Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Long-term mapping of ecosystem services in a river-floodplain system Tomscha, Stephanie Anne 2015

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

Item Metadata

Download

Media
24-ubc_2016_february_tomscha_stephanie.pdf [ 4.43MB ]
Metadata
JSON: 24-1.0220636.json
JSON-LD: 24-1.0220636-ld.json
RDF/XML (Pretty): 24-1.0220636-rdf.xml
RDF/JSON: 24-1.0220636-rdf.json
Turtle: 24-1.0220636-turtle.txt
N-Triples: 24-1.0220636-rdf-ntriples.txt
Original Record: 24-1.0220636-source.json
Full Text
24-1.0220636-fulltext.txt
Citation
24-1.0220636.ris

Full Text

LONG-TERM MAPPING OF ECOSYSTEM SERVICES IN A RIVER-FLOODPLAIN SYSTEM by  Stephanie Anne Tomscha  B.A., Luther College, 2007  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  November 2015  © Stephanie Anne Tomscha, 2015 ii  Abstract Humans derive a wide range of benefits from ecosystems, known as ecosystem services (ES). At the nexus of land and water, floodplains are particularly important for providing ES. Recently, problematic declines in ES have motivated research to better understand their spatial distributions. However, the temporal dynamics of critically important floodplain-specific ES remain poorly understood. These spatial and temporal dynamics as well as trade-offs that occur when management enhances one ES at the expense of others are particularly germane as a warming climate alters river flows. Landscape history is foundational to elucidating these dynamics. Here, I explore the importance of landscape history for understanding the historical, contemporary, and future distributions of ES in the Wenatchee watershed, central Washington State. Using several widely-used datasets in novel ways, my dissertation has five primary objectives (1) quantify the relative importance of different landscape positions for frontier settlers, (2) map change in ES from 1949-2006 using high-resolution imagery, (3) enhance understanding of ES interactions by incorporating change in ES over time, (4) explore the spatial distribution of floodplain-specific ES, and (5) conceptualize shifts in ES under future climates. I found riparian zones and floodplains were disproportionately important for frontier settlement, setting the stage to explore floodplain-specific ES in more detail. ES were dynamic from 1949-2006, largely driven by increasing urbanization and forest densification. Next, I showed how history can provide important insights into ES interactions. Finally, I also found floodplain ES varied considerably with floodplain position. Analyses over broad time frames and at fine spatial scales greatly enhance our understanding of ES dynamics, highlighting the need for long-term monitoring for ES, especially as ES continue to interact under future climates.  iii  Preface This dissertation is comprised of four scientific papers of which I am the first author. The project scope was originally defined by my advisor, Sarah Gergel. Historic and modern aerial photography was provided by Matt Tomlinson et al. 2011. I performed all research, data analyses, interpretation of results and manuscript preparation. My co-authors provided advice on methodology and editorial changes.  One publication has result from this research (reprinted with permission from publishers):  Chapter 2: Tomscha SA, Gergel SE (2015) Historic land surveys present opportunities for reconstructing frontier settlement patterns in North America. Landscape Ecology. In addition, Chapters 3-5 were co-written by myself and my advisor SE Gergel and will be submitted to leading scientific journals. Chapter 6 was co-written by myself, Sarah Gergel, and Timothy Beechie.  iv  Table of contents  Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iii Table of contents .......................................................................................................................... iv List of tables....................................................................................................................................x List of figures ............................................................................................................................... xii List of abbreviations ................................................................................................................. xvii Acknowledgements .................................................................................................................. xviii Dedication ................................................................................................................................... xix Chapter 1: Introduction ................................................................................................................1 1.1 Spatial approaches for understanding ES ....................................................................... 3 1.2 Long-term approaches needed ........................................................................................ 7 1.3 Approaches and objectives ............................................................................................. 8 Chapter 2: Historically, how does frontier settlement vary with landscape position? ......... 11 Chapter 3: How has land cover and ES capacity changed over time? .................................. 11 Chapter 4: How do baselines improve our understanding of ES interactions?..................... 12 Chapter 5: How can river-floodplain concepts improve our understanding of the spatial variability of ES capacity? .................................................................................................... 12 Chapter 6 (Conclusion): How might floodplain ES interact under diminishing snowpack conditions? ............................................................................................................................ 12 Chapter 2: Historic land surveys present opportunities for reconstructing frontier settlement patterns in North America........................................................................................14 v  2.1 Background ................................................................................................................... 14 2.1.1 Historically, which anthropogenic linear disturbances co-occur? ............................ 15 2.1.2 Did historical disturbances vary with landscape position? ....................................... 16 2.2 Methods......................................................................................................................... 16 2.2.1 Study site ................................................................................................................... 16 2.2.2 Data entry .................................................................................................................. 18 2.2.3 Fraudulent surveys .................................................................................................... 20 2.2.4 Landscape position defined by topographic wetness index (TWI) ........................... 20 2.3 Statistical analysis ......................................................................................................... 22 2.3.1 Historically, which anthropogenic linear disturbances co-occur? ............................ 22 2.3.2 Did historical disturbances vary with landscape position? ....................................... 24 2.4 Results ........................................................................................................................... 25 2.4.1 Historically, which anthropogenic linear disturbances co-occur? ............................ 25 2.4.2 Did historical disturbances vary with landscape position? ....................................... 25 2.5 Discussion ..................................................................................................................... 29 2.5.1 Ecological consequences of linear anthropogenic disturbance ................................. 29 2.5.2 Legacies of historical floodplain fragmentation ....................................................... 31 2.6 Conclusion .................................................................................................................... 32 Chapter 3: Long-term changes in the capacity of a river-floodplain to provide ecosystem services ..........................................................................................................................................33 3.1 Introduction ................................................................................................................... 33 3.1.1 Aerial photography interpretation as a tool to track change in ES ........................... 35 3.2 Methods......................................................................................................................... 36 vi  3.2.1 Study Site .................................................................................................................. 36 3.2.2 Aerial photographs and floodplain delineation ......................................................... 38 3.2.3 Land cover classification .......................................................................................... 39 3.2.4 Mapping ES capacity ................................................................................................ 44 3.2.4.1 Orchard and forage production ......................................................................... 44 3.2.4.2 Above-ground carbon storage ........................................................................... 45 3.2.4.3 Paddle routes ..................................................................................................... 46 3.2.4.4 Fish Capacity Index .......................................................................................... 47 3.2.4.4.1 Normalized wood importance index ........................................................... 48 3.2.4.4.2 Fish habitat structures.................................................................................. 49 3.3 Analyses ........................................................................................................................ 50 3.3.1 Land cover change and accuracy assessment ........................................................... 50 3.3.2 ES capacity change ................................................................................................... 51 3.4 Results ........................................................................................................................... 52 3.4.1 How have land cover and ES capacity changed over time? ..................................... 52 3.4.1.1 Changes in land cover ....................................................................................... 52 3.4.1.2 Changes in ES capacity ..................................................................................... 54 3.5 Discussion ..................................................................................................................... 57 3.5.1 Legacies of historical ES .......................................................................................... 61 3.6 Conclusion .................................................................................................................... 62 Chapter 4: Ecosystem service interactions misunderstood without landscape history .........64 4.1 Introduction ................................................................................................................... 64 4.1.1 Trade-offs or synergies may go undetected .............................................................. 68 vii  4.2 Methods......................................................................................................................... 69 4.2.1 Study site ................................................................................................................... 69 4.2.2 Mapping ES capacity ................................................................................................ 70 4.2.3 ES interaction analyses ............................................................................................. 70 4.3 Results ........................................................................................................................... 72 4.3.1 High agreement between space-for-time at different timeframes............................. 72 4.3.2 Little agreement between space-for-time and change-over-time approaches ........... 72 4.3.3 Missed trade-offs and synergies using space-for-time approach .............................. 73 4.3.4 Interactions detected where none occur over time .................................................... 73 4.3.5 ES positively spatially autocorrelated ....................................................................... 73 4.3.6 Spatial autocorrelation decreased number of significant interactions ...................... 74 4.4 Discussion ..................................................................................................................... 78 4.4.1 Different landscape snapshots, different ES interactions characterized ................... 78 4.4.2 Landscape history matters: Even deeper baselines needed ....................................... 79 4.4.3 Space-for-time approach can misconstrue ES interactions ....................................... 80 4.4.4 Data aggregation can obscure trade-offs................................................................... 83 4.5 Approaches for incorporating spatial autocorrelation into ES interactions research .... 83 4.6 Implications................................................................................................................... 84 4.6.1 Missed synergies are missed opportunities ............................................................... 84 4.6.2 Missed trade-offs may result in unexpected ES declines.......................................... 84 4.7 Conclusion .................................................................................................................... 85 Chapter 5: River-floodplain concepts enhance understanding of spatial variability in ecosystem services ........................................................................................................................86 viii  5.1 Methods......................................................................................................................... 89 5.1.1 Study site ................................................................................................................... 89 5.1.2 Aerial photographs and floodplain delineation ......................................................... 91 5.1.3 Reach type identification .......................................................................................... 91 5.1.4 Longitudinal river-floodplain position ...................................................................... 92 5.1.5 Land-cover classification .......................................................................................... 92 5.1.6 ES quantiles .............................................................................................................. 93 5.2 Analyses ........................................................................................................................ 93 5.3 Results ........................................................................................................................... 94 5.3.1 Individual ES vary with longitudinal position .......................................................... 94 5.3.2 Individual ES vary with reach type ........................................................................... 96 5.3.3 Diversity of ES capacity varies with longitudinal position and reach type .............. 98 5.4 Discussion ................................................................................................................... 100 5.4.1 River-floodplain concepts can advance understanding of ES spatial distributions 101 5.5 Implications for river-floodplain restoration .............................................................. 103 5.6 Linking ecosystem service and process-based approaches ......................................... 104 5.7 How spatial patterns of ES can inform process-based restoration .............................. 105 5.8 Conclusion .................................................................................................................. 106 Chapter 6: Conclusions and future research directions .........................................................107 6.1 Caveats and cautions: Limitations of historical datasets and aerial photography ...... 109 6.1.1 Coarse-resolution GLO datasets ideal for describing general settlement patterns . 109 6.1.2 Limited information on linkages between landscape features and ES ................... 109 ix  6.2 Future research directions and applications: ES dynamics and interactions under future drivers of change ..................................................................................................................... 111 6.3 Climate change and anticipated shifts in hydrologic regimes .................................... 113 6.4 Impacts of climate change on river and floodplain ES interactions ........................... 117 6.5 Trade-offs among ES .................................................................................................. 117 6.6 Cascading effects ........................................................................................................ 119 6.7 Reinforcing feedbacks ................................................................................................ 121 6.8 Geographical shifts in ES............................................................................................ 124 6.8.1 Time lags necessitate long-term management strategies ........................................ 125 6.9 In summary ................................................................................................................. 126 Bibliography ...............................................................................................................................127 Appendix .....................................................................................................................................145  x  List of tables  Table 1.1 Examples of the many types of benefits that ecosystems provide. The Millennium Ecosystem Assessment categorizes these services into provisioning, regulating, supporting, and cultural services. Provisioning services are products and goods from ecosystems. Regulating services are the benefits from stable ecosystem processes. Supporting ecosystem services provide services necessary for all other ecosystem services.  Cultural ecosystem services are the nonmaterial benefits that we gain from ecosystems (MA 2005). ................................................... 1 Table 2.1 Topographic Wetness index was mapped across the Wenatchee watershed and divided into five landscape positions based on natural breaks. I explain percent of landscape in of each landscape position. My descriptions are analogous to those used in aerial photograph interpretation (MOF 2011). ........................................................................................................... 22 Table 3.1 Examples of land-cover classification. Land-cover classes were manually digitized at a 1:4000 scale. Patches less than 0.5 ha were not distinguished. The red circles are characteristic examples of land-cover types at each timeframe. For forest and shrub subclasses, the aerial photographer interpreter gave an estimation of their certainty: 1= Certain, 2 = Fairly Certain, 3 = Uncertain. The mixed conifer and broadleaf classes in 1949 were most uncertain. ..................... 40 Table 3.2 Above-ground carbon storage for different land-cover types. ...................................... 46 Table 3.3 A transition matrix converted to an annual time-step shows the persistence (in bold) of different land-cover types in the Wenatchee floodplain based on 5000 random points. Urban land-cover was the most persistent followed by forests. Fields were the most likely to be converted for urban uses followed by orchards. MC—Mostly Conifer, MCB—Mixed conifer broadleaf. ...................................................................................................................................... 54 xi  Table 3.4 Changes in ES summarized at the reach level (Wilcoxon paired-sample test). Significant results are shown in bold. ES were normalized from 0-1 using the maximum and minimum values from both time periods. All changes were significant at the reach level with the exception of paddle route quality and slow/stagnant channels. .................................................... 57 Table 4.1. ES were positively spatially autocorrelated at each time frame. ES change was also positively autocorrelated, but to a lesser degree. *Significant at 0.05 .......................................... 74 Table 4.2 The types of interactions identified using two mapping methods: space-for-time substitution (at both 1949 and 2006) and change-over-time (∆ES). Pairwise Spearman’s rank correlation result. T—Trade-off, S—Synergy, N—No interaction. *Significant at 0.05 ............. 75 Table 5.1 Shows weighted mean landscape position for each ecosystem service in 2006 in the first column. Orchards have the lowest median position ponds and wetlands have the highest weighted mean. Shows if the mean longitudinal position along the floodplain is different among pairs of services (Significance of pairwise Wilcoxon rank-sum test). ES that are significantly different are shown in bold. .......................................................................................................... 94 Table 5.2 Shows differences in fluvial-geomorphic reach type by individual service (Kruskal-Wallis rank sum test followed by post-hoc, Pairwise Wilcoxon rank-sum test with Bonferroni corrections). I=Island braided, M= Meandering, PB = Plane bed, PR= Pool/Riffle. Significant differences are shown in bold. ...................................................................................................... 97 Table 5.3 Shows p-values for multiple pair-wise comparisons (Wilcoxon rank-sum test with Bonferroni corrections) for differences in diversity (Simpson’s Index), as well as the median diversity by reach type in the first column. Significant differences are highlighted. ................. 100  xii  List of figures Figure 1.1 Shows the importance of high-resolution imagery in capturing river-floodplain variability. Panel A and C show mid-resolution Landsat imagery (30m resolution) at the scale of 1:50,000 and 1:4,000 respectively, while panels B and D show 1-m aerial photography at the same scales. In panel D, vegetated islands in the river channel and riparian vegetation can easily be distinguished from agricultural areas. This contrasts panel C where very little can be discerned. ........................................................................................................................................ 5 Figure 1.2 Longitudinal and lateral flows drive the distribution of river-floodplain patterns. ....... 6 Figure 1.3 Timeline and diagram of my approach, and the overall land cover transitions (orange) associated with each chapter and its driving question. ................................................................. 10 Figure 2.1 The Wenatchee watershed, a sub-watershed of the Columbia River basin, in central Washington State. ......................................................................................................................... 18 Figure 2.2 Sampling design for disturbance data, showing how surveyors would have recorded a road transecting a township. ......................................................................................................... 19 Figure 2.3 Grid shows railway and road traversing a landscape co-occurring on one section line. This diagram represents 84 section lines. A contingency table for this diagram would show 69 lines with no disturbances, 9 with only roads, 9 with railway disturbances and 1 with co-occurring roads and railways. Section line-level data is used in the analysis. .............................. 24 Figure 2.4 Map of Wenatchee watershed identifying landscape positions defined by the topographic wetness index. ........................................................................................................... 26 Figure 2.5 Visualizes how disturbances co-occurred in the early-settlement Wenatchee landscape. Significantly co-occurring disturbances are connected with a solid black line (p<0.005). Variance ratio values are shown for co-occurring disturbances. Values greater > 1 xiii  mean disturbances co-occur. Roads were associated with three other disturbances while trails were not associated with any other disturbance. ........................................................................... 27 Figure 2.6 Difference in disturbance percentages by landscape position versus the percent surveyed area by landscape position. More disturbances than expected occurred in the riparian zones and terraces while fewer than expected disturbances occurred in lower slopes, upper slopes, and ridges. ......................................................................................................................... 28 Figure 3.1 The Wenatchee system floodplain situated in central Washington State. ................... 37 Figure 3.2 Timeline of important land use and resource development events in the Wenatchee floodplain. ..................................................................................................................................... 38 Figure 3.3 Land-cover change in hectares across the Wenatchee system floodplain. The greatest total change was an increase urban land-cover type, followed by a decrease in total orchard, fields, shrub cover. While overall forest area only declined slightly (118 ha), mixed conifer broadleaf forests declined, and conifer forests increased. Very minor changes in area in water (17 ha) and rock/snow (19 ha) were also detected (not shown). ......................................................... 53 Figure 3.4 Changes in ES capacity from 1949-2006 across the Wenatchee system floodplain. The river-floodplain’s capacity to provide four of five services examined declined. Carbon storage increased only slightly. ................................................................................................................. 55 Figure 4.1 Hypothesized interactions among floodplain ES. Different drivers can lead to interactions such as trade-offs and synergies among ES. Adapted from Bennett et al. 2009. ...... 64 Figure 4.2 Boxes represent scatter plots showing the relationship between two ES. Positively correlated ES are considered synergies, negatively correlated ES are trade-offs, and uncorrelated ES do not interact. S-Synergy, T-Trade-off, N-No interactions. Comparing space-for-time and change-over-time approaches, a trade-off can be missed when the space-for-time approaches xiv  show no interaction or a synergy, but the change-over-time approach shows a trade-off (Panel A). Synergies missed occur when space-for-time approaches show either a trade-off or no interaction, but the change-over-time approach shows a synergy (Panel B). ............................... 67 Figure 4.3  Approach for comparing spatial correlations of two different ecosystem services (ES) at one year with change-over-time approaches (∆ES). I correlate the spatial distributions of ES at the reach level and compare these results to those correlating map differences in ES from 1949 to 2006 (∆ES). ................................................................................................................................... 71 Figure 4.4 Interaction diagrams for five ecosystem services. Correlations between ES were determined using two different approaches: Space-for-time substitution (left) and change-over-time (right). Circles represent different ecosystem services (ES) and significant Spearman’s Rank correlation coefficients (p < 0.05). Blue lines indicate ES synergies, red lines trade-offs. ES not connected by lines were not significantly correlated. Line width is scaled based on strength of correlation. .................................................................................................................................... 76 Figure 4.5 Interaction diagrams for five ecosystem services. Correlations between ES were determined using two different approaches: Space-for-time substitution (left) and change-over-time (right). Circles represent different ecosystem services (ES) and significant Dutilleul’s modified t-test coefficients (p < 0.05). Blue lines indicate ES synergies, red lines trade-offs. ES not connected by lines were not significantly correlated. Line width is scaled based on strength of correlation. ................................................................................................................................ 77 Figure 4.6 Boxes represent scatter plots showing the relationship between two ES. Positively correlated ES shown in boxes above are considered synergies, negatively correlated ES are trade-offs, and uncorrelated ES do not interact. S-Synergy, T-Trade-off, N-No interactions. xv  Interactions detected where none occur are examples of when space-for-time approaches showing a synergy or a trade-off, but change-over-time showing no interaction. ....................... 82 Figure 5.1 The Wenatchee watershed in central Washington State where the floodplains of the Chiwawa River, White River, Little Wenatchee River, Nason Creek and the Wenatchee main stem comprise the Wenatchee system floodplains examined in this work. .................................. 90 Figure 5.2 Longitudinal patterns in ES capacity across the Wenatchee system floodplain in 2006. My maps support my hypothesis that ES capacity varies along the river-continuum. ................. 95 Figure 5.3 ES diversity (Simpson’s index) varied significantly with reach type (Kruskal-Wallis X2= 55.40, df =4, p <0.01). Pairwise comparisons show island braided reaches and meandering reaches had higher diversity in ES capacity than all other reach types, but did not differ from each other. Straight reaches were more diverse than plane bed and pool/riffle reaches. ............. 98 Figure 5.4 Diversity in ES capacity (Simpson’s Index) according to river position. Diversity of ES capacity and distance to headwaters are negatively correlated (Spearman’s rho= -0.704, p <0.01) with distance from headwaters. ......................................................................................... 99 Figure 6.1 .Examples of snowmelt, transition, and rain-dominant hydrologic regimes based on historical monthly mean flows (m3s-1) for three rivers in the Columbia River Basin from 1929-2011............................................................................................................................................. 114 Figure 6.2 The Columbia River Basin (outlined in black) drains approximately 673,400 km2 and crosses the American and Canadian borders. It is likely to experience hydrologic regime shifts and trade-offs among floodplain ES under future climate. Colors indicate predicted shifts in spatial distribution of different hydrologic regimes in rivers throughout the Columbia River basin from 1976-1999 to 2070-2099 (Beechie et al. 2012). Future work should consider shifting hydrologic regimes in the Columbia River Basin, and its impact to ES..................................... 116 xvi  Figure 6.3 Trade-off between irrigated agriculture and hydropower under likely future climate scenarios (Elsner et al. 2010). Because water availability is expected to decline, maintenance of current levels of one service (e.g., hydropower) may lead to declines in the other. ................... 119 Figure 6.4 Complex cascade of interactions among ecosystem ES that may occur as a result of diminishing snowpack. Declines in summer water availability may lead to reductions in water levels for recreation directly (Hamlet and Lettenmaier 1999). Indirectly, declines in erosion control may cause soil nitrogen loss through erosion (Hatfield and Follett 2008), reducing in-stream water quality, and resulting in declines in water-based recreation. Future work should explore these potential pathways of decline. .............................................................................. 121 Figure 6.5 Complex cascading interactions among ES could cause indirect feedback loops. Though multiple pathways (declines in forest productivity and formation of salmon habitat), an indirect feedback loop may further cause declines in salmon, riparian forest productivity, coarse woody debris, and riparian nitrogen availability (Helfield and Naiman 2001). Future work should explore and quantify such pathways to feedback loops drawing on historical data to quantify time lags. ............................................................................................................................................. 123  xvii  List of abbreviations CRB: Columbia River Basin DEM: Digital Elevation Model DOI: Department of the Interior ES: Ecosystem Services GLO: General Land Office ha: hectare km: kilometers US: United States    xviii  Acknowledgements This research was supported by the Faculty of Forestry, the Namkoong Family Fellowship in Forest Sciences, VanDusen Graduate Fellowship in Forestry, CANFOR Corporation Fellowship in Forest Ecosystem Management, Pacific Institute for Climate Solutions, and NSERC. Baseline data for chapters 3-5 was created by Matt Tomlinson.  I thank Matt Tomlinson and Timothy Beechie for photo acquisition and interpretation of habitat features, funded by NOAA Fisheries.  I thank the students, faculty, and staff of UBC and the Faculty of Forestry for inspiring me and providing a nurturing environment for learning. I thank my committee, Nicholas Coops, Lori Daniels, and Simon Donner, for their kind support and advice throughout the project. I also thank my current and former labmates, Ian Eddy, Tanya Gallagher, Ira Sutherland, Jennifer Selgrath, Michelle Jackson, Kirsten Dales, and Helene Marcoux who have provided valuable feedback and have made this experience fun. I thank my technicians, Melanie Hilborn, Emmeline Topp, and Alex Berthin for their tireless work.   A special thanks to my parents who have believed and me and supported me throughout my entire education. I thank my siblings who are always there for me and inspire me with their accomplishments. I thank my wonderful husband who has been by my side throughout my PhD. Most of all, I thank my adviser, Sarah Gergel, for seeing me through the struggles and achievements over the past five years. I couldn’t have done it without the support of so many great people!  xix  Dedication To my grandfather, Carvel Johnson,  who also loved the land. 1  Chapter 1: Introduction Humans are inextricably linked to the environment and receive a wide variety of benefits from ecosystems (Daily 1997). Ecosystems provide many tangible benefits such as fresh water, food, and fibre. Forested ecosystems, in particular, store carbon in vegetation, support healthy soils by reducing soil erosion (which helps maintain clean drinking water supplies) and maintain soil organic matter (which aids farming and food production). Forested ecosystems support agriculture by providing habitat for crop pollinators and nutrient retention. Ecosystems also provide a wide range of non-material and cultural benefits including recreational, educational, and spiritual benefits (MA 2005). The array of benefits provided by ecosystems are collectively known as ecosystem services (ES) (MA 2005) (Table 1.1).  Table 1.1 Examples of the many types of benefits that ecosystems provide. The Millennium Ecosystem Assessment categorizes these services into provisioning, regulating, supporting, and cultural services. Provisioning services are products and goods from ecosystems. Regulating services are the benefits from stable ecosystem processes. Supporting ecosystem services provide services necessary for all other ecosystem services.  Cultural ecosystem services are the nonmaterial benefits that we gain from ecosystems (MA 2005). Provisioning services  Food  Fibre   Fuel wood  Water  Medicinal resources Regulating services  Climate regulation  Air purification  Disease regulation  Pollination Supporting services  Soil formation  Nutrient cycling  Primary production  Habitat for species Cultural services  Recreation  Spiritual values  Education  Cultural heritage  Inspiration  Aesthetics  2  ES are clearly vital for human well-being, and there is a long history of recognizing their importance, albeit under different terms (Daily 1997). Among the earliest recorded observations of ES interactions are Plato’s account of deforestation in Greece and the resulting increases in erosion and drying of springs (Mooney and Ehrlich 1997). More than two millennia later in 1864, George Perkins Marsh was among the first to assert that America’s resources were limited and attributed the decline in the previously fertile and thriving civilizations, such as the Roman Empire, to environmental degradation (Mooney and Ehrlich 1997). More recently (1960-1970’s), utilitarian approaches were used to garner public support for biodiversity conservation, setting the stage for explicit measurement, mapping, and economic valuation of ES. For example, the foundational paper “How much are Nature’s Services Worth?” discussed the importance of incorporating “nature’s services” into decision making (Westman 1977). While the history of recognizing the importance of nature benefits has long roots, only in the past few decades have scientist developed methods and common terminology to quantify, value, and map ES explicitly.  Ecologically diverse floodplains are particularly important for ES and provide habitat for fish, store large amounts of carbon, and leave legacies of fertile soils for farming (Tockner and Stanford 2002). Floodplains provide more than 25% of terrestrial ES but comprise only 1.4% of the global land surface area (Tockner and Stanford 2002). Floodplains are areas of low relief adjacent to rivers which may periodically connect to adjacent rivers through overbank inundation during floods (Junk and Welcomme 1990; Tockner and Stanford 2002). In pristine environments, floodplain processes are largely driven by flood disturbances which alter vegetation and geomorphic dynamics. Floodplain-adapted flora and fauna depend on periodic flooding; reproduction of floodplain plants and fish spawning often coincide with prevailing flood regimes 3  (Satake et al. 2001; Bailly et al. 2008). Nutrients and sediments from upstream areas are deposited during floods, making floodplains some of the most productive ecosystems in the world (Megonigal et al. 1997). As a result of their productivity, floodplains are also disproportionately important for ES and contribute to production of both aquatic and terrestrial ES (Tockner and Stanford 2002).  Floodplains have also been heavily modified by humans in order to access the wide diversity of ES available at the aquatic-terrestrial interface (Postel and Carpenter 1997). Fertile floodplains are often cultivated, while rivers are strategically dammed for hydropower production. River-floodplain systems may also form the backbone of transportation networks which aid access and distribution of ES (Postel and Carpenter 1997). Highly connected dendritic river networks support transport via barges and riverboats; whereas the gentle relief of adjacent floodplains are suitable for trails, roads and railways (Blanton and Marcus 2009). Despite the importance and development of floodplains for ES, rarely has the relative importance of this ecosystem (relative to terrestrial areas) been quantitatively demonstrated. 1.1 Spatial approaches for understanding ES ES vary considerably across space and such variability is crucial for management (Daily 2000). In terrestrial systems, the importance of understanding the spatial distributions of ES was recognized early. Efforts to map ES have been numerous with a focus on mapping multiple ES and their overlap (or concordance) with biodiversity (Chan et al. 2006; Egoh et al. 2009; Maes et al. 2012). Concordance among ES was deemed important from a conservation planning perspective (Chan et al. 2006). Thus, mapping spatial concordance among ES is of great interest 4  and has been used to determine how ES interact (Raudsepp-Hearne et al. 2010a). Despite this progress, several gaps remain in ES mapping efforts. Firstly, standards and spatial scales for mapping ES are not well-established (Crossman et al. 2013). The Natural Capital Project has pursued standardized, widely adaptable methods for mapping ES (Tallis et al. 2011). Furthermore, ES are often mapped across very broad areas at coarse spatial resolution. From a management perspective, broad-scale maps are limited for implementing local restoration. Rarely have ES been mapped at fine-grained spatial resolutions. A fine-scale approach is particularly key in ecosystems with a high degree of heterogeneity, such as floodplain ecosystems (Figure 1.1) (Gergel et al. 2007). Thus, while spatial assessments of ES are increasingly common, mapping of river-floodplain ES is rare, in part, due to generally very limited work using high resolution imagery map to ES  5   Figure 1.1 Shows the importance of high-resolution imagery in capturing river-floodplain variability. Panel A and C show mid-resolution Landsat imagery (30m resolution) at the scale of 1:50,000 and 1:4,000 respectively, while panels B and D show 1-m aerial photography at the same scales. In panel D, vegetated islands in the river channel and riparian vegetation can easily be distinguished from agricultural areas. This contrasts panel C where very little can be discerned. Mapping of river-floodplain ES may require specialized approaches. The distinct upstream-downstream connections of river networks are important to consider (Figure 1.2) (Wiens 2002). Hydrologic connectivity to upstream and lateral ecosystems is a defining feature of river-floodplains (Tockner and Stanford 2002). Riverscape patterns have spurred specialized 6  management strategies emphasizing environmental flows or process-based approaches (Poff et al. 1997; Naiman et al. 2010), as well as much exploration of how these patterns influence species distributions and ecosystem processes (e.g., Nilsson et al. 1989; Renofalt et al. 2005; Beechie et al. 2010). Interestingly, while the importance of floodplains for ES is appreciated (Felipe-Lucia et al. 2014), rarely have the unique linear connectivity patterns and spatial characteristics of floodplain systems been incorporated into ES assessments.   Figure 1.2 Longitudinal and lateral flows drive the distribution of river-floodplain patterns.  Mapping the spatial dynamics of ES is also important for identifying interactions among ES. ES can interact in a number of ways (e.g., feedbacks, trade-offs, synergies, cascades), but the mostly widely characterized ES interactions include synergies and trade-offs. A trade-off occurs when 7  increases in one ES results in declines in another (e.g., increases in agricultural driving declines in carbon storage (Turner et al. 2014)), while a synergy occurs when two ES are enhanced simultaneously (e.g., carbon storage and forest recreation (Raudsepp-Hearne et al. 2010a)). Unexpected interactions among ES can lead conflict between user groups (e.g., long-standing debates among farmers using irrigation and groups advocating instream water rights for fish in the Pacific Northwest (Moore and Mulville 1991)). Assessments of multiple ES are crucial for identifying interactions, yet methods for doing so are underdeveloped. Most studies use ES spatial distributions to infer their interactions. Locations where ES occur in low levels are presumed to represent a trade-off, while areas with high amounts of an ES are used to represent synergies (Raudsepp-Hearne et al. 2010a).  Yet spatial distributions of ES do not always represent ES interactions. ES may occur in different locations, and thus appear to trade-off using spatial approaches, but enhancing one of these spatially disparate ES may lead to improvements in the other. For example, while areas important for fish habitat and paddle routes may occur in different location, such ES might both improve in response to a shared attribute, such as forest cover. Such inconsistencies mean new spatial methods are needed.  1.2 Long-term approaches needed  New perspectives on temporal dimensions of ES are also needed. Humans have a long history of modifying floodplains, which is well documented in Europe and North America (Decamps et al. 1988; Crawford et al. 1998; Stinchcomb et al. 2011). In North America, European settlement often took place in floodplains (Baker et al. 1993; Tomscha and Gergel 2015). In western montane North America, many changes in ES accompanied the agricultural expansion of the early 20th century. This initial settlement, and the accompanying roads, railways, etc., potentially 8  created the foundation for subsequent ES use across many landscapes. Settlement infrastructure, while enhancing and providing access to certain ES such as agriculture, hydropower, and recreation, may also have unintentionally degraded some ES (Raudsepp-Hearne et al. 2010b). Roads and dams prevent fish migration by acting as barriers, and road increase sediment in streams increasing turbidity and reducing productivity of aquatic ecosystems (Trombulak and Frissell 2000). The ES distributions of contemporary floodplain landscapes are the result of long-term use and degradation.  Long-term changes in riverine systems have been explored using aerial photography (e.g., Large and Petts 1996; Connor et al. 2003; Tomlinson et al. 2011; Wan et al. 2014); however, rarely has aerial photography been used to examine patterns of ES. At broader scales, Public Land Surveys from the United States General Land Office (circa mid-1800’s to early 1900’s) have been used to reconstruct vegetation and carbon storage (e.g.,White and Mladenoff 1994; He et al. 2000; Rhemtulla et al. 2009a; Hanberry et al. 2014); such records could potentially be used to understand other ES. Historical data are generally under-utilized as a source of information about long-term dynamics and baseline conditions (Morgan et al. 2010; Tomscha and Gergel 2015). Understanding ES can benefit greatly from new approaches which enable long-term perspectives to account for changes in ES over time. 1.3 Approaches and objectives Given this need for new spatial and temporal approaches to understanding ES, in this dissertation, I quantitatively explore a river-floodplain system through development of multiple, innovative tools across a long time frame. My overall objective is to characterize ES spatio-temporal dynamics in a way which better informs our understanding of ES interactions. I focus 9  on a multi-use watershed and floodplain in central Washington State—the Wenatchee River watershed—throughout settlement transitions seen in many floodplains globally. This watershed has undergone a typical settlement sequence from initial frontier settlement through a period of agricultural expansion, followed by urbanization and agricultural intensification (Foley et al. 2005). These transitions, and the accompanying timeframes, supports a wide range of geospatial approaches and reveal changes in ES and their interactions beyond what a single snapshot provides (Figure 1.3).  In my concluding chapter, I demonstrate how conceptualizing ES interactions for restoration and in order to avoid problematic declines under future climates. 10    Figure 1.3 Timeline and diagram of my approach, and the overall land cover transitions (orange) associated with each chapter and its driving question.  11   Chapter 2: Historically, how does frontier settlement vary with landscape position? Despite the importance and development of floodplains for ES, rarely has the relative importance of this ecosystem (relative to terrestrial areas) been quantitatively demonstrated. Beginning with a look at early-settlement landscape patterns, I focus on a historical reconstruction of linear anthropogenic disturbances, such as trails, roads and railways, which lay the groundwork for accessing ES across the landscape. Exploring these ideas requires reconstruction of settlement using land surveys from the General Land Office (GLO) USA, which reflect frontier access to different parts of the landscape and to various ES. This dataset has been widely-used to reconstruct vegetation patterns, but I innovate by using it to reconstruct settlement patterns. This chapter lays the foundation for understanding for contemporary ES patterns. Chapter 3: How has land cover and ES capacity changed over time?  In the subsequent chapters (3, 4, and 5), I establish innovative ES mapping approaches, using high-resolution aerial photography to assess long-term changes in land cover and ES. While many ES mapping studies use Landsat imagery to derive land cover, I use high-resolution ( < 1m resolution) approaches necessary for mapping riparian and floodplain ES, such as different agricultural types and vegetated islands (Figure 1.1) (Zhao et al. 2004; Tomlinson et al. 2011). My third chapter provides a rare look at how land-cover dynamics have contributed to ES change, during a time when agricultural valley bottoms in the Wenatchee system were rapidly urbanizing. Emphasizing combined contemporary and long-term drivers of change in ES, I discuss how history can continue to influence ES provision for decades. This chapter introduces methods for subsequent chapters. 12  Chapter 4: How do baselines improve our understanding of ES interactions?  Typical approaches for understanding multiple ES correlate ES across space to infer their interactions. In this chapter, I critique spatial approaches for identifying ES interactions and demonstrate the importance of incorporating baseline information (derived from aerial photography) to understand the dynamics of ES. I show how spatial correlations between ES differ from temporal correlations between ES which has major implications for understanding ES interactions. My novel use of long-term approaches to understand ES interactions helps avoid problematic ES trade-offs and identify opportunities for ES synergies. Chapter 5: How can river-floodplain concepts improve our understanding of the spatial variability of ES capacity?  Here I explore the shortcomings of several well-established approaches for mapping ES in floodplains and suggest how river-floodplain concepts, such as the River Continuum Concept, might help us better understand spatial variability of ES. I map and compare patterns of ES along longitudinal and fluvial-geomorphic gradients, revealing the relevance and potential for river-floodplain concepts to inform ES mapping projects and river-floodplain restoration.  Chapter 6 (Conclusion): How might floodplain ES interact under diminishing snowpack conditions?  In my concluding chapter, I build on the key results and documented interactions among ES to explore the implications of my work for understanding future ES interactions. I also hypothesize using a literature review how floodplain ES may interact under future climate (with a focus on the Interior Columbia River basin), highlighting the importance of long-term datasets for 13  understanding ecosystem service interactions. I suggest a more nuanced view of ES interactions may aid restoration and help avoid problematic declines in ES. Modern patterns of ES use are heavily influenced by historical settlement patterns, especially in fertile floodplains. Overall, I explore long-term utilization of varying landscape positions by frontier settlers. I document the dynamics of floodplain-specific ES over broad timeframes and at fine scales, showing how long-term data is necessary for understanding ES interactions. Long-term approaches can aid efforts to enhance multiple ES simultaneously. I suggest river-floodplain concepts can provide key insights into the distribution of floodplain ES. Floodplain-specific ES may thus require specialized management and mapping approaches. Finally, I hypothesize how ES might interact in floodplains under future hydrologic conditions. Through this, my work demonstrates novel applications for aerial photography and GLO survey data. Such historical datasets are traditionally used for reconstructing forest structure and landscape features but can also benefit efforts to understand ES.   14  Chapter 2: Historic land surveys present opportunities for reconstructing frontier settlement patterns in North America 2.1 Background Human settlement patterns emerge from interacting environmental and social drivers (Bürgi et al. 2004), and efforts to understand drivers of settlement patterns in North America have emerged from several academic disciplines spanning distinct timeframes. Archeologists have explored how physical, environmental, and cultural variables influenced historic settlement patterns (Anschuetz et al. 2001), demonstrating a shift of settlements from lower valley bottoms during dry periods to higher elevations during wet eras in the American Southwest (A.D. 1-1400) (Dean et al. 1985). Rivers and floodplains are often deemed important but primarily in qualitative and non-spatially detailed terms. Landscape ecologists have used aerial photography (1930’s), satellite imagery (1970’s), and census data to identify drivers of recent development (Chen 2002; Morgan and Gergel 2010) which is often influenced by environmental amenities, road access, and economic factors. Early settlement patterns in North America (from 1800’s-1900’s), however, the foundations of recent, continuous settlement often occurred prior to the proliferation of aerial and satellite imagery, yet long after millennial timeframes requiring archeological approaches. Very little work has quantified frontier settlement patterns in North America, despite its critical importance in initiating patterns seen in contemporary landscapes. Thus, quantifying historical settlement patterns and disturbance agents can help us understand modern and historical social-ecological systems.  15  Systematic US General Land Office (GLO) land surveys provide extensive and spatially explicit data for exploring landscapes from the late 1700’s to the 1920’s (Silbernagel et al. 1997; Schulte and Mladenoff 2001). Prior to extensive settlement, GLO surveyors identified profitable areas for logging or agriculture (Radeloff et al. 1999). Ecologists have used these surveys extensively to explore historical vegetation (Brown et al. 1997; He et al. 2000; Bouldin 2003; Schulte et al. 2007; Rhemtulla et al. 2009a; Hanberry et al. 2010, etc.), but surveys have been underutilized as a resource for quantifying frontier settlement patterns since their initial use by Silbernagel et al. (1997). Here, I draw on historical GLO surveys from Washington State, to understand patterns of especially persistent linear disturbances, including roads, railways, trails, fences, and irrigation ditches. This approach helps bridge the gap between archeological studies (often thousands of years ago) and aerial photography reconstructions (several decades ago). Similar land surveys exist throughout the US and Canada providing continental-wide historical information.  I ask two key questions. 2.1.1 Historically, which anthropogenic linear disturbances co-occur? Fragmentation has impacted ecosystems over the past century, yet little is known about the extent and configuration of historical landscape fragmentation. Individually, linear disturbances cause a variety of impacts: Railways, roads, and trails may alter watershed sediment dynamics (Trombulak and Frissell 2000), fences impede the large mammal movement (Boone and Hobbs 2004), irrigation can result in waterlogged and/or saline soils and affect water quality (Wichelns and Oster 2006). Distinguishing disturbance types allows for nuanced interpretation of impacts, as different disturbance combinations may have distinct implications for ecosystems and future settlement. Co-occurring disturbances may have interacting, confounding, or synergistic effects. 16  Here, I map and distinguish linear disturbances such as roads, railways, fences, irrigation ditches, and trails to quantify which co-occurred spatially, and thus simultaneously impacting the landscape. 2.1.2 Did historical disturbances vary with landscape position? Landscape position influences a variety of social-ecological phenomena. In vegetation research using GLO data, topographic variables have been shown to control vegetation distribution and natural disturbance regimes (Whitney 1986; Abrams and Ruffner 1995; Dyer 2001). Lower slope positions, with higher soil organic carbon, nitrogen, and phosphorous concentrations influence agriculture (Schimel et al. 1985; Swanson et al. 1988). Soil wetness influences the distribution of ecological communities (Iverson et al. 1997; Hutchinson et al. 1999; Besnard et al. 2013).  Here, I map topographic wetness index across a watershed to define landscape positions based on wetness thresholds. I determine how settlement infrastructure (trails, roads, etc.) varied with landscape position to better understand environmental drivers foundational to frontier-settlement landscapes.  2.2 Methods 2.2.1 Study site  The Wenatchee watershed in central Washington State drains into the Columbia River (Figure 2.1). The rain shadow of the Cascade Mountains results in high precipitation in the west and low precipitation in the east. Topographically diverse, western slopes are covered with glaciers, moist alpine meadows, open forests at high elevations, and dense forests at low elevations. Eastern slopes are drier and covered with glaciers, alpine grasslands, and drought-tolerant conifer forests 17  (Jorgensen et al. 2009; Tomlinson et al. 2011). Contemporary land uses include urban, orchard-agriculture, forestry, and wilderness. The U.S. Forest Service manages 75% of the 3440 km2 Wenatchee watershed. Historically, the semi-nomadic Wenatchi people inhabited the watershed with an estimated population of 1000 (mid 1800’s) (Ray 1974). At the peak of the salmon season, 2000-3000 people from other regional tribes would visit the confluence of Icicle Creek and the Wenatchee River (Chalfant and Ray 1974). By the time European settlers arrived in large numbers, the population of the Wenatchi had already been greatly reduced from small pox epidemics (Chalfant and Ray 1974). Railway completion in 1892 spurred population growth, lumber mills, and orchard agriculture (Chalfant and Ray 1974; Dimitri 2001).   18  Figure 2.1 The Wenatchee watershed, a sub-watershed of the Columbia River basin, in central Washington State. 2.2.2 Data entry Land surveys in the Wenatchee watershed took place from 1890-1922 (U.S. Department of the Interior). Portions of the watershed included in the Wenatchee National Forest (est. 1908) were not intended for sale, and thus not surveyed, as were areas of steep topography. Earlier surveys took place in the eastern watershed, while western areas with more mountainous terrain were surveyed later. Surveyors divided the land into 6x6 mile townships and further into thirty-six 1x1 mile sections (Schulte and Mladenoff 2001). Surveyors would note disturbances along section lines (Figure 2.2). My final dataset contains data from 42 townships, 821 sections, and 1462 individual section lines, many of these townships were partially surveyed. Linear disturbances (e.g., roads, fences, trails, railroads, and irrigation ditches) were derived from GLO surveys. I recorded 619 linear disturbances and categorized them based on surveyor descriptions (135 roads (including highways), 361 trails, 44 railroads, 57 fences, 22 irrigation ditches). I ignored non-linear disturbances (houses, fields, etc.), because they rarely occurred directly on section lines making records of their spatial locations less reliable.  19  .   Figure 2.2 Sampling design for disturbance data, showing how surveyors would have recorded a road transecting a township. Scanned GLO survey notes and shapefiles of cadastral section lines were downloaded from US Department of the Interior (US DOI), Bureau of Land Management website. Mostly in the form of original, handwritten field notes, I transcribed notes to corresponding sections lines in ArcMap 10.1, using the GCS_North_American_1983 coordinate system and D_North_American1983 datum. Positional accuracy assessments of GLO corner points in Wyoming found errors ranged from 5.6m to 63.4m (mean=22.5m, sd =15.9), thus I assumed similar positional error (Bouldin 2003).  20  2.2.3 Fraudulent surveys GLO land surveys, particularly between 1875-1898, must be carefully scrutinized for evidence of fraud (Puter and Stevens 1908). During this period, an organized crime ring (the Benson syndicate) fabricated GLO surveys for compensation (Puter and Stevens 1908). Corroborating evidence (such as absence of major rivers, topographic features, or bearing trees in “resurveys”) can help identify fraudulent surveys. Furthermore, historians have documented characteristics of individual surveyors implicated in Benson syndicate lawsuits (Olson 2008). Wenatchee surveyors implicated include “Charles C. Holcomb” and “George J. Gardiner” (Olson 2008). I removed their surveys from my analysis and confirmed remaining data by verifying major stream crossings. Where fraudulent surveys were subsequently resurveyed, I used “resurveys” completed before 1922. Data from 2300 km of surveys (78% of section lines) were deemed of good quality for use in my analysis (Figure 4). 2.2.4 Landscape position defined by topographic wetness index (TWI) In river-floodplains, a critical signature of landscape position is potential soil wetness, which can be measured using digital elevation models. Topographic control on irrigation, trees, and crop species means that soil wetness influences societies dependent on agricultural and forest harvest (Bajat et al. 2011). I modeled topographic wetness index in the Wenatchee watershed using a 30m DEM (Data available from the US Geological Society) in ArcMap 10.1 (USGS 1993; ESRI 2012).                                                       (1)  𝑇𝑊𝐼 = ln(𝐴𝑠 / tan(𝑆𝑙𝑜𝑝𝑒))                                                                21  As represents the upslope contributing area estimated using a flow accumulation algorithm in the ArcMap 10.1 (ESRI 2012). Slope (angle) was modeled from a 30 x 30m DEM, as in recent GLO vegetation studies, and then converted to radians (Hanberry et al. 2012; Hanberry et al. 2014). In order to reduce raster noise and account for potential positional error in GLO disturbances the TWI map was smoothed using a 3 x 3 low pass filter, thus ensuring disturbances fell within a locally accurate TWI. Based on a smoothed TWI map, I identified five potential wetness categories defined as landscape positions: ridge, upper slope, lower slope, terrace, riparian zone (Table 2.1), based on natural breaks in Topographic Wetness index, which unlike other breakings (quantile, equal interval, etc.) distinguished riparian zones from terraces (ESRI 2012). These landscape positions are analogous to those used in aerial photography interpretation, but more quantitative instead of manually interpreted (MOF 2011).   22  Table 2.1 Topographic Wetness index was mapped across the Wenatchee watershed and divided into five landscape positions based on natural breaks. I explain percent of landscape in of each landscape position. My descriptions are analogous to those used in aerial photograph interpretation (MOF 2011).  Landscape Position Topographic Wetness Index  Description % of watershed Uplands Ridge 6.22 - 7.59 Highest locations on landscape with low contributing upstream area, often steep slopes on either side 2.6  Upper Slope 7.59- 9.03 Next highest locations on landscape, often adjacent to ridges, steep slopes 43.8  Lower Slope 9.03 - 10.73 Adjacent to terraces (see below), moderate slope 37.7 Floodplains Floodplain terrace 10.73 - 13.80 Bench lying on each side of a river, geologic-scale floodplain, adjacent to riparian zone, flat, gullies  13.3 Riparian zone 13.80 - 22.89 Riparian zone, river channel, active floodplains, wetlands, lakes 2.6   2.3 Statistical analysis 2.3.1 Historically, which anthropogenic linear disturbances co-occur? Disturbances were considered to co-occur locally when two or more were present on the same section line (Figure 2.3). Variance ratio and a W statistic were compared with the chi-square distribution to determine if local disturbance co-occurrences happened more than expected by chance (Schluter 1984).  (2) 𝑉𝑅 =  𝑠𝑇2/ ∑ 𝜎𝑖2                                       23  Here VR measures the ratio of the variance in disturbances per site ( 𝑠𝑇2), where Τ represents an individual site to the sum of variances per disturbance type ( ∑ 𝜎𝑖2), where i represents each disturbance type. A value > 1 indicates presence/absence of disturbances covary positively (i.e., tend to co-occur).   (3)      𝑊 = 𝑉𝑅 𝑥𝑁                 Here, N represents the total number of transects sampled. The calculated W statistic was compared to the chi-square distribution (df = 1461, α = 0.05) to test the significance of the variance ratio. Degrees of freedom represent the total number of section lines. Post-hoc, pair-wise chi-square with Bonferroni corrections were used to determine which disturbances specifically co-occurred (e.g., fence-road, fence-railroad, fence-trail, etc.) for a total of ten pairs. All section lines in my study area were used to make this comparison. Co-occurrence values may differ based on underlying choice of extent (Kallio et al. 2011). Because I was interested in disturbance co-occurrence throughout the watershed, I used all surveyed section lines (including those with no disturbances) for my analysis.  24   Figure 2.3 Grid shows railway and road traversing a landscape co-occurring on one section line. This diagram represents 84 section lines. A contingency table for this diagram would show 69 lines with no disturbances, 9 with only roads, 9 with railway disturbances and 1 with co-occurring roads and railways. Section line-level data is used in the analysis. 2.3.2 Did historical disturbances vary with landscape position? First, I determined expected numbers of disturbances in each landscape position accounting for survey effort/extent in each landscape position based on relative amount. I used a chi-square test to determine if disturbances occurred in different landscape positions more than would be expected by chance. Finally, I used a post-hoc chi-square with Bonferoni corrections to determine how specific disturbances varied with landscape position. 25  2.4 Results Fraudulent data affected the eastern Wenatchee watershed, whereas the northern watershed was not surveyed (USFS land) (Figure 2.4). Topographic Wetness Index is highly variable across the watershed (Figure 2.4). 2.4.1 Historically, which anthropogenic linear disturbances co-occur? Many disturbances co-occurred spatially. Variance ratio (1.281) and W statistic (1873, df= 1461, p < 0.001) indicated disturbances were positively associated. Pair-wise chi-square tests show co-occurrence varied with disturbance type (Figure 2.5). Roads were associated with fences, irrigation ditches, and railways. Irrigation ditches and fences were also associated. Railways were positively associated with roads but not with fence or irrigation ditches (Figure 2.5). Trails were not associated with any other disturbance. 2.4.2 Did historical disturbances vary with landscape position? The spatial distribution of disturbances differed from random (X2=455.3183, df =4, p < 0.001) and were concentrated in riparian zones and terraces. Over 45% of disturbances occurred in these areas, which comprised less than 16% of the study area. In the riparian zone, 8.1% more disturbances occurred than expected, while in terraces 21.1% more than expected occurred. Lower slopes and ridges showed 3.6% and 2.3% fewer disturbances than expected respectively, while upper slopes showed 23.3% fewer than expect (Figure 2.6). For all linear disturbance types, fewer than expected occurred in lower slopes, upper slopes, and ridges. 26   Figure 2.4 Map of Wenatchee watershed identifying landscape positions defined by the topographic wetness index. 27   Figure 2.5 Visualizes how disturbances co-occurred in the early-settlement Wenatchee landscape. Significantly co-occurring disturbances are connected with a solid black line (p<0.005). Variance ratio values are shown for co-occurring disturbances. Values greater > 1 mean disturbances co-occur. Roads were associated with three other disturbances while trails were not associated with any other disturbance. 28    Figure 2.6 Difference in disturbance percentages by landscape position versus the percent surveyed area by landscape position. More disturbances than expected occurred in the riparian zones and terraces while fewer than expected disturbances occurred in lower slopes, upper slopes, and ridges.29   2.5 Discussion My work provides a rare quantification of the co-occurrence of frontier disturbances and their impacts to varying landscape positions. Riparian zones and terraces were highly impacted by co-occurring linear disturbances historically. Despite occupying 84% of the landscape area, only 55% of disturbances take place in upland areas (i.e., lower slopes, upper slopes, and ridges). In contrast, more than 45% of disturbances occurred in riparian zones and terraces, which comprised under 16% of landscape area. This concentration of all disturbance types in floodplains quantifies the extent to which riparian zones and terraces were disproportionately selected for settlement for a wide variety of purposes. The spatial distribution of disturbance types reflect socio-economic drivers: The co-occurrence of fences, irrigation ditches, and roads imply land conversion for farmland, while railway-road co-occurrence suggests development for resource distribution (lumber, fruit, etc.). Quite notably, trails did not co-occur with any other disturbances. Several trails were described as “Indian Trails” in survey notes suggesting their Native American origin. The lack of co-occurrence among trails and other disturbances indicates different access (and probable different use) of various floodplain locations.  2.5.1 Ecological consequences of linear anthropogenic disturbance My work provides evidence that valley bottoms have been prone to lateral disconnection from roads and railways for more than a century, potentially resulting in a suite of ecological consequences (Blanton and Marcus 2009). Roads and railways impact hydrology and geomorphology, including soil density, temperature, soil water content and patterns of water 30  runoff (Forman and Alexander 1998; Trombulak and Frissell 2000). Debris flows are initiated by roads, and in hydrologic and soil erosion models, roads and trails are shown to be disproportionately active (Harden 1992; Jones et al. 2000). Dirt roads, prevalent in the early 1900’s, are particularly prone to sediment runoff (Reid and Dunne 1984). Such impacts are evident today as roads and railways can disconnect rivers from their floodplains reducing side channels, ponds, and refugia habitats, negatively impacting fish populations (Tomlinson et al. 2011; Blanton and Marcus 2013). Additionally, counts of spring-summer chinook salmon and steelhead were negatively related to road density in Idaho (Thompson and Lee 2000).  Agricultural development resulted in co-occurrence of fences and irrigation ditches in the Wenatchee watershed. Early settlement fencing fragmented landscapes by breaking large tracts of land into islands (Woodroffe et al. 2014). Animal populations isolated by fences are prone to local extinction because access to resources is constrained, lowering an area’s carrying capacity (Woodroffe et al. 2014). Historically, early irrigation ditches provided water for mining operations (e.g., Blewett gold mine in Wenatchee watershed), and later, water transport methods were adapted for irrigation for agriculture (Lemly 1994). A suite of ecological issues are associated with dry-land agriculture in water-scarce regions. Draining, conversion, or removal of water from streams and rivers alters streamflow and is associated with wetland drainage (Lemly and Kingsford 2000). Furthermore, downstream effects of irrigation, such as contamination by pesticides can negatively impact fish and migratory bird populations (Ohlendorf et al. 1986; Lemly et al. 1993; Lange et al. 2014).  The origin and purpose of trail establishment may create different impacts. Some were likely established prior to the arrival of European settlers, initially blazed by Native Americans or 31  potentially following wildlife trails. The initial construction of trails can disproportionately impact ecosystems (Monz et al. 2013). Thus, frontier trail blazing by European settlers would have initiated a suite of impacts to biota. For example, recent work showed altered bird species composition adjacent to trails in Colorado with specialist species found further from trails and generalist species found closer to trails (Miller et al. 1998). Woody and delicate herb species have been found further from trails, while graminoids closer to trails in Shenandoah National Park (Hall and Kuss 1989).  2.5.2 Legacies of historical floodplain fragmentation Historical floodplain fragmentation may continue to influence contemporary floodplain ecosystems as legacies of historical disturbances on terrestrial and aquatic systems are widespread (Harding et al. 1998; Foster et al. 2003). For example, species composition in post-agricultural, secondary forests differ from primary forests after more than a century (Bellemare et al. 2002). Land-use effects can persist for millennia with contemporary plant diversity  influenced by Roman agriculturalists (Dambrine et al. 2007). Such persistent historical land-use effects suggest a need for further quantification of historical fragmentation legacies. Thus, quantifiable indicators of historical impacts may be useful to understand biodiversity patterns in fragmented landscapes, and can be useful to test hypotheses about future floodplain changes (Lunt and Spooner 2005).  Historical disturbances likely shaped vegetation seen today, and GLO records can provide such historical disturbance information. In fact, GLO land survey records have been proposed as references for restoration (McAllister 2008). At the turn of the 20th century, most productive landscapes were already occupied by Native Americans and early settlers creating impacts on 32  vegetation. Thus, in addition to elucidating frontier settlement patterns, characterizing early-settlement patterns may improve interpretation of GLO-derived vegetation maps. Our results suggest historical vegetation patterns should be interpreted with care, particularly in floodplain and riparian environments where vegetation may have been highly impacted by anthropogenic disturbances for long periods of time. GLO interpretations should consider the vegetation communities a result of interacting environmental and anthropogenically-driven processes and not necessarily representative of pristine conditions. 2.6 Conclusion I used GLO land survey data to determine how linear settlement disturbances varied across the historical landscape. I found that landscape position shaped early settlement patterns and that every type of anthropogenic disturbance was concentrated in riparian zones and floodplain terraces. Because many disturbance types were shown to co-occur, cumulative impacts to floodplains have been happening for over a century. Underutilized historical land survey data can provide valuable insight into early settlement patterns and likely impacts to floodplain landscapes. Such historical legacies likely continue to shape development patterns on the landscape today.   33  Chapter 3: Long-term changes in the capacity of a river-floodplain to provide ecosystem services 3.1 Introduction Landscapes are inherently dynamic, as is their capacity to provide ecosystem services (ES). Humans routinely intervene to enhance certain ES, often at the expense of other ES. Such historical modifications can leave biophysical and cultural legacies which continue to shape contemporary landscapes. Multiple, interacting layers of historical human intervention and biophysical processes underpin contemporary landscape patterns and ES (Bürgi et al. 2004). While historical context is fundamental for understanding modern ES, rarely have the historical dynamics of ES been explored. Perhaps because long-term tracking of ES is so especially challenging, only recently have ES frameworks incorporated temporal shifts (Villamagna et al. 2013).    Distinguishing between flows and stocks of ES is critical for understanding their temporal dynamics, yet to date, little ES research has made this distinction (Burkhard et al. 2009; Burkhard et al. 2012). Conceptually, flows refer to ES which actually reach people, whereas the stocks of ecosystems providing such flows are a distinct entity (Burkhard et al. 2014). ES by definition require human utilization (Millenium Ecosystem Assessment (MA) 2005). One challenge is that, with ever-increasing human populations, mapping ES over time may show ES have increased simply because ES reach more people. Furthermore, any depletions or changes in ES stocks could be missed by assessments focused solely on ES flows. Understanding such distinctions and addressing changes in ES capacity is necessary to determine if ES demand can 34  be met by existing ecosystems or whether alternatives are needed (Villamagna et al. 2013). Recent ES frameworks have highlighted the critical relevance of this distinction (Villamagna et al. 2013), noting that differentiating these distinct concepts requires clarifying the unit of measurement in ES assessments. In addition to ES capacity, the stocks of ecosystems which provide ES have also been termed potential ES supply, natural capital, etc. (Kienast et al. 2009; Burkhard et al. 2012); here, I also refer to stocks as ES capacity (Villamagna et al. 2013).  While efforts to understand temporal dynamics of ES are generally in their infancy, several studies have tracked temporal changes in ES (Lautenbach et al. 2011; Dearing et al. 2012; Bürgi et al. 2015). For example, paleoecological proxies have been used to explore long-term dynamics of three regulating services (soil stability, sediment regulation, and water purification) in the Yangtze river basin over the past two centuries, showing declines since the 1980’s largely due to agricultural expansion (Dearing et al. 2012). Changes in the types of ES used over time have also been documented (Bürgi et al. 2015). An evaluation of changing ES in Switzerland determined that new ES such as tourism, education, and hydropower have become more important based on being mentioned in alpine cadasters at different timeframes; however, changes in capacity were not examined (Bürgi et al. 2015).  The long-term dynamics of multiple ES remain poorly understood in comparison to the magnitude of literature on spatial aspects of multiple ES (Chan et al. 2006; Egoh et al. 2009; Raudsepp-Hearne et al. 2010a; Crossman et al. 2013). Where changes in ES capacity have been tracked, studies are generally limited to one ES, such as carbon storage, and are reconstructed at broad scales (Rhemtulla et al. 2009a). Few have used high-resolution, spatially explicit 35  approaches. The primary goal of this work is to demonstrate how the temporal dynamics of multi-ES can be better understood when using spatially explicit perspective. 3.1.1 Aerial photography interpretation as a tool to track change in ES  Historical data sources, such as aerial photography are vastly underutilized as a source of information about baseline conditions of ecosystems and landscapes (Morgan et al. 2010; Morgan and Gergel 2013). Such information can be useful in reconstructing historical land cover (Tomlinson et al. 2011) and in quantifying historical  ES capacity, especially when used in combination with other geospatial data (such as digital elevation models). Aerial photography provides some of the best information available to estimate long-term change in ES capacity, because it is spatially explicit and available over timeframes much longer than other types of remote sensing (circa 1930’s vs. 1970’s) (Morgan et al. 2010). Relating land-cover to ES is an approach often used to approximate ES, making an aerial photography approach amenable to other ES studies (Burkhard et al. 2009).  Here, I showcase a mapping approach designed to explore long-term ES capacity dynamics. Using high-resolution aerial photography along with additional geospatial information, I track land cover transitions and identify drivers of ES capacity change. I also discuss the historical social-ecological context likely contributing to these multi-ES capacity dynamics. The two broad objectives are to: (1) Quantify transitions among land cover classes which underlie the production of ES and (2) demonstrate a feasible approach for the difficult task of measuring change in ES capacity.  36  I use an urbanizing river-floodplain landscape to exemplify these issues. River-floodplains are often disproportionately important for providing ES and attract development over centuries and millennia (Tomscha and Gergel 2015). As such, river-floodplains have been greatly impacted by land cover change (Tockner and Stanford 2002). As much as 90% of North American floodplains have been cultivated to enhance food production, and thus have become functionally extinct from an ecological perspective (Tockner and Stanford 2002). With many urban centers originally established along rivers and within floodplains, such areas are gradually shifting away from agricultural uses to more urban uses with consequences for ES and ES capacity. These changes mirror dynamics across the globe with landscapes transition from frontier expansion (Chapter 2) through phases of agricultural intensification and urbanization (Foley et al. 2005). 3.2 Methods 3.2.1 Study Site The Wenatchee River System is a tributary of the Columbia River in central Washington State (Figure 3.1) undergoing rapid urbanization since the mid-20th century. The valley bottom of the Wenatchee main stem has been largely converted for orchard agriculture, while the floodplains of its tributaries remain forested but are traversed by road networks. A popular tourism destination has evolved in the Wenatchee system floodplain attracting recreationists and agricultural tourism. The total area of the floodplain is 210 km2, comprised of the Chiwawa River, White River, Little Wenatchee, Nason Creek, and the Wenatchee main stem floodplains. 37   Figure 3.1 The Wenatchee system floodplain situated in central Washington State. Economic and social changes over the past century in the Wenatchee system floodplain have resulted in shifts in the structure and function of ecosystems, and thus changes in ES capacity. Here, the dominant land cover changes during the timeframe examined resulted from declines in valley-bottom logging and increasing resident population as well as tourism (Figure 3.2). Prior to the mid-1900s, the regional economy was based on forest harvest and other extractive industries such as gold mining. Following economic slumps, economic revival efforts included redesign of Leavenworth, WA into a Bavarian-themed village in 1960. Tourism subsequently increased following this transformation. These diverse sectoral and land use changes over time allows 38  examination of a wide range of relevant ES which include: orchard and forage production, carbon storage, paddle routes, and fish production capacity.   Figure 3.2 Timeline of important land use and resource development events in the Wenatchee floodplain. 3.2.2 Aerial photographs and floodplain delineation I mapped historical and modern ES capacities using aerial photography with a minimum resolution of 1 m from 1949 and 2006. Imagery from September 1949 (1:20,000 scale) was acquired from the National Archives and Records Administration, scanned at high resolution (1200 dpi) using an Epson Expression 1640 (Epson America, Long Beach California, USA), and orthorectfied using Alta Photogrametric Suite (APS) version 7.x (Groupe Alta, Quebec City, Quebec, Canada). Imagery from July 2006 (1:40,000) was acquired from USDA National Agricultural Imagery Program (USDA 2006) (Tomlinson et al. 2011). Because the historical floodplain was indiscernible using modern floodplain maps (due to dams and levees), the 39  floodplain extent was manually digitized using hillshade derivative of USGS 10m DEM and 1:24,000 topoquads (Tomlinson et al. 2011). I mapped ES capacity only within this valley bottom. For further details on orthorectification and floodplain mapping, see Tomlinson et al., 2011.  3.2.3 Land cover classification  Using existing land cover maps of the region (Tomlinson et al. 2011) as base maps, I digitized agricultural types distinguishing orchards from fields in 1949 and 2006 (digitized at 1:4,000 scale) (Table 3.1). I classified land cover into five main categories: orchard, field, urban, forest, and water. From there, I further distinguished forest into density classes (high, moderate, and low), and into vegetation types (mostly conifer, mixed conifer and broadleaf, wet shrub, scrub, rock-snow). Land cover types were used as proxies for several different ES (Burkhard et al. 2009).40  Table 3.1 Examples of land-cover classification. Land-cover classes were manually digitized at a 1:4000 scale. Patches less than 0.5 ha were not distinguished. The red circles are characteristic examples of land-cover types at each timeframe. For forest and shrub subclasses, the aerial photographer interpreter gave an estimation of their certainty: 1= Certain, 2 = Fairly Certain, 3 = Uncertain. The mixed conifer and broadleaf classes in 1949 were most uncertain. Land-cover Subclass Description Mean Certainty   (1 = Certain,  2 = Fairly Certain,  3 = Uncertain) Example 1949 2006 1949 2006 Urban  Dense buildings and roads, geometric patterns, regularly spaced NA NA    41  Orchard Regularly spaced circular crowns of trees which are black-dark grey in the historical photos and dark green in modern photos.  NA NA    Field White or grey (1949) or yellow (2006), geometric patches throughout the landscape NA NA     Forest/shrub Mostly Conifer • Continuous Forest Cover < 75% closed canopy  • Distinguishable as pointed, cone shaped objects  • Shadow indicates coniferous shape  • More homogenous shades • Generally darker than broadleaf species   1.5 1.3   42   Mixed Conifer Broadleaf • Continuous Forest  • Cover < 75% closed canopy • Appears in clumps and large patches   • Dark grey to black on 1949 and green to black on 2006  • More heterogeneous shades indicating lighter-coloured broadleaf  • Clearly visible shadow indicates height of tree as opposed to low lying shrub  • Some rounded shape, some cone shape • Any visible deciduous forest characterized as mixed conifer broadleaf 1.6 1.4   Wet shrub • Less than 10% canopy cover  • Lush low-lying vegetation, smaller or no shadow  • Found in riparian zones  • Bright green (2006) and darker grey (1949)  • Presence of water (e.g. pools), as black (not in pre-classified river) 1.1 1.0   Dry scrub • Less than 10% tree cover and barely vegetated, dry sandy soils • Scrub includes grasses which are low height and characterized by lack of shadow on aerial photos • Appears to be cleared areas or sites of disturbance such as logging • Found within range of elevation zones and the wetland 1.1 1.0   43  zone, but more frequently on slopes (often at sites of drainage) • Paler in 1949 photos, dry yellow/white in 2006 data Water  • Dark with very even texture.  • Usually black in historical photos, but sometimes grey or light grey, especially in headwaters.  • Usually dark in modern photos as well, but sometimes blue to light blue.  NA NA     Rock/snow • No vegetation cover • At high elevations, found in the headwaters • Extremely white reflectance with almost no other obvious texture or shading • Also appears as exposed soils on steep slopes • White in 1949 data, pale grey and brown to white in 2006 data    1.5 1.0   44   3.2.4 Mapping ES capacity Land-cover classes derived from aerial photography were combined with other geo-spatial data to quantify ES capacity. I mapped capacity for five ES (orchard production, forage production, carbon storage, paddle route quality, and fish capacity). Aggregating ES data is a standard procedure and here, I aggregated ES at the scale of a river reach. A river reach is an ecologically important unit for river-floodplains and was defined as a river-floodplain segment 10-20 river widths in length and included adjacent floodplain areas. ES capacity was normalized by reach area to account for differing reach sizes and to ensure better representation of local ES hotspots (i.e., without normalizing by reach area, ES importance would simply be a function of reach size). My unique mapping design used aerial photography from two time periods to ensure that ES capacity at each time period is distinct, rather than an amalgamation of various timeframes common in ES spatial assessments (Morgan et al. 2010; Tomlinson et al. 2011; Holland et al. 2011). 3.2.4.1 Orchard and forage production  Orchard production capacity was estimated based on area (ha) classified as orchard in my land-cover maps. Reaches most important for orchard production were those with the largest area of orchards normalized by the reach area. Similarly, forage production capacity was estimated based on area in fields using land-cover maps.  45  3.2.4.2 Above-ground carbon storage I assigned above-ground carbon storage values to forest cover composition and density and land cover types based on estimates from local FIA (Forest Inventory and Analysis) data plots and COLE (Carbon OnLine Estimator) (O’Connell et al. 2014; Van Deusen and Heath 2014). Using all FIA conifer plots within the Wenatchee watershed weighted by the number of plots in each species class, I used the upper 3rd quartile of carbon storage value for dense conifer forests and the lower 1st quartile carbon storage value for moderately dense conifer forests. For mixed conifer broadleaf forests, I assumed a 50% conifer 50% broadleaf composition.  Because there is only one hardwood FIA plot found in my study area, my area incorporated hardwood plots found in Chelan County to estimate hardwood carbon storage. Replicating my approach for conifer stands, I used the upper 3rd quartile value for dense, mixed conifer broadleaf, and the lower 1st quartile value for moderately dense mixed conifer broadleaf. Above-ground carbon storage values for other land-cover types (including, orchards, urban, fields, and shrubs) were derived from published estimates (Table 2). Carbon storage was totaled, normalized by reach area, and reported in tC/ha, thus distinguishing reaches with the highest areal carbon storage. While large reaches might store more total carbon, normalizing (on a per hectare basis) accounts for local hotspots of carbon storage. Although below-ground carbon storage is likely important (and highly variable) in my study area, I did not include below-ground carbon storage in the analysis, in part because soil characteristics are challenging to distinguish using aerial photography. 46  Table 3.2 Above-ground carbon storage for different land-cover types. Land-cover Above-ground Carbon storage (tC/ha) Source Urban 25.1  (Nowak and Crane 2002) Orchard 63 (Penman et al. 2003) (IPCC) Field 5 (Ruesch and Gibbs 2008) (IPCC) Water 0 Not applicable Conifer forest—Dense  136.9 COLE (Van Deusen and Heath 2014) Conifer forest—Moderately dense 93.0 COLE (Van Deusen and Heath 2014) Mixed conifer broadleaf—Dense  132.8 COLE (Van Deusen and Heath 2014) Mixed conifer broadleaf—Moderately dense 67.4 COLE (Van Deusen and Heath 2014) Shrub (wet shrub and dry scrub) 7.4 (Ruesch and Gibbs 2008) (IPCC) Rock/Snow 0 Not applicable  3.2.4.3 Paddle routes  Quality paddle routes are influenced by multiple landscape characteristics, which can be seen in aerial photography and supplemented by other data sources. Documented paddle routes were identified using American Whitewater, a popular website for paddlers (http://www.americanwhitewater.org/ 2014), and then manually digitized. If no paddle routes were noted on a reach, this portion was considered unsuitable for paddling.  Most parts of the river contained documented paddle routes with the exception of the upper reaches which likely are inaccessible or have low water levels. I determined whether accessibility to paddle routes changed by mapping all roads within 0.5 km of paddle route start locations in each time period. I found route accessibility did not change, and thus I used identical paddle routes for each time frame.  47  To determine if the quality of paddle routes changed over time, I identified the natural vegetation and land-cover types (including forests, shrubs, wetlands, rock/snow), flanking the paddle routes in each reach. While aesthetic preferences in recreational landscapes are complex and vary among cultural, occupational, and user groups (Gomez-Limon and de Lucio Fernandez 1999), European and American adults generally prefer natural land cover to urban covers (Ulrich 1986). Forested cover and natural cover are also preferred to intensified agricultural cover (Arriaza et al. 2004), yet agricultural land, particularly traditional agricultural land, has been shown to be aesthetically appealing (Bergstrom et al. 1985; Brady 2006). Thus, I considered documented paddle routes with the highest percentage of natural cover to be “high quality” paddle routes, whereas agricultural cover types, followed by urban, were considered of lower quality from an aesthetic perspective. Longer river pathways were incorporated into my index by multiplying percent covers by paddle route length, accounting for the greater amount of travel time spent in such reaches by paddlers. The equation below shows how I integrated multiple landscape characteristics contributing to paddle route quality in a unitless index ranging from 0 to 190,780 before normalizing the values from 0-1.  𝑃𝑎𝑑𝑑𝑙𝑒 𝑟𝑜𝑢𝑡𝑒 𝑞𝑢𝑎𝑙𝑖𝑡𝑦 = 𝑃𝑎𝑑𝑑𝑙𝑒 𝑟𝑜𝑢𝑡𝑒 ∗ (%𝑁𝑎𝑡𝑢𝑟𝑎𝑙 𝐶𝑜𝑣𝑒𝑟 +  0.5(%𝐴𝑔𝑟𝑖𝑐𝑢𝑙𝑡𝑢𝑟𝑒)) ∗ 𝐶ℎ𝑎𝑛𝑛𝑒𝑙 𝑙𝑒𝑛𝑔𝑡ℎ 𝑃𝑎𝑑𝑑𝑙𝑒 𝑟𝑜𝑢𝑡𝑒 = {  1 0 𝑓𝑜𝑟 𝑃𝑎𝑑𝑑𝑙𝑒 𝑟𝑜𝑢𝑡𝑒𝑓𝑜𝑟 𝑁𝑜 𝑝𝑎𝑑𝑑𝑙𝑒 𝑟𝑜𝑢𝑡𝑒 3.2.4.4 Fish Capacity Index I mapped a suite of ecological characteristics important to fisheries in my study area to estimate the capacity of each reach to provide habitat for fish. The landscape characteristics important for 48  fish include the provision of wood for habitat formation, wetlands/ponds, and slow/stagnant channels. The rationale for each of these characteristics is explained next, in further detail. 3.2.4.4.1 Normalized wood importance index Wood in streams is associated with formation of habitat for both fish and micro-invertebrates. For example, higher volumes of wood has been linked with greater numbers of juvenile salmonids in winter (Murphy et al. 1984; Beechie and Sibley 1997). Invertebrates rely on woody debris for habitat and biofilms which form on woody surfaces (Benke and Wallace 2003). A suite of forest and geomorphologic factors influence the ability of a reach to contribute large wood including tree species, channel confinement, and channel sinuosity. For example, coniferous wood remains in stream longer than deciduous wood (Hyatt and Naiman 2001). I combined a suite of factors including forest type, % forest area, and stream sinuosity to account for differences among reaches in their capacity to produce large wood. For forest type, I used a multiplier to account for faster rate of decay (and shorter instream residence time) for deciduous tree species. These factors form the basis of my Wood Importance Index for each reach.  The capacity to contribute wood for habitat formation was estimated by using forested area within 75m of streams (flanking 150m total with 75m on each side of the river, without exceeding the extent of the floodplain). Seventy-five m is the minimum buffer width required by the Washington State Department of Ecology (2013) for eastern Washington for surface waters in agricultural land, and thus, I used a 75m buffer for my normalized wood importance index. Forested land cover types (both mostly conifer and mixed conifer and broadleaf) were considered important for contributing woody. Mostly conifer forest cover types contained no easily distinguishable broadleaf species, while mixed conifer and broadleaf forest contained at 49  least one visible patch of broadleaf trees. Conifer species are generally larger and more decay resistant (Naiman et al. 2000; Hyatt and Naiman 2001; Hart et al. 2013). Thus, the importance of coniferous forests were weighted more than mixed conifer deciduous forests by multiplying the importance value of mixed forested areas by 0.75.  Sinuous channels are disproportionately important for wood production (Nakumura and Swanson 1994), so I multiplied importance value by stream sinuosity.  𝑊𝑜𝑜𝑑 𝐼𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑐𝑒 𝐼𝑛𝑑𝑒𝑥 = (%𝐶𝑜𝑛𝑖𝑓𝑒𝑟 + 0.75(%𝑀𝑖𝑥𝑒𝑑 𝐶𝑜𝑛𝑖𝑓𝑒𝑟 𝐵𝑟𝑜𝑎𝑑𝑙𝑒𝑎𝑓)) ∗ 𝑆𝑖𝑛𝑢𝑜𝑢𝑠𝑖𝑡𝑦 Longer reaches (in the lower portions of the floodplain) with wider channels require larger trees and wood volume for geomorphic effects (Gurnell and Pie 2002).  Thus, from a habitat formation perspective, large vs. small reaches are comparable even though total area in forest might larger in larger reaches. I accounted for this by using % forest cover within the 75m buffer in each reach. Percent forest cover allows us to account for differing reach areas. I normalize wood importance (see equation above) from 0 to 1 based on the max observed at both timeframes to create a normalized wood importance index.  3.2.4.4.2 Fish habitat structures Fish reared in floodplain habitats (such as slow/stagnant channels) can have higher growth and survival rates than those in the main channel (Sommer et al. 2001; Jeffres et al. 2008). Ponds and wetlands are critical refugia during times of low flow (Robinson et al. 2002). Mapping specific habitat features allows us to identify reductions in them, and thus identify reaches where the capacity of floodplains to produce fish has likely declined. Habitat characteristics important for salmonids, such as ponds/wetlands and slow/stagnant channels, were digitized at each time frame 50  (Tomlinson et al. 2011). The importance of each habitat feature was found by dividing the habitat area by reach area (e.g., area dry channel/reach area), this value was scaled from 0-1 using the maximum and minimum for both timeframes. These habitat characteristics were combined into a (unitless) fish capacity index calculated as follows.  The final values of the index ranged from 0 - 1.05 with a max possible of 4.00. 𝐹𝑖𝑠ℎ 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝐼𝑛𝑑𝑒𝑥= 𝑁𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑 𝑤𝑒𝑡𝑙𝑎𝑛𝑑 𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑐𝑒+ 𝑁𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑 𝑠𝑙𝑜𝑤 𝑎𝑛𝑑 𝑠𝑡𝑎𝑔𝑛𝑎𝑛𝑡 𝑐ℎ𝑎𝑛𝑛𝑒𝑙 𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑐𝑒+ 𝑁𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑 𝑑𝑟𝑦 𝑐ℎ𝑎𝑛𝑛𝑒𝑙 𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑐𝑒 + 𝑁𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑 𝑤𝑜𝑜𝑑 𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑐𝑒 3.3 Analyses 3.3.1 Land cover change and accuracy assessment I determined the change in areal extent (ha) of each land cover type. Then, I used a transition matrix to quantify rates of conversion, using a stratified random sample design with 500 points per land cover classes. I assessed the accuracy of the 2006 land cover classes (forest/shrub, orchard, field, urban) using ground-truthing. Field data for ground-truthing was collected with a Trimble GPS Pathfinder ProXT receiver. My accuracy points include field data from Tomlinson et al. (2008), as well as new points collected in 2011 to assess accuracy of orchards vs. fields. A total of 238 points were sampled over three days in August 2011 in orchard and field covers geographically representative of agricultural areas across the Wenatchee system floodplain. Such points supplemented Tomlinson’s 406 points spread across the Wenatchee system floodplain  Determining the accuracy of historical maps is challenging (Schulte and Mladenoff 2001), because reference data is rarely available or inappropriate in scale (Manies and Mladenoff 2000). 51  Here, my results represent a best-case scenario for historical land cover as accuracy assessments are unfeasible with my historical imagery (Tomlinson et al. 2011). The accuracy of forest types (density classes and conifer vs deciduous) in modern and historical photos were not assessed with field data but were cross-validated by two aerial photography interpreters to estimate accuracy (See appendix).  3.3.2 ES capacity change I summarized changes in ES capacity for the entire floodplain and determined the percent change for each ES. % 𝐶ℎ𝑎𝑛𝑔𝑒 =  𝐸𝑆𝑓𝑖𝑛𝑎𝑙 − 𝐸𝑆𝑖𝑛𝑖𝑡𝑖𝑎𝑙𝐸𝑆𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑋 100 I determined if changes in ES capacity at the reach level were significant using a paired t-test of Wilcoxon signed ranked scores at the reach level. For this analysis, I included only reaches which contained the ES in at least one of the time periods. Bonferroni corrections were unnecessary, because each of these tests were examining patterns of individual services. For ES capacity based on multiple characteristics (paddle route quality and fish capacity), I also determined whether any sub-characteristics differed over time.  52  3.4 Results 3.4.1 How have land cover and ES capacity changed over time? 3.4.1.1 Changes in land cover Land cover in the Wenatchee system floodplain changed considerably from 1949 to 2006 (Figure 3.3). The largest overall change was an increase in urban cover by 1846 ha (+502%). In contrast, area in orchard and fields decreased by 764 ha (-20%) and 518 ha (-24%), respectively. In general, fields were most likely to be converted for urban uses (Table 3). Overall forested areas declined only 118 ha (1%), while shrubby areas declined by 469 ha (-19 %). Conifer forest increased by 1102 ha (17%), while mixed confer broadleaf forests decreased by 1226 ha (-24%). Forests became denser over time. Forested areas classified as high density increased by 470 ha (5%), while low density forests (including shrubby areas) decreased by 489 ha (25%), and moderately dense forests also decreased by 588 ha (31%).  53   Figure 3.3 Land-cover change in hectares across the Wenatchee system floodplain. The greatest total change was an increase urban land-cover type, followed by a decrease in total orchard, fields, shrub cover. While overall forest area only declined slightly (118 ha), mixed conifer broadleaf forests declined, and conifer forests increased. Very minor changes in area in water (17 ha) and rock/snow (19 ha) were also detected (not shown).   Urban land cover was the most persistent (annual probability of persistence = 1.000) (Table 3.3). Orchards were also relatively persistent (0.994), but were converted to fields (annual probability of transition = 0.003) or urban areas (0.003). Subsequent rates reflect annual probability of transition. Fields were consistently converted to urban cover (0.005) and to orchards (0.003). Shrubby land cover frequently transitioned to high density conifer forest (0.003) and high density mixed conifer broadleaf forests (0.003). Moderate density conifer forests very frequently transitioned to high density conifer forests (0.008). High density mixed conifer broadleaf forests often transitioned to high density conifer forests (0.003), while moderate density mixed forest 54  often transitioned to high density mixed forests (0.004) and high density conifer forests (0.005) (Table 3.3).  Table 3.3 A transition matrix converted to an annual time-step shows the persistence (in bold) of different land-cover types in the Wenatchee floodplain based on 5000 random points. Urban land-cover was the most persistent followed by forests. Fields were the most likely to be converted for urban uses followed by orchards. MC—Mostly Conifer, MCB—Mixed conifer broadleaf.  Urban Orchard Field Shrub MC-High density MC-Moderate density MCB- High density MCB- Moderate density Water Rock/snow Urban 1          Orchard 0.003 0.994 0.003        Field 0.005 0.003 0.991 0.001       Shrub 0.001 0.001 0.001 0.987 0.003 0.002 0.003 0.001 0.001  MC-High density    0.001 0.994 0.001 0.003 0.001   MC-Moderate density 0.001   0.001 0.008 0.986 0.003 0.001   MCB-High density    0.001 0.007 0.001 0.989 0.001 0.001  MCB-Moderate density 0.001   0.002 0.005 0.002 0.004 0.985 0.001  Water    0.001 0.002  0.002 0.001 0.994  Rock/snow    0.01 0.004 0.002   0.001 0.983  3.4.1.2 Changes in ES capacity Overall, ES capacity was reduced for four of five services quantified. Orchard production was reduced by 20% and forage production by 24%.  Paddle route quality was also reduced by 5%. 55  Fish capacity decreased by 10%. Carbon storage increased slightly (1%), likely due to gains in high-density conifer forests (Figure 3.4)  Figure 3.4 Changes in ES capacity from 1949-2006 across the Wenatchee system floodplain. The river-floodplain’s capacity to provide four of five services examined declined. Carbon storage increased only slightly. While capacity to provide four of five ES was reduced overall, at the reach level different patterns emerged. Changes in ES capacity were significant at the reach level for orchard production, forage production, and carbon storage (Table 3.4). Changes in paddle route quality were not significant at the reach level, but several landscape characteristics used to create this index were lower when compared at the reach level (Table 3.4). For reaches in paddle routes, natural cover decreased significantly at the reach level (p <0.01), as did agricultural cover (p 56  <0.01). Furthermore, channel length decreased significantly (p < 0.01). Fish capacity did not change significantly at the reach level (p = 0.22), but several landscape characteristics contributing to this index did. Wood Importance Index for habitat formation increased significantly (p = 0.02). Furthermore, ponds and wetlands increased significantly over time (p = 0.02). Dry channels decreased significantly (p = 0.001). While the area of slow/stagnant channels and the degree of channel sinuosity decreased over time, such changes were insignificant at the reach level (p= 0.65 and p = 0.26 respectively).  57  Table 3.4 Changes in ES summarized at the reach level (Wilcoxon paired-sample test). Significant results are shown in bold. ES were normalized from 0-1 using the maximum and minimum values from both time periods. All changes were significant at the reach level with the exception of paddle route quality and slow/stagnant channels. Ecosystem Service n (number of reaches with an ES at either timeframe) Mean normalized initial importance (1949) Mean normalized final importance (2006) Mean change Significance (p) Orchard production 27 0.669 0.594 -0.075 0.05 Forage production 76 0.249 0.196  -0.054  0.01  Carbon Storage 424 0.670  0.704   0.035 0.02 Paddle Route quality 198 0.400  0.380  -0.021  0.30  Natural cover 198 0.852 0.830 -0.0212 0.00026 Agricultural cover 69 0.412  0.346  -0.00660 7.9e-05 Channel Length 198 0.243 0.232 -0.0109  0.0014  Documented paddle route 198 -- -- -- No change Fish capacity index 424 0.087  0.078  0.009  0.22 Large wood for streams 424 0.075  0.081  0.006  0.02 Pond/Wetlands 75 0.060 0.080  0.019  0.02  Slow/stagnant channels 55 0.165  0.135  -0.03  0.65  Dry channels 59 0.232  0.149  -0.0832  0.01  Sinuosity 424 0.0473  0.0469  -3.95e-04  0.26   3.5 Discussion Urban expansion was the main driver of change, and I identified both direct and indirect effects of urbanization on ES capacity. Fields and orchards overwhelmingly transitioned to urban uses reducing local agricultural capacity for both orchards and forage production. This also had 58  indirect impact on the quality of the river for recreational paddling, reducing aesthetics and potentially contributing to shorten channel lengths. The experience paddlers are having in the Wenatchee River System is shortening (due to shorter river lengths), and is less aesthetically appealing; however, such effects are not yet apparent at the reach level. Furthermore, fish habitat features also declined in area, which has concerning implications for habitats used for juvenile rearing (slow and stagnant and dry channels). Declines in dry channels, sinuosity, and slow/stagnant channels likely contributed to the overall decline in my fish capacity index. Simultaneously, the Wenatchee system floodplain purportedly gained refugia habitat features including ponds and wetlands, however, given the seasonal differences between the two sets of air photos, and glare in the upper reaches of the historic aerial photography, further investigation of this result is warranted (Tomlinson et al. 2011). Finally, slight gains in carbon storage and potential wood for habitat formation are likely due to the increase in shrubby areas transitioning to forests and the densification of both conifer and mixed conifer broadleaf forests.  My results demonstrate losses in the capacity of my landscape to provide four of the five ES examined, yet likely, these losses would have gone undetected measuring ES flows alone. While information regarding ES flows was not available at this fine scale, I hypothesize ES flows of many ES increased over this timeframe. Carbon sequestration (an ES flow) likely increased due to altered fire regimes resulting in increasing forest density. In addition, agricultural yields (also an ES flow) likely increased due to increases in fertilizer and technology for high-yielding orchards. This would reflect yield increases seen globally (Wik et al. 2008). Furthermore, recreational paddling likely increased due to a higher number of visitors to the Wenatchee system floodplain, a simple function of larger human populations. Fish populations likely 59  changed with reductions in wild fish and increases in hatchery-reared fish. Furthermore, fish consumption likely increased in the area due to larger populations of people. Thus, an ES flows approach might suggest improvement in ES across the Wenatchee system floodplain. Future work should examine ES flows and ES capacity concomitantly, which can determine if ES flows are sustainable given an area’s ES capacity. While comparing capacity and demand for individual ES has been attempted (Villamagna et al. 2014), a multi-service approaches exploring flows and capacity has yet to be explored.  For our first objective, I quantified rates of conversion among land cover classes. My work suggests that land cover overwhelming transitioned from agricultural cover to urban cover, while forests are becoming denser and more dominated by conifers. The lower portion of the Wenatchee system floodplain was almost entirely converted to orchard and forage production as of 1949 (~55% of the Wenatchee main stem). Though threatened by urban expansion, this legacy persisted with ~40% of the Wenatchee main stem floodplain remaining in agricultural production in 2006 (Tomlinson et al. 2011). Extensive floodplain cultivation is not unique to the Wenatchee system floodplain. Floodplains have been used for cultivation across the globe, and the legacies of cultivation will likely be pervasive and persist for centuries. In addition to cultural legacies of preferences for certain crop types over decades, ecological legacies may mean ES may differ in formerly cultivated land due to differences in soil nutrient content for millennia (Dambrine et al. 2007).  Our second objective demonstrated an approach using aerial photography to measure spatially explicit change in the capacity of a floodplain to provided ES and revealed some compelling differences. ES capacity declined with increasing urbanization, and the impacts of historical 60  forest management and agriculture may still be evident after nearly a century. Broadly, the floodplain showed a trend of ES capacity decline from 1949 to 2006 with losses in four of five ES capacities quantified across the Wenatchee system floodplain. Two of these losses and one gain were statistically significant when compared using paired t-tests at the reach (local) level.  Urbanization was the primary driver of contemporary change. The transition of field and orchards to urban cover was pervasive throughout the lower Wenatchee floodplain, as in other floodplains throughout North America and in other areas of the world (Maizel et al. 1998; López et al. 2001; Foley et al. 2005). Fertile soils underlying agricultural production are a non-renewable resource, providing a suite of functions for people including food production, habitat, biomass production, etc. (Gardi et al. 2014). Increases in impervious area precludes soil from providing many of its multiple functions (Gardi et al. 2014). Losses in productive farmland resulting from urban expansion can have major implications for local and global food security (López et al. 2001; Chen 2007). Globally, the loss of productive farmland is concerning as total populations are rapidly increasing (Amundson et al. 2015). Furthermore, because such changes are irreversible over practical timeframes, this loss of fertile soil mostly to urban cover is particularly problematic. Additionally, local farmland contributes to other ES, including agri-tourism, aesthetics, and carbon storage (Ma and Swinton 2011; Amundson et al. 2015). One limitation of my approach was not including soil carbon pools. Incorporating soil carbon pools would further advance our understanding of how urban expansion is changing ES such as carbon storage and climate regulation, meaning losses in capacity to store carbon is likely greater than demonstrated here. Expanding urban areas may degrade the very ES which has resulted in increasing populations.  61  3.5.1 Legacies of historical ES The traces of historical preferences for ES remain evident throughout the dynamic Wenatchee system floodplain including historical agricultural expansion, and possibly frontier logging and 20th-century fire suppression. Legacies of early-century logging and fire suppression are common across the Pacific Northwest (Hessburg et al. 2005). Here, they are evident in densifying forests becoming more dominated by conifer species. Showing typical successional patterns transitioning from deciduous forests to dense conifer stands, changes in forest structure have also influenced ES in the Wenatchee in the second half of the 20th century. While research has suggested forests have become denser in uplands in the Eastern Cascades (Hagmann et al. 2014), rarely have the density characteristics of riparian zones in this region been explored. Such changes have implications for a suite of services including water availability (from increased transpiration), carbon storage, etc. Further work on structural composition of forests is warranted to determine how such changes may influence carbon storage, provision of large wood, and water availability. My timeframe could be capturing a forest stand exclusion phase after recovering from logging in the early 20th century, and/or forests in the valley could be impact by fire exclusion as typical throughout the western USA. In either scenario, forests are likely denser than typical of pre-settlement forests. Even longer baselines are needed to detect such changes.  As a regulating ES, conceptualizing the benefits that carbon storage and the carbon cycle provides for people remains a challenge (Law et al. 2015). An ES occurs when the carbon cycle provides a stable and regular climate for people. While denser forests may mean forests currently store more carbon, this may also mean forests are closer to reaching their maximum capacity of carbon storage. If additional carbon storage by forests is needed to regulate climate, dense forest 62  may be less capable of providing this additional carbon storage. While carbon storage in this area increased, providing a greater amount of climate regulating ES, this also means that the capacity of this area to provide further carbon storage has actually decreased.  Historical context and tracking changes in the capacity of landscapes to provide ES over time can provide important lessons for understanding ES. An exercise in understanding change in ES over time leads to a clear need for distinguishing ES and ES capacity, as ES measures flows to people and ES capacity measures stocks. Measuring changes in flows alone may be misleading as increases in populations will certainly increase flows of ES. As contemporary landscapes are a snap shot of combined social and ecological processes, a long-term approach may reveal important trends in sustainability and environmental degradation. Furthermore, long-term approaches may detect slow changes in ES capacity unnoticed over short timeframes. An historical approach to ES capacity is essential, because the physical template formed from multiple layers of landscape history constrain contemporary and future options for ES capacity. 3.6 Conclusion Widespread changes in land cover in the Wenatchee system floodplain resulted in declines in its capacity to provide multiple ES. Urbanization and forest densification were the main drivers of change. Such dynamics are relevant across North America and the globe as urbanization is widespread, and forest recovery from frontier logging and resource extraction. These maps represent some of the longest, reconstructions of ES capacity in a spatially explicit manner to date. Losses in multiple ES over the past century suggest historical reconstructions are invaluable for understanding ES dynamics. The legacies of frontier settlement economies (resource extractions and agricultural production) are still seen on the landscape today. Future work should 63  explore corresponding changes in ES capacity and flows, allow us to assess the sustainability of modern ES flows.   64  Chapter 4: Ecosystem service interactions misunderstood without landscape history 4.1 Introduction People rely on a wide range of ecosystem services (ES) from landscapes and ecosystems. Ensuring landscapes provide multiple benefits to diverse user groups is an ongoing challenge, because ES are interrelated and can interact in complex and unexpected ways (Bennett et al. 2009). When ES respond to shared drivers such as land conversion or restoration, changes in one ES can impact other ES directly as well as indirectly (Bennett et al. 2009). Two commonly characterized interactions are trade-offs and synergies (Figure 4.1). A trade-off occurs when an increase in one ES leads to a decline in another ES (such as crop production reducing fish habitat or impacting water quality). A synergy occurs when enhancement of one ES increases another ES simultaneously. Floodplain restoration for fish habitat enhancing recreational paddling is one example.   Figure 4.1 Hypothesized interactions among floodplain ES. Different drivers can lead to interactions such as trade-offs and synergies among ES. Adapted from Bennett et al. 2009. 65  To avoid problems resulting from ES interactions, governments and managers throughout the world are adopting an ES perspective. For example, the Intergovernmental Panel for Biodiversity and Ecosystem Services was established to assess global biodiversity and ES in 2012; whereas the European Union aims to halt ES and biodiversity loss by 2020 (European Commission 2011). Compliance with the United States National Environmental Policy Act now requires ES monitoring, and land managers are currently developing guidelines for such efforts (Bear 2014). Despite increased integration of ES into policy and decision making, ES monitoring standards are still evolving. While monitoring individual ES is challenging, monitoring multiple ES and their interactions is particularly difficult, in part because standards for understanding ES interactions are poorly established and major conceptual and methods gaps still exist.  Understanding ES interactions at different temporal scales is one key challenge (Birkhofer et al. 2015). ES interactions may involve time lags whereby historical decisions influence current provisioning of ES (Dallimer et al. 2015). A landscape baseline can be used as a reference point against which a current landscape ES can be measured and compared. Ecosystem baselines have been used in landscape ecology to better understand changes in ecosystems over time and under different scenarios (Bull et al. 2014). Such approaches may also be useful for understanding ES interactions. Different landscape contexts and history also likely impact ES interactions. More progress has been made assessing ES over space whereby mapping is used to infer ES interactions (Raudsepp-Hearne et al. 2010a; Qiu and Turner 2013). Typically, multiple ES are identified across broad regions and any spatial overlap (or lack thereof) is assumed to signify ES interactions. The spatial overlap is often quantified using Spearman’s rank correlations where positively correlated ES are assumed to be synergistic whereas negatively correlated ES are 66  presumed to trade-off (Lautenbach et al. 2010; Raudsepp-Hearne et al. 2010a). This approach is analogous to a space-for-time substitution in that spatial relationships are used to infer dynamics over time. Space-for-time substitution, widely used in ecology where long-term records are unavailable, is based on the assumption that temporal and spatial variability are equal (Pickett 1989). Unfortunately, using broad-scale spatial correlations among ES to infer tradeoffs and synergies ignores several fundamental assumptions of a space-for-time approach.  To address this limitation, I explore a contrasting approach, incorporating actual temporal changes in ES to clarify the nature of ES interactions. I create ∆ES maps which show differences in ES from 1949 to 2006. Using these ∆ES maps, I correlate pairs of ∆ES across a river-floodplain landscape. This approach allows for identification of which ES change concomitantly (i.e., trade-offs or synergies). I contrast a ∆ES (change-over-time) approach to the standard space-for-time approach in both 1949 and 2006 where ES are mapped across a landscape and static correlations are used to infer ES interactions. Correlational direction of the two approaches might disagree in several ways (Figure 4.2), with consequences for understanding and characterizing ES interactions, explained next. 67    Figure 4.2 Boxes represent scatter plots showing the relationship between two ES. Positively correlated ES are considered synergies, negatively correlated ES are trade-offs, and uncorrelated ES do not interact. S-Synergy, T-Trade-off, N-No interactions. Comparing space-for-time and change-over-time approaches, a trade-off can be missed when the space-for-time approaches show no interaction or a synergy, but the change-over-time approach shows a trade-off (Panel A). Synergies missed occur when space-for-time approaches show either a trade-off or no interaction, but the change-over-time approach shows a synergy (Panel B). 68   4.1.1 Trade-offs or synergies may go undetected  Space-for-time approaches may miss ES tradeoffs which are more easily detected by change-over-time approaches (Figure 4.1, Panel A). Trade-offs may be missed when ES occur in locally similar locations and compete for space. Competition among different agricultural crops within the same fertile floodplain zone may exemplify this issue. Two crop types might appear to co-occur in space (i.e., a synergy), but actually be competing for space when viewed over time. Thus, increases in one crop will lead to decreases in the other crop over time (i.e., a trade-off). Such undetected trade-offs are likely when ES data are aggregated by geographical or political units (e.g., watersheds or counties), as is common in ES mapping (Raudsepp-Hearne et al. 2010a). Individual ES may also be spatially autocorrelated, which can make identifying ES interactions challenging.  Similarly, a synergy over time might not be evident using a space-for-time approach. When ES rely on common landscape features or attributes, yet are spatially disparate, such ES appear potentially synergistic over time. For example, recreational activities and wildlife habitat may both be linked to forest cover attributes yet not occur in the same specific locations. ES maps created at a single timeframe may suggest a tradeoff but when viewed over time, a clear synergy is evident among such ES (Figure 4.1, Panel B).  Here, my goal is to determine if ES interactions determined from a map difference (change-over-time) approach provide additional insights to a spatial (space-for-time) approach. I explore multiple globally-relevant ES, including: fish production, agricultural production, recreation, and 69  carbon storage. I ask two primary questions: Does examining ES change-over-time (∆ES) yield a different understanding of ES interactions than a space-for-time approach? Does the degree of spatial autocorrelation among ES influence detection of interactions? I explore this question by comparing correlational direction among pairs of ES mapped at one timeframe to correlational direction incorporating prior historical ES and their change over time (∆ES) using historical and contemporary aerial photography. I explore ES interactions in a tributary river-floodplain system to the Columbia River, the Wenatchee system floodplain.  4.2 Methods 4.2.1 Study site The Columbia River Basin (CRB) is among the world’s most managed and regulated river basins, resulting in a number of unintended ES interactions. To exemplify these issues, which require high-resolution approaches, I focus on one of the Columbia River’s tributary river-floodplain system, the Wenatchee system floodplain. The 210 km2 Wenatchee floodplain system, is comprised of the Chiwawa River, White River, Little Wenatchee River, Nason Creek, and the Wenatchee mainstem floodplains (Chapter 3, Figure 3.1). Draining the eastern side of the Cascade Mountains, its variable soils, vegetation, as well as west to east precipitation gradient make this floodplain ideal for studying variability in ES interactions. The watershed’s diverse user-groups and ecosystems make understanding its ES interactions critical. This river-floodplain has been rapidly urbanizing since the mid-20th century. The Wenatchee mainstem floodplain has been primarily converted to orchard agriculture and urban development, while its tributary floodplains are primarily forested and traversed by road networks. The Wenatchee system floodplain is a popular destination for recreation and agritourism. 70  4.2.2 Mapping ES capacity Land-cover classes derived from aerial photography were combined with other geo-spatial data to quantify ES capacity (See Chapter 3). I mapped capacity of the floodplain to provide five ES (orchard production, forage production, carbon storage, paddle route quality, and fish capacity) in both 1949 and 2006. I aggregated ES at the scale of a river reach, as aggregating ES is a typical approach for mapping ES. See Chapter 3.2.4 for details on land cover and ES mapping.  4.2.3 ES interaction analyses I mimicked the general approaches of research quantifying ES interactions (e.g., (Raudsepp-Hearne et al. 2010a; Wu et al. 2013; Turner et al. 2014) by determining correlations between amounts of ES of river reach segments.  Then, I compared the direction and strength of correlations calculated using a static ES approach for both time frames 1949 and 2006, as in (Raudsepp-Hearne et al. 2010a) with correlations calculated using map differences from 1949 and 2006 (∆ES). I refer to these contrasting analyses as the space-for-time approach and change-over-time approach (∆ES) respectively. I did this for each pair of ES. Due to non-normality of ES capacity distributions, I used Spearman’s rank correlations, a standard approach for evaluating ES interactions (Raudsepp-Hearne et al. 2010a). Negatively correlated ES (p <0.05) were considered trade-offs, while positively correlated ES were considered synergistic. Non-significant correlation coefficients indicated non-interacting ES. To better understand the impact of spatial autocorrelation on our analyses, we performed Global Moran’s I (binary connectivity based on shared boundaries) for each ES. If ES were spatially autocorrelated, we also used a Dutilleul modified t-test to determine how spatial autocorrelation may influence the outcomes of ES interactions analyses.71   Figure 4.3  Approach for comparing spatial correlations of two different ecosystem services (ES) at one year with change-over-time approaches (∆ES). I correlate the spatial distributions of ES at the reach level and compare these results to those correlating map differences in ES from 1949 to 2006 (∆ES). 72  4.3 Results 4.3.1 High agreement between space-for-time at different timeframes A space-for-time approach, used to analyze mapped ES in 1949 and 2006, characterized ES interactions similarly (Table 4.2). Trade-offs among orchards/carbon, orchards/fish capacity, forage/carbon storage, and carbon storage/paddle routes were detected in 1949 and in 2006. In 1949, one additional tradeoff was detected between forage/fish capacity. Synergies detected in both years were orchard/forage, orchard/paddle routes, and forage/paddle routes. An additional synergy (carbon storage/fish capacity) was detected in 1949.  4.3.2 Little agreement between space-for-time and change-over-time approaches  Space-for-time and change-over-time approaches rarely agreed in their characterizations of ES interactions. For all possible pairwise interactions, space-for-time results agreed with change-over-time results in two instances for 2006 and three instances in 1949. Both approaches (comparing all years) identified a synergistic relationship between orchard production and paddle routes and also characterized a lack of interaction between forage production and fish capacity (Table 4.2). Change-over-time approaches and space-for-time in 1949 also both identified a synergy between carbon storage and fish capacity.  Different approaches consistently differed in their characterization of all trade-offs, detecting trade-offs among different ES pairs (Figure 4.4). The space-for-time approach and change-over-time approach detected different numbers of trade-offs (Table 4.2). The space-for-time approach identified more trade-offs (five in 1949 and four in 2006) than the change-over-time approach (two for ∆ES) (Table 4.2).  73  4.3.3 Missed trade-offs and synergies using space-for-time approach Two trade-offs with forage production were missed using the space-for-time approach (forage/orchard production and forage/paddle routes). In all cases, the space-for-time approach identified these pairwise interactions as synergies whereas the change-over-time approach identified them as trade-offs. The space-for-time approach consistently missed two synergies found using change-over-time (carbon storage/paddle routes and paddle route/fish capacity). For the missed synergies with fish capacity (involving carbon storage and paddle routes), the space-for-time (2006) results suggested no pairwise interaction. The space-for-time approach also identified a trade-off between carbon storage/paddle routes, while change-over-time showed these ES were synergistic. 4.3.4 Interactions detected where none occur over time The space-for-time approach (2006) detected three ES trade-offs (orchard production/carbon storage, orchard production/fish capacity, forage production/carbon storage) where no interactions occurred according to the change-over-time approach (Table 4.2).  Additionally, space-for-time approaches in 1949 identified a trade-off between forage production and fish capacity that was not detected in 2006 and not detected over time.  4.3.5 ES positively spatially autocorrelated ES were all significantly positively spatially autocorrelated (Table 4.1). Moran’s I ranged from 0.32-0.881 (p < 0.01). Spatial autocorrelation was higher for static maps of ES, while ∆ES spatial autocorrelation was generally lower. We performed a Dutillel modified t-test to acknowledge the influence of spatial autocorrelation on ES interaction detection (Table 4.2.) 74  Table 4.1. ES were positively spatially autocorrelated at each time frame. ES change was also positively autocorrelated, but to a lesser degree. *Significant at 0.05  1949 2006 ∆ES Ecosystem Service Moran’s I Moran’s I Moran’s I Orchard production *0.881 *0.883 *0.393 Forage production *0.738 *0.697 *0418 Carbon Storage *0.708 *0.677 *0.499 Paddle routes *0.820 *0.794 *0.355 Fish  capacity *0.478 *0.402 *0.320  4.3.6 Spatial autocorrelation decreased number of significant interactions Accounting for spatial autocorrelation using dutilleul’s modified t-test resulted in fewer interactions for both space-for-time approaches and change over time approaches. Here, 1949 and 2006 both detect two synergies (paddle route quality/forage production, and orchard production/forage production) and two trade-offs (carbon storage/orchard production, carbon storage/forage production) (Figure 4.5). Change-over-time approaches detected two synergies (paddle routes/orchard production and paddle routes/carbon storage) and one trade-off (forage production and paddle route quality). 75  Table 4.2 The types of interactions identified using two mapping methods: space-for-time substitution (at both 1949 and 2006) and change-over-time (∆ES). Pairwise Spearman’s rank correlation result. T—Trade-off, S—Synergy, N—No interaction. *Significant at 0.05 Ecosystem Service Pair Interaction (1949) Spearman’s rank correlation coefficient (1949) *p<0.05 Dutilleul modified t-test Interaction (2006) Spearman’s Rank Correlation coefficient (2006) *p<0.05 Dutilleul modified t-test Interaction over time Spearman’s Rank Correlation coefficient (∆ES) Dutilleul modified t-test Orchard production Forage production S *0.57 *0.32 S *0.64 *0.26 T *-0.12 -0.10 Orchard production Carbon Storage T *-0.36 *-0.34 T *-0.37 *-0.36 N 0.093 0.091 Orchard production Paddle Route S *0.36 0.32 S *0.31 0.27 S *0.25 *0.55 Orchard production Fish Capacity T *-0.33 -0.12 T *-0.34 -0.11 N -0.03 -0.009 Forage production Carbon Storage T *-0.36 *-0.36 T *-0.45 *-0.42 N 0.066 0.037 Forage production Paddle Route S *0.47 *0.36 S *0.40 *0.29 T *-0.23 *0.22 Forage production Fish Capacity T *-0.13 -0.084 N -0.20 -0.06 N 0.076 -0.007 Carbon Storage Paddle Route T *-0.20 -0.22 T * -0.27 -0.26 S *0.076 *0.16 Carbon Storage Fish Capacity S *0.30 0.11 N 0.23 0.049 S *0.28 0.096 Paddle Route Fish Capacity N -0.036 0.01 N -0.018 0.052 N -0.013 -0.019 76    Figure 4.4 Interaction diagrams for five ecosystem services. Correlations between ES were determined using two different approaches: Space-for-time substitution (left) and change-over-time (right). Circles represent different ecosystem services (ES) and significant Spearman’s Rank correlation coefficients (p < 0.05). Blue lines indicate ES synergies, red lines trade-offs. ES not connected by lines were not significantly correlated. Line width is scaled based on strength of correlation. 77   Figure 4.5 Interaction diagrams for five ecosystem services. Correlations between ES were determined using two different approaches: Space-for-time substitution (left) and change-over-time (right). Circles represent different ecosystem services (ES) and significant Dutilleul’s modified t-test coefficients (p < 0.05). Blue lines indicate ES synergies, red lines trade-offs. ES not connected by lines were not significantly correlated. Line width is scaled based on strength of correlation.78   4.4 Discussion My results demonstrate that analyses using space-for-time versus change-over-time can lead to very different characterizations of ES interactions. Importantly, the space-for-time approach differed most from the change-over-time approach in their characterizations of pairwise trade-offs; the two approaches always identified different trade-offs. Furthermore, the space-for-time approach (2006) identified fewer synergies than the change-over-time approach. Violating or ignoring the basic assumptions of a space-for-time approach may lead to misunderstandings of ES interactions.  4.4.1 Different landscape snapshots, different ES interactions characterized History plays a key role in shaping landscapes, and thus, ES interactions (Rhemtulla and Mladenoff 2007). A measure of baseline conditions is important in showing how ES interactions change over time. Globally, land-use change often follows typical transitions from natural ecosystems to frontier expansion followed by agricultural and urban development (Foley et al. 2005).  Capturing only a snapshot of ES at different transitions may lead to contrasting characterizations of ES interactions. For example, the space-for-time approach for 1949 detected one additional synergy (fish capacity/carbon storage) and one additional trade-off (fish capacity/forage capacity). This demonstrates that not only can using change-over-time approaches capture different ES interactions, but interactions may vary with different landscape timeframes. 79  4.4.2 Landscape history matters: Even deeper baselines needed Typical patterns of ES interactions have emerged using standard space-for-time approaches. Crop production is generally negatively correlated with water quality (Qiu and Turner 2013), carbon, nature appreciation, soil organic carbon (Turner et al. 2014), and habitat conservation (Wu et al. 2013). A trade-off between carbon and agriculture (forage and orchards) was confirmed by my space-for-time approach but not by my change-over-time approach, which showed no interaction between these ES. These differing outcomes may mean: (1) initial agricultural land clearing occurred where above-ground carbon storage was low, resulting in no interaction when viewed over time but a negative interaction when viewed spatially and/or (2) the timeframe considered in the change-over-time approach was too short to capture initial losses in carbon storage from land conversion, resulting in a missed trade-offs.  Contemporary spatial distributions of ES may result from underlying ecological differences rather than an ES decline. Frontier settlers disproportionately selected floodplain and riparian zones (Tomscha and Gergel 2015). Frontier settlers may have also cleared areas with low above-ground carbon storage. Land clearing would involve removing large trees and stumps which would have been challenging with historical technology. For example, historical prairie/savannah with low initial above-ground carbon storage were likely to be converted and persist in agriculture, whereas high-carbon forests were logged, but left to regenerate across Wisconsin (Rhemtulla et al. 2009a).  Alternatively, space-for-time approaches may underestimate the magnitude of ES trade-offs. Early settlers may have cleared areas with the largest trees as underlying soils may be more productive and lumber could have been used for building (suggested in Oregon’s Willamette 80  Basin (Sedell and Froggatt 1984). Similar dynamics might be apparent with wetland losses and reductions of fish capacity—most wetlands were already lost by 1949 (Tomlinson et al. 2011). While my ~60-year timeframe is the one of the longest reconstructions of multiple ES interactions to date, capturing initial losses in ES due to frontier clearing requires even longer timeframes. Unless baseline reconstructions include pre-settlement conditions, these questions will remain unanswered. ES trade-offs vary in temporal scale, taking place rapidly or slowly (Rodríguez et al. 2006). Trade-offs among carbon storage and agriculture are likely rapid and persistent as long as agricultural cover persists. Selecting an appropriate baseline landscape is key to capturing land-cover conversions resulting in brief and strong ES interactions. Such baselines must encompass landscape conditions prior to wide-spread land cover conversion for ES of interest, this is especially challenging for studies exploring interactions among multiple ES. Furthermore, trade-offs among ES in pristine landscapes may appear less severe than those in highly modified landscapes. Agricultural ES will only trade-off if they are competing for space. If there is ample space for agricultural expansion, such ES will not interact or will interact weakly. In my examples, agricultural expansion had already occurred throughout the valley by 1949, in fact agriculture had declined by 2006 (Chapter 3). Thus, this dynamic is not capture in any of my analyses, and multiple baselines may be needed 4.4.3 Space-for-time approach can misconstrue ES interactions  Space-for-time substitution is common in ecological research where temporal records are unavailable and is based on the notion that spatial and temporal variability are equivalent. Thus, it is considered most robust in locations where landscape history is unimportant and where the 81  same drivers remained important throughout the landscape (Pickett 1989). Such assumptions are unlikely to withstand across the extent of most broad-scale ES mapping studies, particularly when landscape heterogeneity is high (e.g., variable terrain, soil fertility, precipitation). In areas with underlying spatial heterogeneity, the original spatial distributions of ES were never likely uniform. However, using a space-for-time approach to infer ES interactions implicitly presumes all locations were originally equally suitable for ES production (e.g., (Raudsepp-Hearne et al. 2010a); thus, ES at low levels in contemporary landscapes are presumed to be the result of prior declines. (Figure 4.6). If ES are spatially disparate, influenced by contrasting underlying landscape features or gradients, a trade-off may be detected using space-for-time approaches. However, a negative spatial correlation does not necessarily indicate a trade-off. For example, different crop types best suited for different soil, terrain, or moisture conditions may rarely be co-located (Baudry and Thenail 2004). When historical baselines are used, different crop types may change independently, showing no interaction over time. Despite widely acknowledged limitations and assumptions of space-for-time substitutions, rarely are their implications explicit in ES research.  82   Figure 4.6 Boxes represent scatter plots showing the relationship between two ES. Positively correlated ES shown in boxes above are considered synergies, negatively correlated ES are trade-offs, and uncorrelated ES do not interact. S-Synergy, T-Trade-off, N-No interactions. Interactions detected where none occur are examples of when space-for-time approaches showing a synergy or a trade-off, but change-over-time showing no interaction. ES and their drivers vary spatially. Most landscape-scale ES interactions studies do not account for this. In this case study, drivers in upland floodplains included road building and forestry, while drivers in fertile lowlands included agriculture and urban development, violating the fundamental assumption of the space-for-time approach that drivers are the same throughout the study area (Pickett 1989). Paired with inherent spatial variability in ES, this likely led to missed synergies in my space-for-time approach (fish capacity/carbon storage, fish capacity/paddle routes, and paddle routes/carbon storage). These ES varied in their spatial locations meaning no interaction was detected using space-for-time approaches, but their temporal changes were linked. For example, paddle routes and fish capacity may differ in spatial distribution because 83  road building increases recreational paddle routes while negatively impacting fish capacity. However, both are improved by reforestation meaning such ES are potentially synergistic when viewed over time. 4.4.4 Data aggregation can obscure trade-offs  Different ways of data aggregation influence characterizations of interactions (Anderson et al. 2009). Incorporating historical data and determining correlations in ∆ES may moderate these issues, because correlating mapped ∆ES shows temporal relationships in addition to spatial relationships. Often, total amounts of ES are spatially aggregated (into watersheds, counties, grids, pixels, etc.) prior to analysis of cumulative or summary values (e.g., percent orchard, total carbon storage, etc.) (Raudsepp-Hearne et al. 2010a; Wu et al. 2013; Turner et al. 2014). Such aggregated information obscures interactions where ES compete for space. For example, agricultural types competing for productive soil will be spatially concurrent in aggregated datasets, thereby showing a synergistic relationship. In the study site examined here, orchard and forage production appear synergistic using space-for-time substitution, but temporally, these ES competed for space (i.e., a clear trade-off). Two trade-offs were likely misidentified as synergies using the space-for-time approach with reach-scale data aggregation: orchards-forage production and paddle routes-forage production. 4.5 Approaches for incorporating spatial autocorrelation into ES interactions research Accounting for spatial autocorrelation in river-floodplain systems requires specialized approaches, which are rapidly developing(Isaak et al. 2014).  However, understanding the spatial properties of river-floodplain ES is complex, because they involved both directional river-84  floodplain processes as well as human-scale use and access processes. Here, I include the results of spatial autocorrelation to illustrate the importance of further understanding ES interactions. Typical approaches for exploring interactions among ES also do not account for this and this study is meant to critique typical approaches. At best, ES studies spatial aggregation is measured with Moran’s I, but not included in subsequent identification of trade-offs and synergies (Raudsepp-Hearne et al. 2010a; Qiu and Turner 2013). Unaccounted for spatial autocorrelation may result in an increase in type I error (Haining 1991; Duncan et al. 2014). 4.6 Implications 4.6.1 Missed synergies are missed opportunities ES synergies present opportunities to enhance multiple ES simultaneously (Anderson et al. 2009; Raudsepp-Hearne et al. 2010a; Qiu and Turner 2013). If land managers depend on space-for-time ES approaches, opportunities for synergies may be missed. Space-for-time approaches missed three important synergies (fish capacity/carbon storage, fish capacity/paddle routes, and paddle routes/carbon storage). For instance, riparian restoration may simultaneously improve fish capacity, carbon storage, and paddle routes, but this potential may not have been realized using only a space-for-time approach.  4.6.2 Missed trade-offs may result in unexpected ES declines Space-for-time substitution missed two trade-offs involving agriculture, instead identifying them as synergies: orchard production/forage production and paddle routes/forage production. When ES trade-offs go undetected, and especially when they are mistaken for synergies, unintentional 85  ES losses may occur. Incorporating baselines may be crucial for correctly characterizing ES trade-offs, especially when using aggregated ES data.  4.7 Conclusion The benefits of using historical information to characterize ES interactions are numerous. Space-for-time substitution may be inadequate for exploring ES interactions, particularly across heterogeneous landscapes affected by multiple drivers. As approaches for ES measurement and mapping evolve, acknowledging the value of baseline information in such assessments is key. The importance of long-term monitoring to adequately capture complex long-term ES interactions cannot be over-stated; it can help us avoid or minimize trade-offs and adequately identify synergies which can simultaneously support multiple ES.     86  Chapter 5: River-floodplain concepts enhance understanding of spatial variability in ecosystem services  River-floodplains are among the most highly modified ecosystems, potentially reflecting their disproportionate importance for ecosystem services (ES) (Tockner and Stanford 2002). Occupying less than 1.4 % of the globe, floodplains provide more than 25% of terrestrial ES (Tockner and Stanford 2002). Floodplains provide a wide range of ES by forming fertile soils for agriculture, habitat for fish, and providing carbon storage (Tockner and Stanford 2002). Some of these ES are threatened by urbanization or endeavors to enhance the provision of ES such as agriculture and hydropower (Sweeney et al. 2004; Foley et al. 2005; Eigenbrod et al. 2011). Because of their importance, efforts to understand spatial trends and patterns of floodplain-specific ES have increased in recent years (Felipe-Lucia et al. 2014); however, the hydro-ecological concepts that underpin floodplain systems are rarely incorporated into ES assessments. Spatial assessments of ES are sorely in need of better approaches for understanding floodplain ES. River-floodplain landscapes are unique in their upstream-downstream connectivity and linear characteristics, and thus require innovative approaches to characterize their spatial distributions. Understanding floodplain-specific ES may benefit greatly from drawing on classic river-floodplain principles. For example, the River-Continuum Concept contributed to decades of ecological research exploring spatial and temporal patterns of river-floodplain ecosystems (Vannote et al. 1980; Thorp et al. 2006). The concept suggests that stream flow, organisms, and ecological communities change in a relatively predictable manner from upstream to downstream (Vannote et al. 1980). For example, the ratio of organism types (shredders, grazers, predators, 87  collectors, microbes) vary with physical variables such as stream flow and temperature (Vannote et al. 1980). While the specifics of this qualitative concept have been debated for decades, and certainly has many limitations, the River Continuum Concept was instrumental in advancing my understanding of spatiotemporal variability in river-floodplains ecosystems (Statzner and Higler 1985; Thorp et al. 2006).  Another important theme in riverine ecology relates the physical template of rivers to their ecological characteristics. Fluvial-geomorphic processes underlie ecological habitat formation and destruction in river floodplains (Fryirs and Brierley 2012), resulting in different fluvial geomorphic reach types (e.g., island-braided, meandering, etc.) and ecological communities. Varying considerably in structure and function, floodplain habitats and the legacies of fluvial-geomorphic processes may influence and constrain ES capacity. For example, reach type may affect spawning habitat (Moir et al. 2004) and contribute to agricultural and recreational capacity, yet details of such ES in relation to fluvial morphology are poorly understood. While frequently assessed in geomorphology research, reach types have not been considered in multi-floodplain ES assessments. Many ES mapping efforts use a combination of land-cover indicators derived from moderate resolution imagery (e.g., Landsat), census information, and modeling to determine important locations for ES (Crossman et al. 2013). However, many floodplain characteristics crucial for ES provisioning can only be detected at fine scales (Gergel et al. 2007). For example, variability in detailed habitat structures of narrow riparian and floodplain areas are indiscernible in moderate resolution or coarse-scale imagery such as Landsat and MODIS (Muller 1997).  In contrast, aerial photography (and other high-resolution satellites such as QuickBird) can provide such 88  detailed information (Gergel et al. 2007; Morgan et al. 2010; Tomlinson et al. 2011). Many ES (e.g., habitat for salmonids, fruit production) and pressures driving their decline (e.g., urban expansion) are only detectable at fine spatial scales, and thus high-resolution mapping is necessary to map ES capacity (Tomlinson et al. 2011). Despite their value, high-resolution ES reconstructions are rare, likely because their creation is time-consuming and may require extensive training (Morgan et al. 2010). Such high resolution approaches are particularly germane for river-floodplain complexity.  Viewing ideas from the River Continuum Concept and fluvial geomorphology through an ES lens, I expect ecosystems, and thus their capacity to provide ES, to vary with floodplain position. I draw on these ideas to explore longitudinal distributions and fluvial-geomorphic relationships with ES in a rapidly urbanizing floodplain. I use high-resolution aerial photography to map ES capacity in a highly modified floodplain landscape and summarize these changes at local reach scale. The River Continuum Concept and fluvial-geomorphic characteristics are easily amenable to ES mapping projects, which typically use remote sensing information. Such widely-researched concepts of river-floodplain patterns have yet to be integrated with research exploring ES spatial patterns. Mapping the floodplain’s capacity to provide ES representing multifaceted values of the Wenatchee system floodplain including recreation, agriculture, fish production, and carbon storage, I ask:  Along a river-floodplain, how do ES capacity and diversity of ES capacity vary with longitudinal position and reach type? 89  5.1 Methods 5.1.1 Study site The Wenatchee River watershed (3440 km2) is a sub-watershed of the Columbia River Basin draining the eastern side of the Cascade mountain range. The diverse topography, vegetation, and strong west to east precipitation gradient of the Wenatchee watershed make it ideal for studying inherent variability in ES capacity along a river-floodplain gradient. The region is a popular destination for regional tourists who visit for paddling (canoeing, kayaking, white water rafting) and agritourism. This river-floodplain has been rapidly urbanizing since the mid-20th century (Tomlinson et al. 2011). While the floodplain of the Wenatchee mainstem has been largely converted to orchards, tributary floodplains remain forested yet traversed by extensive road networks. I focus my work on the floodplain portions of the Wenatchee river system (210 km2 of floodplain, comprised of the Chiwawa River, White River, Little Wenatchee, Nason Creek, and the Wenatchee mainstem floodplains), not only because river-floodplains are often disproportionately important for ES, but also because very little is known about variability in ES within the floodplain (Figure 5.1). 90   Figure 5.1 The Wenatchee watershed in central Washington State where the floodplains of the Chiwawa River, White River, Little Wenatchee River, Nason Creek and the Wenatchee main stem comprise the Wenatchee system floodplains examined in this work.  91  5.1.2 Aerial photographs and floodplain delineation ES capacities were mapped using aerial photography in order to discern floodplain-specific features relevant to ES capacity using orthomosaics (and other geodata) from previous work in the Wenatchee floodplain (Tomlinson et al. 2011). Imagery from July 2006 (1:40,000) was acquired from United States Department of Agriculture National Agricultural Imagery Program (USDA 2006). I mapped ES capacity only within the floodplain valley bottom; however the extent of the original historical floodplain is not discernable on modern floodplain maps which depict changes from dams and levees. As such, the floodplain extent was digitized using a hillshade derived from USGS 10m DEM and 1:24,000 topoquads (Tomlinson et al. 2011). ES capacities were summarized at the reach scale, an ecologically meaningful unit for floodplain analysis. Reach segment lengths were digitized as 10-20 times local stream width (Tomlinson et al. 2011). For further details on floodplain mapping, see Tomlinson et. al., 2011.  5.1.3 Reach type identification Fluvial-geomorphic characteristics vary considerably along river-floodplains. I draw on reach types identified in Tomlinson et al. (2011) to determine if different reach types were associated with different ES. Defined by confinement, sinuosity, gradient, and visual characteristics, I distinguished five reach types (straight, meandering, island braided, pool/riffle, and plane bed) (Tomlinson et al. 2011). Reaches were mapped at the 1:4000-1:5000 scale. Unconfined reaches included straight, meandering, and island braided reaches. Straight reaches include those with a single thread and low sinuosity < 1.5. Meandering reaches also had a single thread, but a sinuosity >1.5. Island-braided reaches contained multiple threads and islands with permanent vegetation. Confined reaches mapped included plane bed reaches (gradient = 1.1-3.0%, with 92  relative featureless bed) and pool/riffle reaches (gradient ≤ 1.0%, with undulating bed creating bars, pools, and riffles) (Tomlinson et al. 2011).  5.1.4 Longitudinal river-floodplain position  Reaches were ranked based on distance to the Wenatchee River’s mouth on the Columbia River (rank = 1 at the Wenatchee/Columbia River confluence and rank = 164 farthest upstream). Of the 424 reaches, 98 reaches were tied with at least one other reach in their rank. I used these reach rankings to determine how river-floodplain position influenced ES distribution and dynamics. This approach was used rather than the classic stream order approach, because many of the 1st and 2nd order streams did not have floodplains, and I was interested in ES capacities of the floodplain specifically. 5.1.5 Land-cover classification  Building on the generalized land-cover maps from previous work (Tomlinson et al. 2011), I further classified land cover into five main categories: orchard, field, urban, forest, and water. To do so, I used details of agricultural cover types such as regularly spaced trees, dark green color, and coarse textures which helped distinguish orchards from yellow/green, smooth textured fields at the 1:4,000 scale. I further distinguished forest into vegetation types (mostly conifer, mixed conifer and broadleaf, wet shrub, scrub, rock-snow) and density classes (high, moderate, and low) (see Chapter 3). Land-cover types were used as proxies for a variety of ES capacity. I used methods described in Chapter 3 to map ES capacity, but focused only on contemporary ES capacity to explore spatial patterns.  93  5.1.6 ES quantiles For some analyses, I separated ES capacity values into quantiles to describe the importance of each reach for a particular ES to account for differences in the data distributions among ES. Reaches where a particular ES was not present, I assigned a value of 0. For reaches where an ES was present, the lowest 20% were assigned value of 1, while reaches in the 20-40% were assigned a value of 2, etc. Finally, reaches in the highest percentile were given a value of 5. I mapped ES capacities across the Wenatchee system floodplain to visualize their longitudinal variability 5.2 Analyses Weighted mean river position for each ES was determined, by multiplying ES quantile values (see above), by river reach ranking. To determine if the occurrence of ES differed with longitudinal position, I used a Friedman’s test (due to non-normality of ES, and unequal variances). From there, I used a pairwise Wilcoxon rank-sum test with Bonferroni corrections to determine which specific pairs of services differed with river-floodplain position. To describe spatial variability in ES capacity diversity, I quantified Simpson’s Index of diversity for each reach (based on unscaled ES values) to account for ES capacity richness and evenness. I then used Spearman’s rank correlation to determine if ES capacity diversity was associated with longitudinal position. I assessed how individual services (using quantiles) and ES diversity (unscaled ES values) varied with reach type using a Kruskal-Wallis test (due to non-normality of ES, and unequal variances) and pairwise Wilcoxon rank-sum test with Bonferroni corrections to determine which pairs differ.  94  5.3 Results 5.3.1 Individual ES vary with longitudinal position Locations important for ES capacity differed with ES type (Figure 5.2). I found median floodplain position differed among services (Friedman’s X2 = 1264.514, df = 4, p < 0.01). Pair-wise comparisons showed 9 of the 10 pairs occurred in significantly different locations (p < 0.05) (Table 5.1). Carbon storage and fish capacity were both distributed throughout the floodplain and did not differ significantly in longitudinal position, but all other ES differed from each other (Figure 5.2, Table 5.1). Carbon storage and fish capacity were concentrated in the upper reaches, while the lower-middle reaches are important for high quality paddling routes (Figure 5.2). Longitudinal position of orchard and forage production differed from every other ES. Carbon storage also differed from paddle routes. Paddle routes also differed in longitudinal position from locations important for ponds/wetlands and locations important for large wood.  Table 5.1 Shows weighted mean landscape position for each ecosystem service in 2006 in the first column. Orchards have the lowest median position ponds and wetlands have the highest weighted mean. Shows if the mean longitudinal position along the floodplain is different among pairs of services (Significance of pairwise Wilcoxon rank-sum test). ES that are significantly different are shown in bold.   Differences in mean longitudinal position Weighted mean longitudinal position Ecosystem Service Orchard production Forage production Carbon storage Paddle routes 12.3 Orchard Production     43.8 Forage production 0.002    97.9 Carbon storage < 0.01 < 0.01    61.9 Paddle routes < 0.01 < 0.01 < 0.01  86.7 Fish Capacity < 0.01 < 0.01 1.00 <0.01  95    Figure 5.2 Longitudinal patterns in ES capacity across the Wenatchee system floodplain in 2006. My maps support my hypothesis that ES capacity varies along the river-continuum. 96   5.3.2 Individual ES vary with reach type Kruskal-Wallis rank-sum tests showed significant differences in ES capacity by reach type for all ES (Table 5.2). Post-hoc pairwise Wilcoxon rank-sum tests showed that for orchard production, island braided reaches were significantly higher in orchard production capacity than plane bed (p = 0.01) and straight reaches (p = 0.03). For forage production, plane-bed reaches were significantly lower in capacity than island-braided (p < 0.01), meandering (p = 0.01), and straight reaches (p = 0.01). For carbon storage, meandering reaches were lower in capacity than pool/riffle reaches (p = 0.03). While my Kruskal-Wallis test suggested differences among paddle route quality by reach, my pairwise Wilcoxon rank sum test was too conservative to detect significant differences. For fish capacity, island braided reaches are significantly higher than meandering (p = 0.02) and plane bed reaches (p = 0.01). Meandering reaches are significantly higher in fish capacity than plane bed (p < 0.01), pool riffle (p < 0.01), and straight reaches (p < 0.01). Overall, ES diversity was higher in unconfined reaches than in confined reaches (Figure 5.3).  97  Table 5.2 Shows differences in fluvial-geomorphic reach type by individual service (Kruskal-Wallis rank sum test followed by post-hoc, Pairwise Wilcoxon rank-sum test with Bonferroni corrections). I=Island braided, M= Meandering, PB = Plane bed, PR= Pool/Riffle. Significant differences are shown in bold.  Ecosystem Service Reach type   I M PB PR Orchard production (Kruskal-Wallis X2 = 19.43, df = 4, p < 0.01)   I 0.111     M 0.001 0.924    PB 1.000 1.000 --   PR 0.030 1.000 0.263 1.00 Forage production (Kruskal-Wallis X2 = 21.27, df = 4, p < 0.01)  I 1.000     M 3.4e-05 0.001    PB 0.613 1.000 --   PR 0.334 1.000 0.010 1.000 Carbon storage (Kruskal-Wallis X2 = 12.45, df = 4, p < 0.01)  I 1.000     M 1.000 1.000    PB 0.141 0.033 0.426   PR 0.727 0.155 1.00 1.000 Paddle route quality (Kruskal-Wallis X2 = 16.88, df = 4, p < 0.01)  I 1.000     M 0.323 0.383    PB 0.054 0.069 1.000   PR 0.332 0.231 1.000 0.125 Fish capacity (Kruskal-Wallis X2 = 77.91, df = 4, p < 0.01)  I 0.02     M 0.01 <0.01    PB 1.00 <0.01 1.00   PR 0.50 <0.01 0.06 1.00 98     Figure 5.3 ES diversity (Simpson’s index) varied significantly with reach type (Kruskal-Wallis X2= 55.40, df =4, p <0.01). Pairwise comparisons show island braided reaches and meandering reaches had higher diversity in ES capacity than all other reach types, but did not differ from each other. Straight reaches were more diverse than plane bed and pool/riffle reaches.      5.3.3 Diversity of ES capacity varies with longitudinal position and reach type Lower reaches of the floodplain had a higher diversity in ES capacity (Simpson’s Index) than upper reaches in the floodplain (Spearman’s rho = -0.704, p <0.01) with highest variability in 99  middle reaches (Figure 5.4). Furthermore, diversity in ES capacity varied with reach type (Kruskal Wallis X2 = 45.76, df = 4, p <0.01). Lower portions of the floodplains were most diverse, owing to the greater amount of agriculture. Variability in ES capacity appears to be highest in middle reaches. Multiple pair-wise, post-hoc comparisons (Wilcoxon rank-sum with Bonferroni corrections) suggest island braided reaches have a higher ES capacity diversity than plane bed (p < 0.01), pool/rifle (p < 0.01), and straight reaches (p = 0.03) (Table 5.3).  Meandering reaches also had significantly higher diversity than plane bed (p < 0.01), pool/rifle (p < 0.01), and straight reaches (p < 0.13).  Straight reaches had higher ES capacity diversity than plane bed reaches (p = 0.015) (Figure 5.4).  Figure 5.4 Diversity in ES capacity (Simpson’s Index) according to river position. Diversity of ES capacity and distance to headwaters are negatively correlated (Spearman’s rho= -0.704, p <0.01) with distance from headwaters.  100   Table 5.3 Shows p-values for multiple pair-wise comparisons (Wilcoxon rank-sum test with Bonferroni corrections) for differences in diversity (Simpson’s Index), as well as the median diversity by reach type in the first column. Significant differences are highlighted. Mean diversity of ES capacity  (Simpson’s Index) Reach type Island braided Meandering Straight Plane bed 0.418 Island braided     0.427 Meandering 1.00    0.300 Straight < 0.01 < 0.01   0.186 Plane bed < 0.01 <0.01 1.00  0.087 Pool/riffle 0.028 < 0.01 0.015 0.065  5.4 Discussion I examined how river-floodplain concepts contributed to ES capacity and found compelling patterns. Drawing on classic river-floodplain approaches, the River Continuum Concept and fluvial-geomorphic forms revealed spatial patterns of floodplain-specific ES. Many ES differed in longitudinal position along the Wenatchee system floodplain with the exception of several fish habitat attributes and carbon storage. Furthermore, ES occurred in differing fluvial-geomorphic reach types. Unconfined reaches (island braided, meandering, and straight) had the highest ES capacity. However, even among unconfined reaches, diversity of ES capacity differed, with meandering and island braided reaches providing a greater diversity of ES than straight reaches. While certainly limited in its predictive capacity, the River Continuum Concept may shed light on longitudinal spatial characteristics of floodplain-specific ES. Understanding how ES capacity differs with fluvial-geomorphological and river floodplain position has a range of implications. I highlight implications for understanding the spatial distributions of ES and their interactions, as well as for blending process- and ES-based restoration approaches. 101  5.4.1 River-floodplain concepts can advance understanding of ES spatial distributions Species diversity has been found to increase with downstream position, and I found downstream positions also have higher level of ES diversity. Adding to efforts addressing spatial patterns of river-floodplains, my work suggests that the rich body of literature examining biological patterns in river-floodplains is applicable to maps of river-floodplain ES. Characterizing biological patterns of river-floodplains remains an active area of interest (Dodds et al. 2015). Since publication of Vannote’s seminal River Continuum Concept (Vannote et al. 1980), more than 326 papers have characterized patterns of species diversity in river channels (Tornwall et al. 2015). For example, hydrologic connectivity to the main channel is a key predictor of species richness of invertebrates (Starr et al. 2014). Fish communities in lower reaches were more diverse across tropical and temperate rivers (Araújo et al. 2009; Ibañez et al. 2009). Plant diversity is higher immediately downstream of confluences (Osawa et al. 2010), and meandering reaches provide important habitat for rare and threatened species (Osawa et al. 2011). Plant species richness increases with downstream position in river networks (Kuglerová et al. 2015). My results reflect similar patterns with downstream positions having greater diversity of ES; however, most ES research has found that hotspots of ES and biodiversity are not concordant (Naidoo et al. 2008; Egoh et al. 2009), this work suggests that hotspots of biodiversity and ES diversity may occupy similar landscape positions. Landscape baselines might help clarify the relationship between ES diversity and biodiversity as modification to produce ES may drive biodiversity declines. Also driven and formed by fluvial-ecological processes, the physical structures of river floodplains mean certain locations are inherently more likely to produce specific ES. Fertile soil 102  in wide floodplain valley bottoms meant such areas are critical for agricultural production. In contrast, narrower floodplains in upper parts of the river-floodplain system often have less frequent, but more severe flood disturbances likely making such areas crucial for growth of large trees and their subsequent contribution to in-stream habitat formation (Junk et al. 1989). Hydromorphic and geomorphic characteristics influence fish spawning patterns (Moir et al. 2004). Such fundamental differences in the capacity for different river-floodplain characteristics to provide ES should be further considered in ES mapping studies Spatial perspectives on ES are in need of more nuanced approaches. My work shows promise in integrating such hydro-ecological concepts into ES capacity spatial assessments. Exploring the relationship of ES within the context of other river-floodplain concepts including the Flood Pulse Concept, the Serial Discontinuity Concept, and the more recent Stream Biome Gradient Concept is also warranted (Junk et al. 1989; Ward and Stanford 1995; Benda et al. 2004; Dodds et al. 2015). Such concepts may further advance spatial predictions of ES distributions and mechanistic relationships among ES. The Flood Pulse Concept suggests variations in timing, magnitude, and predictability of flooding along a river floodplain drives interactions among biota (Junk et al. 1989). Biota and their habitat patterns may also correlate to patterns of ES capacity. The Serial Discontinuity Concept creates space for encompassing human structures such as dams into theoretical frameworks, crucial components of ES accessibility and capacity as impoundments represent pervasive barriers to longitudinal resources along river floodplains (Ward and Stanford 1995). The Stream Biome Gradient Concept may be amenable to exploring broad-scale patterns of ES capacity. This concept suggests that biota vary along precipitation and climatic gradients due to their impacts on hydrology and geomorphology (Dodds et al. 2015). 103  The depth and breadth of river-floodplain concepts means there is a rich body of work to draw on for exploring patterns of river-floodplain ES.  Spatial patterns of  ES have been used to indicate interactions (Raudsepp-Hearne et al. 2010a). Trade-offs occur when one ES is enhanced at the expense of another, while synergies occur when two ES increase simultaneously. Trade-offs and synergies have been shown to vary across space (Qiu and Turner 2013), yet causes of spatial variability in ES interactions are poorly understood. The underlying ecological characteristics contributing to the ES capacity in a given a reach also vary with spatial distribution. These underlying patterns might also contribute to variability in ES interactions in river-floodplains. Identifying how the spatial locations of ES differ and where the ranges of these ES overlap can distinguish tension zones among pairs of ES. Such zones of overlap may be susceptible to trade-offs when two ES cannot co-occur or represent opportunities for synergies for ES depending on similar ecological characteristics.  5.5 Implications for river-floodplain restoration Because river-floodplains are critical for ES and often degraded by agriculture and urbanization, restoration of river-floodplains is a priority for maintaining human well-being. However, best practices, approaches, and objectives for river-floodplain restoration are still evolving (Palmer et al. 2014a). Two approaches have been espoused recently: an ES-based approach and a process-based approach. A process-based approach seeks to halt the root cause of ecosystem degradation, often an anthropogenic disturbance (Beechie et al. 2010), while an ES-based approach seeks to improve human livelihood and well-being (Trabucchi et al. 2012). Recently, ES-based targets for restoration are being adopted widely by governments and agencies, often for social or legal reasons (Palmer and Filoso 2009; Palmer et al. 2014a). The push to adopt an ES-based approach 104  for restoration warrants consideration of how process-based approaches rooted and developed in river-floodplain ecology and ES-based targets are complementary. To date, integration of ES approaches for restoration with decades of work on processes-based restoration for maintaining ecosystems are rare (Palmer and Filoso 2009). By documenting the influence of hydro-geomorphic processes on spatial patterns of ES, my results suggest process-based approaches remain crucial for conserving ES and the ecological systems that provide them. 5.6 Linking ecosystem service and process-based approaches Foundational approaches for restoration have been developed, applied, and refined in river-floodplain systems, in particular, a process-based approach for restoration. Process-based principles for restoration suggest that in order to address the source of ecosystem degradation, the normative rates of ecosystem processes should be re-established (Beechie et al. 2010). This approach improves traditional restoration methods which restore ecosystem structure, anticipating ecosystem processes will follow (Palmer and Filoso 2009). A key benefit of processes-based restoration is that it allows the ecosystem to respond to natural disturbances and drivers by system-alteration of biological and physical variability (Beechie et al. 2010). This contrasts approaches engineering static habitats, and thus attempting to constrain biophysical processes (Palmer et al. 2014a). An ES-based approach appeals to restoration ecologists because of its potential for stakeholder support as well as its potential to aid financing of restoration projects through ES markets (Palmer and Filoso 2009). However, the linkages between process-based approaches and their success in enhancing ES is poorly understood, in part because techniques for restoring processes as well as ES aren’t well established (Palmer et al. 2014b). Standards for measuring ES are still 105  developing (Bear 2014), and the state of restoration science is such that ecosystem restoration is unlikely to regain lost ES (Palmer and Filoso 2009). While restoration can increase certain ES, ES levels in restored systems are often below reference system levels (Benayas et al. 2009).  5.7 How spatial patterns of ES can inform process-based restoration Establishing restoration targets is a fundamental challenge in restoration ecology, which must include a suite of ecological and social factors (Palmer et al. 2005). At the nexus of social-ecological systems, an ES approach has gained traction as a way to identify targets for restoration (Bullock et al. 2011). For ES, their inherent spatial patterns may limit success of their restoration, especially when targets for ES are outside the capacity for a certain site. I show ES typically vary with river-floodplain position. Understanding the spatial patterns of ES may help identify the suite of ES typically produced by certain floodplain positions, enabling restoration ecologists to identify context-specific targets for restoration. Identifying locations that fall within the typical spatial range of ES, but currently have low levels of ES, might ensure greater ES restoration success than selecting sites that fall outside the typical spatial range of ES for restoration. For example, a hierarchical maps were used to prioritize restoration by identifying the spatial distribution of ES in a river in Spain (Trabucchi et al. 2013), such approaches may support process-based approaches which require watershed-level coordination to reduce background levels of sedimentation (Beechie et al. 2010). Understanding where ES typically appear can help us identify areas with uncharacteristically low ES. Using process-based principles to restore the full suite of ES produced in certain river-floodplain positions, may enhance multiple ES (Beechie et al. 2010). Landscape structures and ecosystems are formed by biophysical processes, some of which are associated with higher amounts of ES 106  than others. Restoring the biophysical processes which result in particular landscape structures and ecosystems can result in higher levels of ES. For example, restoring recruitment of large wood may also enhance perceived riverine aesthetic quality, as aesthetic preferences corresponded more than expected with riverine eco-morphological quality (Junker and Buchecker 2008). Furthermore, restoring connectivity of rivers and their floodplains may also enhance ES diversity. Unconfined reach types had much higher ES capacity diversity than confined reach types, and by reconnecting artificially confined rivers to their floodplains diversity in ES capacity could increase.  5.8 Conclusion Spatial concepts in river-floodplain ecology can contribute to my understanding of the spatial distribution of floodplain specific ES. Ecological concepts, such as the River Continuum Concept and fluvial-geomorphic dynamics represent opportunities to explore the nuances of spatial variability among different ES. ES capacity varied longitudinally along the Wenatchee floodplain, which parallels proposals of longitudinal ecological variability from the River Continuum Concept.  Furthermore, ES capacity varied considerably with fluvial-geomorphic reach types. Such classic river-floodplain concepts may invigorate thinking on ES spatial dynamics. Future work exploring mechanistic relationships among flow and ES capacity can further improve my understanding of variability in ES capacity along the river-floodplain continuum. My work suggests a process-based approach will support restoration of a suite of ES typical of a specific river-floodplain position, as patterns of ES in the Wenatchee system floodplain reflect patterns of ecosystem and fluvial-geomorphic processes.  107  Chapter 6: Conclusions and future research directions The overall objective of this dissertation was to characterize historical and contemporary ecosystem service (ES) dynamics using a floodplain landscape to exemplify history’s relevance to ES interactions. Using several novel datasets, including land survey data and aerial photography, I explored how ES and frontier settlement varied with landscape position and mapped changes in ES over broad timeframes. Several major spatial and temporal patterns emerged from the four data chapters in my dissertation. I conclude by describing these key patterns and the limitations of my approach. I also suggest directions for future research which include clarifying individual response to drivers as well as ES interactions. I include a detailed example of this future work in the broader region of the Wenatchee System Floodplain, the Columbia River Basin, focusing on the importance of understanding ES dynamics and interactions under warming climatic conditions. Chapter 2 quantified the relative importance of landscape position for settlement and found floodplains and riparian zones were disproportionately important in a frontier context. This novel use of General Land Office (GLO) surveys showed different disturbance types co-occurred, which suggests different parts of the floodplain were used for different purposes. My results have implications for using GLO datasets as reference conditions for floodplain restoration, as patterns seen on historical landscapes should be considered the result of interacting anthropogenic and ecosystem processes. This chapter highlights the long history of people modifying floodplains and supports the exploration of floodplain-specific ES.  108  Chapter 3 highlighted land cover and ES change within the floodplain specifically, during a time of urban expansion. This chapter emphasizes the importance of distinguishing ES flows (i.e., ES that actually reach people) from ES capacity (i.e., ES that could reach people). Furthermore, it describes the novel use of aerial photography to map changes in ES capacity. Landscape features seen in high-resolution aerial photography underpin my analysis of ES change. In my study area, increasing populations have resulted in urban expansion into fertile floodplain soils, and legacies of past ES preferences were seen in densifying floodplain forests. Chapter 4 demonstrated how baselines can inform characterizations of ES trade-offs and synergies. Trade-offs occur when one ES increases at the expense of another while synergies occur when ES increase concomitantly. Space-for-time and change-over-time approaches resulted in vastly different characterizations of ES interactions. For trade-offs, the two approaches always differed. Missed synergies and tradeoffs may lead to overlooked opportunities for win-win restoration or unexpected declines in ES. Long-term monitoring is key for capturing interactions among ES, and such monitoring programs should be established and maintained as governments adopt an ES approach to management and conservation. Chapter 5 explored how spatial characterizations of ES in floodplains can benefit from incorporating riverine and floodplain principles. River-floodplain ES varied considerably with longitudinal position as with fluvial geomorphic reach type. Such patterns can be used to advance understanding of river-floodplain ES. Furthermore, the association of process-driven patterns and ES distribution suggest processes-based restoration may complement ES-based conservation. 109  6.1 Caveats and cautions: Limitations of historical datasets and aerial photography This dissertation represents a rare spatially-explicit reconstruction of ES over broad timeframes; however, using historical data for these purposes is not without limitations. I describe some of the limitations of using these historical datasets for reconstructing ES and settlement patterns next.  6.1.1 Coarse-resolution GLO datasets ideal for describing general settlement patterns The strength of GLO datasets lies in its consistency and wide-spread availability for reconstructing historical settlement (Rhemtulla and Mladenoff 2010). The limitations of GLO data for reconstructing vegetation information are well-explored (Schulte and Mladenoff 2001; Rhemtulla and Mladenoff 2010); however, there is no work that quantitatively explores the limitations of using GLO data for reconstructing historical settlement patterns. Many of the limitations are similar to those for reconstructing vegetation, including variability in survey quality and coarse resolution of the dataset (Schulte and Mladenoff 2001). Such coarse resolution data is best used to describe general patterns of settlement, as done in this dissertation. Future work should explore supplementary datasets such as historical plat maps and re-surveys to determine how accurate GLO surveys are for reconstructing historical settlement patterns.  6.1.2 Limited information on linkages between landscape features and ES  Aerial photography provides a high-resolution dataset for exploring changes in ES over broad timeframes, and here, I used a high-resolution imagery and associated visible landscape attributes with various ES. Established relationships among landscape features and ecosystem processes and ES were used to create indices of ES importance (Large and Gilvear 2015). This novel approach, while currently the best available for reconstructing multiple river-floodplain ES 110  which require fine-scale information, is not without limitations. Landscape features are assumed to have an inherent set of properties suitable for providing ES, and subsequent ES analyses may be sensitive to such assumptions (Large and Gilvear 2015). For recreational capacity, large wood is widely considered a major hazard to recreational river-users (Biedenweg et al. 2012), however, recreational users have been shown to preferentially float in locations with large amounts of large wood, and perspectives of large wood in streams is diverse. Because of these inconclusive and contrary conclusions,  we did not consider large wood as a negative co-efficient in our model (Biedenweg et al. 2012). An understanding of ES distributions and interactions may be improved through clarifying linkages between landscape feature and ES relationships.  Similarly, the set of landscape characteristics used here to map an ES may not be comprehensive in explaining ES distributions. Unknown relationships among landscape features and ES including attributes that cannot be detected with aerial photography (thermal properties, chemical properties, etc.) might also be important. For example, our maps of above-ground carbon storage do include important below-ground carbon pools, because independent soil carbon samples were unavailable for both timeframes. Soil carbon likely played an important role in the ES dynamics of the region, but such dynamics remain uncaptured with above-ground carbon storage methods. For salmonids, water chemistry characteristics may also play a key role in determining habitat quality, however, spatially explicit data from both time frames was also unavailable. For example, the suitability of fish habitat is also influenced by the ephemerality of floodplain habitats (Jeffres et al. 2008); however flooding dynamics were not incorporated. Future work could elucidate this relationship further, and identify additional relationships warranting 111  consideration in ES mapping. Better causal linkages between landscape characteristics and ES would improve my approach and accuracy of ES maps.  6.2 Future research directions and applications: ES dynamics and interactions under future drivers of change The data of my dissertation is limited to historical and contemporary scope, but future research should explore potential ES response to different drivers as well as ES interactions. Drawing on historical information to understand long-term dynamics will be essential for this. Characterizing landscape change, ES interactions, and ES spatial distributions is particularly germane as ES continue to be dynamic in response to drivers of change such as future climatic conditions. To conclude, I place my work within the broader context of the region and explore the implications of an important driver of regional change: a warming climate. I use this to exemplify how a driver might influence ES and their interactions. I briefly summarize anticipated effects of climate change on stream flows in the Columbia basin, focusing on expected shifts in the timing and magnitudes of stream flows. Synthesizing literature from the CRB as well as other regions, I discuss the importance and foundations for future work examining how hydrologic changes might influence a variety of ES provided by rivers and their floodplains (1) trade-offs among ES; (2) complex cascades of processes that may lead to unanticipated or compounded effects on ES; (3) feedbacks among processes and ES; and (4) potential landscape-scale shifts in service locations. My dissertation provides important context for approaching issues related to avoiding future ES interactions. Residents of the intensively managed Columbia River Basin (CRB) depend upon its river-floodplain systems for food production (from orchards and salmon fisheries), hydropower, flood 112  attenuation, recreation, and ES contributing to a thriving economy and culture. Many ES in the CRB have undergone massive change as a result of historical decisions emphasizing large dam construction for hydropower, irrigation, and flood control (e.g., under the Columbia River Treaty, the Columbia Basin Project, and the Bonneville Power Act). Improvement of these ES was achieved at the expense of other ES such as use of floodplains for agriculture or conservation of productive salmon habitats. For example, the Columbia River Treaty, ratified by the US and Canada in 1964, led to the construction of three new dams in Canada for flood protection and hydropower production which resulted in flooding of 60,000 ha of productive floodplains in British Columbia, relocation of over 2,000 people (Loo 2004) and destruction of fish migratory routes (Huizen 2010). The hydrology of the CRB will likely undergo substantial changes under a warming climate, and future research should address ES interactions in this context. Climate models indicate warming temperatures will reduce snowpack, decrease summer water availability, and increase flood risk in the CRB over coming decades (Arnell 1999; Lehner et al. 2006; Adam et al. 2009; Tohver and Hamlet 2010). It is appreciated that these climate change effects will impact hydropower, agriculture, and irrigation throughout the region (Hamlet et al. 2010; Stockle et al. 2010; Vano et al. 2010). However, less is understood about potential impacts to other river and floodplain ES such as flood attenuation or creations of fish habitat that historically supported abundant salmon runs. Furthermore, a solid understanding of potential interactions among some ES is limited by a lack of understanding of long-term physical and ecological processes which might be influenced by a warming climate.  113  6.3 Climate change and anticipated shifts in hydrologic regimes Changes in future climate are expected to influence hydrologic regimes of the CRB’s rivers, affecting both flood risk and the seasonality of peak flows (Adam et al. 2009). Three characteristic hydrologic regimes are distinguished by the timing and form of precipitation, which is largely driven by temperature. These include: rain-dominant, transitional, and snowmelt-dominant hydrologic regimes (Figure 6.1) (Elsner et al. 2010; Mote and Salathe 2010; Tohver and Hamlet 2010). In rain-dominant basins, the majority of precipitation falls as winter rain with corresponding peaks flows in winter. For snow-dominant regimes, precipitation falls primarily as snow in the winter and subsequent melting of snowpack results in spring and summer peak flows. Intermediate between these dynamics are transitional hydrologic regimes. Under such regimes, peak flows occur in both winter and summer due to a mix of winter precipitation falling as both rain and snow (Tohver and Hamlet 2010). Such hydrologic regime shifts may influence ES interactions and should be considered in future ES research.  114   Figure 6.1 .Examples of snowmelt, transition, and rain-dominant hydrologic regimes based on historical monthly mean flows (m3s-1) for three rivers in the Columbia River Basin from 1929-2011. Future research should also target watersheds in the CRB at greatest risk of undergoing hydrologic regime shifts. Climate models predict declines in winter snowpack, and such declines have already been documented in recent decades (Mote et al. 2005; Adam et al. 2009; Mote and Salathe 2010; Tohver and Hamlet 2010). Reductions in snowpack will shift many snow-dominant watersheds to transitional regimes, and many transitional watersheds to rain-dominant regimes (Figure 6.2) (Tohver and Hamlet 2010), with corresponding reductions in summer peak flows and increased winter flows (Elsner et al. 2010; Tohver and Hamlet 2010). While flood risk is generally expected to increase in the CRB, particularly late in the 21st century, changes in the 115  timing of floods will also be important (Tohver and Hamlet 2010). Responses may be particularly complex in transitional basins. For example, flood risk may decline in cooler transitional watersheds (watersheds experiencing mean winter temperatures between -6 and -3°C). However, in some warmer transitional watersheds (mean winter temperatures > -3°C), flood risk may increase as receding snowlines increase the proportion of drainage basins receiving rainfall rather than snow (Tohver and Hamlet 2010). Watersheds near the boundaries of regions with large amounts of snowmelt-driven runoff are at greatest risk of reductions in snowmelt flows (Arnell 1999; Adam et al. 2009), as well as greatest risk of corresponding reductions of summer low flows and increases in winter flood flows (Merritt et al. 2006; Elsner et al. 2010; Tohver and Hamlet 2010). Lastly, interactions with land use change may also be significant as channel straightening, agriculture, and urbanization could reduce natural flood attenuation in some regions (Changnon and Demissie 1996; Bronstert et al. 2002), compounding the effects of increased flood magnitudes due to climate change.  116   Figure 6.2 The Columbia River Basin (outlined in black) drains approximately 673,400 km2 and crosses the American and Canadian borders. It is likely to experience hydrologic regime shifts and trade-offs among floodplain ES under future climate. Colors indicate predicted shifts in spatial distribution of different hydrologic regimes in rivers throughout the Columbia River basin from 1976-1999 to 2070-2099 (Beechie et al. 2012). Future work should consider shifting hydrologic regimes in the Columbia River Basin, and its impact to ES.  117  6.4 Impacts of climate change on river and floodplain ES interactions Changes in hydrologic regimes and water availability are expected to have profound effects on ES provided by rivers and their floodplains, including municipal water supplies, irrigation, flood attenuation and health of aquatic ecosystems, and a warming climate will likely resulted in ES interactions. Future work should identify and characterize additional pathways for ES interactions characterized to avoid problematic declines in ES. In addition to examining trade-offs under future climate, future research should characterize and measure more complex interactions, such as cascading effects and feedbacks,  as well as the underlying process resulting in ES declines. Cascading effects are those that occur through complex pathways. Feedbacks occur when a change in one process (or service) affects another process which then reinforces the direction of change in the original process. Cascading effects and feedbacks which occur through multi-step pathways, or over broad temporal scales, may greatly confound efforts to clearly detect trade-offs. Inadequate identification of trade-offs may inhibit informed consideration of future planning for multiple ES.  6.5 Trade-offs among ES Future work should measure and model the impacts of hydrologic shifts on trade-offs, especially in water-limited regions such as the CRB. Trade-offs are inherent in interactions among water-related ES (Brauman et al. 2007) and historically, trade-offs among competing water uses and floodplain ES were routine in arid portions of the CRB. Furthermore, the CRB has been highlighted as a region with low storage capacity (Barnett et al. 2005), which means many water-related ES are heavily dependent on timing of flows. While summer water availability is currently relatively high due to snowpack-derived flows, future reductions in snowpack will 118  likely exacerbate existing trade-offs among water for irrigated agriculture, hydropower, and stream flow levels necessary for salmonids. Diminishing snowpack may lead to 2.0-3.4% decreased production of hydroelectricity annually, the primary and least expensive form of energy in the region (Hamlet et al. 2010). Decreases in summer water flows (coupled with increasing crop water demand due to hotter temperatures) may lead to declines in water availability for irrigated agriculture (Neilsen et al. 2006; Vano et al. 2010). This may also exacerbate tensions between irrigators and other stakeholders needing increasing water for in-stream uses, such as hydropower production or improving juvenile salmon survival on their migration to sea. Because water demands are expected to increase for all users, trade-offs will likely intensify under low flows as in-stream flows used for one purpose cannot be used for other purposes in the same location (Figure 6.3) (Rodríguez et al. 2006). Future work should focus on better understanding the potential magnitude and locations of such tradeoffs in the context of historical ES and drives of ES change.  119   Figure 6.3 Trade-off between irrigated agriculture and hydropower under likely future climate scenarios (Elsner et al. 2010). Because water availability is expected to decline, maintenance of current levels of one service (e.g., hydropower) may lead to declines in the other. 6.6 Cascading effects Declines in river and floodplain ES may occur through indirect and potentially unexpected pathways, particularly when ES require multiple fluvial and ecological processes to support their production (MA 2005). Thus, identifying the various pathways through which diminishing snowpack affects ES and processes is essential for identifying potentially problematic interactions; however these pathways are poorly understood. Future research should explore cascading and indirect interactions among ES. For example, diminishing snowpack may lead to declines in water-based recreation via direct and indirect pathways (Figure 6.4). A direct impact is a likely decrease in the reliability of high water levels necessary for recreational boating in the 120  Grand Coulee reservoir (Hamlet and Lettenmaier 1999), as well as changes in the suitability of river flows for paddlers in Alberta, Canada (Rood et al. 2006). Potential indirect cascading effects of diminishing snowpack may involve water quality declines due to increasing stream temperature and pollutant concentrations. Higher water temperatures can cause declines in fish populations (Mantua et al. 2010), with cascading impacts on riverine food webs, facilitating harmful algal growth (Power 1990). Lower stream flows may also result in insufficient dilution of pollutants, particularly during higher summer loadings of pesticides and fertilizers used in nearby watersheds (Whitehead et al. 2009). Leaching of nitrogen may also affect in-stream water quality. Loss of watershed nitrogen through erosion (driven by shifting hydrologic regimes and land use conversion) is an important pathway of nitrogen loss from terrestrial systems to rivers (Hatfield and Follett 2008). A decline in water quality (by any of these mechanisms) could potentially negatively impact salmon production thus reducing recreational fishing opportunities. Moreover, declines in water quality may also have economic impacts which include lower property values (Carson et al. 1992; Leggett and Bockstael 2000). These hypothesized linkages should be tested and quantified. 121   Figure 6.4 Complex cascade of interactions among ecosystem ES that may occur as a result of diminishing snowpack. Declines in summer water availability may lead to reductions in water levels for recreation directly (Hamlet and Lettenmaier 1999). Indirectly, declines in erosion control may cause soil nitrogen loss through erosion (Hatfield and Follett 2008), reducing in-stream water quality, and resulting in declines in water-based recreation. Future work should explore these potential pathways of decline.  6.7 Reinforcing feedbacks Self-reinforcing feedbacks among ES and processes can potentially lead to unexpected declines in some ES (Bennett et al. 2009), but such interactions are poorly understood. Future research exploring feedbacks is particularily germane as such interactions may lead to unexpected ES 122  declines. Interactions among salmon, riparian nitrogen cycling, forest productivity, coarse woody debris, and salmon habitat creation involve a complex cascade of interaction, potentially creating negative reinforcing feedbacks (Figure 6.5). Salmon carcasses provide important marine-derived nutrients which contribute to floodplain forest productivity. Trees growing along salmon-bearing streams in the Pacific Northwest have triple the rate of growth (mean annual basal area increment) than trees along streams without salmon (Helfield and Naiman 2001; Naiman et al. 2010). In some cases, more of the nitrogen in early successional vegetation is attributable to salmon carcasses than to N-fixation by Alnus (Morris and Stanford 2011). Many rivers in the CRB are (or were) salmon-bearing; thus, declines in salmon populations (due to shifting hydrologic regimes and many other factors) will likely decrease floodplain forest productivity (Schindler et al. 2003). Declining salmon populations which reduce forest productivity could potentially reduce inputs of coarse woody debris to streams, a critical component for the formation of salmon habitat (Helfield and Naiman 2001).  Recognizing feedbacks and cascading interactions may be challenging as climate change effects and ecological responses may occur over a variety of timeframes (Carpenter et al. 2009; Raudsepp-Hearne et al. 2010a). Future work should explore these time lags, incorporating historical info into characterizations of complex ES feedbacks and interactions. While shifts in the timing of peak flows have already occurred and may become more apparent in the mid-21st century, flood risk will not likely increase until the late 21st century (Mote et al. 2005; Tohver and Hamlet 2010). Moreover, subsequent ecosystem responses to these hydrologic changes may occur over different temporal scales (Brauman et al. 2007). For example, annually varying salmon populations impact nitrogen budgets for riparian vegetation over unknown timeframes 123  (Helfield and Naiman 2001), and any reductions of tree productivity and growth or coarse woody debris driven by loss of salmon-derived nitrogen may not become apparent for decades or centuries (Hyatt and Naiman 2001). Hence, because time lags in the response of processes and ES may lead to delayed feedbacks or cascades of change, today’s management decisions must learn how to avoid negative impacts to ES over the long term.   Figure 6.5 Complex cascading interactions among ES could cause indirect feedback loops. Though multiple pathways (declines in forest productivity and formation of salmon habitat), an indirect feedback loop may further cause declines in salmon, riparian forest productivity, coarse woody debris, and riparian nitrogen availability (Helfield and Naiman 2001). Future work should explore and quantify such pathways to feedback loops drawing on historical data to quantify time lags.  124  6.8 Geographical shifts in ES Another important direction for future work is exploring geographical shifts in ES, especially in relation to climate-driven shifts in hydrologic regimes. Following shifts in the spatial distribution of hydrologic regimes, the geographic range of some ES may also shift. For example, loss of snowmelt peaks may lead to declines in summer water-based recreational activities (such as boating) in the mid-elevation parts of the CRB (Hamlet and Lettenmaier 1999; Vano et al. 2010). Crops with large irrigation needs (such as apples and cherries) may become unprofitable in regions losing snowmelt peak flows, leading to northward shifts in regional crops and associated agri-tourism. In addition, production of hydropower during the summer months may be adequate only in transitional and snowmelt basins. Salmon (Onchorhynchus spp.) are at considerable risk of range reductions in light of expected regional warming by several mechanisms: increased summer stream temperatures, increased intensity and frequency of winter floods, and reduction of springtime snowmelt (Mantua et al. 2010). There are substantial unknowns, however, in predicting the biological responses of such a wide-ranging and adaptable species. Most predictions are based on assumptions that salmon populations cannot adapt, do not stray from natal rivers, and that physiological tolerances to temperature do not vary among populations throughout the region. However, it is well known that none of these assumptions are true (Beechie et al. 2006; Reed et al. 2011). While specific predictions may be accurate for certain populations in certain parts of the region, a lack of sufficient data on localized environmental attributes and variability in biological adaptations throughout the region prevent extrapolations of finer-scale predictions to salmonid responses 125  throughout the region (Beechie et al. 2012). Future work should explore such dynamics and spatial patterns. Exploring the cumulative impacts of land and water management on ES is also an important direction for ES research. Land and water management often inadvertently creates worse habitat conditions than those predicted as a result of climate change. As such, further degradation (or restoration) of salmon habitat could play a major role in determining the future distribution of salmon populations. For example, some streams have long been over-allocated for water abstraction; thus stream flows are currently too low to support healthy salmon runs. In addition, opportunities to restore key habitat attributes in certain areas are generally not accounted for in developing future scenarios (Waples et al. 2009), nor are the potential spatial shifts in salmon populations. Thus, regional-scale shifts in land management, water management, and salmon populations, may be an important mosaic upon which ES interactions should be considered. 6.8.1 Time lags necessitate long-term management strategies  My long-term perspective sets the stage for future work exploring ES interactions under warming climate conditions. A long-term perspective is essential for understanding, managing, and restoring river and floodplain systems as many important habitat conditions are maintained by long-term processes. Furthermore, because ES may interact through multiple pathways over different timeframes, time lags may obscure detection of problematic interactions. Such time lags may also have the potential to delay the benefits accrued through restoration. Coupling long-term management perspectives with strategies that address the root causes of degradation ensures that restoration actions address the processes that have led to degradation, thereby leading to long-term recovery (Beechie et al. 2010). Future work should combine long-term approaches and 126  datasets with future monitoring to better understand time-lags involved in identifying ES interactions 6.9 In summary Understanding interactions among river and floodplain ES is critical to informed management decisions that must balance trade-offs among competing ES. My concluding review of future research directions highlights the importance of identifying interactions among ES, and suggests that in some cases complex cascading effects or feedback loops may complicate accurate understanding of effects of climate change on ES. Hence, scientists will play an important role in advancing my understanding of how climate change may alter ES over the long term, thereby informing management choices among competing ES. This will require a better mechanistic understanding of potential cascades and feedbacks that may influence ES today and in a climate-altered future, as well as concise explanations of trade-offs among those ES. In this summary of future work, I provide only a small sub-set of potential trade-offs, cascades, and feedbacks, under only one potential driver of change: a warming climate. It is important that scientists and managers broaden this list to include other interactions among ES as well as other drivers of ES change with the goal of incorporating such knowledge into long-term river management and restoration plans. Understanding ES interactions can help managers devise strategies to reduce landscape-scale losses and problematic interactions among ES under future climatic conditions. The long-term approaches explored in this dissertation lay the foundations and are critical for further understanding ES interactions.   127  Bibliography Abrams MD, Ruffner CM (1995) Physiographic analysis of witness-tree distribution 1765-1798 and present forest cover through north central Pennsylvania. Can J For Res 25:659–667. Adam JC, Hamlet AF, Lettenmaier DP (2009) Implications of global climate change for snowmelt hydrology in the twenty-first century. Hydrol Process 23:962–972. doi: 10.1002/hyp.7201 Amundson R, Berhe AA, Hopmans JW, et al (2015) Soil and human security in the 21st century. Science (80- ) 348:1261071–1261071. doi: 10.1126/science.1261071 Anderson BJ, Armsworth PR, Eigenbrod F, et al (2009) Spatial covariance between biodiversity and other ecosystem service priorities. J Appl Ecol 46:888–896. doi: 10.1111/j.1365-2664.2009.01666.x Anschuetz KF, Wilshusen RH, Scheick CL (2001) An Archaeology of Landscapes: Perspectives and Directions. J Archaeol Res 9:157–211. Araújo FG, Pinto BCT, Teixeira TP (2009) Longitudinal patterns of fish assemblages in a large tropical river in southeastern Brazil: Evaluating environmental influences and some concepts in river ecology. Hydrobiologia 618:89–107. doi: 10.1007/s10750-008-9551-5 Arnell NW (1999) The effect of climate change on hydrological regimes in Europe: a continental perspective. Glob Environ Chang Policy Dimens 9:5–23. doi: 10.1016/s0959-3780(98)00015-6 Arriaza M, Cañas-Ortega JF, Cañas-Madueño JA, Ruiz-Aviles P (2004) Assessing the visual quality of rural landscapes. Landsc Urban Plan 69:115–125. doi: 10.1016/j.landurbplan.2003.10.029 Bailly D, Agostinho AA, Suzuki HI (2008) Influence of the Flood Regime on the Reproduction of Fish Species with Different Reproductive Strategies in the Cuiaba River, Upper Pantanal, Brazil. River Res Appl 24:1218–1229. doi: 10.1002/rra Bajat B, Hengl T, Kilibarda M, Krunić N (2011) Mapping population change index in Southern Serbia (1961–2027) as a function of environmental factors. Comput Environ Urban Syst 35:35–44. doi: 10.1016/j.compenvurbsys.2010.09.005 Baker RG, Schwert DP, Bettis EA, Chumbley CA (1993) Impact of Euro-American settlement on a riparian landscape in northeast Iowa, midwestem USA: an integrated approach based on historical evidence, floodplain sediments, fossil pollen, plant macrofossils and insects. The Holocene 3:314–323. doi: 10.1177/095968369300300403 128  Barnett TP, Adam JC, Lettenmaier DP (2005) Potential impacts of a warming climate on water availability in snow-dominated regions. Nature 438:303–9. doi: 10.1038/nature04141 Baudry J, Thenail C (2004) Interaction between farming systems, riparian zones, and landscape patterns: A case study in western France. Landsc Urban Plan 67:121–129. doi: 10.1016/S0169-2046(03)00033-1 Bear D (2014) Integration of Ecosystem Services Valuation Analysis into National Environmental Policy Act Compliance Legal and Policy Perspectives.  Beechie T, Buhle E, Ruckelshaus M, et al (2006) Hydrologic regime and the conservation of salmon life history diversity. Biol Conserv 130:560–572. doi: 10.1016/j.biocon.2006.01.019 Beechie T, Imaki H, Greene J, et al (2012) Restoring salmon habitat for a changing climate. River Res Appl. doi: 10.1002/rra Beechie TJ, Sear D a., Olden JD, et al (2010) Process-based Principles for Restoring River Ecosystems. Bioscience 60:209–222. doi: 10.1525/bio.2010.60.3.7 Beechie TJ, Sibley TH (1997) Relationships between Channel Characteristics, Woody Debris, and Fish Habitat in Northwestern Washington Streams. Trans Am Fish Soc 126:217–229. Bellemare J, Motzkin G, Foster DR, Forest H (2002) Legacies of the agricultural past in the forested present: an assessment of historical land-use effects on rich mesic forests. J Biogeogr 29:1401–1420. Benayas JMR, Newton AC, Diaz A, Bullock JM (2009) Enhancement of Biodiversity and Ecosystem Services by Ecological Restoration: A Meta-Analysis. Scand J Manag 8:1121–1124. Benda L, Poff NL, Miller D, et al (2004) The Network Dynamics Hypothesis : How Channel Networks Structure Riverine Habitats. Bioscience 54:413–427. Benke AC, Wallace B (2003) Influence of Wood on Invertebrate Communities in Streams and Rivers Wood-Created Habitat. Am Fish Soc Symp 37:149–177. Bennett EM, Peterson GD, Gordon LJ (2009) Understanding relationships among multiple ecosystem services. Ecol Lett 12:1394–1404. doi: 10.1111/j.1461-0248.2009.01387.x Bergstrom JC, Dillman BL, Stoll JR (1985) Public Environmental Amenity Benefits of Private Land: The Case of Prime Agricultural Land. 139–150. Besnard AG, La Jeunesse I, Pays O, Secondi J (2013) Topographic wetness index predicts the occurrence of bird species in floodplains. Divers Distrib 19:955–963. doi: 10.1111/ddi.12047 129  Biedenweg K, Akyuz K, Skeele R (2012) Balancing riparian management and river recreation: methods and applications for exploring floater behavior and their interaction with large wood. Environ Manage 50:283–95. doi: 10.1007/s00267-012-9876-8 Birkhofer K, Diehl E, Andersson J, et al (2015) Ecosystem services--current challenges and opportunities for ecological research. Front Ecol Evol 2:1–12. doi: 10.3389/fevo.2014.00087 Blanton P, Marcus WA (2009) Railroads, roads and lateral disconnection in the river landscapes of the continental United States. Geomorphology 112:212–227. doi: 10.1016/j.geomorph.2009.06.008 Blanton P, Marcus WA (2013) Transportation infrastructure, river confinement, and impacts on floodplain and channel habitat, Yakima and Chehalis rivers, Washington, U.S.A. Geomorphology 189:55–65. doi: 10.1016/j.geomorph.2013.01.016 Boone RB, Hobbs NT (2004) Lines around fragments: effects of fencing on large herbivores. African J Range Forage Sci 21:147–158. doi: 10.2989/10220110409485847 Bouldin J (2003) Location and Accuracy of a Sample of General Land Office Survey Corners in a Portion of Western Yosemite National Park: a Summary.  Brady E (2006) The Aesthetics of Agricultural Landscapes and the Relationship between Humans and Nature. Ethics, Place, Environ 9:1–19. doi: 10.1080/13668790500518024 Brauman KA, Daily GC, Duarte TK, Mooney HA (2007) The Nature and Value of Ecosystem Services: An Overview Highlighting Hydrologic Services. Annu Rev Environ Resour 32:67–98. doi: doi:10.1146/annurev.energy.32.031306.102758 Bronstert A, Niehoff D, Bürger G (2002) Effects of climate and land-use change on storm runoff generation: present knowledge and modelling capabilities. Hydrol Process 16:509–529. doi: 10.1002/hyp.326 Brown DG, Ecology SP, Brown G (1997) Mapping historical forest types in Baraga County Michigan, USA as fuzzy sets. Plant Ecol 134:97–111. Bull JW, Gordon A, Law EA, et al (2014) Importance of baseline specification in evaluating conservation interventions and achieving no net loss of biodiversity. Conserv Biol 28:799–809. doi: 10.1111/cobi.12243 Bullock JM, Aronson J, Newton AC, et al (2011) Restoration of ecosystem services and biodiversity: Conflicts and opportunities. Trends Ecol Evol 26:541–549. doi: 10.1016/j.tree.2011.06.011 Bürgi M, Hersperger AM, Schneeberger N (2004) Driving forces of landscape change – current 130  and new directions. Landsc Ecol 19:857–868. doi: 10.1007/s10980-004-0245-8 Bürgi M, Silbernagel J, Wu J, Kienast F (2015) Linking ecosystem services with landscape history. Landsc Ecol 30:11–20. doi: 10.1007/s10980-014-0102-3 Burkhard B, Kandziora M, Hou Y, Müller F (2014) Ecosystem service potentials, flows and demands-concepts for spatial localisation, indication and quantification. Landsc Online 34:1–32. doi: 10.3097/LO.201434 Burkhard B, Kroll F, Müller F (2009) Landscapes‘ Capacities to Provide Ecosystem Services – a Concept for Land-Cover Based Assessments. Landsc Online 1–22. doi: 10.3097/LO.200915 Burkhard B, Kroll F, Nedkov S, Müller F (2012) Mapping ecosystem service supply, demand and budgets. Ecol Indic 21:17–29. doi: 10.1016/j.ecolind.2011.06.019 Carpenter SR, Mooney HA, Agard J, et al (2009) Science for managing ecosystem services: Beyond the Millennium Ecosystem Assessment. Proc Natl Acad Sci U S A 106:1305–1312. doi: 10.1073/pnas.0808772106 Carson RT, Mitchell RC, Hanemann WM, et al (1992) A Contingent Valuation Study of Lost Passive Use Values Resulting From the Exxon Valdez Oil Spill.  Chalfant SA, Ray VF (1974) Ethnohistorical reports on aboriginal land use and occupancy: Spokan Indians, Palus Indians, Columbia Salish, Wenatchi Salish. Garland Pub, Inc. Chan KMA, Shaw MR, Cameron DR, et al (2006) Conservation planning for ecosystem services. PLoS Biol 4:e379. doi: 10.1371/journal.pbio.0040379 Changnon SA, Demissie M (1996) Detection of changes in streamflow and floods resulting from climate fluctuations and land use-drainage changes. Clim Change 32:411–421. doi: 10.1007/bf00140354 Chen J (2007) Rapid urbanization in China: A real challenge to soil protection and food security. Catena 69:1–15. doi: 10.1016/j.catena.2006.04.019 Chen K (2002) An approach to linking remotely sensed data and areal census data. Int J Remote Sens 23:37–48. doi: 10.1080/01431160010014297 Commission E (2011) Our life insurance, our natural capital:an EU biodiversity strategy to 2020. Brussels Connor JEO, Jones MA, Haluska TL (2003) Flood plain and channel dynamics of the Quinault and Queets Rivers, Washington, USA. Geomorphology 51:31–59. Crawford GW, Smith DG, Desloges JR, Davis AM (1998) Floodplains and Agricultural Origins: A Case Study in South-Central Ontario, Canada. J F Archaeol 25:123–137. doi: 131  10.1179/009346998792005496 Crossman ND, Burkhard B, Nedkov S, et al (2013) A blueprint for mapping and modelling ecosystem services. Ecosyst Serv 4:4–14. doi: 10.1016/j.ecoser.2013.02.001 Daily GC (1997) Nature’s services: societal dependence on natural ecosystems. Island Press Daily GC (2000) Management objectives for the protection of ecosystem services. Environ Sci Policy 3:333–339. Dallimer M, Davies ZG, Diaz-porras DF, et al (2015) Historical influences on the current provision of multiple ecosystem services. Glob Environ Chang 31:307–317. doi: 10.1016/j.gloenvcha.2015.01.015 Dambrine E, Dupouey JL, Laüt L, et al (2007) Present forest biodiversity patterns in france related to former Roman agriculture. Ecology 88:1430–9. Dean JS, Euler RC, Gumerman GJ, et al (1985) Human Behavior , Demography , and Paleoenvironment on the Colorado Plateaus. Soc Am Archaeol 50:537–554. Dearing JA, Yang X, Dong X, et al (2012) Extending the timescale and range of ecosystem services through paleoenvironmental analyses, exemplified in the lower Yangtze basin. Proc Natl Acad Sci U S A 109:E1111–20. doi: 10.1073/pnas.1118263109 Decamps H, Fortun M, Gazelle F, et al (1988) Historical influence of man on the riparian dynamics of a fluvial landscape.  Dimitri C (2001) Contract Evolution and Institutional Innovation: Marketing Pacific-Grown Apples from 1890 to 1930. J Econ Hist 62:189–212. Dodds WK, Gido K, Whiles MR, Daniels MD (2015) The Stream Biome Gradient Concept: factors controlling lotic systems across broad biogeographic scales. Freshw Sci. doi: 10.1086/679756. Duncan D, Kawachi I, Kum S (2014) A Spatially Explicit Approach to the Study of Socio-Demographic Inequality in the Spatial Distribution of Trees across Boston Neighborhoods. Spat Demogr 1:1–29. Dyer JM (2001) Using witness trees to assess forest change in southeastern Ohio. Can J For Res 31:1708–1718. doi: 10.1139/cjfr-31-10-1708 Egoh B, Reyers B, Rouget M, et al (2009) Spatial congruence between biodiversity and ecosystem services in South Africa. Biol Conserv 142:553–562. doi: 10.1016/j.biocon.2008.11.009 Eigenbrod F, Bell VA, Davies HN, et al (2011) The impact of projected increases in urbanization 132  on ecosystem services. Proc. R. Soc. B Biol. Sci.  Elsner M, Cuo L, Voisin N, et al (2010) Implications of 21st century climate change for the hydrology of Washington State. Clim Change 102:225–260. doi: 10.1007/s10584-010-9855-0 ESRI (2012) ArcGIS Desktop: Release 10.1.  Felipe-Lucia MR, Comin FA, Bennett EM (2014) Interactions Among Ecosystem Services Across Land Uses in a Floodplain Agroecosystem.  Foley JA, DeFries R, Asner GP, et al (2005) Global consequences of land use. Science (80- ) 309:570–574. doi: 10.1126/science.1111772 Forman RTT, Alexander LE (1998) Roads and their Major Ecological Effects. Annu Rev Ecol Syst 29:207–231. Foster D, Swanson F, Aber J, et al (2003) The Importance of Land-Use Legacies to Ecology and Conservation. Bioscience 53:77–88. doi: doi:10.1641/0006-3568(2003)053[0077:TIOLUL]2.0.CO;2 Fryirs KA, Brierley GJ (2012) Geomorphic Analysis of River Systems: An Approach to Reading the Landscape. Wiley-Blackwell Gardi C, Panagos P, Van Liedekerke M, et al (2014) Land take and food security: assessment of land take on the agricultural production in Europe. J Environ Plan Manag 1–15. doi: 10.1080/09640568.2014.899490 Gergel SE, Stange Y, Coops NC, et al (2007) What is the Value of a Good Map? An Example Using High Spatial Resolution Imagery to Aid Riparian Restoration. Ecosystems 10:688–702. doi: 10.1007/s10021-007-9040-0 Gomez-Limon J, de Lucio Fernandez JV (1999) Changes in use and landscape preferences on the agricultural-livestock landscapes of the central Iberian Peninsula (Madrid, Spain ). Landsc Urban Plan 44:165–175. Gurnell AM, Pie H (2002) Large wood and fluvial processes. Freshw Biol 47:601–619. Hagmann RK, Franklin JF, Johnson KN (2014) Historical conditions in mixed-conifer forests on the eastern slopes of the northern Oregon Cascade Range, USA. For Ecol Manage 330:158–170. doi: 10.1016/j.foreco.2014.06.044 Haining R (1991) Bivariate Correlation with Spatial Data. Geogr Anal 23:210–227. doi: 10.1111/j.1538-4632.1991.tb00235.x Hall CN, Kuss FR (1989) Vegetation Alteration along Trails in Shenandoah National Park. Biol 133  Conserv 48:211–227. Hamlet A, Lee S-Y, Mickelson K, Elsner M (2010) Effects of projected climate change on energy supply and demand in the Pacific Northwest and Washington State. Clim Change 102:103–128. doi: 10.1007/s10584-010-9857-y Hamlet AF, Lettenmaier DP (1999) Effects of climate change on hydrology and water resources in the Columbia River basin. J Am Water Resour Assoc 35:1597–1623. Hanberry BB, Fraver S, He HS, et al (2010) Spatial pattern corrections and sample sizes for forest density estimates of historical tree surveys. Landsc Ecol 26:59–68. doi: 10.1007/s10980-010-9533-7 Hanberry BB, Kabrick JM, He HS (2014) Densification and State Transition Across the Missouri Ozarks Landscape. Ecosystems 17:66–81. doi: 10.1007/s10021-013-9707-7 Hanberry BB, Kabrick JM, He HS, Palik BJ (2012) Historical trajectories and restoration strategies for the Mississippi River Alluvial Valley. For Ecol Manage 280:103–111. doi: 10.1016/j.foreco.2012.05.033 Harden CP (1992) Incorporating roads and footpaths in watershed-scale hydrologicand soil erosion models. Phys Geogr 13:368–385. doi: 10.1080/02723646.1992.10642463 Harding JS, Benfield EF, Bolstad P V, et al (1998) Stream biodiversity: the ghost of land use past. Proc Natl Acad Sci U S A 95:14843–7. Hart SK, Hibbs DE, Perakis SS (2013) Riparian litter inputs to streams in the central Oregon Coast Range. Freshw Sci 32:343–358. doi: 10.1899/12-074.1 Hatfield JL, Follett RF (2008) Nitrogen in the environment: sources, problems, and management. Elsevier He HS, Mladenoff DJ, Sickley TA, et al (2000) GIS interpolations of witness tree records (1839-1866) for northern Wisconsin at multiple scales. J Biogeogr 27:1031–1042. Helfield JM, Naiman RJ (2001) Effects of salmon-derived nitrogen on riparian forest growth and implications for stream productivity. Ecology 82:2403–2409. Hessburg PF, Agee JK, Franklin JF (2005) Dry forests and wildland fires of the inland Northwest USA: Contrasting the landscape ecology of the pre-settlement and modern eras. For Ecol Manage 211:117–139. doi: 10.1016/j.foreco.2005.02.016 Huizen P Van (2010) Building a Green Dam: Environmental Modernism and the Canadian-American Libby Dam Project. Pac Hist Rev 79:418–453. doi: 10.1525/phr.2010.79.3.418 Hutchinson TF, Boerner REJ, Iverson LR, et al (1999) Landscape patterns of understory 134  composition and richness across a moisture and nitrogen mineralization gradient in Ohio (USA) Quercus forests. Plant Ecol 144:177–189. Hyatt TL, Naiman RJ (2001) The Residence Time of Large Woody Debris in the Queets River, Washington, USA. Ecol Appl 11:191–202. Ibañez C, Belliard J, Hughes RM, et al (2009) Convergence of temperate and tropical stream fish assemblages. Ecography (Cop) 32:658–670. doi: 10.1111/j.1600-0587.2008.05591.x Isaak DJ, Peterson EE, Ver Hoef JM, et al (2014) Applications of spatial statistical network models to stream data. Wiley Interdiscip Rev Water 1:277–294. doi: 10.1002/wat2.1023 Iverson LR, Dale ME, Scott CT, Prasad A (1997) A GIS-derived integrated moisture index to predict forest composition and productivity of Ohio forests (USA). Landsc Ecol 12:331–348. Jeffres CA, Opperman JJ, Moyle PB (2008) Ephemeral floodplain habitats provide best growth conditions for juvenile Chinook salmon in a California river. Environ Biol Fishes 83:449–458. doi: 10.1007/s10641-008-9367-1 Jones JA, Swanson FJ, Wemple BC, Snyder KU (2000) Effects of Roads on Hydrology, Geomorphology, and Disturbance Patches in Stream Networks. Conserv Biol 14:76–85. doi: 10.1046/j.1523-1739.2000.99083.x Jorgensen JC, Honea JM, McClure MM, et al (2009) Linking landscape-level change to habitat quality: an evaluation of restoration actions on the freshwater habitat of spring-run Chinook salmon. Freshw Biol 54:1560–1575. doi: 10.1111/j.1365-2427.2009.02207.x Junk WJ, Bayley PB, Sparks RE (1989) The Flood Pulse Concept in River-Floodplain Systems. Can Spec Publ Fish Aquat Sci 106:110–127. Junk WJ, Welcomme R. (1990) Floodplains. In: Pattern B, Al. E (eds) Wetlands and Shallow Continental Water Bodies. SPB Academic Publishers, The Hague, The Netherlands, pp 491–524 Junker B, Buchecker M (2008) Aesthetic preferences versus ecological objectives in river restorations. Landsc Urban Plan 85:141–154. doi: 10.1016/j.landurbplan.2007.11.002 Kallio A, Puolamäki K, Fortelius M, Mannila H (2011) Correlations and co-occurrences of taxa : the role of temporal, geographic, and taxonomic restrictions.  Kienast F, Bolliger J, Potschin M, et al (2009) Assessing landscape functions with broad-scale environmental data: insights gained from a prototype development for Europe. Environ Manage 44:1099–120. doi: 10.1007/s00267-009-9384-7 135  Kuglerová L, Jansson R, Sponseller RA, et al (2015) Local and regional processes determine plant species richness in a river-network metacommunity. Ecology 96:140817224931000. doi: 10.1890/14-0552.1 Lange K, Townsend CR, Gabrielsson R, et al (2014) Responses of stream fish populations to farming intensity and water abstraction in an agricultural catchment. Freshw Biol 59:286–299. doi: 10.1111/fwb.12264 Large ARG, Gilvear DJ (2015) Using Google Earth, A Vertual-Globe Imaging Platform, For Ecosystem Services-Based River Assessment. River Res Appl 7:189. doi: 10.1002/rra Large ARG, Petts GE (1996) Historical channel-floodplain along the River Trent. Appl Geogr 16:191–209. Lautenbach S, Kugel C, Lausch A, Seppelt R (2011) Analysis of historic changes in regional ecosystem service provisioning using land use data. Ecol Indic 11:676–687. doi: 10.1016/j.ecolind.2010.09.007 Lautenbach S, Volk M, Gruber B, et al (2010) Quantifying Ecosystem Service Trade-offs. In: Swayne DA, Yang W, Voinov AA, et al (eds) International Congress on Environmental Modelling and Software Modelling for Environment’s Sake. Ottawa, Canada,  Law EA, Bryan BA, Torabi N, et al (2015) Measurement matters in managing landscape carbon. Ecosyst Serv In Press:1–10. doi: 10.1016/j.ecoser.2014.07.007 Leggett CG, Bockstael NE (2000) Evidence of the Effects of Water Quality on Residential Land Prices. J Environ Econ Manage 39:121–144. doi: 10.1006/jeem.1999.1096 Lehner B, Döll P, Alcamo J, et al (2006) Estimating the Impact of Global Change on Flood and Drought Risks in Europe: A Continental, Integrated Analysis. Clim Change 75:273–299. doi: 10.1007/s10584-006-6338-4 Lemly AD (1994) Irrigated agriculture and freshwater wetlands: A struggle for coexistence in the western United States. Wetl Ecol Manag 3:3–15. doi: 10.1007/BF00177292 Lemly AD, Kingsford RT (2000) Irrigated Agriculture and Wildlife Conservation : Conflict on a Global Scale. Environ Manage 25:485–512. doi: 10.1007/s002679910039 Lemly DA, Finger SE, Nelson MK (1993) Sources and Impacts of Irrigation Drainwater Contaminants in Arid Wetlands. Environ Toxicol Chem 12:2265–2279. Loo T (2004) People in the way: Modernity, environment, and society on the arrow lakes. B C Stud 161–196. López TM, Aide TM, Thomlinson JR (2001) Urban expansion and the loss of prime agricultural 136  lands in Puerto Rico. Ambio 30:49–54. Lunt ID, Spooner PG (2005) Using historical ecology to understand patterns of biodiversity in fragmented agricultural landscapes. J Biogeogr 32:1859–1873. doi: 10.1111/j.1365-2699.2005.01296.x MA (2005) Millennium Ecosystem Assessment: Ecosystems and HumanWell-Being: Synthesis.  Ma S, Swinton SM (2011) Valuation of ecosystem services from rural landscapes using agricultural land prices. Ecol Econ 70:1649–1659. doi: 10.1016/j.ecolecon.2011.04.004 Maes J, Paracchini ML, Zulian G, et al (2012) Synergies and trade-offs between ecosystem service supply, biodiversity, and habitat conservation status in Europe. Biol Conserv 155:1–12. doi: 10.1016/j.biocon.2012.06.016 Maizel M, White RD, Gage S, et al (1998) Historical interrelationships between population settlement and farmland in the conterminous United States 1790 to 1992.  Manies K, Mladenoff D (2000) Testing methods to produce landscape-scale presettlement vegetation maps from the U.S. public land survey records. Springer Netherlands Mantua N, Tohver I, Hamlet A (2010) Climate change impacts on streamflow extremes and summertime stream temperature and their possible consequences for freshwater salmon habitat in Washington State. Clim Change 102:187–223. doi: 10.1007/s10584-010-9845-2 McAllister LS (2008) Reconstructing historical riparian conditions of two river basins in eastern Oregon, USA. Environ Manage 42:412–25. doi: 10.1007/s00267-008-9127-1 Megonigal PJ, Conner WH, Kroeger S, Sharitz RR (1997) Aboveground Production in Southeastern Floodplain Forests: A Test of the Subsidy-Stress Hypothesis. Ecology 78:370–384. Merritt W, Alila Y, Barton M, et al (2006) Hydrologic response to scenarios of climate change in sub watersheds of the Okanagan basin, British Columbia. J Hydrol 326:79–108. doi: 10.1016/j.jhydrol.2005.10.025 Miller SG, Knight RL, Miller CK (1998) Influence of recreational trails on breeding bird communities. Ecol Appl 8:162–169. MOF (2011) Vegetation Resources Inventory: Photo Interpretation Procedures. Victoria, BC Moir HJ, Gibbins CN, Soulsby C, Webb J (2004) Linking channel geomorphic characteristics to spatial patterns of spawning activity and discharge use by Atlantic salmon (Salmo salar L.). Geomorphology 60:21–35. doi: 10.1016/j.geomorph.2003.07.014 Monz CA, Pickering CM, Hadwen WL (2013) Recent advances in recreation ecology and the 137  implications of different relationships between recreation use and ecological impacts. Front Ecol Environ 11:441–446. doi: 10.1890/120358 Mooney HA, Ehrlich PR (1997) Ecosystem services: A fragmentary history. In: Daily GC (ed) Nature’s Services: Societal Dependence on Natural Ecosystems. Island Press, Washington, D.C., pp 11–19 Moore MR, Mulville A (1991) Water Allocation in the American West : Endangered Fish Versus Irrigated Agriculture.  Morgan J, Gergel S (2010) Quantifying historic landscape heterogeneity from aerial photographs using object-based analysis. Landsc Ecol 25:985–998. doi: 10.1007/s10980-010-9474-1 Morgan JL, Gergel SE (2013) Automated analysis of aerial photographs and potential for historic forest mapping. Can J For Res 710:699–710. Morgan JL, Gergel SE, Coops NC (2010) Aerial Photography: A Rapidly Evolving Tool for Ecological Management. Bioscience 60:47–59. doi: 10.1525/bio.2010.60.1.9 Morris MR, Stanford JA (2011) Floodplain succession and soil nitrogen accumulation on a salmon river in southwestern Kamchatka. Ecol Monogr 81:43–61. Mote PW, Hamlet AF, Clark MP, Lettenmaier DP (2005) Declining Snowpack in Western North America. Bull Am Meteorol Soc 86:39–49. doi: 10.1175/BAMS-86-1-39 Mote PW, Salathe EP (2010) Future climate in the Pacific Northwest. Clim Change 102:29–50. doi: 10.1007/s10584-010-9848-z Muller E (1997) Mapping riparian vegetation along rivers : old concepts and new methods. 58:411–437. Murphy ML, Koski K V., Heifetz J, et al (1984) Role of large organic debris as winter habitat for juvenile salmonids in Alaska streams. Proc Annu Conf West Assoc Fish Wildl Agencies 64:251–262. Naidoo R, Balmford  a, Costanza R, et al (2008) Global mapping of ecosystem services and conservation priorities. Proc Natl Acad Sci U S A 105:9495–9500. doi: 10.1073/pnas.0707823105 Naiman R, Bechtold J, Beechie T, et al (2010) A Process-Based View of Floodplain Forest Patterns in Coastal River Valleys of the Pacific Northwest. Ecosystems 13:1–31. doi: 10.1007/s10021-009-9298-5 Naiman RJ, Bilby RE, Bisson PA (2000) Riparian Ecology and Management in the Pacific Coastal Rain Forest. Bioscience 50:996. doi: 10.1641/0006-138  3568(2000)050[0996:REAMIT]2.0.CO;2 Nakumura F, Swanson FJ (1994) Distribution of coarse woody debris in a mountain stream western Cascade Range, Oregon. Can J For Res 24:2395–2403. Neilsen D, Smith CAS, Frank G, et al (2006) Potential impacts of climate change on water availability for crops in the Okanagan Basin, British Columbia. Can J Soil Sci 86:921–936. Nilsson C, Grelsson G, Johansson M, Sperens U (1989) Patterns of Plant Species Richness Along Riverbanks. Ecology 70:77–84. Nowak DJ, Crane DE (2002) Carbon storage and sequestration by urban trees in the USA. Environ Pollut 116:381–9. O’Connell BM, LaPoint EB, Turner JA, et al (2014) The Forest Inventory and Analysis Database: Database description and user guide version 6.0.1 for Phase 2.  Ohlendorf HM, Hoffman DJ, Saiki MK, Aldrich TW (1986) Embryonic mortality and abnormalities of aquatic birds: apparent impacts of selenium from irrigation drainwater. Sci Total Environ 52:49–63. Olson J (2008) General Land Office Surveyors of Washington. In: Detail. Work All Surv. Assoicated with Washingt. Surv. Gen. Off. Contract Era, 1851-1910. http://www.olsonengr.com/sites/default/files/surveying-history/files/glosurveyorsdetails_0.pdf.  Osawa T, Mitsuhashi H, Niwa H, Ushimaru A (2011) The role of river confluences and meanderings in preserving local hot spots for threatened plant species in riparian ecosystems. Aquat Conserv Mar Freshw Ecosyst 21:358–363. doi: 10.1002/aqc.1194 Osawa T, Mitsuhashi H, Ushimaru A (2010) River confluences enhance riparian plant species diversity. Plant Ecol 209:95–108. doi: 10.1007/s11258-010-9726-9 Palmer M, Hondula K, Koch BJ (2014a) Ecological restoration of streams and rivers: shifting strategies and shifting goals. Annu Rev Ecol Evol Syst 45:247–272. doi: 10.1146/annurev-ecolsys-120213-091935 Palmer MA, Bernhardt ES, Allan JD, et al (2005) Standards for ecologically successful river restoration. J Appl Ecol 42:208–217. doi: 10.1111/j.1365-2664.2005.01004.x Palmer MA, Filoso S (2009) Restoration of ecosystem services for environmental markets. Science (80- ) 325:575–576. doi: 10.1126/science.1172976 Palmer MA, Filoso S, Fanelli RM (2014b) From ecosystems to ecosystem services: Stream restoration as ecological engineering. Ecol Eng 65:62–70. doi: 139  10.1016/j.ecoleng.2013.07.059 Penman J, Gytarsky M, Hiraishi T, et al (2003) IPCC-Good Practice Guidance for Land Use, Land-Use Change and Forestry. Institute for Global Environmental Strategies, pp 69–104 Pickett STA (1989) Space-for-Time Substitution as an Alternative to Long-Term Studies. In: Likens GE (ed) Long-term Studies in Ecology. Springer-Verlag, pp 110–135 Poff NL, Allan JD, Bain MB, et al (1997) The natural flow regime. Bioscience 47:769–784. doi: 10.2307/1313099 Postel S, Carpenter S (1997) Freshwater Ecosystem Services. In: Daily GC (ed) Nature’s Services. Island Press, Washington, D.C., pp 195–214 Power ME (1990) Effects of Fish in River Food Webs. Science (80- ) 250:811–814. Puter SAD, Stevens H (1908) Looters of the Public Domain. Portland Printing House Publishers, Portland, OR. Qiu J, Turner MG (2013) Spatial interactions among ecosystem services in an urbanizing agricultural watershed. Proc Natl Acad Sci U S A 110:12149–54. doi: 10.1073/pnas.1310539110 Radeloff VC, Mladenoff DJ, He HS, Boyce MS (1999) Forest landscape change in the northwestern Wisconsin Pine Barrens from pre-European settlement to the present. Can J For Res 29:1649–1659. doi: 10.1139/x99-089 Raudsepp-Hearne C, Peterson GD, Bennett EM (2010a) Ecosystem service bundles for analyzing tradeoffs in diverse landscapes. Proc Natl Acad Sci U S A 107:5242–7. doi: 10.1073/pnas.0907284107 Raudsepp-Hearne C, Peterson GD, Tengö M, et al (2010b) Untangling the Environmentalist’s Paradox: Why Is Human Well-being Increasing as Ecosystem Services Degrade? Bioscience 60:576–589. doi: 10.1525/bio.2010.60.8.4 Ray VF (1974) Ethnohistorical Notes on the Columbia, Chelan, Entiat, and Wenatchee Tribes. Interior Salish and Eastern Washington Indians IV. Garland Pub, Inc., New York & London, pp 377–437 Reed TE, Schindler DE, Hague MJ, et al (2011) Time to evolve? Potential evolutionary responses of fraser river sockeye salmon to climate change and effects on persistence. PLoS One 6:e20380. doi: 10.1371/journal.pone.0020380 Reid LM, Dunne T (1984) Sediment Production From Forest Road Surfaces. Water Resour Res 20:1753–1761. 140  Renofalt BM, Nilsson C, Jansson R (2005) Spatial and temporal patterns of species richness in a riparian landscape. J Biogeogr 32:2025–2037. doi: 10.1111/j.1365-2699.2005.01328.x Rhemtulla JM, Mladenoff DJ (2007) Why history matters in landscape ecology. Landsc Ecol 22:1–3. doi: 10.1007/s10980-007-9163-x Rhemtulla JM, Mladenoff DJ (2010) Relative consistency , not absolute precision , is the strength of the Public Land Survey : response to Bouldin. Ecol Appl 20:1187–1189. Rhemtulla JM, Mladenoff DJ, Clayton MK (2009a) Historical forest baselines reveal potential for continued carbon sequestration. Proc Natl Acad Sci 106:6082–6087. doi: 10.1073/pnas.0810076106 Rhemtulla JM, Mladenoff DJ, Clayton MK (2009b) Legacies of historical land use on regional forest composition and structure in Wisconsin, USA (mid-1800s-1930s-2000s). Ecol Appl 19:1061–78. Robinson CT, Tockner K, Ward JV (2002) The fauna of dynamic riverine landscapes. Freshw Biol 47:661–677. Rodríguez JP, Beard TD, Bennett EM, et al (2006) Trade-offs across Space, Time, and Ecosystem Services.  Rood SB, George C, George B, Tymensen W (2006) Instream flows for recreation are closely correlated with mean discharge for rivers of western North America. River Res Appl 22:91–108. doi: 10.1002/rra.888 Ruesch A, Gibbs HK (2008) New IPCC Tier-1 Global Biomass Carbon Map For the Year 2000. Oak Ridge National Laboratory, Oak Ridge, Tennesse,  Satake A, Sasaki A, Iwasa Y (2001) Variable timing of reproduction in unpredictable environments: adaption of flood plain plants. Theor Popul Biol 60:1–15. doi: 10.1006/tpbi.2001.1528 Schimel D, Stillwell M. A., Woodmansee RG (1985) Biogeochemistry of C, N, and P in a Soil Catena of the Shortgrass Steppe Author. Ecology 66:276–282. Schindler DE, Scheuerell MD, Moore JW, et al (2003) Pacific salmon and the ecology of coastal ecosystems. Front Ecol Environ 1:31–37. doi: 10.1890/1540-9295(2003)001[0031:psateo]2.0.co;2 Schluter D (1984) A Variance Test for Detecting Species Associations, with Some Example Application. Ecology 65:998–1005. Schulte LA, Mladenoff DJ (2001) The Orginal US Public Land Survey Records: Their Use and 141  Limitations in Reconstructing Presettlement Vegetation. J For 99:5–10. Schulte LA, Mladenoff DJ, Crow TR, et al (2007) Homogenization of northern U.S. Great Lakes forests due to land use. Landsc Ecol 22:1089–1103. doi: 10.1007/s10980-007-9095-5 Sedell JR, Froggatt JL (1984) Importance of streamside forests to large rivers: The isolation of the Willamette River, Oregon, U.S.A ., from its floodplain by snagging and streamside forest removal. Verh Internat Verein Limnol 22:1828–1835. Silbernagel J, Martin SR, Gale MR, et al (1997) Prehistoric, historic, and present settlement patterns related to ecological hierarchy in the Eastern Upper Peninsula of Michigan, USA. Landsc Ecol 12:223–240. Sommer TR, Nobriga ML, Harrell WC, et al (2001) Floodplain rearing of juvenile chinook salmon: evidence of enhanced growth and survival. Can J Fish Aquat Sci 58:325–333. doi: 10.1139/cjfas-58-2-325 Starr SM, Benstead JP, Sponseller R a. (2014) Spatial and temporal organization of macroinvertebrate assemblages in a lowland floodplain ecosystem. Landsc Ecol 29:1017–1031. doi: 10.1007/s10980-014-0037-8 Statzner B, Higler B (1985) Questions and Comments on the River Continuum Concept.  Stinchcomb GE, Messner TC, Driese SG, et al (2011) Pre-colonial (A.D. 1100-1600) sedimentation related to prehistoric maize agriculture and climate change in eastern North America. Geology 39:363–366. doi: 10.1130/G31596.1 Stockle CO, Nelson RL, Higgins S, et al (2010) Assessment of climate change impact on Eastern Washington agriculture. Clim Change 102:77–102. doi: 10.1007/s10584-010-9851-4 Swanson FJ, Kratz TK, Caine N, Woodmansee RG (1988) Landform Effects of Ecosystem Patterns and Processes. Bioscience 38:92–98. Sweeney BW, Bott TL, Jackson JK, et al (2004) Riparian deforestation, stream narrowing, and loss of stream ecosystem services. Proc Natl Acad Sci U S A 101:14132–14137. doi: 10.1073/pnas.0405895101 Tallis HT, Ricketts T, Guerry AD, et al (2011) InVEST 2.1 beta User’s Guide. The Natural Capital Project, Stanford Thompson WL, Lee DC (2000) Modeling relationships between landscape-level attributes and snorkel counts of chinook salmon and steelhead parr in Idaho. Can J Fish Aquat Sci 57:1834–1842. doi: 10.1139/f00-135 Thorp JH, Thoms MC, Delong MD (2006) The riverine ecosystem synthesis: biocomplexity in 142  river networks across space and time. River Res Appl 22:123–147. doi: 10.1002/rra.901 Tockner K, Stanford JA (2002) Riverine flood plains: present state and future trends. Environ Conserv 29:308–330. doi: 10.1017/s037689290200022x Tohver I, Hamlet AF (2010) Impacts of 21st century climate change on hydrologic extremes in the Pacific Northwest region of North America. Final Report for the Columbia Basin Climate, Change Scenarios Project, Climate Impacts Group, Center for Science in the Earth System. Chapter 7. Seattle Tomlinson MJ, Gergel SE, Beechie TJ, McClure MM (2011) Long-term changes in river-floodplain dynamics: implications for salmonid habitat in the Interior Columbia Basin, USA. Ecol Appl 21:1643–1658. Tomscha SA, Gergel SE (2015) Historic land surveys present opportunities for reconstructing frontier settlement patterns in North America.  Tornwall B, Sokol E, Skelton J, Brown B (2015) Trends in Stream Biodiversity Research since the River Continuum Concept. Diversity 7:16–35. doi: 10.3390/d7010016 Trabucchi M, Comín FA, O’Farrell PJ (2013) Hierarchical priority setting for restoration in a watershed in NE Spain, based on assessments of soil erosion and ecosystem services. Reg Environ Chang 13:911–926. doi: 10.1007/s10113-012-0392-4 Trabucchi M, Ntshotsho P, O’Farrell P, Comín F a. (2012) Ecosystem service trends in basin-scale restoration initiatives: A review. J Environ Manage 111:18–23. doi: 10.1016/j.jenvman.2012.06.040 Trombulak SC, Frissell CA (2000) Review of Ecological Effects of Roads on Terrestrial and Aquatic Communities. Conserv Biol 14:18–30. doi: 10.1046/j.1523-1739.2000.99084.x Turner KG, Odgaard MV, Bøcher PK, et al (2014) Bundling ecosystem services in Denmark: Trade-offs and synergies in a cultural landscape. Landsc Urban Plan 125:89–104. doi: 10.1016/j.landurbplan.2014.02.007 U.S. Department of the Interior B of LM Land Status and Cadastral Survey Records. http://www.blm.gov/or/landrecords/survey/ySrvy1.php.  Ulrich RS (1986) Human Responses to Vegetation and Landscapes. Landsc Urban Plan 13:29–44. USGS (1993) Data Users Guides. Reston, Virginia Van Deusen P, Heath L. (2014) COLE web applications suite. In: NCASI USDA For. Serv. North. Res. Station. http://www.ncasi2.org/COLE/.  143  Vannote RL, Minshall GW, Cummins KW, et al (1980) The River Continuum Concept. Can J Fish Aquat Sci 37:130–137. Vano J, Scott M, Voisin N, et al (2010) Climate change impacts on water management and irrigated agriculture in the Yakima River Basin, Washington, USA. Clim Change 102:287–317. doi: 10.1007/s10584-010-9856-z Villamagna AM, Angermeier PL, Bennett EM (2013) Capacity, pressure, demand, and flow: A conceptual framework for analyzing ecosystem service provision and delivery. Ecol Complex 15:114–121. doi: 10.1016/j.ecocom.2013.07.004 Villamagna AM, Mogollón B, Angermeier PL (2014) A multi-indicator framework for mapping cultural ecosystem services: The case of freshwater recreational fishing. Ecol Indic 45:255–265. doi: 10.1016/j.ecolind.2014.04.001 Wan Y, Wan C, Hedgepeth M (2014) Elucidating multidecadal saltwater intrusion and vegetation dynamics in a coastal floodplain with artificial neural networks and aerial photography. Ecohydrology n/a–n/a. doi: 10.1002/eco.1509 Waples RS, Beechie T, Pess GR (2009) Evolutionary history, habitat disturbance regimes, and anthropogenic changes: What do these mean for resilience of Pacific salmon populations? Ecol Soc. doi: 3 Ward JV, Stanford JA (1995) The Serial Discontinuity Concept: Extending the Model to Floodplain Rivers. Regul Rivers Res Manag 10:159–168. Westman WE (1977) How Much Are Nature’s Services Worth? Science 197:960–4. doi: 10.1126/science.197.4307.960 White MA, Mladenoff DJ (1994) Old-growth forest landscape transitions from pre-European settlement to present. Landsc Ecol 9:191–205. doi: 10.1007/BF00134747 Whitehead PG, L. WR, Battarbee RW, et al (2009) A review of the potential impacts of climate change on surface water quality. Hydrol Sci J 54:101–123. doi: 10.1623/hysj.54.1.101 Whitney GG (1986) Relation of Michigan’s Presettlement Pine Forests to Substrate and Disturbance History. Ecology 67:1548–1559. Wichelns D, Oster JD (2006) Sustainable irrigation is necessary and achievable, but direct costs and environmental impacts can be substantial. Agric Water Manag 86:114–127. doi: 10.1016/j.agwat.2006.07.014 Wiens JA (2002) Riverine landscapes: taking landscape ecology into the water. Freshw Biol 47:501–515. doi: 10.1046/j.1365-2427.2002.00887.x 144  Wik M, Pingali P, Broca S (2008) Global Agricultural Performance: Past Trends and Future Prospects.  Woodroffe R, Hedges S, Durant SM (2014) To Fence or Not to Fence. Science (80- ) 344:46–48. doi: 10.1126/science.1246251 Wu J, Feng Z, Gao Y, Peng J (2013) Hotspot and relationship identification in multiple landscape services: A case study on an area with intensive human activities. Ecol Indic 29:529–537. doi: 10.1016/j.ecolind.2013.01.037 Zhao B, Kreuter U, Li B, et al (2004) An ecosystem service value assessment of land-use change on Chongming Island, China. Land use policy 21:139–148. doi: 10.1016/j.landusepol.2003.10.003 (2014) American Whitewater. http://www.americanwhitewater.org/.  (2013) Funding Guidelines State Fiscal Year 2015. Olympia, Washington   145  Appendix Table A.1 Confusion matrix of 1949 and 2006 classifications as performed by two different aerial photography interpreters 1949 Interpreter 1 Accuracy Interpreter 2  MC MCB %  MC 23 9 71 MCB 7 21 75 Accuracy % 77 70 73  2006 Interpreter 1 Accuracy Interpreter 2  MC MCB % MC 28 5 85 MCB 2 25 92 Accuracy % 93 83 88  Table A.2 Confusion matrix of 1949 density classifications as performed by two different aerial photography interpreters. Overall agreement among density classes is 81% with the greatest confusion occurring between moderate and high classes, with Interpreter 2 more likely to identify forests as moderate in density. Interpreter 1 was also more likely to classify vegetation as low density.  1949 Interpreter 1 Accuracy Interpreter 2  High Moderate Low % High 34 4 0 89 Moderate 16 46 9 64 Low 0 0 41 100 Accuracy % 68 92 82 81    146  Table A.3 Confusion matrix of 2006 density classifications as performed by two different aerial photography interpreters. Overall agreement among density classes is 83%, with the greatest confusion occurring between low and moderate classes with Interpreter 2 more likely to classify what Interpreter 1 considers low as moderate. 2006 Interpreter 1 Accuracy Interpreter 2  High Moderate Low % High 44 4 1 90 Moderate 6 46 15 69 Low 0 0 34 100 Accuracy % 88 92 68 83  Table A.4 Land-cover accuracy assessment. I used a total of 367 groundtruthing points.  Several issues include the five-gap from when the photos were taken to the time of the accuracy assessment. Overall accuracy shown with (*).   Groundtruthing User’s Accuracy Air photo classification   Forest/Shrub Urban Orchard Field % Forest/Shrub 76 1 1 2 95.0 Urban 8 73 5 5 80.2 Orchard 13 1 91 3 84.3 Field 11 1 9 64 75.3 Producer’s Accuracy % 70.4 96.1 85.8 86.5 82.8*   

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:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0220636/manifest

Comment

Related Items