 3,12 20871 < 0.001 4th > 2nd > 1st > 3rd WR 2,8 1.70 0.24 –      λm     MKRF 3,7.58 232.52 < 0.001 4th > 1st > 2nd > 3rd TLW 3,12 21782 < 0.001 4th > 2nd > 1st > 3rd WR 2,8 1.82 0.22 – Notes: MKRF, Malcolm Knapp Research Forest; TLW, Turkey Lakes Watershed; WR, White River. Site was treated as a random effect for each region. The significance of ‘year’ effects was estimated using an approximate F-test based on the Kenward-Roger approach. Significant pairwise differences between individual years (i.e., 1st to 3rd/4th year) were determined by Tukey’s HSD post-hoc tests. 185  A.5 Stream shredder assemblage structure in litter bags in study regions in British Columbia and Ontario    Figure A.5.1 Non-metric multidimensional scaling ordination of shredder assemblages across all years and study regions, including (a) Malcolm Knapp Research Forest (MKRF), (b) Turkey Lakes Watershed (TLW), and (c) White River (WR). Dashed ellipses represent 95% confidence intervals for the centroids of the groups of years. Note that NMDS axis scores are not comparable between regions. 186  Appendix B  Supporting information for Chapter 4   Figure B.1 Geographic locations of study streams within (a) Malcolm Knapp Research Forest (MKRF), British Columbia, and (b) Turkey Lakes Watershed (TLW), Ontario, Canada.187  Table B.1 Catchment- and reach-scale characteristics of the study streams in Malcolm Knapp Research Forest (MKRF), British Columbia, and Turkey Lakes Watershed (TLW), Ontario, Canada. Measurements of wetted width, mid-channel depth, and water characteristics were conducted at the time of litterbag retrieval. Values were averaged across two years for sites with incubations of all litter pairs. Measurements of canopy openness were obtained using either spherical densiometer or digital hemispherical pictures taken by fisheye lens (see Yeung et al. 2017) in both regions in 2014.  Region and site Location Upstream watershed area (ha) Elevation (m) Stream order Reach gradient (%) Wetted width (m) Mid-channel depth (m) Latitude Longitude MKRF         E10-1 49°18'20"N 122°34'16"W 6 371 1st 14 1.3 0.12 G ‡ 49°16'19"N 122°33'34"W 74 226 2nd 4 ~1.5 ~0.1 G-4 † 49°17'44"N 122°35'48"W 28 257 1st 31 0.9 0.07 Mike 49°16'40"N 122°32'46"W 30 314 1st 5 1.3 0.09 Spring 49°17'41"N 122°34'2"W 38 340 3rd 5 2.4 0.13 R20-4 † 49°19'58"N 122°31'55"W 34 577 1st 37 0.9 0.10 Upper East 49°17'3"N 122°33'43"W 36 306 2nd 58 1.9 0.10  TLW         TLW34 47°3'27"N 84°24'59"W 68 391 2nd 5 2.3 0.16 TLW96 47°4'39"N 84°24'39"W 71 362 2nd 2 1.4 0.10 TLW97 † 47°4'34"N 84°24'59"W 37 363 2nd 9 1.5 0.11 AY1-1 46°58'17"N 84°17'59"W 987 290 3rd 1 4.1 0.14 AY3-1 ‡ 46°59'11"N 84°18'57"W 314 268 3rd 5 1.4 0.13 AY4-1 47°0'28"N 84°18'41"W 219 248 3rd 6 1.7 0.15 AY13 ‡ 47°2'34"N 84°23'11"W 361 384 3rd 3 1.8 0.15 † Incubation of alder litter pair only. ‡ Incubation of maple and cedar litter pairs only.  188  Table B.1 (Cont’d)  Region and site Within-litterbag water temperature during incubation (°C) DO (mg L-1) pH Conductivity (µS cm-1) NO3- concentrations measured as total oxidised nitrogen (mg L-1) MKRF      E10-1 9.7 11.8 6.0 10.4 0.142 G ‡ 9.4 11.9 6.3 10.4 0.174 G-4 † 10.0 11.6 6.0 11.1 0.315 Mike 9.7 11.6 6.3 13.3 0.499 Spring 9.3 11.9 6.6 13.1 0.054 R20-4 † 8.4 11.8 6.6 15.2 0.003 Upper East 9.4 12.1 6.6 13.8 0.050             TLW      TLW34 8.4 11.7 7.1 32.3 0.161 TLW96 8.5 11.6 6.9 24.6 0.231 TLW97 † 8.6 11.8 6.7 17.8 0.213 AY1-1 7.9 12.6 6.8 22.0 0.092 AY3-1 ‡ 7.3 11.7 6.4 15.8 0.048 AY4-1 8.7 12.6 6.6 17.0 0.026 AY13 ‡ 7.2 12.1 6.7 21.6 0.068 † Incubation of alder litter pair only. ‡ Incubation of maple and cedar litter pairs only. 189  Table B.2 Summary of principal component (PC) loadings of 8 litter chemistry traits of 6 plant litter species in British Columbia and Ontario, Canada. Significant Pearson correlation coefficients for relationships between the variables and PC are in bold typeface (* P < 0.05; ** P < 0.01). The contribution of variables with significant correlations to the respective PC (in %) is shown in parentheses.    PC1 PC2 Eigenvalue 3.87 2.08 Variation explained 48.3 26.0    Chemical trait:   % N 0.93 ** (22.1) 0.24 % Lignin -0.28 0.85 * (34.9) % Ca -0.13 -0.77 % K 0.42 -0.66 % Mg 0.68 0.09 C:N -0.98 ** (24.9) -0.02 N:P 0.65 0.47 Lignin:N -0.94 ** (23.0) 0.20 190  Table B.3 Average (with 1 SE in parentheses) temperature-corrected fragmentation rate (λF), breakdown rate in fine-mesh bags (kf), ratio of λF to breakdown rate in coarse-mesh bags (kc) (in geometric mean; geometric SD in parentheses), algal and fungal biomass on leaf discs in coarse- and fine-mesh bags, and shredder density, taxonomic richness and community evenness of two or three litter pairs incubated in the study streams of British Columbia and Ontario, Canada. For each study region, it is indicated whether the litter species is derived from (H), or away from (A) that particular region.    Variable   British Columbia     Ontario   Alder   Alnus rubra (H) Alnus rugosa (A)  Alnus rubra (A) Alnus rugosa (H)  λF (10-3 degree-day-1) 0.66 (0.56) 1.03 (0.63)  0.51 (0.33) 0.57 (0.37)  kf (10-3 degree-day-1)  1.72 (0.01) 0.52 (0.01)  1.27 (0.12) 0.54 (0.11)  λF/kc  0.16 (1.66) 0.38 (1.43)  0.16 (2.43) 0.34 (1.45)  Algal biomass (mg m-2) Coarse: 2.62 (0.86) 4.18 (1.05)  4.86 (1.24) 3.81 (1.54)  Fine: 3.40 (0.56) 2.66 (0.59)  5.77 (1.62) 2.33 (0.68)  Fungal biomass (mg g AFDM-1) Coarse: 58.21 (14.33) 41.74 (7.81)  89.54 (11.08) 59.86 (11.19)  Fine: 64.90 (11.26) 41.26 (10.10)  103.89 (5.24) 47.10 (4.11)  Shredder density (no. g AFDM-1)  13.32 (2.45) 7.41 (1.67)  11.31 (3.46) 5.97 (2.27)  Shredder taxonomic richness  3.72 (0.36) 3.42 (0.58)  1.13 (0.24) 1.80 (0.30)  Shredder community evenness  0.47 (0.05) 0.44 (0.05)  0.58 (0.21) 0.55 (0.14) Maple   Acer circinatum (H) Acer saccharum (A)  Acer circinatum (A) Acer saccharum (H)  λF (10-3 degree-day-1)  0.47 (0.23) 0.23 (0.006)  0.47 (0.22) 0.007 (0.003)  kf (10-3 degree-day-1)  0.92 (0.13) 0.38 (0.004)  1.19 (0.007) 0.58 (0.005)  λF/kc  0.29 (1.59) 0.33 (1.58)  0.19 (1.08) 0.19 (1.11)  Coarse: 0.91 (0.20) 0.23 (0.10)  1.29 (0.27) 0.46 (0.13) 191    Variable   British Columbia     Ontario    Algal biomass (mg m-2) Fine: 1.66 (0.49) 0.23 (0.09)  0.76 (0.17) 0.58 (0.09)  Fungal biomass (mg g AFDM-1) Coarse: 134.99 (30.29) 80.99 (10.99)  133.44 (14.92) 78.19 (12.36)  Fine: 128.01 (10.31) 88.17 (17.12)  137.10 (5.94) 96.67 (9.37)  Shredder density (no. g AFDM-1)  12.12 (2.04) 3.72 (0.98)  10.39 (2.82) 3.82 (1.86)  Shredder taxonomic richness  2.03 (0.19) 1.60 (0.40)  1.90 (0.35) 1.17 (0.33)  Shredder community evenness  0.61 (0.13) 0.80 (0.13)  0.48 (0.09) 0.76 (0.13) Cedar   Thuja plicata (H) Thuja occidentalis (A)  Thuja plicata (A) Thuja occidentalis (H)  kf (10-3 degree-day-1)  0.50 (0.005) 0.41 (0.004)  0.56 (0.003) 0.97 (0.008)  Fungal biomass (mg g AFDM-1)  51.43 (4.47) 56.86 (3.95)  40.21 (4.92) 28.67 (4.13) 192  Table B.4 Observed and estimated (from resampling iterations) home-field advantage index (HFAI) for temperature-corrected fragmentation rate (λF), breakdown rate in fine-mesh bags (kf), algal and fungal biomass on leaf discs in coarse- and fine-mesh bags, and shredder density, taxonomic richness and community evenness of two or three litter pairs incubated in the study streams of British Columbia and Ontario, Canada. The lower and upper 95th percentile confidence limits (95% CL) of estimated HFAI are given in parentheses.  Variable (unit)   Observed HFAI   Estimated HFAI  Alder Maple Cedar  Alder Maple Cedar     Median 95% CL Median 95% CL Median 95% CL λF (10-3 degree-day-1)  -15.8 -30.1 –  -17.9 (-63.5, 89.4) -11.4 (-47.7, 21.1) – – kf (10-3 degree-day-1)  14.2 7.9 45.2  13.1 (-5.6, 36.3) 7.2 (-12.2, 29.3) 44.8 (24.7, 69.3) Algal biomass (mg m-2) Coarse: -29.7 20.7 –  -29.2 (-63.2, 15.5) 19.5 (-25.6, 109.8) – –  Fine: -26.2 91.1 –  -27.1 (-51.1, 12.0) 92.6 (17.2, 224.0) – – Fungal biomass (mg g AFDM-1) Coarse: -3.3 1.1 –  -3.4 (-36.1, 33.1) 0.8 (-25.3, 41.4) – –  Fine: -14.2 -1.1 -19.6  -14.3 (-35.8, 12.7) -1.8 (-18.8, 19.1) -19.9 (-33.9, -1.5) Shredder density (no. g AFDM-1)  -2.4 3.1 –  -3.2 (-38.0, 63.6) 2.8 (-33.4, 76.5) – – Shredder taxonomic richness   31.7 -11.4 –   14.1 (-19.0, 55.6) -17.5 (-38.7, 14.2) – – Shredder community evenness  0.9 9.3 –  3.5 (-25.9, 46.3) 8.9 (-22.9, 50.1) – – 193  Table B.5 Summary statistics of standardised effect sizes (Hedges’ d), including their associated lower and upper 95% confidence limits (95% CL), for the effects of litter origin on stream litter breakdown rates (λF, kf), algal and fungal biomass on leaf discs in coarse- and fine-mesh bags, and shredder density, taxonomic richness and community evenness of two or three litter pairs incubated in the study streams of British Columbia and Ontario, Canada. Significantly positive effect sizes (i.e., 95% CI not overlapping zero) are in bold typeface. 194  Variable Alder Maple Cedar Hedges' d SE N 95% CL Hedges' d SE N 95% CL Hedges' d SE N 95% CL Ontario:              λF 0.106 0.679 5 (-1.176, 1.388) -1.170 0.616 6 (-2.754, 0.413) – – – – kf  -2.377 1.035 5 (-5.252, 0.497) -3.779 1.303 6 (-7.128, -0.429) 2.673 1.018 6 (0.054, 5.291) Algal biomass Coarse: -0.296 0.380 5 (-1.350, 0.759) -1.157 0.772 6 (-3.140, 0.827) – – – – Fine: -1.143 0.795 5 (-3.349, 1.063) -0.471 0.483 6 (-1.712, 0.769) – – – – Fungal biomass Coarse: -1.077 0.641 5 (-2.857, 0.703) -0.799 0.510 6 (-2.109, 0.512) – – – – Fine: -4.852 1.801 5 (-9.850, 0.147) -1.057 0.646 6 (-2.718, 0.604) -0.954 0.574 6 (-2.431, 0.523) Shredder density -0.737 0.682 5 (-2.629, 1.155) -1.151 0.520 6 (-2.488, 0.185) – – – – Shredder taxonomic richness 0.818 0.510 5 (-0.597, 2.234) -0.945 0.428 6 (-2.046, 0.157) – – – – Shredder community evenness -0.047 0.559 5 (-1.599, 1.504) 0.945 0.647 6 (-0.718, 2.607) – – – – British Columbia: λF  -0.356 0.248 6 (-0.994, 0.282) 0.561 0.352 4 (-0.560, 1.682) – – – – kf  6.388 2.074 6 (1.055, 11.722) 2.489 1.229 5 (-0.923, 5.902) 0.775 0.500 5 (-0.613, 2.163) Algal biomass Coarse: -0.584 0.293 6 (-1.337, 0.170) 1.821 0.989 5 (-0.923, 4.566) – – – – Fine: 0.490 0.576 6 (-0.992, 1.972) 1.701 0.901 5 (-0.800, 4.203) – – – – Fungal biomass Coarse: 0.541 0.625 6 (-1.066, 2.148) 1.600 0.670 5 (-0.260, 3.459) – – – – Fine: 0.790 0.281 6 (0.067, 1.513) 1.720 0.662 5 (-0.117, 3.558) -0.516 0.381 5 (-1.572, 0.541) Shredder density 1.036 0.548 6 (-0.372, 2.444) 2.078 0.937 5 (-0.524, 4.680) – – – – Shredder taxonomic richness -0.172 0.471 6 (-1.384, 1.040) 0.293 0.779 5 (-1.868, 2.455) – – – – Shredder community evenness 0.245 0.505 6 (-1.054, 1.544) -0.599 0.433 5 (-1.799, 0.602) – – – –  195  Appendix C  Supporting information for Chapter 5    Figure C.1 (a) Hydrograph of East Creek in Malcolm Knapp Research Forest (MKRF), British Columbia, during 1998. Closed circles refer to daily discharge values identified as individual peak flows in this study, which were manipulated in the heuristic modelling of stream coarse particulate organic matter (CPOM) standing stocks. Open circles refer to unmanipulated daily discharge values. (b) Daily precipitation recorded at the climate station in MKRF, British Columbia (station name: Haney UBC RF Admin; climate ID: 1103332).196    Figure C.2 Modelled effect size of logging-associated increases in stream temperature (T) on average daily stream shredder biomass (i.e., ln[shredder biomass at the increased temperature level/shredder biomass at temperature under unharvested (control) conditions]) and average daily shredder consumption of CPOM, in East Creek, British Columbia. Effect size for each heuristic scenario of temperature increase averaged across ten model simulations, and its associated 95% confidence interval (represented by error ranges), are shown. Effect sizes are all significant as their corresponding error bars do not overlap zero (the solid line).197  Table C.1 Summary of changes in stream peak flows after forest harvesting in rain-dominated, low-elevation watersheds in North America.  Country Site location Latitude Elevation (m) Harvest type (% of watershed area harvested) Change in peak flows (%) Time after harvest (year) Reference USA Catchment I, Holly Springs National Forest, Mississippi 35°N ~100-300 CC (100) 29 0-2 Ursic (1991); Sun et al. (2004) USA Catchment III, Holly Springs National Forest, Mississippi 35°N ~100-300 CC (100) 46 0-2 Ursic (1991); Sun et al. (2004) USA Watershed 1, 2 and 3, Alto Experimental Watersheds, Texas 35°N ~90-131 CC (100); with shearing of vegetation 345 1 Blackburn et al. (1986); McBroom et al. (2003) 367 2 250 3 240 4 USA Watershed 5, 7 and 9, Alto Experimental Watersheds, Texas 35°N ~90-131 CC (100); with roller chopping of vegetation 127 1 Blackburn et al. (1986); McBroom et al. (2003) 33 (n.s.) 2 50 (n.s.) 3 20 (n.s.) 4 USA Watershed 1, Fernow Experimental Forest, Virginia 39°N 655-869 CC (86) 7 0-3 Reinhart et al. (1963) USA BAN, Caspar Creek, California 39°N 37-320 CC (95) 21 † 0-2 Henry (1998); Ziemer (1998) USA CAR, Caspar Creek, California 39°N 37-320 CC (96) 19 † 0-2 Henry (1998); Ziemer (1998) USA EAG, Caspar Creek, California 39°N 37-320 CC (100) 27 † 0-2 Henry (1998); Ziemer (1998) USA GIB, Caspar Creek, California 39°N 37-320 CC (100) 39 † 0-2 Henry (1998); Ziemer (1998) 198  Country Site location Latitude Elevation (m) Harvest type (% of watershed area harvested) Change in peak flows (%) Time after harvest (year) Reference USA KJE, Caspar Creek, California 39°N 37-320 CC (97) 28 † 0-2 Henry (1998); Ziemer (1998) USA Andrews 1, H. J. Andrews Experimental Forest, Oregon  44°N 460-990 CC (100) 31 0-10 Jones (2000) USA Andrews 3, H. J. Andrews Experimental Forest, Oregon  44°N 490-1070 PC (25) 19 0-10 Jones (2000) USA Andrews 6, H. J. Andrews Experimental Forest, Oregon  44°N 880-1010 CC (100) 25 0-10 Jones (2000) USA Andrews 7, H. J. Andrews Experimental Forest, Oregon  44°N 910-1020 SC (50) 25 0-10 Jones (2000) USA Andrews 10, H. J. Andrews Experimental Forest, Oregon  44°N 425-700 CC (100) 37 0-10 Jones (2000) USA Coyote 1, Umqua National Forest, Oregon  44°N 730-1065 SC (50) 36 0-5 Jones (2000) USA Coyote 2, Umqua National Forest, Oregon  44°N 730-1065 PC (30) 32 0-5 Jones (2000) USA Coyote 3, Umqua National Forest, Oregon  44°N 730-1065 CC (100) 45 0-5 Jones (2000) USA Fox 1, Mount Hood National Forest, Oregon  44°N 840-925 PC (25) 12 0-10 Jones (2000) USA Fox 3, Mount Hood National Forest, Oregon  44°N 840-950 PC (25) 1 0-10 Jones (2000) USA Deer Creek, Alsea Watershed, Oregon 44°N 134-488 PC (25) 4 (n.s.) 0-2 Harr et al. (1975) 199  Country Site location Latitude Elevation (m) Harvest type (% of watershed area harvested) Change in peak flows (%) Time after harvest (year) Reference USA Deer Creek 2, Alsea Watershed, Oregon 44°N 134-488 PC (30) 19 (n.s.) 0-2 Harr et al. (1975) USA Deer Creek 3, Alsea Watershed, Oregon 44°N 134-488 PC (65) 44 0-2 Harr et al. (1975) USA Deer Creek 4, Alsea Watershed, Oregon 44°N 134-488 CC (90) 29 (n.s.) 0-2 Harr et al. (1975) USA Needle Branch, Alsea Watershed, Oregon 44°N 134-488 CC (82) 23 0-2 Harr et al. (1975) Canada Basin 6, Ruisseau des Eaux-Volées Experimental Watershed, Quebec 47°N 850 PC (31) n.s. 0-8 Plamondon et al. (1998) Canada Watershed 7A, Ruisseau des Eaux-Volées Experimental Watershed, Quebec 47°N Mean: 871 CC (85) 54 ‡ 0-4 Guillemette et al. (2005) Canada Watershed 7.2, Ruisseau des Eaux-Volées Experimental Watershed, Quebec 47°N ~830-940 CC (50) n.s. 0-2 Tremblay et al. (2008) Canada Watershed 7.3, Ruisseau des Eaux-Volées Experimental Watershed, Quebec 47°N ~830-940 CC (50) n.s. 0-2 Tremblay et al. (2008) Canada Watershed 7.5, Ruisseau des Eaux-Volées Experimental Watershed, Quebec 47°N ~830-940 CC (50) n.s. 0-2 Tremblay et al. (2008) 200  Country Site location Latitude Elevation (m) Harvest type (% of watershed area harvested) Change in peak flows (%) Time after harvest (year) Reference Canada Watershed 7.7, Ruisseau des Eaux-Volées Experimental Watershed, Quebec 47°N ~830-940 CC (50) n.s. 0-2 Tremblay et al. (2008) Canada Watershed 1, UBC Malcolm Knapp Research Forest, British Columbia 49°N 145-455 CC (71) -22 0-2 Cheng et al. (1975) USA Marcell Experimental Forest, Minnesota 48°N 428-438 CC (71) 71 † 0-9 Verry et al. (1983) Canada Subbasin B, Carnation Creek, British Columbia 48°N 8-884 CC (40; over 6 years) n.s. 0-4 Hetherington (1982) Canada Subbasin H, Carnation Creek, British Columbia 48°N 152-305 CC (90) 21 § 0-4 Hetherington (1982) Canada F5, Flume Creek, British Columbia 50°N 505-850 SS (51) 86 1 Hudson (2001) 123 2-3 Canada F4, Flume Creek, British Columbia 50°N 505-850 VR (82) 233 1 Hudson (2001) 230 2-3 Canada Jamieson Creek, British Columbia 51°N 305-1310 CC (19; over 6 years) 13.5 (winter); n.s. (summer) 0-8 Golding (1987) Notes: CC, clearcut; PC: patch cut; SS: strip shelterwood cut; VR: variable-retention cut; n.s.: not significant. † This refers to storm peak events with a return period of 2 years. ‡ This refers to bankfull discharge events with a return period of 1.5 years. § Computed as the difference between the reference stream and the average of two streams in harvested catchments. 201  Table C.2 Summary of changes in stream summer temperature after forest harvesting in rain-dominated, low-elevation watersheds in North America. When specified, summer months during when temperature responses were recorded by individual studies are shown.  Country Site location Latitude Elevation (m) Aspect Bankfull width Harvest type (% of watershed area harvested) Riparian buffer width (m) Temperature variable Change in temperature (°C) Time after harvest (year) Reference USA Flat Branch, Tallulah River Watershed, Georgia 35°N ~900 SW Not given SH (not given) N/A Maximum monthly summer T (June to September) 0.1 0-1 Clinton et al. (2010) USA Leading Ridge Watershed 2, Pennsylvania 39°N 274-442 SSE Not given CC (20) 0 Maximum monthly summer T (June to August) 3.3 0-2 Sopper and Lynch (1970) USA BeeBy stream, Hinkle Creek basin, Oregon 43°N 400-1250 NNW Not given CC (31) 0 Maximum daily summer T 0.7 1 Kibler et al. (2013) USA Clay stream, Hinkle Creek basin, Oregon 43°N 400-1250 NNE Not given CC (38) 0 Maximum daily summer T 1.1 1 Kibler et al. (2013) USA Fenton stream, Hinkle Creek basin, Oregon 43°N 400-1250 N Not given CC (65) 0 Maximum daily summer T -1.6 1 Kibler et al. (2013) USA Russell stream, Hinkle Creek basin, Oregon 43°N 400-1250 NW Not given CC (10) 0 Maximum daily summer T 0.6 1 Kibler et al. (2013) USA Brome, Oregon  43°N 450 N Not given CC (not given) 0 Maximum daily summer T (mid-June to mid-September) -0.7-3.8 1-5 Cole and Newton (2013) 202  Country Site location Latitude Elevation (m) Aspect Bankfull width Harvest type (% of watershed area harvested) Riparian buffer width (m) Temperature variable Change in temperature (°C) Time after harvest (year) Reference USA South Fork Hinkle Creek basin outlet, Oregon 43°N 400-1250 NW Not given CC (14) 0 Maximum daily summer T -0.1 (n.s.) 1 Kibler et al. (2013) USA Deer Creek, Alsea Watershed, Oregon 44°N ~134-488 SW Not given CC (25) 30 (on 1 side only) Maximum monthly summer T (April to October) 2.0 1 Harr et al. (1975); Harris (1977) USA Needle Branch, Alsea Watershed, Oregon 44°N 220 (mean) S Not given CC (82) 0 Maximum monthly summer T (April to October) 5.5 1 Harr et al. (1975); Harris (1977) USA Needle Branch, Alsea Watershed, Oregon 44°N 220 (mean) S Not given CC (60-100; data from three sub-catchments) 15 Maximum daily summer T (7-day moving mean; July to September) 0.3-0.8 (n.s.) 0-3 Bladon et al. (2016) USA NB6, Needle Branch, Alsea Watershed, Oregon 44°N 220 (mean) S Not given CC (89) 15 Maximum daily summer T (7-day moving mean; July to September) n.s. 0-3 Bladon et al. (2017) USA Watershed 1, H.J. Andrews Experimental 44°N 460-490 WNW Not given CC (100) 0 Maximum weekly summer T 5.8 1 Johnson and Jones (2000); 6.2 2 6.4 3 203  Country Site location Latitude Elevation (m) Aspect Bankfull width Harvest type (% of watershed area harvested) Riparian buffer width (m) Temperature variable Change in temperature (°C) Time after harvest (year) Reference Forest, Oregon (June to September) 5.4 4 Jones (2000) USA Watershed 2, H.J. Andrews Experimental Forest, Oregon 44°N 490-1070 NW Not given PC (25) 0; debris flows removed riparian vegetation Maximum weekly summer T (June to September) 3.5 1 Johnson and Jones (2000); Jones (2000) 5.2 2 5.1 3 Not given 4 USA Kibby, Pierce 1, Skinner 1, Maine 45°N 469-724 SSE (Kibby), NW (Pierce 1, N (Skinner 1) 1.9-3.1 CC (not given) 0 Maximum daily summer T (June to August) 3.6 1 Wilkerson et al. (2006) 3.3 2 USA Bald Mt., Caratunk, Skinner 2, Maine 45°N 345-676 NNW (Bald Mt.), SE (Caratunk), NW (Skinner 2) 2.0-3.9 CC (not given) 11 Maximum daily summer T (June to August) 1.5 (n.s.) 1 Wilkerson et al. (2006) 0.4 (n.s.) 2 USA Big Rock, Oregon 45°N 150 W Not given CC (not given) 0 Maximum daily summer T (mid-June to mid-September) -0.1-3.8 1-4 Cole and Newton (2013) USA GS3, Gus, Trask Watershed, Oregon 45°N ~275-1100 W Not given CC (94) 0 Maximum daily summer T (mid-June 3.9 0-3 Bladon et al. (2017) 204  Country Site location Latitude Elevation (m) Aspect Bankfull width Harvest type (% of watershed area harvested) Riparian buffer width (m) Temperature variable Change in temperature (°C) Time after harvest (year) Reference to mid-September) USA GS4, Gus, Trask Watershed, Oregon 45°N ~275-1100 SW Not given CC (91) 0 Maximum daily summer T (mid-June to mid-September) 3.4 0-3 Bladon et al. (2017) USA Marys, Oregon 45°N 400 S Not given CC (not given) 0 Maximum daily summer T (mid-June to mid-September) -0.1-3.4 1-5 Cole and Newton (2013) USA Mass 2, Roxbury, Sanderson, Maine 45°N 371-700 NNE (Mass 2), WNW (Roxbury), E (Sanderson) 2.4-3.8 CC (not given) 23 Maximum daily summer T (June to August) 0.3 (n.s.) 1 Wilkerson et al. (2006) -0.3 (n.s.) 2 USA Mass 1, Pierce 2, UpCup, Maine 45°N 436-672 SSE (Mass 1), W (Pierce 2), S (UpCup) 2.0-4.5 PH (not given); residual basal area reduced by an average of 38% N/A Maximum daily summer T (June to August) 0.9 (n.s.) 1 Wilkerson et al. (2006) 1.0 (n.s.) 2 USA PH1, Pothole, Trask Watershed, Oregon 45°N ~275-1100 S Not given CC (40); RC (37) 17 Maximum daily summer T (mid-June 0.8 0-3 Bladon et al. (2017) 205  Country Site location Latitude Elevation (m) Aspect Bankfull width Harvest type (% of watershed area harvested) Riparian buffer width (m) Temperature variable Change in temperature (°C) Time after harvest (year) Reference to mid-September) USA PH2, Pothole, Trask Watershed, Oregon 45°N ~275-1100 S Not given CC (78) 11 Maximum daily summer T (mid-June to mid-September) 0.6 0-3 Bladon et al. (2017) USA PH4, Pothole, Trask Watershed, Oregon 45°N ~275-1100 SE Not given CC (92) 12 Maximum daily summer T (mid-June to mid-September) 1.0 0-3 Bladon et al. (2017) USA UM2, Upper Mainstem, Trask Watershed, Oregon 45°N ~275-1100 SW Not given CC (83) 8 (25% of stream) Maximum daily summer T (mid-June to mid-September) 2.4 0-3 Bladon et al. (2017) USA UM3, Upper Mainstem, Trask Watershed, Oregon 45°N ~275-1100 W Not given CC (56) 8 (60% of stream) Maximum daily summer T (mid-June to mid-September) 3.3 0-3 Bladon et al. (2017) Canada Sub-catchment 1, Hayward Brook Watershed, New Brunswick 45°N 25-230 W ~1 CC and SW (17) 30-60 Daily mean summer T (June to September) 0.7 1 Bourque and Pomeroy (2001) 206  Country Site location Latitude Elevation (m) Aspect Bankfull width Harvest type (% of watershed area harvested) Riparian buffer width (m) Temperature variable Change in temperature (°C) Time after harvest (year) Reference Canada Sub-catchment 4, Hayward Brook Watershed, New Brunswick 45°N 25-230 SW ~1 CC (25) 60 Daily mean summer T (June to September) 1.9 1 Bourque and Pomeroy (2001) Canada Sub-catchment 5, Hayward Brook Watershed, New Brunswick 45°N 25-230 NW 2-4 CC (3) 30 Daily mean summer T (June to September) 0.3 1 Bourque and Pomeroy (2001) Canada Sub-catchment 6, Hayward Brook Watershed, New Brunswick 45°N 25-230 NE 0.5-1.5 CC (15) 30 Daily mean summer T (June to September) 0.4 1 Bourque and Pomeroy (2001) Canada A Creek, UBC Malcolm Knapp Research Forest, British Columbia 49°N ~110 S 2.3 CC (21) 0 Maximum daily summer T (July to August) 2.1 0-4 Gomi et al. (2006); Lecerf and Richardson (2010) Canada A Creek, UBC Malcolm Knapp Research Forest, British Columbia 49°N 110-160 S 4 CC (21) 0 Maximum summer T 5.0 0-4 Moore et al. (2005b) 207  Country Site location Latitude Elevation (m) Aspect Bankfull width Harvest type (% of watershed area harvested) Riparian buffer width (m) Temperature variable Change in temperature (°C) Time after harvest (year) Reference Canada B Creek, UBC Malcolm Knapp Research Forest, British Columbia 49°N ~110 SSE 1.1 CC (24) 0 Maximum daily summer T (July to August) 0.4 0-4 Gomi et al. (2006); Lecerf and Richardson (2010) Canada C Creek, UBC Malcolm Knapp Research Forest, British Columbia 49°N ~110 S 2.4 CC (21) 10 Maximum daily summer T (July to August) 1.0 0-4 Gomi et al. (2006); Lecerf and Richardson (2010) Canada D Creek, UBC Malcolm Knapp Research Forest, British Columbia 49°N ~180 E 2.1 CC (22) 30 Maximum daily summer T (July to August) -0.2 0-4 Gomi et al. (2006); Lecerf and Richardson (2010) Canada E Creek, UBC Malcolm Knapp Research Forest, British Columbia 49°N ~170 S 0.5 CC (53) 0 Maximum daily summer T (July to August) 0.4 0-4 Gomi et al. (2006); Lecerf and Richardson (2010) Canada Griffith Creek, UBC Malcolm Knapp Research 49°N ~365-405 S 1.3-1.5 PH (12); basal area reduced by 50% N/A Maximum daily summer T (July to August) 1.6-3.0 0-1 Guenther et al. (2014); Lecerf and 208  Country Site location Latitude Elevation (m) Aspect Bankfull width Harvest type (% of watershed area harvested) Riparian buffer width (m) Temperature variable Change in temperature (°C) Time after harvest (year) Reference Forest, British Columbia Richardson (2010) Canada Creek H, UBC Malcolm Knapp Research Forest, British Columbia 49°N ~205 S 4 CC (22) 30 Maximum daily summer T (July to August) 0.4 0-4 Gomi et al. (2006); Lecerf and Richardson (2010) Canada Creek I, UBC Malcolm Knapp Research Forest, British Columbia 49°N ~255 SSE 1.9 CC (21) 0 Maximum daily summer T (July to August) 3.9 0-4 Gomi et al. (2006); Lecerf and Richardson (2010) Notes: CC, clearcut; PC: patch cut; PH: partial harvest; SH: shelterwood harvest; n.s.: not significant 209  Table C.3 Summary of changes in riparian litterfall after forest harvesting in watersheds in North America.  Country Site location Latitude Elevation (m) Harvest type (% of watershed area harvested) Change in litterfall (%) Time after harvest (year) Reference USA Sawmill Branch WS6, North Carolina 35°N Midstream: 770 CC (12%) -28 † 16 Webster et al. (1988); Webster et al. (1990) USA Big Hurricane Branch WS7, North Carolina 35°N Midstream: 853 CC (100%) -98 2 Webster and Waide (1982); Webster et al. (1983); Webster et al. (2014b) -83 3 -52 4 37 6 32 16      13 26  USA Upper Deschutes River, Washington 46°N ~1200-1400 CC (100%) -93 8 Bilby and Bisson (1992); Manga (1996) -93 9 Canada 33, Turkey Lakes Watershed, Ontario 47°N ~400 SeH (N/A) 8 1 Kreutzweiser et al. (2004) -3 2 -2 3 Canada 34M, Turkey Lakes Watershed, Ontario 47°N ~400 SH (N/A) 7 1 Kreutzweiser et al. (2004) 4 2 -1 3 Canada 34L, Turkey Lakes Watershed, Ontario 47°N ~400 DLH (N/A) -41 1 Kreutzweiser et al. (2004) -62 2 2 3 Canada H6, H7, H8, H9, H10, H12, H13, H14, H15, H16, White River Forest 48°N 341-519 CC (26-59%); with minimum 30-m riparian buffers ~ -6 (n.s.) 7-17 Musetta-Lambert et al. (2017) 210  Country Site location Latitude Elevation (m) Harvest type (% of watershed area harvested) Change in litterfall (%) Time after harvest (year) Reference Management Area, Ontario Canada Streams B, E, and I, UBC Malcolm Knapp Research Forest, British Columbia 49°N 110-320 CC (33%) ~ -91 1 Kiffney and Richardson (2010) ~ -78 2 ~ -79 6 ~ -47 7 ~ -11 8 Canada Streams C, F, and G, UBC Malcolm Knapp Research Forest, British Columbia 49°N 110-325 CC (23%); with 10-m riparian buffers ~ -2 1 Kiffney and Richardson (2010) ~ 6 2 ~ -14 6 ~ 6 7 ~ 37 8 Canada South Creek and Stream H, UBC Malcolm Knapp Research Forest, British Columbia 49°N 175-320 CC (18%); with 30-m riparian buffers ~ 11 1 Kiffney and Richardson (2010) ~ 44 2 ~ 14 6 ~ -6 7 ~ 74 8 Notes: CC, clearcut; DLH: diameter limit harvest (all trees >10 cm felled; average of 89% basal area removed); SeH: selection harvest (average of 29% basal area removed); SH: shelterwood harvest (average of 42% basal area removed). † Computed as the difference between the reference stream and the average of two streams in harvested catchments. 211  Table C.4 Summary statistics of the distributions of changes in peak flows (in %) and stream summer temperature (°C) due to forest harvesting in low-elevation watersheds in North America. Multiple post-harvest measurements taken at each site (or set of sites) were averaged. Post-harvest changes that were determined to be insignificant were assigned zero values. Post-harvest response values of peak flows were first grouped by percentage ranges of watershed area harvested (0-40%; 41-80%; 81-100%), and response values of stream summer temperature measured in the presence of riparian buffers were excluded, prior to obtaining the summary statistics of distributions. Number of sites in each distribution is shown in parentheses.    Peak flows Stream summer temperature (n = 22)  Percentage of watershed area harvested (%)   0-40 (n = 8) 41-80 (n = 6) 81-100 (n =19) Minimum value 0.0 -22.0 0.0 -1.6 25th percentile 0.0 27.8 22.0 0.6 50th percentile 6.5 40.0 29.0 2.0 75th percentile 14.9 64.3 45.5 3.8 Maximum value 32.0 104.5 301.0 6.0 212  Appendix D  Supporting information for Chapter 6  Table D.1 Proportion of presence of shredder taxa in sites within each forest treatment, which were collected in autumn 2006 (i.e., data from Lecerf and Richardson (2010)) and 2013 (i.e., present study). A blank cell indicates the absence of taxon from all reaches of the same treatment. Note that all leuctrid and nemourid stoneflies were identified to the genus level in 2013.  Taxon Reference 30-m reserves 10-m reserves No reserve Thinning 2006 2013 2006 2013 2006 2013 2006 2013 2006 2013 Capniidae (early instars) 2/3 – – – 1/3 – 2/4 – 2/3 – Leuctridae Despaxia augusta 3/3 3/3 1/3 1/3 1/3 2/3 1/4 4/4 3/3 3/3 Leuctridae Moselia 2/3 – – – – – – – 3/3 – Leuctridae (juvenile) 3/3 – 1/3 – 2/3 – 2/4 – 3/3 – Nemouridae Soyedina 3/3 – 3/3 – 2/3 – 2/4 – 3/3 – Nemouridae Visoka cataractae 1/3 – – – – – – – 1/3 – Nemouridae Zapada 3/3 3/3 2/3 3/3 2/3 2/3 1/4 2/4 3/3 2/3 Nemouridae (early instars) 3/3 – 3/3 – 2/3 – 2/3 – 3/3 – Peltoperlidae Yoraperla 3/3 3/3 2/3 – – – 1/4 2/4 3/3 1/3 Pteronarcyidae Pteronarcys 1/3 – 1/3 – 1/3 – 1/4 – – – Limnephilidae Cryptochia 1/3 1/3 1/3 – 1/3 – – – – 1/3 Limnephilidae Onocomoecus – – – 1/3 – – – – – – Lepidostomatidae Lepidostoma 3/3 3/3 2/3 3/3 1/3 3/3 1/4 3/4 2/3 3/3 Tipulidae Tipula 3/3 1/3 – 1/3 1/3 –  1/4 2/4 3/3 1/3 213  Table D.2 Mean (± SE) daily and temperature-corrected decomposition of alder leaves, shredder density and rarefied taxonomic richness between studies conducted in 2006 (i.e., data from Lecerf and Richardson (2010)) and in 2013 (i.e., present study). Values presented in the table are identical to those illustrated in Fig. 6.1.  Variable Reference 30-m reserves 10-m reserves No reserve Thinning 2006 2013 2006 2013 2006 2013 2006 2013 2006 2013 k (day-1) 0.0088 (0.0020) 0.0066 (0.0016) 0.0021 (0.0002) 0.0056 (0.0003) 0.0040 (0.0010) 0.0070 (0.0008) 0.0038 (0.0007) 0.0058 (0.0010) 0.0038 (0.0005) 0.0089 (0.0017) k (degree day-1) 0.0023 (0.0006) 0.0013 (0.0003) 0.0004 (0.00004) 0.0011 (0.0001) 0.0008 (0.0002) 0.0014 (0.00008) 0.0007 (0.0001) 0.0011 (0.0002) 0.0008 (0.00009) 0.0014 (0.0002) Shredder density (no. m-2) 823.5 (169.8) 82.2 (27.8) 203.7 (143.1) 63.0 (41.1) 344.4 (302.5) 40.7 (18.7) 112.9 (71.2) 72.2 (31.1) 1123.5 (571.1) 200.0 (81.8) Rarefied shredder richness 5.0 (0.1) 3.4 (0.2) 3.7 (0.4) 2.1 (0.1) 3.8 (0.5) 3.0 (0.0) 3.3 (1.3) 3.0 (0.2) 5.2 (0.8) 2.7 (0.1) 214    Fig. D.1 Comparison of the effects of forest treatments on unrarefied shredder richness between studies conducted in 2006 (i.e., data from Lecerf and Richardson (2010)) and in 2013 (i.e., present study). Letters indicate homogeneous groups derived from ANOVA and planned contrasts uniquely for each study (see Table 6.2). Error bars denote standard errors. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0376538/manifest

Comment

Related Items