UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Responses of macroinvertebrate community composition to changes in stream abiotic factors after streamside… Fuchs, Shirley A. 1999

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

Item Metadata

Download

Media
831-ubc_2000-0059.pdf [ 5.44MB ]
Metadata
JSON: 831-1.0089348.json
JSON-LD: 831-1.0089348-ld.json
RDF/XML (Pretty): 831-1.0089348-rdf.xml
RDF/JSON: 831-1.0089348-rdf.json
Turtle: 831-1.0089348-turtle.txt
N-Triples: 831-1.0089348-rdf-ntriples.txt
Original Record: 831-1.0089348-source.json
Full Text
831-1.0089348-fulltext.txt
Citation
831-1.0089348.ris

Full Text

RESPONSES OF MACROINVERTEBRATE COMMUNITY COMPOSITION TO CHANGES IN STREAM ABIOTIC FACTORS AFTER STREAMSIDE CLEAR-CUT LOGGING IN THE CENTRAL INTERIOR OF BRITISH COLUMBIA by SHIRLEY A. FUCHS B.Sc, Dalhousie University, 1993 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER S OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Zoology) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA December 1999 © Shirley A. Fuchs, 1999 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of o & l o The University of British Columbia Vancouver, Canada Date - ' X o \C/<j DE-6 (2/88) Abstract Effects of a clear-cut timber harvest of the riparian zone were studied on 18 stream systems in watersheds influenced by a continental climate regime in the central interior of British Columbia. Streams selected in the Stuart-Takla and Willow-Bowron watersheds were sampled between mid-September and late October 1995. Study streams were classified according to stages of clear-cut harvest in the riparian zone: streams within unharvested reaches, streams without canopy cover (timber harvest less than 5 years old) or streams with a deciduous canopy cover (timber harvest more than 20 years old). Measurements were taken of topographic and habitat attributes, of watersheds and productivity and the benthic macroinvertebrate community. 'Uncut' stream channels had higher gradients and narrower widths than were found in 'old cut' streams suggesting that stream channel dimensions of streams in logged areas may have changed over time. More specifically, 'recent cut' streams had the lowest canopy cover and the highest chlorophyll a biomass. However, in 'old cut' streams with dense canopy cover, chlorophyll a biomass was similar to levels in 'uncut' streams. In addition, although density of benthic macroinvertebrates did not appear to have changed over time, total biomass increased soon after timber harvest. After more than 20 years, biomass measurements of macroinvertebrates were similar to levels found in unharvested streams. Density and biomass of the scraper functional guild, in particular, were lowest in 'recent cut' streams. These streams also had the highest densities of predators and invertebrate parasites and the highest biomass of collector functional guilds and leeches. Higher biomass of the two collector guilds in 'recent cut' streams may have been linked to an increase in the quality of their food source, i.e., FPOM, although this was not measured in this study. Higher biomass of scrapers and collectors found soon after the timber harvest may have attracted resident fish into these streams, and so increased fish parasite biomass. After more than 20 years following timber harvest, biotic and abiotic factors may have ii returned to levels found prior to timber harvests. Future work is required to determine the factors that influence changes in benthic community composition and instream habitat in stream ecosystems following timber harvest in this region. iii Table of Contents Abstract ii Table of Contents iv List of Tables vi List of Figures xi List of Appendices xv Acknowledgements xvi Dedication xvii 1. Introduction 1 1.1 Disturbance and Recovery of Stream Ecosystems 1 1.2 Effects of Riparian Harvest on Streams 4 2. Methods 9 ; 2.1 Study site description 9 2.2 Field collection 14 2.3 Sample processing 16 2.3.1 Periphyton 16 2.3.2 Benthic Macroinvertebrates 17 2.3.3 Substrate 18 2.4 Data Organization 18 2.5 Statistical Analysis 20 3. Results. 25 3.1 Environmental variables 25 3.1.1 Landscape characteristics 25 3.1.1.1 Watershed descriptors 25 3.1.1.2 Topography descriptors '..33 3.1.2 Physical habitat characteristics 33 3.1.2.1 Instream 33 3.1.2.2 Instream chemistry/productivity descriptors 46 3.2 Macroinvertebrate variables 54 3.2.1 Macroinvertebrate functional guild density 54 3.2.2 Macroinvertebrate taxa density 66 3.2.3 Macroinvertebrate functional guild biomass 73 3.3 Macroinvertebrate relationships with instream habitat and productivity 81 iv 4.0 Discussion 93 4.1 Macroinvertebrate density and biomass 98 4.1.1 Total macroinvertebrate density and biomass 98 4.1.2 Functional guilds density and biomass 100 4.1.3 Taxa density 105 4.2 Environment-macroinvertebrate relationships 107 4.3 Comparison of streams influenced by continental and maritime climates Ill 4.4 Study limitations 115 Literature cited 119 v List o f Tables Table 1. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=18 variables) of the watershed dataset .26 Table 2. Table 3. Table 4. Table 5. F-value approximations and associated P-values for M A N O V A , contrasting three groups (TJ'- 'uncut' streams, 'R'- 'recent cut' streams and '0'- 'old cut' streams) based on watershed, topography, instream habitat and instream chemical/productivity datasets. Differences among groups were assessed by canonical variates analysis. Those groups whose centroids differ are indicated by a separate letter and groups with similar centroids have their letters attached .28 The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis based on the watershed variables of 18 streams : A N O V A results contrasting three groups ('U'- 'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the watershed dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r 2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached .30 .32 Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=18 variables) of the topographic dataset. .34 v i Table 6. F-value approximations and associated P-values for A N O V A , contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the topography dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r 2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached 36 Table 7. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=27 variables) of the instream habitat dataset 38 Table 8. The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis based on the instream habitat variables of 27 stream sites : 42 Table 9. F-value approximations and associated P-values for A N O V A , contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the instream habitat dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r 2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached. 44 Table 10. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=18 variables) of the instream chemical/productivity dataset 48 Table 11. The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis.based on the instream chemical/productivity variables of 18 stream sites 50 vn Table 12. F-value approximations and associated P-values for A N O V A , contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the instream chemical/productivity dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r 2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached 52 Table 13. F-value approximations and associated P-values for A N O V A s contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on total benthic macroinvertebrate biomass and density datasets. The degrees of freedom (df), Mean Square term (MS), R-square (r 2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by Tukey's Studentized Range Test, whose means differ are indicated by a separate letter. Those groups with similar means have their letters attached 55 Table 14. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=27 variables) of the functional guild density dataset .58 Table 15. F-value approximations and associated P-values for M A N O V A , contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on functional guild density and biomass datasets. The degrees of freedom (df), Mean Square term (MS), R-square (r 2), F-value approximations (F) and associated P values of significance (P) are listed. Those groups whose centroids differ are indicated by a separate letter and groups with similar centroids have their letters attached 61 Table 16. The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis based on the functional guild density variables (n= 27 sites) 62 viii Table 17. F-value approximations and associated P-values for ANOVA, contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the functional guild density dataset. The degrees of Table 18. F-value approximations and associated P-values for ANOVA, contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the benthic macroinvertebrate taxa density dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached 67 Table 19. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=27 variables) of the Table 20. The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis based on the functional guild biomass freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached 64 functional guild biomass dataset. 74 variables of 27 stream sites 78 ix Table 21. F-value approximations and associated P-values for A N O V A , contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the functional guild biomass dataset. Differences among groups were assessed by the Tukey's Studentized test. The degrees of freedom (df), Mean Square term (MS), R-square (r 2), F-value approximations (F) and associated P values of significance (P) are listed. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached 79 Table 22. Pearson correlation between the first two axes from the P C A on functional guild density data and first two axes from the PCA' s on instream and chemical datasets. Probability of significance is in brackets 82 Table 23 . Pearson correlation between the first two axes from the P C A on functional guild biomass data and first two axes from the PCA' s on instream and chemical datasets. Probability of significance is in brackets 83 Table 24. Results of the Canonical correlation analysis. ' R C i ' and ' R C 2 ' represent the canonical coefficients. ' V and ' W represent their respective canonical variates (see Methods section). Pearson correlations and probability showing the major trends between functional guild density and instream variables of the canonical coefficients are also shown 92 x List of Figures Fi gure 1. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the watershed dataset. Polygons delineate streams belonging to 'uncut', 'recent cut' and 'old cut' treatments. 27 Figure 2. Plot of Canonical Variate 2 (CAN2) versus Canonical Variate 1 ( C A N ! ) for the watershed dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 29 Figure 3. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the topography dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 35 Figure 4. Histograms ( ± 1 standard error) of In transformed percent channel gradient variables in 'uncut', 'recent cut' and 'old cut' streams 37 Figure 5. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the instream habitat dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 40 Fi gure 6. Plot of Canonical Variate 2 (CAN2) versus Canonical Variate 1 (CAN1) for the instream habitat dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 41 Figure 7. Histograms ( + 1 standard error) of In transformed instream habitat variables in 'uncut', 'recent cut' and 'old cut' streams 45 Figure 8. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the instream chemical/productivity dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 49 Fi gure 9. Plot of Canonical Variate 2 (CAN2) versus Canonical Variate 1 (CAN1) for the instream chemical/productivity dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 51 xi Figure 10. Histograms ( ± 1 standard error) of ln transformed percentage of canopy cover variables in 'uncut', 'recent cut' and 'old cut' streams 53 Figure 11. Histograms ( ± 1 standard error) of ln transformed total taxa density variables in the riffle habitat of the 'uncut', 'recent cut' and 'old cut' streams 56 Figure 12. Histograms ( ± 1 standard error) of ln transformed total taxa biomass variables in the riffle habitat of the 'uncut', 'recent cut' and 'old cut' streams 57 Figure 13. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for functional guild density. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 59 Figure 14. Plot of Canonical Variate 2 (CAN2) versus Canonical Variate 1 (CAN1) for the functional guild density dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 63 Figure 15. Histograms ( ± 1 standard error) of ln transformed functional guild density of collector-gatherers, scrapers, predators, invertebrate parasites and fish parasites in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets 65 Figure 16. Histograms ( ± 1 standard error) of ln transformed density of taxa within the collector guild, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets 68 Figure 17. Histograms ( ± 1 standard error) of ln transformed density of taxa within the predator guild, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets 69 Figure 18. Histograms ( ± 1 standard error) of ln transformed density of taxa within the parasite guild, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets 70 xn Figure 19. Histograms ( ± 1 standard error) of In transformed density of taxa within the scraper guilds, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance-is in brackets 71 Figure 20. Histograms ( ± 1 standard error) of In transformed density of taxa within the shredder-detritivore guild, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets 72 Figure 21. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the functional guild biomass dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 75 Figure 22. Plot of Canonical Variate 2 (CAN2) versus Canonical Variate 1 (CAN1) for the functional guild biomass dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 77 Figure 23. Histograms ( ± 1 standard error) of In transformed biomass of the collector-gatherer, scraper, predator, invertebrate parasite and fish parasite functional guilds, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets 80 Figure 24. Plot of the functional guild density data Principal Component axis 2 versus the physical habitat data Principal Component axis 1. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 85 Figure 25. Plot of the functional guild density data Principal Component axis 2 versus the physical data Principal Component axis 2. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 86 Figure 26. Plot of the functional guild density data Principal Component axis 1 versus the chemical Principal Component axis 1. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 88 xiii Figure 27. Plot of the functional guild density data Principal Component axis 2 versus the chemical Principal Component axis 2. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 89 Figure 28. Plot of the functional guild biomass data Principal Component axis 2 versus the chemical data Principal Component axis 2. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams 91 xiv List of Appendices Appendix 1. General physical characteristics of the study locations in the Stuart-Takla and Willow Bowron watersheds, September to October 1995 125 xv Acknowledgements M y appreciation goes to my supervisor, Dr. Scott Hinch of the U B C Institute for Research and Environment and Forest Sciences Department, for his invaluable advice, guidance and constant support throughout my project and amongst many other things, his help particularly with respect to unfamiliar statistical procedures. Special thanks to my committee members, Dr. J. Stevenson Macdonald of the Department of Fisheries and Oceans and Dr. John Richardson of the U B C Forest Sciences Department for their support and direction in this project. I am also indebted to Eric Mellina of the U B C Forest Sciences Department for his invaluable assistance in the design of this study and collection of data. Thaddeus Siedler was especially helpful in the collection of the data and processing of macroinvertebrate samples. Funding for this project was provided by a: N S E R C - D F O Science Subvention Grant, N S E R C Research Grant, B . C . Ministry of Forests, B . C . Ministry of Environment, Lands and Parks, and Forest Renewal B . C . The stream physical habitat data presented in this thesis forms part of Eric Mellina's PhD thesis and are used by permission. The publication of data analysis in this thesis format in no way prohibits the use of the same data in theses or scientific publications by others. xvi Dedication This thesis is dedicated to my father Horst, whose love and support for this endeavour has been constant supply. XVII 1. Introduction 1.1 Disturbance and Recovery of Stream Ecosystems The concepts of ecosystem disturbance and recovery have been given considerable attention in the literature (see Bender et al. 1984; Sousa 1984 as cited in Wallace 1990; White and Pickett 1985; Resh et al. 1988; Wallace 1990; Yount and Niemi 1990; Detenbeck et al. 1990; Niemi et al. 1990; and Poff 1992). In the context of population-dynamics context Sousa (1984) described "disturbance" as a "discrete, punctuated, killing, displacement or damaging of one or more individuals (or colonies) that directly or indirectly creates an opportunity for new individuals (or colonies) to become established". In a broader context, White and Pickett (1985) described "disturbance" as "any relatively discrete event in time that disrupts ecosystem, community, or population structure, and that changes resources, availability of substratum, or the physical environment." In the context of stream ecosystems, a disturbance can be natural (e.g. flooding, fire) or anthropogenic (e.g. pollution, dredging, forestry) in origin. Recovery is a process that "moves" the ecosystem towards the previously existing, pre-altered state (Yount and Niemi 1990). When a disturbance activity occurs in a stream ecosystem the resultant changes in the benthic macroinvertebrate community structure (abundances, biomass, productivity) may persist from several weeks to several decades. Rates of recovery of functional guilds (i.e., abundances, biomass and/or production of functional feeding groups) and taxonomic recovery (individual species or taxa) often differ (Wallace 1990). Several factors may affect the recovery rate (resilience) of the benthic macroinvertebrate community structure to the pre-altered state. Characteristics of the disturbance activity itself (i.e., the spatial scale, intensity, frequency and duration) can potentially determine ecological response and recovery (Kelly and Harwell 1990). Bender et al. (1984) first used the terms pulse and press to categorize the "nature of a disturbance" into two general types based on the duration of their effects on the structure and function of a community (Stone and Wallace 1998). A pulse 1 perturbation describes the effect of a discrete, sometimes catastrophic, disturbance event in which the effects continue for a short "limited and easily definable duration" (Niemi et al. 1990). In stream ecosystems, point-source inputs of sediment or contaminants or short-lived hydrologic events are examples of a pulse disturbance (Detenbeck et al. 1990) as are high peak rainfall or discharge events. In general, recovery of the benthic macroinvertebrate community often begins immediately after the cessation of the disturbance (Detenbeck et al. 1990) and is often due to "re-colonization from nearby undisturbed areas or internal refugia" (Yount and Niemi 1990). Stream macroinvertebrate communities often recover from pulse effects of a disturbance event in a few months to five years (Wallace 1990; and Niemi et al. 1990). In contrast, the effects of a press perturbation on an ecosystem can continue over a relatively long period of time after the initial discreet disturbance event (Niemi et al. 1990). The "sustained alteration" (Bender et al 1984) of the community's function and structure may be due to long-term changes in "stream thermal regime" (Wallace 1990), "stream productivity", "habitat integrity" and the "persistence of the stressor" (Niemi et al. 1990) after the initial disturbance event. In stream ecosystems, the sustained alteration of the community's structure and function can be found after an event that reduces the stability of stream banks or reduces the quality of physical instream habitat (Niemi et al. 1990). The benthic community structure may adjust to long-term changes in habitat through the elimination or reduction of susceptible species (Yount and Niemi 1990), although many studies on press disturbances show no major losses or gains in specific taxa (Wallace 1990). There may also be increases in local densities and biomass of taxa in those functional feeding guilds that are able to benefit from these temporary habitat modifications (Hawkins et al. 1982). In some cases, however, the physical instream habitat does not recover to the pre-stressed state so the long-term effects on the community continues. The benthic community structure then, may proceed to form an alternative or new, relatively stable state of community composition, i.e., one better adapted to the new physical habitat conditions, 2 instead of returning to their pre-stressed state. Recovery of macroinvertebrate density or biomass may not occur until the physical habitat of the stream returns to a semblance of its pre-stressed state (Niemi et al. 1990). However, once physical habitat has recovered, the response of the biota appears to be relatively rapid (Niemi et al. 1990). The recovery rate of the community is influenced by the specific effects of the disturbance and recovery rates for each organism within the community, controlled by any number of biological, instream productivity and physical factors (Niemi et al. 1990). Specific life history traits (including generation time, emergence time, mobility and propensity to disperse) of the organisms influence susceptibility and recovery (Niemi et al. 1990). Similarly, the time of year the disturbance activity occurs can have a strong influence on the successful re-colonization of individual species or total populations. Presence of, and distance from, refugia (e.g. nearby epicentres as sources of organisms) also may affect recovery of the community composition (Niemi et al. 1990). For example, as the distance of the refugia from the disturbed area increases, the importance in the vagility of the adults and life-history traits (e.g., generation time and fecundity) to the rate of recovery also increases (Niemi et al. 1990). The history of disturbance activity may also contribute to the resilience of the macroinvertebrate community structure in a lotic ecosystem. Previous exposure to a particular disturbance or disturbance regime may modify such components of organisms within the community as the specific life history trait or certain physiological resistances (Wallace 1990). These modifications, then, may aid in shaping the resilience of a community structure and response of the community to subsequent natural or anthropogenic disturbance activity. 3 1.2 Effects of Riparian Harvest on Streams Timber harvesting often causes long-term physical alteration of stream habitats (Wallace 1990), sometimes throughout the entire watershed (Detenbeck et al. 1990), and is generally categorized as a press disturbance. These habitat alterations can change the biotic community structure to an alternative state because "the magnitude of the critical driving forces (e.g. nutrient inputs and volume of watershed runoff)" and the "transformation of energy inputs (e.g. the absorption of radiant energy by an overhead canopy) for these ecosystems have been excessively altered" (Detenbeck et al. 1990). In the section below, I will review the effects of riparian harvest on streams and their macroinvertebrate fauna. Riparian timber harvest can alter streams through the introduction of sediment and debris, by altering flow and temperature regimes and by destabilizing streambanks (Kiss 1984). Perhaps one of the most important influences with respect to benthic macroinvertebrates, is the removal of riparian vegetation and canopy cover. These changes can alter the nature, quantity and timing of energy and nutrient inputs into streams (Mahoney and Erman 1980; Gurtz and Wallace 1984; Noel et al. 1986; and Breschta et al. 1987). Immediately following a clear-cut timber harvest of the riparian zone, the canopy cover is opened and the amount of sunlight into the stream increases (Gregory et al. 1987). Subsequently, the summer stream temperature may also increase. With the increase in light, temperature, and dissolved nutrient concentrations (Murphy and Hall 1981; Noel et al. 1986; Gregory et al. 1987; Webster 1990), aquatic primary productivity may increase (Murphy and Hall 1981; Murphy et al. 1981; Murphy et al. 1986; Breschta et al. 1987; Webster 1990; and Anderson 1992). The growth in primary productivity (for example periphyton biomass; Hansmann and Phinney 1973) often leads to increased benthic macroinvertebrate density (Thedinga 1989), biomass and diversity (Mahoney and Erman 1980; Newbold et al. 1980; Murphy et al. 1981; and Murphy and Hall 1981), assuming summer temperatures are sub-lethal for the 4 macroinvertebrates. As successional occur in the riparian zone, deciduous shrubs and plants become dominant, increasing both the terrestrial nutrient uptake and density of the canopy cover (Murphy et al. 1986; and Anderson 1992). Upper story species follow increasing the canopy closure and cause a decrease in the amount of sunlight reaching the stream. A decline in primary productivity occurs (although there is an associated seasonal increase in the allochthonous food source due to increased leaf litter input) and the density and biomass of benthic macroinvertebrates may also decline. Benthic macroinvertebrate diversity and richness have been found to decrease below levels found in streams of old growth forests after the initial 10-20 year increase (Murphy et al. 1986; and Anderson 1992). The link between the successional changes of the riparian vegetation, the instream resources and the subsequent alterations of the benthic macroinvertebrate community composition after a disturbance by a clear-cut timber harvest disturbance is well supported in the literature (Newbold et al. 1980; Murphy and Hall 1981; Hawkins et al. 1982; Culp and Davies 1983; Murphy et al. 1986; Noel et al. 1986; Carlson et al. 1990; Webster 1990; and Anderson 1992). Most of the research investigating the influence of disturbances related to timber harvesting on streams and their biota, has been conducted in mountainous regions with maritime climates (i.e., high winter storm frequency and high precipitation rates; Heede 1984). In these regions, hillslope and stream gradients as well as precipitation volume affect timing and magnitude of stream discharge (Culp and Davies 1983) and many other factors. Stream ecosystems within clear-cut timber harvest catchments, with this combination of climate and topography are often exposed to large amounts of hillslope erosion via landslides. The discharge of these streams increase dramatically. With the subsequent loss of large woody debris streams and erosion of the stream banks, the streams quickly become highly simplified, shallower and develop wider stream channels (Swanson 1979). There is much evidence to show that dramatic changes can occur in the benthic macroinvertebrate community composition within these 5 streams in response to habitat modifications after a clear-cut timber harvest of the riparian zone (Newbold et al. 1980; Murphy and Hall 1981; Hawkins et al. 1982; Culp and Davies 1983; Murphy et al. 1986; Webster 1990; and Anderson 1992). Stream ecosystems may not return to their pre-cut structure until the instream habitat recovers. Depending upon the terrestrial conditions such as forest type which influences such factors as the rate of forest succession and disturbance regime, the instream habitat may not recover to the pre-stressed state for more than 200 years after the timber harvest (Gregory et al. 1987). Alternatively, stream systems in watersheds that encounter continental climates (i.e., low storm frequencies and low precipitation) and which typically also have more moderate stream and hillslope gradients, may not experience the intensity, magnitude or duration of the press disturbance regimes found in maritime systems following clear-cut timber harvest in the catchment. Channels may be more unstable and hillslopes less prone to erosion (i.e., sediment intrusion) than in maritime systems and potentially a faster recovery of the physical instream habitat and their biota may occur. Few studies have been conducted on the effects of a clear-cut timber harvest on the physical instream habitat and benthic macroinvertebrate communities of streams in catchments with continental climates and a gentler topography (see Silsbee and Larson 1983; Noel et al. 1986; Carlson et al. 1990; and Stone and Wallace 1998). Alteration of the primary productivity (i.e., through removal of canopy cover) and the sediment intrusion regime in these systems appear to be the factors most affected by this disturbance and that subsequently have the most effect on the stream ecosystem (Murphy and Hall 1981). In general, I suggest that the interior stream systems of British Columbia, influenced by continental climates (i.e., low storm frequencies and low precipitation) and moderate stream and hillslope gradients, will experience less channel instability and less hillslope erosion after a clear-cut timber harvest of the riparian zone over time-since-cut than in the stream systems found in the more maritime systems. Therefore, the changes that may occur in the benthic 6 macroinvertebrate community structure over time may be those associated with shifts in primary productivity influenced by the density of the stream's canopy cover. Richardson (1994) predicted that in the interior streams overall productivity of the benthic community may increase for several years after an initial timber harvest due to an increase in sunlight and nutrients to the stream. He also predicted that in these streams of the interior region, after several years primary productivity would decline as nutrient yields slow and sunlight is blocked to the streambed with the re-establishment of successional vegetation. Thus benthos productivity would also decline. Another factor that may have an detrimental effect on benthos productivity is an increase in fine sediment intrusion following timber harvest activities which has been documented in this region (Slaney et al. 1977; Brownlee et al. 1988; and Larkin et al. 1998). The soils in this region are primarily lacustrine and very erodable (Slaney et al. 1977; and Larkin et al. 1998) so it could have a short-term effect on the physical habitat of these streams (Slaney et al. 1977; Brownlee et al. 1988; and Larkin et al. 1998). Fine sediment intrusion has been found to have a negative effect on the benthic macroinvertebrate community abundances (Lemly 1982) however, increases in primary productivity from the removal of canopy cover have been found to override or mask effects of sedimentation (Murphy et al. 1981; and Hawkins et al. 1982) over the short-term. The principal objective of this study was to examine the effects of a clear-cut timber harvest of the riparian buffer zone on the benthic macroinvertebrate community composition and the watershed, topographic, instream productivity and physical attributes of streams in the Central-Interior region of British Columbia. The study is primarily exploratory and provides a preliminary basis for future work in these regions. Specifically I will investigate the relationships within the benthic macroinvertebrate community and environmental variables among stream sites and stream treatments; to discern whether changes in the benthic community composition could be associated with variation in physical stream habitat and productivity; and to evaluate 7 whether benthic community composition and the environmental factors change over time since the riparian buffer zones were harvested (recently cut and older cut) with respect to the unlogged streams. To accomplish the above objective, the benthic macroinvertebrate community structure and the watershed, topographic, instream productivity and physical attributes were assessed in streams that drain uncut, recently cut (less than 5 years old) and older cut (greater than twenty years old) watersheds. The riparian zones of all harvested streams were clear-cut to the streamside. The successional vegetation in the harvested streams were in various stages of successional recovery. 8 2. Methods 2.1 Study site description The present study was conducted in sites within the Stuart-Takla region, described in detail by Macdonald et al. (1992) and the Wil low-Bowron watersheds, described in detail by Larkin et al. (1998). The Stuart-Takla region is located in the most northern area of the Fraser watershed among the Omineca mountains (Lat. 55 °.00' N . Long. 125°.50'W.) (Macdonald et al. 1992) in the Interior Plateau of British Columbia. Glacial t i l l deposits intermixed with lacustrine clays were important sources of sediment input into the streams of this region (Macdonald et al. 1992). The W i l l o w and Bowron watersheds have experienced much greater levels of logging than the Stuart-Takla. They are situated south-east of the Stuart-Takla region. Both watersheds drain northwards into the Fraser River (Lat. 53 °.45' N . Long. 122°.00'W.), just west of the Cariboo mountains in the Interior Plateau of British Columbia. The soil in the watersheds consist primarily of glaciolacustrine and sandy glaciofluvial deposits (Larkin et al. 1998). Extensive clear-cut harvesting occurred in the W i l l o w and especially the Bowron River watersheds in the late 1970's and early 1980's due to a beetle infestation. More recent harvesting had been done in the higher elevations, in the watersheds of the tributary streams of these two rivers prior to and during my study. The Stuart-Takla region and W i l l o w and Bowron watersheds within the Sub Boreal Spruce biogeoclimatic zone. The Sub Boreal Spruce zone (SBS) characterizes the lower elevations of the Stuart-Takla region and the W i l l o w and Bowron watersheds. The characteristic trees in these regions include the Lodgepole pine (Pinus contorta var. latifolia Engelm.)White Spruce (Picea glauca) and alpine fir (subalpine fir) (Abies lasiocarpa var. lasiocarpa). In the higher elevations, the watersheds fall into the Engelmann spruce - subalpine fir zone (ESSF), characterized by alpine fir and Engelmann spruce (Picea engelmannii). In addition to the 9 conifers, streamside vegetation consists primarily of small alders (Alnus sp.) and Willow (Salix sp.). The Interior Plateau region of BC is located in the Maritime mountain rainshadow (Heede 1984) where most of the annual precipitation occurs as snowfall (Breschta et al. 1987). Although timing and thaw vary according to the location and elevation (Heede 1984), the greatest discharge commonly occurs in the Spring from mountain snowmelt (Breschta et al. 1987). A second discharge occurs in autumn due to an increase in rainfall. Ice scour, a relatively short growing season and low precipitation rates are habitat features that were common to interior stream systems and are absent from maritime stream systems. The soils in the Interior Plateau region are composed primarily of fine, highly erodable sediments. As previously mentioned, some watersheds in this study contain glacial till interspersed with extensive deposits of lacustrine clays (Macdonald et al. 1992), while others contain glaciolacustrine and sandy glaciofluvial deposits (Larkin et al. 1998). Previous work on the benthic macroinvertebrate community in the stream ecosystems of the interior regions of B.C. consists of four years of drift sampling (Choromanski et al. 1994) in the Stuart-Takla region and Surber and drift sampling in creeks within the Slim Creek watershed (Slaney et al. 1977). Only the investigation in the Slim Creek watershed examined the potential effects of timber harvesting on benthic macroinvertebrate community composition using surber samples and drift samples. Sedimentation studies were also done in experimental channels within the Slim Creek watershed to determine any effects on the benthic community (Slaney et al. 1977). The sites for this present study were selected from 45 potential stream sites within the Stuart-Takla, Willow and Bowron watersheds. During July and August 1995, a stream survey was done to determine accessibility and suitability of the streams in order that logged and uncut 10 streams could be paired according to a "stream pair design" (Mellina and Hinch 1995). Biogeoclimatic zone, channel width, slope and aspect were the initial characteristics to be held consistent between pairs. Variability in width and slope corresponds with variations in channel cross sections, bed materials and sediment transport (Frissel et al. 1986). The direction of the stream (aspect) in relation to the direction of the sun may also have an effect on the stream's water temperature regime. The treatment streams only included those streams with clear-cut riparian harvesting to the bank on both sides of the stream, to maximize stream exposure due to harvesting. Due to the immense environmental variability found among the study streams, the inaccessibility to several suitable streams and the requirement of finding a suitable harvesting treatment, the "stream pair design" (Mellina and Hinch 1995) proved unfeasible. Streams were then chosen which remained similar in biogeoclimatic zone and aspect, otherwise varying in other measurements. Stream treatments were then organized into "logged" or "uncut" categories. "Uncut" sites included those with no logging in the watershed either adjacent to the stream or upstream of the section of the stream sampled. "Logged" sites included those with a clear cut timber harvest of the riparian zone ranging from harvests less than 5 years old to those harvests older than 20 years, but not more than 35 years old. During preliminary data analysis, the "logged" streams were tentatively divided into those streams with riparian zones that had been cut recently (less than 5 years old) and those streams with riparian zones with older cuts (older than 20 years). The streams with the recently harvested riparian zones ("recent cut") had no canopy cover. In these riparian zones, substantial riparian re-growth had not yet been established. Those streams categorized as having older cut riparian zones ("old cut") had substantial amounts of riparian deciduous and coniferous re-growth. 11 Eighteen streams were eventually selected for intensive examination. Three of the four streams examined in this study (O'ne-eil, Gluskie and Forfar) have experienced little anthropogenic disturbance in the past (Macdonald et al. 1992). The fourth stream, Bivouac Creek has experienced some logging in the past, in the lower reaches of the stream (Macdonald et al. 1992) and a new clear-cut in the upper reaches. The Baptiste Creek watershed has also experienced a fairly recent clear-cut timber harvest. The study site "Baptiste tributary one" has been logged below the study site, while the entire watershed of the "Baptiste tributary two" study site remained entirely uncut at the time of study. Four of the streams in this study were found in the Willow River watershed and five in the neighbouring Bowron River watershed. General descriptions of the streams are summarized in Appendix 1. Eight "uncut" streams and two "recent cut" streams were in the Stuart-Takla watershed. The Willow and Bowron watersheds provided the remaining four "recent cut" and six "old cut" streams. Each stream represented one "site" with the exception often streams in which two instream sample sites were established. I consider these 24 sites as my instream sampling sites. Within each category, each stream was classified according to specific instream physical characteristics, i.e., bankfull width, gradient and riparian terrain description (Flat, Moderate or Steep) and stream classification according to the Forest Practices Code (Appendix 1.). The riparian terrain descriptions describe the quality and quantity of the material entering the stream from the riparian zone. Lake-fed streams were excluded, as a lake will influence stream water conditions. By grouping the streams in this manner, the variability of each stream's biotic and abiotic characteristics can be analyzed by stream treatment category ("uncut", "recent cut" and "old cut"). The watershed dimensions were measured using maps and satellite photographs. The watershed drainage areas of the streams and the length of the streams were measured using 12 1:50 000 scale maps. If these were not available 1:100 000 scale maps were used. Recent satellite photographs of 1:75 000 of the Stuart-Takla and 1:100 000 of the Willow-Bowron regions were also used to measure the proportion of clear-cut areas within each watershed, where applicable. 13 2.2 Field collection A l l samples were collected from mid-September to mid October in 1995. A number of physical stream measurements were recorded as part of joint effort between this study and the Ph.D. study by Eric Mel l ina (Department of Forest Sciences, University of British Columbia) on the effects of logging practices on physical and biological changes in streams of the Prince George region. Aspect was measured using a compass. Stream and hillslope gradients of each stream were estimated using a clinometer. Three measurements were taken for canopy cover per section measured, using a densiometer, following Lemmon (1957). Riparian overstory and streamside vegetation were described visually and photographed. Actual measurements o f the latter two measurements were not taken, due to time constraints. A number of instream variables and structures were also measured. A hand held conductivity meter was used to both measure total suspended ions (conductivity) and temperature at all study sites. Pools, riffles and other physical variables (bankfull width, bankfull height and wetted width) were measured with a folding ruler. The number of pieces of large organic debris ( N L O D ; at least 0.1 m in diameter and three metres in length; Ralph et al. 1994) in the channel was counted and the percentage of the area covered by L O D was estimated within each stream. A scaling factor of the number of L O D per 100 m of the stream, was used. Five replicates were taken for each stream section length of bankfull width, bankfull height and wetted width measurements. Discharge was estimated on a single point in the sampling section with the water velocity meter and a depth gauge. The number, area and ratio o f pools and riffles were measured in each stream section to estimate the amount of erosional and depositional habitats available to the aquatic macroinvertebrates within the reach. One periphyton sample was collected from a riffle at each site as an index of instream primary productivity. To accomplish this, the surface of six cobbles collected in each stream were scrubbed with a toothbrush. The residue collected was placed into sample bottles 14 containing buffered formalin. Buffered formalin can prevent the cellular chlorophyll a molecules from degrading. Breakdown of the molecule can often be found when the residue is added to the acidic environment of the unbuffered formalin. This often can interfere with the estimation of chlorophyll a concentration within a sample of the biofilm (Steinman and Lamberti 1996). The length and width of each cobble was measured to give a surface area estimate. Within each stream a Surber sampler with a 0.35 mm nitex net was used to collect manually dislodged benthic macroinvertebrates. The substrate was agitated by hand to a depth of 10 cm within the sampler's quadrat frame which was 0.09 m2. Larger rocks, i.e., cobbles and gravel, were wiped by hand to ensure removal of larger organisms. Each sample was placed into a container with 10% buffered formalin and labeled for subsequent sorting and analysis. From each site a sample was collected from each riffle, resulting in 3-6 samples of benthic macroinvertebrates per site. Other measurements were also taken from the riffle habitat of the benthic macroinvertebrates. From each riffle sampled for invertebrates, substrate sections of the riffle were used to estimate the substrate composition using the method outlined by Mellina and Rasmussen (1994). The substrate from each riffle sampled for invertebrates was removed from the streambed to the depth of 10 cm. The substrate was placed onto an tarp, divided into groups by size according to the Wentworth classification method. The substrate was then photographed with a ruler for scale for later analysis. Depth and velocity were also measured in the middle of each riffle. Velocity was measured at four percent of the stream depth from the stream bottom of each riffle to obtain a correct velocity measure. 15 2.3 Sample processing 2.3.1 Periphyton Periphyton (standing stock) biomass per riffle was measured to give a relative estimate of primary productivity of each stream. Initially, the periphyton samples were all standardized to a 1 L volume with distilled water. The sample was shaken to bring periphyton into suspension. Fifty ml samples were then filtered through an ashed and pre-weighed glass filter (pore size of 0.1 pm). Periphyton biomass (Ash-free Dry Mass or AFDM) was estimated by the method outlined by Aloi (1990). The filter was placed into an aluminum boat and weighed. The filter was then dried for 24 hours in a standardized temperature in a drying oven. The temperature was approximately 60°C to prevent the loss of volatile organic compounds. Because several of the filters could not be removed intact the boats and filters were weighed together. The filters should ideally be weighed without the boats due to potential changes in the weight of the boat after ashing. After weighing the boats, the filters were ashed at 550°C for another 24 hours; 550°C was used because the organics were not removed at lower temperatures (Aloi, 1990). Relative measurements were also made of the chlorophyll a biomass. Dudgeon (1994) suggested that after storing periphyton in buffered formalin, some of the chlorophyll a would be lost over time. Therefore, only relative measurements can be derived from this analysis. The samples were all stored in a similar fashion and therefore it can be assumed that any degradation of the chlorophyll a was consistent over all the samples. My method of processing the periphyton in this study to determine biomass is the standard method used by the Oceanography Department of the University of British Columbia in their periphyton analyses. Fifty ml of each periphyton sample was filtered, then the filter was folded in half and stored in a labeled plastic ziploc bag. The bags were then frozen for one week to fracture the periphyton cell walls (plastids). The filter was placed into a test tube with a 90% 16 acetone solution. The sample was then sonicated for 15 minutes to split the remaining plastids, then covered and stored in the refrigerator for 24 hours, to extract the remaining chlorophyll a. After 24 hours the tubes were centrifuged. The supernatant was poured into a new test tube and placed into a fluorometer for a first reading. Three drops of HC1 were then dropped into the tube and after 10 seconds, a second reading taken. The two readings were taken to determine the pigment phaeophytin already present in the supernatant and the amount actually extracted from the cells. Biomass was measured according to the corrected formula: Chi a (ugL-')= [1.974X(Fn - F 1 )xl/0.9489] x vV(L) % of total volume used where F 0 = first reading F4= second reading v= volume of solvent used V = volume filtered 2.3.2 Benthic Macroinvertebrates The insect taxonomic groups in this study were identified to genus using the taxonomic key by Merritt and Cummins (1996a). Non-insect invertebrate groups were identified to more general groupings such as Phylum Nematoda and the leeches, Class Hirudinea. The invertebrate taxa were identified, sorted, counted and put into sample bottles with isopropanol and proper labeling. Invertebrates were later sorted into functional guild groupings using the tables in Merritt and Cummins (1996a). Taxa that belonged to two functional guilds, were divided in half and each half added to the appropriate functional guild (Hawkins et al. 1982). Those taxa sorted into more than two functional guilds were placed into a "generalist" category. To obtain functional guild biomass, relative wet mass measurements were calculated. Taxa were grouped into their 17 appropriate guilds then drained onto a 350 pm nitex filter. The insects were then centrifuged for 10-15 minutes to remove excess liquid, then air dried for 15 minutes. The insects were then scraped onto a pre-weighed dish and weighed to one hundredth of a gram. 2.3.3 Substrate For an estimate of the substrate size ranges within the sampled macroinvertebrate riffle habitat, the mean size of each subgroup of substrate within each photograph was measured according to the Wentworth scale. The proportion of each subgroup within a group was then visually estimated. The proportion of the specific substrate size for each sample was then estimated. The substrate composition of each of the macroinvertebrate samples could then be estimated. It should be noted that the proportion of fine material less than 0.35 pm may be less accurate than the re-examining of substrate particles, due to potential losses incurred while sampling for macroinvertebrates or collection of the substrate sample. Sediment particles may be suspended in the water flow and carried out of the substrate sample during collection. However, fine substrate material is often cited in the literature to have potential effects on the aquatic macroinvertebrate community (e.g. Culp et al. 1983), so it was included in this study. However, I expect to find a potential loss of fines. Thus, the sample will be biased. 2.4 Data Organization Benthic macroinvertebrate density and biomass (total number and mass of individuals per m 2 of riffle habitat), functional guild density (number of individuals within each guild per m 2 of riffle habitat) and biomass (relative weight of individuals within each guild in grams per m 2 of riffle habitat) were calculated by site (some streams contained two sections). Eighty-four Surber samples of macroinvertebrates (1 season x 1-2 sites/stream x 3-6 samples per site) were collected and averaged to provide one estimate of abundance for each site, which were then analyzed. The environmental data were divided and assigned to one of four main groups. The 18 "instream productivity" dataset consisted of measurements pertaining to primary productivity within the stream, such as canopy cover, conductivity, water temperature, periphyton biomass (ash-free dry mass) and chlorophyll a biomass. The "watershed" dataset described the larger scale drainage basin of each stream (stream length, percent of stream length clear-cut, drainage area and percentage of drainage area clear-cut). The "topographic" dataset outlined the general characteristics of the stream's surrounding terrain such as elevation, hill gradient, stream gradient and maximum stream power (D). Maximum stream power was estimated from the measurement (cm) of the largest substrate particle in the stream section that had been moved into the site by the maximum stream velocity of the stream over the past year. The selected pieces of substrate had no moss and were non-angular. The presence of moss would indicate that the rock had not moved recently. Angularity of the rock would have indicated that it had been present for only a short time prior to measurement. Finally, the instream habitat dataset addressed the instream physical habitat characteristics such as bankfull width, bankfull height, wetted width, number of large organic debris, pool area, riffle area, riffle depth, riffle velocity, proportion of riffle substrate as pebble (diameter 16-64 mm) and proportion of fine material (diameter < 2 mm). The instream productivity, topographic and watershed datasets were divided by stream, because only one to three measures per stream was taken. The instream habitat variables, such as the invertebrate abundance datasets, were divided by stream section, because separate measurements were taken per stream section. The instream habitat variables were then used to describe macroinvertebrate habitat in more detail. The respective streams and stream sections were further divided by stream treatment ('uncut', 'recent cut', and 'old cut'). 2.5 Statistical Analysis The data analyses in this study were primarily exploratory in nature. Prior to analysis, 19 benthic macroinvertebrate, functional guild, instream productivity, topographic and instream habitat data were log transformed [log10(x+l)] to ensure that data being examined were normally distributed (Manly 1986). Transformation of the data also allowed for the comparison of the parameters within these datasets whose measurements differed by orders of magnitude. In some cases, when no more than two environmental sample units lacked data for a variable, the data were estimated by regression. If more samples were missing, the entire variable was removed from the dataset. All statistical analyses were performed using SAS computer software (SAS Institute Inc. 1988). A probability level of P<0.05 was used to determine significance in all statistical tests. Trends among sample sites by stream treatments were explored using Principal Component Analyses (PCA) on each dataset. PCA was used to identify sets of correlated variables measured in the respective benthic macroinvertebrate and environmental datasets that reflect the differences among sites. The sites were then described in PCA plots, as graphs have been found to be "especially valuable in exploratory data analysis" (SAS Institute Inc. 1988). Because both the benthic macroinvertebrate and environmental datasets had been log transformed to normalize the data, covariance matrices were used in the PCA to maximize the amount of variance in the transformed data (Pimentel 1979). Finally, to examine potential relationships between benthic macroinvertebrate and environmental variables, PCscores of each of the biotic datasets was correlated with the scores of each of the four environmental datasets (see example in Hinch and Collins 1993). When analyses showed strong correlations between any of the biotic and abiotic datasets, these relationships were plotted, then examined more closely through Canonical Correlation Analysis. Benthic macroinvertebrate abundance and environmental datasets were next examined using analysis of variance (ANOVA) statistics. ANOVAs were utilized to determine whether 20 benthic macroinvertebrate abundance and productivity and instream habitat and productivity variables differed among stream treatments. The reference stream sites were compared with streams draining 'recent cut' (less than 5 years old) and 'older cut' (greater than 20 years old) riparian zones. A one way ANOVA (Zar 1996) was used to test the null hypothesis that the benthic community abundance (density and biomass) and physical stream measurements (watershed, topographic, instream habitat and instream productivity) did not differ between stream treatments. Because the numbers of sites within each treatment group were not equal, the analysis was done using SAS procedure GLM (SAS Institute Inc. 1988) to compensate for the unbalanced design. Tukey's Studentized test was used to examine differences among the means of the treatments. Often, despite finding significant effects of stream treatment using univariate statistics, the way in which the groups actually differed often cannot be discerned using Tukey's test. For this reason, multivariate analyses were also used in this study. Multivariate analyses were able to determine how the three groups of stream 'treatments' differed within the respective datasets. The method used in this study, Canonical Discriminant Analysis (also known as 'Canonical variates analysis' or 'CVA'), recognized each individual stream or stream site within the respective dataset as belonging to one of the three stream 'treatments', then examined the variations between the three groups (Pimentel 1979). This method contrasted with PC A which examined all individual streams and stream sites as a single group within a population (Pimentel 1979). CVA discerned how well the stream and stream sites could be separated by the three stream 'treatments' within each dataset, and identify those variables that most strongly influenced the difference between these groups (Manly 1986). More specifically, CVA was used to determine, within each of the datasets, whether the groups of stream 'treatments' differed, which groups differed and how the stream 'treatment' 21 variables were interrelated (Pimental 1979). A M A N O V A was first used to test whether a significant difference among the three stream 'treatments' existed within the respective datasets. Although this analysis defined whether a significant difference between the stream 'treatments' existed, it could not determine how different one stream 'treatment' was from the others (Pimental 1979). Therefore, to establish which of the stream 'treatments' within the dataset differed from each other, the Mahlanobis distance was then calculated on the basis of a squared distance of the means of each variable in the dataset from the centroid of each of the three stream 'treatments' (Pimental 1979). The Mahlanobis distance demonstrated how well each stream or stream site could be allocated to the closest stream 'treatment' and whether or not it was the stream 'treatment' from which it was derived (Manly 1986). The resultant percentage of correct allocations showed how well the stream 'treatments' could be separated from each other, using the respective datasets (Manly 1986). These analyses, although able to define whether a significant difference between the stream 'treatments' existed and how different one stream 'treatment' was from the others, they could not determine how the stream 'treatment' variables were interrelated within the datasets (Pimental 1979). C V A described how the variables within the respective datasets contrasted between the three stream 'treatments' by identifying the sets of correlated variables measured within the respective benthic macroinvertebrate and environmental datasets (Pimental 1979). A linear combination of the variables within the dataset and those variables that showed the highest number of possible correlations between the groups was first extracted and labeled as the first canonical correlation. The same procedure was repeated to obtain the second and following canonical correlations (SAS Institute Inc. 1988). The correlated variables of these correlations explained the differences between the three stream treatments. Finally, a Canonical Correlation Analysis ( C A N C O R R ) was used to examine whether 22 changes in the benthic macroinvertebrate community (density and abundance) could be correlated with modifications in their environment (instream productivity and instream habitat). This technique searched for a linear combination from each of the two selected datasets that extracted the maximum correlation between these two datasets (SAS Institute Inc. 1988). Each of these two canonical variates had the maximum fitted relationship to both its own and the other dataset (SAS Institute Inc. 1988). These two covariates were then combined in the analyses to form the first canonical coefficient (Rc). The canonical coefficient was then the combination of the two canonical variates best describing the relationship, in this study, between the selected macroinvertebrate and environmental datasets. Only those canonical coefficients that proved to be significant, were interpreted. As previously mentioned, in the CANCORR the two selected canonical variates included in each canonical coefficient were the best fitting linear combination from each dataset (SAS Institute Inc. 1988). The canonical variates were also best described by specific variables from both their "own" dataset from which they were extracted and the "other" dataset, to which they had been fit. Thus, the significant canonical coefficients could be described by the macroinvertebrate and environmental variables that most strongly described both of the canonical variates. Cases where macroinvertebrate and environmental variables were strong in both canonical variates suggested some interaction between the functional guild and the environmental factor. Relationships between these two types of variables could then be observed and interpreted. In this study, the macroinvertebrate canonical variate with correlated variables were listed under canonical variate ' V , while the environmental variables and their correlated variables were listed under canonical variate 'W. Only those variables that appeared to have had the greatest influence on both canonical covariates and subsequently on the canonical coefficient, were selected. 23 It should be noted that in CANCORR, because dimensions were chosen to maximize a correlation, a spurious association between sets can occur to create a meaningless association, one perhaps not found in nature (Pimentel 1979). Thus, this technique can only suggest possible relationships between the data sets and it cannot be assumed that it was how one data set fits to another (Pimentel 1979). However, it was still useful to examine the relationships as predictors or building hypotheses for further study (Pimentel 1979). 24 3. Results 3.1 Environmental variables 3.1.1 Landscape characteristics 3.1.1.1 Watershed descriptors Trends among streams in this study were related to watershed characteristics, in particular, the stream length and size of the watershed (Table 1). Principal component analyses described 96% of the variation in the watershed data in the first two axes. The first PC axis described almost all of the variation (91%), with the second axis expressing 5%. When the streams were roughly grouped by the time since their watersheds had been clear-cut harvested (Figure 1), each cluster described the wide range of sizes, in particular with respect to watershed area and stream length. The clusters do not separate very clearly on either of the two axes and show almost equal distributions along the first axis (Figure 1). However, the cluster of the 'recent cut' sites (less than 5 years old) included only streams in the lower distribution of the streams included in the other two treatments, i.e., those with smaller watershed sizes and shorter stream lengths. I used a MANOVA to establish that I controlled for the amount of timber harvesting in the selection of the streams for this study. As expected, the 'uncut' group of streams (unlogged streams) differed significantly from the other two groups of streams (Table 2). The first canonical axis explained 95% of the variance among the three groups of streams (Figure 2; Table 3), primarily in the percentage of watershed and stream length cut. The stream groups separated 25 Table 1. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=18 variables) of the watershed dataset. PCI PC2 Watershed area 0.980 (0 0001) 0.179 (0.478) Percent watershed area cut -0.204 (0 417) 0.591 (0.010) Stream length 0.975 (0 0001) -0.191 (0.448) Percent stream length cut -0.101 (0 691) 0.491 (0.038) (% explained variance) 91 PC3 PC4 Watershed area -0.089 (0.724) -0.005 (0.983) Percent watershed area cut 0.665 (0.003) 0.408 (0.093) Stream length 0.111 (0.662) 0.008 (0.975) Percent stream length cut 0.797 (0.0001) -0.336 (0.172) (% explained variance) 3 0.6 26 Figure 1. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the watershed dataset. Polygons delineate streams belonging to 'uncut', 'recent cut' and 'old cut' treatments. Percent watershed cut Percent stream length cut CM O -3 0. 2 George trib Bivouac U A - * ; Gluskie2 _ 1 Baptiste2 + Baptiste1+ ' + U N C U T • R E C E N T CUT x O L D CUT Thursday " * - _ Forfar. . . George • . 18 mife~" " -,~x — r > ^ ' • y e - - - x ' -2-Kyflocl1 h l y yUenflel PC1 (91%) Watershed area Stream length 27 Table 2. F-value approximations and associated P-values for MANOVA, contrasting three groups ('U'- 'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on watershed, topography, instream habitat and instream chemical/productivity datasets. Differences among groups were assessed by canonical variates analysis. Those groups whose centroids differ are indicated by a separate letter and groups with similar centroids have their letters attached. Differences DATASET DF F P among groups Watershed 15 4.366 0.002 U O-R Topographic 15 1.754 0.137 Instream habitat 24 1.954 0.047 U-R R-0 Instream chemical/productivity 15 3.005 0.015 R U-0 28 Figure 2. Plot of Canonical Variate 2 (CAN2) versus Canonical Variate 1 (CAN1) for the watershed dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Stream length in CM z < o Watershed area + .'-3 --2 • ^ 1 -2 -3 -4-+UNCUT • RECENT CUT xOLD CUT \ / \ / \ 1 « 2 -v-3 * C A N 1 (95%) Percent watershed cut Percent stream length cut 29 Table 3. The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis based on the watershed variables of 18 streams. CAN1 CAN2 Watershed area 0.356 (-)0.630 Percent watershed area cut 1.415 (-)0.581 Stream length 0.656 1.379 Percent stream length cut 0.774 0.395 % variance explained 94.86 0.05 30 along this axis only. The 'uncut' streams were clustered along the lower end of the axis. These streams were found higher along the axis (Figure 2) and significantly different from the next group of 'recent cut' streams (p=0.04), and the 'old cut' streams (greater than 20 years old; MANOVA, p=0.0004). The 'old cut' streams aggregated the farthest along the first axis and did not differ in the percentage of watershed and stream length cut from the 'recent cut' streams (MANOVA, p=0.164). These results showed that the 'uncut' streams have the lowest percentage of stream length and watersheds cut. The percentage of the stream lengths and watersheds cut do not appear to differ between the 'old cut' and 'recent cut' streams. Univariate ANOVA also determined that there were treatment differences between the watershed variables of the 'percentage of watershed cut' and 'percentage of stream length cut'fTable 4). The multiple comparison test between means (the Tukey's Studentized Range or 'Tukey's test'), again, revealed that the streams in the 'recent cut' and 'old cut' treatments with respect to percentage of watershed cut, did not differ from each other. However, both groups of streams differed from the 'uncut' streams. The Tukey test of the 'percentage of stream length cut' measurement found that although streams in the 'uncut' and 'old cut' treatments differ, neither are significantly different from streams in the 'recent cut' treatments. 31 Table 4. ANOVA results contrasting three groups ('U'- 'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the watershed dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached. Differences Parameter Source df MS r2 F P among groups Watershed area Group 2 3.71 0.160 1.43 0.271 O-U-R Error 15 19.51 Percent watershed cut Group 2 0.59 0.604 11.47 0.001 O-R U Error 15 0.39 Stream length Group 2 4.01 0.192 1.78 0.203 O-U-R Error 15 16.90 Percent stream Group 2 0.56 0.511 7.84 0.005 O-R R-U length cut Error 15 0.54 32 3.1.1.2 Topography descriptors Trends among the streams were related to topographical characteristics, in particular, the hill and stream gradients and to a lesser extent stream power (D). PCA described 73% of the variation in the first axis, attributed primarily to an increase in hill gradient, stream gradient and stream power (Table 5). The second axis (18% of variation) described a trend of increasing stream power, stream gradient and elevation. The streams in the topographic dataset, when clustered by the time since their watersheds had been harvested, showed a lack of distinctiveness between clusters, as with the watershed dataset (Figure 3). MANOVA (Table 2) found that the streams, grouped by treatment could not be distinguished from each other. Univariate ANOVAs on the individual variables found that the only crucial difference among treatments was in 'channel gradient'. Tukey's tests showed that 'uncut' and 'old cut' treatments differed, but neither significantly varied from 'recent cut' streams (Table 6; Figure 4). 3.1.2 Physical habitat characteristics 3.1.2.1 Instream A PCA on the Instream habitat dataset revealed that most instream variables were highly, positively correlated with each other (Table 7). Bankfull widths, wetted widths, pool areas, riffle areas, and riffle depths were all positively correlated with each other (Pearson's, p<0.021 for each variable). These variables were weakly, negatively correlated with the 'percent of fines'. Riffle velocity was positively correlated with wetted width, number of large organic debris pieces (NLOD) and riffle depth (Pearson's, p<0.021 for each variable) and had a negative correlation with the percent of fines. 33 Table 5. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=18 variables) of the topographic dataset. PCI PC2 Stream power 0.444 (0.065) 0.847 (0.0001) Elevation -0.400 (0.100) 0.593 (0.009) Hill gradient 0.994 (0.0001) -0.065 (0.799) Stream channel gradient 0.546 (0.019) 0.649 (0.004) Explained variance (%) 73 18 PC3 PC4 Stream power -0.287 (0.248) -0.060 (0.814) Elevation 0.698 (0.001) 0.014 (0.958) Hill gradient 0.090 (0.723) -0.002 (0.995) Stream channel gradient -0.344 (0.162) 0.403 (0.098) Explained variance (%) 8 0.5 34 Figure 3. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the topography dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Stream power Stream gradient Elevation ~ -2.5 CM O CL 1.5 s . s *3rg s +._ - -"Bpf1" 0.5 -0-2 -1.5 x * Wendel G r 9 t -1 +JA"-Q.5 V 0 -0.5 -1 + -1.5 -2 + -2T5 + UNCUT • R E C E N T CUT x OLD CUT . . - - •TtTu'rs^ . PF U B ? JBRP GIs -\ r 0.5» ll Bv ( s s s » .5 \ 2 18M Gls2 s s ' UD PC1 (73%) Hill gradient Stream gradient Stream power 35 Table 6. F-value approximations and associated P-values for ANOVA, contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the topography dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached. Differences Parameter Source df MS r 2 F P among grouj Stream power Group 2 0.77 0.201 1.88 0.186 U-R-0 Error 15 0.41 Elevation Group 2 0.005 0.001 0.01 0.993 R-O-U Error 15 0.376 Hill gradient Group 2 1.445 0.092 0.76 0.484 U-R-0 Error 15 1.9 Channel gradient Group 2 0.265 0.405 5.1 0.020 U-R R-0 Error 15 0.051 36 Figure 4. Histograms (± 1 standard error) of ln transformed percent channel gradient variables in 'uncut', 'recent cut' and 'old cut' streams. Table 7. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=27 variables) of the instream habitat dataset. PCI PC2 Bankfull width 0.920 (0.0001) 0.072 (0.721) Bankfull height 0.705 (0.0001) -0.228 (0.252) Wetted width 0.880 (0.0001) 0.258 (0.193) NLOD -0.050 (0.804) 0.981 (0.0001) Pool area 0.898 (0.0001) 0.045 (0.824) Riffle area 0.805 (0.0001) -0.1841 (0.358) Riffle depth 0.549 (0.003) 0.255 (0.200) Riffle velocity 0.408 (0.035) 0.460 (0.016) Percent pebbles 0.207 (0.299) -0.1297 (0.519) Percent fines -0.236 (0.236) -0.1189 (0.555) Explained variance (%) 57 24 PC3 PC4 Bankfull width -0.143 (0.477) 0.205 (0.306) Bankfull height -0.249 (0.150) 0.161 (0.422) Wetted width -0.218 (0.274) 0.141 (0.484) NLOD 0.170 (0.398) -0.075 (0.712) Pool area -0.297 (0.133) -0.300 (0.128) Riffle area 0.563 (0.002) -0.000 (0.100) Riffle depth -0.211 (0.290) 0.705 (0.0001) Riffle velocity 0.150 (0.457) 0.156 (0.438) Percent pebbles -0.003 (0.987) -0.286 (0.148) Percent fines 0.239 (0.230) -0.211 (0.291) Explained variance (%) 11 , 7 38 The first two axes of the PCA explained 78% of the total variation in the instream dataset (Figure 5). The first axis (54% of total variation) organized the streams by increasing stream size variables. The second axis explained 24% of the variation in the data with a positive correlation between the number of pieces of LOD per 100 m and riffle velocity. The 'uncut' and 'old cut' groups contained streams with a wide range in sizes, riffle depths and riffle velocities (Figure 5). These two sets of streams overlap, somewhat, with each other, although the 'old cut' streams vary particularly widely with respect to size, riffle depth and riffle velocity. In contrast, the 'recent cut' streams were of a smaller size. Note that George tributary ('Grgt'), the smallest stream in the 'old cut' streams cluster, was a tributary of the largest stream, George Creek ('Grg'; Figure 5). In general, these results imply that timber harvesting in this region in the five years before the study, was done in the riparian zones of the smallest sized streams. These streams appeared to have had lower channel gradients than found in the larger, but steeper gradient 'uncut' streams. A MANOVA on the streams grouped by time since logging revealed notable differences (Table 2). The first canonical variate axis (66%) separated groups by bankfull width and bankfull height (Figure 6; Table 8). The group of 'uncut' streams were found to be associated with the highest bankfull heights but low bankfull widths. In contrast, the group of 'old cut' streams differed from 'recent cut' streams (MANOVA, p=0.044) and were characterized by lower bankfull heights and greater bankfull widths. The 'recent cut' streams, clustered in between these two groups, but did not significantly differ from either the group of 'uncut' streams (MANOVA, p=0.195) or the 'old cut' streams (MANOVA, p=0.194). 39 Figure 5. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the instream habitat dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Number of pieces of LOD Riffle velocity CM o CL Gls2 1 JJD~ TUB I '/ GIs UA + / UA/ + • -2 l,Grgt \ \ -1 • -Bv \ / \ ,' Bpt2 \ W Bv \ \ B p M ^ • Bpt1 + lBpt2 UC Fly X ' ' F F GIs Wendel I F F / / / •,?8M ' *Thurs ^ d-"t)n ^ On PC1 (54%) • - -x Grg +UNCUT • RECENT CUT X OLD CUT Bankfull width Bankfull height Wetted width Pool area Riffle area Riffle depth Riffle velocity 40 Figure 6. Plot of Canonical Variate 2 (CAN2) versus Canonical Variate 1 (CAN1) for the instream habitat dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Bankfull height Wetted width Percent fines Riffle area CO CN I* +-+ + 1 +. + / * -T x JtL 1 \ 2 ; 3 / \ \ \ \ \ + UNCUT • RECENT CUT X OLD CUT Bankfull height CAN1 (66%) Bankfull width 41 Table 8. The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis based on the instream habitat variables of 27 stream sites. CAN! CAN2 Bankfull width 2.542 -0.421 Bankfull height -1.968 0.884 Wetted width -1.064 0.707 Number of LOD -1.035 0.320 Pool area 0.524 -0.037 Riffle area -0.758 0.627 Riffle depth 0.358 -0.299 Riffle velocity 0.953 -0.150 Percent pebbles 0.420 0.147 Percent fines 0.713 0.655 Explained variance (%) 66 34 42 The ANOVAs demonstrated several important differences among treatments with regard to bankfull width, bankfull height, pool area, riffle area, riffle to pool area ratio and riffle velocity (Table 9). The groups of 'recent cut' and 'old cut' streams differed most consistently in the ANOVAs. The group of 'old cut' streams had, in general, much greater bankfull widths, pool areas and riffle areas than the 'recent cut' streams. The group of 'uncut' streams did not differ from either of the other two treatments. Bankfull height seemed to be higher in the group of 'uncut' streams compared to the group of 'recent cut' streams, neither group differing from the groups of 'old cut' streams (Figure 7). Riffle velocity was higher in the groups of 'old cut' streams compared to the 'uncut' streams. Riffle to pool area ratio did not differ between the streams of the three different treatments. The distribution of the stream sites on the PCA graph of the 'instream' data set also demonstrated the high within-stream variability in the physical characteristics of the streams sampled in this study. In many cases, the site pairs that came from a single stream and were in close geographic proximity [e.g. some that were sampled next to each other such as Bivouac ("Bv"), Unnamed A ("UA"), Unnamed B ("UB"), Gluskie ("Gls"), Forfar ("FF") and Fly creeks] were not spatially proximate on the PCA graphs (Figure 5). Therefore, sites within a given stream were physically different enough to be considered separate sites for purposes of further analysis (i.e., I will use them as independent observations). 43 Table 9. F-value approximations and associated P-values for ANOVA, contrasting three groups ('U'-'uncut' streams, 'R - 'recent cut' streams and 'O'- 'old cut' streams) based on the instream habitat dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached. Parameter Source df M S Differences among groups Bankfull width Group 2 0.875 0.209 3.18 0.060 O-U U-R Error 24 0.275 Bankfull height Group 2 0.110 0.262 4.26 0.026 Error 24 0.026 U-O O-R Wetted width Group 2 0.600 0.187 2.76 0.083 Error 24 0.218 U-R-0 NLOD Group 2 0.070 0.008 0.09 0.911 Error 24 0.722 U-R-0 Pool Area Group 2 2.220 0.236 3.71 0.039 Error 24 0.598 O-U U-R Riffle Area Group 2 2.445 0.264 4.3 0.025 Error 24 0.000 O-U U-R Riffle:Pool Group 2 0.070 0.009 0.11 0.899 Error 23 0.650 U-R-0 Riffle depth Group 2 0.195 0.060 0.76 0.479 Error 24 0.255 U-R-0 Riffle velocity Group 2 0.040 0.276 4.58 0.021 Error 24 0.009 O-R R-U Percent pebbles Group 2 0.001 0.008 0.09 0.910 Error 24 0.010 U-R-0 Percent fines Group 2 0.010 0.090 1.19 0.322 Error 24 0.009 U-R-0 44 Figure 7. Histograms ( ± 1 standard error) of In transformed instream habitat variables in 'uncut', 'recent cut' and 'old cut' streams. C H A N N E L M E A S U R E M E N T S Bankfull width Bankfull height Wetted Width I N S T R E A M P H Y S I C A L H A B I T A T Pool area Riffle area I N S T R E A M V A R I A B L E S 0 UNCUT (n=13) • R E C E N T (n=6) • OLD(n=8) Riffle velocity 45 3.1.2.2 Instream chemistry/productivity descriptors Chlorophyll a was positively correlated with conductivity (Pearson's p=0.007), water temperature (Pearson's p=0.0504), and periphyton biomass ('AFDM'; Pearson's p=0.007). Periphyton biomass may have been weakly influenced by temperature (Pearson's p=0.062), albeit not significantly. The complete dataset for these parameters can be seen in Appendix 2. Close to 100% of the variance in these data was explained by the first two axes of the PCA (Table 10). The first axis (68%) explained the variation between sites in terms of canopy cover, water conductivity, water temperature, and chlorophyll a biomass. The 'uncut' streams had slightly greater canopy cover, conductivity, temperature and chlorophyll a biomass than the 'old cut' streams. The second axis (31%) described variation in canopy cover and conductivity, and seemed to separate the 'recent cut' streams from the remaining two groups. Along this axis, the 'recent cut' streams were associated with lower canopy cover and greater stream conductivity (Figure 8) than found in the other two groups. A MANOVA revealed differences in time-since logging treatment (Table 2). The treatments were positioned along the first canonical axis (explained 87% of the variance) which attributed differences among treatments for variation in canopy cover, periphyton biomass and chlorophyll a biomass (Table 11). The 'recent cut' streams had with the greatest chlorophyll a biomass and the least canopy cover (Figure 9). This group was distinctly separated from the next cluster, the 'old cut' streams (MANOVA, p=0.015) and the farthest cluster, the 'uncut' streams (MANOVA, p=0.0069). The 'uncut' streams and the 'old cut' streams did not differ (MANOVA, p=0.431). ANOVAs demonstrated that of the five variables in this dataset, 'canopy cover' was the only one that indicated a difference between treatments (Table 12; Figure 10). Tukey's test showed that the amount of canopy cover differed between the 'recent cut' and both the 'uncut' 46 and 'old cut' streams. However, the 'uncut' and 'old cut' streams did not differ from each other with respect to canopy cover. 47 Table 10. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=18 variables) of the instream chemical/productivity dataset. PCI PC2 Canopy cover 0.769 (0.0001) 0.639 (0.004) Conductivity 0.876 (0.0001) -0.481 (0.043) Temperature 0.581 (0.012) -0.267 (0.285) Periphyton biomass 0.355 (0.148) -0.093 (0.713) Chlorophyll a biomass 0.468 (0.050) -0.420 (0.083) Explained variance (%) 68 31 PC3 PC4 Canopy cover -0.002 (0.995) -0.000 (0.999) Conductivity -0.018 (0.944) -0.000 (0.999) Temperature 0.768 (0.0001) -0.044 (0.863) Periphyton biomass 0.332 (0.178) 0.869 (0.0001) Chlorophyll a biomass 0.134 (0.598) 0.421 (0.082) Explained variance (%) 1 0.2 48 Figure 8. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the instream chemical/productivity dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Canopy cover a- -3 2-CM o fi-St ream conductivity 4 --2 -\- -Wendel UD UE :1_ fcrgt + U N C U T • R E C E N T CUT x O L D CUT F FUs2\ . UA+ \ On Thurs B p t l ^ 2 UB -3-P C 1 (68%) Canopy cover, Stream conductivity Temperature Chlorophyll a biomass 49 Table 11. The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis based on the instream chemical/productivity variables of 18 stream sites. CAN! CAN2 Canopy cover 1.533 0.281 Conductivity 0.063 0.944 Temperature -0.464 -0.453 Periphyton biomass 0.885 -0.245 Chlorophyll a biomass -1.175 0.529 Explained variance (%) 87 13 50 Figure 9. Plot of Canonical Variate 2 (CAN2) versus Canonical Variate 1 (CAN1) for the instream chemical/productivity dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Conductivity co CM O \ \ \ 1 - r -3 / -2 \ \ \ / \ / V -1 * — + UNCUT • RECENT CUT x OLD CUT 0 2 3 Chlorophyll a biomass CAN1 (87%) Canopy cover 51 Table 12. F-value approximations and associated P-values for ANOVA, contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the instream chemical/productivity dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached. Differences Parameter Source df MS r2 F P among groups Canopy cover Group 2 3.210 0.553 9.26 0.002 U-0 R Error 15 0.347 Conductivity Group 2 1.615 0.242 2.39 0.126 U-R-0 Error 15 0.675 Temperature Group 2 0.005 0.035 0.27 0.767 R-U-0 Error 15 0.031 Periphyton biomass Group 2 0.001 0.037 0.29 0.754 U-R-0 Error 15 0.003 Chlorophylls biomass Group 2 0.025 0.238 2.35 0.130 R-U-0 Error 15 0.011 52 Figure 10. Histograms (± 1 standard error) of In transformed percentage of canopy cover variables in 'uncut', 'recent cut' and 'old cut' streams. 4.5 T 3.5 4-> U N C U T R E C E N T C U T O L D C U T (n=7) (n=4) (n=7) STREAM TREATMENT 53 3.2 Macroinvertebrate variables ANOVAs on total benthic macroinvertebrate density and biomass indicated a difference between stream treatments (p=0.04; Table 13; Figures 11 and 12). However, Tukey's test did not demonstrate differences between the treatments. Figure 11 illustrated a difference between densities within the 'recent cut' and 'old cut' streams, neither differing from the 'uncut' streams. In contrast, the total macroinvertebrate biomass measurements were found to be considerably higher in the 'recent cut' streams, than from the 'uncut' and 'old cut' streams. Total macroinvertebrate biomass within the 'uncut' and 'old cut' streams did not differ. 3.2.1 Macroinvertebrate functional guild density Density was strongly, positively correlated among all functional guilds (Pearson's p<0.006 for each correlation; r=0.517 to 0.891). The one exception was the fish parasite. The PCA revealed one main trend in functional guild density. The first axis (explained 71% of the variation) ordered the streams by increasing density of collector-gatherers, scrapers, predators, shredder-detritivores and invertebrate parasites (Table 14). The 'old cut' streams tended to cluster in the lower end of the first axis (i.e., lower densities in collector-gatherers, scrapers, predators, shredder-detritivores and invertebrate parasites) (Figure 13). The 'recent cut' streams tended to cluster at the opposite end of the first axis, indicating higher densities of the respective guilds. The 'uncut' streams tended to be clustered intermediately. Among stream variation in fish parasite, invertebrate parasite and collector-filterer guilds was responsible for the second principal component (13% explained variance; Table 14). There were no distinct separations of the stream clusters along the second axis (Figure 13). 54 Table 13. F-value approximations and associated P-values for ANOVAs contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on total benthic macroinvertebrate biomass and density datasets. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by Tukey's Studentized Range Test, whose means differ are indicated by a separate letter. Those groups with similar means have their letters attached. Differences Parameter Source df MS r2 F P value among groups Density Group 2 2.761 0.227 3.53 0.045 R-U U-0 Error 24 0.783 Biomass Group 2 0.567 0.441 9.46 0.001 R U-0 Error 24 0.060 55 Figure 11. Histograms (± 1 standard error) of In transformed total taxa density variables in the riffle habitat of the 'uncut', 'recent cut' and 'old cut' streams. 10 Uncut Recent cut Old cut (n=13) (n=6) (n=8) Stream treatment 56 Figure 12. Histograms (± 1 standard error) of ln transformed total taxa biomass variables in the riffle habitat of the 'uncut', 'recent cut' and 'old cut' streams. 1.4 1.2 H 1 CM E Uncut Recent cut Old cut (n=13) (n=6) (n=8) S t r e a m t rea tmen t 57 Table 14. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=27 variables) of the functional guild density dataset. Collector-gatherers Collector-filterers Scrapers Predators Shredder-detritivores Invertebrate parasites Fish parasites Explained variance (%) Collector-gatherers Collector-filterers Scrapers Predators Shredder-detritivores Invertebrate parasites Fish parasites Explained variance (%) P C I 0.970 (0.0001) 0.882 (0.0001) 0.898 (0.0001) 0.887 (0.0001) 0.747 (0.0001) 0.822 (0.0001) 0.322 (0.102) 71 PC3 0.024 (0.905) -0.266 (0.180) 0.163 (0.416) 0.331 (0.091) 0.441 (0.022) -0.105 (0.601) -0.324 (0.010) 7 PC2 0.035 (0.862) -0.385 (0.047) 0.234 (0.240) -0.010 (0.961) -0.164 (0.414) 0.455 (0.017) 0.744 (0.0001) 13 PC4 -0.052 (0.798) 0.016 (0.936) -0.276 (0.164) -0.181 (0.366) 0.461 (0.016) 0.103 (0.610) 0.175 (0.382) 4 58 Figure 13. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for functional guild density. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. F i s h pa ras i t es Inver tebrate pa ras i t es Co l lec to r - f i l t e re rs UC*"v - -Bpt1 / 2/ --fc<LE Fly C M O -10 Q. -8 -6 G I s ^ ' +Gls . :4 * +. - 2 F l y X _ 0 W e n d e L " i C " " G r g t FF G r g :+-z --6 + U N C U T • R E C E N T C U T X O L D C U T / • " " r & N / B v U A + • -1 J M \ BP t 1 '^.r , - - + x ^ U D 2 4 . • B p t 2 FF+ ^ J ? . . . . . 4 •4-, B p t 2 PC1 (71%) C o l l e c t o r - g a t h e r e r s Co l lec to r - f i l t e re rs S c r a p e r s P r e d a t o r s Sh redde r -de t r i t i vo res Inver tebra te pa ras i t es 59 MANOVA revealed that functional guilds varied among logging treatments (Table 15). The first canonical axis (the only one significantly different from zero) indicated that the 'old cut' streams contained high scraper densities and low densities of predators and invertebrate parasites compared to the 'recent cut' stream group (Table 16; Figure 14). These two groups of streams differed significantly (MANOVA, p=0.0102). The 'uncut' streams lay between the two timber harvesting treatment clusters and only weakly differed (MANOVA, p=0.064) from the 'recent cut' streams and did not differ from the 'old cut' streams (MANOVA, p=0.368). Although the two groups are not statistically different, the plots indicate that the 'old cut' streams are found in the most extreme lower end of the range of the group of 'uncut' streams. ANOVAs revealed that the collector-gatherer, collector-filterer and scraper guild densities differed among treatments (Table 17). Tukey's test, however, was not able to find differences among specific treatments. Graphically, the collector-gatherer and scraper guilds appeared to have lower densities in the 'old cut' streams (Figure 15). Predator density was similar in the 'uncut' and the 'recent cut', but was much lower in the 'old cut' streams. Densities of invertebrate parasites differed among treatments, being highest in 'recent cut' streams, and lowest in 'old cut' and 'uncut' streams, which were not different from each other. Densities of fish parasites also differed significantly over time among treatments. Their densities were higher in the 'recent cut' streams, compared to the 'old cut' streams. However, neither the 'recent cut' and 'old cut' streams differed from the 'uncut' streams (Table 17 and Figure 15). 60 Table 15. F-value approximations and associated P-values for MANOVA, contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on functional guild density and biomass datasets. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Those groups whose centroids differ are indicated by a separate letter and groups with similar centroids have their letters attached. Differences Parameter df F P among groups Functional guild density (no./m2 riffle) 24 2.021 0.045 U-0 R Functional guild biomass (g/m2 riffle) 24 2.868 0.006 U-0 R 61 Table 16. The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis based on the functional guild density variables (n= 27 sites). CAN! CAN2 Collector-gatherers -0.604 -1.597 Collector-filterers 0.240 1.090 Scrapers -1.435 0.954 Predators 1.905 0.292 Shredder-detritivores -0.752 0.637 Fish parasites 0.569 -0.061 Invertebrate parasites 1.546 -0.443 Explained variance (%) 85 15 62 Figure 14. Plot of Canonical Variate 2 (CAN2) versus Canonical Variate 1 (CAN1) for the functional guild density dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Collector-filterers Scrapers CM < -3 O Collector-gatherers 1 / > . •2. * -1 x 0 \ 0 . - V + UNCUT • RECENT CUT xOLD CUT • * • / +, ' ' / 1 / 4 I l v / Scrapers C A M (85%) Predators Invertebrate parasites 63 Table 17. F-value approximations and associated P-values for ANOVA, contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the functional guild density dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached. Parameter Source df MS Differences among groups Collector-gatherer Group Error 2 3.335 0.219 3.37 0.051 24 0.990 U -R-0 Collector-filterer Group Error 2 7.300 0.196 2.92 0.073 24 2.498 U -R-0 Scraper Group 2 3.725 0.219 3.37 0.051 Error 24 1.105 U -R-0 Predator Group 2 4.365 0.288 4.84 0.017 Error 24 0.902 R - U O Shredder-detritivore Group Error 2 1.675 0.128 1.75 0.195 24 0.955 U -R-0 Invertebrate parasite Group Error 2 7.920 0.376 7.24 0.004 24 1.095 R U - 0 Fish parasite Group 2 2.744 0.267 4.38 0.024 U - R U - 0 Error 24 0.626 64 Figure 15. Histograms ( ± 1 standard error) of ln transformed functional guild density of collector-gatherers, scrapers, predators, invertebrate parasites and fish parasites in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets. Collector-gatherers (p=0.0512) Scrapers (p=0.0513) Predators (p=0.0171) Invertebrate parasites (p=0.0035) Fish parasites (p=0.0239) 0 Untagged (n=13) • Recent cut (n=6) • Old cut (n=8) 10 In density (m'2) 65 3.2.2 Macroinvertebrate taxa density The densities of 20 taxa out of the 75 total, appeared to show significant differences between treatments (ANOVA's, Table 18). Nine taxa had higher densities in the 'recent cut' streams and lower densities in the 'uncut' and 'old cut' streams. The taxa included the collectors, Dixa sp., Simulium sp., Pelecypoda; the collector-gatherer/scrapers, Farula sp., Ecclisomyia sp., Orthocladiinae; the predator, Chelifera sp.; the fish parasite, Hirudinea; and the invertebrate parasite, Nematoda (Figures 16, 17, and 18). Densities of the predator Drunella grandis were also highest in the 'recent cut' streams and lowest in the 'uncut' streams, but densities in the 'old cut' streams intermediate to both stream treatments. Five taxa had similar densities in both the 'uncut' and 'recent cut' streams, but densities were notably lower in the 'old cut' streams. These taxa included the collector, Tanytarsini; the scrapers Drunella sp. and Glossosoma sp.; the predator, Rhyacophila sp.; and the invertebrate parasite/predator Acarina (Figures 16,17, 18, and 19). The density of the remaining six taxa varied. Density of the predator, Oreogeton sp. was lower in the 'recent cut' and 'old cut' streams compared to the 'uncut' streams. In contrast, the collector Oligochaeta, was found to be in greater densities in the 'recent cut' and 'old cut' streams compared to the 'uncut' streams (Figures 16 and 17). The densities of the remaining three groups of taxa, could not be clearly separated by stream treatment, although there were some notable differences. The density of the shredder-detritivore Eucapnopsis sp., was seen to be the highest in the 'uncut' streams and lowest in the 'recent cut' streams, but densities in the 'old cut' streams were similar to both groups of streams (Figure 20). Another shredder-detritivore Despaxia sp. had much higher densities also in the 'uncut' streams, but were the lowest in the 'old cut' streams. Density of this taxa in the 'recent 66 Table 18. F-value approximations and associated P-values for ANOVA, contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the benthic macroinvertebrate taxa density dataset. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Differences among groups were assessed by the Tukey's Studentized test. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached. Functional Sub-group Guild Taxon Order Source df MS r 2 F P value Differences among groups Collector gatherer Dixa sp. Diptera Group Error 2 24 3.3 5.9 0.362 6.85 0.004 R U-0 gatherer Oligochaeta Group Error 2 24 22.6 19.0 0.543 14.25 0.0001 R-0 U gatherer/filterer filterer Tanytarsini Simulium sp. Diptera Diptera Group Error Group Error 2 24 2 24 30.2 63.9 34.5 50.7 0.318 0.337 5.67 8.17 0.010 0.002 U-R O R U-0 filterer Pelecypoda Group Error 2 24 8.7 13.1 0.361 7.94 0.002 R O-U Collector-gatherer/scraper Caudatella sp. Ephemeroptera Group Error 2 24 29.6 52.4 0.307 6.79 0.005 O-R R-U Farula sp. Trichoptera Group Error 2 24 22.6 51.0 0.253 5.33 0.012 R U-0 Ecclisomyia sp. Trichoptera Group Error 2 24 15.1 44.4 0.339 4.07 0.030 R U-0 Orthocladiinae Diptera Group Error 2 24 17.9 34.9 0.321 6.14 0.007 R U-0 Scraper Drunella sp. Ephemeroptera Group Error 2 24 22.7 48.8 0.404 5.59 0.010 R-U O Glossosoma sp. Trichoptera Group Error 2 24 39.9 78.6 0.398 6.09 0.007 R-U O Shredder detritivore Despaxia sp. Plecoptera Group Error 2 24 23.2 43.4 0.381 6.4 0.006 U-R R-0 detritivore Eucapnopsis sp Plecoptera Group Error 2 24 13.8 51 0.213 3.25 0.056 U-0 O-R Predator Drunella grandis Ephemeroptera Group Error 2 24 27.6 105.2 0.208 3.15 0.061 R-0 O-U Rhyacophila sp. Trichoptera Group Error 2 24 20.6 29.4 0.412 8.39 0.002 U-R 0 Oreogeton sp. Diptera Group Error 2 24 16.6 33.1 0.333 6 0.008 U R-0 Chelifera sp. Diptera Group Error 2 • 24 19.3 29.0 0.399 7.97 0.002 R U-0 Parasite fish Hirudinea Group Error 2 24 25.2 51.9 0.327 5.82 0.009 R U-0 invertebrate Nematoda Group Error 2 24 17.2 47.0 0.268 4.4 0.024 R U-0 Parasite/ invertebrate predator Acarina Group Error 2 24 17.9 33.9 0.345 6.32 0.006 R-U O 67 Figure 16. Histograms (± 1 standard error) of In transformed density of taxa within the collector guild, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets. Orthocladiinae (Collector-gatherer/ scraper) (p= 0.007) Ecclisomyia sp. (Collector-gatherer/ scraper) (p= 0.03) Farula sp. (Collector-gatherer/ scraper) (p= 0.012) Caudatella sp. (Collector-gatherer/ scraper) (p= 0.005) Tanytarsini (Collector-gatherer/ filterer) (p= 0.010) Pelecypoda (Collector-filterer) (p= 0.002) Simulium sp. (Collector-filterer) (p= 0.002) Oligochaeta (Collector-gatherer) (p= 0.0001) Dixa sp. (Collector-gatherer) (p= 0.004) 0.5 t-M M-l HUnlogged (n = 13) • Recent cut (n = 6) • Old cut (n = 8) 1 1.5 In Abundance (m 2) 2.5 68 Figure 17. Histograms (± 1 standard error) of ln transformed density of taxa within the predati guild, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets. Chelifera sp. (p= 0.002) Oreogeton sp (p= 0.008) Rhyacophila sp. (p= 0.002) Drunella grandis (p= 0.061) HUnlogged (n=13) • Recent cut (n=6) • Old cut (n=8) 0.5 1.5 In Abundance (m ) 69 Figure 18. Histograms ( ± 1 standard error) of ln transformed density of taxa within the parasite guild, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets. Hirudinea (fish) (p= 0.009) Acarina (predator/ invertebrate) (p= 0.006) Nematoda (invertebrate) (p= 0.024) m m HUnlogged (n=13) • Recent cut (n=6) • Old cut (n=8) 0.5 1 1.5 In Abundance (m ) 70 gure 19. Histograms ( ± 1 standard error) of In transformed density of taxa within the scraper guilds, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets. Glossosoma sp. (p= 0.007) Drunella sp. (p= 0.010) 0.5 1 1.5 HUnlogged (n=13) • Recent cut (n=6) • Old cut (n=8) 2.5 In abundance (m") 71 Figure 20. Histograms (± 1 standard error) of In transformed density of taxa within the shredder-detritivore guild, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets. Eucapnopsis sp. (p= 0.056) Despaxia sp. (p= 0.006) EUnlogged (n=13) • Recent cut (n=6) • Old cut (n=8) 0.5 1 1.5 In A b u n d a n c e (m~2) 72 cut' streams was similar to both stream treatments. Densities of the collector-gatherer/scraper Caudatella sp. were highest in the 'old cut' streams and lowest in the 'uncut' streams. The density of this taxon in the 'recent cut' streams was similar to both groups of streams (Figure 16). 3.2.3 Macroinvertebrate functional guild biomass Pearson correlations demonstrated that the biomass of functional guilds was less closely correlated with each other than was the density of other guilds. Collector-gatherer biomass was positively associated with scraper (Pearson's p=0.0001; r=0.899), predator (Pearson's p=0.0001; r=0.686) and fish parasite (Pearson's p=0.001; r=0.623) biomass. Scrapers were positively correlated with fish parasites (Pearson's p=0.0001; r=0.763), and predators (Pearson's p=0.012; r=0.476). Measurements of fish parasite biomass were found to be positively correlated with the invertebrate parasite biomass only (Pearson's p=0.0397; r=0.763). Weak positive relationships appeared between the scrapers and invertebrate parasites (Pearson's p=0.061; r=0.366) and the predators and shredder-detritivore biomass (Pearson's p=0.065; r=0.360). The first component of a PCA axis on functional guild biomass explained 70% of the variation and ordered the stream sites by increasing biomass of the collector-gatherer, scraper, predator and fish parasite guilds (Table 19). The second component (18%) ordered the stream sites by the increasing biomass of scraper and fish parasite guilds. Scraper biomass along this axis includes only those scrapers associated with collector-gatherer biomass, while those along the second axis included only those identified to be solely in the scraper guild. The groups clustered out solely on the first axis (Figure 21). The 'recent cut' stream sites were clustered slightly higher on the first axis than the other two groups of streams and had the greatest range in biomass. This suggested that these streams had both the greatest variation and highest biomass with regard to collector-gatherers, scrapers, predators and fish parasites than found in the other two treatments. 73 Table 19. Pearson correlations (and associated probabilities) between the original observations and the principal component scores (n=27 variables) of the functional guild biomass dataset. Collector-gatherers Collector-filterers Scrapers Predators Shredder-detritivores Invertebrate parasites Fish parasites Explained variance (%) PCI 0.899. (0.0001) -0.083 (0.681) 0.753 (0.0001) 0.931 (0.0001) 0.324 (0.099) 0.342. (0.081) 0.463 (0.015) 70 PC2 0.401 (0.038) 0.112 (0.578) 0.629 (0.0001) -0.355 (0.069) -0.364 (0.062) 0.104 (0.606) 0.605 (0.001) 18 Collector-gatherers Collector-filterers Scrapers Predators Shredder-detritivores Invertebrate parasites Fish parasites Explained variance (%) PC3 0.023 (0.909) 0.582 (0.001) 0.056 (0.783) -0.068 (0.738) 0.803 (0.0001) -0.006 (0.977) -0.040 (0.844) 7 PC4 -0.030 (0.883) 0.801 (0.0001) -0.030 (0.894) 0.050 (0.805) -0.344 (0.079) -0.091 (0.601) -0.105 (0.653) 4 74 Figure 21. Plot of Principal Component 2 (PC2) versus Principal Component 1 (PCI) for the functional guild biomass dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Scrapers Fish parasites -0:6-0.4 Jk2 Thurs k i l I A F l y M ° FF q.Wenpel - 8 J L t 1 + * ^ G r g ^ B p t l C M O 0--0.2 -0.4 H -0.6 -0.8 —1-Bv V*pt2 UD r x ,0.2 0.4 « 0.6 0.8 A V Bpt2 \ UB \ UB ' v. \ x Grgt +UNCUT • RECENT CUT x OLD CUT P C 1 (70%) Collector-gatherers Scrapers Predators Fish parasites 75 The 'old cut' stream sites assembled the lowest along the axis, implying the lowest amount of biomass in the collector-gatherers, scrapers, predators and fish parasites. Functional guild biomass varied among timber harvesting treatments (MANOVA, Table 15). The treatments were segregated by the biomass of collector-gatherers, scrapers, collector-filterers and fish parasites. The first canonical axis (91%; Figure 22) showed that 'recent cut' streams had higher biomass of collectors and fish parasites and lower biomass of scrapers compared to the 'uncut' and 'old cut' streams (Table 20; MANOVA, p=0.002 and p=0.002, respectively).). The 'uncut' and 'old cut' streams (MANOVA, p=0.684) were not different on the first canonical axis. Invertebrate parasite biomass differed weakly among treatments (ANOVA, p=0.070; Table 21), being slightly greater in the 'recent cut' stream sites (Figure 23). Individual biomass of the collector-gatherer, scraper, predator and fish parasite guilds was greatest in the 'recent cut' stream sites. However, their biomass in the 'old cut' stream sites declined to levels found in the 'uncut'. The distribution of the stream sites on the PCA graph of the functional guild density and biomass datasets, as seen with the 'instream' dataset, illustrated the high within-stream variability of the benthic community structure of the streams sampled in this study. The functional guild density and biomass within the streams sampled in this study. In many cases, the site pairs that came from a single stream and were in close geographic proximity [e.g. some that were sampled next to each other such as Bivouac ("Bv"), Unnamed A ("UA"), Unnamed B ("UB"), Gluskie ("Gls"), Forfar ("FF") and Fly creeks] 76 Figure 22. Plot of Canonical Variate 2 ( C A N 2 ) versus Canonical Variate 1 ( C A N 1 ) for the functional guild biomass dataset. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Scrapers 2. CM < -4 o Collector-gatherers 1 -3 -fc H h \ + \ * n4_ 1 7 4,0 + U N C U T • R E C E N T C U T X O L D C U T 1 I 2 3 ^ 4 ^ 5 6 CAN1 (91%) Scrapers Collector-gatherers Collector-filterers Fish parasites 77 Table 20. The total sample standardized canonical coefficients for the first two axes from a canonical variates analysis based on the functional guild biomass variables of 27 stream sites. CAN1 CAN2 Collector-gatherers 1.153 -2.190 Collector-filterers 0.875 0.139 Scrapers -0.867 2.411 Predators 0.669 0.232 Shredder-detritivores -0.497 0.278 Fish parasites 1.005 -0.659 Invertebrate parasites 0.318 0.534 Explained variance (%) 91 9 78 Table 21. F-value approximations and associated P-values for ANOVA, contrasting three groups ('U'-'uncut' streams, 'R'- 'recent cut' streams and 'O'- 'old cut' streams) based on the functional guild biomass dataset. Differences among groups were assessed by the Tukey's Studentized test. The degrees of freedom (df), Mean Square term (MS), R-square (r2), F-value approximations (F) and associated P values of significance (P) are listed. Those groups whose means differ are indicated by a separate letter and groups with similar means have their letters attached. Parameter Source df MS Differences among groups Collector-gatherer Group 2 0.178 0.453 9.95 0.001 Error 24 0.018 R U-0 Collector-filterer Group 2 0.007 0.084 1.10 0.350 Error 24 0.007 R-U-0 Scraper Group 2 0.123 0.411 8.37 0.002 Error 24 0.015 R U-0 Predator Group 2 0.205 0.253 4.07 0.030 Error 24 0.050 R U-0 Shredder-detritivore Group 2 0.002 0.013 0.15 0.858 Error 24 0.011 U-R-0 Invertebrate parasite Group 2 0.001 0.199 2.99 0.070 Error 24 0.000 U-R-0 Fish parasite Group 2 0.003 0.369 7.02 0.004 Error 24 0.000 R U-0 79 Figure 23. Histograms ( ± 1 standard error) of In transformed biomass of the collector-gatherer, scraper, predator, invertebrate parasite and fish parasite functional guilds, in 'uncut', 'recent cut' and 'old cut' streams. Probability of significance is in brackets. Collector-gatherers (p=0.0007) Scrapers (p=0.0018) Predators (p=0.0300) Invertebrate parasites (p=0.0695) Fish parasites (p=0.004) 0Unlogged (n=13) • Recent cut (n=6) • Old cut(n=8) 0.1 0.2 0.3 0.4 0.5 0.6 0.7 In biomass (g/m ) 80 were not spatially proximate on the PCA graphs (Figures 13, 21). Therefore, sites within a given stream were different enough with respect to the benthic community structure to be considered separate sites for purposes of further analysis (i.e., I will use them as independent observations). 3.3 Macroinvertebrate relationships with instream habitat and productivity Trends among study streams were first examined by PCA according to four separate types of environmental variables (i.e., watershed, topography, instream physical and instream chemical/productivity variables). From the PCA plots, it could be seen that the study sites ordered by instream physical and instream chemical/productivity variables revealed patterns based on logging treatment. PCA revealed that the watershed and topographic variables, overlapped almost entirely among treatments, so treatments appeared to have minimal influences on the among-site trends. Further, these cluster similarities indicate that any differences in the benthic macroinvertebrate trends among sites may not be attributable to among site variation in topographic or watershed conditions. Instead, they may be attributable to differences in other factors. Therefore, it was assumed from this analysis, that any trends among the functional guild density and biomass datasets could not be attributed to changes in these variables. Thus, all further analyses examining the relationships between biotic and abiotic data were examined using only the instream physical and instream chemical/productivity datasets and their relationship to the functional guild density and biomass datasets. The first two PCA axes of the instream physical and instream chemical/productivity datasets were then compared to the first two axes of the functional guild density and biomass datasets (Tables 22 and 23). The Pearson correlation between 81 Table 22. Pearson correlation between the first two axes from the PCA on functional guild density data and first two axes from the PCA's on instream and chemical datasets. Probability of significance is in brackets. Physical data axis 1 (n=27) Physical data axis 2 (n=27) Chemical data axis 1 (n=18) Chemical data axis 2 (n=18) Functional guild density Functional guild density PCA axis 1 PCA axis 2 -0.085 (0.673) -0.374 (0.055) 0.062 (0.760) -0.408 (0.035) 0.440 (0.068) -0.091 (0.721) -0.336 (0.173) -0.505 (0.032) 82 Table 23. Pearson correlation between the first two axes from the PCA on functional guild biomass data and first two axes from the PCA's on instream and chemical datasets. Probability of significance is in brackets. Functional guild biomass Functional guild biomass PCA axis 1 PCA axis 2 Physical data axis 1 (n=27) -0.375 (0.054) -0.018 (0.931) Physical data axis 2 (n=27) 0.043 (0.830) 0.010 (0.960) Chemical data axis 1 (n= 18) 0.188 (0.455) -0.105 (0.678) Chemical data axis 2 (n= 18) -0.321 (0.194) -0.447 (0.063) 83 the second PCA axis of functional guild density and the first axis of the instream dataset was only marginally and negatively associated (r=-0.374; p=0.060; Table 22; Figure 24). Sites within large sized streams may have had slightly lower invertebrate parasite and fish parasite densities than the smaller-sized 'recent cut' streams with regards to collector-filterers. 'Recent cut' streams in comparison may have had a slightly wider range in benthic macroinvertebrate density among the relevant guilds (Figure 24). However, these differences were not significant The second PCA axis of functional guild density was more significantly and negatively correlated with the second axis of the instream physical dataset (r= -0.408; p=0.035; Table 22; Figure 25). Stream sites with a large number of LOD and high riffle velocity appeared to have had low densities of collector-filterers, invertebrate parasites and fish parasites. The 'uncut' streams seemed to have had the highest densities of collector-filterers, fish parasites and invertebrate parasites. However, the 'uncut' streams did not seem to differ in the number of pieces of LOD and riffle velocities from the other treatments (Figure 25). The 'recent cut' streams had a lower density of the three relevant guilds but did not differ among the other treatments with respect to NLOD and riffle velocity. The 'old cut' streams, which covered the range in the number of pieces of LOD and the riffle velocities had intermediate densities of the relevant macroinvertebrate guilds (Figure 25). 84 Figure 24. Plot of the functional guild density data Principal Component axis 2 versus the physical habitat data Principal Component axis 1. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Bankfull width Bankfull height Wetted width Pool area Riffle area Riffle depth Riffle velocity jn re .*-» re •o "re o to x re b 0 . \3 \ X X .-1 .+ + -3-+ UNCUT • RECENT CUT x OLD CUT X 0 :___4 Collector-filterers Invertebrate parasites Fish parasites PC axis 2 invertebrate guild density data (13%) 85 Figure 25. Plot of the functional guild density data Principal Component axis 2 versus the physical data Principal Component axis 2. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. N u m b e r p i e c e s L O D Ri f f le ve loc i t y CM ro TJ "co o "(fl >. JZ a CM </> "5 to O CL -3 x J . X . ,+ + + + UNCUT • R E C E N T CUT x OLD CUT + v Ik V. •+-\ 3 _± . ^ PC axis 2 invertebrate guild density (13%) Col lec tor - f i l te rers , F i s h paras i tes , Invertebrate paras i tes 86 The first PCA axis of functional guild density and the first PCA axis of the instream chemical/productivity dataset was only weakly associated (r=0.44; p=0.070; Table 22; Figure 26). Sites, then, with a high percentage of canopy cover, conductivity, stream temperature and chlorophyll a biomass also may have been associated with a high density of collector-gatherers, collector-filterers, scrapers, predators and shredder-detritivores. In contrast, a significant negative correlation existed between the second PCA axis of the functional guild density and the second PCA axis of the instream chemical/productivity variables demonstrated (r=-0.5053; p=0.032; Table 22; Figure 27). This result suggested that those stream sites with a high percentage of canopy cover and low stream water conductivity may have contained higher densities of collector-filterers, invertebrate parasites and fish parasites. In contrast to functional guild densities, functional guild biomasses were less strongly correlated with environmental variables (Table 23). The first PCA axis of functional guild biomass was only marginally and weakly negatively associated with the first axis of the instream dataset (r=-0.375; p=0.054; Table 23). The smaller the sizes of the stream sites may have had slightly higher collector-filterer, invertebrate parasite and fish parasite biomass. 87 Figure 26. Plot of the functional guild density data Principal Component axis 1 versus the chemical Principal Component axis 1. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Canopy cover Conductivity Stream temperature Chlorophyll a biomass re TJ re u E X to O a. -6 .SWc -6 -9-3 ^ + U N C U T • R E C E N T C U T x O L D C U T PC axis 1 invertebrate guild density data (71%) Collector-gatherers Collector-filterers Scrapers Predators Shredder-detrit ivores Invertebrate parasites 88 Figure 27. Plot of the functional guild density data Principal Component axis 2 versus the chemical Principal Component axis 2. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams. Canopy cover ra TJ ra u E £ O CN (A - x re O a- -2 Conductivity 1 -1.5 -3:5-2.5 I.-\ X , 1.5 X 0.5 x \ -0--0.5 -1.5 - 2 - L + UNCUT • RECENT CUT x OLD CUT "0.5 1.5 2.5 Collector-filterers Invertebrate parasites Fish parasites PC axis 2 invertebrate guild density (13%) 89 The second PCA axis oi" the guild biomass was also weakly and negatively associated with the second PCA axis of the instream chemical/productivity dataset (r=-0.447; p=0.063; Table 23; Figure 28), although not significantly. A relatively higher percentage of canopy cover with low conductivity may have been associated with slightly lower biomass of the fish parasites and scrapers. After the initial examination of potential relationships between environmental and functional guild abundance data (density and biomass) using PCA, interdataset relationships were examined using canonical correlation analysis. This method was used to measure how well the datasets of benthic macroinvertebrate variables fit to a selected number of instream habitat or instream productivity datasets. The first canonical correlation revealed the only significant relationship as being between functional guild density and the instream habitat (60%; Wilks' lambda p=0.041; Rcl= 0.902; Table 24). The canonical variables that described the strongest correlation between functional guild density and instream physical habitat was the density of scrapers, their predators and invertebrate parasites. Their densities were positively correlated with the number of LOD pieces within the stream and negatively with bankfull height (Table 24). 90 Figure 28. Plot of the functional guild biomass data Principal Component axis 2 versus the chemical data Principal Component axis 2. Polygons delineate sites belonging to 'uncut', 'recent cut' and 'old cut' streams Canopy cover 00 rs TJ re u E a> € -1 CM CO X re O CL Conductivity i .5 -1 \ ^ \ -1T5-\ 1 \ 6-5 -0.5 -0.5 / / / -1T5-0.5 si x / » \ PC axis 2 invertebrate guild biomass data (14%) + U N C U T • R E C E N T C U T x O L D C U T Fish parasites Scrapers 91 Table 24. Results of the Canonical correlation analysis. 'RC,' and 'RC2' represent the canonical coefficients. 'V and 'W represent their respective canonical variates (see Methods section). Pearson correlations and probability showing the major trends between functional guild density and instream variables of the canonical coefficients are also shown. Rci Rc 2 Pearson correlations 0.902 0.784 Probability 0.041 0.403 (% explained variance) 60 22 Canonical variables VI Wl V2 W2 Collector-gatherer -0.383 -0.346 -0.039 -0.031 Scraper -0.490 -0.443 0.003 0.002 Predator -0.440 -0.397 -0.072 -0.057 Shredder-detritivore 0.106 0.095 -0.043 -0.034 Fish parasite 0.136 0.123 -0.673 -0.528 Invertebrate parasite -0.481 -0.434 -0.473 -0.371 Bankfull width 0.147 0.163 0.576 0.734 Bankfull height 0.506 0.561 0.426 0.543 Wetted width -0.010 -0.011 0.694 0.885 NLOD -0.580 -0.643 0.277 0.353 Pool area -0.013 -0.014 0.494 0.630 Riffle area 0.164 0.182 0.422 0.538 Riffle depth 0.020 0.022 0.358 0.456 Riffle velocity 0.113 0.126 0.465 0.592 92 4.0 Discussion The watersheds of the 'old cut' streams whose riparian zones had been timber harvested more than twenty years before may have been more easily accessible to harvesting activities due to the large watershed size and low gradients of these streams compared to watersheds of the other two treatments. After the larger watersheds in the lower gradient stream channels had been harvested, timber harvesting activity moved into the higher gradient streams found in the smaller and less accessible watersheds. The 'recent cut' stream watersheds were harvested next despite their smaller watershed size compared to the 'uncut' streams. This may have been due to these streams having somewhat lower stream channel gradients than the gradients found in the 'uncut' streams. The proportion of watershed area and stream length that had been completely harvested also differed between the three treatments. The percentages of the watershed and stream length harvested in the 'old cut' and 'recent cut' streams clearly differed from the 'uncut' streams. When examined more closely, my results indicated that the 'uncut' group of streams had a much lower percentage of the watershed and stream length harvested than the other two treatments. This result was expected since the unlogged streams were selected to be in watersheds that had not been altered by timber harvesting. The 'old cut' streams seemed had a greater percentage of their entire stream length harvested than the 'uncut' streams, although the percentage of the stream length harvested in the 'recent cut' streams seemed to be intermediate to the proportion harvested in the other two treatments. This suggests that timber harvesting was not as extensive in the 'recent cut' streams' watersheds at the time of the study as seen with the 'old cut' streams' watersheds. Also, trends in the instream habitat dataset indicated that the smaller size of the 'recent cut' streams showed that changes in recent timber harvesting practices may have allowed for only smaller streams to have been harvested to the streambanks. This was in contrast to 93 earlier timber harvest practices where larger streams were often harvested to the streambank. The streams in this study may show that more than twenty years after the timber harvest of a riparian zone, the streambanks of the 'old cut' streams may be experiencing greater erosion compared to the streams in the other two treatments. The 'recent cut' streams despite having the smallest range in stream sizes differed in bankfull width, which was not unexpected. Also, the 'uncut' streams appeared to be about in the same size range as the 'old cut' streams. Their bankfull width measurements were not significantly wider than the 'recent cut' streams. It appeared in the streams of this region that the greater width of the 'old cut' streams may have been due to the presence of highly erodible lacustrine soils, lower channel gradients and naturally occurring mass wasting events commonly found in the Willow and Bowron watersheds (Larkin et al. 1998). Richards et al. (1996) found that bankfull width and channel cross section dimensions in low relief lacustrine were strongly influenced by geological structure variables and landscape watershed size. Further evidence supported the suggestion that streambank erosion may have occurred in these streams after the timber harvest. The 'uncut' streams appeared to have had the greatest bankfull heights associated with the smallest bankfull widths compared to the 'old cut' sites. The sizes of the bankfull widths and bankfull heights of the 'recent cut' stream sites appeared to be intermediate between the other two treatments. These results suggested that the streambanks remained destabilized even as the riparian zones matured. Bankfull width increase is frequently accompanied with a reduction of channel depth. Even twenty or more years after the timber harvest, the stream channels did not appear to have reached the conditions found in the 'uncut' sites. Streambank destabilization may have created a continuing press perturbation on the stream ecosystem in these streams over time after the riparian zones were harvested through the changes in the physical habitat. Through field observations, the streambanks of many of the 'old cut' 94 streams in the Willow-Bowron region seemed to be wider and shallower streambanks and perhaps experiencing severe streambank erosion. It was possible that after the initial removal of the riparian vegetation, streambanks could easily be eroded by stream flow, particularly due to the presence of the fine, loose lacustrine soils (Slaney et al. 1977; and Larkin et al. 1998) found in these watersheds. In the lower reaches of the study sites in the 'old cut' streams of the Willow-Bowron region, several streambanks were reinforced through human-made constructions. This was done to slow further erosion of the streambanks. Differences in pool and riffle areas within the 'old cut' streams were greater compared to those found in the 'recent cut' streams. This may be due to the larger sizes and greater bottom surface area of the 'old cut' streams compared to the overall, smaller sized 'recent cut' streams. As would be expected, the size of the pool and riffle areas in the 'uncut' streams were similar to both the 'recent cut' and the 'old cut' streams as the 'uncut' streams' sizes were intermediate. More notably, the riffle to pool ratio did not differ between the streams by treatment. In forest channels such as those examined in this studies, the channel type, slope and width and density of LOD has a strong influence on the riffle-pool morphologies (Montgomery et al 1995). The presence and stability of large structures such as LOD allows the persistence of riffle/pool systems by anchoring the pool locations and creates upstream riffles (Brisson and Montgomery 1996). In streams influenced maritime climate, the number of pools eventually decreases without a sustained supply of obstructions such as LOD (Montgomery et al 1995). Evidence in this study did not show a difference in the density of LOD within the stream. I speculate that if the 'recent cut" and 'old cut' streams had similar densities of LOD and riffle to pool ratio before harvest, as did the 'uncut' streams, the density of LOD within the streams had remained persistent over time. Thus, the pool to riffle ratio may not have been altered either through press or pulse perturbations following the timber harvest of the riparian zone. Either the freshet events within 95 these reaches may not be strong enough to displace the LOD or there remained a large enough supply in the riparian zones to ensure a constant supply to replace the LOD that had been displaced or had deteriorated over time. Closer examination of the physical structure of these streams would be necessary to determine if and how the riffle to pool ratio and LOD may actually change over a longer time post logging, than had been examined in this study. Richardson (1994) predicted that following logging, the release of nutrients and the enhancement of the streambed to light in these streams of the interior would stimulate primary productivity. I found that trends of the streams by treatment, indicated that 'recent cut' stream sites had the highest levels of benthic algae, represented in this study by chlorophyll a biomass, correlated with the lowest amount of canopy cover. This is consistent with the work of others (Noel et al. 1980; Murphy and Hall 1981; and Webster 1990). Richardson (1994) also suggested that after several years post logging in these streams, productivity would decrease as the successional vegetation fills in. As expected, more than twenty years after the timber harvest the 'old cut' streams had a greater canopy cover and lower chlorophyll a biomass than I had found in the 'recent' cut streams. The 'old cut' streams had the same densities of canopy cover and levels of chlorophyll a biomass as found in the 'uncut' streams. However, the canopies of the 'old cut' streams were dominated primarily by successional deciduous vegetation. Only further work done in the 'old cut' streams can clarify whether aquatic primary productivity levels will actually remain at this level or if the levels will decline further over time. Other researchers have found that the deciduous canopy in recent second growth riparian, which is thicker than the old growth canopy, can cause additional decline in autochthonous productivity levels below those levels found in 'uncut' streams (Gregory et al. 1987). Giroux (1994) found that continental streams of the Temagami, Ontario that had experienced a timber harvest 15-150 years earlier, had a much denser canopy cover than his streams that were found in old growth forests. My 96 study streams were also dominated by deciduous vegetation. Thus, removal of the canopy cover in these streams can also potentially cause a long-term press perturbation within the continental streams by reducing light penetration to the streambed over the long-term. Further study would need to be done in these streams over a longer period of time to examine whether canopy cover density increases. Another function of canopy cover is as the representative of the presence or absence of riparian trees that control overland flow of water and nutrients. Richardson (1994) predicted that the logging of streams in the interior would result in the early release of overland nutrients into the streams and therefore an increase in stream conductivity. Giroux (1994) found a strong negative correlation between biomass of trees (dbh > 10 cm) in the riparian zone and stream chemistry (total dissolved substances). Patterns in the streams by treatment showed that the 'recent cut' streams in this study had relatively low canopy cover associated with higher stream conductivity than was found in the streams of the other two treatments. This result confirmed the conclusions of others (e.g. Swanson et al. 1982). The riparian vegetation may absorb more of the overland water flow and the associated dissolved nutrients (Gregory et al. 1987) and its removal may allow for this increase into the streams. Past literature strongly suggested that the opening of the canopy and the increase in sunlight and nutrients would substantially increase aquatic primary productivity (Breschta et al. 1987; Webster 1990; and Anderson 1992). Trends between the streams indicated that the 'recent cut' stream sites differed from the other two stream treatments by their much lower canopy cover and higher stream conductivity (Figure 8). Patterns between the streams by treatment, which explained 87% of variability, indicated a strong separation by the 'recent cut' streams from streams in the other two treatments was associated with a much lower canopy cover density and higher chlorophyll a biomass in the 'recent cut' streams. In both observations the variance due to 97 canopy cover, conductivity and periphyton biomass in the 'uncut' and 'old cut' streams were similar. Thus, the loss of the streamside riparian trees may have been associated with increased input of overland organic debris such as fine soils and nutrients into the streams, along with increased streambank erosion. The loss of the canopy cover may also have been associated with increased sunlight onto the streambed, which often stimulates photosynthesis in chlorophyll a production in the portion of periphyton that is algae. My results also showed that only the differences in canopy cover was independently significant between treatments (Table 12). A larger number of cobbles sampled within each streams site and a sample taken within each riffle in each site may have strengthened the evidence for differences between each site. Measurements taken in the more productive summer growing months may also have shown a more significant difference between sites in primary productivity. 4.1 Macroinvertebrate density and biomass 4.1.1 Total macroinvertebrate density and biomass In streams found within watersheds in both continental and maritime climates a complete riparian harvest can change the abundance and/or accessibility to food resources (temporal inputs and quality and quantity), often having a large effect on the macroinvertebrate community structure (Gregory et al. 1987). Changes in stream food resources such as algae, represented in this study by chlorophyll a biomass, after logging activity may have influenced the differences in the total biomass and density of macroinvertebrates between the streams in this study. As previously mentioned, the patterns of streams by treatment suggested a strong association between removal of the canopy cover and higher chlorophyll a biomass in the 'recent cut' streams. Patterns of the total macroinvertebrate biomass by stream treatment also indicated that biomass was much higher in the 'recent cut' streams than in the streams of the other two treatments. The higher levels of biomass in the 'recent cut' streams suggested that some 98 members of various taxa within the community may have been more productive in the streams with lower canopy cover and associated higher chlorophyll a biomass, compared to the streams in the other two treatments. The 'old cut' streams in this study appeared to have had canopy cover and chlorophyll a biomass and total macroinvertebrate biomass levels similar to pre-harvest conditions. Studies of both maritime and continental climate influenced streams have suggested that various taxa within the community initially react quickly to the new habitat conditions often found soon after initial disturbance event, for example with the increase in periphyton quality and/or quantity (Newbold et al. 1980; Murphy et al. 1981; Hawkins et al. 1982; Carlson et al. 1990; and Anderson 1992). Once the riparian vegetation has re-established a dense canopy cover, the productivity of these taxa and periphyton quality decline accordingly, eventually returning to pre-harvest levels. In contrast to these results, Giroux's (1994) study on continental streams in Ontario did not find differences in total benthic macroinvertebrate biomass between any of his three treatments. Macroinvertebrate density in this study did not differ by treatment as seen in the biomass of the macroinvertebrate community. This result contrasts with the past literature dealing with maritime climate streams where the density of macroinvertebrates tended to increase dramatically in the few years after the timber harvest, then declined after the canopy cover grew back over the harvested streams (Newbold et al. 1980; Murphy et al. 1981; Hawkins et al. 1982; Carlson et al. 1990; and Anderson 1992). However, it appeared to agree with Giroux's (1994) study on continental streams that did not find differences in total benthic macroinvertebrate density between any of his three treatments. The changes in the instream habitat conditions soon after a timber harvest and after more than twenty years since the timber harvest may have affected the life history of various taxa in the benthic macroinvertebrate community. A speculative explanation of these results would 99 suggest that the increase in the quality of primary productivity within these streams soon after a timber harvest may have stimulated faster growth and maturation of some of the susceptible taxa of the benthic macroinvertebrate community during the growing season (or alternatively, some taxa may have been replaced by larger sized taxa). In the recently harvested streams, individual taxa may have benefited by the new conditions, increasing biomass and potentially maturity more quickly than in streams with a denser canopy cover. Eventually, after more than twenty years and regrowth of the riparian canopy cover, biomass levels returned to pre-harvest levels in these streams. Further study of these streams is required to determine whether these changes in biomass are occurring over time or if these differences were due to confounding factors. 4.1.2 Functional guilds density and biomass My results suggested strongly that total macroinvertebrate biomass in this region was influenced by stream treatment. Density also seemed to differ between treatments, albeit not as significantly. Past literature had shown there is often an increase in the densities and biomass the taxa within certain functional feeding guilds that are able to benefit from the temporary habitat modifications and increase in food resources after a riparian harvest (Hawkins et al. 1982; Carlson et al. 1990; and Anderson 1992). At the same time, there may also be a decline in local densities and biomass of taxa in other guilds that are not able to withstand these new conditions. I examined the changes in density and biomass of the functional guilds between stream treatments. In this study, the community structure in the streams of this region, with respect to the density and biomass of several guilds, appeared to have differed among the three treatments. The differences in guild density by stream treatment when the guilds were examined individually, were not clear. Invertebrate parasite densities seemed to have been highest in the 'recent cut' stream sites and at comparable levels in 'uncut' and in the 'old cut' streams. Fish 100 parasite density appeared to have been much lower in the 'old cut' streams compared to the 'recent cut' streams. Because the differences in density of the invertebrate and fish parasites are often strongly influenced by the presence of their hosts, the results inferred a difference in the density of their respective invertebrate or fish hosts. Past literature has also shown that there are definite, positive relationships between the densities of predators and their prey (Hawkins and Sedell 1980). In these streams predator density was clearly lower among the 'old cut' stream sites than in the other two treatments which suggests that there may have been a lag time between the rapid increase in the productivity of their prey taxa earlier in the growing season in the 'recent cut' streams. However, no direct evidence was found to determine whether prey productivity had been higher earlier in the growing season. Further analysis was required to examine these ecological relationships more closely. Individual guild biomass appeared to have supported the results found in past studies (Murphy and Hall 1981; and Murphy et al. 1981). Independently, collector-gatherer, scraper, predator and fish parasite biomass were much higher in the 'recent cut' streams compared to the streams in the other two treatments. The higher biomass of the collector-gatherer and scraper guilds in the 'recent cut' streams may have been linked to the higher quality of their food resources, i.e., benthic algae in these streams. The higher biomass of the prey taxa may have been associated with the greater presence of the invertebrate predator biomass in the 'recent cut' streams. The taxa within these guilds must have been smaller sized taxa and may have been dominant in these streams since predator biomass, which is dependent upon small-sized prey with short life cycles, were higher in these streams than in the other two treatments (Merritt and Cummins 19966). Also, fish, attracted by the greater macroinvertebrate biomass to feed upon, may have moved into these small 'recent cut' streams, thereby providing a greater food resource for the fish parasites in these streams. Fish hosts such as trout species may be attracted to these 101 streams by the increase in the productivity of macroinvertebrates as a food source (Anderson et al. 1981). In addition, habitat conditions in the 'old cut' streams may have been similar to those found in the 'uncut' streams. As a result, biomass of these guilds in the 'old cut' streams seemed to have been comparable to the biomass found in the 'uncut' streams. However, the individuals of the predator guild appeared to be larger in size, as their density was the lowest in the 'old cut' streams compared to the other treatments, yet their biomass did not appear to have differed from the 'uncut' streams. Observations of the relationships between the guilds gave a clearer picture of the structure of the benthic community. Patterns between treatments with respect to the guild density suggested that while the 'recent cut' stream sites had the highest densities of predators and invertebrate parasites compared to the 'uncut' and 'old cut' stream sites, they also had the lowest densities of scrapers. Thus, the density of predators and the invertebrate parasites may have been linked to the density of the scraper guild in a potential prey/host relationship. Past literature has shown that there are definite, positive relationships between the densities of predators and their prey (Hawkins and Sedell 1980). Parasites are also often strongly influenced by the presence of their hosts. Patterns between treatments with respect to the guild biomass indicated that a higher biomass of collectors (gatherers and filterers) and fish parasites was also associated with the lowest biomass of scrapers in the 'recent cut' streams compared to the other two treatments. In the 'old cut' stream sites, the biomass levels seemed to be comparable to the levels found in the 'uncut' streams. Scrapers may have been the guild most strongly affecting the benthic community structure with respect to other guilds. Scrapers have been positively associated with high quality food sources, in particular standing chlorophyll a biomass on cobbles (Hawkins and Sedell 1980), more than on food quantity and substrate for feeding (Merritt and Cummins 19966). With 102 the lack of canopy cover and the greater quality of higher quality periphyton in the 'recent cut' streams scraper density and biomass would have been expected to be highest in these streams. One speculation of this trend may be that warmer water temperatures and primary productivity earlier in the growing season may have accelerated scraper productivity through stimulated feeding (Beschta et al. 1987) in the 'recent cut' streams. The higher densities of predators and invertebrate parasites in these streams may have been linked to the hypothetically higher density of scrapers found earlier in the growing season. Densities of the predators and invertebrate parasites of the scraper guild may have been higher because of the potential lag time between scraper productivity and predator and parasite development. If earlier scraper productivity had been higher, the possibly longer lived taxa found in the predator and invertebrate parasite guilds may have built up their density. When the shorter-lived scraper productivity declined, compared to levels found earlier in the season possibly due to lower quantities of food sources, predation pressure from the other two guilds may have driven productivity down. Then, scraper productivity may have declined quickly to levels found in the streams of the other two treatments by the time of the study. The other two guilds may have remained higher in comparison, due to their longer life history. Giroux (1994) suggested that food resources are important but only for immediate responses and development rather than for long-term development. Thus, since conditions in the 'old cut streams appeared to be similar to those in the 'uncut' streams they may have had a similar community structure despite possible changes soon after the timber harvest in their watersheds. The high levels of collector-filterer and gatherer biomass in the 'recent cut' streams indicated that taxa within the guilds of these streams tended to be larger in size than those found in the other two treatments, although their density remained the same. Larger taxa may have replaced smaller ones soon after the harvest (Hawkins et al. 1982) and may be replaced by the 103 original taxa twenty or more years after the timber harvest. Another explanation could be that the 'recent cut' streams may have produced a significantly higher quantity or quality of food resources in their FPOM (Fine particulate matter) and VFPOM (Very fine particulate matter) pools. As a result, their body size may have been increasing steadily earlier in the season, prior to this study. Often, after a timber harvest of the riparian zone there can be a substantial influx of FPOM from the input of organic soils transported overland from the riparian zone into the stream and from the erosion of the streambanks (Merritt and Cummins 19966). Another factor could be the possible increased quality of the FPOM in these streams (Hawkins and Sedell 1980). The lack of canopy in these streams often results in a higher density and sloughing off of higher quality periphyton (Noel et al. 1986) in these streams, usually in the growing season. Also, potentially, there is often an input of herbaceous vegetation, high in nutrient content and low in fibre, that may also increase the quality of the FPOM within 'recent cut' streams (Meehan et al. 1977). Since the collector guilds tend to be longer-lived than scrapers, they may have been able to continue to be present or even growing, even after the increased primary productivity had slowed, perhaps feeding on other sources. The scrapers, which in this study included the short-lived Baetis sp. may have declined compared to the collector guilds, because of the lower availability of periphyton that may have been found in much higher production earlier in the growing season. However, measurements of detritus in these streams were not taken, and summer variables were not measured. Further study of these streams is required to clarify this speculation. The increased biomass in the 'recent cut' streams of the collectors was also strongly associated with a higher biomass of the fish parasites. Trout biomass has been found to be significantly greater in 'recent cut' streams than in forested streams and strongly correlated with biomass of collector-gatherers (Anderson et al. 1981). Thus, the greater biomass of the collector-104 guild in the 'recent cut' streams may have attracted fish to these streams, thereby also increasing fish parasite biomass. This strongly suggested the increased presence in the 'recent cut' streams of fish, compared to the other two treatments. Total biomass of the macroinvertebrates and the density and biomass of several functional guilds appeared to have recovered in the 'old cut' streams to levels found in the 'uncut' streams. Only time will tell whether community composition will continue to change. Changes in the community composition may continue if the habitat will continue to change in the future. An ongoing 'press' perturbation, due to continuing change in instream habitat conditions or productivity, may drive the system even more than twenty years after the timber harvest. Since a benthic invertebrate community structure may not return to the pre-cut condition until recovery of the instream habitat conditions and proportions of food resources had taken place (Niemi et al. 1990), recovery of the community structure would not take place for a longer period than that measured in this study. However, in this study, the habitats may have experienced changes, but those that have, appeared to have changed back. Further study is required over a longer time span in these streams to determine whether these levels in both biotic and abiotic variables stabilized to the pre-cut conditions, exemplified in this study by the 'uncut' streams. 4.1.3 Taxa density Changes in the composition of benthic macroinvertebrate taxa can influence modifications in the overall functional guild measurements (Hawkins et al. 1982). I found that comparisons between treatments with respect to the density of taxa within the functional guilds indicated in many cases that individual taxa did not appear to strongly influence trends found in functional guild density between treatments. Overall, the largest differences between the taxa by treatment occurred between the 'recent cut' and 'old cut' streams. Earlier results indicated that 105 the collector, predator and parasite guilds were the most affected by stream treatment. Examination of the taxa differed slightly. The density of the predator Chelifera sp. larvae and the nematode invertebrate and fish parasites may have increased due to the higher densities of their potential hosts in the 'recent cut' streams. In particular, as seen previously, the fish parasites Hirudinea may have had higher densities due to the availability of their fish hosts in these streams. In addition, Chelifera sp. are commonly burrowed in the sediments in depositional areas in the stream channel. This suggested that the substrate in the 'recent cut' streams may have contained more deposited sediments than the streams in the other two treatments, although my sampling measures did not find differences between treatments. The taxa that had higher densities in the 'recent cut' streams had densities in the 'old cut' streams similar to the 'uncut' streams, suggesting that specific instream conditions for these taxa were similar to those found in the 'uncut' streams. The lower densities of hosts for the parasitic invertebrates appeared also to have been similar to levels found in the 'uncut' sites. However, four taxa that had not shown to be affected by the changes in the 'recent cut' streams, declined substantially in the 'old cut' streams. The densities of the generalist collector Tanytarsini larvae; the scraper nymphs of the Drunella sp. and Glossosoma sp.; the predator larvae of the Rhyacophila sp.; and the predator/parasite Acarina, did not appear to have been affected by the different conditions in the 'recent cut' streams compared to the 'uncut' streams. These results were in contrast to past literature that suggested that the adult emergence densities of the Rhyacophilids and Glossomatids increased from the streams that had been recently harvested (Hawkins et al. 1982). However, they appeared to be influenced by conditions in the 'old cut' streams. The differences in the densities of this group of taxa in the 'old cut' streams indicated that despite the appearance of similar conditions to the 'uncut' streams, for certain taxa the 106 conditions within these streams differed considerably for others. It was also notable that Baetis sp. density has been shown to increase dramatically in streams soon after a timber harvest of the riparian zone soon after a timber harvest then will decline substantially after a few years (Wallace and Gurtz 1984; and Giroux 1994). Notably, Baetid density and biomass were not found to have differed between treatments. In general, these results have suggested that there may have been ongoing alterations in the functional guild densities in the two points in time sampled after the riparian zones had been harvested, compared to the streams which were not harvested. With respect to the individual taxa within each guild, the effects may differ according to their resilience to the prospective changes in their habitat and food resources. Although the densities of the taxa may differ overall, there seems to be no real major losses or gains in the presence or absence of any taxa between any of the stream treatments, allowing for regional differences in a few taxa, most notably Pelecypoda. These results conflict with Giroux (1994) who found a greater taxa richness in streams within less than 15 year old harvested riparian zones. Overall, in the streams of this study only the density of 20 taxa out of the 75 total number of taxa found showed notable differences between treatments. Differences may have been clearer either with using a larger sample size, by sampling more often over the year or by sampling more often after harvest than was done in this study. Alternatively, biomass measurements of these taxa may have provided additional information of the response of the individual taxa to the changes in the stream ecosystem after the timber harvest. 4.2 Environment-macroinvertebrate relationships In this section I examined how the benthic macroinvertebrate community structures differed with respect to instream habitat and productivity factors (Hawkins et al. 1982). Scrapers appeared to play a key role in the benthic community structure in their response to changes in the 107 stream habitat in the streams of this region, as seen in previous results. Further analysis indicated that the density of the scraper guild in association with their invertebrate parasites and predators, appeared to be strongly influenced by the density of LOD and negatively with bankfull height. Input of woody debris from the riparian zone can often have a strong influence on the benthic community structure and their habitat (Stone and Wallace 1998). LOD within a stream is used as a substrate for biological activity (grazing activity and substrate for periphyton growth), the creation of habitat variation (Meehan et al. 1977) and the dissipation of the stream's energy (Silsbee and Larson 1983). Scrapers are grazers and depend on periphyton (Merritt and Cummins 19966) which grows predominantly on rock, wood and plant surfaces (Lowe and La Liberte 1996). Thus, the greater the density of LOD within the instream habitat, the more surface area for growth of periphyton and the more accessible it may have been to grazing by the scraper guild. Scrapers also depend upon a stable substrate within their habitat on which to feed, often associated with higher densities of LOD. Further study of the effects of alterations in factors supporting periphyton may be useful. The lower bankfull heights of the streams, negatively associated with the density of scrapers and LOD, may also have been associated with primary productivity. Often with low water depth within a stream, greater amounts of sunlight can reach the substrate of the streambed. The increase in stream water temperature can also increase the metabolic rate and biological activity of the susceptible benthic macroinvertebrate taxa within the guilds. Unfortunately, neither the direct measure of the quantity (periphyton biomass) nor quality (chlorophyll a biomass) was found to be more directly correlated with scraper density. However, the correlation between scrapers and this food source has been found to be significant in past studies (Hawkins and Sedell 1980). Having only one measurement of periphyton and stream water temperature may not have been able to demonstrate a clear association between primary 108 productivity and scraper density. Sampling had also been done in late summer/early autumn, when perhaps the difference in primary productivity between streams may not have been as distinct as earlier in the growing season. Patterns of data between the physical and functional guild data suggested that changes in some environmental factors seemed to be associated with changes in several guilds. High density of LOD over or within a stream can indicate the retention of sediment, presence of instream heterogeneity (Murphy and Hall, 1981) and instream cover for organisms within the stream, such as fish (Murphy et al. 1981). It can also provide a stable substrate for organisms feeding either on the surface of the substrate (e.g., scrapers), for attaching to the substrate to grow (e.g., periphyton), or from which to feed (e.g., collector-filterers). Fast riffle velocity associated with a large number of LOD within the stream also suggested fast movement of suspended FPOM and VPOM over the substrate and submerged LOD. Associated high riffle velocities may provide fast, continual movement of food particles along the stream for collector-filterers. Collector-filterer density has been more often associated with high current velocity than quantity of detritus (Hawkins et al. 1982). High levels of nutrients (conductivity) and sunlight input from the removal of the canopy cover may have stimulated the growth of high quality periphyton, in turn possibly producing high quality VPOM/FPOM (Murphy and Meehan 1991) for filterers and periphyton for scrapers, a potential host for the invertebrate parasites. In contrast to expected results, the 'recent cut' streams contained the lowest densities of collector-filterers, invertebrate parasites and fish parasites compared to streams in the other two treatments. One explanation might be that the low densities of the collector guild and the hosts of the invertebrate parasites in the 'recent cut' streams may have been due to emergence earlier in the growing season compared to the taxa in the streams of other two treatments. Alternatively, the alteration of other factors in the streams, other than those measured in this study, may have affected these guilds, despite the 109 potentially beneficial conditions measured in these streams. Although substrate compositions of the riffle habitat are often altered after the clear-cut timber harvest in both maritime and continental climate streams (Noel et al. 1986), they were not found to have differed between sampled riffle sites in these streams. The stable substrate, along with the high density of LOD, may have provided food and stable substrate for the scraper guild. This may have allowed access to the scraper hosts by their parasites, the invertebrate parasite guild. Possibly the conditions in the 'recent cut' streams may not have been suitable for the host fish despite the high biomass of macroinvertebrates in these streams. Lethal summer temperatures of the stream water and lack of habitat or access to these streams could also have been a factor. However, although these associations proved significant, the PCA axes used here explained much less of the variation in the data of the biomass, density and physical datasets compared to the first axes. Thus, it may only explain a very minor part of the pattern in the respective datasets. After more than twenty years after the timber harvest, the densities of collector-filterers, fish parasites and invertebrate parasites and biomass of the collector-gatherer, scraper, predator and fish parasite guilds appeared to have returned to levels similar to levels found in the 'uncut' streams. Canopy cover, conductivity, density of LOD and riffle velocities were similar to those levels found in the 'uncut' streams. Only canopy composition differed between 'uncut' and 'recent cut' streams from a primarily coniferous canopy in the 'uncut' streams, to a primarily deciduous one in the 'old cut' streams. At the time of the study, leaf abscission from the riparian vegetation in the 'old cut' streams had not yet begun so canopy cover density was similar to the 'uncut' streams at the time of the study. This result suggested that because the two sets of physical variables were similar to the levels found in the 'uncut' streams, the benthic macroinvertebrate community structure in the 'old cut' streams may have returned to levels 110 found in the 'uncut' streams. Recovery of the instream habitat is strongly influenced by the replacement of adequately sized LOD to these streams. True recovery of the stream ecosystem, assuming that the community structure before the harvest was similar to that found in the 'uncut' streams, depends on the density of dead or dying mature trees falling into the streams from the riparian zone. Thus, instream habitat recovery may not occur for many more years (Giroux 1994) in the streams of this region, and up to one hundred years or more in watersheds in other regions (Gregory et al 1987). Other environmental requirements of the guilds, either not measured in this study or because they were only measured once in only one season, (e.g. stream temperature) may also have had an effect on the life histories of the taxa of these guilds that may not have been realized in this study. Although some speculations were made in this study about changes in functional guild density and biomass in the streams within this region over two points in time since the riparian zone of the streams were harvested, this study was not sufficient to establish whether trends of disturbance and recovery were taking place. Some changes in the density and productivity within the functional guilds between these three stream treatments were observed. Further work is also required in these streams to determine whether the 'press' effects of the timber harvest in the 'old cut' streams would continue after the study was done after more than twenty years, assuming the community structure had originally been similar to those found in the 'uncut' streams. 4.3 Comparison of streams influenced by continental and maritime climates Many studies have been carried out to determine how timber harvest related disturbances affect streams and their biota. The studies have been primarily conducted in streams found in the mountainous regions of the Pacific Northwest regions of the United States and Canada. The 111 watersheds of these streams are affected primarily by maritime climates (i.e., high winter storm frequency and high precipitation rates; Heede 1984). As expected, the streams within the continental climate influenced watersheds of the interior stream systems of B.C. in this study did not seem to demonstrate extreme temporal instream habitat changes as found in more maritime streams. However, there was the suggestion of a difference in the ratio of channel dimensions among the three groups of streams, although not as definitive as seen in more maritime climate influenced streams. The 'uncut' streams had the highest bankfull heights and the narrowest bankfull widths. The 'recent cut' group, in turn had intermediate channel dimensions. At the other extreme, the 'old cut' streams appeared to have had the lowest bankfull heights and the widest bankfull widths between all three treatments. These differences may have been associated with slight dissimilarities in channel gradients among the three groups. However, gradients of the streams by treatment did not differ enough to create a difference between the groups of streams with respect to the substrate composition of the riffle habitats or stream power sampled between each group of streams. The streams may have been becoming wider and shallower more slowly over time after the timber harvest due to streambank erosion and overland transport of sediment. The loss of LOD through seasonal flooding was not extreme in these streams and may occur over a longer period of time. There was no obvious difference in the density of LOD among the three groups of streams. This suggested that sediment storage was being maintained even after more than twenty years after the timber harvest. Thus, definite changes in channel morphology and instream habitat may not be as intensive in the first twenty years or so after the timber harvest in continental streams compared to alterations found in the maritime streams. However, some temporal changes may still have been occurring, so modifications in instream habitat may continue to occur after a longer period of time than sampled in this study, 112 particularly as the LOD disintegrates or is slowly removed through spring freshet activity. Giroux (1994) found that even 15-150 years after the timber harvest in Ontario, the biomass of the primarily coniferous vegetation (dbh > 10 cm) was significantly smaller than found in the old growth riparian zones. The size of LOD entering the stream may be smaller in harvested watersheds, even after more than twenty years. Thus, replacement of adequately sized LOD to these streams to encourage stability and heterogeneity may not occur for many years after the timber had been removed or deteriorated. This may affect instream habitat with respect to channel morphology and/or channel heterogeneity, as seen in the 'old cut' streams. Scouring activity by transported sediment has also been found to remove macroinvertebrates from the substrate (Culp et al. 1985). Sediment intrusion often occurs after a timber harvest in this region where soil is highly erodible (Slaney et al. 1977; Brownlee et al. 1988; Larkin et al. 1998). Continued streambank erosion and terrestrial runoff after the timber harvest may store a significant quantity of sediment within the LOD structures still in the streams after the timber harvest. Sediment intrusion into a stream habitat often occurs when there is no longer sufficiently large LOD to trap sediment or maintain storage of sediments released at the time of the timber harvest. With the gradual loss of LOD over time the stored sediment may enter into the stream habitat. This may create a press perturbation for the stream's biotic community long after the timber harvest. Scouring activity with increased sediment intrusion may have a detrimental effect on the benthic macroinvertebrate community until the lost LOD is eventually replaced by large mature timber from riparian zone. Measurements of suspended sediment and deposited sediments were not taken in this study. In addition, to changes in physical instream morphology, new riparian growth could drive primary productivity down in both the continental and maritime streams. The 'recent cut' streams appeared to be influenced by lack of canopy cover associated with high conductivity. 113 The return of the canopy cover and conductivity during growing season, as seen in the 'old cut' streams, may show the loss of access to sunlight and potential nutrients due to fast growing deciduous riparian vegetation which absorb large amounts of nutrients from the overland runoff. The deciduous vegetation may also produce changes in canopy composition from a primarily coniferous canopy in 'uncut' streams to deciduous canopies in 'old cut' streams. Despite lack of evidence of some habitat changes, the benthic macroinvertebrate community structure appeared to change, although not dramatically as would be expected in maritime streams. The timber harvest on these streams did not appear to be as large a disturbance event on the community as found in the streams within the more maritime climates. Changes in the benthic macroinvertebrate community structure may have been associated more with shifts in primary productivity influenced by the density of the canopy cover and input of nutrients and less by severe habitat degradation. For example, the greater presence of the immature stages of several non-colonizing taxa (Farula sp., Ecclisomyia sp. and Orthocladiinae) in the 'recent cut' streams seemed to have indicated that the streams in this study had not been devastated by the severe scouring disturbances often found in the maritime streams. In addition, only the density of 20 of the 75 taxa in these streams had differed in the streams by 'treatment', and none of the taxa had been lost among any of the treatments. Therefore, there was no loss of macroinvertebrate diversity at the time of the study. Past literature has found that in maritime watersheds after timber harvest, there is a dominant influx of colonizing taxa, in particular collector/scraper guilds such as the Baetis sp. Densities of this taxa emerged at a rate four to five times higher from a stream within a newly clear-cut watershed compared to levels found in old growth sites in some maritime streams (Hawkins et al. 1982). In continental streams, densities of insects emerging averaged 17.6 times the densities found in the unharvested streams (Gurtz and Wallace 1986). Giroux (1994) also 114 found in his study of continental streams, that Baetis sp. had the highest density and biomass in streams with 15-150 year old harvests of their riparian zones. These taxa are small sized and have short, multivoltine lives and high fecundity (Gurtz and Wallace 1986). They often re-colonize streams after a large disturbance event through drift activity (Hawkins et al. 1982; and Anderson 1992) and thereby quickly exploit the increases in primary productivity often found after a clear-cut (Gurtz and Wallace 1986). Diatoms compose the major portion of their diet and after logging, their gut contents indicate a significant increase in diatom consumption (Gurtz and Wallace 1986). However, in this study, I did not find a difference in their density between the groups of streams. The lack of difference in Baetis sp. density between treatments may have been due to the slowing of primary productivity in the autumn season in the 'recent cut' streams. Because of their short lives, any high productivity earlier in the summer growing season would not have been measurable at the time of this study. Notably, the predator guild biomass, which is dependent upon small-sized prey with short life cycles, appeared to be higher in the 'recent cut' streams than in the other two treatments (Merritt and Cummins 19966). This may have indicated that there had been a higher level of productivity earlier in the growing season, not determined in this study. However, differences were not seen in this study. Overall evidence in this study did not find that severe press disturbance events were affecting the overall biological community structure in these streams after the removal of the riparian zone, as seen in many maritime streams after a timber harvest. Changes in the benthic macroinvertebrate community structure may have been associated more with shifts in primary productivity influenced by the density of the canopy cover and input of nutrients and less by habitat degradation at the time of the study. Although changes in the stream ecosystem may not be as extreme as in streams that experience maritime climates, there is some evidence that it may be still occurring. Further and more detailed work is required to examine potential physical and 115 biotic changes over time in these streams. This can be done to examine how long changes in these streams will occur before the habitat recovers to conditions found in streams that have not had a timber harvest in their riparian zone. 4.4 Study limitations This study could have been improved with a long-term pre-disturbance database and an equally long-term post disturbance record (Yount and Niemi 1990) of each of the streams studied. However, because measurements over such a long time span was not feasible for this study, I attempted to address this concern by using a wide range of streams with wide range of instream characteristics over three periods of time since the riparian zone of the stream had been harvested. I also included a control treatment. An ideal scientific field study of ecosystem recovery from disturbance would also include an equally long-term post disturbance record, to determine when and if stream ecosystem recovery occurred (Yount and Niemi 1990). However, the main difficulty in examining impacts on the stream ecosystem from press disturbances is that the "time scales for recovery are too long to document endpoints" (Niemi et al. 1990). In this study, the longest period after the timber harvest of the riparian zones was approximately twenty years. Press disturbances often impact stream ecosystems for long periods of time. Thus, recovery of the instream habitat and benthic community may take a long time. Sampling again after 50 years or longer could give a clearer view of whether recovery of the factors studied in these streams had occurred or if they will continue to change, as suggested by Meehan and Murphy 1991 and Gregory et al. 1987. Thus, unless further study is made, it will remained unknown for these streams whether the stream ecosystem had actually recovered after more than twenty years, or whether they remained in flux. The changes in the benthic community structure may become more extreme before they return to the pre-cut conditions, if they return at all. 116 This study focused on 'snapshot' assessments of stream conditions towards the end of the summer base flow, including periphyton and chlorophyll a biomass, stream temperature, conductivity, riffle velocity, bankfull width, and bankfull height. Stream temperature, conductivity, riffle velocity, bankfull width, and bankfull height are highly variable with respect to previous night flows and storm activity and the area of the stream being measured. Periphyton and chlorophyll a biomass are also highly variable within a stream. These parameters were also measured only once over six weeks. In this period of time the aforementioned variables could alter due to potential differences in the weather conditions. If each group of streams by treatment were measured at the same time or a few times over the growing season conditions may be found to be more attributable to weather conditions as opposed to stream conditions. In the 'old cut' streams, it should be noted that despite the late summer/early autumn season, the amount of leaf litter input and density of the canopy cover had not changed, nor had the abundance and productivity of the shredder-detritivore guild, compared to the levels found in the 'uncut' streams. Thus, although the streams were sampled over the six weeks, they were not found to have been significantly different from the 'uncut' streams. However, it was observed that the autumn conditions were starting to develop. Another factor to be taken into account was that the harvesting of the watersheds of the streams was not done randomly and evenly distributed over the regions. Thus, the 'recent cut' streams were.in the smaller size ranges compared to the streams in the other two treatments, possibly due to changes in harvesting practices. 'Old cut' streams may have had lower gradients than the others did, due possibly to accessibility by older timber harvesting practices. The use of multivariate analyses (i.e., PCA, MANOVA and CANDISC) allows me to extract the relationships between the groups with respect to the variables that best described these relationships. 117 Relative measures of abundance (density) and biomass of each functional guild were limited by the size of the mesh of the surber sampler. The entire size range of the benthic macroinvertebrates could not be collected, so the data collection was biased in having no collection of those organisms or generations that were smaller than the mesh size. However, the collection of insects was consistent between sites. The use of two different people to sample the macroinvertebrate community and to sort out the benthic macroinvertebrates from the samples also could have had an influence on the data. But, I attempted to minimize this bias by having standard techniques for processing. Finally, in doing the analyses, other correlations between functional guild composition and environmental factors were found, but with relationships that appeared to be less clear. This may have been attributable to the small number of streams within each treatment. Using a greater number of sites for each treatment may have provided clearer trends between functional guild composition and environmental data, but access to other streams was difficult in this region. Also, a thorough investigation of the streams of these watersheds showed that only these streams were similar enough with respect to watershed and topographic characteristics to be compared. Overall, the work done in this study was primarily exploratory, to provide a preliminary basis for future work in these regions. More extensive, controlled studies need to be done in these systems for more precise information on the interactions between the stream habitat changes and their effects on the benthic macroinvertebrate community structure. 118 Literature cited Aloi, J.E. 1990. A critical review of recent freshwater periphyton field methods. Can. J. Fish. Aquat. Sci. 47:656-670. Anderson, N.H., 1992. Influence of disturbance on insect communities in Pacific Northwest streams. Hydrobiologia 248: 79-92. Bender, E.A., Case, T.J., and Gilpin, M.E. 1984. Perturbation experiments in community ecology: theory and practice. Ecology 65:1-13. Beschta, R.L., Bilby, R.E., Brown, G.W., Holtby, L.B. and Hofstra, T.D. 1987. Stream temperature and aquatic habitat: fisheries and forestry interactions in Streamside management: forestry and fishery interactions, University of Washington, Institute of Forest Resources, E.O. Salo, and T.W. Cundy (eds.). Bisson, P.A. and Montgomery, D.R. 1996. Valley segments, stream reaches and channel units in Methods in Stream Ecology. Academic Press, Toronto, Ontario. F.R. Hauer and G.A. Lamberti (eds.). Brownlee, M.J., Shepherd, B.G., and Bustard, D.R. 1988. Some effects of forest harvesting practices on water quality in the Slim Creek Watershed in the central interior of British Columbia. Can. Tech. Rep. Fish. Aquat. Sci. #1613: 4lp. Carlson, J.Y., Andrus, CW. and Froelich, H.A. 1990. Woody debris, channel features, and macroinvertebrates of streams with logged and undisturbed riparian timber in Northeastern Oregon, U.S.A. Can. Tech. Rep. Fish. Aquat. Sci. 47:1103-1111. Choromanski, E.M., Macdonald, J.S., and Scrivener, J.C, 1994. Invertebrate production in the Takla tributaries in Proceedings of the Takla Fisheries/Forestry Workshop: a two year review. April 1, 1993, Prince George, B.C. Can. Tech. Rep. Fish. Aquat. Sci. 2007: 104p. J.S. Macdonald (ed.). Culp, J.M., and Davies, R.W. 1983. An assessment of the effects of streambank clear-cutting on macroinvertebrate communities in a managed watershed. Canadian Technical Report of Fisheries and Aquatic Sciences No. 1208. Detenbeck, N., DeVore, P., Niemi, G.J., Lima, A., 1990. Recovery of temperate-stream fish communities from disturbance: a review of case studies and synthesis of theory. Environmental Management 14: 493-515. Dudgeon, D. 1994. The influence of riparian vegetation on macroinvertebrate community structure and functional organization in six New Guinea streams. Hydrobiologia 294:65-85. 119 Frissel, C.A., Liss, W.J., Warren, C.E., and Hurley, M.D. 1986. A hierarchical framework for stream habitat classification: viewing streams in a watershed context. Environmental Management 10: 199-214. Giroux, P. 1994. Correlations between catchment disturbance, stream chemistry, invertebrate and riparian communities in northeastern temperate forests near Temagami, Ontario: a test of the Biomass Accumulation model. MSc thesis, Trent University. Gregory, S.V., Lamberti, G.A., Erman, D.C, Koski, K.V., Murphy, M.L. and Sedell, J.R. 1987. Influence of forest practices on aquatic production in Streamside management: forestry and fishery interactions, University of Washington, Institute of Forest Resources, E.O. Salo, and T.W. Cundy (eds.). Gregory, S.V., Swanson, F., McKee, W.A., and Cummins, K., 1991. An ecosystem perspective of riparian zones. Bioscience 41:540-551. Gurtz, M.E., and Wallace, J.B. 1986. Substrate-mediated response of stream invertebrates to disturbance. Ecology 65: 1556-1569. Hansmann, E.W. and Phinney, H.K. 1973. Effects of logging on periphyton in coastal streams of Oregon. Ecology 54:194-199. Hawkins, CP., Murphy, M.L. and Anderson, N.H., 1982. Effects of canopy, substrate composition, and gradient on the structure of macroinvertebrate communities in Cascade range streams of Oregon. Ecology 63: 1840-1856. Hawkins, CP. and Sedell, J.R. 1980. Longitudinal and seasonal changes in functional organization of macroinvertebrate communities in four Oregon streams. Ecology 62: 387-398. Heede, B.H., 1984. Sediment source areas related to timber harvest on selected Arizona watersheds in Symposium on effects of forest land use on erosion and slope stability: 7-11 May, 1984. Environment and policy institute East-West centre, University of Hawaii, Honolulu, Hawaii, USA. Hinch, S.G. and Collins, N.C, 1993. Relationships of littoral fish abundance to water chemistry and macrophyte variables in central Ontario lakes. Can. J. Fish. Aquat. Sci. 50: 1870-1878. Kelly, J.R. and Harwell, M.A. 1990. Indicators of ecosystem recovery. Environmental Management 14: 527-545. Kiss, L. 1984. A study of five timber harvesting systems used for streamside logging. MSF thesis, University of British Columbia. 120 Larkin, G.A., Slaney, P.A., Warburton, P., and Wilson, A.S. 1998. Suspended sediment and fish habitat sedimentation in central interior watersheds of British Columbia. Province of British Columbia, Ministry of Environment, Lands and Parks, and Ministry of Forests. Watershed Restoration Management Report No. 7:3 lp. Lemly, A.D. 1982. Modification of benthic insect communities in polluted streams: combined effects of sedimentation and nutrient enrichment. Hydrobiologia 87:229-245. Lemmon, P.E., 1957. A new instrument for measuring forest overstory density. J. Forestry 2:314-320. Lowe, R.L. and LaLiberte, G.D. 1996. Benthic stream algae: distribution and structure in Methods in Stream Ecology. Academic Press, Toronto, Ontario. Hauer, F.R. and Lamberti, G.A. (eds.). Macdonald, J.S., Scrivener, J.C, and Smith, G. 1992. The Stuart-Takla Fisheries/Forestry Interaction project: study description and design. Can. Tech. Rep. Fish. Aquat. Sci. 1899: 39p. Mahoney, D.L., and Erman 1980. The role of streamside bufferstrips in the ecology of aquatic biota in California riparian systems - ecology, conservation and productive management. R.E. Warner and K.M. Hendrix (eds.) University of California Press, Berkeley. Manly, B. F. J. 1986. Multivariate Statistical Methods. Chapman and Hall. London, U.K. Meehan, W.R., Swanson, F.J., and Sedell, J.R., 1977. Influences of riparian vegetation on aquatic ecosystems with particular references to salmonid fishes and their food supply in Importance, preservation and management of riparian habitat: a symposium. UDSA Forest Service General Technical Report RM-43, 1977. Mellina, E. and Hinch, S. 1995. Overview of large-scale ecological experimental designs and recommendations for the British Columbia Watershed Restoration Program. British Columbia Ministry of Environment, Lands and Parks, and Ministry of Forests. Watershed Restoration Project Report No. 1: 3 lp. Mellina, E. and Rasmussen, J.B. 1994. Patterns in the distribution and abundance of zebra mussel (Dreissena polymorphd) in rivers and lakes in relation to substrate and other physicochemical factors. Can. J. Fish. Aquat. Sci. 51:1024-1036. Merritt, R. and Cummins, K. 1996a. Aquatic insects of North America. 3rd ed. Kendall/Hunt Publishing Co., Dubuque, Iowa. Merritt, R. and Cummins, K. 19966. Trophic relations of macroinvertebrates in Methods in Stream Ecology. Academic Press, Toronto, Ontario. F.R. Hauer and G.A. Lamberti (eds.). 121 Montgomery, D., Buffington, J., Smith, R., Schmidt, K., and Pess, G. 1995. Pool spacing in forest channels. Water Resources Research 31:1097-1105. Murphy, M.L., and Hall, J.D. 1981. Varied effects of clear-cut logging on predators and their habitat in small streams of the Cascade mountains, Oregon. Can. J. Fish. Aquat. Sci. 38:137-145. Murphy, M.L., Hawkins, CP. and Anderson, N.H., 1981. Effects of canopy modification and accumulated sediment on stream communities. Trans. Amer. Fish. Soc. 110:469-478. Murphy, M.L., Heifetz, J., Johnson, S.W., Koski, K.V., and Thedinga, J. F. 1986. Effects of clear-cut logging with and without buffer strips on juvenile salmonids in Alaskan streams. Can. J. Fish. Aquat. Sci. 43: 1521-1533. Murphy, M.L., and Meehan, W.R. 1991. Stream ecosystems in Influences of forest and rangeland management on salmonid fishes and their habitats. American Fisheries Society Special Publication 19:17-46. Newbold, J.D., Erman, D.C, and Roby, K.B., 1980. Effects of logging on macroinvertebrates in streams with and without buffer strips. Can. J. Fish. Aquat. Sci. 37:1076-1085. Niemi, G.J., DeVore, P., Detenbeck, N., Taylor., D., Lima, A., Pastor, J., Yount, J.D., Naiman, R.J., 1990. Overview of case studies on recovery of aquatic systems from disturbance. Environmental Management 14: 571-587. Noel, D.S., Martin, C.W., and Federer, C.A., 1986. Effects of forest clearcutting in New England on stream macroinvertebrates and periphyton. Environmental Management 10: 661-670. Pimental, R.A. 1979. Morphometries: The multivariate analysis of biological data. Kendall/Hunt Publishing Co. Iowa 276 pp. Poff, N. 1992. Why disturbances can be predictable: a perspective on the definition of disturbance in streams. J.N. Am. Benthol. Soc. 11:86-92. Ralph, S.C, Poole, G.C, Conquest, L.L., and Naiman, R.J. 1994. Stream channel morphology and woody debris in logged and unlogged basins in western Washington. Can. J. Fish. Aquat. Sci. 51:37-51. Resh, V., Brown, A., Covich, A., Gurtz, M., Li, H., Minshall, G., Reice, S., Sheldon, A., Wallace, J., and Wissmar, R., 1988. The role of disturbance in stream ecology. J.N. Am. Benthol. Soc. 7:433-455. 122 Richards, C., Johnson, L., Host, G. 1996. Landscape-scale influences on stream habitats and biota. Can. J. Fish. Aquat. Sci. 53 (Suppl. 1): 295-311. Richardson, J. 1994. Forest-stream interactions: potential effects on productivity and biodiversity of benthic communities in interior British Columbia in Proceedings of the Takla Fisheries/Forestry Workshop: a two year review. April 1, 1993, Prince George, B.C. Can. Tech. Rep. Fish. Aquat. Sci. 2007: 104p. SAS Institute Inc. SAS/STAT User's Guide, Release 6.03 Edition. Cary, NC: SAS Institute Inc., 1988. 1028 pp. Silsbee, D. and Larson, G. 1983. A comparison of streams in logged and unlogged areas of Great Smoky Mountains National Park. Hydrobiologia 102: 99-111. Slaney, P.A., Smith, H.A., and Halsey, T.G. 1977. Physical alterations to small stream channels associated with streamside logging practices in the central interior of British Columbia. Province of British Columbia, Ministry of Recreation and Conservation, Fisheries Technical Report No. 31:41p. Steinman, A. and Lamberti, G. 1996. Biomass and pigments of benthic algae in Methods in Stream Ecology. Academic Press, Toronto, Ontario. F.R. Hauer and G.A. Lamberti (eds.). Stone, M. and Wallace, J. 1998. Long-term recovery of a mountain stream from clear-cut logging: the effects of forest succession on benthic invertebrate community structure. Freshwater Biology 39: 151 -169. Swanson, F.J. 1979. Geomorphology and ecosystems in Forests: fresh perspectives from ecosystem analysis : proceedings of the 40th annual Biology Colloquium, 1979, R.E. Waring (ed.) Swanson, F.J., Gregory, S.V., Sedell, J.R. and Campbell, A.G. 1982. Land-water interactions: the riparian zone. pp. 267-291 in Analysis of coniferous forest ecosystems in the western United States. Hutchinson Ross Publishing Company, R.L. Edmonds (ed.). Thedinga, J.f., Murphy, M.L., Heifetz, J., Koski, K.V., and Johnson, S.W. 1989. Effects of logging on size and age composition of juvenile coho salmon (Oncorynchus kisutch). Can. J. Fish. Aquat. Sci. 46: 1383-1391. Wallace, J.B., 1990. Recovery of lotic macroinvertebrate communities from disturbance. Environmental Management 14: 605-620. Wallace, J.B and Gurtz, M.E. 1984. Response of Baetis mayflies (Ephemeroptera) to catchment logging. Am. Midi. Nat. 115: 25-41. 123 Webster, J.R., Golladay, S.W., Benfield, E.F., D'Angelo, D.J. and Peters, G.T. 1990. Effects of forest disturbance on particulate organic matter budgets of small streams. J. N. Am. Benthol. Soc. 9:120-140. White, P., and Pickett, S. 1985. Natural disturbance and patch dynamics: an introduction, in The ecology of natural disturbance and patch dynamics. Academic Press Inc., New York, P. White and S. Pickett (eds.). Yount, J.D. and Niemi, G.J. 1990. Recovery of lotic communities and ecosystems from disturbance-a narrative review of case studies. Environmental Management 14: 547-569. Zar, J. H. 1996. Biostatistical Analysis. 3rd ed. Prentice Hall, New Jersey. 662 pp. 124 s cS H •c cd C/3 C D 43 0 0 e o o o >v X) 3 1/) u 43 o o H—» o CO VH CS o 13 o '0 un >v ov. -5 ON cS 1-<u O u O -*-» o o '•3 c a a < T3 CJ 43 oo c3 CN S u e CJ t-< H ts « < s ft s~ S o ^ cd x s — u 5 CJ .—• ca fi | b ft E o s '3 ID H o c j a. c _o 13 > cj s B 5b s o cj 3 t a 1 00 22 T3 3 C J .2 OH 00 C O 3 t a C J o l-J CJ u o o o o o o o o o o o o p p © © © d © © © 2 «o ^ vo CN un ca H CN O un r f o © r f o CO o o o o r f r f CN o o o — o o r f m cn r f un cn r f r f VO r f - H ' vo oo CN — CN O O cn r f m 00 OO u-> vo cn m vo cn cn cn C N oo oo 00 00 - - ft H-ca ca E E Z <U 43 00 ^ z z ft H -z cn OV —< C N un od Un C N r f C N cn cn 00 00 00 u C J o "3 13 13 v. V H V H u C J cj T3 -o O o o s s w e w Z Z H—( UJ CN co ov — un CN oo oo CN Ov vo cn un C N O cn o © O r f r- cn VO 0 0 r f m cn m 00 oo ft *n 8 ,2 00 ^ z cn C N 00 00 ca E E W W Z Z •» r f 00 CN CN 00 00 H- ft H-PH 55 fa o o O o o m m O o un m Ov o Sv cn CN CN CN '—' CN o o o o o o o o cn o o o O >/-) CN o un 0 0 o un CN un un CN CN r f VO . - H cn CN CN in un CN CN O cn u-i CN ^ ^ ^ £ £ £ £ o o U l O C N r~ o o un O CN © o o un o C N cn o cn cn U-> CN u-> U l CN — i un o CN o Ov un r f un r f un un un un un un un un un z z z z z z z 3 3 3 3 3 3 3 o o o o o o a c a a c s a a P D CN CN CN CN CN CN c c a> cj O o C J cj Pi Oi a CJ CJ CJ Pi C CJ o Pi 3 o 3 o 3 o T 3 T3 o o 3 o ^ o o o ca 3 ca H ca ca H M M M M cd cd cd b+ \-+ \r+ ca H ca H a o o CQ 3 o o CQ o £ £ £ CN CN 00 oo ft +-55 * w w w Z z z z z z z o un o o o o o o o o un CN un o o un un un un o cn o o r f vo *—i o^ o o CN cn cn cn cn ~ O o o vo o o O o o o o un o o cn un o Ov o o un o p vq oo CN p rn cn Ov p r f vb o r f vo un r f vo cn cn '—< cn '— 1 •— 1 •— 1 r f cn un r f u-i CN CN CN CN T—H CN CN CN CN CN CN CN CN CN CN ^ £ £ £ ^ O O O O un O O O O un cn un Ov un un oo o cn O o CN r f Ov o 0 0 oo cn ui Ov cn cn r f oo C-- un un p CN cn CN cn rn r f cn CN r f r f un cn cn cn cn cn cn cn cn cn un un un un un un un un un un un z z z z z z z z z z z 3 3 3 3 3 C o o u • u !S ^ o o —' CN _ H 3 T3 cj E ca c 3 P 3 3 . o CQ CQ CN 3 .O ' f i cj cfl _ 3 o 5 cd +-» 3 42 ' f i CJ M cfl 3 CQ « 3 cj cj z b o ta T3 CJ s ca 3 3 P o ca 3 o > CQ Q w u T3 T3 CJ e CJ c CJ e c ca 3 ca 3ca 3 3 3 3 3 3 P P P 3 CJ CJ &0 BO O cj o CJ a a Pn cB T3 cn tH 3 cd -*-» 3 43 CJ -a 3 cj 3 o -3 T3 O O 2 e U Q 0 PQ _c <D T3 O o C * H - » cS o ' GO OO cS s cS CD O O CS C/3 VH o P H CD 43 CH-H CD VH CD O U U HH CQ CQ CQ E un 

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.831.1-0089348/manifest

Comment

Related Items