Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Seasonal changes of benthic macroinvertebrate communities in Southwestern British Columbia Dymond, Pamela F. 1998

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

Item Metadata

Download

Media
831-ubc_1998-0154.pdf [ 8.12MB ]
Metadata
JSON: 831-1.0099262.json
JSON-LD: 831-1.0099262-ld.json
RDF/XML (Pretty): 831-1.0099262-rdf.xml
RDF/JSON: 831-1.0099262-rdf.json
Turtle: 831-1.0099262-turtle.txt
N-Triples: 831-1.0099262-rdf-ntriples.txt
Original Record: 831-1.0099262-source.json
Full Text
831-1.0099262-fulltext.txt
Citation
831-1.0099262.ris

Full Text

SEASONAL CHANGES OF BENTfflC MACROINVERTEBRATE COMMUNITIES IN SOUTHWESTERN BRITISH COLUMBIA by P A M E L A F. D Y M O N D B.Sc , University of Calgary, 1995 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE D E G R E E OF M A S T E R OF SCIENCE m THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Zoology) We accept this thesis as conforming to the required standard THE UNIVERISTY OF BRITISH C O L U M B I A April , 1998 © Pamela F. Dymond, 1998 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 The University of British Columbia Vancouver, Canada DE-6 (2/88) Abstract Two main goals of community ecology are to recognise patterns in species composition and to understand the processes affecting those patterns. The first goal of this study was to identify patterns of benthic macroinvertebrate composition between both different classes of streams and seasons in southwestern British Columbia. As well, life cycles and environmental conditions were examined to determine which processes were associated with patterns of community composition. The second goal of this study was to determine whether seasonal changes of the benthic community were large enough to affect the accuracy of predictions made by predictive models used for biomonitoring. Predictive models cannot extrapolate beyond the range of variability included within a reference data set. Because of seasonal variation, these models may only provide accurate predictions for the season when reference samples were collected. In the Fraser River Basin, reference samples to be used for biomonitoring with predictive models were collected only once a year - in the autumn. This study established the applicability of test samples collected outside of the reference-sample season. To address the goals of this study, benthic macroinvertebrate samples and associated environmental data were collected over five sampling dates during the course of one year, 1995-1996 (late spring, summer, autumn, winter, early spring). The samples were collected from eight streams that comprised three different classes of streams: 1) coastal streams, 2) interior plateau streams, and 3) large rivers. Each of the stream classes has a different climate, elevation, riparian vegetation, and discharge regime. Seasonal change of the benthic invertebrate community was small relative to the spatial change i.e., between the three stream classes. Correlation analysis indicated that spatial change of the invertebrate community in southwestern British Columbia was related to the environmental factors channel width, mean depth, maximum depth, maximum velocity, discharge, conductivity, alkalinity, nitrite and nitrate nitrogen, and total Kjeldahl nitrogen. Seasonal change of the community was not directly related to the seasonal change of any of the individual environmental variables measured. It was however, related to changes in the environment through effects of the environment on invertebrate life cycles. Of the iii environmental conditions measured, temperature had the greatest influence on the timing of insect life cycles. Abundance patterns in the small coastal and interior streams were related to the timing of invertebrate life cycles whereas abundance was related to life cycles and the spring freshet in the large rivers. Although seasonal variation of the invertebrate community in ordination space was less than spatial variation, this study found that test samples should be collected during the same season as when the biomonitoring reference samples were collected, or over multiple sampling dates. This strategy reduces the possibility of seasonal shifts or stochastic events leading to erroneous conclusions about the state of a test site. iv Table of Contents Abstract ii List of Tables vi List of Figures vii Acknowledgements x Chapter 1 Introduction 1 The Benthic Community 1 Biomonitoring and Benthic Macroinvertebrates 2 Seasonality and Biomonitoring in the Fraser River Basin 3 Purpose 4 Chapter 2 Life Cycles of Three Lotic Insects Living Under Three Different Discharge Regimes 6 Introduction 6 Study Sites 9 Interior Streams 9 Coastal Streams 12 Large Rivers 12 Methods 14 Benthic Invertebrate Life Cycle and Abundance Data 14 Environmental Conditions 17 Results 19 Invertebrate Life Cycles, Abundance and Growth Rates 19 Environmental Conditions 24 Discussion 28 Influence of Discharge Regime on Life Cycles 28 Influence of Temperatures and Food Resources on Life Cycles 30 Chapter 3 Seasonal and Spatial Changes of the Benthic Macroinvertebrate Community and Associated Environmental Conditions in Southwestern British Columbia 33 Introduction 33 Methods 34 Invertebrate Samples 34 Environmental Conditions 36 Data Analysis 38 Results 42 Invertebrate Composition 42 Environmental Conditions 55 How Changes in the Benthic Community are Related to Changes in the Environment 55 Discussion 58 Spatial Patterns of Benthic Communities 58 Seasonal Patterns of Benthic Communities 61 V Chapter 4 Seasonal Change of the Benthic Macroinvertebrate Community and the Implications for Biomonitoring 65 Introduction 65 Methods 68 Invertebrate Samples 68 Environmental Conditions 69 Data Analysis 69 Grouping of Sites 70 Selection of Predictor Variables 71 Prediction of Seasonal Test Sites to Groups 72 Results 73 Results for Genus/Species Data (0.1% cut-off) 74 Grouping of Sites 74 Selection of Predictor Variables 79 Prediction of Seasonal Test Sites to Groups 79 Effect of Data Censorship Level on the Results 83 Grouping of Sites - Genus/Species Data (0.5% cut-off) 83 Selection of Predictor Variables 85 Prediction of Seasonal Test Sites to Groups 89 Effect of Taxonomic Resolution 89 Grouping of Sites - Family Data (0.5% cut-off) 89 Selection of Predictor Variables 92 Prediction of Seasonal Test Sites to Groups 92 Discussion 96 Seasonal Change at the Test Sites 96 Influence of Data Censorship and Taxonomic Levels 99 Implications for Future Users of the Predictive Model 102 Chapter 5 Comments and Conclusions 105 Summary 105 Recommendations 106 Literature Cited 109 Appendix 1.1: Body length data for invertebrates measured, but which did not fit the criteria used for selecting taxa included in Chapter 2 118 Appendix 2.1: List of taxa identified, functional group classification, and the class(es) of streams in they were found in. C P O M shredder (SH), herbivore shredder (HE), deposit collector (DE), filter collector (FI), grazer (GR), wood gouger (GO), parasite (PA), and predator (PR). * = taxon found in >25% of the samples from a specific stream class (i.e., common taxa). ! = taxon in less than 25% of the samples collected from a class of stream (i.e., rare taxa) 128 Appendix 2.2: Invertebrate taxa included in analysis at the genus/species and family level analyses and their respective class, order, and families 132 Appendix 3.1: Invertebrate taxa included in analysis at the lowest possible taxonomic level (A) and in the analysis at the family level (B), and two censorship levels (0.1% and 0.5%) 133 VI List o f Tables Table 2.1: Physical and environmental characteristics for each seasonal test site stream. Ranges are given for variables that changed seasonally. Stream order is based on a 1:250 000 scale map 13 Table 2.2: Environmental variables measured and the respective abbreviations (adapted from Rosenberg et al. in prep.) 18 Table 2.3: Growth rates between sampling dates for each taxa. For Drunella doddsi, 0.0191 day"1 is the growth rate for the period between October 1995 and April 1996. N/S indicates no sample 20 Table 3.1: Environmental variables used in the analysis 41 Table 3.2: List of taxa found in greater than 25% of the samples from a specific stream class (interior stream, coastal streams, or large rivers). See Appendix 3.1 for a listing of all taxa found in each stream class 43 Table 3.3: Invertebrate diversity at each sampling site and date based on Simpson's index of diversity (1-D). Index ranges from a low diversity value of near 0 to almost 1 54 Table 3.4: Percent variance (of explained variance) explained by season, site, and stream class in the PC ordinations as determined by MANOVA. The environmental data used in the ordination consisted of just those environmental variables that change seasonally 54 Table 3.5: Coefficient of variation used to summarise the seasonal variation in ordination space for each site. Coefficients of variation were determined for the ordinations of the genus/species, family, and functional data and for the ordination of the environmental variables, that change seasonally. The rank order of the first axis coefficients of variation for the genus/species and family data are correlated (Spearman's rank order correlation, rs= 0.9286, p < 0.005), but the rank order of the second axis is not correlated (Spearman's rank order correlation, rs = 0.2857, p > 0.5) 54 Table 3.6: Correlation coefficients (r) and significance values (p) for correlation of PC axes from the invertebrate ordination with the PC axes from the ordination of environmental variables, that change seasonally (a = 0.0125). Bold type indicates significant correlations 56 Table 3.7: Summary of the significant correlations between the seasonal change of the invertebrate community PCA scores and the seasonal change of environmental variables (a = 0.05). Abbreviation details can be found in Table 3.1. Astrix represent probability values *p>0.05, **p<0.05, ***p<0.01, ****p<0.001 58 Table 3.8: Summary of the environmental variables correlated with benthic community composition from this and previous studies. Abbreviations established in this paper for the environmental variables are used 60 Table 4.1: Environmental variables used in the discriminant function model and the rate of site misclassification (error rate). The misclassification rate is based on a cross-validation analysis 80 Table 4.2: Summary of the groups to which the seasonal test sites were predicted based on the autumn sampling, and where the seasons fall in the ordination space. Within the 90% probability ellipse (in), between the 90 - 99% ellipses (>90), between the 99-99.9% (>99), or outside of the 99.9% ellipse (>99.9) 82 V l l List of Figures Figure 2.1: Location of sampling sites in southwestern British Columbia. Al l eight sampling sites are located in the Fraser River Drainage Basin 10 Figure 2.2: Typical hydrographs for interior streams (Beak Creek - A)(source: Environment Canada, Water Survey), coastal streams (East Creek - B)(source: Dr.M. Feller, Forest Sciences,UBC), and large rivers (Thompson River at Spences Bridge - C) (source: Environment Canada, Water Survey). The dates the streams were sampled are indicated by the arrows 11 Figure 2.3: Size distribution of Drunella doddsi at each sampling date. Number of individuals measured is indicated. Thick bars represent standard deviation and thin bars represent upper and lower size ranges. C=coastal, I=interior, and L=large rivers 21 Figure 2.4: Size distribution of Drunella spinifera at each sampling date. Number of individuals measured is indicated. Thick bars represent standard deviation and thin bars represent upper and lower size ranges. C=coastal, I=interior, and L=large rivers 21 Figure 2.5: Size distribution of Zapada cinctipes from coastal (A) and interior (B) streams. Number of individuals measured is indicated. Thick bars represent standard deviation and thin bars represent the size range. Zapada cinctipes body lengths in Mayfly and Spring Creek are off-set for each season to aid identification 22 Figure 2.6: Relative abundance of Drunella doddsi in Beak Creek (A), Drunella spinifera in Mayfly Creek (B), and Zapada cinctipes (C) at each sampling date. Points for each season are slightly off-set because of the slightly different sampling dates 23 Figure 2.7: Average water temperatures for each of the three classes of streams (interior, coastal and large rivers) at each sampling date 25 Figure 2.8: Periphyton chlorophyll-a content for each stream at each sampling date. A=large rivers, B=coastal streams, and C=interior streams. Chlorophyll-a content is presented as wg/5 cm 2 of rock surface area 26 Figure 2.9: Average coarse and fine particulate organic matter content at each site and date. The average organic matter content was determined from the 3, one-minute kicknet samples collected at each sampling. A=large rivers, B=coastal streams, and C=interior streams 27 Figure 3.1: Seasonal pattern of insect abundance in small undisturbed streams as proposed by Hynes (1970) 35 Figure 3.2: Invertebrate abundance at each sampling site and date based on three one-minute kicknet samples. The numbers of taxa found at that site and sampling date is above the bars. Al l eight sampling sites are located in southwestern British Columbia. Note the scales differ. N/S=no sample, A=large rivers, B=coastal streams, and C=interior streams 44 Figure 3.3: (A) Average number of invertebrates collected in the composite three one-minute kicknet samples over all sampling dates. Bars represent + 1 SE. (B) Total number of invertebrate taxa in the composite three-minute kicknet samples at each sampling site over all sampling dates. I=interior streams, C=coastal streams, and L=large rivers 45 vii i Figure 3.4: Relative abundance of the major insect orders and other invertebrate in the composite three one-minute kicknet sample collected at each sampling site and date. L=large rivers, C=coastal streams, and I=interior streams 46 Figure 3.5: Relative abundance of functional groups in each composite three-minute kicknet sample collected from each stream and sampling date. Functional groups are as defined by Merritt and Cummins (1996) and Pennak (1978). L=large rivers, C=coastal streams, and I=interior streams 47 Figure 3.6a: Ordination plot of the seasonal samples for each site along principal component axes 1 and 2 based on log-transformed invertebrate data identified to genus/species, 1% censorship level 50 Figure 3.6b: Direction and magnitude of loadings for taxa with significant loadings along axis 1 and 2 of the ordination shown in Fig. 3.6a 50 Figure 3.7a: Ordination plot of the seasonal samples for each site along principal component axes 1 and 2 based on log-transformed invertebrate data identified to family, 0.5% censorship level 51 Figure 3.7b: Direction and magnitude of loadings for families with significant loadings along axis 1 and/or axis 2 of the ordination illustrated in Fig. 3.7a 51 Figure 3.8a: Ordination plot of the seasonal samples for each site along principal component axes 1 and 2 based on invertebrate data classified by functional group 52 Figure 3.8b: Direction and magnitude of loadings for functional groups with significant loadings 52 Figure 3.9a: Ordination plot of the sampling date and sites along principal component axes 1 and 2 based on environmental conditions which change seasonally (Table 3.1) 53 Figure 3.9b: Direction and magnitude of significant environmental loadings along axis 1 and 2 of the PCA. Abbreviations are listed in Table 3.1 53 Figure 4.1: Dendrogram illustrating reference site groups for genus/species level data (>0.1 % occurrence) 75 Figure 4.2a: Ordination of Fraser River Basin reference sites based on invertebrate data identified to the genus/species level (>0.1 % occurrence), stress level = 0.1859 76 Figure 4.2b: Direction and magnitude of environmental variables with significant loadings on the ordination of Fraser River Basin invertebrate data. The invertebrates were identified to the genus/species level (>0.1 % occurrence), stress level = 0.1859. See Table 2.2 for abbreviation explanations 77 Figure 4.2c: Direction and magnitude of loadings for 16 of the invertebrates that had significant loadings. Invertebrates were identified to the genus/species level, stress level = 0.1859. 1 = Chironominae, 2 = Orthocladiinae, 3 = Paraleptophlebia spp., 4 = Baetidae, 5 = Zapada cinctipes, 6 = Taenionema spp., 7 = Lepidostoma spp., 8 = Sweltsa spp., 9 = Capnia spp., 10 = Heptageniidae, 11= Zapada spp., 12 = Micrasema spp., 13 = Paraleuctra spp., 14 = Rhyacophila spp., 15 = Serratella spp., 16 = Drunella doddsi 78 Figure 4.3: Mean discriminant function scores for each reference group. Groups are based on cluster analysis of invertebrate data identified to the genus/species level, >0.1% occurrence (A) and >0.5 % IX occurrence (B). Triangles represent the discriminant function coordinates of the environmental variables used to discriminate between groups 81 Figure 4.4: Location of seasonal test sites in the ordination space of the group to which the sites were predicted. The sites were predicted to their respective groups based on the autumn (3) sampling of the test sites. The invertebrates were identified to the genus/species level, 0.1 % cut-off. A - the group 1 ordination space, B - the group 3 ordination space. The 90, 99, and 99.9 % probability ellipses indicated are based on the ordination scores of the reference data for the respective groups 84 Figure 4.5a: Ordination of Fraser River Basin reference sites based on invertebrate data identified to the genus/species level (>0.5 % occurrence), stress = 0.209. 86 Figure 4.5b: Direction and magnitude of environmental variables with significant loadings on the ordination of Fraser River Basin invertebrate data. The invertebrates were identified to the genus/species level (>0.5 % occurrence), stress = 0.209. See Table 2.2 for abbreviation explanations 87 Figure 4.5c: Direction and magnitude of loadings for 16 of the invertebrates that had significant loadings. Genus/species level, >0.5 % occurrence, stress = 0.209. 1 = Chironominae, 2 = Orthocladiinae, 3 = Baetidae, 4 = Zapada cinctipes, 5 = Paraleptophlebia spp., 6 = Lepidostoma spp., 7 = Taenionema spp., 8 = Zapada spp., 9 = Hydropsyche spp., 10 = Heptageniidae, 11= Drunella doddsi, 12 = Sweltsa spp Figure 4.6: Location of seasonal test sites in the ordination space of the group to which the sites were predicted. The sites were predicted to their respective groups based on the autumn (3) sampling of the test sites. The invertebrates were identified to the genus/species level, 0.5 % cut-off. A - group 1 ordination space, B - group 2 ordination space, and C - group 3 ordination space. The 90, 99, and 99.9 % probability ellipses indicated are based on the ordination scores of the reference data for the respective groups. 91 Figure 4.7: Mean discriminant function scores for each reference group. Groups are based on cluster analysis of family data, >0.1% occurrence (A) and >0.5 % occurrence (B). Triangles represent the discriminant function coordinates of the environmental variables used to discriminate between groups 93 Figure 4.8: Location of seasonal test sites in the ordination space of the group to which the sites were predicted. The sites were predicted to their respective groups based on the autumn (3) sampling of the test sites. The invertebrates were identified to the family level, 0.5 % cut-off. A - group 1 ordination space, B - group 2 ordination space, and C - group 3 ordination space. The indicated 90, 99, and 99.9 % probability ellipses indicated are based on the ordination scores of the reference data for the respective groups 94 X Acknowledgements I would like to dedicate this thesis to my mom Frances Dymond who was always there supporting, encouraging, and believing in me, and whom I miss dearly. I would like to thank John Richardson for being a great supervisor through this project. I really appreciated his comments, direction, and support. I would also like to thank my supervisory committee members Bill Neill, Trefor Reynoldson, and Dave Rosenberg for their feedback and insight. Many thanks go to the people who helped with my sampling, sample processing and invertebrate identification: Jill Melody, Jeremy Detman, Jessica Kaman, and Craig Logan. Thanks also go to the people who generously donated their time and effort to verify my invertebrate identifications: Art Borkent, Jeff Cumming, Doug Currie, Boris Kondratieff, and Rob Roughly. This project was funded by Environment Canada's Fraser River Action Plan and NSERC. I would also like to thank my sister and friend Caren, whose thoughtful comments and lively conversation kept me on my toes and helped me to focus my ideas. Thanks go to my husband Clint for keeping balance in my life. 1 Chapter One I N T R O D U C T I O N The Benthic Community The benthic community is the assemblage of species populations which occur and interact together in space and time on the bottom of streams and rivers (Begon et al. 1990). Community composition or structure is described by the number of coexisting species and their relative abundance. Benthic macroinvertebrate communities of streams and rivers change both spatially and temporally (Hynes 1970), largely in relation to environmental factors. Three theoretical frameworks predict the relationship between environmental factors and community structure. The habitat templet concept holds that ".. .habitats provide the framework on which evolution forges characteristic species traits.. ."(page 3) and influences community composition (Southwood 1977, 1988; Townsend and Hildrew 1994). The R C C describes the changing structure and function of communities along a river system in response to the changing abiotic environment and upstream processes (Vannote et al. 1980; Minshall et al. 1985). The patch dynamics concept examines the importance of temporal phenomena, history, and chance in organizing communities (Townsend 1989). Field studies and the use of predictive models have provided support and evidence that abiotic factors influence community composition. Community structure is known to be related to environmental factors (e.g., Corkum 1990; Richards et al. 1993; Tate and Heiny 1995). As well, models have been used for the successful prediction of community structure based on physical and chemical features of the environment (e.g. Wright et al. 1984; Reynoldson et al. 1995). However, the most influential abiotic factors vary from system to system and with the spatial scale of the study (Corkum 1990; Cobb et al. 1992; Tate and Heiny 1995). Abiotic factors that are related to community structure include discharge, substrate, dissolved substances, turbidity, riparian vegetation, land use, temperature, altitude, and latitude (Hynes 1970). 2 Many of the environmental conditions that are related to community structure change seasonally. Yet, surprisingly, few studies have related the seasonal change in community structure with changes in the environment (Giller and Twomey 1993). The benthic invertebrate community could change seasonally in association with changes in the environment and as invertebrates move through their life cycles (Hynes 1970). Previous studies have related the timing of life cycle events and size at emergence with temperature, photoperiod, food resources, and discharge (see review by Sweeney 1984; Poff 1989). Seasonal patterns of invertebrate abundance have been proposed and monitored (Hynes 1970; Boulton and Lake 1992; Bothwell and Culp 1993). Studies on the seasonal changes that occur in benthic communities and how they relate to the seasonal change of environmental conditions are needed. The limited amount of information on the pattern and magnitude of seasonal change in benthic communities creates a level of uncertainty when using the communities for biomonitoring. Biomonitoring and Benthic Macroinvertebrates Biomonitoring uses biological responses to measure changes in the environment. In streams, biomonitoring can be done by using benthic invertebrates, fish, algae, or more than one of these groups (Schindler 1987; Munkittrick and Dixon 1989; Norris 1994; Rosenberg and Resh 1993). However, benthic invertebrates have been most commonly used (Rosenberg and Resh 1993), and there are at least six reasons for this preference. First, they tend to move very little and are therefore representative of the area in which they are collected. Second, invertebrate life cycles are relatively long compared to other biota, so temporal changes in abundance and age-structure can be followed. Third, benthic invertebrates live and feed in, on, and around sediments which is where toxins accumulate and persist (Reynoldson 1987; Schindler 1987; Richardson and Levings 1996). As a result of the interaction between benthos and sediments, the benthos themselves may accumulate toxins and pass them up the food chain, or resuspend them in the water column through bioturbation (Reynoldson 1987; Reice and Wohlenberg 1993). Fourth, different taxa have differing sensitivities to stresses and thus provide a gradient of indicators to perturbation (Schindler 1987). Fifth, benthic macroinvertebrate taxonomy is reasonably well understood 3 and sample collection is relatively simple and inexpensive. Finally, benthic invertebrates are important components of the ecosystem. They are the primary food source of many fish, and play a critical role in breaking down organic matter and in nutrient cycling. Disruption of the benthic community will therefore likely disrupt the ecosystem. Benthic macroinvertebrate populations, or species assemblages, can be used in biomonitoring programs as indicators. When communities are used in biomonitoring programs, many different factors can be looked at, including: presence/absence or numerical predominance of indicator organism populations, diversity, and relative species abundance (Rosenberg and Resh 1993). Multivariate models for the prediction of benthic macroinvertebrate communities is a biomonitoring technique that utilizes a relationship between environmental conditions and community structure. Based on the relationship between the environment and community structure from a large number of unimpacted reference sites, multivariate techniques are used to make predictions as to the expected community at test sites. If community composition at the test site is different from that predicted, an environmental impact is inferred. Multivariate models are robust and accurate indicators of expected community structure (Wright et al. 1984; Reynoldson et al. 1995). However, accurate predictions are limited to the range of reference conditions included in the reference data set. A lack of knowledge of the extent of seasonal changes introduces a level of uncertainty into the accuracy of predictions made with multivariate models at times of the year other than when the reference samples were collected. Seasonality and Biomonitoring in the Fraser River Basin In 1994, under the Fraser River Action Plan, Environment Canada began an extensive program to assess and monitor the Fraser River Basin (FRB) using benthic macroinvertebrate communities, associated environmental conditions, and multivariate models. Reference samples were collected once per year - in the autumn - over three years. The effectiveness of multivariate models based on autumn samples in accurately predicting the expected community during other times of the year needed to be determined, because the community changes seasonally as invertebrates move through their life cycles. 4 Knowledge of the seasonal changes of benthic communities in the Fraser River would determine the degree of applicability of model predictions throughout the entire year. The FRB covers over 230,000 km 2 and consists of 10 different ecoregions, which have different climates, vegetation, and in some cases different land use and discharge regimes. The potential exists for the invertebrate communities to change seasonally at different rates with different predictable seasonal signals. Purpose This study addresses how the benthic macroinvertebrate community changes seasonally both taxonomically and functionally, the role of life cycles and environmental conditions in the seasonal change of community composition, and whether the seasonal changes are significant enough to affect the accuracy of predictions made with multivariate models. Benthic community structure and associated environmental conditions were monitored in eight streams from three different classes of streams (large rivers, coastal, and interior streams) each having different climate, riparian vegetation, land use, and discharge regime. For the purpose of this study, the benthic macroinvertebrate community is defined as the assemblage of invertebrate species collected in the benthic samples. Although this may not be the full community of interacting organisms as seen from an invertebrates point of view, it does include species populations which occur together in space and time and interact across trophic levels (Begon et al. 1990). The findings of this study are presented in three main chapters. Chapter 2 examines the life cycles of three invertebrates found in the different classes of streams. Seasonal changes in benthic communities and invertebrate abundance are attributed in part to invertebrate life cycles. The potential for life cycles to differ in the three types of streams because of different environmental conditions means that faunal communities may have different seasonal patterns. The influence on invertebrate life cycles of discharge regime, food resources, and temperatures is determined. Chapter 3 considers how the community changes seasonally and whether the seasonal changes can be attributed to life cycles and/or particular environmental conditions. The chapter also examines 5 how the community differs between the three classes of streams, and which environmental factors are related to the spatial change in community structure. Finally, the relative importance of spatial versus temporal change is considered. Both taxonomic and functional classifications are analyzed. Chapter 4 examines whether invertebrate life cycles and the seasonal changes in community structure described in Chapters 2 and 3 have implications for biomonitoring in the FRB. Faunal communities at the seasonal test sites are compared to reference faunal communities having similar environmental conditions. Two different taxonomic levels and two data censorship levels (level of occurrence below which taxa are removed from the data set) are analyzed to determine what effect these data-handling techniques have on the predictive model and conclusions drawn. Chapter 5 summarizes the findings of this study, and makes recommendations for future research. 6 Chapter Two L I F E C Y C L E S O F T H R E E L O T I C INSECTS L I V I N G U N D E R T H R E E D I F F E R E N T D I S C H A R G E R E G I M E S Introduction Ecology attempts to understand the relationship between the complex array of organisms and the environment (Begon et al. 1990), and it has been theorized that the physical environment provides a templet on which evolution forges characteristic life history strategies (Southwood 1977 1988; Chesson 1986; Townsend and Hildrew 1994). Environmental characteristics involved in habitat templets and thus related to life histories, include 1) habitat favorableness and disturbance frequency and intensity (Southwood 1977,1988), 2) the relative importance of stress, disturbance and competition (Grime 1977), and 3) temporal and spatial variation (Townsend 1989; Townsend and Hildrew 1994). Life history characteristics such as life cycle length, time to first reproduction, number of reproductive events, and number and size of offspring adapted to environmental conditions constitute appropriate ecological strategies for persistence in a habitat. For benthic macroinvertebrates environmental conditions influence life history parameters such as rate of larval growth, size of maturation, and the timing of life cycle events (review by Sweeney 1984). Stream discharge influences many important structural attributes in streams (e.g., channel geomorphology and substrate stability) and varies temporally and spatially within and between lotic systems. Both unusually high and low levels of discharge constitute sources of disturbances in streams. Bankfull discharge is the level of discharge required to move dominant substrates and shape the stream channel (Newbury 1984), creating disturbances in streams (Poff 1992). Bankfull discharges and associated disturbances have a return time of 1-2 years (Newbury 1984). The return time of seasonal flooding and bankfull discharge, the associated disturbances, and discharge variability have the potential to influence benthic invertebrate life cycles. The predictability and variability of discharge, and resulting disturbances, may influence the timing of benthic invertebrate life cycle events (Resh et al. 1988; Poff and Ward 1989). Resh et al. 7 (1988) proposed that benthic invertebrates are adapted to predictable seasonal changes in discharge, but data are still needed to determine whether this is true (Poff and Ward 1989; Poff 1992; Robinson et al. 1992). Poff (1989) identified potential biological and life cycle attributes of lotic invertebrates under different levels of discharge variability and predictability. Adaptations to minimize exposure to predictable seasonal scour associated with bankfull discharge include modifications to life cycles, behavioral responses, or both. Life cycle tactics to avoid the seasonal scour may include alterations to the timing or duration of resting, egg, or adult life cycle stages. Modifications to life cycle timing wil l vary depending on the intensity, frequency and predictability of hydrologic disturbance events (Resh et al. 1988; Poff and Ward 1989). As a result, there is the potential for invertebrate life cycles to differ, within a species, when populations inhabit streams with different discharge regimes. A few studies have looked at the influence of discharge regime on invertebrate life histories and characteristics. For example, life history traits of the snail Juga plicifera are influenced by stream size (Diamond 1982), and thus habitat predictability (Leopold 1962; Ward and Stanford 1980). Snail populations from large streams with less variable discharge, had low individual fecundity, high individual survivorship, and high individual biomass. In contrast, snail populations from small unpredictable streams had high individual fecundity, low survivorship, and low individual biomass (Diamond 1982). Greater genetic variability occurs in invertebrates living in streams with more variable discharge regimes (Robinson et al. 1992). In southwestern British Columbia (Fig. 2.1) there are three different classes of streams with discharge regimes of different predictability. In the interior streams, bankfull discharge occurs with snowmelt in the spring (see Fig. 2.2A). The discharge level achieved and the duration depends on the depth of the snow pack and the speed of melting. The second characteristic discharge regime, found in coastal streams, is characterized by high and variable winter discharge caused by heavy winter rains (Fig. 2.2B). The winter rains begin in late September and last through January and February; bankfull discharge is typically achieved during this period. The third discharge regime, found in large rivers, is quite predictable and is characterized by a unimodal freshet that lasts from early April until late August 8 (Fig. 2.2C). The long freshet of large rivers is a result of the slightly different timing of snowmelt in subdrainage basins. The life cycles of benthic invertebrates common to these three classes of streams may differ if the invertebrates have adapted their life cycles to the seasonal change in discharge. The three stream classes used in this study also have different climates, geology, and riparian vegetation. As a result, the streams differ in water temperature, chemistry, nutrients, primary production, and the type and amount of organic matter inputs. Environmental variables such as temperatures, in-stream primary production, and allochthonous inputs influence life cycle timing, survival rates, and growth rates of benthic invertebrates (review by Sweeney 1984). Temperature may be the single most important factor influencing egg development, growth rates, and timing of emergence (Corkum 1978; Vannote and Sweeney 1980; Sweeney 1984). The quality and quantity of food resources also affects growth rates and size at maturation (Anderson and Cummins 1979). In natural systems, it can be difficult to distinguish between the relative importance of temperature and food resources on growth rates (Sweeney and Vannote 1981; Hawkins 1986). The purpose of this study was to examine the timing of benthic invertebrate life cycles in the three different classes of streams southwestern British Columbia. According to the classification scheme developed by Poff and Ward (1989), the three classes of streams are all perennial and have high flood predictability and low flood frequency. As such, the invertebrates should have synchronous development, and emergence and reproduction should be temporally cued to floods for a given stream type (Poff and Ward 1989). There are two potential strategies the invertebrates may adopt to cope with the seasonal flooding and associated disturbances. First, invertebrates could use emergence periods, or egg stages to avoid seasonal flooding, i.e., they would be absent from the stream during seasonal flooding (Gray 1981). Second, invertebrates may be mobile larvae using behavioural tactics to deal with the scour(Gray 1981). The timing of invertebrate life cycles could also differ between stream classes given the different climates, water chemistry, allochthonous and autochthonous inputs. Invertebrate abundance and growth 9 rates should be greatest in the summer and autumn because warm temperatures, high primary production, and the greatest input of organic matter should occur at this time. Study Sites Benthic invertebrate samples and associated environmental data were collected from three interior streams (Mellin Creek, Glimpse Creek, and Beak Creek), three coastal streams (Spring Creek, Mayfly Creek, and the North Alouette River), and two large rivers (Fraser and Thompson Rivers) all located in southwestern British Columbia (B.C.) (Fig. 2.1). One riffle was sampled in each river. Each study site was chosen because it was the site of previous research, (Spring Creek, Mayfly Creek, Fraser River, Thompson River) or the site was accessible and the stream maintained flow year round, (North Alouette River, and the interior streams). See Table 2.1 for the environmental characteristics of each stream. Interior Streams: The interior streams are located in the Nicola River watershed. The climate of the Nicola drainage basin is arid with an average annual rainfall of 223 mm and snowfall of 88 cm at Merritt (Environment Canada, Historical and Statistical Climate Information). Summer temperatures reach 35C and winter temperatures dip to -30C. Beak Creek was inaccessible in the winter, and could not be sampled. Highest discharge occurs in the spring as a result of snow melt (Fig. 2.2A). Through the summer, there is very little precipitation and many streams in the area become intermittent. The three streams chosen for this study maintained some flow throughout the entire year. The interior streams have the highest elevations of all the streams sampled (Table 2.1). Agriculture is the dominant land use in the Nicola Drainage Basin and most streams, including those sampled for this study, are impacted by cattle. Glimpse Creek is less impacted by cattle than Mellin and Beak Creek. The riparian vegetation for all three interior streams consists of Pseudotsuga menziesii (Douglas fir), Picea glauca (white spruce), and Picea engelmanii (Engelmann spruce). Deciduous Figure 2.1: Location of seasonal test sites in southwestern British Columbia ( ). Al l eight sampling sites are located in the Fraser River Drainage Basin. -Average discharge from 1989-94 -Discharge for 1995 and the beginning of 1996 iiiiiiiiiiiiiiiiiiiiiiii I I I in I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I in inn in I I MI I I I I I I I in in I I in I I in i cd PH '3 M Date < i tN CH I OO CN o I oo tN > o I tN o Q • CN NO CN (D PH I in CN NO CN i in CN Figure 2.2: Typical hydrographs for interior streams (Beak Creek - A) (source: Environment Canada, Water Survey), coastal streams (East Creek - B) (source: Dr.M. Feller, Forest Sciences, UBC), and large rivers (Thompson River at Spences Bridge - C) (source: Environment Canada, Water Survey). The dates the streams were sampled are indicated by the arrows. Note the scales differ. 12 vegetation at Mellin Creek consists of Salix spp. (willow) along with abundant grasses and herbs, and at Beak Creek it consists of Alnus spp. (alder) and very few grasses or herbs. At Glimpse Creek there is almost no deciduous riparian vegetation. Parent rock surrounding the interior streams consists of sedimentary rocks, in particular limestone (Geological Survey of Canada, 1969). The extreme temperatures in the interior increase the rate of weathering of the sedimentary rocks and result in hardwater streams with high alkalinity, high conductivity, and a slightly basic pH (Table 2.1). Coastal Streams: The three coastal streams sampled are located in the Malcolm Knapp Research Forest in the Pitt River catchment. This area is a second growth forest with a closed canopy and some logging in the watersheds. The coastal climate consists of heavy winter rains and correspondingly high and variable winter discharges (Fig. 2.2B). Reduced levels of precipitation through the summer result in lower and less variable discharge. Average annual precipitation at the entrance to the research forest consists of 2016.6 mm rain and 73 cm snow (Environment Canada, Historical and Statistical Climate Information). Air temperatures can range from 30C in the summer to -10C in the winter. The dominant vegetation in the area consists of Tsuga heterophylla (western hemlock), Thuja plicata (western red cedar), and Douglas fir. Red alder (Alnus rubra), Oplopanax horridum (Devil's club), and a variety of ferns can be found along the stream banks and in the understory. These streams have the highest gradients of the three classes of streams and are at intermediate elevations (Table 2.1). The parent rock surrounding the coastal sites consists of quartzite (Geological Survey of Canada, 1957 and 1969). The hard rocks and temperate coastal climate result in very slow weathering of the parent material, so the streams have soft water, low alkalinity, low conductivity, and are slightly acidic (Table 2.1). Large Rivers: The two large-river sampling sites were the Thompson River at Spences Bridge, and the Fraser River at Agassiz. Air temperatures at these sites range from 30C in the summer to -20C in the winter. 13 1-3 c o CO o 3 ia •2 l 3 CD ti CD 3 O < 03 ft oo T* CQ .. wo 0 0 C O I D C N O N C N C N T T o o o Os Tt ^ M so in ,—1 SO 0 0 t--Tt o I D CN Tt CN r—1 CN o I D co CN I D © 1 CN CO SO • oo T - H -8.6 T - H 1 CN Tt i 2-0.1 9-0.1 6-0.0 -3.4 i • I D Os O 1 oo co CN co NO o NO o O CO T 1 I T ) I D CO ON O © © O N o o o Tt © CN d d d d C N C N d C N „ - NO C N Tt ON C N 0 0 SO d C N CO 0 0 ID T—I CO C N OS ° C N ID O N C N Tt ON d CN o 0 0 Tt OO q OO ON NO NO • D ON CO T-H o 2.51 CN TT • Tt I D CN I D d • d • d • NO o 1 I D • * I D 0 0 oo ' o /—s i CN SO o T I Tt sq CN o Ti CN NO o O Tt CN i ON I D ON o ' I o Tt d CO d d 1 o NO ON CO d d d ON Tt r- I D Tt 0 0 r—1 r—1 0 0 CO CN Tt t-- o co q ,_; NO I D Tt o o CN o o I D d • • 1 NO ,_ NO i CN d • NO d • d • o i Tt i T—I co CO T—' I D l 1 ON co o NO q 0 0 CN Tt Tt 1 NO CO o O o o d cn d d d CN I D d d d d d C N NO CN o 0 0 CN 0 0 —^i- CN © NO o Tt 0 0 so ID NO CN ' CN i — I ON © CN 0 0 d d d d d q © Os CO O CO • © 1 ID CN I o CN NO i CN ' Q CN CN co Tt o 1 T—1 oo Tt T—' i— l CN i co Tt © ON ON i o ON 1 o o 1 ID Tt o © © CN Tt Tt T 1 CN d CN d d d CN CN d © © © E Tt n d o „ - Tt NO CO T I o OS ° C N C N ON C N CO O N ON (XJ ON CO ' o\ q i Tt Ti Ti Ti Ti © CO Os C N co o d d CO CO C N © © Tt © © © oo CO CO ON o © H '—1 ON © C N NO co C N © © C N T—1 i C N © • i T 1 Tt CO CO © OO CO SO ON © © © cn NO Ti Tt © d d © C N NO 0 0 CD © ID ON CO © © C N C N NO © ID Tt © © Tt ID Tt NO O co 1 1 od CN oo NO I 1 • l Tt H 1 © ID Tt i Tt C N NO © oo "—1 © Tt ID d C N oo d NO d d d d CO SO q © ID © © 0 0 Tt C N Tt CO NO q I D © CO 0 0 1 C N NO © ID CO CO ON C N 0 0 T-H ID ft C N C N Ti CO C N g „ Tt ID 0 0 © © ID od ID T-H Ti C N 1 1 od 1 O f~N £ 0 0 C N © oo Tt © ON © C N d © CO © © © NO •D Ti Ti i—i i—i © 1—> © © © © ON Tt CO Tt Tt CO CN © C N CO C N 0 0 TH OO CN Ti SO Ti r- O CO O O CN © CN Ti ID Ti Ti ,—i co Tt so 0 0 Tt ON ID CN Tt ID © d Tt od SO ON NO CN i ID © ' 1 CO CN ID ON © 1 Ti Ti CN CN O q q © CN CN © CN ID d © © © © © t-» Tt o o o — 1 r» SO ID Tt Ti ID © CO © © © © © CO ID NO i © 0 0 © ID © CN © CO d © © *"< Tt Os r~ OO ON © SO © CN © © d co ID ID • © 0 0 Tt ro o Tt q CO d © © Tt o C J O CD CO 1U C5j t s. o 111 • 1 1 T> C C J j -S T2 ? T« s s 5 cd u CD ^ & O U a TH u 7 1 M cd a ^ > is s C • T « cc! X o CD C3 .co 6 T - J T J W O Q O O S S Q H u O I D C N i» s "o T C a • H > • H 1 » o c o o 0 0 S o o ca O • i-H C • l-H 6 0 s o ro O bo 0 0 S3 co O T J CO O 13 TS ^ ft J3 SS ft 2 « &, CO •*-» o o H H 14 Land use around Spences Bridge and Agassiz consists of agriculture, some rural development, and forested areas. There are major urban centers and pulp mills upstream of both sampling sites. The geology surrounding the Thompson River sampling site consists of undivided sedimentary and volcanic rocks (Geological Survey of Canada, 1969). The geology surrounding the Fraser River sampling site consists of recently deposited alluvium and glacial deposits (Geological Survey of Canada, 1969). The water at these two large river sites is probably influenced more by the various parent materials and land uses in their whole drainage basins than by the immediate surrounding area. The Fraser and the Thompson have an intermediate alkalinity, and a conductivity and pH similar to the interior streams (Table 2.1). Methods Benthic invertebrate samples and environmental data were collected from all eight streams. The samples were collected over 5 sampling dates spread between April 1995 and April 1996 (Fig. 2.2). The sampling dates were chosen to bracket the major discharge periods. Benthic Invertebrate Life Cycle and Abundance Data Benthic invertebrate samples were collected using triangular kicknets (38.5 cm for each side and 400-wm mesh). Five one-minute kicknet samples were collected at each site and date, in the large rivers, this meant starting at the river edge and moving into the river channel for one minute. In the small streams however, sample collection was achieved by kicking from stream bank to stream bank and then back until one minute had passed. Efforts were made to keep sampling intensity consistent between sites and sampling dates. Samples were taken at least three days after a high discharge event to avoid redistribution of animals and changes in sampling efficiency. The samples were preserved in 4% formaldehyde and then taken to the lab for sorting and identification. Three randomly chosen samples per site per date were processed. 15 If a sample contained less than 500 invertebrates the entire sample was sorted and all the invertebrates were identified. Samples containing more than 500 invertebrates were subsampled (Marchant 1989). The sample was poured into a 35x35x10cm box divided into 100 equal cells. A lid was placed onto the box, and the box was shaken to distribute the sample evenly among the cells. Cells were randomly emptied and the contents were sorted until 200 invertebrates were collected. The cell containing the 200th invertebrate was completed. Organisms in the subsample were identified to the lowest possible taxonomic level, numbers were extrapolated to a full sample. Rosenberg et al. (in prep.) compared the results of subsampling 100, 200, and 300 invertebrates to sorting entire samples, and found no significant difference between the total number of invertebrates estimated and the actual numbers in the whole sample. Rosenberg et al. (in prep.) found no significant difference in the number of families estimated between the subsamples, but observed that fewer families were estimated by the subsamples than were in the entire sample. However, subsampling only missed rare families. Subsampling 200 invertebrates captured all the families that represented >5% of a sample and collected 46 % (± 36%) of the families that represented <1% of the sample (Rosenberg et al. in prep.). Because only taxa common in at least two stream classes were useful for this study, the loss of the rare taxa from the data set was not considered to be a problem. Material not used in the subsample was split into two size fractions (greater and less than 2 mm) and any invertebrates greater than 2 mm long were collected to aid in identification of the smaller invertebrates. Quality assurance /quality control was maintained by resorting and reidentifying invertebrates from the first 20 samples processed, and then randomly selecting 10% of the remaining samples for reexamination. The resorting of the QA/QC samples was carried out by a different person than first sorted the sample. Sample specimens of almost all invertebrates identified were verified by experts (when possible), and a voucher collection was given to the Royal British Columbia Museum, Victoria. Kicknet samples of organic matter permitted an estimate of the type and quantity of food resources available for detritivores. Organic material from the subsamples was split into three size 16 fractions - <lmm (fine particulate organic matter [FPOM]), l-2mm (coarse particulate organic matter [CPOM]) and >2mm [CPOM]. The organic matter content of each size fraction was determined by drying it to a constant weight, weighing it, ashing it at 540C for one hour, and weighing it again. The total organic matter content of the >2mm material remaining after subsampling was also determined so a total >2mm C P O M content could be determined. The length of each invertebrate was determined using a dissecting microscope, digitizing tablet, and the Zoobenthos Biomass Digitizing Program (Hopcroft 1995). Weights were estimated using the length/weight regression slopes and intercepts, for the respective families included in the Zoobenthos Biomass Digitizing Program. Growth rates between sampling dates were calculated using (from Hawkins 1986): G = In (W t / W„) t Where: G = growth rate (day"') W t = weight at time t (mg) W 0 = weight at time 0 (mg) t= time interval (number of days). Invertebrate size, growth pattern, and wing pad development were used to determine invertebrate life cycles. Growth rates were used to determine the maximal growth period for each invertebrate in relation to the different environmental parameters. Correlation analysis (Proc Corr; SAS Institute 1996) was used to determine if invertebrate growth varied with allochthonous or autochthonous resources or water temperatures. The invertebrates included in this study had to meet a number of criteria: 1) taxa had to be found in at least two of the three classes of streams; 2) they had to be identifiable to the species level; 3) they had to be abundant enough to determine what their life cycle was; and 4) univoltine taxa were preferred to facilitate identification of the life cycle. Only three taxa fulfilled the criteria: Drunella doddsi and D. spinifera (Ephemerellidae: Ephemeroptera) and Zapada cinctipes (Nemouridae: Plecoptera). The life cycles of other taxa not included in this study can be found in Appendix 1.1A to I. 17 Environmental Conditions Environmental variables were either measured when the benthic samples were collected or were determined from maps (Table 2.2). Measurement procedures for hydraulic variables such as drainage basin area, distance from source, stream order, stream channel slope, bankfull width, stream width, and discharge, were taken from Newbury (1984) and Newbury and Gaboury (1993). Drainage basin area and distance from source were calculated from 1:50000 scale maps by digitizing. Stream order was determined from 1:250000 scale maps. Slope, bankfull width and channel width were measured at each site. Discharge (m3/s) for the small streams was determined by multiplying current velocity (m/s) with area (m2). Current velocity was measured at regular intervals along the channel width and at 60% of the depth (site of average velocity for a given area) with a Marsh-McBirney electromagnetic current velocity meter. Discharge measurements for the Fraser and Thompson rivers and the annual discharge hydrograph for Beak Creek were supplied by the Water Survey Branch of Environment Canada. Environment Canada has gauging stations at the mouth of Beak Creek (200 m downstream of the sampling site used here) on the Thompson River at Spences Bridge, and on the Fraser River at Hope. There are no major tributaries to the Fraser River between Hope and Agassiz (Fig. 2.1), so there is less than 2% difference in discharge between Hope and Agassiz (Environment Canada, Water Survey). The discharge of a tributary of Spring Creek, East Creek, is continuously recorded, and was used to represent the coastal stream hydrograph (Michael Feller, Forest Sciences, UBC, unpublished data) (Fig. 2.2). Periphyton samples were collected for chlorophyll-a content and periphyton biomass using the protocol adapted by Environment Canada (Rosenberg et al. in prep.). Three rocks from the stream channel were selected at each sampling site. In small streams, one rock was selected from each of the right side of the stream, the left side, and the center while proceeding in an upstream direction. At large river sampling sites, the rocks were selected from the shallow, intermediate, and arms-length depths while proceeding in an upstream direction. Thus, bank-to-bank and upstream-downstream differences CU IS fa. CD C3 16 s e a fa. ~*3 C3 fa. a 5 •3 CJ IF cu •a 43 T3 C O o o Q 6 £ g §3 S T3 > T3 T3 O O CJ c o '5b <u i-c o o CJ CJ 3 3 o PI o o 11 o ^ o--2 o o £ B '3 -O rH oo S • 3 H^ C O & £ 3 « CJ 3 PH J 3 „ cd Q £ B P. CJ T3 (3 cd X cj cd E a [3 P. CJ CJ 00 cd >H CJ > o o >. PH O u 60 C o c o cd -P» PI CJ -5 .2 13 CJ 00 43 CJ O u o cd cj 60 cd u CJ > O CJ CJ & PH O |H o cd E CJ T3 E cd cj CO 3 C O I cd cd 60 O CJ C o N u g PH CJ a o N PI cd 'S PH c P! 60 C O -5 PQ Q IT? ^ I H o -3 s cj cd i i S > ^ ca o o CN s CN CJ cd 43 cj 60 cd C cd & CJ PI o N I cd PH cd CJ o o co E o cfc) CJ o C cd Pi o & & "5 =* Z cj CJ CU C Pi CJ CD PH PH C O C O 3 3 o > rH E £ I cd X ^ ^ cj cd o r^ cd o '-*-> •c CJ -§ CO OO cj C o .JS cd CJ 60 CJ > CO 3 O o cj PI CL) 60 O a xn xn H CO 43 cd T3 r4 Id o co T3 CJ PI CJ PH co CO T3 (U T3 PI U PH CO 3 CO PI xn cd (J I 3 > / \ o M 2 E 4 ^ T3 40 CO in ork (mal dne slop mew< (mal CU mew< trix mew< trix be en fra: ma em cd B B B 60 cd "cd "cd 13 -b -b 43 to CO CO 43 43 43 cd ns ns SU 43 o CJ 60 CJ > CO 3 O 3 T3 'o CJ T 3 T3 t3 19 were accounted for. The rocks selected were representative of the channel substrate. Rocks selected were larger than base stream flow could move, so they were stable. Periphyton samples from the three rocks at a site were combined into one sample, which was taken to represent the primary production for that site and sampling date. The periphyton sample was collected by placing a cylindrical template (11.3 cm2) on the rock and removing the periphyton from within the template boundaries using a toothbrush attached to a drill bit and an electric drill. The composite sample from the three rocks represents a periphyton sample from 34 cm 2 of rock surface. Stream water was added to the periphyton sample to bring the volume up to 680 ml. This solution was blended to create a homogenous mixture. Twenty milliliter subsamples of the solution (1 cm 2 of rock surface area) were then filtered for chlorophyll-a and periphyton biomass analysis. The periphyton samples were dried, and frozen until analysis. Chlorophyll-a and biomass samples were analyzed according to Stainton et al. (1977). Results Invertebrate Life Cycles, Abundance and Growth Rates Drunella doddsi, D. spinifera, and Zapada cinctipes were not equally abundant in each stream or between the different stream classes. Drunella doddsi was most abundant in Beak Creek and was found only occasionally in the North Alouette River, Spring Creek, and in the Thompson River (Fig. 2.3). Drunella spinifera was most abundant in Mayfly Creek and a few specimens were found in the Thompson River and in Beak Creek (Fig. 2.4). Zapada cinctipes occurred in all six small streams, but was only abundant enough for analysis in four of them, and was not found in any of the large river samples. The limited data from the other two streams (Beak Creek and the North Alouette River) were consistent with the data in Figures 2.5A and 2.5B. In the interior streams, Drunella. doddsi emerged in early July, had begun hatching by early July, and had its greatest growth rate between July and October (Fig. 2.3 and Table 2.3). In the Thompson River, D. doddsi appeared in July and was not collected again. In the coastal streams, Drunella doddsi 20 most likely emerged a little earlier and hatched a little earlier than in the interior streams, so D. doddsi in the coastal streams were slightly larger at each sampling date (Fig. 2.3). In the coastal streams, Drunella spinifera was both emerging and hatching in July (Fig. 2.4). Its greatest growth rate was during the period between July and October (Table 2.3). The size ofD. spinifera was similar in the large rivers, interior streams, and the coastal streams at each of the sampling dates (Fig. 2.4), so it appears that the life cycle and its timing of D. spinifera is similar in all three classes of streams. In the coastal streams, Zapada cinctipes began emergence in December, and had begun hatching prior to late March/early April (Fig 2.5A), and had their highest growth rate between October and December (Table 2.3). In the interior streams, Zapada cinctipes did not emerge until late March/early April, hatching was concurrent (Fig. 2.5B), and they had their highest growth rates between July and October (Table 2.3). Table 2.3: Growth rates between sampling dates for each taxa. Vox Drunella doddsi, 0.0191 day"1 is the growth rate for the period between October 1995 and April 1996. N/S indicates no sample Taxa Site Growth Rate (day'1) May - July July - October October - December -December April Drunella doddsi Beak 0.0163 0.0416 0.0191 D. spinifera Mayfly 0.0172 0.0238 0.0067 .0.008 Zapada cinctipes Mayfly -0.0018 0.0295 0.0405 N/S Z.cinctipes Mellin N/S 0.0424 0.0187 0.0057 The abundance of each taxon varied with stream, sampling date, and life cycle stage. Abundance was greatest during the July sampling for all three taxa and in all three classes of streams (Fig. 2.6). Abundance of Drunella doddsi and D. spinifera was greatest during the period when emergence and hatching coincided (Fig 2.6A and B). Abundance of Zapada cinctipes was greatest after most of the eggs had hatched (Fig. 2.6C). 21 w 5 °i VD I . . . .1 CN CN TT I • 1 O ro vo g •S & cd 43 O 2 * 43 o cd -»-» cd fa. CD ^ £ § .a S-f .g >3 co CD -r) « CD CD bO 3 cd u CD N CM o CN CU fa. 3 TO fa T3 CD T3 g M CD P. P. co 3 i-i C - M CD C M fcj) CD ,cd fi co •2 13 3 3 4? 3 B .. .5 CD C M N O cd i_l M • « 43 ra T3 O 5 'C " * CD § .s T3 52 cd O -a cj T f - r — « -3 O -^v CD CJ £ S-2 6 ^ U < 4 ? s vo ro T f ON CD / 0 0 Q vo CN (UIUI) mSiraj Apoq trespvj 43 O cd - M bO 3 a CD •3 C O i—i CD P. M a a 5 <D C O M cd cd 43 CD _v Id H ^ -2 s .S 2 .53 T3 CD M 3 co cd CD s o c o * M 3 B~ -3 cd co M . "3 3 CD > ro CN <u fa. 3 CM O CD bO § M CD N CD O M CD P. P. C O 3 M C - M CD C M bO P- s CD cd M — ' cd J 43  43 •4—' 3 cd 3 O T3 u M " O ' C CD 3 w fir T2 CD h T3 CJ CN tt 'E C O o cn CD .g ft ?! So l-^-l O N OO ~i r r- N O n r C O T f CN XT CD Q o r—I I -I C3 CA 03 O U T t CN I ' I . B . 60 ! ft i C O o C O i n CN ~ i r V O C O I CN CD 0 Q ' f a , CD _P 3 C CD H B CD -b CO m i -o 'C CD CD OJQ a cd rH CD N CD HC a CD CO J J CD -g fe a co cd O a § -a. a o a cd o cd CD O C+H t s •5 •§ C3 N <+H o a o '-4-» a 5 '> _ co T J f+H ^H ^ cd O § S3 to ^ CD a a CD CO C J CD tH 60 ft a CD • £ tH C O r O ^ r ^ C3 .2 >, H >> . cd T J ^ CU f^ H c-> rS 60 C O g .a <u CO ' 1 • 1 CD O 3 - ° cd cu CD • O ( N CD >-3 -*-» c^o K Id o > CS • S N (uiui) mSuaj Xpoq wzsy^ < Sampling date Figure 2.6: Relative abundance oi Drunella doddsi in Beak Creek (A), D. spinifera in Mayfly Creek (B), and Zapada cinctipes (C) at each sampling date. Points for each season are slightly off-set because of the slightly different sampling dates. 24 Environmental Conditions The timing and amount of discharge during the study year were consistent with the average discharge for the previous four - five years (Fig. 2.2); slight differences are a result of year to year variation. Discharge for the fall and winter months (October to January) in the coastal streams appears to be quite different between 1995 and the 4 previous years because of how the data are presented (Fig. 2.2B). The discharge is quite low when the slightly different timing of the spates during the variable fall/winter discharge is averaged out. However, the timing and the magnitude of the variable fall/winter discharge in 1995 is typical of previous years (Michael Feller, Forest Sciences, U B C , unpublished data). The higher-elevation interior streams were generally cooler than the coastal streams and large rivers (Fig. 2.7). The interior streams were frozen over by November and retained some ice cover until mid-April. Although neither chlorophyll-a content or biomass was significantly correlated with water temperatures, for most streams, the greatest amount of biomass and chlorophyll-a coincided with the warmest water temperatures and the greatest amount of light. Both periphyton biomass and chlorophyll-a were greatest in July and October in all streams, except in the Thompson River and Mayfly Creek (Fig. 2.8). The large amount of algae in Mayfly Creek in the April 1995 sample is probably sampling error caused by inclusion of moss in the sample. In the Thompson River, algal biomass and chlorophyll-a were greatest in the early spring (March 1996) - just before the predictable seasonal increase in discharge, which scours the channel bottom. Typically, the interior streams had the highest amount of both C P O M and F P O M (Fig. 2.9) and the large rivers had the least organic matter. The winter sampling generally produced the least amount of organic matter and spring samplings generally produced the most (Fig. 2.9). Growth rates for each taxon were not significantly correlated with allochthonous food resources (CPOM or FPOM), autochthonous food resources (chlorophyll-a or periphyton biomass), or temperatures (a c = 0.05). Figure 2.7: Average water temperatures for each of the three classes of streams (interior, coastal, and large rivers) at each sampling date. 26 1800 - , 1600 / 400 200 A o 400 - , ^ 350 H c CD -4—» o o c o p., 'S 300 H 250 B V A / 50 H 0 Fraser Thompson Chlorophylls Spring Mayfly N. Alouette 300 - , 250 -200 -150 -100 -50 -0 - m m JSL CO Mellin PV I p^  I s 1 Glimpse 4? Beak ~\ «_ *-. o P^  Figure 2.8: Periphyton chlorophyll-a content for each stream at each sampling date. A = Large rivers, B = coastal streams, and C = interior streams. Chlorophyll-a content is presented as um I 5 cm2 of rock surface area 27 5 4 -3 -CPOM(>lmm) | | FPOM (<lmm) Fraser Thompson B — 5o3 Spring Mayfly N. Alouette 15 -10 -5 -0 N A Mellin Glimpse Beak Jo -9 o- / 4f o° / * Figure 2.9: Average coarse (CPOM) and fine particulate organic matter (FPOM) content at each site and date. The avergae organic matter content was determined from the 3, one-minute kicknet samples collected at each sampling. A = large rivers, B = coastal streams, and C = interior streams. 28 Discussion Influence of Discharge Regime on Life Cycles Differences in the timing of emergence, egg hatching, and growth rates appeared to be unrelated to discharge regimes and therefore cannot be explained by the different regimes. Both Drunella doddsi and D. spinifera emerged and hatched in the summer regardless of the discharge regime of the stream they were inhabiting. Zapada cinctipes emerged either during or prior to peak discharge depending on whether they were from coastal or interior streams. The timing of emergence for all three taxa was similar to the range of emergence periods found in other studies regardless of the discharge regime (Allen and Edmonds 1962; Bradford and Hartland-Rowe 1971; Hawkins 1986). There are five potential reasons why D. doddsi, D. spinifera, and Z. cinctipes do not have life cycles adapted to the three discharge regimes in this study. First, there is the potential that the predictability and frequency of the seasonal flood events are not sufficient to promote local adaptation (Poff 1992). Second, other environmental constraints may limit life cycle adaptations (Poff and Ward 1989). Third, these three invertebrates may be generalists - they are able to colonize and survive in any environment. Fourth, the invertebrates may be using behavioural adaptations rather than life cycle ones to avoid the seasonal scouring. Finally, bankfull discharge associated with seasonal flooding may not accurately represent disturbance events at a level meaningful to benthic invertebrates (Townsend et al. 1997). The three discharge regimes included in this study occur naturally in southwestern British Columbia. The timing of seasonal floods caused by snow melt is predictable, although the magnitude and duration of flooding will depend on the depth of the snow pack and the timing of spring rains. The months in which seasonal floods caused by winter rains occur are also predictable. However, the exact timing of flood events within those months cannot be predicted. Seasonal flood events occur once per year or once per generation for D. doddsi, D. spinifera, and Z. cinctipes, and they occur over a 2-4 month period. The frequency and the duration of these presumably harsh conditions should create an environment in which life cycle adaptations would be beneficial. Because life cycle timing was not 29 adapted to the discharge regimes, there must be constraints on the ability to shift timing, or the invertebrates must be using other strategies to deal with the seasonal scour. Environmental conditions such as photoperiod and temperature may regulate the timing of life cycle events including hatching and emergence. Seasonal environmental conditions such as ice, low oxygen levels or food shortages may constrain life cycle adaptations. Thus, adaptations to other environmental signals such as the discharge regime could be limited. This is a plausible explanation for the lack of adaptation in southwestern B.C. The invertebrates included in this study and in other life cycle studies of the same taxa, all have similar emergence and hatching periods (Allen and Edmonds 1962; Radford and Hartland-Rowe 1971; Hawkins 1986; Richardson 1989) despite discharge regimes that differed from those described in this study. Only minor changes occur in the timing of emergence periods and growth rates primarily due to temperatures, food resources, or both (Radford and Hartland-Rowe 1971; Hawkins 1986). The three invertebrate taxa may be able to exist under a wide variety of environmental conditions and possible discharge regimes, i.e., they may be habitat generalists. Zapada cinctipes was found in all streams except the large rivers. Z. cinctipes is a shredder/detritivore (Short and Ward 1981; Merritt and Cummins 1996) and its absence from the large rivers may be simply a result of the lack of C P O M in the large rivers. Drunella doddsi and D. spinifera, however, are less habitat generalists than Z. cinctipes, because they were not common to as many streams. Drunella doddsi have particular habitat requirements such as coarse substrates, steep gradient streams, and dense riparian vegetation (Mangum and Winget 1991). Although the invertebrates used in this study may not be generalists for all environmental conditions, they may use behaviours that allow them to cope with different discharge regimes. Invertebrates may use behavioural adaptations rather than life cycle adaptations to avoid seasonal scouring. These behavioural adaptations could include movement to margins or flood plains, movement down into the substrate, or movements into areas of downwelling. Rempel (1997) found invertebrates in large rivers use the flood plain as a refuge during the annual freshet. It has been demonstrated that areas of the substrate act as flow refugia (Lancaster and Hildrew 1993 a), particularly 30 where there are patches of flow downwelling (Dole-Olivier et al. 1997). Invertebrates move into these refugia during periods of high flow. These behavioural adaptations may provide the invertebrates with sufficient protection from the seasonal scour so that changes to the timing of life cycle events are not required. Life cycle adaptations to bankfull discharge associated with seasonal flooding may not be necessary for benthic invertebrates. Researchers have defined bankfull discharge as an event that scours, disturbs stream channel substrate, and disrupts and removes benthic invertebrates (Newbury 1984; Poff 1992). However, measures of bed movement may be better related to species traits and richness than measures of discharge (Townsend et al. 1997). Influence of Temperatures and Food Resources on Life Cycles Interior streams were generally cooler than the coastal streams and large rivers. These cooler temperatures were a result of a combination of the high elevations and cooler air temperatures in the spring, autumn and winter. By November, the interior streams were frozen over, and they retained some ice cover until mid April. Coastal streams were generally warmer than the interior streams because the coastal climate is moderated by the Pacific Ocean, and the streams are at intermediate elevations. As a result, the streams did not freeze over in the winter. Water temperatures in the large rivers were warm in the spring, summer, and autumn because of the large surface area of water exposed to the sun and air. Water temperatures influenced the timing of insect life cycles in the coastal and interior streams. The greatest growth rate for D. doddsi and D. spinifera occurred between July and October when stream temperatures were the warmest. Warm coastal stream temperatures also resulted in D. doddsi reaching their final instar and emerging earlier than individuals living in the cooler interior streams (Sweeney 1984). Drunella doddsi from the coastal streams were slightly larger at each sampling date than D. doddsi from interior streams, probably because D. doddsi emerged, laid eggs, and hatched a little earlier in coastal streams than in the interior. 31 The greatest growth rate of Zapada cinctipes from interior streams also occurred between July and October when stream temperatures were the warmest. These Z. cinctipes continued to grow through the winter at a reduced rate but sufficient for them to achieve the size required for emergence by late March/early April, after the ice cover had broken up. The greatest growth rate of Z. cinctipes in coastal streams occurred between October and December. Although temperatures are cooler between October and December than between July and October, they are not too cool for the invertebrates to take advantage of the seasonal input of C P O M . In fact, the combination of 'warm' temperatures and abundant food through October and December allowed Z. cinctipes to attain a size large enough for emergence by mid-December. Ice did not form over the coastal streams in the winter and emergence of Z. cinctipes was not delayed. Zapada cinctipes in coastal streams of British Columbia emerge from January to March (Richardson 1989). The quantity and quality of food resources also influences growth rates and life cycles. In temperate streams, the majority of leaf litter falls between September and November (Richardson 1992). In this study, major inputs into the stream channel were not limited to the period of leaf fall. For the interior streams and large rivers, the relatively large amount of C P O M and FPOM in spring and summer coincided with the annual increase in discharge, which can remove organic matter from the stream banks and flood plain bringing it into the channel (Ractliffe et al. 1995; Rempel 1997). For the coastal streams, the majority of the leaf litter input coincided with leaf fall (Richardson 1992) and discharge high enough to entrain the majority of the accessible material. Leaf fall between September and October provided food resources during the major growth period of coastal stream Z. cinctipes. It also provided food resources for over-wintering detritus shredders such as D. spinifera (Hawkins 1984) and interior stream Z. cinctipes (Short and Ward 1981). The major influx of organic matter in the spring, provided a flush of resources for shredders through spring and summer. Periphyton biomass and chlorophyll-a levels were highest in spring and summer for all streams except the Thompson River and Mayfly Creek. In the Thompson River, algal biomass and chlorophyll-a 32 were greatest in spring - just before the predictable seasonal increase in discharge scours the channel bottom. Bothwell and Culp (1993) concluded that the scouring of the channel bottom at the same site in the Thompson River, removed most of the periphyton. Post scouring, the periphyton community rebuilds itself until the next scouring event. This trend, however, was not seen at the Fraser River sampling site, perhaps because of the large amount of suspended particulates in the water column (Table 2.1). As discharge decreases, sediments fall out of suspension, coating the substrate and covering any periphyton growth. As a result, primary production was highest in the summer during high levels of discharge when the sediments were removed from the rock surfaces. Invertebrate abundance varied with stream, sampling date, and life cycle stage. In general, the highest abundance of each taxon was related to the hatching of the new generation. Some of the changes in abundance between sampling dates may have been a result of different sampling intensities at each date, although efforts were made to keep sampling intensity consistent. Drunella doddsi are diatom scrapers and predators (Hawkins 1984), their highest growth rates and abundance occurred during the spring and summer, the interior and coastal streams, and corresponded with high levels of periphyton. In July, the abundance of Zapada cinctipes was lower in the interior streams than in coastal streams, this may be a result of the timing of hatching and the discharge regime. In summary, the timing of D. doddsi, D. spinifera, and Z. cinctipes life cycle events do not appear to be altered with different discharge regimes. The invertebrates may be unable to adapt to the discharge regimes. Conversely, these invertebrates may not need to adapt if behavioural tactics are sufficient to avoid the seasonal scour. Invertebrate growth rates and the timing of emergence are influenced by temperatures and food resources, but it was not possible to separate these effects. 33 Chapter Three S E A S O N A L A N D S P A T I A L C H A N G E S O F B E N T H I C I N V E R T E B R A T E C O M M U N I T I E S A N D A S S O C I A T E D E N V I R O N M E N T A L CONDITIONS I N S O U T H W E S T E R N B R I T I S H C O L U M B I A Introduction The relationship between environmental conditions and benthic invertebrates has been a dominant theme in stream ecology. Combinations of geographic factors, water chemistry, habitat stability, and/or land use influence stream community structure (Wright et al. 1984; Ormerod and Edwards 1987; Corkum 1989; Richards et al. 1993; Death 1995; Tate and Heiny 1995 and others). Many environmental factors change seasonally and there is the potential for the structure of invertebrate communities to change with them. Environmental factors which change seasonally include those which have been shown to have a direct influence on invertebrate life cycles, population dynamics, trophic interactions etc. e.g., temperature, resource abundance, photoperiod, and discharge (Sweeney and Vannote 1981; Sweeney 1984; Hawkins 1986; Robinson et al. 1992; Chapter 2). Studies have related seasonal change of community structure to a limited number of environmental conditions. For example, Hawkins and Sedell (1981) related functional composition to longitudinal and seasonal changes of food resources. Boulton and Lake (1992) and Closs and Lake (1994) related changes in faunal and functional composition to seasonal changes of flow regime and food resources in intermittent streams. Hynes (1970) proposed a pattern of seasonal change in abundance and biomass in small streams based on a review of several studies that sampled benthic communities in small temperate streams through the year, and the life histories of invertebrates. Hynes predicted that with recruitment, total invertebrate abundance would increase from the late spring through the summer, and highest abundance would be achieved in late autumn (Fig 3.1). Abundance would decrease through the winter as a result of deaths and predation, and lowest abundance would occur in the early spring (Fig 3.1). In lotic systems, many environmental factors may change simultaneously. The relationship between seasonal change of invertebrate community structure and changes of co-occurring environmental 34 variables should be determined to facilitate identification of environmental factors that have the greatest influence on the seasonal change of community structure or that interact with other variables to influence community structure. The purpose of this chapter is to examine how benthic invertebrate community composition changes spatially, between streams classes and streams nested within classes, and seasonally in southwestern British Columbia. The three stream classes include coastal streams, interior streams and large rivers. Changes to community composition were assessed at 1) the lowest possible taxonomic level, 2) family level, and 3) functional level. Environmental variables were monitored along with the invertebrate communities. Relationships between invertebrate community structure and environmental characteristics were first determined, followed by ways in which the benthic community changes seasonally, and then whether community changes are related to seasonal changes in the environment. Methods Spatial and temporal changes of benthic macroinvertebrate community structure were assessed for eight rivers in three stream classes in southwestern British Columbia (Fig. 2.1, 2.2). Samples were collected from the eight streams over five sampling dates between April 1995 and April 1996 (Figs. 2.2a,b and c). The sampling dates were selected to bracket seasonal changes in temperature, food resources, and discharge. Invertebrate Samples Benthic invertebrate were collected using kicknets, a detailed description of sample collection and processing was presented in Chapter 2. Three samples per site per date were randomly selected from the five collected, processed, and the data were combined. The loss of rare taxa from the data set as a result of subsampling was not considered to be a problem. Rare taxa create a large number of zero values, noise in the data set, and are therefore typically excluded when multivariate analyses are used. 35 Spring Summer Autumn Winter Spring Figure 3.1: Seasonal pattern of insect abundance in small undisturbed streams as proposed by Hynes (1970). 36 Invertebrates were identified to the lowest practical taxonomic level, usually species and genus levels. However, Ostracoda, Turbellaria, and Copepoda were not identified beyond class, and the Chironomidae were identified to the subfamily level. The resulting data set will hereafter be referred to as the genus/species data set. Family level and functional groups data sets were generated from the genus/species data set. Insects were placed into functional groups based on Pennak (1978) and Merritt and Cummins (1996). Invertebrates using more than one food source were divided equally into each of the appropriate functional groups. Environmental Conditions Environmental variables were measured either when the benthic samples were collected or were determined from maps (Table 2.2). Variables having an association with benthic invertebrate community composition were included (Wright et al. 1984; Corkum and Currie 1987; Ormerod and Edwards 1987; Corkum 1989). The measurement of hydraulic variables and the collection of periphyton samples are described in chapter 2. Conductivity, pH, dissolved oxygen, temperature, depth, and velocity were all measured at each sampling site with the respective meters. Conductivity was measured with a YSI 6000 conductivity meter, dissolved oxygen and pH with Hanna Instruments pH and D.O. meters, temperature with a thermometer, and depth with a ruler. Current velocity was determined by measuring velocity with a Marsh-McBirney electromagnetic current velocity meter at regular intervals along the stream and at 60% of the depth. From those measurements, average velocity was determined by averaging the velocities and maximum velocity was the highest velocity measured. Water chemistry samples (alkalinity, T K N , N 0 3 N 0 2 - N , N H 3 - N and total phosphorous) were collected, sent to the National Water Research Institute (NWRI) for analysis, and analyzed following the methods described in Cancilla (1994). One milliliter of 30% H 2 S 0 4 was used to preserve the water sample for the phosphorous analysis, and all samples were stored at 4 -5C until analyzed. Alkalinity (capacity of water to accept protons as a result of carbonates (C032") and bicarbonates (HC03") was determined by electromertic titration with a minimum detection level of 0.1 mg/L (see Cancilla 1994). 37 Kjeldahl nitrogen is the sum of free ammonia and trinegative nitrogen compounds(Greenberg et al. 1992). T K N was determined using the semi-automated block digestion andphenate-hypocolormetric procedure. This macro-kjeldahl method was used because it is applicable for samples containing either low or high concentrations of organic nitrogen and has a minimum detection level of 0.01 mg T K N / L . The ammonia-nitrogen (NH 3-N) content was determined using the Berthelot Reaction and spectrophotometry (630 nm wavelength), which has a minimum detection level of 0.001 mg NH 3 -N/L. Nitrate-nitrite-nitrogen concentration was determined using the TRAACS 800 automated colormetric cadmium reduction procedure which has a minimum detection level of 0.005 mg N Q N0 2 -N /L . Total phosphorus was determined by hydrolysing all organically-bound and other forms of phosphorous, by acid digestion, to form orthophosphate (P043~). Orthophosphate content was then determined using automated colormetric, stannous chloride technique which was able to detect as little as 0.0002 mg P/L. Samples of total suspended solids and sediment particle size were also sent to the NWRI for analysis. Sediment particle size was determined by sieving the sample, drying and weighing the sand and gravel portion (>0.063mm), and then running the <0.063mm portion through a Sedigraph analyzer to determine silt and clay content (Duncan and LaHaie 1979) Channel bottom substrates were described using a classification system that involved three descriptors: framework, matrix, and embeddedness (Rosenberg et al. In prep). Framework represented the size of dominant substrate on the channel bottom; matrix represented the size of substrate surrounding the framework; and embeddedness represented the depth at which dominant substrates were embedded in the surrounding material. Each of these substrate characteristics was assigned a score depending on the size and class of the substrate. The scoring system for framework and matrix based on the substrate particle size was: 1 = organic cover over 50% of the bottom, 2 = <0.1-0.2cm, 3 = 0.2-0.5cm, 4 = 0.5-2.5cm, 5 = 2.5-5cm, 6 = 5-10cm, 7 = 10-25cm, and 8 = >25cm. Forembeddedness, 1 = totally embedded, 2 = Va embedded, 3 = Vi embedded, 4 = lA embedded, and 5 = unembedded. Vegetation in and around the stream was also described in the data set by categories. Macrophyte categories were assigned based on the percent area covered by macrophytes: 1 = 0%, 2 = 0-25%, 3 = 25-38 50%, 4 = 50-75%, 5 = 75-100%. Each type of riparian vegetation was assigned a number based on whether the different vegetation types (grasses, shrubs, deciduous, coniferous) were present (1) or absent (0). Percent canopy cover for a reach was estimated based on the area covered by a canopy along a transect on one side of the stream channel. Data Analysis Principal Components Analysis (PCA) (Proc Princomp; SAS Institute 1996), an indirect ordination technique, was used to summarize the variation and identify major gradients in both the invertebrate and environmental data sets. Both the invertebrate and environmental data sets used in the PCA contained the seasonal information as well as the spatial information. Ordinations were run for three invertebrate classification levels: genus/species, family, and functional group. P C A assumes a linear response curve (Gauch 1982) and a large number of zero values will create a non-linear response curve, so rare taxa were removed from the data set. Taxa that occurred < 1% of the time were removed from the genus/species data set, and taxa with < 0.5 % occurrence were removed from the family data set (a 1 % cut-off for the family data left too few taxa for the analysis and a 0.5% cut-off for the genus/species data added only 5 taxa). Patterns of composition (rather than just numeric differences) were important, so the faunal data were log transformed (logi0(x+l)), which reduced the influence of the more abundant taxa. A correlation matrix was used because it standardizes the data, thus altering the relative weight of each variable and allowing relative abundance to be examined (Jackson 1993). Two ordinations were run for the environmental data - one on the 33 environmental variables listed in Table 3.1 and one for 21 of the environmental variables that changed seasonally. The correlation matrix was used because it standardizes the data and minimized variation caused by different scales of the environmental variables. Correlation analysis (Proc Corr; SAS Institute 1996) was used to determine which taxa and environmental variables had significant loadings for their respective ordinations. Significance was determined by comparing probability values with a Bonferroni-corrected probability value (p = 39 0.05/number of comparisons made). This step reduced the likelihood of a correlation being significant simply due to chance, and reduced over-interpretation of the data (Rice 1989). Correlation analysis was also used to determine if and how eigenvectors from the invertebrate and environmental ordinations were related. Two different analysis techniques were used to determine how much variation in the ordinations could be explained by season, site, and stream class. Multiple analysis of variance (MANOVA) (Proc G L M ; SAS Institute, 1996) of the axis 1 and 2 principal component (PC) scores was used to determine if the ordination scores for the seasons, sites nested within stream class, and stream class were significantly different, and how much variation each spatial and temporal level explained. Coefficient of variation (CV = standard deviation of seasonal PC scores for a site/average seasonal PC scores for a site) was used to determine how much seasonal variation existed for each stream (Zar 1984). PCA assumes that the distance between a given pair of ordination points is proportional to the measure of similarity between the pair of samples those points represent (Gauch 1992). Thus, the distance in ordination space between seasonal samples for a site represents the (dis)similarity in the communities between seasons. This magnitude of invertebrate seasonal change in ordination space was tested for correlation with those environmental variables that change seasonally to determine whether seasonal changes in the invertebrate community were related to seasonal changes in the environment. The distance between seasonal coordinates for a site was calculated using: Z 2 = X ( a 2 - a i ) 2 + Y ( b 2 - b , ) 2 Where: Z = Distance in ordination space between seasons (seasonal change) a{ - PC axis 1 coordinate at season 1 a 2 = PC axis 1 coordinate at season 2 X = Proportion of variance explained by axis 1 b, = PC axis 2 coordinate at season 1 b 2 = PC axis 2 coordinate at season 2 Y = Proportion of variance explained by axis 2 40 Two measures of discharge were used (Table3.1): peak discharge and total cumulative discharge during the month prior to sampling. Total cumulative discharge was determined from guaging station measurements collected by Environment Canada and Dr. Michael Feller, (Department of Forest Sciences, UBC). Discharge (peak and cumulative), wetted width, mean depth, maximum depth, and mean and maximum velocity were standardized (e.g., Q m a x / Q average for the year) so that the change was relative and inter-system comparisons could therefore be made. Correlation analysis assumes that the variables have a bivariate normal distribution. To meet this assumption, seasonal changes of the environmental variables were tested for normality (Shapiro-wilk test; Statsoft Inc. 1994), and transformations were tested and applied as required. No transformations were required for the distance between the PC scores or the standardized variables because both were already standardized. Transformations were also not required for pH, temperature, and N 0 3 N 0 2 - N. Al l other variables were loge -transformed. The magnitude of change in stream communities between seasons and their relations with environmental variables, or the seasonal difference in those variables, were tested with analysis of covariance (ANCOVA) (Proc G L M ; SAS Institute, 1996). A N C O V A was first used to determine the relationship between the change in invertebrate PC scores between seasons and the change in environmental variables (regression). It was then used to test for a difference between the slopes for streams nested within stream class and stream class (treatments) while controlling for an environmental variable (covariate). Variables used in this analysis were the same as those used in the correlation analysis above. The same transformations were applied to the environmental variables to meet the assumptions of normality, homogenous variances, and distributions of the residuals. Least square means were used to determine which means differed for the various streams and stream classes. Alpha critical for the least square means was Bonferroni-corrected because all possible pairwise comparisons were used, thus controlling for Type I errors. Table 3.1: Environmental variables used in the analyses. Environmental variables Abbreviation Variables which change seasonally Stream order order Drainage basin area D B A Distance from source dist Elevation elev Channel gradient slope Flow state 1 (riffle to pool) flwl Flow state2 flw2 (slow subcritical to hydraulic jump) Bankfull width bnkfl Wetted width width * Mean depth menD * Maximum depth maxD * Discharge at sampling Q * Peak discharge peakQ * Cumulative discharge cummQ * Mean velocity menV * Maximum velocity maxV * Temperature temp * pH pH * Conductivity cond * Total suspended solids TSS * Percent carbon in TSS %CTSS * Alkalinity alkal * Nitrate and nitrite - nitrogen N03-N * Ammonia - nitrogen NH3-N * Total Kjeldahl nitrogen T K N * Total phosphorus totlP * Suspended N suspN * Suspended C suspC * Chlorophyll-a chla * Coarse particulate organic matter C P O M * Fine particulate organic matter FPOM * Canopy coverage %cover Substrate framework frmwrk Substrate matrix matrx Substrate embedbedness embd 42 Results Invertebrate Composition A total of 152 taxa was identified, mostly to genus. Forty-nine of the taxa were found in all three classes of streams, 20 only in the interior streams, 36 only in coastal streams, and 20 only in large rivers (Appendix 2.1). Many of the invertebrates specific to a stream class were found in low numbers and only on one or two occasions. Taxa common to a stream class are defined as those found in > 25 % of the samples. There were nine taxa specific to and common in interior streams, and they are listed in Table 3.2, which also lists the 16 taxa specific to and common in coastal streams and the three taxa specific to and common in the large rivers. The lack of common occurrence by more taxa in the large rivers may have been simply a result of the generally low abundance of most taxa. A few of the exclusive taxa were only specific to a stream class because identifications were made to the genus or species level. Taxon richness, abundance, and diversity changed among stream class, sites, and sampling dates. The large rivers generally had the lowest taxon richness and abundance, whereas the coastal streams had the highest richness and interior streams had the greatest abundance (Figs. 3.2 and 3.3). Simpson's index, of diversity (Krebs 1989) was generally high (Table 3.3), which may have been partly caused by the removal of rare taxa with subsampling. Streams and seasons with the highest diversity typically had the greatest number of taxa (Fig 3.2) or, to a lesser extent, a more even distribution of abundance. Across all eight streams there was no consistency as to which season had either the highest or lowest taxon richness, invertebrate abundance, or diversity. Through all seasons, the large river samples were composed primarily of Diptera (up to 93 %). In the large rivers, combined abundance of Ephemeroptera, Plecoptera, and Trichoptera ranged from 4.7 % in the Thompson River spring 1996 sampling to 66.8% in the winter sampling of the Fraser River. Coleoptera and other invertebrates made up only a small proportion of the total (Fig. 3.4). Naididae (Oligochaeta) were abundant in seasons in which other invertebrates had relatively high abundances in the Fraser River. In contrast, in both the coastal and interior streams Ephemeroptera, Plecoptera, and Trichoptera were typically more abundant than the Diptera (Fig. 3.4). Again, Coleoptera and other 43 invertebrates made up only a small proportion of the interior and coastal benthic communities, except in Mellin Creek in the 1995 spring and summer samplings when tubificids were abundant. Deposit FPOM collectors were the most abundant functional group in the large rivers and across all seasons, making up to 94.8 % of community composition (Fig. 3.5). The second most abundant functional group was herbivores (up to 53 % of the composition). All other functional groups combined made up a maximum of 29.2 % of the composition. The most abundant functional groups in the coastal and interior streams were also the deposit FPOM collectors and herbivores. However, predators and Table 3.2: Taxa found in > 25% of the samples from a specific stream class (interior streams, coastal streams, or large rivers). See Appendix 2.1 for a listing of all taxa found in each stream class. Taxonomic Group Interior Coastal Large Gastropoda Pisidium casertanum + Plecoptera Amphinemura spp. + Plecoptera Megarcys spp. + Trichoptera Brachycentrus americanus + Trichoptera Hesperophylax spp. + Trichoptera Allomyia spp. + Trichoptera Neophlax spp. + Diptera Pericoma spp. / Thelmatoscopus spp + Diptera Hexatoma sp. A + Ephemeroptera Epeorus (Ironopsis) spp. + Ephemeroptera Ironodes spp. + Plecoptera Malenka spp. + Plecoptera Yoraperla brevis + Plecoptera Hesperoperla pacifwa + Plecoptera Setvena bradley + Plecoptera Pteronarcys princeps + Trichoptera Micrasema sp. A + Trichoptera Agraylea spp. + Trichoptera Onocosmoecus unicolor + Trichoptera Wormaldia spp. + Trichoptera Polycentropus spp. + Trichoptera Rhyacophila chilsia grp. + Coleoptera Cleptelmis spp. + Coleoptera Lara spp. + Coleoptera Hexatoma sp. B + Ephemeroptera Stenonema spp. + Trichoptera Brachycentrus occidentalis + Diptera Robackia spp. + 6000 - i 44 15 4000 H 2 0 0 0 -Z o-cu | ia 12000-CD | 1 0 0 0 0 -'s rA 8000 -.s c3 6000 --4—» 4000 -C+H o (U 43 2000 0 25 32 13 23 i —• 12 l 1 28 17 Fraser Thompson 44 46 46 39 39 58 JZl 51 62 46 30 30 17 34 Spring Mayfly N. Alouette c 13 o 16000-14000-12000-10000-8000 -6000 -4000 -2000 -0 -45 30 32 31 3ft JZZL Mellin cf / ^ 31 26 2r' 25 45 41 41 48 N/S Glimpse & # J? & & ^ ^ ^ «9 Beak Figure 3.2: Invertebrate abundance at each sampling site and date based on three one-minute kicknet samples. The number of taxa found at that site and sampling date is above the bars. All eight sampling sites are located in southwestern British Columbia. Note the scales differ. N/S = no sample. L = large rivers, C = coastal streams, and I = interior streams. 45 C/3 C3 I-i •8 14000-1 12000-n £ ioooo-| .a 'o 8000 -1 I 8 6000 -1 U 4000 H 2000 H X X 4? 6* Is. » 6 x s P Streams T / ioo-i B c3 X ccj -*-» <H-H o 43 80" 60 H 40—| 13 +^  o H 20 H T # 6 T / Streams Figure 3.3: (A) Average number of invertebrates collected in the composite three one-minute kicknet samples over all sampling dates. Bars represent + 1 SE. (B) Total number of invertebrate taxa in the three one-minute kicknet samples at each sampling site over all sampling dates. I = interior streams, C = coastal streams, and L = large rivers. Fraser Figure 3.4: Relative abundance of the major insect orders and other invertebrates in the composite 3 X 1 - minute kicknet samples collected at each sampling site and date. L = large rivers, C = coastal streams, I = interior streams, and N/S = no sample. Shredder - C P O M Collector - deposit FPOM Collector - filterer Herbivore Parasites Predators L Fraser Thompson Spring Mayfly N.Alouette lOO—i 80-6 0 -4 0 -2 0 -0 Mellin Glimpse Jj ,«/ A? Beak & *v / 0° J & Figure 3 . 5 : Relative abundance of functional groups in each composite 3 X 1-minute kicknet sample collected from each stream and sampling date. Functional groups are as defined by Merritt and Cummins (1996) and Pennak (1978). L = large rivers, C = coastal streams, I = interior streams, and N/S = no sample. 48 shredders made up larger proportions of the community than in the large rivers (Fig. 3.5). Predators and shredders composed -32 % of the community in coastal streams, and 13.6 and 63.6 % of the community, respectively, in interior streams. Of the 153 taxa identified, 23 represented > 1% of the total organisms collected when identification was to the genus/species level (Appendix 2.2). When identification was to the family level, 24 taxa represented > 0.5% of the total organisms collected (Appendix 2.2). These taxa were used in the PC analyses. Many of the rare taxa exclusive to the three stream classes were thus removed from the data set. The absence of rare taxa did not result in less discrimination between sites in the PCA because results of a PCA of presence/absence data were similar. The first three axes for the PCA done on abundances of genus/species taxa explained 58.7 % of the variation in the data set. Each of the sampling dates for a site clustered together (Fig 3.6a). Large river sites occupy one end of the ordination and the other six sites are spread across the ordination plot. Fourteen taxa were significantly positively correlated with the first axis (p < 0.002). Invertebrate abundance rather than composition dominated the first axis even though the data were log 1 0 (x+1) transformed and the correlation matrix was used. Taxa had both significant positive and negative correlations with axes 2 and 3 indicating that composition was more important along these axes. Tubificidae and Prosimulium sp. were positively correlated with the second axis and Neophylax sp., Zapada oregonensis group, Sweltsa sp., and Cinygmula sp. were negatively correlated with the second axis. The direction and magnitude of invertebrate loadings for taxa with significant correlations are illustrated in Fig 3.6b. The M A N O V A indicated that seasonal variation within a site accounted for < 5 % of the explained variance expressed in the ordination space, whereas the stream accounted for 47.9 and 37.6 % of the variation along axes 1 and 2, respectively, and the stream class accounted for 47.4 and 57.5 % of the variation along axes 1 and 2 (Table 3.4). Coefficient of variation analyses indicated that the benthic community in each stream varied different amounts between seasons. The communities in Mellin and 49 Glimpse creeks varied the most along axis 1 (changes in abundance) and changed very little along axis 2 (change in composition) (Table 3.5). All other sampling sites vary more along axis 2 (change in composition) than along axis one (change in abundance) (see Table 3.5). P C A results were similar for analysis at the genus/species and family levels. For the family level data, the first three axes explained 61.5 % of the variation. Most of the seasons for a site and the sites within the large river and coastal stream classes clustered together (Fig. 3.7a). As in the genus\species ordination, 16 of the 17 taxa with significant correlations (p < 0.002) were positively correlated with the first axis. Naididae were included at this level of analysis, however, and were negatively correlated with the first axis because of their increased abundance in large rivers. Naididae, Psychodidae, Ephemerellidae, Brachycentridae, and Hydropsychidae were positively correlated with the second axis, whereas Glossosomatidae were negatively correlated with the second axis (p < 0.002). The direction and magnitude of loadings for taxa with significant correlations are illustrated in Fig. 3.7b. Seasonal variation in the family data set only explained 5% of the variation in the ordination, similar to M A N O V A analysis at the genus/species level. The stream explained 39 and 75 % of the variation along axes 1 and 2, respectively, and the stream class accounted for 56 and 19.6 % of the variation along axis 1 and 2 (Table 3.4). The coefficient of variation analyses indicated lower seasonal variation at each site when family level data were used than when the genus/species level data were used. At the family level of analysis, Mellin Creek still varied the most along axis 1, followed by the North Alouette River and Glimpse Creek. Mayfly Creek, Mellin Creek, and the Fraser River varied the most along axis 2 (Table 3.5). When functional groups were used in the PCA, the first three axes explained 67.7, 11.4, and 10.2 % of the variation respectively. Certain communities were functionally distinct between sites or regions such as Glimpse Creek versus the Thompson River or the interior streams versus the large rivers (Fig3.8a). Al l functional groups had significant positive correlations with the first axis (p < 0.008). Only filter feeders were significantly correlated with the second axis (Fig. 3.8b). 50 5 2 H CD C '3 1-CD a o 3 <H CN co • 2 H O 0 o A A T -6 -2 • • 1 0 T T T A x i s 1 (31.5 % variation explained) Figure 3.6a: Ordination plot of the seasonal samples for each site along principal component axes 1 and 2 based on log-transformed invertebrate data identified to genus/species, 1% censorship level. OMel l i n • Glimpse A Beak # Spring • Mayfly A N.Alouette ^Frase r 0 Thompson 0.5 T3 CD « % x CD o > cn CN co < o.o--0.5' Tubific Prosimulium /Baetissp.A n P B a e t i s s p B 1 / ///Hydropsyche ' u< .< MMEpeorus ff)l°mjy0r^Z. cinctipes / / / • ^ ^ % ^ ^ Pericoma P- temporalis ^ S ^ ^ — ' = = = = = = ^ % Q B . americanus Chironominae \ \ \ \ Lepidostoma \ \ \ \ . * Heterlimnus \ \ Cinygmula \w Sweltsa \ Neophylax 9 Z. oregonensis grp. -0.50 0.50 -0.25 0.00 0.25 Axi s 1 (31.5 % variation explained) Figure 3.6b: Direction and magnitude of loadings for taxa with significant loadings along axis 1 and 2 of the ordination shown in Fig 3.6a.. 0.75 6 A <D 4-g '3 CD (U o S3 > o x CO Tt r—< CN CO 2H -2H -6-o o • • O n A A 1 A ^ A O A • • • • • T -6 ~ r -2 0 Axis 1 (35.3 % variance explained) F i g u r e 3.7a: Ordination plot of the seasonal samples for each site along principal component axes 1 and 2 based on log-transformed invertebrate data identified to family, 0.5% censorship level. OMel l in • Glimpse A Beak • Spring • Mayfly A N.Alouette ^Frase r @ Thompson 0.5-0.4H T3 CD g ' 3 CD (3 O > o x CO 0.3 H 0.2 H o . H 0.0-- o . n CN '$-02 A -0.3-Hydropsychidae (J) Ephemerellidae Naididae / / % Psychodidae // / M Brachycentridae / / / wCSimuliidae / / B a e t i d a e / /<//y7' Chironomidae / /J/MS Tricladida r ~. Lepidostomatidae ^ S a ^ s a^r— 9 • Y ^ ^ ^ ^ ^ ^ - ^ L e p t o p h l e b i i d a e \ ^ x ^ ^ ^ ^ " ^ ^ Rhyacophilidae \ Lebertnd|e^WHeptageniidae NOstracoo'aX Nemouridae \ Chloroperlidae • Glossosomatidae T -0.3 -0.2 -0.1 0.4 T 0.5 0.6 1 1 1 0.0 0.1 0.2 0.3 Axis 1 (35.3% variation explained) F i g u r e 3.7b: Direction and magnitude of loadings for families with significant loadings along axis 1 and/or axis 2 of the ordination illustrated in Fig 3.7a. 0.7 52 2 H 1 H 0 g a o i > o x TJ; r—I CN < -2 -3 0 0 -0 0 # A A - <• . o • • c f • A o • -r -6 -4 -2 0 2 4 6 Axi s 1 (67.7% variation explained) Figure 3.8a: Ordination plot of the seasonal samples for each site along principal component axes 1 and 2 based on invertebrate data classified by functional group. OMel l in • Glimpse A Beak • Spring • Mayfly • N.Alouette ^ Fraser @ Thompson 1-A Filter feeder / ^ % Gatherer ^ v ^ ^ t ^ Parasite \ ^ • Scraper \ S h r e d d e r • Predator 1 1 1 1 T3 <D '3 o •g °-> x° o x co •1.5 •1.0 -0.5 0.0 0.5 1.0 1.5 A x i s 1 (67.7% variation explained) Figure 3.8b: Direction and magnitude of significant loadings for functional groups. 53 4H § 3H CD c "3 a, x CD S=! o '-3 2H > cN r -© w CN C/3 OH • H -2H -3H -4 • 0 A A 0 • o A o O A • • • D 0 T " -4 T -2 T T 0 2 4 6 Axis 1 (31.5 % variation explained) Figure 3.9a: Ordination plot of the sampling dates and sites along principal component axes 1 and 2 based on environmental conditions that change seasonally (Table 3.1). OMellin • Glimpse A Beak • Spring • Mayfly A N.Alouette ^Fraser Q Thompson 0.6 MenD 1 NO 3NO 2-N ^MaxD r9 o Width ' / / > Q /UMaxV ^-#TSS %CTSS V ^ ^ S u s p C ^ J j SuspN ^ T o t a l ^ F * * Chl a \ S ^ ^ p H N ^ ^ A C o n d XKN A l k a l T 3 CD 8* CD c o d C N < N 0.4H 0.2 H 0.0--0.2 H -0.4 H -0.4 -0.2 0.4 0.6 0.8 0.0 0.2 Axis 1 (31.5% variation explained) Figure 3.9b: Direction and magnitude of significant environmental loadings along axis 1 and 2 of the PCA. Abbreviations are listed in Table 3.1 54 Table 3.3: Invertebrate diversity at each sampling site and date based on Simpson's index of diversity (1-D). Index ranges from a low diversity value of near 0 to almost 1. Stream Season Spring 1995 Summer Autumn Winter Spring 1996 Mellin 0.4874 0.8019 0.8123 0.8722 0.4985 Glimpse 0.5307 0.9153 0.843 0.6419 0.7493 Beak 0.8553 0.9172 0.8722 N/S 0.8972 Spring 0.9254 0.6867 0.9235 0.79 0.9126 Mayfly 0.9251 0.8078 0.8021 0.8106 0.9313 N. Alouette 0.8863 0.9017 0.9029 0.7968 0.5382 Fraser 0.7264 0.5183 0.8149 0.9053 0.8541 Thompson 0.4121 0.8709 0.6586 0.8389 0.2299 Table 3.4: Percent variance (of explained variance) explained by season, site, and stream class in the PCA ordinations as determined by M A N O V A . The environmental data used in the ordination consisted of environmental variables that changed seasonally. Data Analyzed PCA Axis Season Site Stream Class Genus / Species 1 4.7 47.9 47.4 2 4.9 37.6 57.5 Family 1 5 39 56 2 5.3 75.1 19.6 Functional Groups 1 17.9 26.9 55.2 2 7.4 85.1 7.5 Environmental 1 3 6.6 90.4 2 3.6 11.7 84.7 Table 3.5: Coefficient of variation used to summarize the seasonal variation in ordination space for each site. Coefficients of variation were determined for the ordinations of the genus/species, family, and functional data, and for the ordination of the environmental variables that change seasonally. The rank order of the first axis coefficients of variation for the genus/species and family data are correlated (Spearman's rank order correlation, rs= 0.9286, p < 0.005), but the rank order of the second axis is not correlated (Spearman's rank order correlation, r s = 0.2857, p > 0.5). Genus / Species Family Functional Environmental Site Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Mellin 70.87 0.69 3.49 1.88 2.17 0.95 0.18 0.58 Glimpse 11.24 0.22 0.59 0.3 0.64 0.69 0.52 0.32 Beak 0.16 7.26 0.09 0.89 0.21 0.7 0.19 0.7 Spring 0.39 1.21 0.38 1.07 0.8 1.9 0.24 29.47 Mayfly 0.72 2.43 0.45 7.2 1.69 4.75 0.7 2.38 N.Alouette 0.55 3.08 0.69 0.34 0.68 0.45 0.25 0.4 Fraser 0.45 3.77 0.41 1.26 0.87 1.46 0.23 0.34 Thompson 0.28 0.55 0.27 0.67 0.7 0.89 0.49 0.63 55 Seasonal variation in the functional group data set explained 17.9 and 7.4 % of the variation along axes 1 and 2 respectively. The stream explained 26.9 and 85.1% and the stream class accounted for 55.2 and 7.5 % of the variation along axes 1 and 2 (Table 3.4). Results from the coefficient of variation analyses indicated that Mellin and Mayfly creeks varied the most along the first axis or, abundance changed seasonally the most in these streams. The Fraser River and Mayfly Creek varied the most along the second axis, indicating that abundance of filter-feeders changed seasonally the most (Table 3.5). Environmental Conditions Principal component analyses were similar for all the environmental variables measured and just those that varied seasonally. Therefore, only results from the latter ordination are presented. The first three axes explained 66.4 % of the variation in the data set. The coastal sites occupied the left side of the ordination space, the interior streams occupied the lower right corner, and the large rivers occupied the upper right corner (Fig. 3.9a). Fourteen of the environmental variables were significantly correlated with axis 1 and/or axis 2 (p < 0.0024). The direction and magnitude of the loadings are illustrated in Fig 3.9a. Variables such as pH, conductivity, alkalinity, and T K N were high in the interior streams. Wetted width, maximum depth, and discharge were high in the large rivers. Percent carbon of suspended solids were high in the coastal streams. Seasonal variation explained < 4 % of the variation in the ordination space and stream explained only slightly more (MANOVA, Table 3.4). Stream class explained most of the variation (85 - 90 %). Coefficient of variation analyses indicated that seasonal variation at each site was minor. Most of the variation occurred along the second axis (Table 3.5). How Changes in the Benthic Community are Related to Changes in the Environment The first and second PCA axes from the genus/species ordination were significantly correlated with the second axis of the ordination of environmental variables that change seasonally (p < 0.0125) (Table 3.6). The taxa Baetisi sp.A, Baetis sp.B, Hydropsyche sp., Epeorus sp., Zapada cinctipes, 56 Serratella tibialis, Pericoma sp., Paraleptophlebia temporalis, Brachycentrus americanus, Chironominae, Lepidostoma sp., and Heterlimnus sp. from axis 1, and the taxa Cinygmula sp., Sweltsa sp., Neophylax sp., and Zapada oregonensis group from axis 2 all increased in abundance in streams with small channel widths, small discharge levels, low maximum and mean depths, low levels of N Q N 0 2 - N , and a high conductivity, alkalinity and T K N level (small interior streams). Tubificidae and Prosimulium sp. decreased in abundance in rivers which Cinygmula sp., Sweltsa sp., Neophylax sp., and Zapada oregonensis increased in abundance. The first and second PCA axes from the family ordination were also significantly correlated with the second axis from the environmental ordination (Table 3.6). The families Simuliidae, Baetidae, Chironomidae, Tricladida, Lepidostomatidae, Leptophlebiidae, Rhyacophilidae, Elmidae, Perlodidae, Lebertiidae, Heptageniidae, Nemouridae, and Chloroperlidae from axis 1 and Glossosomatidae from axis 2 all increased in abundance in streams with small channel widths, small discharge levels, low maximum and mean depths, and low levels of NO5NO2-N, and a high conductivity, alkalinity, and T K N level (small interior streams). Hydropsychidae, Ephemerellidae, Psychodidae, Brachycentridae, and Naididae were most abundant in rivers with some combination of great widths, depth and discharge, and/or low alkalinity, conductivity, and T K N levels. The first P C A axes from the ordination of functional group data was significantly correlated (p < 0.0125) with the first and second axes from the environmental ordination (Table 3.6). Al l functional groups were least abundant in rivers with the greatest depths, widths, and discharges and were most abundant in the smaller streams. Table 3.6: Correlation coefficients (r) and significance values (p) for correlation of PCA axes from the invertebrate ordination with the PCA axes from the ordination of environmental variables that changed seasonally (a c = 0.0125). Bold type indicates significant correlations. Genus / Species Family Functional Groups Environment Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 r -0.1981 0.0019 -0.245 0.2285 -0.4233 0.1655 P 0.2267 0.9910 0.132 0.1617 0.0072 0.3141 Axis 2 r -0.5386 0.4477 -0.538 0.4903 -0.5784 0.1841 P 0.0004 0.0043 0.0004 0.0015 0.0001 0.2620 57 Seasonal changes in the benthic community at each site were significantly correlated with the seasonal changes of some environmental conditions. The relationship between changes in community structure with environmental variables differed between streams and between invertebrate classification level. A summary of the significant correlations can be found in Table 3.7. For this analysis, probability values were compared to an uncorrected alpha critical value (0.05) because patterns were sought and thus the problem of type I error was not so critical (Rice 1989). To test a common hypothesis, the Bonferroni-corrected alpha value would be occ = 0.0023, resulting in only one significant correlation (Table 3.7). Mellin Creek was the only interior stream that had significant correlations (p<0.05); seasonal change of the invertebrate communities was positively correlated with seasonal change in temperature, and seasonal change of the functional community was positively correlated with the change in cumulative discharge (Table 3.7). Changes in the benthic community in Spring Creek were positively correlated with channel width and suspended carbon with identification to the genus/species level. When identification was to the family level, community change was positively correlated with suspended nitrogen and negatively correlated with peak discharge. Functional community changes were positively correlated with channel width. The benthic community in Mayfly Creek (genus/species level) was positively correlated with suspended nitrogen and negatively correlated with suspended carbon and CPOM. In the North Alouette River, changes in the benthic community (family level), were positively correlated with suspended nitrogen. At the functional level, changes to the community were correlated with changes in suspended nitrogen and carbon. In the Fraser River, changes in the benthic community (family level) were positively correlated with change in peak discharge and negatively correlated with conductivity. A N C O V A s were run for the 3 invertebrate classification levels and 22 environmental variables to determine whether the benthic communities from the different streams and classes of streams changed seasonally with the same magnitude for a given environmental variable and how change in community was related to seasonal change in the environmental variable. While stream and stream class had significant effects there were no significant effects of the environmental variables or of the interaction 58 between the environmental variables and streams or stream classes (a c = 0.05) The seasonal change of benthic community structure was not related to the rate of seasonal change of the environmental variables measured in this study regardless of the invertebrate classification level used. Table 3.7: Summary of the significant correlations between the seasonal change of the invertebrate community PCA scores and the seasonal change of environmental variables (ac = 0.05). For abbreviations see Table 3.1. Asterisk represent probability values, *p<0.05, ** p<0.01, *** p<0.001. Invertebrate Environmental Correlation Coefficient p - value Site Classification Variables Mellin genus/species temp 0.990 * family temp 0.986 * functional cumulative Q 0.989 * Glimpse no significant correlations Beak no significant correlations Spring genus/species width; suspC 0.960 *; 0.952 * family suspN; Peak Q 0.976 *; -0.944 * functional width 0.992 ** Mayfly genus/species suspN; suspC; CPOM -0.978 *; -0.998 ***; -0.957 * N.Alouette family suspN 0.948 * functional suspN; suspC 0.969 *; 0.953 * Fraser family Peak Q, cond 0.950 *; -0.963 * Thompson no significant correlations Discussion Spatial Patterns of Benthic Communities Benthic invertebrate species composition, diversity, and abundance can vary along environmental gradients (e.g. biogeographical features, altitude). The particular gradients of environmental conditions correlated with patterns of community composition can be hierarchically nested across spatial scales of the study and environmental gradients present in the study area. Many of the environmental variables related to community composition in this study are the same as those related to community composition in previous studies (Table 3.8). Environmental variables related to invertebrate composition in this study were channel width, mean depth, maximum depth, maximum velocity, discharge, conductivity, alkalinity, nitrite and nitrate nitrogen, and T K N . 59 Invertebrate abundance was high in the small streams and was low in the large river sites. This pattern of abundance related to stream size dominated the PCA results at all levels of invertebrate classification. Nutrients, autochthonous and allochthonous inputs, and canopy coverage were high in the small streams, presumably providing resources for the invertebrates. Within the small streams Zapada oregonensis group, Sweltsa sp., Cinygmula sp., and Neophylax sp. were most abundant in Glimpse Creek, which had the highest elevation, alkalinity, and conductivity of all the streams; and conifers dominated the riparian vegetation. Ward (1986) found Z. oregonensis, and Sweltsa coloradensis to be euryzonal, although Z. oregonensis was most abundant in high elevation streams with a combination of coniferous and deciduous riparian vegetation. Zapada oregonensis grp. are shredders and Glimpse Creek typically had the highest amounts of CPOM. Cinygmula sp. and Neophylax sp. are scrapers and Glimpse Creek also had one of the highest amounts of periphyton. Cinygmula sp. are dominant taxa in high elevation streams with both open and closed canopies of coniferous and deciduous vegetation (Ward 1986). Prosimulium sp. and Tubificidae were not very abundant in the large river samples as suggested by the correlation analysis however, they were never found in the samples from streams thatZ oregonensis group, Sweltsa sp., Cinygmula sp., and Neophylax sp. were most abundant in. Hydropsychidae, Ephemerellidae, Brachycentridae, and Naididae were found in all stream classes but increased in abundance in rivers with low conductivity, alkalinity and T K N levels. In previous studies, Hydropsychidae were most abundant at deciduous sites (Corkum 1990) or in 4lh and 5"' order streams with deciduous trees and grasses in the riparian vegetation (Ward 1986). Ephemerellid genera were diverse in both the interior streams and large rivers, however Ephemerella inermis/infrequens dominated at the large river sites. Brachycentridae are ubiquitous and can be found throughout the Holarctic region from cold mountain springs to marshy rivers (Wiggins 1996). The genera Brachycentrus occidentalis occurred at the large river sites while B. americanus occurred at the interior stream sites. The River Continuum Concept (RCC) proposed some general characteristics of "the structure and function of communities along a river system" in relation to the abiotic environment (Vannotee^ al. 60 1980). Community composition in pristine river systems should shift from one dominated by shredders (CPOM) and collectors (FPOM) in upstream reaches, to one dominated by collectors and grazers in intermediate reaches, and then to communities primarily composed of collectors in downstream reaches (Vannote et al. 1980; Hawkins and Sedell 1981; Cummins 1988). This change should occur as the relative contribution and sources of autochthonous and allochthonous inputs change. My study supports the R C C in that C P O M shredders were relatively more abundant in the headwaters than downstream in the large rivers. As well, collector-gatherers made up a larger proportion of the community in the large rivers than in the small streams. Contrary to predictions made by the RCC, my study found shredders and grazers on average made up 20 % of the functional composition of the large river communities. The large river samples were collected from the river margins, which could bias composition. However, Rempel (1997) found consistent proportions of functional groups up to 3 m in depth in the Fraser. As well, the river margins are the habitat used primarily by invertebrates (Rempel 1997). Table 3.8: Summary of the environmental variables correlated with benthic community composition from this and previous studies. For abbreviations see Table 2.2. Reference Geographic Region Environmental Variables Furse et al., 1984 Great Britain Ormerod and Edwards, 1987 River Wye Corkum, 1989 northwestern North America Corkum, 1990 Richards et al, 1993 Tate and Heiny, 1995 Dymond, 1998 (this study) eastern deciduous forest biome, southwestern Ontario, Canada Saginaw Bay catchment, Michigan, U S A South Platte River basin, Southern Rocky Mountains to the Great Plains, USA southwestern British Columbia, Canada alkal, N 0 3 N 0 2 - N , substrate, slope, elev, Q, dist, width, and depth pH, total hardness, associated geology, slope and dist. Biogeographical features: latitude, elev, slope Hydrological variables: meanV, meanD land use/riparian vegetation habitat/channel morphology (substrate composition and riparian vegetation). temp, cond, organic nitrogen, NH 4 -N , totalP, N 0 3 N 0 2 - N , slope, and width cond, alkal, T K N , N 0 3 N 0 2 - N , width, meanD, maxD, maxV, Q. 61 Both the R C C and food web theory predict that predator/prey ratios should remain constant. Freshwater invertebrate predator/prey ratios range between a mean of 0.29 in large (species- rich) collections to 0.48 in small (species-poor) collections (Jeffries and Lawton 1985). In my study, invertebrate predator/prey ratios ranged between 0.01 in the Thompson River and 0.46 in the North Alouette River. These ratios were fairly consistent within a site although they were not constant between study sites or stream classes, and they were consistently less than those previously found in the literature (Jeffries and Lawton 1985). There are two possible explanations for this discrepancy. First, vertebrate predators were not considered in this study and they may account for a larger proportion of the predators than in other systems. Second, it is difficult to accurately classify invertebrates into functional groups. Merritt and Cummins' (1996) functional group classifications are only defined for some genera; however, species within a genus may have different food resources (e.g. Hawkins 1984), and taxa may change food sources between sites and seasons, or with life cycle stage (e.g. Chapman and Demory 1963; Fuller and Stewart 1977; Martinson and Ward 1982). Shifts in functional group composition between the various stream classes shown in Fig. 3.4 may be obscured in the ordination (Fig. 3.8a) because of differences in abundance between large rivers and small streams. Abundance of invertebrates in all functional groups increases from the large river sites to the small streams; greatest abundance was attained in Beak Creek. Seasonal Patterns of Benthic Communities Seasonal variation in community composition was less than between-site variation and the variation between stream classes in this study. The relatively small amount of seasonal variation in community composition compared to spatial variation is also common to other studies over a range of spatial scales. Both Hawkins and Sedell (1981) and Corkum (1990) found less seasonal variation than spatial variation when samples were collected from Ist to 1th order streams or when samples were collected over a large area within a biome. In those studies, the relatively small amount of seasonal variation may have been a result of large spatial differences. However, Matthews et a/. (1991) and Death 62 (1995) also found only small amounts of seasonal variation relative to spatial variation when samples were collected within a 4-km stretch of a stream or from streams with similar physical and chemical characteristics. Studies documenting that seasonal variation of the benthic community was greater than spatial variation were conducted in either intermittent streams in Australia (Boulton and Lake 1992; Closs and Lake 1994), or in a Mediterranean stream with distinctly unfavorable seasons (Doledec 1989). In intermittent streams, seasonal change was related to time since drought or flooding; the number of taxa and/or number of individuals increase as the period of constant streamflow increases (Harrel and Dorris 1968; Boulton and Lake 1992; Closs and Lake 1994). In the Mediterranean stream, seasonal changes of benthic composition resulted from taxa able to tolerate strong, sudden spates in the winter, to those able to tolerate elevated water temperatures in the summer, with a transition period between (Doledec 1989). Extreme conditions such as those mentioned above were not found in the streams included in this study, resulting in no drastic changes in composition. In addition, the extent to which the communities in my study can change seasonally may be limited by the short period of warmer temperatures. Within a study site there were seasonal changes in community composition and invertebrate abundance. The variation resulted from changes in relative and absolute abundances as invertebrates moved through their life cycles (see also Giberson and Hall 1988). Hynes (1970) proposed a general pattern of annual insect abundance in small temperate streams (Fig. 3.1). The mechanism of seasonal change in the small streams of my study was similar to that established by Hynes (1970); however, the pattern of abundance differed slightly probably because of local environmental conditions and the fauna specific to each site. According to Hynes (1970), invertebrate abundance should decrease during the spring season with the emergence of many taxa. This loss of numbers may be offset as the eggs of summer species hatch. In the interior streams, however, greatest abundance was achieved in the early spring. This early spring peak corresponded with overlapping generations of ready-to-emerge winter stoneflies and the beginning of recruitment of the next generation (e.g. Zapada cinctipes, Chap. 2). Abundance then 63 decreased later in the spring only to increase again in the summer with recruitment. The difference between Hynes's predictions and the results from the interior streams may be a result of temperatures. Thus, it may be more productive to compare the timing of annual temperature cycles rather than specific seasons. In the coastal streams, abundance was low in the spring samples. Summer was the season of greatest abundance and coincided with a variable peak predicted by Hynes (1970). This peak in abundance was a result of the hatching of eggs of both summer and univoltine taxa (Chap. 2). In the coastal streams, late spring through summer were more important seasons for invertebrate recruitment than was suggested by Hynes (1970). Macroinvertebrate abundance in the large rivers, increased from the late spring through summer and autumn and was maximum in the early spring. This pattern in the large rivers was different from that predicted for small streams by Hynes (1970), but was very similar to that found by Bothwell and Culp (1993) in the Thompson River. Bothwell and Culp (1993) found that many taxa emerged in the spring prior to the onset of the annual freshet and density remained low until after the freshet. Following the freshet, abundance then increased to a secondary annual peak in the autumn, during a period of lower, stable flows and high water temperatures. Annual peak abundance was reached late in the winter, and the insects were large and close to emergence. The seasonal change of the invertebrate community was not directly related to the seasonal change of environmental conditions in this study. The seasonal change of the faunal community was however, indirectly related to environmental conditions such as temperatures and food resources through the influence of the environment on invertebrate life cycles (Chap. 2). Many of the changes in composition observed are a result of increases and decreases in abundance that corresponded to timing of invertebrate life cycles. Mellin Creek provides an example of the relationship between environmental conditions, life cycles, abundance, and community compositon. Changes in community composition in Mellin Creek were positively correlated with changes in temperatures. The largest change in community composition at both the genus/species and family levels occurred between the late spring (1995) and summer samples at a time of maximum increase in water temperatures (Fig 2.7).Most taxa increased in 64 abundance between these dates, especially the tubificids, Prosimulium sp., Serratella tibialis, and Cinygmula sp. Increases in abundance were primarily a result of the hatching of eggs of summer species and summer generations of multivoltine taxa (Hynes 1970). Changes in composition between autumn and winter were slight because of a decline in recruitment and emergence rates. The R C C predicts that functional group composition should remain consistant through the year. (Vannote et al. 1980). Studies have found the functional composition of stream communities change seasonaly. Seasonal change of the functional composition would allow the community to take advantage of the change in energy input sources over time (Statzner and Higler 1985). In small streams with open canopies and deciduous riparian vegetation, shredders should increase in the autumn and winter when C P O M resources are at their greatest. In the summer, the communities in these streams should have the greatest abundance of herbivores because of the high primary production accompanying warm temperatures and high light availability. Shredder abundance in the interior streams did increase in the autumn, however, herbivore abundance was not greatest in the summer. Perhaps the riparian vegetation along the interior streams reduced light penetration to the stream channel. The riparian vegetation along the coastal streams was primarily coniferous, resulting in reduced seasonality of influx rate of C P O M and a fairly consistent abundance of shredders. My study demonstrated that the amount of seasonal variation changed between each site. Differences in the seasonal variation between each site are a result of the different taxa composing each community and differences in environmental conditions. Environmental conditions that influence the timing of life cycles, change seasonally, and differ among the three stream classes include discharge, water temperature, and food resources. The seasonal change in assemblage structure at each site was correlated with changes of few, if any, environmental variables. However, discharge seems to influence invertebrate abundance in the large rivers. The difference in the winter temperatures between the coastal and interior streams influences invertebrate life cycles (Chap.2) and patterns of abundance. The level of each food resource changes seasonally and abundance of each functional group follows the seasonal flux. 65 Chapter Four S E A S O N A L C H A N G E O F B E N T H I C M A C R O I N V E R T E B R A T E C O M M U N I T Y S T R U C T U R E I N S O U T H W E S T E R N B R I T I S H C O L U M B I A A N D T H E I M P L I C A T I O N S F O R B I O M O N I T O R I N G Introduction Biomonitoring can identify changes in the environment and stresses on a system, and it is becoming increasingly popular for detecting environmental impacts. Biological responses can reflect a change in both physical and chemical conditions over an extended period of time, and give ecological relevance to environmental change. Benthic invertebrates, fish, periphyton, and macrophytes have all been used to assess ecosystem conditions and water quality (Schindler 1987; Fausch et al. 1990; Rosenberg and Resh 1993). Benthic invertebrates are the most often recommended and used organisms (Hellawell 1986; Rosenberg and Resh 1993). Benthic invertebrate community composition is useful for biomonitoring because it is strongly influenced by local environmental conditions (e.g., Cummins et al. 1989; Corkum 1990; Cobb et al. 1992; Richards et al. 1993; Tate and Heiny 1995). In fact, benthic invertebrate communities are possibly ".. .the most sensitive tool for quickly and accurately detecting alterations in aquatic ecosystems" (Cairns and Pratt 1993, page 10). Predictive models that require detailed data on benthic macroinvertebrate community composition and associated physical, chemical, and biological conditions from a large number of unimpacted or reference sites can now be applied to biomonitoring. There are advantages to using predictive models for biomonitoring. First, they are fairly accurate in predicting community structure (75 to 100% accuracy) (Wright et al. 1984; Reynoldson et al. 1995; Reynoldson et al. 1995). Second, the models are sensitive to both physical and chemical stresses. Third, these models can be used to establish targets for remediation by comparing the predicted community composition of a test site to the actual community found there. A disadvantage of predictive models is that they can only be accurately applied to test sites that are within the range of environmental variables included in the reference sites. The models cannot extrapolate beyond the range of variability included within the reference data set. Since environmental 66 and community variables may change seasonally, the model may only provide accurate predictions for the season when the samples were collected, unless the extent of the seasonal environmental and community change is negligible or known. Benthic communities change seasonally (Hynes 1970; Boulton and Lake 1992; Death 1995), but how significant these seasonal changes are, and the implications for biomonitoring have not been fully established. Furse et al. (1984) recommended sampling seasonally when using predictive models; their samples, collected over three seasons, increased prediction accuracy. However, Death (1995) found that although benthic communities did change seasonally, the change was minimal. The time commitment and cost of collecting and identifying samples from a large number of reference sites over multiple sampling dates could prohibit use of predictive models for biomonitoring. Therefore, the significance of seasonal changes to the benthic community has to be established to determine whether seasonal sampling is required. In 1994, under the Fraser River Action Plan, Environment Canada began an extensive program to assess and monitor the Fraser River Basin (FRB) using benthic macroinvertebrate communities. Monitoring was to be done using the Benthic Assessment of Sediment (the BEAST) predictive model (Reynoldson et al. 1995). As a result, reference samples were collected from > 200 sites in the Fraser River Basin in the autumn of 1994, 1995, and 1996. Sampling once in a year may not provide enough information to accurately apply the BEAST model in other seasons because the benthic community structure may change seasonally in response to changing environmental factors such as temperature, discharge, and food resources. Understanding the extent of seasonal changes to benthic communities in the FRB would enable determination of the applicability of model predictions throughout the entire year. The Fraser River Basin covers > 230,000 km 2 of British Columbia and includes 10 different ecoregions (Fig. 2.1), which have a different climate, vegetation, and sometimes, a different discharge regime. There is the potential for communities to change seasonally at different rates with the timing of predictable seasonal signals (temperature and discharge). 67 The purpose of this study was to determine whether seasonal changes to benthic macroinvertebrate community structure are sufficient to affect the accuracy of predictions made with the B E A S T model. Without this study, the validity of results obtained from test samples collected outside of reference-sample season could not be assured. In order to ensure the results of this study apply to a number of stream types, seasonal change of community structure was assessed for 8 streams from 3 different classes of streams. The three different stream classes comprised interior plateau streams, coastal streams, and large rivers. Each of these classes has different climates, vegetation, and discharge regimes. The B E A S T model involves using a clustering technique to group the reference sites based on faunal composition. Stepwise discriminant function analysis is then used to identify environmental variables which are best able to discriminate between the reference groups. These environmental variables can then be used to predict membership of a test site to a reference group. An environmental impact is inferred if community composition at the test site differs from the community composition found at reference sites within the group. Although the source of the impact is not directly known, its origins can be hypothesized. Data for this study were analyzed at both the lowest practical taxonomic level (genus and species) and at the family level. The genus/species level was analyzed because identification to species is considered to be critical to biomonitoring (Resh and Unzicker 1975; Resh and McElravy 1993), although not all larvae can be identified to species. Stress tolerance may differ within a family or even within a genus. As a result, species level information is more sensitive and increases the ability to detect subtle changes (Zamora-Munoz and Alba-Tercedor 1996). Seasonal changes may be more detectable at species level. Identification to the family level is cost and time efficient, achievable for most taxa, can be consistently done between laboratories, and would be easier for future users of the predictive model. Whether the family level of invertebrate classification is acceptable for biomonitoring seems to depend on the study, and the questions being addressed (Resh and McElravy 1993). Zamora-Munoz and Alba-Tercedor (1996) found family level of identification to be sufficient for monitoring water quality in 68 streams, but species data were required to determine the exact biological response to stress. Monitoring using family level data has been recommended for marine (Warwick 1993) and lake (Jackson and Harvey 1993) ecosystems. The sensitivity to seasonal changes of family level identifications was determined in this study. Methods Seasonal change of benthic macroinvertebrate community structure was assessed for eight rivers in southwestern British Columbia (Fig. 2.1). The characteristics of each stream are described in chapter 2. These study sites will be referred to as seasonal test sites. Benthic invertebrate samples and associated environmental information were collected from reference sites throughout the Fraser River Basin in the autumn of 1994, 1995, and 1996 by Environment Canada. My study used data only from reference sites sampled in 1994 and 1995. Environment Canada's site selection protocol was described by Rosenberg et al. (in preparation). Al l FRB reference invertebrate samples and environmental condition samples were processed by Environment Canada. Samples were collected from the study streams over five sampling dates between April 1995 and April 1996 to identify seasonal changes of benthic macroinvertebrate community structure (Figs. 2.2a,b and c). The fall sampling date corresponded with Environment Canada's 1995 collection of reference samples from throughout the FRB. My study sites were also sampled by Environment Canada in 1995 and became part of their reference and test site data base. Invertebrate Samples Five one-minute kicknet samples were collected from each site on each sampling date. Sample collection protocols can be found in Chapter 2. The kicknet was triangular (38.5cm per side) and a mesh of 400-wm The size of the mesh to be used and the sampling time were determined by Environment Canada. Although a 200-wm mesh retained a greater number of invertebrates, there was no significant difference 69 in the number of taxa retained between the 200 and 400-wm mesh sizes (Rosenberg et al. in prep.). Thus, the 400-wm mesh was used to reduce sample processing time. One-, 3- and 5-minute sampling times were tested by Rosenberg et al. (in prep.) who found no significant difference in the number of major orders or total number of invertebrates collected (when numbers were standardized to number per unit time). The number of taxa collected was significantly lower in the 1-minute samples than in the 3- and 5-minute samples, but there was no difference in the number of taxa collected between the 3- and 5- minute samples (Rosenberg et al. in preparation). Thus, I collected multiple 1 -minute kicknet samples, processed three samples per site per date and then combined the results (number of in vertebrates/1-minute kicknet sample). The procedures used for sorting and identification are described in chapter 2. Invertebrates were identified to the lowest practical taxonomic level. Chironomidae were not identified beyond subfamily, and Ostracoda, Turbellaria, and Copepoda were not identified beyond class. Environmental Conditions Environmental variables were measured either when the benthic samples were collected or were determined from maps (Table 2.2). The variables included those that are related to benthic invertebrate community composition (Wright et al. 1984; Corkum and Currie 1987; Omerod and Edwards 1987; Corkum 1989). I measured environmental variables the same way as Environment Canada for their collections at reference sites (see Chapters 2 and 3). Data Analysis All data analysis was done at the species or genus level with some exceptions (listed above and in Appendix 3.1A) and the family level (exceptions listed above and in Appendix 3.IB). Rare taxa were removed from the data set because they create noise which obscures patterns in classification analysis. Two levels of data censorship were used: > 0.1 % and > 0.5 % occurrence. These two cut-off levels were used to determine any effect of data censorship on the results. Because patterns of composition (rather 70 than just numeric differences) were deemed to be important, the faunal data were log transformed (logio(x+l)), which reduced the influence of the more abundant taxa. The data analysis used in this study is similar to that used by Wright et al. (1984), Reynoldson et al. (1995), and Parsons and Norris (1996). Cluster analysis of the faunal data from the 127 FRB reference sites produced reference groups as detailed below. Stepwise discriminant function analysis and correlation analysis were then used to select environmental variables, which became potential predictor variables in the discriminant model. Seasonal test sites were predicted to the reference groups using the discriminant function model. Probability ellipses drawn around site scores in the ordination space defined by the reference sites were used to determine how (dis)similar the seasonal test site invertebrate community was to the communities in each reference group. Grouping of Sites Clustering of the faunal data to group the FRB reference sites was done using PATN, a pattern analysis software package developed by CSIRO in Australia (Belbin 1993). The Bray and Curtis association measure was used because it performs consistently well in a variety of tests, and is recommended as the most appropriate association measure for ecological data (Beals 1984; Faith et al. 1987). Clustering was done using an agglomerative hierarchical method with the Ward fusion technique (Cao et al. 1997), which calculates a central point for each possible combination of 2 clusters. The total sum of squared distances from this point to all objects in this hypothetical cluster is then evaluated. The association of the two clusters that results in the smallest sum of squares is then taken to be the new cluster. Dendrograms of the classifications were used to identify reference site groups. The reference site macroinvertebrate data were ordinated using Semi-Strong Hybrid Multi-Dimentional Scaling (SSHMDS; Belbin 1991 and 1993). The relationship between taxon abundance and environmental data to the ordination space was determined using the principal axis correlation (PCC) module in PATN. The PCC module finds a linear combination of the ordination axes that has maximum correlation with each variable. Monte Carlo significance tests (MCAO with 100 permutations; Belbin 71 1993) were performed to test the significance of the environmental and invertebrate correlations with the ordination. A variable was considered to have a significant correlation if the observed correlation was higher than that of all of the 100 random permutations. The locations of the correlation vectors for the significant variables were plotted in the reference data ordination space. Selection of Predictor Variables Environmental variables were selected as potential predictor variables of sites to a group by two techniques: 1) if they had significant correlations with the reference site ordination axes, and 2) through a stepwise discriminant function analysis (DFA) (Proc Stepdisc; SAS Institute, 1996). The stepwise procedure enters one environmental variable at a time and selects those variables that are best able to discriminate between the groups determined from the cluster analysis of the invertebrate data. The selection procedure was stepwise, and the significance levels for variables to enter and to stay in the stepwise D F A were both set at 0.05. An assumption of Discriminant Function Analysis (DFA) is that the discriminator variables have a multivariate normal distribution (Jackson 1983). All environmental variables were tested for a normal distribution to meet this assumption. Each variable was appropriately transformed to achieve normal distributions. Variables requiring no transformations included: order, Julian day, latitude, longitude, pH, temperature, dissolved oxygen, percent silt, percent clay, macrophytes, grass, deciduous, shrub, conifers, framework, matrix, embeddedness, maximum velocity, mean velocity, and elevation. Variables requiring a log-base 10 transformation included: conductivity, alkalinity, bankfull width, channel width, maximum depth and mean depth. Variables requiring a square root transformation included: percent gravel, percent sand, nitrate - nitrogen, T K N , and slope. Arcsin of the square root of percent silt/clay was used to transform this variable. Transformations were applied to the data set prior to analysis. The environmental data set contained both continuous (e.g., temperature, pH, latitude, longitude, etc.) and categorical variables (e.g., framework, matrix, and riparian vegetation). This type of data set is 72 not optimal for linear discriminant function analysis (Norusis 1985). However, D F A performs reasonably well when a mixture of variables are used (Gilbert 1968; Norusis 1985). The final selection of the environmental variables to be used as predictor variables in the discriminant model was done using a D F A (Proc Discrim Crossvalidate; SAS Institute 1996). The cross-validation method was used to test the discriminant model. The cross-validation method removes each site (one at a time) from the data set, predicts each site to a group, and then determines the misclassification rate. The optimum model would have the lowest error rate. A model using the set of environmental variables selected in the stepwise analysis was first tested using DFA. Environmental variables with significant correlations were then added to the analysis and kept or removed as required to achieve the lowest possible error rate. The final discriminant model included those environmental variables that achieved the lowest possible misclassification rate. Prediction of Seasonal Test Sites to Groups The discriminant model and the environmental data were used to predict the 8 seasonal test sites to reference groups (Proc Discrim Testdata; SAS Institute 1996). Invertebrate data for each test site were ordinated (SSHMDS; Belbin 1993) with the reference invertebrate data of the group to which the test site was predicted based on the fall sampling. Probability ellipses (90, 99, and 99.9 %) for the ordination scores of only the reference site data in each group were calculated (SYSTAT 1993) and plotted. The location of the seasonal test sites in relation to the probability ellipses was then examined. The use of a series of probability ellipses for establishing test site impairment enables invertebrate community responses to be viewed along a gradient of potential impact response levels. The decision as to what level of departure from the reference condition constitutes an unacceptable level of impairment is ultimately a subjective one. The choice of a 90% probability ellipse as the first band was based on previous water quality studies (Wright 1995; T.B. Reynoldson unpublished) and the reduction of potential Type II errors (classifying a stream asunimpacted when it is). Because there is no other way of proportionally assigning levels of impairment, two other probability ellipses were selected that created 73 stress bands of equal widths (99% and 99.9% probability ellipses). Sites between the 90 and 99 % probability ellipses are considered to be potentially stressed, since there is 1 chance in 10 that sites will fall in this band as a result of normal variability (Type I error rate). Sites between the 99 and 99.9% probability ellipses are considered stressed, because there is only 1 chance in 100 that the sites are incorrectly described as stressed. Sites which fall outside of the 99.9 % probability ellipse are considered to be severely stressed. Results The initial FRB reference data set was composed of 214 taxa from 127 sites. The seasonal test site data set was composed of 152 taxa from 8 sites. Some of the difference in the number of taxa can be accounted for by the larger area and number of sites included in the reference data set. The reference data set also consisted of most Diptera identified to at least the genus level, whereas Chironomidae were only identified to the subfamily level in the seasonal test site data base. Genus/species data were grouped into higher taxonomic levels (e.g. Heptageniidae, Baetidae, Chironomidae subfamilies, see Appendix 3.1 A) because of discrepancies in identifications between the two data sets. Similar grouping of taxa was not required for the family level analysis. The genus/species taxonomic level having a cut-off of > 0.1% and 0.5% occurrence resulted in 61 and 24 taxa, respectively, in the data sets (Appendix 3.1A). The family data with 0.1 % and 0.5 % cut-off resulted in 32 and 23 taxa used in the analysis (Appendix 3. IB). Analysis of the data at two different taxonomic levels and two different levels of censorship resulted in the cluster analysis forming slightly different groups with the four data sets. To create models with the lowest misclassification rate, different predictor variables were selected for each level of analysis. The lowest achievable error rates were somewhat different between the models. The seasonal test sites were sometimes predicted to different groups with the 4 different levels of analysis. I begin with the model for the genus/species level data (0.1% cut-off) and then consider how the level of censorship and taxonomic resolution affect the models. 74 Results for Genus/Species Data (0.1% cut-off) Grouping of Sites Examination of the dendrogram resulting from the cluster analysis of the genus/species level data (0.1% cut-off) revealed 3 groups (Fig. 4.1). These groups are important because they form the basis of the remaining analysis. A selection of too few groups would result in a low misclassification rate by the DFA, but the model would not be sensitive to slight variations in the data. A selection of too many groups would create a very sensitive model and a high rate of site misclassification. Three groups were selected as an appropriate trade-off between sensitivity and misclassification; there was a sufficient number of reference sites in each group to ordinate with the test sites predicted to each group. Ordination of the reference invertebrate data illustrates the location of the groups in ordination space. Each group is fairly distinct although there is some overlap between them; most of the separation is along axis 1 (Fig. 4.2a). Group 1 was composed primarily of streams from the interior of B.C. Generally, these streams are at high elevations, grass dominates the riparian vegetation, the substrate is embedded, and they have high total Kjeldahl nitrogen (TKN), alkalinity and pH (Fig. 4.2b). Group 3 consisted of coastal streams, upper Fraser River sites (headwaters of the Fraser River located in the Rocky Mountains), and some Fraser River mainstem sites. The Fraser mainstem reference sites were all located upstream of Prince George, the first major urban center and location of pulp mills on the river. The group 3 streams have some combination of the following environmental characteristics: high gradients, large dominant substrates (framework), and high mean and maximum depths and channel and bankfull widths. Nitrate and nitrite nitrogen concentrations are highest in group 3 streams (Fig 4.2b). Group 2 streams are intermediate between group 1 and 3 streams in ordination space and in environmental characteristics. Invertebrate abundance rather than composition drives the ordinations even though the invertebrate data were log transformed. Most of the invertebrates with significant correlations increase in abundance in group 1 and 2 sites (Fig. 4.2c). The exception to this is Taenionema spp. (Taeniopterygidae) and Drunella doddsi (Ephemerellidae), which increase in abundance in group 3 sites (Fig. 4.2c). 75 CHI Oil 94 CHI 0119S NIC04195 NIC05195 CHI 02194 Group 1 Group 2 Group 3 Figure 4.1. Dendrogram illustrating reference site groups for genus/species level data (>0.1 % occurrence). C N C N O £ s i x y . C N C N C U 4= C D -4—» C CD T3 CO A3 CD V, CD > g c o CD to C3 cu 1 0 ."2 0 0 'co ""J CD O S 11 CU T-C S CD ,53 > C « • 1 CD m ^ CD > ^ cd <+H O c a o a C N CH 3 o O 9 cd w •g 13 A VH C D O 3 CD O CX >H 5? O CO 3 s n 00 w « • CM Tt CU • 3 o cd O £ a •s .a ° a C O C O £ 3 « ft o 6 0 • — ' C U I ° _ cd c u e bb cfl c^d 'co ' - M "B. £ g g "S - a ^ cu o to H 'rl •2 ^  > > 7 3 -2 .—. <u rt «d ir; M C b « CJ r O B 2 ^ O > C J g 2 H cj « 1 / 5 5 ^ «o cu oo „ "cd ^ cd b O 6 | II i s s C * co •2 c ^ .£ « g 3 * ° CJ CU C O p M rH eo p-1 o C M A fa O w co 60 g •3 cd o o 'I CJ .ft, u •Cf cd cj •5 cj P H P H C O CO T3 cd 43 td 43 +n co ON Id °° L H H •S d t i || CJ __ > CJ g > CJ •— 1 43 co + H co C+H CJ o 4a L H "P ^ c2 Z C O - a ' s " O CJ Cd O H O co tZl co o =» C3 c5 a a CJ ,. 6 0 II 43 LO P. ^  CJ P. 1 0 2 rS i3 P H ^ + H C O cj u ~ Q PQ S ^ II p_ ON r-- „ p, •§ & g C2 S CO -ft CO sa CJ ST3 60 3 CJ 43 &• cd is B T3 CJ —i T 3 ,2H 'FJ a ^ cd a g il .2 -H CN o 5T CO ik. 2 "I £ £ °°. " II & 2 co co CJ cd .g -3 ca "o o 43 P H C H CO C3 £ CU C>H ? CJ -*-» — H » 6 II ^ 8 £ CN 43 CU CJ CJ L H CJ CO CJ +J 2 5 a 2 ^ a « B I O R CO Q K -Q CD •° tj S • - H Cj C3 ~ o f2 N q .11 s 00 > 11 " r—I VO r—I V© r-H T—< 2 srxy 79 Selection of Predictor Variables The discriminant model that included variables selected by the stepwise analysis plus some of the variables with significant correlations had lower error rates than models with just variables selected by the stepwise analysis alone. The model with the lowest error rate was the one selected for further use. For the genus/species data (0.1 % cut-off), the lowest achievable total error rate was 30.9 %. The predictor variables used in the genus/species model included: T K N , the presence or absence of conifers in the riparian zone, stream channel gradient (slope), pH, latitude, longitude, elevation, and alkalinity (Table 4.1). Figure 4.3a summarizes the separation of the reference groups, the variance explained by each discriminant function axis and the relative contributions of each of the environmental variables in discriminating between the different groups. For groups based on invertebrates identified to the lowest taxonomic level (0.1 % cut-off), the first discriminant function axis explains 79 % of the variation and the second axis explains 21 % of the variation. Of the environmental variables used to discriminate between the groups, T K N , alkalinity, longitude, and elevation discriminate the most along the first axis, whereas, latitude, conifers, slope and pH discriminate the most along the second axis (Fig. 4.3a). Prediction of Seasonal Test Sites to Groups The discriminant model, including the environmental variables chosen as predictor variables, was used to predict the autumn sampling of the seasonal test sites to a reference group (Table 4.2). The seasonal test sites from the Nicola drainage basin were predicted to group 1, which contained reference sites with similar environmental conditions from the interior of B.C. The coastal streams and the large river test sites were both predicted to group 3. These test sites have similar latitudes, longitudes, and elevation, but two of the coastal streams are very different rivers from the rest in terms of size, temperature, and riparian vegetation. Large rivers were not separated from the small streams because channel width, depth, and temperature were not included in the discriminant model. 80 81 H-l 'S o O cu a cd h-l f t o u o CN Q . o • f t CU ; r< CU 4 - 4 O CU a C3 VH PH CU CH o c/5 a o s-0 T • CN r H VO o r H O I CA CN CN CN (% V£Z) Z uoipuru; juBuraiuosia S6.S CN T 3 - B r H ca -GT 9 .2 3 "5 toil' £ c cu o W r J • 4 03 Ai CN O I -0 i c5 cj PH O 5o 4 o CN CV o a • r H . M •CN CN ( % 1 3 ) 3 UOipUTt} JUBUIUIUOSTQ too CQ CJ w 4 0 82 Table 4.2: Summary of the groups to which the seasonal test sites were predicted based on the autumn sampling and where the seasons fall in the ordination space. Within the 90% probability ellipse (in), between the 90 - 99 % ellipses (>90), between the 99 -99.9 % (>99), or outside of the 99.9 % ellipse (>99.9). Site Sample Lowest In/Out Lowest In/Out Family In/Out Family In/Out (>0.1 %) (>0.5 %) (>0.1 %) (>0.5 %) Mellm spring95 1 >99 1 >90 2 >99.9 2 >99.9 Mellin summer 1 >90 1 in 2 in 2 in Mellin autumn >99 >99 2 in 2 in Mellin winter 1 >99.9 1 >99.9 2 >99 2 >99 Mellin spring96 1 >90 1 in 2 in 2 in Glimpse spring95 1 in 1 >90 2 >90 1 >90 Glimpse summer 1 in 1 in 2 >90 1 >90 Glimpse autumn >90 >90 2 in 1 >90 Glimpse winter 1 >90 1 >90 2 >90 1 >90 Glimpse spring96 1 >90 1 in 2 >90 1 in Beak spring95 1 m 1 in 2 in 2 in Beak summer 1 in 1 in 2 in 2 in Beak autumn 1 in in 2 in 2 in Beak spring96 1 in 1 in 2 in 2 in Spring spring95 3 >90 1 in 2 in 2 in Spring summer 3 >90 1 in 2 in 2 in Spring autumn 3 in in 2 >90 2 in Spring winter 3 in 1 in 2 >90 2 >90 Spring spring96 3 >90 1 in 2 in 2 in Mayfly spring95 3 in 1 in 1 in 1 in Mayfly summer 3 >90 1 in 1 >90 1 in Mayfly autumn 3 >90 >90 1 in 1 in Mayfly winter 3 >90 1 in 1 in 1 in Mayfly spring96 3 >90 1 in 1 in 1 in N.Alouette spring95 3 in 2 in 3 in 3 in N.Alouette summer 3 in 2 in 3 in 3 in N.Alouette autumn 3 >90 2 in 3 in 3 >90 N.Alouette winter 3 >90 2 >90 3 in 3 >90 N.Alouette spring96 3 in 2 in 3 in 3 in Fraser spring95 3 >90 3 >99 3 >90 3 >90 Fraser summer 3 in 3 in 3 in 3 in Fraser autumn 3 >90 3 >90 3 >90 3 >90 Fraser winter 3 in 3 in 3 in 3 in Fraser spring96 3 in 3 in 3 in 3 in Thompson spring95 3 >90 3 >90 3 in 3 >90 Thompson summer 3 >90 3 >90 3 >90 3 >90 Thompson autumn 3 >90 3 >90 3 >90 3 >90 Thompson winter 3 >90 3 >90 3 >90 3 >90 Thompson spring96 3 >90 3 >90 3 in 3 >90 83 To assess the significance of seasonal changes to the invertebrate community, seasonal test site data were ordinated with reference site data from the group to which the test site data were predicted based on the autumn sampling (Fig. 4.4). The 90, 99, and 99.9 % probability ellipses are drawn around the reference site invertebrate data and the seasonal test sites were added to the plots. Table 4.2 summarizes the location of the seasonal test sites in the ordination space of the group to which the test site was predicted based on the autumn sampling. A seasonal test site sample had to be located within the 90 % probability ellipse for all axes to be considered the same as the reference condition. Difference of the test site from the reference condition depends on how far out from the 90 % ellipse the test site falls. Each test site benthic community changed seasonally, in ordination space, in different directions and with different magnitudes. This seasonal change caused some test site sampling dates to fall outside of the 90 % probability ellipse, (Fig. 4.4 or Table 4.2) (e.g. Glimpse, Spring, Mayfly Creek and the North Alouette River). These streams had sampling dates that fell outside the 90% ellipse, but had no apparent impacts. Mellin Creek and the Fraser River varied seasonally more than the other streams, and some sampling dates fall outside the 90 % and even 99 % ellipses (Fig. 4.4). Beak Creek is the only stream in which all sampling dates fell within the 90 % probability ellipse. Seasonal changes in the invertebrate community affect the results of this analysis, but the results change with each stream. Effect of Data Censorship Level on the Results Grouping of Sites - Genus/Species Data (0.5% cut-off) Examination of the dendrogram for the genus/species data (0.5 % cut-off) revealed 3 groups. Many of the sites included in the groups were the same as included in the 0.1 % cut off. However, some sites were placed into different groups. In general, group 1 consisted of interior stream sites, group 2 consisted of Fraser River sites, and group 3 consisted of coastal stream sites. A plot of the reference invertebrate data illustrates the location of the groups in ordination space (Fig 4.5a). The results are similar to the analysis at the 0.1 % cut-off level. Group 1 sites are generally distinct from group 3 sites. Group 2 sites are intermediate between, and overlap with, group 1 and 3 sites 84 -— I " 1 1 ' 9 0 % / 9 9 % 7A99.9%_ l i i / \ \ \ \ 1 1 ^ 3 \ V / I 1 -2.5 1.5 Y 0.5 h -0.5 Y •1.5 Y -2.5 -1.5 -0.5 0.5 1.5 2.5 -2.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 0.5 h -0.5 Y 2.5 1.5 ro 1 0.5 -0.5 -1.5 h -2.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 Axis 1 Axis 1 Figure 4.4: Location of seasonal test sites in the ordination space of the group to which the sites were predicted. The sites were predicted to their respective groups based on the autumn (3) sampling of the test sites. The invertebrates were identified to the genus/species level, 0.1 % cut-off. A - group 1 ordination space, B - group 3 ordination space. The 90, 99, and 99.9 % probability ellipses indicated are based on the ordination scores of the reference data for the respective groups. OBeakCk, O Glimpse Ck, V Mellin Ck, • Mayfly Ck, A N . Alouette R., •fr Spring Ck, < Fraser R., * Thompson R. 8 5 (Fig. 4.5a). Group 1 sites are primarily from interior streams with high elevations and T K N levels, and grass dominates the riparian vegetation (Fig 4.5b). Group 2 and 3 streams have some combination of these environmental characteristics: large dominant substrates, high maximum and mean velocities, warm temperatures, large bankful and channel widths, deeper mean and maximum depths, and a high concentration of nitrite and nitrate nitrogen. Most invertebrates increase in abundance in the group 1 sites. Only Taenionema spp. (Taeniopterygidae) and Hydropsyche spp. (Hydropsychidae) increase in abundance in group 2 and 3 streams (Fig. 4.5c). Selection of Predictor Variables The cluster analysis of the genus/species data (0.5 % cut-off) resulted in a slightly different combination of reference sites in each group than analysis at the 0.1 % cut-off level, so the optimal predictive model was slightly different for each analysis level. Environmental variables selected for the predictive model included T K N , conifers, stream channel gradient, pH, latitude, longitude, channel width, bankfull width, size of the dominant substrate, and mean depth (Table 4.1). The lowest achievable misclassification rate was 23.2 %, 7.7 % better than when the analysis was done with the 0.1 % cut-off (Table 4.1). Figure 4.3b summarizes the separation of the reference groups, the variance explained by each discriminant function axis, and the relative contributions of each of the environmental variables in discriminating between the groups. For groups based on invertebrates identified to the lowest taxonomic level, (0.5 % cut-off), the first discriminant function axis explains 74.6 % of the variation and the second axis explains 25.4 % (Fig.4.3b). The environmental variables used to discriminate between the groups and their relative contributions are shown. T K N , bankful width, channel width and mean depth, and framework discriminate the most along the first axis. Latitude, longitude, conifers, slope, and pH discriminate the most along the second axis. X < o & 3 ^ S CN PH o O a T3 cd J3 t; CJ > S ex O T3 g a 42 cj o g <N c § ° ^ 11 cj 'cn J3 cd CO PQ ^ CJ ^ 53 § o cd S i cd w • 5 13 T3 > cj 0 -• • . u "TJ cj 1 S CN -a CN ^ O HH CN Z S J X V CN 'a bo P . " PH CD Lo CD <3 -2 II II a O Cd « a Si IS CO "o T3 O o PH PH cd CD K O 1 PH PH CD V, CD > a °1 I  1 C H II 8*2 CD A3 o ND '-' OS fe © c2 « C O © s> II .a co T 3 co cd <D o a ^ H a, <+H CD O a I o o o NO PH II PH CD 3 1 cd .6 a o -a; C H a a. in CD cd 32 'e CD cd PH CD s CD > L H CD £ '§ CD I-s WD CD a PH . O , co ~ CD JS H I N 33 3 .gf S I I fa O 'cf OS 89 Prediction of Seasonal Test Sites to Groups The groups to which the seasonal test sites were predicted based on the autumn sampling are summarized in Table 4.2. Seasonal test sites from the Nicola drainage basin and two of the coastal streams (Mayfly and Spring Creek) were predicted to group 1. Prediction of the coastal streams to group 1 rather than group 3 (as in the 0.1 % cut-off analysis) was caused by the use of channel and bankfull width, mean depth, and conifers as predictor variables in the DFA. Group 1 reference sites were primarily interior streams. The North Alouette River was predicted to group 2; the reference sites in this group are primarily upper Fraser and Fraser mainstem sites. The Fraser and Thompson river test sites were predicted to group 3, which was primarily coastal streams. Figure 4.6 and Table 4.2 summarize the location of the seasonal test sites in the ordination space of the group to which the test site was predicted based on the autumn sampling. As in the analysis of the 0.1 % cut-off data, each of the test sites changed seasonally in the ordination space in different directions and with different levels of magnitude. However, at the higher cut-off level of 0.5 % there were fewer dates which fell outside the 90 % probability ellipse (Table 4.2). The higher cut-off level resulted in less variation being expressed in the data set and a less sensitive analysis. Grouping of Sites - Family Data (0.5% cut-off) Family level data were used following the same procedure as above (0.5% cut-off level) to test the effect of taxonomic resolution. Cluster analysis of this data set again divided the reference sites into 3 groups. Many of the sites in each of the groups differed from those included in the groups as defined by the genus/species data (0.1 and 0.5 % cutoff). The reference sites included in groups 1 and 2 were primarily from interior streams whereas the reference sites in group 3 were primarily from coastal streams, the upper Fraser, and the Fraser mainstem. A plot of the SSHMDS scores from the reference invertebrate data illustrates the location of the groups in ordination space (Fig. not shown). Group 1 sites were generally distinct from group 3 sites. Group 2 were intermediate sites and overlapped with group 1 sites. Group 1 and 2 sites were primarily 90 Figure 4.6: Location of seasonal test sites in the ordination space of the group to which the sites were predicted. The sites were predicted to their respective groups based on the autumn (3) sampling of the test sites. Invertebrates were identified to the genus/species level, 0.5 % cut-off. A - group 1 ordination space, B - group 2 ordination space, and C - group 3 ordination space. The 90, 99, and 99.9 % probability ellipses indicated are based on the ordination scores of the reference data for the respective groups. O Beak Ck, O Glimpse Ck, V Mellin Ck, • Mayfly Ck, A N. Alouette R., # Spring Ck, < Fraser R., * Thompson R. 92 interior stream sites with high elevations, embedded substrates. Group 1 sites have high conductivity, and shrubs and conifers dominate the riparian vegetation. Group 2 sites have high T K N and alkalinity levels and grass dominates as the riparian vegetation. As in the previous analyses, group 3 sites have some combination of these environmental characteristics: large dominant substrates, high maximum and mean velocities, warm temperatures, large bankfull and channel widths, deep mean depths, and a high concentration of nitrite and nitrate nitrogen. Most invertebrates increased in abundance in group 1 and 2 sites. Only Taeniopterygidae, Hydropsychidae, and Naididae had high abundances in group 3 streams. Selection of Predictor Variables The optimal predictive model for the family data (0.5 % cut-off) was slightly different from the previous models. Environmental variables selected for the predictive model included T K N , conifers, shrubs, grass, stream channel gradient, pH, conductivity, and channel width (Table 4.1). Error rates were similar between analysis at the genus/species level and family level for a given cut-off value (Table 4.1). Figure 4.7 summarizes the separation of the reference groups, the variance explained by each discriminant function axis, and the relative contributions of each of the environmental variables in discriminating between the different groups. For the 0.5 % cut-off, the first discriminant function axis explains 84.4 % of the variation and the second axis explains 15.6 % (Fig.4.7b). T K N , pH, grass, and channel width discriminated the most along the first axis. Conductivity, conifers, and shrubs discriminated the most along the second axis. Prediction of Seasonal Test Sites to Groups The reference groups to which the seasonal test sites were predicted to are summarized in Table 4.2. Mellin and Beak Creek (interior streams) and Spring Creek (coastal stream) were predicted to group 2. The third interior stream (Glimpse Creek) was predicted to group 1 along with Mayfly Creek (coastal stream). Glimpse had a high conductivity, which placed it in group 1. The North Alouette river was 93 CN ,<D-2 C M 3 ° « a s • o 0 •4 o 3 O CJ r o s ° • O CN (% 9'Sl) 2 uoipunj lUBijTTiiuosiQ CN o >-o a cd 4 CD & § S CJ (_; cu 3 -a • i-H 4 CO a 3 cd L H o •4 3 O • i—i - M cd > ~ A3 W 3 00 o L . M C M '3 o u - 4 -CN (%S'C,\)Z uoipimj juRuraiuosiQ oo a o ca CO 00 o CD o co 3 AH O cd 3 5 cS cj 3 - M cd C O SJ 3 •£ o g § ' S W C O •s * A-> - M 8 « C O O CJ CJ l i CJ O •» -5 CJ g o 9 u H cj CJ ^ Jo 8 ™ | CJ CJ 5 8 S 3 CD O -3 8 2 C M CD M A3 o cd c2 CD ^ -2 cn 3 C O cj O o co '\ .2 § .S s ^ > <B T3 CJ cd e 3 C O W •3 so 3 CJ 6 3 O L H ' > 3 CJ <D C M O ^ *3 I) »- A^ -3 S '2 H 2P S 2 Figure 4.8: Location of seasonal test sites in the ordination space of the group to which the sites were predicted. The sites were predicted to their respective groups based on the autumn (3) sampling of the test sites. The invertebrates were identified to the family level, 0.5 % cut-off. A - group 1 ordination space, B - group 2 ordination space, and C - group 3 ordination space. The indicated 90, 99, and 99.9 % probability ellipses are based on the ordination scores of the reference data for the respective groups. OBeakCk, O Glimpse Ck, V Mellin Ck, * Mayfly Ck, A N. Alouette R., •fr Spring Ck, < Fraser R., * Thompson R. 96 predicted to group 3 along with the Thompson and Fraser test sites. Prediction of the seasonal test sites into different groups than in the prior analysis reflects the different groupings of the reference site data. The seasonal test site invertebrate data was ordinated with reference site data (Fig. 4.8, Table 4.2). Mellin Creek had more sampling dates inside the 90 % probability ellipse at the family level than at the genus/species level. Beak and Mayfly did not have sampling dates that fell outside of the 90 % probability ellipse. There is no difference in the results at the different levels of analysis for the Fraser and Thompson rivers. Discussion Seasonal Change at the Test Sites Stream benthic invertebrate community composition changes seasonally (Hynes 1970; Furse et al. 1984; Death 1995), although these seasonal changes are not large (Hawkins and Sedell 1981; Death 1995; Zamora-Munoz and Alba-Tercedor 1996). My study also found that the invertebrate communities at the seasonal test sites changed between sampling dates, depending on the site and the season. Seasonal changes in community composition occur as invertebrates move through their life cycles (Chapter 2; Butler 1984), and because of short-term exclusion of taxa as a result of redistribution of invertebrates with disturbance events such as floods (Reice et al. 1990). Short-term exclusions may also result from sampling, identification, and data entry errors (Furse et al. 1984). Each test site community changed seasonally in the ordination space in different directions and with different levels of magnitude. The seasonal change resulted in some test sites falling outside of the reference site 90 % probability ellipses and even 99.9 % probability ellipses for some seasons, resulting in a site falsely being classified as impacted. Further investigation is required to determine whether differences between the reference community and the test site community were a result of anthropogenic impacts. Glimpse Creek, Spring Creek, Mayfly Creek, and the North Alouette River are all streams that have no apparent impacts. Yet, each of these sites had sampling dates that fell outside their respective 90 97 % ellipses. Movement of the seasonal test samples in and out of the 90 % probability ellipses was a result of seasonal changes in the invertebrate community. There was no consistency as to which sampling dates fell outside of the ellipses. There was also no consistency as to how much the sites varied seasonally and in what direction, so there is no way of knowing what influence seasonal change will have on future test sites. If a future test site sample was collected outside of the season when the reference samples were collected and the test sample differed from the reference condition, there would be no way of knowing how much of the difference was a result of the time of year the sample was collected. As a result, a single test sample collected at a time of year other than when the reference samples were collected is not recommended. Mellin Creek and the Fraser River were the two rivers in which seasonal change to the community resulted in sampling dates falling outside of the 99 and 99.9 % probability ellipses. For Mellin Creek, the autumn sampling and other seasons fell outside of the 99% ellipse, probably because of a combination of actual stresses and seasonal change. Cattle were using the stream as a source of water, they were moving through the stream, and defecating in it at the time of the autumn sampling,. The model apparently detected this impact. The model was more sensitive to the impact of the cattle when genus/species level data were used, and placed the autumn sampling further from the reference condition, than when family level data were used. The model also detected the effect of seasonal change at Mellin Creek. The spring 1995 and winter sampling dates consistently fell outside of the 99 % probability ellipse. The discharge was above bankfull in Mellin Creek when the spring 1995 samples were collected. This high discharge would have scoured the stream channel and redistributed many of the invertebrates (Reice et al. 1990). The low abundance of all taxa caused the spring 1995 sampling to fall outside of the 90 and 99.9 % probability ellipses. When the winter samples were collected, Mellin Creek had a very low discharge, ice covered the channel, and much of the ice at the sampling site was anchored to the substrate. Invertebrates may have avoided the sampling riffle because of the anchor ice, or they may have been dislodged during the ice removal prior to sampling. Invertebrates were not abundant and the site fell outside of the 99 % 98 probability ellipses. Collection of test samples after unusually high flows, spate, or winter conditions is not recommended (see also Wright 1995). Summer and spring 1996 samples from Mellin Creek were placed within the 90 and 99 % ellipses. The invertebrate community apparently recovered from the disturbances (Reice et al. 1990), and this recovery between sampling periods has implications. First, a site that suffered seasonal impacts could be classified as unstressed depending on when the test samples were collected. Conversely, a healthy stream could be classified as impacted if the test samples were collected immediately following a natural disturbance. Therefore I recommend that multiple sampling dates be used (Furse et al. 1984). Second, the model seems to be able to detect recovery of benthic communities between sampling dates, indicating that it should be useful in detecting recovery of impacted sites after remediation. Each large river is unique in its physical and biological structure (Dodge 1989), so there are no other large rivers that can be used as reference sites. The Fraser and the Thompson Rivers are both large rivers, with drainage basin areas (at the sampling site) of 217,000 and 54,900 km 2 and average annual discharges of 2100 and 781 m3/s (Environment Canada, Water Survey Branch, pers. comm.). These two large rivers suffer from anthropogenic impacts such as industrial and agricultural use, sewage, and water extraction. There are few large rivers of this size, if any, that do not suffer from human impacts. Some reference condition is needed with which to compare current and future biological and environmental conditions of the Fraser and Thompson rivers with. The FRB reference data set does not include reference sites from the Thompson River or from the lower mainstem and impacted reaches of the Fraser River. However, other reference sites from the FRB can potentially be used as reference sites (e.g. sites on the upper Fraser mainstem and other large tributaries draining into the Fraser). However, only a few large sites (5th - 9 th order) exist, and they are still smaller than the large river test sites. Thus, comparing the Fraser and Thompson seasonal test samples to the reference data base involves extrapolation from smaller rivers, and caution must be used when concluding how large river test sites compare to the reference sites. It is important to determine first if FRB sites can be used as large river reference sites and then whether the test sites have more significant seasonal changes than the reference sites. 99 Both the Fraser and Thompson River seasonal test sites fell within the range of reference conditions, indicating that reference sites included in the model are somewhat representative of conditions found at the large river seasonal test sites. Because Fraser and Thompson River sampling dates fell outside the 90 % ellipse, the model may be detecting the difference between reference sites and the large river test sites, and possibly impacts. Influence of Data Censorship and Taxonomic Levels The level of data censorship and taxonomic level used influenced the grouping of the reference sites, the sensitivity of the analysis, and the results. The placement of reference sites into groups varied with each censorship level and taxonomic level. The environmental variables that best discriminated between the groups also changed with the change of reference sites included in each group. These changes resulted in slightly different models and predictions made. No one model is necessarily more correct than the others but it is critical to determine how various decisions about analysis affected the model and the predictions made. In general, the cluster analysis grouped the reference site invertebrate data by biogeographical regions. Group 1 was always composed of interior streams, group 3 was composed of coastal streams and some Fraser River sites, and the sites in group 2 varied between interior streams to upper Fraser and Fraser mainstem sites located in mountain regions. The separation of the invertebrate communities into three different ecoregions; interior, mountain, and coastal is similar to what has been described by Corkum (1989) and Tate and Heiny (1995). These studies found distinct differences in the benthic communities in mountain and plain regions of northwestern North America and in the South Platte River Basin. Most invertebrate taxa increased in abundance in the interior streams. However, there were a few exceptions. For the genus/species level data, (0.1 % cut-off) Taenionema spp. (Taeniopterygidae) and Drunella doddsi (Ephemerellidae) increased in abundance in the coastal, upper Fraser and Fraser mainstem sites. These sites generally have large substrates, steep gradients, low levels of alkalinity, 100 coniferous riparian vegetation, and some have large channel widths. Similar environmental conditions have been described or the occurrence of Taenionema sp. and D. doddsi (Ward 1986; Mangum and Winget 1991). Data analysis at genus/species level (0.5 % cut-off) found Hydropsyche spp. (Hydropsychidae) and Taenionema sp. to increase in abundance in coastal, upper Fraser, and Fraser mainstem, sites. Hydropsyche spp. distribution overlaps with Taenionema spp. and extends into large rivers with warm temperatures (Ward 1986). This same pattern was seen when the data were analyzed at the family level (0.5 % cut-off). The level of censorship used had a greater effect on the genus/species level than on the family level. The censorship level affected the predictive model and site misclassification rates. At the genus/species level, use of the 0.5% cut-off reduced the predictive model error rate by 7.7 % over analysis with the 0.1 % cut-off level. The higher cut-off level reduced the amount of variability in the reference data set allowing stronger groups to be formed in the cluster analysis. Better discriminator environmental variables were selected and used for prediction with greater accuracy with clearer grouping of the reference communities. At the family level, use of the 0.5 % cut-off improved the error rate by only 4.5%, but some of the variation in the data set was eliminated. Thus, the higher cut-off level had less of an effect on misclassification rates. The level of censorship affected which variables were used as predictor variables, and as a result, the groups to which the seasonal test sites were predicted and their location within the reference probability ellipses. For example, Spring Creek and Mayfly Creek were predicted into different groups with the different levels of analysis. For the genus/species level data (0.1 % cut-off), most of the seasonal test samples for Spring and Mayfly Creek fell outside of the 90 % ellipses. At the 0.5 % cut-off level fewer seasonal samples fell outside of the 90 % probability ellipse (Table 4.2), and the predictive model included environmental variables such as channel width, bankfull width and size of the dominant substrate (framework). Thus, Spring Creek and Mayfly Creek were placed into more appropriate groups and fewer sampling dates fell outside of the ellipses for these unstressed sites. The higher censorship level resulted in a better predictive model and more appropriate placement of test sites into groups. 101 The sensitivity of the predictive model was affected by the censorship level when genus/species data were used. For example, two of the Mellin Creek sampling dates that fell outside of the ellipses at the 0.1 % level fell inside the 90 % ellipse at the 0.5 % level. Less variation was expressed in the data set, at the 0.5 % censorship level, which resulted in a less sensitive analysis. A more sensitive analysis would be better at detecting stress, but will also detect more natural variation, which could lead to erroneous conclusions about a test site. In general, the 0.5% level of censorship detected the stressed sites and seasons and only a few unstressed seasons fall outside the 90 % ellipses. The number of taxa included with each taxonomic level of analysis resulted in the genus/species data containing more information than the family data. The grouping of the reference invertebrate communities and the variables used to best discriminate between the groups also changed with the taxonomic level used. However, there was very little difference in the predictive model error rates between the genus/species and family models for a given censorship level. As a result, there is no advantage in using one taxonomic level over the other based on model error rates. Genus/species level data were more sensitive to natural and seasonal variations than the family level data. The genus/species data detected minor seasonal changes in the invertebrate community because of invertebrate life-cycles and environmental conditions. The family level data, to a certain extent, masked seasonal patterns (Corkum 1990). Species in a family can have different life cycles, e.g., different hatching, growth and/or emergence periods (Sweeney and Vannote 1981), so species-level data will reflect seasonal changes caused by life cycle changes, whereas family level data will not. The genus/species level data were also more sensitive to perturbations in the streams than the family data. For example, the autumn sampling dates for Mellin Creek fells outside the 99 % ellipses for the genus/species data and inside the 90 % ellipses for the family data. Mellin Creek would be considered impaired based on genus/species data, but not based on family data. The impact of the cattle on the system must have affected some species and/or genera but not all individuals representing the families. With different species tolerances, the genus/species data were more sensitive to this stress than the family data (Wright 1995; Zamora-Munoz and Alba-Tercedor 1996). The family level data did however 102 detect that Mellin Creek was impacted in the spring 1995 and winter 1995 sampling. So, the family data did detect major seasonal events, such as flooding and anchor ice, which altered the invertebrate community. Implications for Future Users of the Predictive Model The invertebrate communities at the seasonal test sites changed seasonally, and these changes were enough to place some sampling dates outside of the 90 % probability ellipses. Such placement does not necessarily mean an unimpacted test site will be classified as an impacted site, but it creates uncertainty. The uncertainty comes from the inability to separate typical seasonal changes from anthropogenic changes. It is an inability to predict the magnitude and direction of seasonal change in a stream. The seasonal changes of the seasonal test site communities were significant enough to affect how the predictive model should be used in the FRB. Therefore, future test site samples should be collected in the same season as the reference samples were collected (autumn), or over multiple sampling dates to minimize the chance of missclassifying a stream. The collection of test samples in the same season as the reference samples should account for differences caused by invertebrate life cycles. The collection of test samples over two or more sampling seasons (e.g., spring, summer, fall), should account for taxa excluded because of stochastic events or sampling errors. Sampling over multiple dates will also help to determine whether the invertebrate community can recover from disturbance events. The usefulness of test samples collected over two or more sampling dates may outweigh the time and effort required to collect and process the samples. Caution and common sense should be used when relating single test samples collected outside of the season when reference samples were collected. No unusual disturbance events should precede sampling and since seasonal variation may cause unimpacted sites to fall outside of the ellipses. Further investigation of the sites may be required. 103 The appropriate level of data censorship for future users of the FRB predictive model will depend on the desired sensitivity of the model and the taxonomic level used. The 0.1 % censorship level data set included more taxa than the 0.5% level and as a result it also included more information and variation. Therefore, the model created with the 0.1 % level of data was more sensitive to changes in the invertebrate community caused by impacts and/or seasonal variation. This lower cut-off level should be used if a more sensitive analysis is desired. Higher censorship levels (0.5 % cut-off) should be used if models are desired which have robust relationships between the invertebrate data and environmental predictor variables. Use of high censorship levels will result in a lower site misclassification rate. The level of censorship had a greater influence on the sensitivity of analysis at the genus/species level than on analysis at the family level because the genus/species data set contained many rare taxa prior to censorship, whereas several of the rare invertebrate taxa were grouped into higher taxonomic levels in the family data set. The most appropriate taxonomic level for future users of the FRB predictive model will depend on the level of expertise of the biologists involved, on the sensitivity of analysis required, and whether seasonal samples were collected. It would be more appropriate to use higher taxonomic levels than to incorrectly identify test site invertebrate data. Identification to the family level may also be more appropriate when samples contain many early instar larvae, which are difficult, if not impossible, to identify. Family level identifications are typically consistent between labs (Cranston 1990). Misidentification of taxa may lead to poor representation of the test site community. As a result, a site may be classified as impacted when it is not. Analysis with the family level data set was not as sensitive to seasonal changes to the invertebrate community as with the genus/species level data. Because the genus/species level data were more sensitive to slight changes in the community, species level data should be used for biomonitoring if a sensitive model is required. However, the sensitivity of the genus/species level data to seasonal changes in the community may just create noise and obscure patterns useful for biomonitoring. Analysis at the family level filtered out the noise associated with minor seasonal changes and stochastic events. As a 104 result, if seasonal or stochastic events are suspected, and it is desired to filter them out, then family level data may be more appropriate. 105 Chapter Five C O M M E N T S A N D C O N C L U S I O N S Summary This study has shown that benthic invertebrate communities of southwestern British Columbia streams change both seasonally and spatially. Seasonal changes of the fauna were related to timing of invertebrate life cycles, which were influenced by local environmental conditions, in particular, temperature and food resources. Based on the data reported here, there was no evidence to support the hypothesis that the life cycles oi Drunella doddsi, D. spinifera, and Zapada cinctipes were influenced by discharge regimes; timing of life cycles may be constrained by other environmental conditions. As a result, these invertebrates may use strategies other than the timing of emergence to deal with seasonal increases in discharge, or discharge regime may be an inappropriate measure of disturbance for life cycle adaptations. This study did not rule out the possibility that other taxa, life history tactics, or genetic variability may be influenced by disturbance events. Seasonal variation of faunal composition and abundance was less than spatial variation. Spatial variation was related to environmental variables reported in previous studies, including channel width, mean depth, maximum depth, maximum velocity, discharge, conductivity, alkalinity, nitrite and nitrate nitrogen, and total Kjeldahl nitrogen. Seasonal changes of the benthic community were related to changes in the environment in part through their effects on invertebrate life cycles. There were no direct relationships between changes of invertebrate composition and the rate of changes of environmental conditions. In the small streams, the pattern of invertebrate abundance was similar to the seasonal pattern proposed by Hynes (1970) with an increase in abundance through the summer because of hatching and a decrease through the winter and early spring because of deaths and emergence. The seasonal pattern of invertebrate abundance in large rivers was related to the discharge regime and so differed from the 106 pattern in small streams. A decrease in abundance with an increase in discharge may be related to life cycles or to invertebrate use of refugia to avoid the seasonal scour. Although seasonal variation was small relative to spatial variation, it was still sufficient to warrant caution when interpreting results from test site samples collected other than in autumn. The magnitude and direction of seasonal change varied with each stream and season and could not be predicted from the present study. I recommend that test samples be collected over multiple sampling dates or restricted to autumn to avoid stochastic events and seasonal shifts. Appropriate taxonomic and censorship levels used for biomonitoring will depend on the accuracy and sensitivity desired. Low taxonomic and censorship levels contained more information and models were more sensitive to variation than when high taxonomic and censorship levels were used. Higher taxonomic and censorship levels reduced variability and resulted in lower misclassification rates. Recommendations 1) To thoroughly establish insect life cycles, samples should be collected at least once per month and as frequently as every two weeks. Sampling over multiple years would also be desirable since year-to- year variation would be accounted for and clearer life cycle and abundance patterns could be established. Sample collection from only five sampling dates and the lack of emergence traps limited my ability to accurately determine timing of insect life cycles. The sampling regime used in this study provided a coarse look at invertebrate life cycles for comparison between stream classes. Future studies might examine life history tactics in relation to measures of disturbance such as measures of bed movement which may be more ecologically meaningful to the invertebrates. Studies may also look at the composition of life cycles in communities with different disturbance regimes. In stable streams, a large proportion of the community may have synchronous development, whereas with unpredictable regimes, a large proportion of the community may have asynchronous or delayed development. 107 2) Biomass and production estimates as well as abundance measures would provide a more complete understanding of the seasonal changes occurring within each stream, differences between the streams, and associations between the community and environmental conditions. Within each small stream, biomass should follow the same pattern as abundance but with a slight time delay (Hynes 1970). Biomass should increase through the summer and autumn as insects that hatched in the spring / summer grow. Biomass should decrease through the spring with emergence and hatching of small individuals. In large rivers biomass should increase with time since peak freshet, and as abundance and size of invertebrates increase. Secondary production should change seasonally and be at its maximum at times of the year when food resources are also at their maximum. The differences in abundance between the stream classes may be explained by biomass and production. Abundance may be under or overestimating the roles of organisms in stream ecosystems (Lugthart and Wallace 1992). 3) Use of discharge regime without a measure of disturbance was a limitation of this study. Although it has been argued that high discharge levels and possibly seasonal bankfull floods constitute disturbances, Townsend et al. (1997) have convincingly demonstrated that high levels of discharge do not represent a disturbance event at a level meaningful to benthic invertebrates. Measures of stream bed ^ movement would have been a better measure. Understanding whether the seasonal increase in discharge is significant for invertebrates would help understand why timing of life cycles was not related to discharge in a simple way and why abundance patterns differed between the large rivers and small streams. Future studies might look at life history tactics (e.g. the number of weedy taxa with univoltine or bivoltine life cycles versus the number of taxa with specific habitat requirements and life cycles longer than one year [number of r- versus K-selected taxa]) in relation to different discharge regimes. Studies may also look at the composition of life cycles in communities with different discharge regimes. A large proportion of the community may have synchronous development in stable streams or in streams with predictable discharge regimes, whereas, a large proportion of the community may have asynchronous or delayed development in streams with unpredictable regimes. 108 4) Further study into invertebrate behaviour and life cycles in large rivers is required. Few studies have looked at invertebrate characteristics in large rivers (Dodge 1989). It is currently unclear how the increase in abundance with time-since-peak-discharge is related to life cycles and invertebrate movement out of refugia. Bothwell and Culp (1993) suggested the decrease in abundance prior to freshet was primarily a result of emergence, and that abundance increased with lower, stable flows. Lancaster and Hildrew (1993) proposed the hyporheos as a refugium in small streams and Giberson and Hall (1988) found early instars to use the hyporheos as a nursery. Rempel (1997) found invertebrates to use lateral margins of large rivers as refugia; the importance of backwaters and side channels as refugia has not been fully explored. 5) Further work should be done on the relationship between seasonal change of the benthic community and biomonitoring. Other studies on the seasonal change of the fauna may identify predictable patterns. Future work may also consider the magnitude of seasonal change in unimpacted streams versus impacted streams. 109 Literature Cited Allen, R.K. and G.F. Edmunds. 1962. A revision of the genus Ephemerella (Ephemeroptera, Ephemerellidae) V. The subgenus Drunella in North America. Entomological Society of America, Special publications 3:147 - 179. Anderson, N.H., and K.W. Cummins. 1979. The influence of diet on the life histories of aquatic insects. Journal of the Fisheries Research Board of Canada 36:335-342. Baldy, V. , M.O. Gessner and E. Chauvet. 1995. Bacteria, fungi and the breakdown of leaf litter in a large river. Oikos 74:93-102. Beals E.W. 1984. Bray-Curtis ordination: An effective strategy for analysis of multivariate ecological data. Advances in Ecological Research 14:1-55. Begon, M . , J.L. Harper, and C.R. Townsend. 1990. Ecology individuals, populations and communities. 2n d edition. Blackwell Scientific Publications Cambridge, M A . Belbin, L . 1991. Semi-strong hybrid scaling, a new ordination algorithm. Journal of Vegetation Science 2:491-496. Belbin, L . 1993. PATN, pattern analysis package. Division of Wildlife and Ecology, CSIRO, Australia. Boulton, A.J. , and P.S. Lake. 1992. The ecology of two intermittent streams in Victoria, Australia. III. Temporal changes in faunal composition. Freshwater Biology 27:123-138. Bothwell, M . L . and J.M. Culp. 1993. Sensitivity of the Thompson River to phosphorus: studies in trophic dynamics. National Hydrological Research Institute contribution No. 93006, 45p. Butler, M . G . 1984. Life histories of aquatic insects. In: D .M. Rosenberg and V . H . Resh (editors). Freshwater biomonitoring and benthic macroinvertebrates. Chapman and Hall, New York. Cairns, J.Jr., and J.R. Pratt. 1993. A history of biological monitoring using benthic macroinvertebrates. In: D . M . Rosenberg and V . H . Resh (editors). Freshwater biomonitoring and benthic macroinvertebrates. Chapman and Hall, New York. Cancilla, D. 1994. Manual of analytical methods. Volume 1. National Laboratory for Environmental Testing, Canada Centre for Inland Waters, Environment Canada, Burlington, Ont. (unpaginated). Cao, Y , A.W. Bark, and W.P. Williams. 1997. A comparison of clustering methods for river benthic community analysis. Hydrobiologia 347:25-40. Chance, M . M . , and D.A. Craig. 1986. Hydrodynamics and behaviour of Simuliidae larvae (Diptera). Canadian Journal of Zoology 64:1295-1309. 110 Chapman, D.W. and R.L. Demory. 1963. Seasonal changes in the food ingested by aquatic insect larvae and nymphs in two Oregon streams. Ecology 44:140-146. Chesson, P.L. 1986. Environmental variation and the coexistence of species. Pages 240-256 in J.M. Diamond and T.J. Case (editors). Community ecology. Harper and Row, New York. Ciborowski, J.J.H., and D.A. Craig. 1989. Factors influencing dispersion of larval black flies (Diptera: Simuliidae): effects of current velocity and food concentration. Canadian Journal of Fisheries and Aquatic Science 46:1329-1341. Closs, G.P., and P.S. Lake. 1994. Spatial and temporal variation in the structure of an intermittent-stream food web. Ecological Monographs 64:1-21. Cobb, D.G. , T.D. Galloway, and J.F. Flannagan. 1992. Effects of discharge and substrate stability on density and species composition of stream insects. Canadian Journal of Fisheries and Aquatic Science 49:1788-1795. Corkum, L.D. 1978. The nymphal development of Paraleptophlebia adoptiva (McDunnough) and Paraleptophlebia mollis (Eaton) (Ephemeroptera: Leptophlebiidae) and the possible influence of temperature. Canadian Journal of Zoology 56:1842-1846. Corkum, L.D. , 1989. Patterns of benthic invertebrate assemblages in rivers of northwestern North America. Freshwater Biology 21:195-205 Corkum, L.D. 1990. Intrabiome distributional patterns of lotic macroinvertebrate assemblages. Canadian Journal of Fisheries and Aquatic Science 47:2147-2157. Corkum, L.D. , and D.C. Currie, 1987. Distributional patterns of immature Simuliidae (Diptera) in northwestern North America. Freshwater Biology 17:201-221. Cranston, P.S. 1990. Biomonitoring and invertebrate taxonomy. Environmental Monitoring and Assessment 14:265-273. Cummins, K.W. 1988. The study of stream ecosystems: a functional view. In: Pomeroy, L.R. and Alberts, J.J. (eds), Concepts of ecosystem ecology. Springer, New York, Cummins, K.W., M.A. Wilzbach, D .M. Gates, J.B. Perry, and W.B. Taliaferro. 1989. Shredders and riparian vegetation. Bioscience 39:24-30. Death, R.G. 1995. Spatial patterns in benthic invertebrate community structure: products of habitat stability or are they habitat specific. Freshwater Biology 33:455-467. Diamond, J.M. 1982. Stream geomorphology and benthic habitat predictability as determinants of the population dynamics and life history of the snail Juga plicifera. Journal of Freshwater Ecology 1:577-588. Dodge, D.P. 1989. Proceedings of the international large river symposium (LARS), 1986. Canadian Special Publication of Fisheries and Aquatic Sciences 106. Department of Fisheries and Oceans, Ottawa. I l l Doledec, S. 1989. Seasonal dynamics of benthic macroinvertebrate communities in the lower Ardeche River (France). Hydrobiologia 182:73-89. Dole-Olivier, M.J. , P. Marmonier, and J.L. Beffy. 1997. Response of invertebrates to lotic disturbance: is the hyporheic zone a patchy refugium? Freshwater Biology 37:257-276. Duncan, G.A. and G.G. LaHaie. 1979. Size analysis procedures used in the Sedimentology Laboratory. Manual, Hydraulics Division, National Water Research Institute, Burlington, Ontario. Faith, D.P., P.R. Minchin, and L. Belbin. 1987. Compositional dissimilarity as a robust measure of ecological distance. Vegetatio 69:57-68. Fausch, K.D. , J.Lyons, J.R. Karr, and P.L. Angermeier. 1990. Fish communities as indicators of environmental degradation. American Fisheries Society Symposium 8:123-144. Fuller, R.L. , and K.W. Stewart. 1977. The food habits of stoneflies (Plecoptera) in the Upper Gunnison River, Colorado. Environmental Entomology 6:293-302. Furse, M . T . , D. Moss, J.F. Wright, and P.D. Armitage. 1984. The influence of seasonal and taxonomic factors on the ordination and classification of running-water sites in Great Britain and on the prediction of their macro-invertebrate communities. Freshwater Biology 14:257-280. Gauch, H.G. 1982. Multivariate analysis in community ecology. Cambridge University Press. New York. Geological Survey of Canada , 1969. Geological Atlas, sheet 92. Fraser River, British Columbia - Washington, map no. 1386A. Survey and Mapping Branch. Geological Survey of Canada , 1957. Victoria - Vancouver, British Columbia, map no. 1069A. Department of mines and technical surveys, British Columbia. Giberson, D.J. and R.J. Hall. 1988. Seasonal variation in faunal distribution within the sediments of a Canadian shield stream, with emphasis on responses to spring floods. Canadian Journal of Fisheries and Aquatic Science, 45:1994-2002. Gilbert, E.S. 1968. On discrimination using qualitative variables. Journal of the American Statistical Association 63:1399-1412. Giller, P.S. and H. Twomey. 1993. Benthic macroinvertebrate community organization in two contrasting rivers - between-site differences and seasonal patterns. Biology and environment: Proceedings of the Royal Irish Academy Vol. 93B: 115-126. Gray, L.J. 1981. Species composition and life histories of aquatic insects in a lowland Sonoran Desert stream. The American Midland Naturalist 106(2):229-242. 112 Greenberg, A . E . , L.S. Clesceri, and A.D. Eaton. 1992. Standard methods for the examination of water and waste water. 18th edition. American Public Health Association, American Water Works Association and Water Environment Federation. Grime, J.P. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. American Naturalist 111:1169-1194. Harrel, R.C. and T.C. Dorris. (1968). Stream order, morphometry, physico-chemical conditions, and community structure of benthic macroinvertebrates in an intermittent stream system. American Midland Naturalist 80:220-251. Hawkins, C P . 1984. Food habits of species of ephemerellid mayflies (Ephemeroptera: Insecta) in streams of Oregon. The American Midland Naturalist 113:343-353. Hawkins, C P . 1986. Variation in individual growth rates and population densities of ephemerellid mayflies. Ecology 76:1384-1395. Hawkins, C P . , and J.R. Sedell. 1981. Longitudinal and seasonal changes in functional organization of macroinvertebrate communities in four Oregon streams. Ecology, 62:387-397. Hellawell, J.M. 1986. Biological indicators of pollution and environmental management. Elsevier, London. Hopcroft, R. 1995 . Zoobenthos Biomass Digitizing Program. Department of Zoology, University of Guelph, Ontario. Hynes, H.B.N. 1970. The ecology of running waters. University of Toronto Press. Ontario, Canada. Jackson, B.B. 1983. Multivariate data analysis, an introduction. Richard D. Irwin Inc, Illinois. Jackson, D.A. 1993. Multivariate analysis of benthic invertebrate communities: the implications of choosing particular data standardizations, measures of association, and ordination methods. Hydrobiologia 268:9-26. Jackson and Harvey. 1993. Canadian Journal of Fisheries and aquatic science, 50:2641-2651. Jeffries, M.J. and J.H. Lawton. 1985. Predator-prey ratios in community of freshwater invertebrates: the role of enemy free space. Freshwater Biology 15:105-112. Krebs, C J . 1989. Ecological methodology. HarperCollinsPublishers, Inc. New York. Lancaster, J. and A. G. Hildrew. 1993a. Characterizing in-stream flow refugia. Canadian Journal of Fisheries and Aquatic Science 50:1663-1675. Lancaster, J. and A. G. Hildrew. 1993b. Flow refugia and the microdistribution of lotic macroinvertebrates. Journal of the North American Benthological Society 12:385-393. 113 Legendre, L . and P. Legendre. 1983. Numerical Ecology. Elsevier Scientific Publishing, Amsterdam. Leopold, L .B. 1962. Rivers. American Scientist 50:511-537. Lugthart, G.J. and Wallace, J.B. 1992. Effects of disturbance on benthic functional structure and production in mountain streams. Journal of the North American Benthological Society 11:138-164. Mangum, F.A., and R.N. Winget. 1991. Environmental profile oi Drunella (Eatonella) doddsi (Needham) (Ephemeroptera:Ephemerellidae). Journal of Freshwater Ecology 6:11-22. Marchant, R. 1989. A subsampler for samples of benthic invertebrates. Bull Australian. Soc. Limnology 12:49-52. Marchant, R. 1990. Robustness of classification and ordination techniques applied to macroinvertebrate communities from the La Trobe River, Victoria. Australian Journal of Marine and Freshwater Research 41:493-504. Martinson, R.J. and J.V. Ward. 1982. Life history and ecology of Hesperophylax occidentalis (Banks) (Trichoptera: Limnephilidae) from three springs in the Piceance Bason, Colorado. Freshwater Invertebrate Biology 1:41-47. Matthews, G.B., R.A. Matthews, and B. Hachmoller. 1991. Mathematical analysis of temporal and spatial trends in the benthic macroinvertebrate communities of a small stream. Canadian Journal of Fisheries and Aquatic Science 48:2184-2190. Merrit, R.K. and K.W. Cummins. 1996. An introduction to the aquatic insects of North America. 3rd edition. Kendall/Hunt Publishing Company, Iowa. McElravy, E.P., G.A. Lamberti, and V . H . Resh. 1989. Year-to-year variation in the aquatic macroinvertebrate fauna of a northern California stream. Journal of the North American Benthological Society 8:51-63. Minshall, G.W., K.W. Cummins, R.C. Perersen, C E . Cushing, D.A. Bruns, J.R. Sedell, and R.L. Vannote. 1985. Developments in stream ecosystem theory. Canadian Journal of Fisheries and Aquatic Science 42:1045-1055. Munkittrick, K.R. and D.G. Dixon, 1989. Use of White Sucker (Catostomus commersoni) populations to assess the health of aquatic ecosystems exposed to low-level contaminant stress. Canadian Journal of Fisheries and Aquatic Science 46:1455-1462. Newbury, R.W. 1984. Hydrologic determinants of aquatic insect habitats, p.323-357. In: V . H . Resh and D . M . Rosenberg (eds). The ecology of aquatic insects. Praeger, New York. Newbury, R.W., and M.N. Gabory, 1993. Stream analysis and fish habitat design: A field manual. Newbury hydraulics Ltd. Gibsons, BC. 256pgs. 114 Norris, R.H. 1994. Rapid biological assessment, natural variability, and selecting reference sites. In: Many Uys (editor) Proceedings of Joint South African Australian workshop: Classification of rivers and environmental health indicators pp 129 - 166. Norris, R.H. 1995. Biological monitoring: the dilemma of data analysis. Journal of the North American Benthological Society 14:440-450. Norusis, M.J. 1985. SPSSX Advanced Statistics Guide. McGraw-Hill Book Co., New York. Ormerod, S.J., and R.W. Edwards. 1987. The ordination and classification of macroinvertebrate assemblages in the catchment of the River Wye in relation to environmental factors. Freshwater Biology 17:533-546. Parsons, M . and R.H. Norris. 1996. The effect of habitat-specific sampling on biological assessment of water quality using a predictive model. Freshwater Biology36:419-434. Pennak, R.W. 1978. Fresh-water invertebrates of the United States, 2n d edn. John Wiley and Sons, New York. Poff, N.L. 1992. Why disturbances can be predictable: a perspective on the definition of disturbance in streams. Journal of the North American Benthological Society 11:86-92. Poff, N.L. , and J.V. Ward. 1989. Implications of streamflow variability and predictability for lotic community structure: a regional analysis of streamflow patterns. Canadian Journal of Fisheries and Aquatic Sciences 46:1805-1818. Ractliffe, S.G., B.R. Davies, B.A. Stewart and C D . Snaddon. 1995. The influence of discharge on entrainment of bank litter in a headwater stream. Archive fur Hydrobiology 134:103-117. Radford, D.S. and R. Hartland-Rowe. 1971. The life cycles of some stream insects (Ephemeroptera, Plecoptera) in Alberta. The Canadian Entomologist 103:609-617. Reice, S.R., and M . Wohlenberg. 1993. Monitoring freshwater benthic macroinvertebrates and benthic processes: Measures for assessment of ecosystem health. In: Rosenberg, D . M . and V . H . Resh (eds) Freshwater biomonitoring and benthic macroinvertebrates. Chapman and Hall, New York. Reice, S.R., R.C. Wissmar, and R.J. Naiman. 1990. Disturbance regimes, resilience, and recovery of animal communities and habitats in lotic ecosystems. Environmental Management 14:647-659. Rempel, L . L . 1997. Habitat variation due to seasonal flooding in the lower Fraser River and the influence on the macroinvertebrate community. MSc Thesis, Department of Zoology, University of British Columbia, Canada. Resh, V . H . , and E.P. McElravy. 1993 Contemporary quantitative approaches to biomonitoring using benthic macroinvertebrates. In: D . M . Rosenberg and V . H . Resh (editors). Freshwater biomonitoring and benthic macroinvertebrates. Chapman and Hall, New York. 115 Resh, V . H . , and J.D. Unzicker. 1975. Water quality monitoring and aquatic organisms: the importance of species identification. Journal of Water Pollution Control Fed. 47:9-19 Resh, V . H . , A . V . Brown, A.P. Covich, M.E . Gurtz, H.W. LI, G.W. Minshall, S.R. Reice, A . L . Sheldon, J.B. Wallace and R.C. Wissmar. 1988. The role of disturbance in stream ecology. Journal of the North American Benthological Society 7:433-455. Reynoldson, T.B., 1987. Interactions between sediment contaminants and benthic organisms. Hydrobiologia 149:53-66. Reynoldson, T.B., R.C. Bailey, K . E . Day and R.H. Norris. 1995 . Biological guidelines for freshwater sediment based on benthic assessment of sediment (the BEAST) using a multivariate approach for predicting biological state. Australian Journal of Ecology 20:198-219. Reynoldson, T.B., H.H. Norris, V . H . Resh, K . E . Day, and D.M. Rosenberg. 1997. The reference condition: a compariaon of multimertic and multivariate approaches to assess water-quality impairment using benthic macroinvertebrates. Journal of the North American Benthological Society 16(4):833-852. Rice, W.R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225. Richards, C , G.E. Host, and J.W. Arthur. 1993. Identification of predominant environmental factors structuring stream macroinvertebrate communities within a large agricultural catchment. Freshwater Biology 29:285-294. Richardson, J.S. 1989. Seasonal food limitation of detritivorous insects in a montane stream. PhD Thesis, Department of Zoology, University of British Columbia, Canada. Richardson, J.S. 1991. Seasonal food limitation of detritivores in a montane stream: an experimental test. Ecology 72:873-887. Richardson, J.S. 1992. Coarse particulate detritus dynamics in small, montane streams of southwestern British Columbia. Canadian Journal of Fisheries and Aquatic Science 49(2):337-346. Richardson, J.S. and C D . Levings. 1996. Chlorinated organic contaminants in benthic organisms of the lower Fraser River, British Columbia. Water Quality Research Journal of Canada 31:153-162. Robinson, C.T. L . M . Reed, and G.W. Minshall. 1992. Influence of flow regime on life history, production, and genetic structure of Baetis tricaudatus (Ephemeroptera) and Hesperoperla pacifica (Plecoptera). Journal of the North American Benthological Society 11:278-289. Rosenberg, D .M. , T.B. Reynoldson, and V . H . Resh. Design of a regional benthic biomonitoring program: I. Development of protocols for establishing refemce conditions in the Fraser River catchment, British Columbia. In preparation. Rosenberg, D .M. , and V . H . Resh. 1993. Freshwater biomonitoring and benthic macroinvertebrates. Chapman and Hall, New York. 116 SAS Institute. 1996. SAS/STAT Users' Guide, Release 6.11 ed. SAS Institute, Cary NC. Schindler, D.W. 1987. Detecting ecosystem responses to anthropogenic stress. Canadian Journal of Fisheries and Aquatic Science 44:6-25. Short, R.A., and J.V. Ward. 1981. Trophic ecology go three winter stoneflies (Plecoptera). American Midland Naturalist 105:341-347. Southwood, T.R.E. 1988. Tactics, strategies and templets. Oikos 52:3-18. Southwood, T.R.E. 1977. Habitat, the templet for ecological strategies. Journal of Animal Ecology 46:337-365. Stainton, M.P., M.J. Capel, and F.A.J Armstrong. 1977. The chemical analysis of fresh water. 2nd edition. Canadian Fisheries and Marine Service Misc. Special Publication, 25. 180p. Statsoft, Inc. 1994. Statistica Manual Volume I: Conventions and Statistics I. Tulsa, Oklahoma, Stearns, S.C. 1976. Life-history tactics: a review of the ideas. The Quarterly review of Biology 51(l):3-47. Stonebruner, D.L. , and L . A . Smock. 1979. Seasonal fluctuations of macroinvertebrate drift in a South Carolina piedmont stream. Hydrobiologia 63:49-56. Sweeney, B.W. 1984. Factors influencing life-history patterns of aquatic insects. Pages 56-85. In: V . H . Resh and D . M . Rosenberg (editors). The ecology of aquatic insects. Praeger, New York. Sweeney, B.W. and, R.L. Vannote. 1981. Ephemereila mayflies of White Clay Creek: bioenertetic and ecological relationships among six coexisting species. Ecology 62:1353-1369. SYSTAT 1993. SYSTAT Release 5.03 for Windows. SYSTAT, Inc. Evanston, IL. Tate, C M . , and J.S. Heiny. 1995. The ordination of benthic invertebrate communities in the South Platte River Basin in relation to environmental factors. Freshwater Biology 33:439-454. Townsend, C.R. 1989. The patch dynamics concept of stream community ecology. Journal of the North American Benthological Society 8:36-50. Townsend, C.R, and A . G . Hildrew. 1994. Species traits in relation to a habitat templet for river systems. Freshwater Biology 31:265-275. Townsend, C.R., M.R. Scarsbrook, and S. Doledec. 1997. Quantifying disturbance in streams: alternative measures of disturbance in relation to macroinvertebrate species traits and species richness. Journal of the North American Benthological Society 16:531-544. Vannote, R.L. , and B.W. Sweeney. 1980. Geographical analysis of thermal equilibria: a conceptual model for evaluating the effect of natural and modified thermal regimes on aquatic insect communities. American Naturalist 115:667-695. 117 Vannote, R.L. , G.W. Minshall, K.W. Cummins, J.R. Sedell, and C E . Cushing. 1980. The river continuum concept. Canadian Journal of Fisheries and Aquatic Science 37: 130-137. Ward, J.V. 1986. Altitudinal zonation in a Rocky Mountain stream. Arch. Hydrobiol./Supplement 74:133-199. Warwick, R .M. 1993. Environmental impact studies on marine communities: pragmatical considerations. Australian Journal of Ecology 18:63-80. Wiggins, G.B. 1996. Larvae of North American caddisfly genera (Trichoptera), 2n d edition. University of Toronto Press, Toronto. Winget, R.N. 1993. Habitat partitioning among three species of Ephemerelloidea. Journal of Freshwater Ecology 8(3):227-233. Wright, J. F. 1995. Development and use of a system for predicting the macroinvertebrate fauna in flowing waters. Australian Journal of Ecology 20:181-197. Wright, J.F., D. Moss, P.D. Armitage, and M.T. Furse. 1984. A preliminary classification of running-water sites in Great Britain based on macroinvertebrate species and the prediction of community type using environmental data. Freshwater Biology 14:221-256. Zamora-Munoz, C. and J. Alba-Tercedor. 1996. Bioassessment of organically polluted Spanish rivers, using a biotic index and multivariate methods. Journal of the North American Benthological Society 15:332-352. Zar, J.H. 1984. Biostatistical analysis. Prentice Hall, New Jersey. 118 Appendix 1.1 This study examined the life cycles of invertebrates common to, at minimum, both the interior and coastal streams. As a result, body length measures were taken for more taxa than the three included in this chapter. The measurements taken but not used are included in this appendix. The following taxa are included: Rhyacophila narvae (Rhyacophilidae: Trichoptera), R. acropedes group, Calineuria californica (Perlidae: Plecoptera), Zapada oregonensis group (Nemouridae: Plecoptera), Ephemerella inermis/infrequens (Ephemerellidae: Ephemeroptera), and Paraleptophlebia temporalis (Leptophlebiidae: Ephemeroptera). Most of the figures included in this appendix consist of raw data - measurements for each individual at a sampling date and site. Raw data are presented because it was not possible to separate the different generations. For R. narvae, however, mean body lengths are presented for each sampling date and site because the size classes were distinct enough to identify the different generations. 119 12 - i 11 ~\ • Mayfly Ck (C) A Beak Ck (I) 10 H 9 H 4=1 i 6 o 7 H 0> 5 H 4 H 3 H .19 11 2 H I: 13 I T T Date <7 Figure 1.1A: Life cycle plot for Rhyacophila narvae. Numbers next to the means are the number of individuals measured. Thick bars represent standard deviation and thin bars represent upper and lower size ranges. C = coastal streams and I = interior streams. 120 12 - i Mayfly Ck Spring Ck i o H f a o CQ 6 H 4 H * A 2 H Date Figure L I B : Raw life cycle data for Rhyacophila acropedes group from coastal streams. 121 Figure 1.1C: Raw life cycle data for Rhyacophila acropedes group from interior streams. 30 - i 25 H 20 H 122 • Spring Ck (C) • North Alouette R. (C) A Beak Ck (I) a 15 s o PQ io H 5 H • I i I 1 x Date Figure 1.1D: Raw life cycle data for Calineuria californica from both coastal (C) and interior (I) streams. 123 Figure 1.1E: Raw life cycle data for Zapada oregonensis group from Glimpse Creek (interior stream). 124 7 H Mayfly Ck (C) Spring Ck (C) Beak Ck (I) 6 H 5 H I , 4 ^ el o W 3 H i 2 H l H A A Date Figure 1.1F: Raw life cycle data for Zapada oregonensis group from coastal (C) and interior (I) streams 125 7 H Mayfly Ck (C) Spring Ck (C) Beak Ck (I) I I 6 H 5 H t 4 H T3 O CQ 3 H 2 H l H A I i I ! T Date Figure 1.1G : Raw life cycle data for Ephemerella inermis/infrequens from coastal (C) and interior (I) streams. 126 Figure 1.1H: Raw life cycle data for Ephemerella inermis / infrequens from large rivers. 127 10 - i • N . Alouette R. (C) • Fraser R. (L) 9 H A Thompson R.(L) • Beak Ck.tt) O Mayfly Ck(C) 7 H 6 H O O O O O O e 5 H 4 H 3 H 2 H l H O O O O O O O O 9 o 5 o x Date Figure 1.11: Raw life cycle data for Paraleptophlebia temporalis from coastal streams (C), interior streams (I), and large rivers (L). 128 Appendix 2.1: List of taxa identified, functional group classification, and the class(es) of streams in which they were found. C P O M shredder (SH), herbivore shredder (HE), deposit collector (DE), filter collector (FI), grazer (GR), wood gouger (GO), parasite (PA), and predator (PR) * = taxon found in > 25% of the samples from a specific stream class (i.e, common taxa). ! = taxon in less than 25% of the samples collected from a specific class of stream (i.e., rare taxa). T A X O N O M I C G R O U P FUNCTIONAL G INTERIOR COASTAL LARGE Coelentrata Hydra polyps ! P R - - + Turbellaria Tricladida P R + + + Nematoda P A + + + Gastropoda Valvatidae Valvata sincera ! GR - + -Plecypoda Sphaeriidae Psidium cassertanum * FI + - -Crustacea Amphipoda Hyalella azteca ! GR - - + Stygobromus sp. ! GR + - -Ostracoda GR + + -Oligochaeta Lumbriculidae GR + + -Naididae GR + - + Tubificidae GR/DE + + + Arachnida Araneae ! PR - + -Anisitsiellidae PA + + + Ataridae Albiinae ! PA - + -Axonopsinae ! PA + - -Aturidae Frontipodopsis sp. ! PA - + -Hydryphantidae Prozia sp. ! PA - + -Lebertiidae Lebertia sp. PA + + + Limnesiidae Limnesia sp. PA + + + Tyrreliinae PA + + + Sperchonidae Sperchon sp. PA + + + Torrenticolidae Torrenticola sp. PA + + + Unidentified mite ! PA + - -Pseudoscorpiones PR - + -Insecta Collembola DE + + + Ephemeroptera Ameletidae Ameletus sp. H E / D E + + + Baetidae Baetis sp. A H E / D E + + + Baetidae Baetis sp.B HE/DE + + + Baetidae Callibaetis sp. ! + - -Ephemerellidae Caudatella sp. HE/DE + - + Ephemerellidae Drunella doddsi H E / D E + _+ + Ephemerellidae Drunella flavilinea H E / D E + + + Ephemerellidae Drunella grandis ingens H E / D E + - + Ephemerellidae Drunella spinifera H E / D E + + + 129 Plecoptera Trichoptera Ephemerellidae Ephemerella inermis/infrequens D E + + + Ephemerellidae Serratella tibialis D E + + -Heptageniidae Cinygma sp. H E / D E + + -Heptageniidae Cinygmula sp. H E / D E + + + Heptageniidae Epeorus (Ironopsis) sp. * H E / D E - + -Heptageniidae Epeorus (Iron) sp. H E / D E + + + Heptageniidae Ironodes sp. * H E / D E - + -Heptageniidae Rhithrogena sp. H E / D E + + + Heptageniidae Stenonema sp. * H E / D E - - + Leptophlebiidae Paraleptophlebia temporalis D E + + + Tricorythidae Tricorythodes minutes ! D E - - + Capniidae Capnia sp SH + + + Capniidae Eucapnopsis brevicauda ! SH - + -Capniidae Mesocapnia sp. ! SH - - + Chloroperlidae Haploperla sp. PR + + -Chloroperlidae Kathroperla sp. ! D E - + -Chloroperlidae Plumiperla sp. + + -Chloroperlidae Suwallia sp. PR + + + Chloroperlidae Sweltsa sp. PR + + + Leuctridae Despaxia augusta SH + + + Leuctridae Perlomyia collaris SH + + -Nemouridae Amphinemura sp. * SH + - -Nemouridae Malenka sp. * SH - + -Nemouridae Ostrocerca sp. ! SH - + -Nemouridae Podmosta sp. SH + + + Nemouridae Visoka cataractae + + + Nemouridae Zapada cinctipes SH + + + Nemouridae Zapada oregonensis SH + + + Peltoperlidae Yoraperla brevis * SH - + -Perlidae Calineuria califomica PR + + -Perlidae Doroneuria baumanni PR + + -Perlidae Hesperoperla pacifica * PR - + -Perlodidae Isogenoides sp. ! PR - - + Perlodidae Isoperla sp. PR + + + Perlodidae Kogotus sp. ! PR - - + Perlodidae Megarcys sp. * PR + - -Perlodidae Setvena bradley * PR - + -Perlodidae Skwala sp. PR + + + Pteronarcyidae Pteronarcys princeps * SH - + -Taeniopterygidae Taenionema sp. SH + + + Apataniidae Apatania zonella HE/DE + + -Brachycentridae Amiocentrus aspilus ! D E + - -Brachycentridae Brachycentrus americanus * FI + - -Brachycentridae Brachycentrus occidentalis * FI - - + Brachycentridae Micrasema sp. A * SH - + -Glossosomatidae Glossosoma sp. H E + + + Hydropsychidae Parapsyche elsis FI Hydropsychidae Hydropsyche sp. FI Hydropsychidae Hydropsyche "British Columbia" FI Hydropsychidae Hydropsyche centra ! FI Hydropsychidae Hydropsyche oslari ! FI Hydroptilidae Hydroptilidae H E Hydroptilidae Agraylea sp. * H E Hydroptilidae Hydroptila sp. ! H E Leptoceridae Ceraclea sp. ! H E Lepidostomatidae Lepidostoma sp. SH Limnephilidae Limnephilidae SH Limnephilidae Chyranda centralis ! SH Limnephilidae Clostoeca sp. ! SH Limnephilidae Ecclisomyia sp. H E / D E Limnephilidae Hesperophylax sp. * SH/HE Limnephilidae Onocosmoecus unicolor * SH Limnephilidae Allomyia sp. * SH/DE Philopotamidae Wormaldia sp. * FI Philopotamidae Dolophilodes sp. ! FI Polycentropodidae Polycentropus sp. * PR Rhyacophilidae Rhyacophila sp. PR Rhyacophilidae Rhyacophila acropedes grp. PR Rhyacophilidae Rhyacophila angelita grp. PR Rhyacophilidae Rhyacophila chilsia grp. * PR Rhyacophilidae Rhyacophila grandis ! PR Rhyacophilidae Rhyacophila narvae PR Rhyacophilidae Rhyacophila norcuta grp. ! PR Rhyacophilidae Rhyacophila vagrita PR Uenoidae Neophylax sp. * SH/HE Coleoptera Amphizoidae Amphizoa sp. ! PR Dytiscidae Agabus sp. ! PR Dytiscidae Dytiscidae (adult) ! PR Dytiscidae Brachyvatus sp. ! PR Dytiscidae Desmopachria sp. (adult) PR Dytiscidae Laccornis sp. ! PR Elmidae Ancronyx sp. (adult). HE/DE Elmidae Cleptelmis sp. * H E / D E Elmidae Dubiraphia sp. ! H E / D E Elmidae Heterlimnius sp. H E / D E Elmidae Narpus sp. (adult) ! D E Elmidae Lara sp. * GO Elmidae Ordobrevia sp. (adult) H E / D E Elmidae Optioservus sp. (adult) * D E Elmidae Zaitzevia sp. (adult) ! HE/DE Tenebrionidae ! Hemiptera Corixidae ! 131 Odonata Diptera Coenargrionidae ! - - + Athericidae Atherix sp. ! PR - - + Blephariceridae Blepharicera sp. ! HE/DE - + -Ceratopogonidae Ceratopogoninae PR + + + Ceratopogonidae Atrichopogon sp. ! H E / D E - + -Chironomidae Chironominae HE/DE + + + Constempellina sp. DE/FI + + + Robackia sp. * - - + Tanytarsus sp. DE/FI + + + Chironomidae Orthocladiinae H E / D E + + + Corynoneura sp. H E / D E + + + Chironomidae Prodiamesinae ! H E / D E + - -Chironomidae Tanypodinae R/DE/F + + + Dixidae Dixa sp. ! D E - + -Dolichopodidae Hydrophorus sp. ! PR + - -Empididae Chelifera sp. PR + + -Empididae Hemerodromia sp. R/DE/F - + + Empididae Oreogeton sp. PR + + -Psychodidae Pericoma sp./Thelmatoscopus sp. * D E + - -Simuliidae Prosimulium sp. FI + + + Simuliidae Simulium sp. FI + + + Tipulidae Antocha monticola D E + + + Tipulidae Dicranota sp. PR + + + Tipulidae Hexatoma sp. A * PR + - -Tipulidae Hexatoma sp. B * PR - + -Tipulidae Rhabdomastix sp. - + + Tipulidae Tipula sp. ! SH/DE + - -Appendix 2.2: Invertebrate taxa included in genus/species and family level analyses and their respective class, order, and family. T A X O N O M I C C L A S S I F I C A T I O N S G R O U P S I N E A C H A N A L Y S I S C L A S S O R D E R F A M I L Y GENUS/SPECIES F A M I L Y Turbellaria Tricladida Ostracoda Oligochaeta Arachnida Acari Insecta Ephemeroptera Plecoptera Trichoptera Coleoptera Diptera Naididae Tubificidae Lebertiidae Torrenticolidae Baetidae Baetidae Ephemerellidae Ephemerellidae Heptageniidae Heptageniidae Leptophlebiidae Capniidae Chloroperlidae Nemouridae Nemouridae Nemouridae Perlodidae Brachycentridae Glossosomatidae Hydropsychidae Lepidostomatidae Rhyacophilidae Uenoidae Elmidae Chironomidae Chironomidae Psychodidae Simuliidae Tubificidae Torrenticola sp. Baetis sp.A Baetis sp.B Ephemerella inermis/infrequens Serratella tibialis Cinygmula sp. Epeorus (Iron) sp. Paraleptophlebia temporalis Capnia sp. Sweltsa sp. Podmosta sp. Zapada cinctipes Zapada oregonensis Brachycentrus americanus Hydropsyche sp. Lepidostoma sp. Neophylax sp. Heterlimnius sp. Chironominae Orthocladiinae Pericoma sp. / Thelmatoscopus sp. Prosimulium sp. Tricladida Ostracoda Naididae Tubificidae Lebertiidae Torrenticolidae Baetidae Ephemerellidae Heptageniidae Leptophlebiidae Capniidae Chloroperlidae Nemouridae Perlodidae Brachycentridae Glossosomatidae Hydropsychidae Lepidostomatidae Rhyacophilidae Uenoidae Elmidae Chironomidae Psychodidae Simuliidae 133 Appendix 3.1: Invertebrate taxa included in analysis at the lowest possible taxonomic level (A) and in the analysis at the family level (B), and two censorship levels (0.1% and 0.5%). CLASS ORDER usrracoda Turbellaria Tricladida Plecypoda Oligochaeta Arachnida Acari Insecta Ephemeroptera FAMILY LOWEST LEVEL DATA (>0.1%) LOWEST LEVEL DATA (>0.5%) Plecoptera Trichoptera Sphaeriidae Sphaeriidae Sphaeriidae Naididae Naididae Lebertiidae Sperchonidae Torrenticolidae Torrenticolidae Ameletidae Baetidae Ephemerellidae Ephemerellidae Ephemerellidae Ephemerellidae Ephemerellidae Ephemerellidae Ephemerellidae Heptageniidae Heptageniidae Heptageniidae Leptophlebiidae Capniidae Chloroperlidae Chloroperlidae Leuctridae Nemouridae Nemouridae Nemouridae Nemouridae Nemouridae Peltoperlidae Perlidae Perlodidae Taeniopterygidae Brachycentridae Brachycentridae Glossosomatidae Glossosomatidae Hydropsychidae Hydropsychidae Hydroptiliidae Pisidium casertanum Pisidium casertanum Pisidium spp Sphaerium striatum Nais alpina Nais behningi Lebertia spp Sperchon spp Testudacarus spp Torrenticola spp Ameletus spp Baetidae Baetidae Drunella doddsi Drunella doddsi Drunella grandis ingens Drunella spinifera Ephemerella infrequens Ephemerella spp Ephemerella spp Serratella spp Serratella spp Serratella tibialis Serratella tibialis Epeorus spp Epeorus spp Heptageniidae Heptageniidae Rhithrogena spp Rhithrogena spp Paraleptophlebia spp Paraleptophlebia spp Capnia spp Capnia spp Plumiperla diversa Sweltsa spp Sweltsa spp Paraleuctra spp Podmosta spp Zapada cinctipes Zapada cinctipes Zapada columbiana Zapada columbiana Zapada oregonensis Zapada spp Zapada spp Yoraperla mariana Calineuria californica Isoperla spp Taenionema spp Brachycentrus americanus Micrasema sp Glossosoma spp Protoptila spp Hydropsyche morosa Hydropsyche spp Hydroptila spp Taenionema spp Hydropsyche spp Coleoptera Diptera Hydroptiliidae Lepidostomatidae Rhyacophilidae Uenoidae Elmidae Elmidae Elmidae Elmidae Chironomidae Chironomidae Chironomidae Chloroperlidae Diamesinae Empididae Empididae Psychodidae Simuliidae Tipulidae Tipulidae Stactobiella spp Lepidostoma spp Rhyacophila spp Oligophlebodes spp Heterlimnius spp Optioservus spp Zaitzevia parvula Zaitzevia spp Chironominae Orthocladiinae Tanypodinae Haploperla brevis Diamesinae Chelifera spp Clinocera spp Pericoma/Thelmatoscopus Simulium spp Antocha monticola Dicronota spp Lepidostoma spp Rhyacophila spp Optioservu s spp Chironominae Orthocladiinae Diamesinae spp Simulium spp B C L A S S u o p q j o d ^ Ostracoda Turbellaria Plecypoda O R D E R F A M I L Y L E V E L D A T A (>0.1%) Copepoda lamily = Ostracoda family Tricladida family Sphaeriidae F A M I L Y L E V E L D A T A (>0.5%) Tricladida Ostracoda family Tricladida family Sphaeriidae Oligochaeta Arachnida Insecta Acari Ephemeroptera Enchytraeidae Naididae Lebertiidae Sperichontidae Torrenticolidae Ameletidae Baetidae Ephemerellidae Naididae Baetidae Ephemerellidae Heptageniidae Heptageniidae Plecoptera Leptophlebiidae Capniidae Chloroperlidae Leptophlebiidae Capniidae Chloroperlidae Leuctridae Nemouridae Nemouridae Trichoptera Peltoperlidae Perlodidae Taeniopterygidae Brachycentridae Perlodidae Taeniopterygidae Brachycentridae Glossosomatidae Hydropsychidae Hydroptiliidae Lepidostomatidae Hydropsychidae Hydroptiliidae Lepidostomatidae Coleoptera Diptera Limnephilidae Rhyacophilidae Uenoidae Elmidae Chironomidae Empididae Psychodidae Simuliidae Tipulidae Rhyacophilidae Elmidae Chironomidae Empididae Simuliidae Tipulidae 

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.831.1-0099262/manifest

Comment

Related Items