@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix dc: . @prefix skos: . vivo:departmentOrSchool "Forestry, Faculty of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Decker, Andrew Scott"@en ; dcterms:issued "2009-06-16T18:01:44Z"@en, "1999"@en ; vivo:relatedDegree "Master of Science - MSc"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """In British Columbia and the Pacific Northwest, construction of off-channel habitat including artificial ponds and groundwater-fed channels, figures prominently in restoration efforts intended to benefit coho salmon (Oncorhynchus kisutch). To examine how juvenile coho production was effected by autotrophic productivity in these systems, I estimated the average size and abundance of juvenile coho during late summer and early spring over two years in three pairs of physically similar channels and one pair of ponds with one member of each pair having relatively high periphyton biomass and the other member having low periphyton biomass. Differences in total alkalinity between paired sites suggested that nutrient availability was a strong determinate of periphyton biomass on artificial substrate. However, periphyton biomass did not appear limited by phosphorus alone; phosphorus was relatively abundant in the groundwater channels, and as a result, nitrogen and possibly other nutrients were also important. Average coho fry size and biomass per unit area were greater by late summer in the periphyton-rich sites, but fry density was not. Physical habitat variability also failed to explain high variation in coho densities among the groundwater channels. Biomass and taxonomic composition of benthic invertebrates also appeared to be related to periphyton biomass, but a strong interaction between invertebrate abundance or distribution and current velocity was a confounding factor. In the subsequent spring, pre-smolt coho were still larger in the periphyton-rich site in the majority of cases. However, water temperature during the winter had a much greater effect on average pre-smolt size than growth rate the previous summer. In the groundwater channels, consumption of chum salmon (O. keta) eggs, fry, and carcasses may have also contributed to high coho growth rates during the winter. The abundance of spring pre-smolts in the channels and ponds was not related to either periphyton biomass or average fry size the previous summer. Favorable growing conditions during the winter may have lessened size-dependent overwinter mortality, but the benefit of nutrient addition to coho production in off-channel habitat remains uncertain. Pre-smolt abundance in the off-channel sites was strongly influenced by small differences in water depth and velocity in winter and by fry density the previous summer. Physical habitat differences between the channels and ponds also influenced abundance: pre-smolts densities were consistently higher in the deeper, more structurally complex ponds. Large, density-dependent declines in coho abundance in the relatively warm and hydrologically stable groundwater channels from September to March contradicts traditional emphasis on starvation and adverse physical conditions (e.g., freshets) as primary sources of coho winter mortality. My results suggest that other factors such as predation or energetics warrant attention."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/9267?expand=metadata"@en ; dcterms:extent "6386541 bytes"@en ; dc:format "application/pdf"@en ; skos:note "EFFECTS OF PRIMARY PRODUCTION AND OTHER FACTORS ON THE SIZE AND ABUNDANCE OF JUVENILE COHO SALMON IN ARTIFICIAL OFF-CHANNEL HABITAT by ANDREW SCOTT DECKER B.S.F., The University of British Columbia, 1993 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Forest Sciences) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA May 1999 © Andrew Scott Decker, 1999 ln presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, 1 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 In British Columbia and the Pacific Northwest, construction of off-channel habitat including artificial ponds and groundwater-fed channels, figures prominently in restoration efforts intended to benefit coho salmon (Oncorhynchus kisutch). To examine how juvenile coho production was effected by autotrophic productivity in these systems, I estimated the average size and abundance of juvenile coho during late summer and early spring over two years in three pairs of physically similar channels and one pair of ponds with one member of each pair having relatively high periphyton biomass and the other member having low periphyton biomass. Differences in total alkalinity between paired sites suggested that nutrient availability was a strong determinate of periphyton biomass on artificial substrate. However, periphyton biomass did not appear limited by phosphorus alone; phosphorus was relatively abundant in the groundwater channels, and as a result, nitrogen and possibly other nutrients were also important. Average coho fry size and biomass per unit area were greater by late summer in the periphyton-rich sites, but fry density was not. Physical habitat variability also failed to explain high variation in coho densities among the groundwater channels. Biomass and taxonomic composition of benthic invertebrates also appeared to be related to periphyton biomass, but a strong interaction between invertebrate abundance or distribution and current velocity was a confounding factor. In the subsequent spring, pre-smolt coho were still larger in the periphyton-rich site in the majority of cases. However, water temperature during the winter had a much greater effect on average pre-smolt size than growth rate the previous summer. In the ii groundwater channels, consumption of chum salmon (O. keta) eggs, fry, and carcasses may have also contributed to high coho growth rates during the winter. The abundance of spring pre-smolts in the channels and ponds was not related to either periphyton biomass or average fry size the previous summer. Favorable growing conditions during the winter may have lessened size-dependent overwinter mortality, but the benefit of nutrient addition to coho production in off-channel habitat remains uncertain. Pre-smolt abundance in the off-channel sites was strongly influenced by small differences in water depth and velocity in winter and by fry density the previous summer. Physical habitat differences between the channels and ponds also influenced abundance: pre-smolts densities were consistently higher in the deeper, more structurally complex ponds. Large, density-dependent declines in coho abundance in the relatively warm and hydrologically stable groundwater channels from September to March contradicts traditional emphasis on starvation and adverse physical conditions (e.g., freshets) as primary sources of coho winter mortality. My results suggest that other factors such as predation or energetics warrant attention. iii TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iv LIST OF TABLES vi LIST OF FIGURES ix ACKNOWLEDGEMENTS xv GENERAL INTRODUCTION 1 CHAPTER 1: METHODS 5 STUDY AREAS 5 Groundwater channels 5 Off-channel ponds 8 FIELD METHODS 10 Periphytic algae 10 Temperature and discharge 11 Water chemistry 12 Benthic invertebrates 13 Juvenile coho salmon 14 CHAPTER 2: SUMMER ECOLOGY 19 INTRODUCTION .19 DATA ANALYSIS 21 RESULTS 24 Periphyton accrual 24 Water quality 26 Benthic invertebrates 29 Coho salmon 30 DISCUSSION ; 34 Periphyton accrual 34 iv Water quality 36 Benthic invertebrates 38 Coho salmon 42 CHAPTER 3: WINTER ECOLOGY 74 INTRODUCTION 74 DATA ANALYSIS 77 RESULTS 83 Temperature and discharge 83 Fish behaviour 84 Seasonal changes in fish size 85 Seasonal changes in fish abundance 90 DISCUSSION 93 Seasonal changes in fish size 93 Seasonal changes in fish abundance 100 Effect of habitat 101 Factors important to overwinter survival 104 CHAPTER 4: CONCLUSIONS AND MANAGEMENT IMPLICATIONS 127 BIBLIOGRAPHY 13 v LIST OF TABLES Table 2.1. Summary of means, standard errors and sample sizes for temperatures and current velocities in three pairs of groundwater channels and the inlet channels of a pair of ponds (Coquitlam) during June-October 1995. Pairs are identified by the names of the parent rivers given in the left margin. Each pair consists of a site with relatively high periphyton biomass (HP) and a site with relatively low biomass (LP).49 Table 2.2. Summary of means, standard errors and sample sizes for total alkalinity, soluble reactive phosphorus and dissolved inorganic nitrogen for three pairs of groundwater channels and the inlet channels of a pair of ponds (Coquitlam) during June-October 1995. Pairs are identified by the names of the parent rivers given in the left margin. Each pair consists of a site with relatively high periphyton biomass (HP) and a site with relatively low biomass (LP). Underlined values indicate significant differences between paired sites 50 Table 2.3. Mean atomic N:P ratios (dissolved inorganic nitrogen : soluble reactive phosphorus) among three pairs of groundwater channels and a pair of ponds (Coquitlam) during June - October, 1995. Values in brackets are standard errors, underlined values indicate significant differences in N:P ratios between paired sites.51 Table 3.1. Percent increases in the mean weight of coho fry from paired groundwater channels and ponds during 1995-96 and 1996-97. Sites are paired by river. Each pair consists of a site with high periphyton biomass (HP) and a site with low biomass (LP). Values reported in the table are the changes (%) in mean weight between the sample dates indicated. In 1996-97, sampling occurred in September and March only 108 Table 3.2. Multiple regression analysis of coho pre-smolt weight, density and biomass with fall fry density, mean winter water temperature, and habitat variables derived vi from principal component analysis (see Table 3.6). Variables were log 10 transformed prior to analysis. Models a and b, include groundwater channels and ponds as sample units. Models d and e, include only groundwater channels. Model c includes 20 m groundwater channel reaches as sample units (three per channel). Partial r2 values are the squared values for the correlation between the given independent variable and the dependent variable controlling for the other independent variables in the model. For a particular model, total r2 (coefficient of determination) andp (probability) are cumulative in that these values are re-calculated with the inclusion of each new independent variable. Slope is the coefficient for each independent variable in a model. Only variables that significantly accounted for variation in the dependent variable were included in each model. Data were checked for multi-collinearity and logio transformed prior to analysis 109 Table 3.3. Sample sizes, mean total lengths (cm) and associated standard errors for samples of coho fry from off-channel (LP, HP) and mainstem sites in four rivers during September 1995. LP (low productivity) and HP (high productivity) refer to the relative amount of periphyton biomass between off-channel pairs. Underlined values for length indicate significant differences between sites from each river (ANOVA, bonferroni adjusted pair-wise comparison, logged values, p < 0.05) 110 Table 3.4. Mean total lengths of marked and unmarked coho fry from three pairs of groundwater channels and a pair of ponds during November, January, and March 1995-96. Pairs are identified by the name of the parent river. Each pair consists of a site having high periphyton biomass (HP) and a site having low biomass (LP). Values in brackets are standard errors Ill Table 3.5. Mean densities and biomass per unit area of juvenile coho from three pairs of groundwater channels and a pair of ponds (Coquitlam). Pairs are identified by the name of the parent river. Each pair consists of a site haying high periphyton biomass (HP) and a site having low biomass (LP). Values in brackets are standard errors and underlined values indicate significant differences between paired sites 112 vii Table 3.6 a-b. Correlations of habitat variables with principal components (PC's) derived from factoring the correlation matrix. Data are from 18 groundwater channel reaches (three 20 m reaches in each of the six channels). A description of the variables is given in Field methods in Chapter 1. The proportion of the total variance accounted for by each PC is given at the bottom of each table. In Table 3.6a, the variables are not rotated, whereas in Table 3.6b, the loading of variables on the PC's was improved using an orthogonal factor rotation (Orfhomax; SYSTAT 1997) 113 viii LIST OF FIGURES Figure 1.1. Locations of the four study rivers within the Fraser and Squamish River Watersheds in southern coastal British Columbia. The solid circles show the location of the artificial groundwater channels and ponds (Coquitlam) selected for paired observation 18 Figure 2.1. Mean weekly chlorophyll a density on artificial substrates from three pairs of groundwater channels and a pair of ponds (Coquitlam) during July, September and March 1995-96. Pairs are identified by the name of the parent river given in the upper margin of each graph. For each pair, the site with higher peak chlorophyll a density is classed as the HP (high productivity) member whereas the site with lower peak density is classed as the LP (low productivity) member. Numbers along the x-axis refer to the number of weeks from the time of initial placement during each period 52 Figure 2.2. Mean weekly temperatures from three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96. Pairs are identified by the name of the parent river given in the left-hand margin. Each pair consists of a site with relatively high periphyton biomass (HP) and a site with relatively low biomass (LP). Solid lines and shaded circles denote mean temperatures, dashed lines denote maximum and minimum temperatures observed 54 Figure 2.3. Total alkalinity concentrations from three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96. Each pair is identified by the name of the parent river given in the left-hand margin. Solid lines and shaded diamonds denote the member of each pair having high periphyton biomass (HP) and dashed lines and unshaded squares denote the member having low biomass (LP) 56 Figure 2.4. Soluble reactive phosphorus (SRP) concentrations from three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96. Each pair is ix identified by the name of the parent river given in the left-hand margin. Solid lines and shaded diamonds denote the member of each pair having high periphyton biomass (HP) and dashed lines and unshaded squares denote the site having low biomass (LP) 58 Figure 2.5. Dissolved inorganic nitrogen (DIN) concentrations from three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96. Each pair is identified by the name of the parent river given in the left-hand margin. Solid lines and shaded diamonds denote the member of each pair having high periphyton biomass (HP) and dashed lines and unshaded squares denote the site having low biomass (LP) 60 Figure 2.6. Proportional composition of benthic invertebrates by mean numbers of taxa from three pairs of groundwater channels and a pair of ponds (Coquitlam) during September 1995. Pairs are identified by the name of the parent river given in the upper margin. Each pair consists of a site having high periphyton biomass (HP) and a site having low biomass (LP) 62 Figure 2.7. Logarithmic mean biomass of benthic invertebrates from three pairs of groundwater channels and the inlets of a pair of ponds during September 1995. Pairs are identified by the names of the parent rivers given in the lower margin of each graph. Shaded bars denote values for the site of each pair having high biomass of periphyton and unshaded bars denote values for the site having low biomass. Error bars indicate ± one standard error and stars indicate significant differences between pairs (n = 3, p < 0.05) ....63 Figure 2.8. Mean biomass of benthic invertebrates in off-channel habitat sites during September 1995 in relation to peak chlorophyll a density (average of values for July, September, and March 1995-96). Note the log scales on both axes. Values for r2 and p are the coefficient of determination and probability, respectively, for the linear regression of chlorophyll a on invertebrate biomass 64 Figure 2.9. Mean biomass of benthic invertebrates in relation to current velocity in the immediate vicinity of the colonization baskets during September 1995. Note the log scale on the Y axis. Values for r2 and p are the coefficient of determination and probability, respectively, for the linear regression of velocity on invertebrate biomass (groundwater channels only) 65 Figure 2.10. Mean weight of juvenile coho during September 1995 in relation to peak chlorophyll a density (average of values for July, September, and March 1995-96). Note the log scales on both axes. Values for r2 andp are the coefficient of determination and probability, respectively, for the linear regression of chlorophyll a on coho weight (groundwater channels only) 66 Figure 2.11. Mean weight of juvenile coho in relation to benthic invertebrate mean biomass during September 1995. Note the log scales on both axes. Values for r2 and p are the coefficient of determination and probability, respectively, for the linear regression of invertebrate biomass on coho weight (groundwater channels only) 67 Figure 2.12. Mean weight, biomass, and density of coho fry during September from three pairs of groundwater channels and a pair of ponds (Coquitlam) in 1995 and 1996. Pairs are identified by the names of the parent rivers given in the lower margin of each graph. Shaded bars denote values for the member of each pair having high periphyton biomass (HP) and unshaded bars denote values for the member having low biomass (LP). The ponds were sampled as whole units. Error bars indicate ± one standard error and stars indicate significant differences in mean density between pairs (n = 3, p < 0.05) 68 Figure 2.13. Length-frequency histograms for coho fry populations from three pairs of groundwater channels or ponds (Coquitlam) during September 1995 and 1996. Pairs are identified by the names of the parent rivers given in the left-hand margin. For each pair in each year, the site with low periphyton biomass (LP) appears on the right and the high biomass site (HP) appears on the left. The x-axis is labeled xi identically among figures, but is only shown for the figure in the bottom left hand corner 70 Figure 2.14. Length vs. weight linear regression predicted values for coho fry populations from sites with low periphyton biomass (broken lines) and sites with high biomass (solid lines) during September 1995 and 1996. Pairs are identified by the names of the parent rivers given in the left-hand margin 72 Figure 3.1. Responses of empirical quantile-quantile (QQ) plots to combinations of size-selective mortality of small and large individuals and to growth models in which the incremental change in length of small individuals is equal to or less than that of large individuals. The dashed line is the 1:1 line. The open arrows indicate the direction of the translation due to growth effects, the size of the arrows indicating the relative magnitude of the response. The curved arrows indicate the rotation due to size-selective mortality. The a, b and c lines illustrate how combinations of size-selective mortality and size-selective growth can result in conflicting interpretations of QQ plots, (adapted from Post and Evans 1989) 114 Figure 3.2. Annual thermographs of four rivers and associated pairs of groundwater channels or ponds (Coquitlam). In each graph, the river is represented by shaded stars and a solid line, the channel or pond with high periphyton biomass (HP) by shaded circles and a long-dashed line, and the channel or pond with low biomass (LP) by open circles and a short-dashed line. Trend lines were drawn using LOWESS (SYSTAT 1997), a running average smoothing function. Data points for each channel or pond are mean weekly water temperatures recorded during 1995-96. Temperature data for each river were obtained from historic water temperature records (Environment Canada 1977) 116 Figure 3.3. Weekly estimates of discharge from three pairs of groundwater channels and the inlet channels of a pair of ponds (Coquitlam) during 1995-96. Each pair is xii identified by the name of the parent river given in the left hand margin. LP and HP denote relatively low and high periphyton biomass, respectively 117 Figure 3.4. Mean weight of coho fry in three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96 and 1996-97. Each pair is identified by the name of the parent river given in the left hand margin. Shaded bars denote values for the member of each pair having high biomass of periphyton and unshaded bars denote values for the member having low periphyton biomass. Error bars indicate ± one standard error 119 Figures 3.5, 3.6 Empirical quantile-quantile plots derived from the length-frequency distributions of coho fry in September (x-axis) and March (y-axis) for four pairs of groundwater channels or ponds (Coquitlam) during 1995-96 (a) and 1996-97 (b). Data points represent quantiles 5, 10, 25, 50, 75, 90 and 95 of the length distribution from the two periods. Dashed lines are 1:1 lines. Pairs are identified by the names of the parent rivers given in the left margin. Right-hand-side graphs correspond to the member of each pair having low biomass of periphyton (LP), and left-hand-side graphs correspond to the member having high biomass (HP). Trend-lines were fitted by linear regression. Sample sizes («) are indicated in cases where the number of fish measured in March was less than 100 121 Figure 3.7. Logarithmic mean densities (fish-m\") of juvenile coho in September and March in six groundwater channels (solid lines) and two ponds (broken lines) during 1995-96 and 1996-97. Geometric mean densities, sample sizes and standard errors are given in Table 3.5 124 Figure 3.8. Seasonal declines in CPUE for coho fry in six groundwater channels during 1995-96. For a given channel on a particular sample date, CPUE represents the total number of fish captured during one pass of electro-fishing in three 20 m permanent sample reaches. Values for the Vedder LP channel in July and both Vedder channels in September are not given. On these occasions, electro-fishing was conducted xiii throughout the Vedder sites rather than in just the 20 m reaches (see Field methods in Chapter 1) 126 Figure 3.9 a-d. Abundance of coho pre-smolts in March in six groundwater channels in relation to: a) late summer fry density, b) depth-velocity (see Data Analysis), c) average winter water temperature (Nov. 1 - March 31), and d) fry density the previous September. Note the log scale on the Y axis. The scatter-plots include data from both 1995-96 and 1996-97 127 xiv ACKNOWLEDGEMENTS I am grateful to my supervisor, Scott Hinch, for his patience and support. I thank the other members of my advisory committee, Carl Walters and Steve McDonald, for their advice with regard to the design of my study and for their comments on an earlier draft. Matt Foy's experience with habitat restoration and his enthusiasm and willingness to test my results in practical applications were a tremendous source of encouragement. I am also indebted to Stephen Riley for being both a friend and mentor. His editing skills and insight regarding stream ecology are noted. Joda Takarabe's hard work and patience as a field assistant were a great asset. His presence was a wonderful display of friendship. Thaddeus Seidler, Steve Ruskey, Pier van Dishoech and Andrew Lotto also helped in the field and in the laboratory. Additional field assistance was provided by volunteers that generously donated their time. Several colleagues at U.B.C., including Sean Cox, Guillermo Giannico, Joel Sawada and Lisa Thompson, also provided advice and assistance. Bob Brown, Maurice Coulter-Boisvert, Rheal Finnigan, Josh Korman, Sara Mouldey, and Brian Toth contributed as well. Ken Ashley, Pat Slaney and Bruce Ward helped secure funding from Forestry Renewal B.C., and provided equipment and expertise. Goff Longworth secured additional funding from B.C. Hydro. Technical support and advice were also provided by Michael Bradford, Michael Healey, Tom Johnston and Bob Land. Finally, I would like to thank Jessica Bratty for her critical input, but more importantly, for her perspective, love and support. xv GENERAL INTRODUCTION The decline of coho salmon (Oncorhynchus kisutch) and other anadromous salmonid stocks along the Pacific coast of North America has become one of the leading conservation issues in Canada and the United States (Brown et al. 1994; Nehlsen et al. 1991; Nickelson and Lawson 1998; Slaney et al. 1996). While overfishing and poor ocean conditions have contributed to stock depletions (Holtby and Kadowaki 1998; Walters 1993), a reduction in freshwater rearing capacity as a result of habitat degradation has been assumed to play a major role (Beechie et al. 1994; Frissell 1993; Nehlsen et al. 1991). Attempts to mitigate reduced freshwater salmon production through hatchery propagation have largely failed to reverse the trend (Coronado and Hillborn 1998; Nickelson et al. 1986), and in some cases have contributed to the erosion of wild populations (Goodman 1990; Hillborn 1992; Walters and Cahoon 1985). Watershed restoration has been lauded by many as an alternative means of restoring the productive potential of freshwater ecosystems. However, the potential benefits of restoration may only be realized in the long-term; it may take decades for the full recovery of watershed structure and function following natural or anthropogenically induced perturbations (Lewis et al. 1996). Stream habitat improvement is one component of watershed restoration and refers to efforts within the stream channel and riparian zone to enhance rearing conditions for fish in the short-term (Hartman et al. 1996). Although first started during the 1930's in North America, stream habitat improvement has increased dramatically over the past two decades at a substantial cost to the public. 1 However, there is growing concern over the lack of rigorous evaluation of the various techniques (Keeley and Walters 1996; Minns et al. 1996; Rabeni and Sowa 1996; Riley and Fausch 1995; White 1996). Predicting the response of fish populations to habitat improvement is often limited by our relatively poor understanding of the factors that constrain the abundance of each species or life-history stage (Bisson et al. 1992; Reeves etal. 1991). Among Pacific salmon, coho may benefit the most from habitat improvement because of their relatively lengthy freshwater residency. Juvenile coho spend one or sometimes two years in freshwater before migrating seaward as smolts. Freshwater production is thought to be limited in part by low survival during the winter (Hartman et al. 1996), and there is considerable evidence to suggest that overwinter survival is influenced by habitat quality (Nickelson et al. 1992; Quinn and Peterson 1996). Suitable habitat may provide refuge from adverse hydrologic conditions or predators during winter (see Cunjak 1996), but the mechanisms underlying the relationship between habitat and smolt production remain unclear. Attempts to improve winter habitat for coho salmon include restructuring degraded stream channels, placement of woody debris, and construction of artificial groundwater-fed off-channel habitat (Reeves et al. 1991). Artificial groundwater channels and ponds may be more effective than in-channel structures because they are less prone to failure in destabilized, high energy coastal streams (Frissell and Nawa 1992; Reeves et al. 1991). In addition, both artificial and natural off-channel habitat (e.g., riverine ponds, ephemeral tributaries and swamps, groundwater tributaries) have been shown to support relatively high proportions of 2 overwintering juveniles in some streams (Brown and Hartman 1988; Cederholm and Scarlett 1982; Decker 1998; Everest et al. 1986). Despite this, little effort has been made to identify the factors affecting fish productivity in off-channel areas. Coho overwinter survival and smolt production may also be affected by body size at the end of summer (Holtby and Hartman 1982). Because primary production is often limited by phosphorus or nitrogen in temperate streams (Hynes 1970), the addition of nutrients (Perrin et al. 1987; Peterson et al. 1993) or salmon carcasses (Bilby et al. 1996; Wipfli et al. 1998) may positively influence fish growth in the summer by increasing biomass at lower trophic levels (Deegan and Peterson 1992; Johnston et al. 1990). However, the effect of primary production on overwinter survival remains uncertain and likely depends on additional factors such as temperature, the severity or duration of winter conditions, and habitat characteristics (Cunjak 1996; Hunt 1969; Mason 1976; Oliver et al. 1979). Artificial groundwater channels and ponds differ markedly from surface water streams in several ways: temperature is colder during the summer and warmer during the winter, flows are relatively stable, and habitat is less heterogeneous both within and among sites. This contrast provides interesting research opportunities. For example, if habitat acts primarily to reduce displacement of individuals during severe winter conditions (e.g., winter freshets; Tschaplinski and Hartman 1983), then smolt production in stable groundwater systems should be less dependent on physical habitat. Furthermore, if primary production is a strong determinant of overwinter survival, this 3 should be more evident in artificial channels and ponds where variability in habitat quality is relatively low. This thesis addresses such gaps in our understanding of the freshwater ecology of coho by investigating some of the factors influencing juvenile coho production in off-channel habitat. My basic approach was to measure abundance, growth and survival in an experimental set of groundwater channels and ponds constructed in southern coastal British Columbia. In Chapter 1,1 describe my study sites and the methods I used to collect the field data. In Chapter 2,1 trace the 'bottom-up' effect of autotrophic productivity during the summer by comparing benthic invertebrate biomass, and fish size and abundance between paired off-channel sites in which the standing crop of periphyton differed. In Chapter 3,1 examine the relationship between coho size and abundance at the end of summer and abundance the following spring. I also consider whether this relationship depends on temperature, food availability and habitat conditions during the winter. I also speculate on the major sources of overwinter mortality in juvenile coho based on differences in the seasonal pattern of abundance I observed among study sites, and on differences between these sites and surface water streams. Finally, in Chapter 4,1 summarize my findings and provide several suggestions for fisheries managers. 4 CHAPTER 1: METHODS STUDY AREAS Three pairs of artificial groundwater-fed channels and one pair of artificial ponds were selected as off-channel habitat sites for this study. The paired sample design was intended to contrast the mean size and standing crop of juvenile coho in channels where autotrophic productivity differed. Each pair of groundwater channels or ponds was selected from a different river system. I attempted to select regional pairs for which other abiotic and biotic parameters including size, construction date, proximity, physical habitat, discharge, and riparian vegetation were as similar as possible. Groundwater channels The three groundwater channel pairs are located in the lower reaches of the Cheakamus, Mamquam and Chilliwack-Vedder Rivers. These high-order streams are glacial; snow; and rain-fed and have mean annual flows of 32, 26, and 67 m3-s\"', respectively (Environment Canada 1991). The Cheakamus and Mamquam Rivers are major tributaries of the Squamish River and rise within the southern range of the Coast Mountains. Both rivers join the Squamish within five km of Howe Sound, 60 km west of Vancouver, British Columbia (Fig. 1.1). The Chilliwack-Vedder River flows from the Cascade Mountains into the Fraser River 100 km upstream of Georgia Strait (Fig. 1.1). In the lower reaches of the three rivers where the groundwater channel pairs are situated, gradient is low (< 0.5 %), and mainstem channels are confined by extensive systems of dikes. Riparian areas have been previously logged and are now dominated by rural 5 (Cheakamus, Mamquam) or agricultural (Chilliwack-Vedder) development. Prior to settlement on the flood plain and the ensuing demand for flood control, the lower portions of each river were likely characterized by unconfined and often braided channels. Natural off-channel habitats including wetlands, cut-off meanders and wall-base groundwater channels, although common historically in Pacific coastal drainages (Beechie et al. 1994; Peterson and Reid 1984), are relatively rare. The groundwater channels included in this study were constructed during the 1980's, mainly to provide spawning habitat for chum salmon (Bonnell 1991). They were created by excavating a portion of the floodplain parallel to the river mainstem. Each channel originates from a blind upstream end, and flows into its parent river either directly or through an existing natural channel. Only the excavated portion of each channel was included in the study. The channels are protected from mainstem flooding by set-back dykes. Flow within each channel is derived entirely from groundwater upwelling or sub-surface flow along the adjacent river's gravel fan. The channels are confined and relatively straight with little variance in morphology (width, depth), substrate composition, or current velocity (they lack the riffle-pool morphology typical of small streams). Among the channels, average width ranges from 5.0 to 7.0 m, and average depth from 25 to 55 cm. The banks of the channels are lined with large angular rip-rap boulders which provide the principle source of fish hiding cover (Sheng et al. 1990). Mid-channel cover such as boulders, deep pools or woody debris is uncommon. Adjacent vegetation is generally composed of deciduous shrubs or red alder (Alnus rubra) saplings. Light levels are high except for the Cheakamus channels which are 6 partially shaded by the riparian canopy. Substrate in the upstream excavated portion of each channel is composed of either native or introduced gravels (size range, 1.6-10.2 cm; Bonnell 1991), whereas in the lowermost reach, substrate is generally dominated by sand or silt. Coho salmon (Oncorhynchus kisutch) and chum salmon (O. keta) are present in all the groundwater channels. Pink salmon (O. gorbuscha), steelhead trout (O. mykiss), cutthroat trout (O. clarki), three-spine stickleback {G aster osteons aculeatus), pacific lamprey (Entosphenus tridentatus) and sculpin (cottus spp.) are present in some of the channels, but in very low abundance. Coho fry present during summer are the progeny of wild adults that spawned within the channels the previous fall and winter (Sheng et al. 1990). Adult chum also spawn in the groundwater channels (mean spawner density often exceeds 0.8 females-m\" ; Bonnell 1991) during November and December, and chum fry are abundant during March and April prior to their downstream migration. During other periods of the year, juvenile coho salmon constituted the bulk of fish biomass. The Upper Paradise and Lower Paradise channels enter the Cheakamus River at approximately 5 and 3 km upstream of the Squamish River confluence, respectively (Fig. 1.1). The excavated portion of the Upper Paradise channel is 397 m long. Below this reach, the channel joins with another excavated channel and drains into the Cheakamus River, 100 m downstream. The excavated portion of the Lower Paradise channel is 305 m long, and drains into the Cheakamus through a natural channel that is approximately 1.5 km in length. The design of this channel differs from that of the other groundwater 7 channels in that an intake from the main river was added following construction. The intake provides additional surface water flow to the channel throughout the year. As a result, temperature and turbidity are more variable than in other groundwater channels. Mashiter and Brennan channels enter the Mamquam River about 2 km upstream of the Squamish River confluence (Fig. 1.1). Brennan channel is 315 m long and drains directly into the Mamquam River. The excavated portion of Mashiter channel is 248 m long, and it enters the Mamquam River through a 1 km long natural channel. Hopedale and Barrett channels enter the Chilliwack-Vedder River approximately 10 km upstream of the Fraser River confluence (Fig. 3.1). The study reach of Hopedale channel extends 390 m downstream from the blind upstream end, while the study reach of Barrett channel extends 369 m upstream of a small weir. Both channels enter the Chilliwack-Vedder River through natural groundwater side channels. Off-channel ponds The ponds studied are located in the upper Coquitlam River below Coquitlam River Dam. The Coquitlam River is a fourth order stream that flows south from the southern Coast Mountains into the Fraser River 8 km east of New Westminister (Fig. 1.1). The Coquitlam River differs physically from the other three river systems in the following: present mean annual discharge is much lower (1.7 m3-s-1) due to a hydroelectric diversion project (Ministry of Environment 1978), flow originates mostly from rain and snow melt rather than glacial run-off, and gradient and average substrate size are considerably greater. The fish fauna is similar to that in the groundwater channels, although chum salmon are not common in the upper river. 8 Grant's Tomb Pond is located on the west side of Coquitlam River about 17 km upstream of the Fraser River (Fig. 1.1). The site was constructed in 1993, and consists of a 3300 m rearing pond and a 103 m long outlet channel. A 52 m long inlet channel was added in 1995 following the first summer of the study. Or Creek Pond is located on the east side of Coquitlam River, about 2 km downstream of Grant's Tomb Pond (Fig. 1.1). This project was completed in 1994. Or Creek Pond is 3936 m2 and the inlet and outlet channels are 68 and 112 m long respectively. The inlet and outlet channels were not included in the study. Grant's Tomb and Or Creek Ponds were created by the construction of earthen dikes within low-lying areas of the flood plain and the diversion of water from the Coquitlam River. Although groundwater seepage is present at each site, surface water diversions from the Coquitlam River also contribute to flow. Unlike the groundwater channels, the ponds contain large amounts of woody debris including whole trees and rootwads. This material either existed in the sites prior to enhancement or was added during construction. Both ponds are bordered by early serai forests and riparian canopy cover is low. The substrate of each pond is the former forest floor. The ponds range in depth from 0.2 to 2.5 m, with 70 % and 40% of the area of Grant's Tomb and Or Creek, respectively, being less than 1 m in depth. The inlet channels were designed as spawning habitat to ensure adequate fry recruitment to the ponds. 9 FIELD METHODS Periphytic algae I measured chlorophyll a densities on artificial substrates as an index of periphyton biomass (Perrin et al. 1987). For the purpose of the study, I considered differences in periphyton biomass between paired channels and ponds to be a reflection of relative differences in autotrophic productivity. Artificial substrates consisted of a 0.30m x 0.30m x 0.0075 m styrofoam sheet attached to a plexiglass plate that was bolted to a concrete block. Two blocks were placed in close proximity at a depth of about 0.4 m near the longitudinal midpoint of each groundwater channel and in the inlet channel of the two Coquitlam ponds. Sampling occurred during four six week periods: mid-July to early September 1995, mid-August to late September 1995, mid-October to early December 1995, and early March to late April 1996. New sheets were attached at the beginning of each sample period (sheets from the first two overlapping sampling periods were attached to the blocks side-by-side). Data for the October to December period were not used because the artificial substrate was damaged by the spawning activity of chum salmon. Duplicate samples (one from each plate) were collected weekly from each channel during a sample period; samples were not collected in some weeks due to logistic constraints. The paired channels from each river system were generally sampled on the same day, although different pairs were sometimes sampled during different weeks within a sample period. 10 A sample of chlorophyll a from a channel consisted of a randomly punched core from one of the styrofoam sheets stored in a seven dram plastic vial (2.7 cm diameter, 5.73 cm area). Each sample was drained of any water and silica gel was added as a desiccant. The vial was then wrapped in aluminum foil to prevent exposure to light and it was frozen within 24 hours at -4° C. Chlorophyll a densities (mg • m\") from the styrofoam samples were measured using acetone extraction and spectrophotogammetry (Strickland and Parson 1972). Temperature and discharge Water temperature was monitored weekly from June 1995 to April 1996 using minimum-maximum thermometers placed at the longitudinal midpoint of each groundwater channel and in the inlet channels of the Coquitlam River ponds. Thermometers were secured to metal posts and positioned under the rip-rap banks to reduce exposure to direct sunlight. Minimum and maximum water temperatures were recorded during each visit to a channel, which was generally weekly from July to September and bi-monthly from October to April. Temperature readings from the minimum-maximum thermometers were calibrated on each occasion with a hand-held thermometer accurate to +/- 0.1 °C. Average weekly temperatures were calculated as the means of calibrated minimum and maximum weekly temperatures. Depth and current velocity in each channel were estimated as the mean of 5-10 measurements (at 0.6 depth with a propeller-type current meter) made at uniform intervals across a fixed transect near the periphyton collection plates. Discharge for each 11 channel was estimated using the velocity-area method (Gore 1996). These measurements were repeated across a range of observed staff heights throughout the year. I developed staff-discharge relationships by performing linear regressions of staff height against discharge for each channel. Regression equations were then used to convert weekly staff gauge heights to predicted discharge (Gore 1996). Water chemistry Water samples were collected bimonthly from July to September, and monthly from October to December 1995. The paired channels from each river system were sampled on the same day, and during each sampling period, all channels were sampled within a seven day span. Only single samples were collected from each channel due to the cost of water chemistry analysis. Samples were analyzed for dissolved inorganic nitrogen (NO3 + NO2 or DIN), dissolved ortho-phosphate phosphorus (soluble reactive phosphorus or SRP), total alkalinity, and turbidity. I collected water samples at a depth of 30 cm at the longitudinal midpoint of each groundwater channel and in the inlet channel of the Coquitlam ponds. Water samples analyzed for concentrations of DIN and SRP were field-filtered through a portable, 0.45-um membrane filter with a 60 ml syringe pump into a 100 ml sterilized glass bottle (Mouldey and Ashley 1996). Samples used to measure other water quality parameters were obtained by filling 1 litre sample bottles directly from the stream after several rinses with stream water. All samples were immediately placed in coolers with freezer packs 12 and transported to Zenon Environmental Labs within 24 hours. The handling and analysis of the samples were conducted according to standard methods (APHA 1985). Benthic invertebrates I used the colonization of artificial substrates by benthic invertebrates as an index of relative biomass of invertebrates in the channels (Slaney and Ward 1993). The artificial substrate consisted of a cylindrical, open-sided plastic basket (22 cm in diameter, 13 cm deep, 0.04 m in area and 0.05 m in volume) filled with washed, screened gravel (4 cm diameter average). Six baskets were placed in two groups of three upstream and downstream of the longitudinal midpoint of each groundwater channel in areas where current velocity was relatively fast and depth averaged about 0.30 m (in the Coquitlam sites, baskets were in the inlet channels of the ponds). Baskets were initially placed during mid-August 1995, and then replaced with clean gravel during late September 1995, and again during mid-September 1996. During each period, baskets were sampled after being immersed in a channel for six weeks (August 15 to September 31, 1995). To minimize disturbance of the invertebrates as the baskets were removed from the stream for sampling, a plastic pail was placed downstream of each basket and dipped upstream as the basket was placed inside. Individual stones were scrubbed by hand while immersed in the water. The water was then filtered through a 0.177 mm brass sieve to separate the invertebrates. This process was repeated three times for each sample with additional water being added to the pail and agitated each time. The sample was then 13 rinsed from the sieve into a 250 ml glass jar containing an 80 % isopropyl alcohol solution. Three of the six samples collected for each channel were randomly selected for analysis. Invertebrates were sorted by hand at 10 x magnification, identified to order and counted. Samples were then dried at 55°C for 24 hours, and dry weight was measured using an electronic balance accurate to 0.0001 g. An index of invertebrate biomass in each site was calculated as the mean dry weight of the three analyzed samples divided by the gravel-filled volume of a colonization basket (0.05 m3). Juvenile coho salmon At least 95% of juvenile fish biomass in the channels consisted of young-of-the-year coho (ages were estimated from length frequency histograms). To quantify differences in fish production among channels and ponds, I estimated the densities and mean sizes of coho fry at each site during early September and mid March in both 1995-96 and 1996-97. In 1995-96,1 also collected size data in mid July, mid November and late January. On all occasions, fish were anesthetized with MS-222, counted, weighed (nearest 0.1 g), and measured (total length to the nearest mm). During the second year of the study, a random sub-sample of coho fry from a channel reach or pond were weighed and measured if total capture exceeded 200 fish I estimated coho population abundance in each groundwater channel by conducting electrofishing removals from permanent sampling reaches. For each channel, three 14 reaches were selected systematically: the channel was divided into three reaches of equal length, and a 20 m long sample reach was located at the midpoint of each reach. Wetted surface area and volume of each reach were calculated on each sampling occasion from average wetted width and depth across three transects. Total area and volume of each channel were also estimated using similar methods. Depending on the area of each channel, sampling intensity ranged from 15 to 24%. The exceptions were Barrett and Hopedale channels in the Vedder River in September 1995 when both channels were sampled in their entirety (each channel was divided into several closed sections). Three pass removals were conducted at each site by a three-person crew equipped with a Smith Root® model 12B back-pack electro-shocker in 1995-96. In 1996-97 a fourth pass was added and each pass was expanded to include a beach seine sweep prior to electrofishing. Sample sites were completely enclosed using 10 mm stretched-mesh seines before electrofishing to ensure population closure. Nets were installed as quickly as possible to minimize disturbance to fish. Each pass consisted of one downstream sweep with a beach seine (1996-97 only) followed by a thorough upstream electrofishing search. To standardize effort among passes as much as possible, the time spent electrofishing was recorded for each pass. In July and November 1995 and January 1996, only one electrofishing pass was conducted in each site. On all other occasions, a population estimate for each sample site was computed using a maximum likelihood algorithm (Warren 1994). 15 Physical habitat was examined in each groundwater channel during August 1996. Mean depths and current velocities were measured at five points along each of three width transects located at the middle, top, and bottom of each 20 m sample reach. To quantify the relative amount of cover provided by the rip-rap banks (mid-channel cover was not present in any of the reaches), a 20 m tape was laid out along each bank and the following three measurements were made at every 2 m interval: 1) water depth at the bank, 2) width of the rip-rap bank, and 3) horizontal depth to which a metre stick could penetrate the rip-rap bank (eg. if a rip-rap boulder was present at 4.0 m, the measurement would equal zero). Additional data collected for each reach included an estimate of substrate composition by size class (i.e, silt, sand, gravel, cobble), and the proportion of each bank for which overhead cover was present. Because the channels were morphologically stable, I considered the results of the survey to represent habitat conditions throughout both 1995-96 and 1996-97. However, current velocities and water depths did vary somewhat with seasonal discharge. For this reason, 'summer' values for each variable were measured in August when flows were relatively low, and 'winter' values were measured in November when flows were relatively high. Electrofishing was not possible in the Grant's Tomb and Or Creek ponds (Coquitlam River) due to excessive water depths, and I estimated coho populations in the ponds by mark-recapture using Gee® wire mesh minnow traps (42 cm long x 22 cm wide, with 5 mm mesh and 15 mm openings in funnels at each end). On each occasion, 100 traps 16 baited with two grams of preserved salmon roe were allowed to soak 24 hours in each pond. Although the same capture method was used during both the marking and recapture periods, the number offish initially marked (mean = 521, range = 219-967) and the recapture rate (mean = 0.21, range = 0.11 to 0.48) were relatively high on all sampling occasions. Thus, it is unlikely that population estimates were seriously biased by differing capture probabilities between marked and unmarked fish due to 'trap-happy ' or 'trap-shy' behaviour (Ricker 1975). Fish captured during the first set were given a small caudal fin clip. After three to four days, fish were again trapped and examined for marks. Population size was estimated using the modified Petersen mark-recapture equation Chapman (1951): M - Marked population C - Total captures on second occasion R - Recaptures Population = (M+1)(C+1) / (R+l) Var(Population) = N2(C-R) / (C+l)(R+2) The wetted area of each pond was estimated from a closed traverse survey of its perimeter using a hand compass and surveyor's chain. I calculated wetted volume as the product of wetted area and mean depth; the latter value was estimated from measurements of depth across transects of the longtitudal axis of each pond. 17 Figure 1.1. Locations of the four study rivers within the Fraser and Squamish River Watersheds in southern coastal British Columbia. The solid circles show the location of the artificial groundwater channels and ponds (Coquitlam) selected for paired observation. 18 CHAPTER 2: SUMMER ECOLOGY INTRODUCTION Primary production is thought to be an important 'bottom-up' control of the abundance and growth of fish in streams (Bilby and Bisson 1992; McFadden and Cooper 1962). Periphytic algae and aquatic invertebrates provide the principal trophic pathway by which energy from autochthonous sources reaches insectivorous fish species (Peterson et al. 1993; Power 1993), and positive correlations between fish size or abundance and standing crop of invertebrates (Bolby and Roff 1986; Warren et al. 1964; Murphy et al. 1981) or periphyton biomass (Thedinga et al. 1989) are commonly observed. Autotrophic productivity is often limited by nutrient availability in small temperate streams (Hart and Robinson 1990; Peterson et al. 1983; Stockner and Shortreed 1978). There is also evidence of positive relationships between nutrient availability and the mean size (Gibson and Haedrich 1988) or standing stock of fish (Hoyer and Canfield 1991; McFadden and Cooper 1962). While fish growth and abundance may be influenced by primary production, both trophic interactions (Abrams 1993) and physical habitat limitations (Bjornn and Reiser 1991; Mason 1976, Nickelson et al. 1992) may also be important. In two whole-stream fertilization trials (Deegan and Peterson 1992; Slaney and Ward 1993) that included spatial and temporal controls, the response of fish to increased biomass at lower trophic levels was evident but difficult to predict. Deegan and Peterson (1992) found that size and growth of Arctic grayling Thymallus arcticus increased in response to increased 19 abundance of periphyton and benthic invertebrates following stream fertilization, but increases in standing stock were not observed. Slaney and Ward (1993) also found an increase in the biomass of periphyton, invertebrates and salmonids at the end of summer following nutrient enrichment, but not an increase in fish abundance. Case studies in small streams suggest that the mean size of coho fry in late summer is both food-limited and density-dependent (Bilby and Bisson 1987; Mason 1976; Scrivener and Andersen 1984). However, a relationship between fish density and food abundance is less evident (Bilby and Bisson 1992; Mason 1976; Scrivener and Andersen 1984; but also see Thedinga et al. 1989). Investigations of the effect of primary production on juvenile coho have been conducted mainly in small streams, yet coho are known to utilize a diverse range of habitats for summer rearing. This may have important implications with regard to production. For instance, higher rates of coho survival and/or growth have been noted for estuaries (Koski and Kirchhofer 1984; Tschaplinski 1982) and lakes (Irvine and Ward 1989; Quinn and Peterson 1996) compared to nearby streams. Groundwater-fed off-channel habitat may also provide an advantageous rearing environment. For example, temperature and discharge regimes tend to be more stable (Bonnell 1991; Cunjak and Power 1986) and nutrient concentrations can be relatively high compared to run-off streams (Ford and Naiman 1989; Hynes 1970; Labaugh et al. 1995). Higher growth rates (Brown and Hartman 1988; Peterson 1982; Swales and Levings 1989), densities (Swales et al. 1986) and survival (Brown and Hartman 1988; Swales and Levings 1989) have been reported for juvenile coho in off-channel habitat compared to adjacent streams. 20 While salmonids are often excluded from naturally occurring off-channel habitat during summer months because of low water levels, artificial groundwater channels and ponds may remain accessible year-round (Lister and Finnigan 1997; Sheng et al. 1990). However, despite the apparent importance of summer growing conditions to overwinter survival and smolt production in streams (Scrivener and Andersen 1984), the influence of primary production on the growth and abundance of coho fry in groundwater-fed off-channel habitat during the summer is not known. I address this question by comparing the mean size and abundance of juvenile coho in late summer from three pairs of groundwater-fed channels and one pair of ponds where periphtyon standing crops differed. My prediction was that differences in periphyton biomass between paired sites would be reflected by differences in nutrient concentrations, benthic invertebrate biomass and fish size and abundance. DATA ANALYSIS Analyses in this chapter are based on data collected from three pairs of groundwater channels and one pair of ponds during 1995-96 and 1996-97. Complete descriptions of the field and lab methods used are given in Chapter 1. Initially, I selected pairs of channels and ponds on the basis of visible differences in standing crops of periphytic algae. During the course of the study, I then compared relative levels of periphyton biomass between pairs using chlorophyll a density on artificial substrate as an index. Chlorophyll a is considered a reliable index of periphyton 21 growth and biomass (Bothwell 1988; Perrin et al. 1987), and of relative productivity among aquatic systems (Hynes 1970; Rieman and Myers 1992). Because periphyton growing on artificial substrates became unstable as biomass increased and often broke away at peak density, chlorophyll a densities were highly variable both among replicate substrates and over time. Therefore, I describe relative chlorophyll a abundance for each channel as the average of seasonal peak accruals (July 1995, September 1995, March 1996). In order to assess both the influence of water quality on periphyton biomass and the effect of periphyton biomass on benthic invertebrate and fish populations, the following approaches were used. The first approach was to pool all eight sites and look for relationships between peak chlorophyll a and measures of invertebrate and fish abundance in late summer using scatterplots and simple linear regression (1995 only). Estimates of the mean weight, biomass and density of coho fry at each site were calculated from the September removal (channels) or mark-recapture (ponds) data. The number offish sampled in each site ranged from 303 to 1309. Sample sizes for estimates of benthic invertebrate biomass, coho density and biomass and water quality data appear in the tables and figures at the end of the chapter. Mean values for total alkalinity, phosphorus, nitrogen, and stream temperature were estimated from biweekly data for June to October, 1995. Average current velocities within channels and within invertebrate colonization basket sites were estimated as the means of velocity measurements made in August 1996. For all significant regressions, Cook's distances (Cook 1977) were plotted against the estimates offish density to check for outliers or 22 data points with large leverage. Prior to all statistical testing, data were checked for normality (Kolmogorov-Smirnov test, or scatterplots of variances against means if sample size was small) and for homogeneity of variances (probability plots), and then log 10 transformed when necessary. The second approach was to use t-tests to compare individual pairs in terms of benthic invertebrate biomass (1995) and coho size, biomass and abundance (1995 and 1996). For these analyses, the member of each channel or pond pair with relatively high chlorophyll a densities was considered the high productivity treatment (HP) and the member with low chlorophyll a, the low productivity treatment (LP). Because chlorophyll a data were not collected in 1996,1 assumed that differences in chlorophyll a between paired sites would be similar for each year. Relative differences in chlorophyll a between channels were generally similar among the three sample periods during the initial year, and there was no evidence to suggest significant differences between years in light levels, discharge or temperature. Two-way ANOVA without replication (Sokal and Rohlf 1981) was used to test for within-pair differences in concentrations of phosphorus, nitrogen and alkalinity. For these analyses, productivity was considered as a treatment effect, and sample date as a source of variation. In all cases where pairs were tested independently, observed significance levels were adjusted using the bonferroni method to reflect the fact that multiple comparisons were made (Zar 1974). I compared the relationship between coho size distributions and chlorophyll a densities within pairs by visually comparing histograms of fish length data for September 23 1995 and 1996. Previous studies have suggested that increased competition for food due to greater fish density or lesser food abundance, may lead to a shift in juvenile salmonid populations from normal length distributions to skewed distributions consisting of large numbers of small fish and small numbers of larger 'dominant' fish (Keeley 1998; Mason 1976). Analysis of covariance was used to compare fish 'condition' between paired sites (Cone 1989). Tests were performed with logio transformed weight as the dependent variable, logio transformed length as the covariate, and channel productivity (high or low) as the treatment effect. Comparisons of fish condition were only made for pairs that had length-weight regressions with similar slopes. For each of these pairs, a significant difference in the intercepts of the length-weight regressions indicated a productivity effect (high or low) on condition. RESULTS Periphyton accrual Large and consistent differences in peak periphyton accrual were evident between the paired sites in three of the rivers during each of the three sample periods (Fig. 2.1). Depending on the time of year, peak chlorophyll a density was 2 to 4 times greater in Hopedale channel compared to Barrett channel, 2 to 6 times greater in Mashiter channel compared to Brennan channel, and 3 to 15 times greater in Or Creek pond inlet compared to Grant's Tomb pond inlet. The rate of chlorophyll a accrual was similar to that for peak density within each pair; chlorophyll a accumulated more rapidly at the site where peak density was greater in each period (Fig. 2.1). Differences in peak chlorophyll a density were smaller and less consistent between the Cheakamus River pair. Chlorophyll a 24 density was 1.5 and 2 times greater in Upper Paradise channel in September and March, respectively. However, chlorophyll a density appeared greater in Lower Paradise channel during August. Conflicting results for this pair could be a result of measurement error in either channel; periphyton samples collected from Upper Paradise channel in the third week of July (when peak accrual likely occurred) were destroyed during transport, and peak density observed in the Lower Paradise Channel in July was unusually high compared to all other measurements. Based on differences in peak chlorophyll a, each site within a river pair was categorized as either high productivity (HP) or low productivity (LP) (Fig. 2.1). I observed no consistent seasonal trend in peak density of chlorophyll a between each pair (Fig. 2.1). Chlorophyll a density was highest after five weeks in July in half of the channels, highest in March in the other half, and lowest in most channels in September. Chlorophyll a also peaked sooner in July compared to both September and March. This was particularly true for channels where peak density was consistently greater than 10 pgm\"2 (Hopedale, Mashiter, Upper and Lower Paradise, and Ore Creek). In July peak density in these channels occurred after four weeks on average, while in September and March, peak density was often not observed after 5 or 6 weeks. This suggests that periphyton accrual rates were highest during the summer, somewhat lower during the spring, and lowest in early fall. While I did not identify algae to taxonomic unit, algal biomass appeared to be composed mostly of diatoms in the majority of the channels throughout the year. 25 However in the HP channels of the Vedder and Mamquam pairs where peaks in chlorophyll a density were highest dense mats of chlorophytes were present throughout each channel from early summer until late fall when they were dislodged by the spawning activity of chum salmon. A large portion of the Vedder HP channel was also colonized by dense growths of water cress (Nasturtium officinale) during the summer. Filamentous algae was also visible (but to a lesser extent) in shallow portions of the Coquitlam HP pond. Lower light levels may have prevented chlorophyte blooms in the Cheakamus HP channel; chlorophyll a densities there were similar compared to the other HP sites. Water quality Differences in mean summer temperature (July-October) between paired channels or ponds were significant (bonferroni adjusted t-tests, p < 0.001 for each case; Table 2.1), but not consistent with relative differences in chlorophyll a density. Mean summer temperature was highest in the HP channel of the Vedder pair and the HP pond of the Coquitlam pair, and highest in the LP channel of the Mamquam and Cheakamus pairs. When compared across all sites, peak chlorophyll a density was not correlated with mean summer temperature (logged values, r = 0.18, n = 8, p = 0.67). Average weekly temperatures were less variable in the channels than in the ponds (Fig. 2.2). Absolute temperature differences observed in the groundwater channels ranged from 5 to 12°C compared to 14 to 16°C in the two Coquitlam ponds and the Cheakamus LP channel where flow was augmented by surface water diversions. With the exception of the Vedder HP channel, mean weekly temperatures in the groundwater 26 channels were below 12°C during the summer whereas temperatures in the ponds and the Cheakamus LP channel were generally greater than 12°C. Weekly variation in temperature was roughly similar for the channels and ponds (Fig. 2.2). By contrast, temperatures in the groundwater channels were generally 3 to 5 °C warmer in January and 3 to 6 0 C cooler in August compared to their parent rivers (Environment Canada 1977). Mean current velocities in the groundwater channels and the inlet channels of the ponds ranged from 0.02 to 0.30 m-s\"1 (Table 2.1). Current velocity was similar for three of the pairs (Mamquam, Cheakamus, and Coquitlam, bonferroni adjusted t-tests, logged values, p > 0.013 for each case), but significantly higher for the HP channel of the Vedder River pair (p < 0.0001). Differences in peak chlorophyll a were not correlated with velocity (logged values, r = 0.21, n = 8, p = 0.62). Turbidity was very low in all sites (< 1.0 NTU) except the Cheakamus LP channel where levels were moderately elevated (3.0-13.5 NTU) by glacially silted water from the Cheakamus River. Peak chlorophyll a density was weakly correlated with mean total alkalinity (logged values, r = 0.62, n = 8, p = 0.10), and SRP (r = 0.70, p = 0.06), but not with DIN (r = 0.08, p = 0.86). Total alkalinity was generally higher in the HP channel or pond of each pair (Fig. 2.3). However, differences in concentrations of SRP and DIN between pairs were less consistent (Figs. 2.4, 2.5, respectively). Interactions between site and sample date were not significant for alkalinity (ANOVA, p = 0.11), SRP (p = 0.55), or DIN (p = 0.95). Therefore, two-way ANOVA without replication (bonferroni adjusted) was used 27 to test for within-pair differences in the three water quality parameters (Sokal and Rohlf 1981). Total alkalinity was significantly greater in the HP channel compared to the LP channel of the Mamquam (p = 0.0001), and Cheakamus pairs (p = 0.004), but alkalinity was not greater in the HP pond of the Vedder (p = 0.04) or Coquitlam pairs (p = 0.12). SRP concentrations were significantly higher in the HP channel only for the Mamquam (p = 0.007; Table 2.2) and Coquitlam pairs (p = 0.011), but not for the Cheakamus (p = 0.18) or Vedder pairs (p = 0.30). DIN concentrations were significantly higher in the HP channel for the Mamquam (p < 0.0001; Table 2.2) and Cheakamus pairs (p = 0.011), but were similar between the Coquitlam pair (p = 0.31) and higher for the LP channel of the Vedder pair (p = 0.003). Because concentrations of SRP and DIN were not consistently higher in the HP channels, I also considered the importance of nitrogen to phosphorus ratios (DIN: SRP). I expected periphyton growth to be limited by phosphorus at N:P ratios (in moles) greater than 30:1, nitrogen at ratios less than 5:1, and both nutrients at ratios between the two values (Murphy and Meehan 1991; OECD 1982). Mean atomic N:P ratios were relatively similar between paired sites (Table 2.3). Periphyton biomass in the Coquitlam ponds was likely limited by phosphorus; N:P ratios were considerably higher compared to those in the groundwater channels (Table 2.3), and modestly elevated SRP concentrations in the HP pond compared to its LP pair (means = 2.5, 1.3 [ig-m3, respectively) coincided with much higher chlorophyll a density. Periphyton biomass in the groundwater channels may have been limited by both phosphorus and nitrogen (Table 2.3). In the Cheakamus pair, higher chlorophyll a densities in the HP channel could be 28 attributed to higher DIN concentrations. In the Mamquam pair, the HP channel had higher concentrations of both DIN and SRP. However, for the Vedder groundwater channel pair, SRP concentrations were similar between channels and DIN levels were higher in the LP channel. Benthic invertebrates Although Diptera were generally most numerically abundant followed by Ephemeroptera and Plecoptera, there were no apparent trends in the taxonomic composition of benthos samples among river pairs or among HP and LP sites (Fig. 2.6). For example, Ephemeroptera, while abundant in the majority of channels, were entirely absent from samples collected from the Vedder LP and the Mamquam HP channels. In the latter channel, annelids represented 90 and 83% of benthos numbers and biomass, respectively, in the colonization baskets. However, the proportional abundance of taxa was similar within samples for each channel. Differences in the relative proportion of individual taxa among sites were not correlated with differences in SRP or DIN concentrations, current velocity, water temperature, or substrate composition (logged values, n = 8, r < 0.5, p > 0.05 for all cases). When pairs were compared individually, biomass of invertebrates in the gravel baskets (all taxa combined) was not consistently greater for the channels where chlorophyll a densities were higher. Estimated mean biomass was 4.8 fold greater in the HP channel of the Vedder pair (bonferroni adjusted t-tests, logged values, p < 0.007; Fig. 2.7) and possibly 2.2 fold greater in the HP channel of the Mamquam pair, but not 29 significantly so (p = 0.15). Estimated biomass was similar between the Coquitlam pair (p = 0.64) and higher in the LP channel of the Cheakamus pair, but this difference was also not significant (p = 0.02). A weak correlation existed between logarithmic mean biomass of invertebrates and logarithmic mean peak chlorophyll a density across all channels and ponds (r = 0.70, n = 8, p = 0.12; Fig. 2.8). However, current velocity may have been a confounding factor. Invertebrate biomass in the baskets was strongly correlated with current velocity in the immediate vicinity (r = 0.98, n = 6, p < 0.001; Fig. 2.9), but not with average velocity in the channels (r = 0.71, p = 0.12). In considering the effect of current velocity on invertebrate biomass, I excluded data for the two Coquitlam sites. Although the baskets were placed in the inlet channels rather than in the ponds, velocities were not measured in the immediate vicinity of the baskets. The inlet channels of the Coquitlam ponds had the highest mean water velocities, and were clear outliers when logarithmic mean biomass of invertebrates versus velocity was plotted for each channel (Fig. 2.9). Invertebrate biomass was not correlated with other factors including water temperature (r = 0.27, p = 0.52), alkalinity (r = 0.30, p = 0.35), and fish density (r = 0.02, p = 0.84; logged values, n = 8 for all cases). Coho salmon There was a significant correlation between logged values for the mean weight of fish in September and average peak chlorophyll a density in the groundwater channels in 1995 (r = 0.78, n = 6, p = 0.02; Fig. 2.10). A positive, although not significant relationship was also found between fish size and invertebrate biomass (r = 0.73, p = 0.09; Fig. 2.11). Mean weight was not correlated with summer stream temperature (r = 30 0.25, p = 0.64), nor did it appear to be associated with fish density in either 1995 or 1996 (r = 0.37, p = 0.47; r = 0.56, p = 0.25, respectively). The Coquitlam ponds were not included in these correlation analyses for two reasons: first, coho populations from the ponds shared the same rearing environment for the majority of the summer growing period (I stocked the LP pond with coho fry from the HP pond six weeks prior to the September 1995 sampling period because coho fry densities in the LP pond were extremely low in 1995; the subsequent construction of an inlet spawning channel facilitated natural seeding of the pond in 1996), and second, a large increase in mean size offish from July to September in the LP pond (see Fig. 3.4 in Chapter 3) probably reflected the fact that stocked fish gained access to a largely unexploited and vulnerable aquatic invertebrate community (Werner 1986). The Coquitlam LP pond was a clear outlier when included in scatter-plots offish weight vs. chlorophyll a density (Fig. 2.10.) and fish weight vs. benthic invertebrate biomass (Fig. 2.11). The mean weights of late summer fry in the high chlorophyll a (HP) sites were 1 to 2.5 fold greater than those in the LP sites in September 1995 and 1996 (bonferroni adjusted t-tests, p < 0.001 for all cases; Fig. 2.12). The one exception was the Coquitlam ponds in 1995, where fish were significantly larger in the LP pond (p < 0.0001). Again, this result was likely influenced by the transfer of fish from the HP to the LP pond. By comparison, fish from the same cohort were 18% larger in weight in the HP pond by the end of winter, (see Results in Chapter 3), while fish from the subsequent cohort were 42% heavier in the HP pond in September 1996. Fish were larger at each site in 1996 compared to 1995 (t-tests, p < 0.013 for each case) except for the Vedder HP channel and 31 the Coquitlam LP pond (p > 0.013 for each case). Greater mean weights in the Mamquam and Cheakamus LP channels and the Coquitlam HP pond in 1996 compared to 1995 were associated with lower estimated fish densities (Fig. 2.12). However, greater mean weights in 1996 compared to 1995 in the other sites were associated with either similar (Mamquam HP), or higher fish densities (Vedder LP, Cheakamus HP; Fig. 2.12). The relationship between periphyton biomass and fish growth during the summer was also evident by the difference in length frequency distributions between HP and LP members of each pair in September (Fig. 2.13). Coho populations in the less productive sites tended to consist of a large number of small fish and a few larger fish, whereas length distributions in the more productive sites were more normalized. Also, in 1996 length distributions for each site were less skewed to the right compared to the previous year. This was associated with larger fish size in most sites and with lower fish density in several of the sites compared to 1995. Within the Coquitlam sites in 1995, the distribution of lengths in the LP pond was less skewed compared to that in the HP pond, but this was consistent with results from other pairs because food abundance was likely greater in the previously Ashless LP site. Size distributions were skewed in both ponds in June (not shown), but the distribution in the LP pond was nearly normal by early September whereas the distribution in the HP pond remained highly skewed (Fig. 2.13). Fish condition was also related to periphyton biomass. Length-weight regressions (logio transformed values) were highly significant for each site (p < 0.001, r2 > 0.87 for all cases). The slopes of the regressions for weight on total length (logged values) 32 differed between the HP and LP sites for the Vedder and Mamquam pairs in both 1995 and 1996 (ANCOVA, p < 0.01 for each case), but the differences in slopes within pairs and between years in each site were not consistent (Fig. 2.14). Regression slopes were similar between the HP and LP sites of the Cheakamus and Coquitlam pairs in both years (ANCOVA, p > 0.05 for each case; Fig. 2.14). In 1995, length-weight regressions for the HP site of both the Cheakamus and Coquitlam pairs had greater intercepts than the corresponding LP site (ANCOVA, p < 0.001 and p = 0.04, respectively; Fig. 2.14). In 1996, the length-weight regression for the HP pond of the Coquitlam pair had a greater intercept than the LP pond (ANCOVA, p < 0.001; Fig. 2.14), but the intercepts of the regressions were similar between the Cheakamus channels (ANCOVA, p = 0.96). This indicated that fish were heavier for their length in the site where periphyton biomass was higher for the Coquitlam pair in both years and the Cheakamus pair in 1995, but not in 1996. Although late summer abundance of coho fry varied substantially among sites in 1995 (Fig. 2.12), neither fish density (r = 0.57, p = 0.24) nor biomass (r = 0.52, p = 0.19) were correlated with peak chlorophyll a density (logged mean values, n = 8 for each case). When pairs were compared individually, differences in mean density between paired HP and LP groundwater channels were also not significant in 1995 or 1996 (bonferroni adjusted t-tests, logged values, p > 0.013 for each case) except for the Cheakamus pair in 1996 when fish density was greater in the HP channel (p < 0.013). In 1995, estimated mean fry biomass was about 60% greater in the Mamquam and Cheakamus HP channels compared to their respective LP pairs, but similar between the 33 Vedder channels (Fig. 2.12). In 1996, mean biomass was greater in the HP member of each pair by 17 to 267%. However, none of these differences were statistically significant (logged values, t-tests, p > 0.013 for all cases). The Cheakamus pair in 1996 was again the exception (p < 0.013). Fry density and biomass were both similar between the Coquitlam HP and LP ponds in both years, but values were much lower and less variable between years (ranges = 0.5 to 0.8 fish-m\"2 and 1.1 to 1.7 g-m\"2, respectively; Fig. 2.12) compared to those in the groundwater channels (ranges = 1.2 to 5.9 fish-m\"2 and 2.2 to 13.4 g-m\"2, respectively). Neither fish density or biomass were correlated with benthic invertebrate biomass, summer stream temperature, discharge or velocity in 1995 (logged mean values, r < 0.60, n = 8, p > 0.05 for all cases). DISCUSSION Periphyton accrual Peak chlorophyll a density was generally greater in the HP member of each pair throughout the year. Differing periphyton biomass between paired sites was not likely influenced by water velocity or turbidity. Current speed was relatively low in all the channels and pond inlets and similar within three of the four pairs. Turbidity was also low in the majority of sites. Although not measured quantitatively, variation in light penetration among channels and ponds was also an unlikely contributor to differences in chlorophyll a density within pairs; aspect was similar between paired sites and only the Cheakamus channels had well developed overhead canopies. 34 Peak chlorophyll a density was greater in summer compared to early fall and late winter in the Coquitlam ponds and in two of the six groundwater channels where temperature varied substantially during the year. Perrin et al. (1987) also observed greater peak chlorophyll a density during the summer compared to the fall in a surface water stream. In the four groundwater channels where temperature was relatively stable, peak chlorophyll a density was either similar among seasons or highest in late winter. However, the rate of chlorophyll a density in these sites appeared to be somewhat higher in summer compared to other periods. Nevertheless, the relatively high amount of periphyton accrual during the winter in these channels was unexpected considering the importance of light to autotrophic productivity (Hynes 1970; Rand et al. 1992). Bilby and Bisson (1992) found that winter gross primary production in two small headwater streams occurred at only 15% of the rate measured during the summer. High periphyton accrual in the groundwater channels in late winter was likely a result of warm water temperatures, increased levels of DIN and SRP compared to summer months, and the absence of scouring flows. In seven of the eight sites, peak chlorophyll a density was lowest in the September sampling period despite similar or relatively high levels of nutrients, light and temperature compared to other periods. A possible explanation is control of periphyton by invertebrate grazers during this period. Invertebrates were observed grazing on the artificial substrates used to measure periphtyon accrual in the channels. Numerous researchers have observed grazer control of periphyton in experimental channels (Hart and Robinson 1990; Rosemond et al. 1993) and streams (Bothwell et al. 1992; Peterson 35 et al. 1993; Power 1990). Hart and Robinson (1990) argued that the strong top-down effect of grazers that occurred in their study was influenced by stable discharge because moderate flows allowed invertebrate populations to reach maximum densities. It is probable that populations in the groundwater channels were also near capacity; flow and temperature in the channels were stable and average substrate sizes were comparable to those reported to have supported maximum invertebrate densities in other streams (2.5 to 3.5 cm, Rabeni and Minshall 1977; 2.4 cm, Williams and Mundie 1978). Water quality Nutrient availability was an important factor in the production of periphyton in the off-channel sites. Total alkalinity, which has been shown to be a reliable index of nutrient abundance (Brylinsky 1980), was consistently higher in the channel of each pair where chlorophyll a density was greater. However, consistent differences in the concentration of phosphorus and nitrogen between each pair were less evident. Correlations between phosphorus and chlorophyll a concentrations have been noted in temperate streams of diverse order and geographic location (Van Nieuwenhuyse and Jones 1995), and corroborated by numerous phosphorus enrichment experiments (Hart and Robinson 1990; Slaney and Ward 1993; Peterson et al. 1983). Lack of a strong chlorophyll a-phosphorus relationship in this study may reflect the relative abundance of SRP in the groundwater channels. SRP concentrations in the channels ranged from 5 to 18 jig-1compared to less than 1 ugT1 for typical coastal B.C. streams (Paul et al. 1996; Slaney and Ward 1993; Toth et al. 1996). In contrast to phosphorus, nitrogen (DIN) concentrations in the channels (45 to 184 ug-i\"1) were similar to those observed in Pacific 36 coastal streams (Bilby et al. 1996; Perrin et al. 1997; Stockner and Shortreed 1978). Result from streams (Murphy and Meehan 1991; Stockner and Shortreed 1978) and lakes (OECD 1982) suggest that algal growth is usually limited by phosphorus concentrations at N:P ratios (in moles) greater than 30:1, or nitrogen concentrations at ratios less than 5:1. Phosphorus control of chlorophyll a density may have been less pronounced in the groundwater channels because N:P ratios were relatively low (8:1 to 34:1) compared to those in surface water streams (e.g., 20:1 to 60:1, Johnston et al. 1990; > 100:1, Hart and Robinson 1990). Relatively high phosphorus concentrations in the channels were likely a result of the mineral-richness of groundwater (Ford and Naiman 1989; Hynes 1970; Labaugh et al. 1995). Input of chum salmon carcasses (Bilby et al. 1996; Wipfli et al. 1998) during the fall and winter may have also contributed to elevated SRP levels. However, carcasses were likely less important to year-round nutrient abundance in the channels compared to groundwater. In both relatively eutrophic tributary streams of Lake Ontario (Rand et al. 1992) and oligitrophic headwater streams in the Pacific Northwest (Bilby et al. 1996), salmon carcasses did not contribute substantially to annual nutrient budgets. Among the groundwater channels, higher DIN concentrations may have contributed to greater periphyton biomass in the HP channel of the Cheakamus and Mamquam pairs. However, peak chlorophyll a density was also several times higher in the HP channel of the Vedder pair despite similar SRP concentrations between channels and greater DIN 37 concentrations in the LP channel. Higher total alkalinity in the HP channel may reflect a difference in the abundance of some other nutrient or mineral. Benthic invertebrates Comparisons of invertebrate biomass among sites may have been confounded by limited replication (variation among samples in each channel was often high) or the sampling technique. For example, greater benthos biomass in the colonization baskets from the HP channels of the Vedder and Mamquam pairs may have been a result of higher current velocity rather than greater periphyton biomass (Table 2.1). Among all the channels, invertebrate biomass and velocity were not significantly correlated, but there was a highly significant correlation between invertebrate biomass and velocity in the immediate vicinity of the colonization baskets. Higher velocities in some locations may have aided invertebrates in colonizing the artificial substrate. Alternatively, small-scale variation in current velocity may have had a strong effect on invertebrate distribution within the groundwater channels. The lack of a correlation between chlorophyll a density and current velocity argues against the possibility that greater invertebrate biomass at the high velocity sites was the result of localized food abundance. However, greater biomass of invertebrates in the higher velocity sites could have been a result of either habitat selectivity or higher food delivery rates; occupancy of specific ranges of water velocity is common in some species (Hynes 1970; Rosenfeld 1997), and an apparent preference for high-velocity microsites has been noted for certain taxa (Kennedy 1967; Kimble and Wesche 1975). 38 Among the pairs, biomass of benthic invertebrates in the gravel baskets was significantly higher in the HP channel in only one of four cases (Vedder). While a lack of a significant difference in invertebrate biomass within the Mamquam pair may have been the result of low replication and relatively high variance among samples, invertebrate biomass was not greater in gravel baskets from the HP sites of the Coquitlam or the Cheakamus pairs. For the Cheakamus pair, differences in chlorophyll a densities were less pronounced between channels, and invertebrate biomass may have been higher in the channel with lower chlorophyll a densities. Between the Coquitlam ponds, invertebrate biomass in the inlet channels was similar despite chlorophyll a densities being 3 to 15 times higher in the HP site. The lack of a positive relationship between benthos and chlorophyll a density in these pairs contradicted results from other studies (e.g., Hart and Robinson 1990; Mundie et al. 1991; Murphy et al. 1981; Peterson et al. 1985; Slaney and Ward 1993). Besides nutrients and current velocity, invertebrate abundance in lotic environments can also be affected by temperature (Hynes 1970) and fish density (Power 1990). Temperature was significantly higher, and fish density significantly lower, in the Cheakamus LP channel compared to its HP pair. These factors may have led to better growing conditions or a lower predation risk, respectively, thereby resulting in a higher invertebrate standing crop despite lower autotrophic productivity. Current velocity and temperature were both similar between the Coquitlam pair. However, prior to the introduction of coho fry in July, fish biomass was very low in the Coquitlam LP pond in 1995. Relatively high invertebrate biomass in the LP pond during the summer may have been influenced by low predation prior to July. By September, fish that had been taken from the HP pond and placed in the LP pond were larger than 39 those that remained in the HP pond. In contrast, invertebrates in the LP pond in 1996 had been subject to coho predation for 14 months. In that year, coho were significantly smaller in the LP pond by September. The rapid summer growth of fish in the LP pond in 1995 (but not in 1996) suggests the presence of a largely unexploited aquatic invertebrate community. It has been shown that growth rate and ultimate size often increase when fish are transplanted to a richer environment (Werner 1986). The composition of the benthic community also differed in some of the HP channels. Acoelomates (flatworms) and Nematodes (roundworms), while not present in baskets from the two LP channels, represented 33 and 83% of benthos biomass in baskets from the Vedder and Mamquam HP channels, respectively. Acoelomates and Nematodes are considered to be tolerant of eutrophication compared to more typical stream taxa such as Plecoptera, Emphemeroptera and Diptera (Hynes 1970; Matthews et al. 1991). A more divergent invertebrate community in the HP channels of the Vedder and Mamquam pairs may have also been influenced by differences in the physical structure of theie algae and aquatic plant communities compared to other sites. Whereas algal biomass was limited mostly to diatoms in the majority of sites, algal biomass in the Mamquam and Vedder HP sites also included thick mats of chlorophtyes, and in the case of the Vedder HP channel, dense growth of a macrophyte (watercress). Stream invertebrates are known to readily colonize filamentous algae and stream macrophytes (Berg 1950; Hynes 1970), and high densities of some taxa have been previously associated with water cress in groundwater channels (Sheng et al. 1990). 40 Finally, top-down control of invertebrate biomass by fish provides an alternative hypothesis to explain the lack of a clear relationship between periphyton and invertebrates in this study. Some studies have shown that fish are capable of significantly altering invertebrate community structure by selectively preying on specific macroinvertebrate taxa (Bechara and Moreau 1992; Feltmate and Williams 1989; Power 1990). In studies of three-tiered trophic food webs, it is commonly observed that nutrient input to the lowest trophic level usually increases the abundance of the highest trophic level while leaving the middle trophic level unchanged (Abrams 1993). The pattern of trophic interaction in the off-channel sites appeared to meet this prediction: higher periphyton biomass in the more nutrient-rich sites led to an increase in juvenile coho biomass, but did not lead to an increase in benthic invertebrates biomass. However, other research contradicts this by demonstrating that fish predation has little effect on invertebrate standing crops in streams (Allan 1982; Culp 1986; Deegan et al. 1997; Holomuzki and Stevenson 1992). Allan (1982) argued that this apparent insensitivity of invertebrate communities to fish presence was likely the result of highly evolved predator-avoidance traits such as small size, cryptic coloration and nocturnal activity. Dill et al. (1981) found that coho predation on invertebrates was limited primarily to those drifting in the water column. Given that invertebrate drift represents only a small proportion of total invertebrate biomass (Hynes 1970), enrichment would be expected to increase the biomass of both invertebrates and fish through '\"donor-controlled\" predator-prey dynamics (Abrams and Walters 1996). 41 Coho salmon The size of coho fry in late summer in the ponds and groundwater channels was positively related to periphyton biomass. In both years, fish were larger in the member of each pair where peak chlorophyll a densities were higher, which suggests that productivity was a strong determinant of summer growth in the off-channel sites. Similar to the results for invertebrate biomass, the magnitude of between-pair differences in fish weight also reflected the relative difference in chlorophyll a densities. The disparity in mean weight was greatest between the Mamquam pair where chlorophyll a densities differed the most, and least between the Cheakamus pair where chlorophyll a densities differed the least. Differences in fish condition between paired sites also supported the hypothesis that autotrophic productivity influenced summer growing conditions. In cases where the slopes of the length-weight regressions for paired sites were similar (Cheakamus and Coquitlam), fish were generally heavier for their length in the more periphyton-rich channel or pond of each pair. There was no ecological explanation for differences in slopes between the other paired sites. Greater periphyton biomass also shifted the distribution of fish size from skewed towards normal. Length-frequency distributions were less skewed towards larger individuals in the HP member of each pair. In stream-rearing coho, skewed size distributions appear to be a result of territorial behavior creating unequal access to food resources (Chapman 1962; Dill et al. 1981). Competition for food may have been lower in the more productive off-channel sites, thus lessening the intensity of size-dependent 42 growth. This contention is supported by a food manipulation experiment conducted by Mason (1976) who demonstrated that the distribution of size in young-of-the-year coho was strongly influenced by food availability and fish density. Mason found that the distribution of size normalized with both increased food and reduced density. Late summer fry mean weight was, at best, weakly correlated to benthic invertebrate biomass. The prediction that differences in invertebrate abundance would strongly influence fish growth between sites was supported only by the Vedder and Mamquam groundwater pairs. Differences in fish size and periphyton biomass between the Cheakamus channels were small relative to the other pairs. The fact that invertebrate biomass and fish size were inversely related for this pair probably reflects a lack of contrast in primary production and the questionable accuracy of the benthos sampling technique used. Lack of differences between the Coquitlam pair could be explained by benthic samples being collected in the inlet channels rather than the ponds where the fry resided or by lower fish predation on invertebrates in the LP pond prior to stocking in July. Despite the uncertainty of the fish-invertebrate interaction, my results lend support to other studies that have clearly linked fish size with productivity at lower trophic levels (Bolby and Roff 1986; Deegan and Peterson 1992; Johnston et al. 1990; Toth et al. 1996). I attribute larger fish size in the HP sites to an invertebrate food resource that was elevated by greater periphyton biomass. This bottom-up effect has been shown for coho across a diverse range of habitats including small streams (Bilby and Bisson 1992; Mason 43 1976; Thedinga et al. 1989), estuaries, (Koski and Kirchhofer 1984; Tschaplinski 1982) and off-channel ponds (Peterson 1982). My results extend this finding to juvenile coho in artificial ponds and groundwater channels. There is considerable evidence of density-dependent growth among stream-rearing coho populations during the summer (Bilby and Bisson 1992; Holtby and Hartman 1982; Mason and Chapman 1965). In my study, the effect of fish density on summer growth was uncertain. Within each year, a relationship between fish size and density among sites was either not apparent, or was overridden by differences in food abundance. I also failed to observe a consistent size-density relationship between years in each site. In three sites, a decrease in fish density from 1995 to 1996 was associated with an increase in mean weight, but in the other sites, a change in density between years failed to influence late summer size, or both density and mean weight increased in 1996. Greater fish biomass and average size in 1996 compared to 1995 in the majority of sites suggests that better growing conditions in 1996 may have masked density-dependent growth. However, relatively high coho densities in the groundwater channels may explain smaller average size there in September 1995 compared to average coho size in the adjacent river mainstems (see Results in Chapter 3). Average density of coho fry in the groundwater channels in late summer was 4.1 fish-m\"2. By comparison, average summer coho density for a dataset of B.C. streams was 0.35 fish-m-2 (n = 281, CV = 1.65; Ptolemy 1993), and average coho density in mainstem riffles of the Cheakamus River in October 1995 was estimated to be less than 0.01 fish-m-2 (Riley and Korman 1995). 44 Summer water temperatures were similar between the off-channel sites and the adjacent river mainstems, and did not likely contribute to differences in average size (see Fig. 3.2 in Chapter 3). The fact that primary production appeared to overshadow the importance of temperature in influencing fish size despite considerable variation in temperature among the off-channel sites is also significant. Studies that examined the impact of forest harvesting on juvenile coho showed that removal of the forest canopy increased both stream temperature and primary production, but it was believed that temperature had a much stronger influence on annual variation in coho size at the end of summer because of its effect on fry emergence timing and length of the growing season (Holtby and Hartman 1982; Thedinga et al. 1989). There was no apparent relationship between the abundance of coho fry and nutrient concentrations, periphyton biomass or benthic invertebrate biomass in the off-channel sites. During the summer, stream-rearing coho feed primarily on invertebrate drift which they compete for by defending individual territories or 'feeding stations' (Chapman 1962; Hartman 1965). Territory size has been shown to decrease with increasing food abundance in artificial stream channels (Dill et al. 1981; Mason and Chapman 1965). Primary production may therefore influence abundance because individuals require smaller feeding territories to maximize net energy gain when food is more abundant (Grant and Noakes 1987; Marschall and Crowder 1998). Yet, territory size also depends on body size because larger individuals require greater resources in order to meet higher 45 metabolic demands (Keeley and McPhail 1998). Comparing fish biomass rather than numbers addresses the potential confounding effect of fish size. Yet, a relationship between fish biomass and periphyton biomass was also not apparent across sites. However, differences in fish biomass were apparent between individual pairs in many cases. Two to fifteen times greater chlorophyll a densities in the HP sites was associated with, on average, 1.3 times greater invertebrate biomass and 1.3 times greater fish biomass in 1995, and 2.0 times greater fish biomass in 1996 (invertebrate data was not collected in 1996). The magnitude of differences in chlorophyll a densities and fish biomass between the HP and LP sites were comparable to the response of two B.C. coastal rivers to phosphorus enrichment (Slaney and Ward 1993). In their study, Slaney and Ward found that chlorophyll a increased 5 to 10 fold, invertebrate abundance, 2 to 7 fold, and fish biomass, 1.8 fold, when SRP concentrations were increased from less than 1 to 5-10 u-g-L\"1 following stream fertilization. Several studies that involved experimental manipulations of food abundance and/or fish density also found a relationship between biomass of late summer coho fry and food availability, but similar to my results, greater biomass was a result of larger fish size rather than higher density (Bilby and Bisson 1992; Mason 1976; Slaney and Ward 1993). This, together with evidence of a relationship between food availability and the skewness of length-frequency distributions in both my study and Mason's (1976), suggests that intra-specific competition is more likely to result in an unequal distribution of resources rather than the forced emigration of unsuccessful competitors. However, there are 46 exceptions; Thedinga et al. (1989) found that both the size and density of coho fry in 18 southeast Alaskan streams varied with autotrophic productivity. In the groundwater channels, a combination of relatively high fish abundance, low current velocity, and a lack of instream structure may have reduced the benefit of territory maintenance and therefore, the ability of coho to regulate their numbers in accordance with food availability. At sufficiently high densities, stream-rearing fish have been shown to switch from territorial to schooling behaviour in both laboratory experiments (Kalleberg 1958) and in field observations (Magnuson 1962; Kawanabe et al. 1957). Schooling was commonly observed in the groundwater channels during the summer (see Results in Chapter 3). Current velocity is important because it strongly affects invertebrate drift rates (Smith and Li 1983). Current speed was very low among the channels (mean = 0.09 m-s_I, range = 0.02 to 0.22 m-s\"1) and approached zero in the ponds. Smith and Li found that drift was negligible at velocities below 0.1 m-s\"1. Low current velocity in the channels and ponds may have favored fish that chose to actively search for prey rather than defend a feeding territory. Finally, a lack of heterogeneity in both morphology and current velocity may have also reduced the profitability of territory defense in the groundwater channels. In natural streams, habitat complexity is thought to aid in territory delineation by creating discrete feeding stations or by visually isolating individuals from one another (Dolloff 1983; Shrivell 1990). Ruggles (1966) examined the effect of channel morphology on juvenile coho behaviour in artificial stream channels that were physically similar to the groundwater channels in my study. He found that aggressive behaviour and territoriality were lower in relatively deep channels with low 47 current velocity compared to shallower channels with higher velocity. A lack of territorial behaviour was also reported for coho rearing in stream pools compared to higher velocity habitats (i.e., glides and riffles; Pucket and Dill 1985). 48 Table 2.1. Summary of means, standard errors and sample sizes for temperatures and current velocities in three pairs of groundwater channels and the inlet channels of a pair of ponds (Coquitlam) during June-October 1995. Pairs are identified by the names of the parent rivers given in the left margin. Each pair consists of a site with relatively high periphyton biomass (HP) and a site with relatively low biomass (LP). Site Mean summer temperature (°C) Mean current velocity (cm-s* Current velocity at invertebrate baskets (cm-s\"1) estimate SE n estimate SE n estimate SE n Vedder LP 10.2 0.14 13 2 0.5 14 2 1.1 13 HP 13.1 0.41 10 22 7.4 10 24 2.3 10 Mamquam LP 9.3 0.18 11 7 1.6 14 7 0.4 11 HP 8.5 0.13 11 5 1.0 10 15 2.2 11 Cheakamus LP 13.1 0.35 8 10 2.4 11 17 1.6 8 HP 10.0 0.12 10 11 1.9 15 10 0.6 10 Coquitlam LP 13.8 0.20 14 30 2.3 14 - - -HP 15.5 0.43 11 29 4.2 11 - - -49 Table 2.2. Summary of means, standard errors and sample sizes for total alkalinity, soluble reactive phosphorus and dissolved inorganic nitrogen for three pairs of groundwater channels and the inlet channels of a pair of ponds (Coquitlam) during June-October 1995. Pairs are identified by the names of the parent rivers given in the left margin. Each pair consists of a site with relatively high periphyton biomass (HP) and a site with relatively low biomass (LP). Underlined values indicate significant differences between paired sites. Site Vedder LP HP Mamquam LP HP Cheakamus LP HP Coquitlam LP HP Alkalinity (mg-L\") SRP (ug-L\") DIN (ug-L'1) estimate 26 36 14 20. 16 12 3 5 SE 2J. L3 QJ. 06 LO 1 2 0.3 1.4 estimate 6 5 1 19 5 7 2 4 SE 1.4 0.4 L i 22 0.7 0.5 06 0.9 estimate 125 81 47 185 20 4J 70 82 SE &5 9_Ji 6& 8,2 82 5A 6.7 13.7 50 Table 2.3. Mean atomic N:P ratios (dissolved inorganic nitrogen : soluble reactive phosphorus) among three pairs of groundwater channels and a pair of ponds (Coquitlam) during June - October, 1995. Values in brackets are standard errors, underlined values indicate significant differences in N:P ratios between paired sites. Site Cheakamus Mamquam Vedder Coquitlam LP 8.0 (3.3) 17.0 (9.6) 34.3 (4.3) 114.2 (-29.41 HP 9.3 (1.3) 13.0 (0.9) 23.3 (3.9) 51.5 51 Figure 2.1. Mean weekly chlorophyll a density on artificial substrates from three pairs of groundwater channels and a pair of ponds (Coquitlam) during July, September and March 1995-96. Pairs are identified by the name of the parent river given in the upper margin of each graph. For each pair, the site with higher peak chlorophyll a density is classed as the HP (high productivity) member whereas the site with lower peak density is classed as the LP (low productivity) member. Numbers along the x-axis refer to the number of weeks from the time of initial placement during each period. 52 Figure 2.2. Mean weekly temperatures from three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96. Pairs are identified by the name of the parent river given in the left-hand margin. Each pair consists of a site with relatively high periphyton biomass (HP) and a site with relatively low biomass (LP). Solid lines and shaded circles denote mean temperatures, dashed lines denote maximum and minimum temperatures observed. 54 High Productivity Low Productivity May-95 Aug-95 Oct-95 Jan-96 Apr-96 May-95 Aug-95 Oct-95 Jan-96 Apr-96 u O S H May-95 Aug-95 Oct-95 Jan-96 Apr-96 May-95 Aug-95 Oct-95 Jan-96 Apr-96 May-95 Aug-95 Oct-95 Jan-96 Apr-96 May-95 Aug-95 Oct-95 Jan-96 Apr-96 May-95 Aug-95 Oct-95 Jan-96 Apr-96 May-95 Aug-95 Oct-95 Jan-96 Apr-96 Date 55 Figure 2.3. Total alkalinity concentrations from three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96. Each pair is identified by the name of the parent river given in the left-hand margin. Solid lines and shaded diamonds denote the member of each pair having high periphyton biomass (HP) and dashed lines and unshaded squares denote the member having low biomass (LP). 56 Vedder River Mamquam River 40 30 20 \\-10 -OA E 0 40 R 30 20 10 h 0 a w - - 0 \" 0 - 0 ' Jun-95 Sep-95 Nov-95 Feb-96 Jun-95 Sep-95 Nov-95 Feb-96 o H 40 30 20 10 0 Cheakamus River u- - • -40 R 30 h 20 10 0 Coquitlam River Jun-95 Sep-95 Nov-95 Feb-96 Jun-95 Sep-95 Nov-95 Feb-96 Date 57 Figure 2.4. Soluble reactive phosphorus (SRP) concentrations from three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96. Each pair is identified by the name of the parent river given in the left-hand margin. Solid lines and shaded diamonds denote the member of each pair having high periphyton biomass (HP) and dashed lines and unshaded squares denote the site having low biomass (LP). 58 Vedder River Mamquam River -OX S3-© a (/> o .= a; & 30 25 20 15 10 5 0 Jun-95 30 25 20 15 10 5 0 Jun-95 Sep-95 Nov-95 Cheakamus River • 30 25 20 15 10 5 0 • -p.-n Feb-96 Jun-95 Sep-95 Nov-95 Feb-96 Coquitlam River Sep-95 Nov-95 Feb-96 Jun-95 Sep-95 Nov-95 Feb-96 Date 59 Figure 2.5. Dissolved inorganic nitrogen (DIN) concentrations from three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96. Each pair is identified by the name of the parent river given in the left-hand margin. Solid lines and shaded diamonds denote the member of each pair having high periphyton biomass (HP) and dashed lines and unshaded squares denote the site having low biomass (LP). 60 61 Vedder Mamquam Cheakamus Coquitlam Legend d3 Diptera @ Plecoptera • Ephemeroptera • Trichoptera • Gastropoda 53 Acoelomate, Nematoda Figure 2.6. Proportional composition of benthic invertebrates by mean numbers of taxa from three pairs of groundwater channels and a pair of ponds (Coquitlam) during September 1995. Pairs are identified by the name of the parent river given in the upper margin. Each pair consists of a site having high periphyton biomass (HP) and a site having low biomass (LP). 62 £ © © 6X1 ( « « s o BX) O --0.5 -1.5 -2.5 -3.5 ;fi;-si 1 MI s i Vedder Mamquam Cheakamus River Coquitlam Figure 2.7. Logarithmic mean biomass of benthic invertebrates from three pairs of groundwater channels and the inlets of a pair of ponds during September 1995. Pairs are identified by the names of the parent rivers given in the lower margin of each graph. Shaded bars denote values for the site of each pair having high biomass of periphyton and unshaded bars denote values for the site having low biomass. Error bars indicate ± one standard error and stars indicate significant differences between pairs (n = 3, p < 0.05). 63 Figure 2.8. Mean biomass of benthic invertebrates in off-channel habitat sites during September 1995 in relation to peak chlorophyll a density (average of values for July, September, and March 1995-96). Note the log scales on both axes. Values for r2 andp are the coefficient of determination and probability, respectively, for the linear regression of chlorophyll a on invertebrate biomass. 64 0.01 1 1 ' 1 1 0 0.1 0.2 0.3 0.4 Current Velocity (ms1) Figure 2.9. Mean biomass of benthic invertebrates in relation to current velocity in the immediate vicinity of the colonization baskets during September 1995. Note the log scale on the Y axis. Values for r 2 and p are the coefficient of determination and probability, respectively, for the linear regression of velocity on invertebrate biomass (groundwater channels only). 65 • Pond • Groundwater channel Log(Weight) = -1.39 + 0.60 • Log(Chl a) 8 15 23 30 Chlorophyll a (mg-m\"2) Figure 2.10. Mean weight of juvenile coho during September 1995 in relation to peak chlorophyll a density (average of values for July, September, and March 1995-96). Note the log scales on both axes. Values for r2 and p are the coefficient of determination and probability, respectively, for the linear regression of chlorophyll a on coho weight (groundwater channels only). 66 Figure 2.11. Mean weight of juvenile coho in relation to benthic invertebrate mean biomass during September 1995. Note the log scales on both axes. Values for r2 and p are the coefficient of determination and probability, respectively, for the linear regression of invertebrate biomass on coho weight (groundwater channels only). 67 Figure 2.12. Mean weight, biomass, and density of coho fry during September from three pairs of groundwater channels and a pair of ponds (Coquitlam) in 1995 and 1996. Pairs are identified by the names of the parent rivers given in the lower margin of each graph. Shaded bars denote values for the member of each pair having high periphyton biomass (HP) and unshaded bars denote values for the member having low biomass (LP). The ponds were sampled as whole units. Error bars indicate ± one standard error and stars indicate significant differences in mean density between pairs (n = 3, p < 0.05). 68 69 Figure 2.13. Length-frequency histograms for coho fry populations from three pairs of groundwater channels or ponds (Coquitlam) during September 1995 and 1996. Pairs are identified by the names of the parent rivers given in the left-hand margin. For each pair in each year, the site with low periphyton biomass (LP) appears on the right and the high biomass site (HP) appears on the left. The x-axis is labeled identically among figures, but is only shown for the figure in the bottom left hand corner. 70 IL Frequency Coquitlam Cheakamus Mamquam Vedder o o K) I 1 | 1 | 1 I 1 Figure 2.14. Length vs. weight linear regression predicted values for coho fry populations from sites with low periphyton biomass (broken lines) and sites with high biomass (solid lines) during September 1995 and 1996. Pairs are identified by the names of the parent rivers given in the left-hand margin. 72 WD WD o DX O J-> -1 1.0 S S S i V • i h 1.0 1.5 2.0 -1 h 1.5 2.0 2.5 1.0 1.5 2.0 2.5 3 1 -1 -3 2.5 1.0 1.5 2.0 2.5 1.0 1.5 2.0 2.5 1.0 1.5 2.0 2.5 1.0 1.5 2.0 -3 2.5 1.0 1.5 2.0 2.5 Log 1 0 Length (cm) 73 CHAPTER 3: WINTER ECOLOGY INTRODUCTION In temperate streams, primary production generally declines during the winter in response to reduced temperature and light (Bilby et al. 1996; Hynes 1970). In addition, periphytic algae and benthic invertebrate biomass may be reduced by adverse winter stream conditions such as ice formation or winter freshets (Hynes 1970). Low water temperature also affects fish directly by reducing both swimming performance (Griffiths and Alderdice 1972) and energy requirements (Metcalfe and Thorpe 1992). Some argue that the combined effect of low food availability, reduced feeding (or assimilation) efficiency and decreased appetite causes fish to adopt a strategy of minimizing their energy expenditure during the winter (Cunjak 1996; Metcalfe and Thorpe 1992). For example, juvenile salmonids conserve energy at low temperatures by occupying low-velocity areas within the streambed or in the lee of woody debris or other cover (e.g., Atlantic salmon: Allan 1940, Cunjak 1988a; chinook salmon (Oncorhynchus tshawytscha): Hillman et al. 1987, Taylor 1988; coho salmon: Taylor 1988, Bustard and Narver 1975; cutthroat trout: Bustard and Narver 1975; rainbow trout (O. mykiss): Reimers 1963). During the winter, it is also common for growth to cease or decline considerably (Cunjak 1988b; Holtby and Hartman 1982; Metcalfe and Thorpe 1992) The importance of primary production to fish growth in streams during the winter has received little attention. However, relatively high growth rates have been reported for coho overwintering in groundwater-fed ponds (Brown 1985; Peterson 1982) and artificial 74 groundwater channels (Sheng et al. 1990). Periphyton biomass, temperature and discharge were relatively stable throughout the year in the ponds and groundwater channels in my study. If conditions in these sites represent an opportunity for winter growth, it may be advantageous for fish to continue actively foraging rather attempting to conserve energy. In Chapter 2, average fish size in late summer was shown to be strongly correlated with periphyton biomass. My first objective here was to determine the influence of fish size in late summer on size the following spring just prior to smolting. My second objective was to determine if significant growth occurred during the winter, and if so, to consider its relative contribution to spring size. Smolt size is an important life-history parameter because it may influence subsequent survival and fitness. For example, marine survival is often correlated with smolt size (e.g., chum salmon: Healey 1982; chinook salmon: Neilson and Geen 1986; coho salmon: Holtby et al. 1990; Thedinga and Koski 1984; sockeye salmon (Oncorhynchus nerki): Peterman 1982; steelhead trout: Ward and Slaney 1988). In coho salmon, smolt size may also affect the timing of seaward migration (Irvine and Ward 1989), and this can also influence marine survival (Bilton et al. 1982; Labelled al. 1997). Through its effect on summer growth rate, primary production may also influence survival in freshwater. Winter is often a period of high mortality for young-of-the year fish (see Smith and Griffith 1994 for a review). Annual variation in overwinter survival has been positively correlated with mean body size at the end of summer (Holtby and 75 Hartman 1982; Hunt 1969). Relative survival rates of individuals may also be size-dependent during winter (Oliver et al. 1979; Post and Evans 1989; Quinn and Peterson 1996). In cases where there is little opportunity for winter feeding, larger fish may be less likely to starve because the ratio of energy stored to basal metabolic rate increases with size (Shuter et al. 1980). Others have suggested that size may be an advantage during the winter in avoiding predators or adverse stream conditions (Cunjak 1996). However, winter growth rate and survival may differ for juvenile coho rearing in off-channel habitats compared to those rearing in streams (Brown and Hartman 1988), and this may in turn affect size-dependent mortality (Brown 1985; Peterson 1985). My third objective was to examine the relationship between coho size in the channels and ponds in late summer and the abundance of pre-smolts the following spring. Coho smolt production is correlated with available rearing space (Marshall and Britton 1990; Bradford et al. 1996), and annual variation in smolt abundance is relatively low compared to variation in abundance at other life-history stages (Bradford 1995). Because of this, and the influence of habitat on the distribution of juveniles during the winter (Bustard and Narver 1975; Tschaplinski and Hartman 1983; Nickelson et al. 1992), some have suggested that smolt production is determined primarily by the amount of suitable winter habitat available in a stream (Hartman et al. 1996 ; Nickelson et al. 1992). The idea of a winter 'bottleneck' to freshwater production (Hall and Field-Dodgson 1981; Nickelson et al. 1993) is supported by studies that suggest a lack of a relationship between spring smolt abundance and the number of fry present the previous summer (e.g., Holtby and Hartman 1982; Lestelle et al. 1993; Mason 1976), providing 76 that fry numbers meet or exceed the winter 'carrying capacity' of the stream (Hartman et al. 1996). Although the groundwater channels studied here were selected to have similar physical habitat, relatively small differences in factors such as water depth and velocity appeared to influence the distribution of fish in the channels during the winter. My fourth objective was to examine the effect of both summer fry abundance and habitat quality on smolt production. I also considered the effect of habitat quality on smolt production in off-channel areas by contrasting seasonal fish abundance between the groundwater channels and ponds. DATA ANALYSIS Analyses in this chapter are based on data collected from three pairs of groundwater-fed channels and one pair of ponds during 1995-96 and 1996-97. Descriptions of the field and lab methods used are given in Field methods in Chapter 1. In Chapter 2, it was shown that for each pair, late summer fry (September) were larger at the site with high periphyton biomass site (HP). Therefore, in the analyses presented in this chapter, the treatment effect (i.e., HP or LP) reflects greater chlorophyll a densities and/or larger fry size in September. Throughout this chapter, the term 'fry' is used to describe juvenile coho sampled in September, and 'pre-smolt' is used to describe those sampled in March prior to seaward migration. 77 I used two approaches to assess the effect of primary production and other factors on the mean weight, density and biomass of pre-smolts. First, I used t-tests to detect differences in pre-smolt mean weight, density and biomass between paired HP and LP sites. Analysis of covariance was used to compare length-weight relationships for juvenile coho (see Data Analysis in Chapter 2). Fish condition was considered to be differ between paired sites if the slopes of the length-weight regressions were similar and the intercepts were significantly different. Differences in length-weight relationships could not be interpreted for pairs where the slopes of the length-weight relationships were not similar. In all cases where pairs were tested independently, observed significance levels were adjusted using the bonferroni method to reflect the fact that multiple comparisons were made (Zar 1974). My second approach was to pool the sites and test the relative influence of independent variables on pre-smolt size and abundance by building multiple linear regression models. For each case, I combined data from 1995-96 and 1996-97 if ANCOVA indicated that the slopes and intercepts of the regression were not significantly different between years. If the slopes or intercepts were different, data from 1995-96 and 1996-97 were tested separately. For all significant regressions, Cook's distances (Cook 1977) were plotted against the estimates of fish density to check for outliers or data points with large leverage. For multiple regressions, the data were also checked for multi-collinearity. Prior to testing, all data were checked for normality (Kolmogorov-Smirnov test, or scatterplots of variances against means if sample size was small) and for 78 homogeneity of variances (probability plots), and then logio transformed when necessary. Percentage data were arcsin transformed as suggested by Zar (1974). In addition to being quite variable among channels, densities of pre-smolts in March were often highly variable among the three 20 m sample reaches within each groundwater channel. Water depth and the width of the rip-rap bank tended to decrease, and current velocity to increase in a linear manner from the upper to the lower end of each channel. As a result, upstream reaches were often deeper and lower in velocity compared to downstream reaches. I examined whether fish density was more strongly correlated to habitat at this finer scale (i.e., 20 m reaches as opposed to whole channels) by treating each reach as an independent observation and building multiple regression models to find the best fit between the habitat variables and fish abundance. A description of the habitat information collected is provided in Chapter 1. I addressed the problem of autocorrelation between the habitat variables by initially summarizing the habitat data into principal components (PC's) derived from the correlation matrix. These PC's were then used in place of the habitat variables in the regression models. In cases where a scatter plot did not suggest a relationship between a particular habitat variable and pre-smolt density, that variable was not included in the PCA. Prior to regression analysis, the loading of factors on the PC's was improved using an orthogonal factor rotation (Orthomax; SYSTAT 1997). Overwinter increases in estimates of mean size could result from immigration of larger fish from the parent river mainstems to the off-channel sites during the fall, rather 79 than from growth of resident fish. The effect of immigration on overwinter size increase was investigated by marking all fish captured in the off-channel sites in September 1995 and comparing the lengths of marked (resident) and unmarked fish (resident or immigrant) captured in the off-channel sites in November, January and March. Fish were marked by complete removal of the left pelvic fin. In September 1995, 100 coho were also collected and sampled (weights and lengths) from the mainstem of each of the four parent rivers near the outlets of the channels in order that late summer fish sizes in off-channel and mainstem habitats could be compared. I assumed that immigration of mainstem fish had influenced the mean size of fish in an off-channel site if the mean length of the unmarked portion was greater than that of the marked portion. I did not consider the change in mark rate over time as a measure of immigration to the off-channel sites for several reasons. Firstly, mark rate was sensitive to fish movement within each groundwater channel and also between each pond and its inlet and outlet channels because only fish from limited areas of the channels and ponds were marked. Secondly, an unknown rate of mark loss occurred from September to March (i.e. pelvic fins that were not completely removed tended to grow back). The pattern of overwinter decline in fish abundance in the groundwater channels was also of interest. Densities were estimated in September and March only (sampling was conducted in July, November and January 1995-96, but only one electrofishing pass was completed in each sample reach). Therefore, to obtain an index of seasonal abundance, I compared catch-per-unit-effort (CPUE) among the five sampling dates in 1995-96. In this case, CPUE was the sum of all fish captured in the three sample reaches on the first 80 pass. Several studies have shown strong correlations between first-pass catches and population estimates for stream salmonids (Crozier and Kennedy 1994; Lobon-Cervia and Utrilla 1993; Jones and Stockwell 1995). Effort was standardized among dates by using the same number of people and repeating the same pattern and duration for each electrofishing pass. I compared capture probabilities (estimated as the ratio of the first pass capture total to the total capture for three passes) for the September and March sampling periods to determine if differences in temperature, fish behaviour or other factors biased the CPUE data. Size increases during the winter could have also resulted from size-selective mortality or emigration of smaller individuals (Ricker 1969). To test for this, I compared length distributions of fish sampled in September and March using empirical quantile-quantile (Q-Q) plots (Chambers et al. 1983). This method allows a comparison between two distributions by describing their shape with quantiles. Post and Evans (1989) demonstrated that this technique could be used to identify over-winter size-selective mortality among young-of-the-year yellow perch (Perca flavescens) in field enclosures. The technique involves plotting the quantiles of the length distribution at the end of summer (to) against the quantiles of the length distribution at the end of winter (ti). For example, if the median of the length distribution at ti (y coordinate) is plotted against the median at to (x coordinate), the data point would be (Qx 0.5, Q y 0.5). In order to reduce a univariate frequency distribution to a linear relation, I plotted quantiles 5, 10, 25, 50, 75, 90, 95 as they are distributed evenly along the length axis for a normal distribution (Post and Evans 1989). In my study, a reduction in the proportion of smaller fish would have 81 the same effect on size regardless of whether it was a result of mortality or emigration. Therefore, I did not attempt to distinguish between the two. Figure 3.1 illustrates the responses of Q-Q plots to various combinations of uniform and size-dependent growth and size-selective mortality. If the length distributions from September and March are identical, the Q-Q plot is a straight line with an intercept of 0 and slope of 1 (Fig. 3.1, Case 1). In the absence of overwinter growth, size-selective mortality of small individuals produces a slope of < 1 with a positive deviation from the 1:1 line for the lowermost quantiles only (Case 2). If mortality is not size-selective and if absolute growth is similar among small and large fish, the Q-Q plot will shift upward with no change in slope (Case 4). If mortality is not size-selective but larger fish grow faster than smaller ones, the Q-Q plot will have both a positive slope and a positive deviation from the 1:1 line (Case 7). Cases 5 and 8 show responses of the Q-Q plot to size-selective mortality of smaller fish under scenarios of uniform and size-dependent growth, respectively. Cases 3, 6 and 9 illustrate the response of Q-Q plots to size-selective mortality of larger fish. This latter scenario would not contribute to overwinter increases in fish size and was therefore not considered. For my study, it was also necessary to consider the response of the Q-Q plots to the immigration of larger fish. This would produce a response similar to that of size-selective mortality of smaller fish (i.e., a decrease in the slope of the Q-Q plot; Case 2) because the proportion of the length distribution represented by smaller fish would decrease. Unfortunately, certain responses cannot be separated using the simple slope and position diagnostics of the Q-Q plots (Post and Evans 1989; Sawada 1993). For example, cases 7 and 8a are similar as are 82 cases 4 and 8b. However, Post and Evans (1989) maintain that this approach is useful in revealing potential patterns of response particularly when a consistent pattern is evident among sites and among years. RESULTS Temperature and discharge Mean weekly water temperatures in the groundwater channels were considerably warmer during the November to April period compared to temperatures in the associated river mainstems (Fig. 3.2). During the December to February period, water temperature generally remained above 6°C in all of the channels whereas temperatures recorded for the rivers were usually less than 5°C and often approached 0°C. Winter water temperatures were colder in the Coquitlam ponds and the Cheakamus LP channel compared to the other off-channel sites because surface water contributed substantially to flow (Fig. 3.2; see Study Areas in Chapter 1). Temperatures in the warmer groundwater-fed channels were also less variable. For example, in the Vedder LP and both Mamquam channels, annual variation in mean weekly temperature was less than 4°C. In contrast, annual temperatures in two Coquitlam ponds and the Cheakamus LP channel varied over a range of 9 to 12°C. During the summer, water temperatures in all of the off-channel sites were similar to those in the adjacent river mainstems. The Vedder LP channel was exceptional in that temperature there remained close to 10°C year-round (Fig. 3.2). Discharge in the off-channel sites was low (less than 0.2 m3-s\"'), and remained relatively stable throughout the year (Fig. 3.3). The exceptions were the Vedder HP and 83. Cheakamus LP channels where discharge during the winter was both higher (0.2 to 0.6 m^ s\"1) and more variable. Surface water input from the Cheakamus River contributed to the variability of both flows and temperature in the Cheakamus LP channel. Flow in Grant's Tomb pond was regulated - an increase in discharge commencing on August 22 and continuing throughout the remainder of the study was the result of an adjustment to the intake control valve. Variable temperature and discharge in the Vedder HP channel suggest that sub-surface flow along the gravel fan of the Vedder river had a relatively large influence on flow in this site compared to the other channels. In 1995-96, flows in the channels and ponds were lowest in late summer and early spring and highest in November. Overall, weekly variation in discharge did not exceed one order of magnitude for any of the sites. Actual peak discharge may have been somewhat higher, but not by much. For example, on November 30, 1995 discharge in the Cheakamus and Mamquam groundwater channels was not substantially greater than on other occasions despite the occurrence of a storm the previous day that produced over 80 mm of rain in 12 hours and raised nearby streams to bankfull height. Fish behaviour Coho fry were highly visible in the groundwater channels from shortly after emergence until mid-November and were often aggregated in large schools. Fish were also visible in the ponds until November, but observation was made difficult by greater water depths and woody debris cover. Refuge cover in the groundwater channels consisted primarily of the interstitial spaces in the rip-rap banks. With the exception of the Vedder LP site, fish remained within the rip-rap during daylight hours after mid 84 November and were rarely observed. In the relatively warm Vedder LP site (see Fig. 3.2), juvenile coho were observed schooling in mid-channel areas throughout the winter. Despite an apparent lack of daytime activity, it was evident that fish were feeding throughout the winter in the groundwater channels. Fish captured during the winter had noticeably distended stomachs, and occasional samples of stomach contents were dominated by chum salmon eggs, fry, and carcasses. Diet samples suggested that coho were feeding primarily on eggs and aquatic invertebrates in November, carcasses in January and alevins, fry and carcasses in March. I visually (and non-statistically) estimated that chum spawner densities in the channels were between one and six fish per linear metre during the peak of spawning in November. While low numbers of adult coho were observed in the inlets and outlets of the Coquitlam ponds, chum salmon were not present. The stomach contents of fish sampled in the ponds during the winter included aquatic invertebrates, but no fish parts or egg sacs. Seasonal changes in fish size The mean weight of coho fry increased markedly from September to March at all of the off-channel sites in both 1995-96 and 1996-97 (Fig. 3.4). Among the groundwater channels, increases in mean weight between September and March ranged from 190 to 457% in 1996-97 compared to 103 to 399% in 1995-96 (Table 3.1). Increases in mean length between September and March ranged from 35 to 74% in 1996-97 and 29 to 64% in 1995-96. In the Coquitlam HP and LP ponds, overwinter increases in mean weight were also greater in 1996-97 (119 and 121%, respectively) compared to 1995-96 (23 and 85 96%, respectively), but were less than in the groundwater channels in both years (Table 3.1). In the HP and LP ponds, mean length increased 31 and 27%, respectively, in 1996-97 and 33 and 11%, respectively, in 1995-96. Seasonal patterns of 'growth' differed between groundwater channels where winter water temperatures were relatively warm and sites where temperatures were lowered by surface water input. In the latter sites where mid-winter temperatures generally remained below 5°C (Coquitlam ponds and the Cheakamus LP channel), increases in weight (both absolute and proportional) from November to March were less than those during July -November (Table 3.1; Fig. 3.4). However with the exception of the Vedder HP channel, weight increases were roughly equal between the two periods for the remaining sites. Across all sites, differences in mean water temperature in winter (Nov. 1 - March 31) explained 83% of the variation in the mean weight of pre-smolts in March 1996 (regression, logged values, n = 8, r2 = 0.83, p = 0.002; Table 3.2a) and 84% of the variation in pre-smolt weight in March 1997 (r2 = 0.84, p = 0.004; Table 3.2b). Inclusion of late summer fry size in the model explained an additional 12% of the variation in pre-smolt weight in March 1996 (multiple regression, logged values, n = 8, r2 = 0.95, p = 0.001; Table 3.2a), but did not explain additional variation in March 1997 (Table 3.2b). During both years, pre-smolt weight was not correlated with fish abundance in either March or September (regression, n = 8, p > 0.05 for all cases). For all analyses, data from each year were examined independently because the mean weight of pre-smolts was greater in 1997 than in 1996 (two-way ANOVA, logged values, p < 0.001). 86 When the paired sites were examined independently, a relationship between size in September and March was apparent. Fish from the HP site of each pair were generally larger than those from the LP site throughout 1995-96 and 1996-97, but relative differences in mean weight between paired sites in March were either similar or less compared to differences the previous September (Fig. 3.4). Within the Mamquam pair, coho fry were significantly larger in the HP channel on all sampling dates (bonferroni adjusted t-tests, logged values, p < 0.002 for all cases). Fish were larger in the HP channel of the Cheakamus pair for all dates (p < 0.002 for each case) except November 1995 (p = 0.15) and March 1997 (p = 0.17). Within the Vedder pair, fish were larger in the HP channel from July to January in 1995-96 and in September 1996 (p < 0.002 for all cases), but a difference in mean weight between channels was not apparent by March in 1996 (p = 0.64) or 1997 (p = 0.06). For the Coquitlam ponds, coho in the HP pond were smaller in September 1995 than those stocked in the LP pond (p < 0.002; see Results in Chapter 2). However by November, mean weights were similar between sites (p = 0.17), and by March, fish were larger in the HP pond (p = 0.001). In 1996-97, fish were 40% larger in the HP pond in both September (and March (p < 0.002 for both cases). While the mean weights of pre-smolts in March were often greater in the HP sites, there was no evidence of better 'condition' compared to that in the LP sites. I could not compare differences in condition between the HP and LP sites of the Coquitlam and Cheakamus pairs in 1995-96 because in both pairs, the slopes of the regressions for weight on total length differed (ANCOVA, logged values, p < 0.01 for each case), but not in a consistent manner. In all other cases, regression slopes were similar between pairs 87 (ANCOVA, p > 0.05). For these latter cases, the intercepts did not differ between paired sites (ANCOVA, p > 0.01 for all cases). This indicated that fish from the HP and LP member of each pair were similar in weight for a given length. The exception was the Vedder pair in 1995 where the intercept of the length-weight regression was greater for the LP channel indicating better condition compared to that in the HP channel (ANCOVA, p< 0.001). Mean lengths of coho fry from the groundwater channels were significantly less than those from the adjacent river mainstem in September 1995 (ANOVA, bonferroni post-hoc comparisons, p < 0.006 for all cases; Table 3.3). Fish from the Coquitlam LP pond were also smaller than those from the mainstem (p < 0.006; Table 3.3), but fish from the HP pond were similar in size to those from mainstem (p = 0.86). While sample sizes for the marked populations were often small, particularly for the January and March sample periods, the mean lengths of marked fish and unmarked fish were not significantly different for any of sites in November, January or March (bonferroni adjusted t-tests, logged values, p > 0.002 for all cases; Table 3.4). Quantile-quantile plots of the September and March length distributions showed a positive deviation in both the lower and upper quantiles compared to the 1:1 line for each site in each year (mean increment = 2.0 cm in 1995-96 and 2.4 cm in 1996-97; Figs. 3.5, 3.6). In other words, the entire length distribution shifted towards larger size in March compared to September. In addition to having greater intercepts compared to the 1:1 line, the plots for all the LP sites in both years (Figs. 3.5a-d; Fig. 3.6a-d) and the Coquitlam 88 HP pond in 1996-97 (Fig. 3.6h) had slopes greater than or equal to one (9 of 16 cases). Q-Q plots with positive intercepts and slopes greater than 1 diagnose faster growth of large individuals in the absence of size-selective mortality (Fig. 3.1, scenario 7). The slopes of the Q-Q plots for the LP member of each pair were generally greater those for the HP member which suggested that size-dependent growth was more likely in the less productive sites (Figs. 3.5, 3.6). Plots for the Vedder and Mamquam HP channels in 1995-96 (Figs. 3.5e-f) and the Cheakamus HP channels in 1995-96 and 1996-97 (Figs. 3.5g, 3.6g) diagnose either size-independent growth and mortality (Fig. 3.1, scenario 4), or size-selective growth for larger fish combined with size-selective mortality for smaller fish (Fig. 3.1, scenario 8a). The Q-Q plot for the Coquitlam HP pond in 1995-96 had both a positive intercept and a slope less than one (Fig. 3.5h) as did the plots for the Vedder and Mamquam HP channels in 1996-97 (Fig. 3.6e,f). These Q-Q plots diagnose overwinter growth in the presence of size-selective mortality of smaller fish or immigration of larger non-resident fish (Fig. 3.1, scenario 5). However, in the case of the Coquitlam HP pond, a slope of less than one for the Q-Q plot was a result of a relatively large positive shift for quantiles near the middle of the length distribution (Fig. 3.5h). Deviations from the 1:1 line were similar between the uppermost and lowermost quantiles suggesting that mortality was not necessarily higher for smaller fish. The Q-Q plots for the Vedder and Mamquam HP channels in 1996-97 should also be interpreted with caution; low sample sizes in March (n = 20 and 40, respectively) may have contributed to contrary results for these channels compared to other sites. 89 Seasonal changes in fish abundance Overwinter mortality and/or emigration of coho fry was very high among the groundwater channels. From September to March, abundance declined 86-99% during 1995- 96 and 80-99% during 1996-97 (Table 3.5). Mean densities of pre-smolts among the channels were generally less than 0.5 fish-m\"2 in (March) 1996, and less than 0.2 fish-m\"2 in 1997. Virtually no fish remained in the Cheakamus LP and Vedder HP channels by the end of winter in both years. Overwinter declines in fish biomass were also high, ranging from 60 to 98% among the channels in 1995-96 and 38 to 98% in 1996- 97 (Table 3.5). In contrast, spring pre-smolt densities were roughly equal to late summer fry densities in the Coquitlam HP pond in both years, and pre-smolt densities in the LP pond were only 18 and 23% less than fry densities in 1995-96 and 1996-97, respectively; (Table 3.5). Fish biomass in the HP pond increased during both winters (94 and 118% in 1995-96 and 1996-97, respectively), while biomass in the LP pond increased during the 1996-97 winter only (69%). The slopes of the lines for mean coho densities (logged values) plotted for March versus the previous September were relatively similar among the groundwater channels in 1995-96 (Fig. 3.7), suggesting that overwinter mortality and/or emigration in that year was not density-dependent. However, in 1996-97 the slopes of the lines were more strongly negative for channels where fish densities were relatively high in September, which suggests that density-dependent mortality and/or emigration may have occurred in that year (Fig. 3.7). The Vedder LP channel was an exception to the trend in 1996-97; the rate of overwinter decline in fish density was lower in this site compared to most of 90 the other channels despite the fact that fish densities at the Vedder LP site were relatively high in September. In sharp contrast to the plots for the groundwater channels, the plots of March versus September fish densities in the ponds had slopes near zero, suggesting that overwinter mortality and/or emigration was lower in the ponds than the channels (Fig. 3.7). The 1995-96 CPUE data suggested that a large proportion of the total decrease in coho abundance in the groundwater channels between July and March occurred during the September to November period (Fig. 3.8). The sharp decline in CPUE from September to November coincided with declines in water temperature and observed daytime fish activity and with the arrival of large numbers of adult chum salmon. Declining CPUE from November onward did not appear to be a result of reduced susceptibility to electro-fishing. In 1995-96, estimated capture probabilities were significantly higher in March (0.66) compared to September (0.45) (t-test, df = 34, p = 0.001). Pre-smolt abundance differed substantially among sites during March 1996 and 1997 (Table 3.5), but was not correlated with either peak chlorophyll a density (r = 0.44, n = 16, logged values, p = 0.19) or the mean weight of fry the previous September (r = 0.14, n = 16, logged values, p = 0.60; Fig. 3.9a). For these analyses, data from 1996 and 1997 were pooled because the slopes and intercepts of the regressions of chlorophyll a and mean size on pre-smolt density did not differ significantly between years (ANCOVA, p > 0.05 for each case). When I compared each pair individually, results were similar; the 91 HP member of each pair did not necessarily support greater numbers or biomass (Table 3.5) of pre-smolts compared to the LP member. Pre-smolt densities were significantly higher in the HP channel of the Cheakamus pair in both years and in the HP pond of the Coquitlam pond in 1995-96 (bonferroni adjusted t-test, logged values, p < 0.013 for each case; Table 3.5). However, for other pairs, pre-smolt densities were either greater in the LP site (Vedder pair in both years, p < 0.013 for each case; Table 3.5), or similar between sites (Mamquam, Coquitlam, p > 0.013 for all cases; Table 3.5). Coho pre-smolt densities tended to be greater in groundwater channel reaches that were relatively deep and slow-moving with larger substrate and porous rip-rap banks. PC 2 (which I will call 'depth-velocity'; Table 3.6b) was most important in explaining variation in pre-smolt density, followed by PC 1 (which I will refer to as 'substrate'; Table 3.6b). PC 3 (which I will refer to as 'bank cover'; Table 3.6b) did not explain a significant proportion of the variation in pre-smolt density. Prior to regression analysis, I pooled all reaches in both years because the slope and intercept of the regression for each PC on pre-smolt density did not differ between years (ANCOVA, p > 0.05 for all cases). PC's 1 and 2 together explained 28% of the variation in fish density among the 18 groundwater channel reaches (Table 3.2c). When the groundwater channels were compared as whole units, logio transformed pre-smolt density was positively correlated with logged values for depth-velocity (Fig. 3.9b), temperature (Fig. 3.9c), and the abundance of fry the previous summer (Fig. 3.9d). Prior to testing, I pooled data from both years. This was done because both the slope and 92 intercept of the regression for each independent variable on pre-smolt density did not differ significantly between years (ANCOVA, p > 0.05 for all cases). At the channel scale, depth-velocity and late summer fry density together explained 81% of the variation in pre-smolt density (Table 3.2d). However, the inclusion of temperature in the model did not significantly improve the fit (regression, n = 16, r2 = 0.84, p = 0.002). Depth-velocity and late summer fry density also explained 82% of the variation in pre-smolt biomass (Table 3.2e). DISCUSSION Seasonal changes in fish size The average size of fish in the off-channel sites increased markedly from September to March during both years of the study. This change in size was most likely due to overwinter growth. Immigration by larger fish and size-selective mortality of smaller fish were rejected as alternative explanations for overwinter size increases. Immigration by larger mainstem fry during the fall and winter did not appear to be an important contributor to increased size because the mean lengths of marked (summer resident) and unmarked fish (summer resident or fall-winter immigrant) were similar during November, January and March in 1995-96. Size-selective mortality (or emigration) of smaller fish also did not appear to have an effect on mean size because the Q-Q plots suggested that overwinter size increases were a result of growth. The Q-Q plots also did not suggest immigration of larger mainstem fish as a cause of overwinter size increases. Among the LP sites, overwinter growth appeared 93 to be size-dependent with larger fish growing more than smaller fish. In contrast, Q-Q plots for the HP sites diagnosed either non-selective growth or size-selective mortality of smaller fish. However, if size-selective mortality was contributing strongly to changes in the distribution of lengths during the winter, it would seem unlikely that this would occur only in the more productive sites. From the results of other studies, one would have expected size-dependent mortality to be more pronounced in the less productive sites where mean fish size was smaller. For example, in small-mouthed bass (Micropterus dolomieui; Oliver et al. 1979) and yellow perch (Post and Prankevicius 1987) size-selective mortality was found to increase with decreasing mean size in young-of-the-year populations. There were differences in the pattern and magnitude of size increases over time in the Vedder River channels, suggesting that immigration of larger mainstem fry may have had an influence on fish size at these sites. For example, mean weights in the Vedder LP and HP channels increased 169 and 118%, respectively, from September to November, compared to 16 to 67% for the other sites. Furthermore, the mean weight of fry actually declined in the Vedder LP channel from November to January (-18%) and in the Vedder HP channel from January to March (-4%). At other sites, changes in mean weight over time were always positive. Results from other studies conducted in groundwater channels within the same river systems lend support to the contention that overwinter size increases were due to growth (G. Giannico, University of British Columbia, Vancouver, unpublished data; Sheng et al. 94 1990). In a study of two channels in the Mamquam and Cheakamus rivers, the mean weight of individually marked coho fry increased 138% (CV = 0.54) on average between November and May (G. Giannico, University of British Columbia, Vancouver, unpublished data). Giannico also monitored upstream and downstream movement of coho fry in the upper 100 m reach of each channel during the winter. While it was not known whether upstream immigrants to the study areas originated from downstream areas of the channel or from the river mainstem, immigrants to the upstream study reaches were larger on average than emigrants. Thus, size-dependent movement could contribute to overwinter increases in mean size. However, Giannico found that upstream immigrants represented only 8.5% (CV = 0.63) of smolt outmigrants from the study reaches the following spring. Assuming that winter movement rates were similar in my study, this suggests that the large observed changes in size may have been due primarily to growth rather than size-dependent movement. In two other studies of artificial groundwater channels, spring pre-smolt populations also appeared to consist mostly of fish that were present the previous summer. An earlier study conducted in a groundwater channel in the Mamquam River found that smolt production was dependent largely on summer fry recruitment; smolt densities averaged 3.2 fish-m\"2 over two years when overwintering populations consisted of both summer residents and fall-winter immigrants compared to 0.1 fish-m\"2 for two years when fry were not present the previous summer (Sheng et al. 1990). Low or highly variable smolt production in several artificial off-channel ponds in the Clearwater River in Washington has also been linked to low summer recruitment (Peterson 1985). In contrast, 95 overwintering populations of coho in natural off-channel habitat may consist mostly of fish that emigrated from the mainstem during periods of high flow in the fall and early winter (Brown and Hartman 1988; Cederholm and Scarlett 1982). My results reinforce previous findings that opportunities for winter growth are relatively favorable in off-channel habitat compared to surface water streams (Brown 1985; Peterson 1982; Swales and Levings 1989). Peterson (1982) found that the average size of coho overwintering in two groundwater-fed ponds increased 49% and 94% from November to April. Overwinter weight increases in the Coquitlam ponds were similar to this (23-121%), but increases in the relatively warm groundwater channels were substantially higher (103-457%). However, larger coho from the adjacent rivers in September compared to those from the groundwater channels suggests that summer growth rates were higher in the former habitat type (see the Discussion in Chapter 2). There was a strong positive correlation between temperature and the size of pre-smolts at the end of winter in the channels and ponds. Temperature-related differences in growth were also found for fish overwintering in groundwater tributaries and swamps; growth occurred only during periods when water temperature exceeded 5-6°C (Brown 1985). Reduced foraging and increased photonegative behaviour concurrent with temperature declines below 5 - 10°C is a common behaviour among stream-dwelling salmonids (Allan 1940; Bustard and Narver 1975; Cunjak 1988a; Hillman et al. 1987; Taylor 1988). Winter temperatures in the ponds and groundwater channels (5.8 - 10.3°C) were warmer than those in the adjacent river mainstems, and overlapped the range of 96 temperatures for which a shift in feeding behaviour often occurs in streams. Because optimal feeding and assimilation efficiency for coho increases with temperature up to at least 12°C (Konecki et al. 1995), temperature likely acted directly on foraging ability or metabolism rather than indirectly on food abundance because chum salmon flesh, eggs and fry provided a seemingly unlimited source of food in the groundwater channels during the majority of the winter (see below). The presence of chum salmon, combined with higher water temperatures, may explain greater overwinter size increases in the channels compared to the ponds. Furthermore, higher metabolic demands in the relatively warm off-channel sites probably necessitated active feeding regardless of the potential for winter growth, because stored energy reserves would have been insufficient to maintain fish through the winter (Cunjak 1996). While fish generally reduce their foraging activity in response to declining temperatures during the fall in surface water streams, feeding may also continue to some degree at considerably lower temperatures (Cunjak 1988a; Cunjak and Randall 1993; MacMahon and Hartman 1989), but often only at night (Cunjak 1988a; Heggenes et al. 1993). This may also have been the case in the ponds and groundwater channels. In a previous study of the Cheakamus HP channel, Sheng et al. (1990) observed juvenile coho feeding on chum salmon carcasses at night during the month of January, but they did not observe any juveniles during daylight hours. In my study , coho appeared to exhibit strong photonegative behaviour from November to March in most of the sites concurrent with a decline in water temperatures below about 7-8°C. During daytime hours, fry remained visible only in the relatively warm (= 10°C) Vedder LP channel. 97 Chum salmon may be a major contributor to the winter diet of coho fry in the groundwater channels. Occasional diet samples that I collected during the winter contained mainly chum salmon eggs, adult carcass flesh and fry. This observation is supported by Sheng et al. (1990) who found that adult carcasses represented over 95% of juvenile coho stomach content samples by volume in January. In March, Sheng et al. found that carcasses represented 20%, fry and alevins, 60%, and invertebrates 20% of stomach samples, and in May, they found that fry represented 95% of stomach samples, while carcasses and invertebrates together represented 5%. The importance offish in the diet of juvenile coho has been suggested by other studies as well. Using stable isotope analysis, Bilby et al. (1996) found that adult coho spawners contributed over thirty percent of the carbon ( C) and nitrogen ( N) to juvenile coho fry in a coastal headwater stream, mostly as a result of fry consuming the eggs, carcasses and flesh of adult coho spawners. In a subsequent experiment, Bilby et al. (1998) found that the placement of salmon carcasses had a positive effect on the abundance, growth and condition of juvenile coho and steelhead during the winter. Finally, Ruggerone and Rogers (1992) found that yearling coho fed almost exclusively on sockeye fry during the spring and early summer in the Chignik Lakes, Alaska. They estimated that annual consumption of sockeye was between 24 and 78 million fish. Research concerning the growth of fish in streams is often limited to summer and often fails to consider energy sources derived from pathways other than autochthonous production. These types of studies may not provide meaningful insight into freshwater production in cases where growth continues through the winter and is influenced by 98 marine-derived energy sources. In the ponds and groundwater channels, autotrophic productivity was important in determining the size of coho fry in late summer, and to a lesser extent, the size of pre-smolts the following spring. However, pre-smolt size was consistently greater in the site with relatively high periphyton biomass only in the case of the Coquitlam ponds where winter growth appeared to be relatively slow, and the Mamquam channels where differences in fish size between sites in September were relatively large. Overall, temperature had a greater influence on pre-smolt size than fry size the previous summer. Because late summer fry size was reflected to some extent in pre-smolt size at the end of winter, summer growth may influence survival during subsequent life-history phases. For example, smolt size may influence the timing of seaward migration with larger individuals entering the ocean earlier (Irvine and Ward 1989; Sawada 1993). Marine survival has been correlated with both smolt size and the timing of seawater entry (i.e., higher survival with larger size and delayed migration; Bilton et al. 1982; Labelle et al. 1997; Mathews and Ishida 1989; Thedinga and Koski 1984). However, the influence of freshwater growth rate on marine survival may also depend on ocean conditions; Holtby et al. (1990) found that larger smolts survived better only during years when marine survival was relatively low. Given recent variability in ocean survival, consideration for smolt size as well as numbers is warranted (Bisbal and McConnaha 1998; Walters and Post 1993). 99 Seasonal changes in fish abundance I found no evidence that smolt production in the ponds and groundwater channels was related to fish size prior to the winter. Pre-winter fish size did not explain variation in the abundance of spring pre-smolts among sites nor did it explain differences in pre-smolt density between pairs. This suggests that overwinter survival was not strongly influenced by body size. Size-selective mortality was also not apparent within populations. Quantile-quantile plots of length-frequency data did not indicate a higher rate of mortality for smaller fish. In contrast, overwinter survival of young-of-the-year brook trout in Lawrence Creek, Wisconsin (Hunt 1969) and coho in Carnation Creek, B.C. (Holtby and Hartman 1982) was strongly dependent on average body size the previous fall. Increased fish size in late summer as a result of nutrient addition led to an increase in coho smolt production in both a stream in British Columbia (Slaney and Ward 1993) and a lake in Alaska (Kyle 1994), but this appeared to result from a reduction in freshwater residency rather than an improvement in overwinter survival (i.e., faster growth led to an increase in the proportion of age-1 smolts compared to age-2 smolts). Duration of freshwater residency was not a factor in my study because very few fish remained an additional year in the channels and ponds regardless of growth rate. My results suggest that caution should be exercised in ascribing correlations between growth rate and smolt (or pre-smolt) abundance to variation in overwinter survival. While size-dependent survival was apparent for coho in Carnation Creek as a whole, this was not observed in off-channel areas of the creek's floodplain (Brown 1985). Peterson's (1985) assessment of several natural groundwater channels in the Clearwater 100 River, Washington also did not suggest size-selective mortality within populations. An important similarity between the latter studies and mine was that substantial growth occurred during the winter. The studies by Hunt (1969) and Holtby and Hartman (1982) were conducted in streams where water temperature was colder during the winter. Coho in Carnation Creek grew very little during the winter (Holtby and Hartman 1982), and while it was not indicated whether overwinter growth occurred in Lawrence Creek, size-dependent mortality was offset in years when water temperature during the winter was relatively warm (Hunt 1969). In other relatively cold environments where overwinter survival was influenced by temperature, starvation appeared to be a significant mortality agent (e.g., Oliver et al. 1979; Shuter et al. 1980; Smith and Griffith 1994). In the warmer off-channel sites, temperature, despite its effect on winter growth, did not appear to strongly influence the number of fish remaining in early spring. The risk of starvation may have been relatively low in these sites because feeding opportunities remained during the winter. Winter growth also appeared to alleviate size-dependent mortality in Big Beef Creek, Washington: although survival was influenced by size, relatively small fish that overwintered in a lake in the upper reach of the stream grew significantly more and survived better than larger fish overwintering in the lower reach (Quinn and Peterson 1996). Effect of habitat In surface water streams, the distribution of juvenile coho during the winter may be strongly influenced by channel morphology, specifically water depth and velocity, and the amount of available cover (Bustard and Narver 1975; Heifetz et al. 1986; MacMahon 101 and Hartman 1989; Nickelson et al. 1992; Shrivell 1990). Whereas channel morphology in streams can vary dramatically in response to differences in factors such as gradient, discharge or riparian vegetation (Church 1992), morphology in the artificial groundwater channels was relatively homogenous both within and among sites. In addition, variation in discharge in the channels was exceptionally low compared to that typically occurring in streams (Hynes 1970). Despite these differences, fish abundance at the end of winter in the channels was positively correlated with depth and negatively correlated with velocity. Almost no fish remained by March in the two channels where depth-velocity was most unfavorable. This finding reinforces the importance of channel morphology as a determinate of winter habitat suitability for coho (MacMahon and Hartman 1989; Ruggles 1966). At the smaller spatial scale of channel reaches, cover and substrate also appeared to influence pre-smolt abundance, but to a lesser extent than depth and velocity. In natural streams, relatively high overwintering densities of coho were associated with relatively deep, low-energy habitat, but also with the presence of cover (Bustard and Narver 1975; Heifetz et al. 1986; MacMahon and Hartman 1989). Hiding cover in the groundwater channels was limited to the rip-rap boulders lining the banks. The fact that coho appear to prefer woody debris to other types of cover (Bustard and Narver 1975; MacMahon and Hartman 1989; Tschaplinski and Hartman 1983) may explain its lesser importance in the groundwater channels. Ruggles (1966) examined the effect of depth, velocity and cover (in this case, overhead plywood sheets) on coho smolt production in artificial stream channels that had similar morphology to the groundwater channels in my study. He also 102 found that smolt production was higher in relatively deep channels with greater availability of low velocity water. However, he did not find that smolt production increased with the addition of cover. The importance of habitat to overwintering success in off-channel systems was also suggested by the differing seasonal pattern of fish abundance between the surface water ponds and the groundwater channels. In comparison to the channels, the ponds were relatively deep, free of current, and rich in large woody debris. Recalling that in September the abundance of fish in the ponds was much lower than that in the channels, greater and less variable pre-smolt densities in the ponds in March suggests that overwinter survival was greater. Higher overwinter survival or carrying capacity in relatively deep, ponded habitat (e.g. groundwater ponds, swamps, lakes) compared to adjacent stream channel habitat has also been reported in several previous studies (Everest et al. 1986; Quinn and Peterson 1996; Swales et al. 1986; Tschaplinski and Hartman 1983). I did not find a relationship between the abundance of fry in late summer and depth or velocity. This was expected; coho are thought to be less rigid in their choice of summer habitat compared to winter habitat (Bustard and Narver 1975; Nickelson et al. 1992; Mason 1976; Tschaplinski and Hartman 1983). While coho may prefer deeper, slower habitat during summer (Bugert et al. 1991; Nickelson et al. 1992), their swimming performance and critical holding velocity are positively related to water temperature (Griffiths and Alderdice 1972). Thus, they are able to exploit a larger proportion of 103 available habitat during the summer. Depths and velocities in the groundwater channels were within the range preferred by coho during the summer (Bjornn and Reiser 1991; Frasen et al. 1993). Therefore, it is unlikely that either of these factors limited abundance. Factors important to overwinter survival Overwinter mortality is often high in young-of-the-year coho (Mason 1976; Holtby 1988; Quinn and Peterson 1996) and other stream-dwelling salmonids (see Smith and Griffith 1994 for examples). Many researchers have attributed winter mortality to physical displacement during freshets or starvation. However, overwinter declines in coho abundance among the groundwater channels (80-99%) were generally higher than those observed in streams (14-75%; reviewed by Murphy et al. 1984) despite the lower risk of displacement or starvation in the channels, suggesting that other mortality agents may be important. The assertion that winter flow conditions strongly influence overwinter survival is not supported by findings for coho or other species. For example, overwinter survival of juvenile coho was not correlated with flow conditions in Carnation Creek, a stream where 100-fold fluctuations in discharge are common during winter freshets (Holtby and Hartman 1982; Tschaplinski and Hartman 1983). It is likely that coho are highly adapted to the dynamic flow regimes of coastal watersheds. Flow experiments suggest that coho respond to increased discharge by seeking low velocity refuge areas, thus minimizing their risk of displacement (MacMahon and Hartman 1989; Shirvell 1990). 104 The fact that depth-velocity and cover influenced the distribution and abundance of coho pre-smolts in the relatively stable groundwater channels suggests that predation (Dolloff 1993) or energetic costs associated with high flow and low temperature (MacMahon and Hartman 1989) may have been more important than physical displacement as sources of winter mortality. Given that available refuge cover would likely be saturated at some level of fish abundance, one would expect overwinter mortality to be density-dependent. Overwinter mortality and/or emigration appeared to higher in the groundwater channels that had higher initial densities of coho in September in 1996-97, but this was not the case in 1995-96. Density-dependent overwinter mortality may have been masked in 1995-96 by a lack of contrast in September fish densities among channels compared to that 1996-97. When both years' data from all six groundwater channels were plotted together, coho density in March appeared to log-linearly related to density the previous September (Fig. 9d), which again suggests density-dependent overwinter mortality or emigration. The fact that overwinter declines in abundance were relatively low in the ponds, which had much lower densities in September compared to the channels, also suggests density-dependence. However, this is uncertain because differences in fish species assemblage or habitat between the channels and ponds may have also affected overwinter survival or emigration. During much of the winter, coho biomass in the groundwater channels was insignificant compared to chum biomass (i.e. < 0.1%). Coho predation may have been lower in the ponds because chum were not present to attract predators, or because greater abundance of woody debris and deeper water provided better hiding cover. 105 While predation rates on overwintering salmonids have generally not been quantified, fish may be particularly vulnerable to endothermic predators during the winter because their swimming performance is reduced at lower water temperatures (Cunjak 1996). For example, the removal of common merganser (Mergus merganser) from a stream in eastern Canada resulted in a 300-500% increase in the abundance of Atlantic salmon smolts (Elson 1962). There is also evidence that predation may account for a large proportion of mortality in juvenile coho (Dolloff 1993; Wood 1997). Faced with high discharge, ice formation or turbidity in surface water streams, predators may increase their foraging effort in off-channel habitat during the winter. The lack of structure in the groundwater channels coupled with favorable water depth and clarity appear to provide optimal foraging conditions for piscovores (Wood 1997). In my study, great blue heron (Ardea herodias), common merganser, and American dipper (Dolichonynx oryzivorus) were frequently observed foraging for coho fry during the winter. The CPUE data suggest that the majority of the September-March decline in coho abundance in the groundwater channels occurred prior to the November sampling period. Density-dependent emigration may have been a factor in this. Juvenile salmonids often emigrate from stream reaches that contain unsuitable winter habitat during the fall in response to falling water temperature (Bjornn 1971; Cunjak and Randall 1993; Hillman et al. 1987). The fact that pre-smolt abundance varied with depth, velocity and temperature suggests that fish occupying the shallower, higher velocity channels (or reaches) may have left in search of better winter habitat, particularly if temperatures in those channels 106 were also relatively cold. In contrast, more suitable habitat in the ponds may have increased the retention of summer resident fish and also encouraged immigration of mainstem fish during the fall (Peterson 1982). However, previous work in artificial groundwater channels did not suggest emigration by large numbers of coho during the fall or winter (G. Giannico, University of British Columbia, Vancouver, unpublished data). My results suggest that coho smolt production in off-channel habitat is strongly influenced by the quantity and quality of available winter habitat, similar to findings for surface water streams (Hartman et al. 1996, Nickelson et al. 1992). However, habitat quality did not act as a single limiting factor or 'bottleneck'. In contrast to other studies (e.g., Holtby and Hartman 1982; Knight 1980; Mason 1976), late summer fry abundance was equally important in explaining variation in pre-smolt densities among the groundwater channels. This relationship has previously been observed only in cases where summer rearing space was severely reduced by drought (Smoker 1955), or parent spawner abundance was very low (Andersen and Scrivener 1992). Neither of these conditions occurred in the channels. Although this study does not provide a clear explanation for high overwinter mortality, it does provide insight as to the relative importance of various factors. Large overwinter declines in abundance in the groundwater channels were not likely caused by physical displacement or starvation. However, predation and emigration may have been important and warrant further investigation. 107 Table 3.1. Percent increases in the mean weight of coho fry from paired groundwater channels and ponds during 1995-96 and 1996-97. Sites are paired by river. Each pair consists of a site with high periphyton biomass (HP) and a site with low biomass (LP). Values reported in the table are the changes (%) in mean weight between the sample dates indicated. In 1996-97, sampling occurred in September and March only. River Type of Habitat Period % Increase in Mean Weight Between Sample Periods 1995-96 LP HP 1996-97 LP HP Vedder groundwater channel July-Sept Sept-Nov Nov-Jan Jan-Mar Sept-Mar 42 169 -18 128 399 78 118 29 -4 170 457 270 Mamquam groundwater channel July-Sept Sept-Nov Nov-Jan Jan-Mar Sept-Mar 46 77 65 37 300 42 32 52 1 103 214 190 Cheakamus groundwater channel July-Sept Sept-Nov Nov-Jan Jan-Mar Sept-Mar 73 67 18 20 136 35 30 68 17 155 248 225 Coquitlam off-channel pond July-Sept Sept-Nov Nov-Jan Jan-Mar Sept-Mar 42 16 1 4 23 3 50 17 12 96 121 119 108 Table 3.2. Multiple regression analysis of coho pre-smolt weight, density and biomass with fall fry density, mean winter water temperature, and habitat variables derived from principal component analysis (see Table 3.6). Variables were login transformed prior to analysis. Models a and b, include groundwater channels and ponds as sample units. Models d and e, include only groundwater channels. Model c includes 20 m groundwater channel reaches as sample units (three per channel). Partial r values are the squared values for the correlation between the given independent variable and the dependent variable controlling for the other independent variables in the model. For a particular model, total r2 (coefficient of determination) and p (probability) are cumulative in that these values are re-calculated with the inclusion of each new independent variable. Slope is the coefficient for each independent variable in a model. Only variables that significantly accounted for variation in the dependent variable were included in each model. Data were checked for multi-collinearity and logio transformed prior to analysis. Model Dependent Variable Data Grouping Independ. variable n Partial r 2 Total r 2 slope P a. Pre-smolt mean weight By site 1. Temperature 8 0.96 0.83 1.90 0.002 1995-96 2. Summer fry size 0.79 0.95 0.30 0.001 b. Pre-smolt mean weight By site 1. Temperature 8 0.88 0.84 1.74 0.001 c. Pre-smolt density By section PC 2 (depth-velocity) 36 0.19 0.21 -0.05 0.008 Both years pooled PC 1 (substrate) 0.16 0.28 0.05 0.005 d. Pre-smolt density By site 1. Depth-velocity 12 0.53 0.53 -0.29 0.007 Both years pooled 2. Summer fry density 0.51 0.81 0.12 0.0005 e. Pre-smolt biomass By site 1. Depth-velocity 12 0.56 0.57 -0.74 0.005 Both years pooled 2. Summer fry density 0.51 0.84 0.28 0.0003 109 Table 3.3. Sample sizes, mean total lengths (cm) and associated standard errors for samples of coho fry from off-channel (LP, HP) and mainstem sites in four rivers during September 1995. LP (low productivity) and HP (high productivity) refer to the relative amount of periphyton biomass between off-channel pairs. Underlined values for length indicate significant differences between sites from each river (ANOVA, bonferroni adjusted pair-wise comparison, logged values, p < 0.05). River Site n Length SE Vedder LP 1309 5,6 0.04 HP 862 63 0.07 Mainstem 100 2A 0.09 Mamquam LP 477 5J) 0.04 HP 393 62 0.05 Mainstem 100 6£ 0.09 Cheakamus LP 270 5£ 0.06 HP 305 0.04 Mainstem 100 6S 0.07 Coquitlam LP 87 5J> 0.08 HP 100 6.4 0.09 Mainstem 100 6.4 0.09 110 Table 3.4. Mean total lengths of marked and unmarked coho fry from three pairs of groundwater channels and a pair of ponds during November, January, and March 1995-96. Pairs are identified by the name of the parent river. Each pair consists of a site having high periphyton biomass (HP) and a site having low biomass (LP). Values in brackets are standard errors. River Type of Site Date \" Mean Total Length (cm) Habitat marked unmarked marked unmarked Vedder groundwater LP November 27 209 6.9 (0.3) 6.9 (0.1) channel January 12 81 6.5 (0.4) 6.3 (0.1) March 22 212 8.5 (0.3) 8.5 (0.1) groundwater HP November 37 148 8.3 (0.1) 8.2 (0.1) channel January 4 29 8.8 (0.3) 9.3 (0.2) March 3 29 9.3 (0.9) 9.2 (0.2) groundwater LP November 37 219 5.3 (0.2) 5.3 (0.1) channel January 11 125 6.6 (0.3) 5.9 (0.1) March 17 397 7.1 (0.3) 6.9 (0.1) groundwater HP November 15 70 6.5 (0.3) 6.4 (0.1) channel January 6 67 6.5 (0.5) 6.9 (0.1) March 7 115 7.9 (0.2) 7.5 (0.1) groundwater LP November 18 107 6.0 (0.2) 5.9 (0.1) channel January 2 26 6.3 (0.6) 6.1 (0.1) March 2 14 7.1 (0.6) 6.7 (0.4) groundwater HP November 32 167 6.0 (0.1) 5.7 (0.1) channel January 8 83 6.8 (0.3) 6.5 (0.1) March 16 209 7.6 (0.3) 7.4 (0.1) off-channel LP November 33 129 7.1 (0.1) 7.3 (0.1) pond January 28 77 7.1 (0.1) 7.5 (0.1) March 46 129 7.3 (0.1) 7.7 (0.1) off-channel HP November 19 119 7.0 (0.2) 7.0 (0.1) pond January 9 78 6.9 (0.4) 7.7 (0.1) March 9 142 7.9 (0.3) 8.1 (0.1) 111 Table 3.5. Mean densities and biomass per unit area of juvenile coho from three pairs of groundwater channels and a pair of ponds (Coquitlam). Pairs are identified by the name of the parent river. Each pair consists of a site having high periphyton biomass (HP) and a site having low biomass (LP). Values in brackets are standard errors and underlined values indicate significant differences between paired sites. Site n 1995-96 1996-97 September March September March Vedder LP 3 4.52 (0.30) 0.25 (0.18) 5.92 (0.84) 0.66 (0.381 HP 3 2.45 (0.64) 0-02 amy 5.64 (0.37) 0.03 (0.031 , ui.qi Mamquam , ui.qi LP 3 5.60 (0.80) 0.56 (0.33) 2.43 (0.57) 0.07 (0.02) in (S HP 3 3.60 (2.13) 0.22 (0.11) 3.51 (3.05) 0.09 (0.07) isity | Cheakamus isity | LP 3 2.52 (0.72) 0.03 (0.01) 1.22 (0.23) 0.04 (0.011 Q HP 3 3.54 (0.21) 0.39 (0.06) 3.98 IL42) 0.19 (0.061 Coquitlam1 LP - 0.61 (0.04) 0.50 (0.04) 0.60 (0.08) 0.46 (0.09) HP - 0.77 (0.12) 0.76 (0.051 0.52 (0.07) 0.52 (0.06) Vedder LP 3 6.80 (0.43) 1.83 (1.32) 11.40 (1.62) 7.08 (10.12) HP 3 6.01 (1.57) 0.10 (0.05) 13.45 (0.82) 0.30 (0.27) S Mamquam LP 3 4.87 (0.70) 1.95 (1.28) 4.19 (0.98) 0.36 (0.11) V) HP 3 7.74 (4.58) 0.95 (0.47) 8.29 (7.20) 0.63 (0.49) « a Cheakamus S O LP 3 3.39 (0.97) 0.08 (0.03) 2.03 (0,38) 021 (0.03) « HP 3 5.42 (0.32) 1.70 (0.24) 7.45 (2.66) 1.17 (0.341 Coquitlam1 LP - 1.74 (0.11) 1.75 (0J2) 1.10 (0.14) 1.87 (0.38) HP - 1.62 (0.24) 3.13 (0.20) 1.35 (0.17) 2.96 (0.33) 1 The Coquitlam ponds were sampled as whole units, with fish abundance being estimated by mark-recapture. 112 Table 3.6 a-b. Correlations of habitat variables with principal components (PC's) derived from factoring the correlation matrix. Data are from 18 groundwater channel reaches (three 20 m reaches in each of the six channels). A description of the variables is given in Field methods in Chapter 1. The proportion of the total variance accounted for by each PC is given at the bottom of each table. In Table 3.6a, the variables are not rotated, whereas in Table 3.6b, the loading of variables on the PC's was improved using an orthogonal factor rotation (Orthomax; SYSTAT 1997). A. Habitat Variable PC 1 PC 2 PC 3 % Sand -0.802 0.393 0.3668 % Cobble 0.6416 -0.6707 -0.304 Velocity -0.6492 -0.629 0.1821 Depth 0.6152 0.5598 -0.4096 Bank penetration 0.7949 -0.0166 0.5201 Bank width 0.7739 0.0077 0.5763 % Variance 51.4 21.9 17.2 B. Habitat Variable PC 1 PC 2 PC 3 % Sand -0.8884 0.2921 -0.2257 % Cobble 0.9627 0.0059 0.1713 Velocity -0.0063 0.904 -0.2648 Depth 0.2301 -0.9041 0.0825 Bank penetration 0.2278 -0.1794 0.9053 Bank width 0.1621 -0.1671 0.9368 % Variance 31.6 30.4 28.2 113 Figure 3.1. Responses of empirical quantile-quantile (QQ) plots to combinations of size-selective mortality of small and large individuals and to growth models in which the incremental change in length of small individuals is equal to or less than that of large individuals. The dashed line is the 1:1 line. The open arrows indicate the direction of the translation due to growth effects, the size of the arrows indicating the relative magnitude of the response. The curved arrows indicate the rotation due to size-selective mortality. The a, b and c lines illustrate how combinations of size-selective mortality and size-selective growth can result in conflicting interpretations of QQ plots, (adapted from Post and Evans 1989). 114 Vedder Mamquam Figure 3.2. Annual thermographs of four rivers and associated pairs of groundwater channels or ponds (Coquitlam). In each graph, the river is represented by shaded stars and a solid line, the channel or pond with high periphyton biomass (HP) by shaded circles and a long-dashed line, and the channel or pond with low biomass (LP) by open circles and a short-dashed line. Trend lines were drawn using LOWESS (SYSTAT 1997), a running average smoothing function. Data points for each channel or pond are mean weekly water temperatures recorded during 1995-96. Temperature data for each river were obtained from historic water temperature records (Environment Canada 1977). 115 Figure 3.3. Weekly estimates of discharge from three pairs of groundwater channels and the inlet channels of a pair of ponds (Coquitlam) during 1995-96. Each pair is identified by the name of the parent river given in the left hand margin. LP and HP denote relatively low and high periphyton biomass, respectively. 116 High Productivity Low Productivity 0) u o TS TS > s CQ 0.8 0.6 0.4 0.2 0.0 May-95 Aug-95 Oct-95 Jan-96 Apr-96 0.8 May-95 Aug-95 Oct-95 Jan-96 Apr-96 0.8 0.6 0.4 0.2 s J2 \"« .tS 0.4 S CT O U 0.0 May-95 Aug-95 Oct-95 Jan-96 Apr-96 0.8 0.6 0.2 0.0 May-95 Aug-95 Oct-95 Jan-96 Apr-96 May-95 Aug-95 Oct-95 Jan-96 Apr-96 0.8 May-95 Aug-95 Oct-95 Jan-96 Apr-96 0.8 May-95 Aug-95 Oct-95 Jan-96 Apr-96 0.8 0.6 h 0.4 0.2 h 0.0 May-95 Aug-95 Oct-95 Jan-96 Apr-96 Date 1 1 7 Figure 3.4. Mean weight of coho fry in three pairs of groundwater channels and a pair of ponds (Coquitlam) during 1995-96 and 1996-97. Each pair is identified by the name of the parent river given in the left hand margin. Shaded bars denote values for the member of each pair having high biomass of periphyton and unshaded bars denote values for the member having low periphyton biomass. Error bars indicate ± one standard error. 118 1995-96 1996-97 u ? 4 > 2 0 July Sept Nov Jan Mar 10.7 July Sept Nov Jan Mar OX WD •FN o 3 cr E 03 1/3 s S « el itL July Sept Nov Jan Mar 6 4 2 0 8 6 4 E July Sept Nov Jan Mar July Sept Nov Jan Mar July Sept Nov Jan Mar •2 s 4 cr a * 1 July Sept Nov Jan Mar 1 July Sept Nov Jan Mar Month 119 Figures 3.5,3.6 Empirical quantile-quantile plots derived from the length-frequency distributions of coho fry in September (x-axis) and March (y-axis) for four pairs of groundwater channels or ponds (Coquitlam) during 1995-96 (a) and 1996-97 (b). Data points represent quantiles 5, 10, 25, 50, 75, 90 and 95 of the length distribution from the two periods. Dashed lines are 1:1 lines. Pairs are identified by the names of the parent rivers given in the left margin. Right-hand-side graphs correspond to the member of each pair having low biomass of periphyton (LP), and left-hand-side graphs correspond to the member having high biomass (HP). Trend-lines were fitted by linear regression. Sample sizes (n) are indicated in cases where the number of fish measured in March was less than 100. 120 1995-96 11 9 7 5 3 Low Productivity High Productivity a. jf 11 e. ^ 9 7 _ * 5 n = 29 1 1 t t 11 11 s ox a 0> o I u a c U ft CZ5 C5 S a\" E 3 £ 5 h 7 h 5 h c. 9 7 n= 14 5 n = 56 10 6 8 10 4 6 8 Late summer fry length (cm) 10 121 1996-97 Low Productivity High Productivity Figure 3.7. Logarithmic mean densities (fish-m\"2) of juvenile coho in September and March in six groundwater channels (solid lines) and two ponds (broken lines) during 1995-96 and 1996-97. Geometric mean densities, sample sizes and standard errors are given in Table 3.5. 123 0.8 0.6 0.4 0.2 0 1 0.8 0.6 0.4 0.2 1996-97 •Vedder HP -B—Vedder LP • Mamquam HP - X — Mamquam LP - * — Cheakamus HP - A — Cheakamus LP • - - Coquitlam HP -X- - - Coquitlam LP •Vedder HP -B—Vedder LP - • — Mamquam HP - X — Mamquam LP - * — Cheakamus HP -it— Cheakamus LP • - - Coquitlam HP -X- - • Coquitlam LP September March Time of Year 124 1200 - « - - Vedder LP July Sept Nov Jan March Date Figure 3.8. Seasonal declines in CPUE for coho fry in six groundwater channels during 1995-96. For a given channel on a particular sample date, CPUE represents the total number of fish captured during one pass of electro-fishing in three 20 m permanent sample reaches. Values for the Vedder LP channel in July and both Vedder channels in September are not given. On these occasions, electro-fishing was conducted throughout the Vedder sites rather than in just the 20 m reaches (see Field methods in Chapter 1). 125 Figure 3.9 a-d. Abundance of coho pre-smolts in March in six groundwater channels in relation to: a) late summer fry density, b) depth-velocity (see Data Analysis), c) average winter water temperature (Nov. 1 - March 31), and d) fry density the previous September. Note the log scale on the Y axis. The scatter-plots include data from both 1995-96 and 1996-97. 126 CHAPTER 4: CONCLUSIONS AND MANAGEMENT IMPLICATIONS During the summer, autotrophic productivity may be an important determinant of both average size and the degree of size-selective growth for juvenile coho populations rearing in artificial off-channel habitat. However, autotrophic productivity does not appear to have a strong influence on summer coho abundance in groundwater channels and ponds. Habitat quality is also not likely to be a strong determinant of fish abundance during the summer. However, my results do suggest that groundwater channels may be capable of supporting higher summer fish densities than ponds. Through its effect on summer growth rate, primary production may also influence the size of pre-smolts the following spring, but to a much lesser extent than mean water temperature during the winter. Winter growth rates in the ponds and groundwater channels were substantially higher than those observed in streams and this appeared to be the result of relatively warm water. The presence of chum salmon may have also been important to winter growth. In contrast to results from streams, my results suggest that pre-smolt abundance in groundwater channels is positively related to fish abundance the previous summer. Body size in late summer does not appear to have a strong effect on coho smolt production in groundwater channels and ponds. The assumption that overwinter survival can be improved by enhancing summer growth through nutrient addition in groundwater-fed off-channel habitat was unsubstantiated. It is possible that favorable winter growing 127 conditions in these areas alleviated size-dependent overwinter mortality. Stream enrichment studies show that nutrient addition may increase smolt production in populations where increased growth is sufficient to reduce the length of freshwater residency (Kyle 1994; Slaney and Ward 1993). However, its effect on overwinter survival remains uncertain. Nutrient addition may not be effective in restoring southern coho stocks which generally spend only one year in freshwater (Bradford et al. 1996). The abundance of coho pre-smolts in the groundwater channels was strongly influenced by apparently small differences in physical habitat. In addition, densities of pre-smolts in the deeper, more structurally complex ponds were generally greater than those in the channels. This reinforces both the importance of winter habitat to coho smolt production (Nickelson et al. 1992, 1993) and the importance of scale in attempting to re-create this habitat (Lewis et al. 1996). Fish abundance declined greatly during the winter in all the groundwater channels despite the absence of high flows. This pattern is similar to that occurring in surface water streams, which raises questions regarding the significance of winter freshets as a source of overwinter mortality in juvenile salmonids. Future research and management efforts should focus on other effects of winter habitat such as reducing predation risk or energetic costs. My results suggest that winter habitat requirements for coho rearing in relatively low energy off-channel areas are similar to requirements in high energy coastal streams (i.e. low velocity, adequate depth and structural complexity). The design of artificial off-channel habitat for coho could emphasize high quality winter habitat similar to that 128 preferred by coho in streams and naturally occurring off-channel areas. Ideally, off-channel projects should consist of a groundwater flow source, if available, and a series of connected channels and ponds. This study indicates that channel habitat could provide high quality spawning and summer rearing areas, while pond habitat could serve to maximize the retention and survival of juveniles through the winter. If managers also wanted to increase fish size to potentially improve marine survival (Bilton et al. 1982), or reduce the likelihood of size-dependent overwinter mortality (Quinn and Peterson 1996), directly increasing winter food abundance through the addition of salmon carcasses may be more effective than enhancement of primary production during the summer. Ideally, habitat restoration for coho should also include restoration of important prey species such as chum, pink or sockeye salmon. 129 BIBLIOGRAPHY Abrams, P. A. 1993. Effect of increased productivity on the abundance of trophic levels. Am. Nat. 141: 351-371. Abrams, P.A., and C.J. Walters. 1996. Invulnerable prey and the paradox of enrichment. Ecology 77(4): 1125-1133. Allan, K.R. 1940. Study on the biology of the early stages of the salmon (Salmo salar). 1. growth in the River Eden. J. Anim. Ecol. 9: 1-23. Allan, P. A. 1982. The effect of reduction in trout density on the invertebrate community of a mountain stream. Ecology 93: 1444-1455. Anderson, B.C., and J.C. Scrivener. 1992. Fish populations of Carnation Creek 1987-1990. Can. Data Rep. Fish. Aquat. Sci. 890. APHA. 1985. Standard methods for the examination of water and wastewater. 16th edition. Amer. Public. Health. Assoc. 1268 p. Beechie, T., E. Beamer, and L. Wasserman. 1994. Estimating coho salmon rearing habitat and smolt production losses in a large river basin, and implications for habitat restoration N. Amer. J. Fish. Mgmt. 14: 797-811. Bechara, J.A., and G. Moreau. 1992. Top-down effects of brook trout (Salvelinus fontinalis) in a boreal forest stream. Can. J. Fish. Aquat. Sci. 49: 2093-2103. Berg, CO. 1950. Biology of certain Chironomidae reared from Potamogen. Ecol. Monogr. 20(2). Bilby, R.E., B.R. Fransen, P.A. Bisson, and J.K. Walter. 1998. Response of juvenile coho salmon {Oncorhynchus kisutch) and steelhead {Oncorhynchus mykiss) to the addition of salmon carcasses to two streams in southwestern Washington, USA Can. J. Fish. Aquat. Sci. 55: 1909-1918. Bilby, R.E., B.R. Fransen, and P.A. Bisson. 1996. Incorporation of nitrogen and carbon from spawning coho salmon into the trophic system of small streams: evidence from stable isotopes. Can. J. Fish. Aquat. Sci. 53: 164-173. Bilby, R.E., and P.A. Bisson. 1992. Allochthonous versus autochthonous organic matter contributions to the trophic support of fish populations in clear-cut and old-growth streams. Can. J. Fish. Aquat. Sci. 49: 540-551. 130 Bilby, R.E., and P. A. Bisson. 1987. Emigration and production of hatchery coho salmon (Oncorhynchus kisutch) stocked in streams draining an old-growth and a clear-cut watershed. Can. J. Fish. Aquat. Sci. 44: 1397-1407. Bilton, H.T., D.F. Alderdice, and J.T, Schnute. 1982. Influence of time and size at release of juvenile coho salmon (Oncorhynchus kisutch) on returns at maturity. Can. J. Fish. Aquat. Sci. 39: 426-447. Bisbal, G.A., and W.E. McConnaha. 1998. Spatial and temporal factors affecting survival in coho salmon (Oncorhynchus kisutch) in the Pacific Northwest. Can. J. Fish. Aquat. Sci. 55: 2178-2186. Bisson, P. A., T.P. Quinn, G.H. Reeves, and S.V. Gregory. 1992. Best management practices, cumulative effects, and long-term trends in fish abundance in Pacific Northwest river systems. Pages 189-232 In: R.J. Naiman [ed.]. Watershed management: balancing sustainability and environmental change. Spronger-Verlag, New York. Bjornn, T.C. 1971. Trout and salmon movements in two Idaho streams as related to temperature, food, stream flow, cover, and population density. Trans. Amer. Fish. Soc. 100: 423-438. Bjornn, T.C, and D.W. Reiser. 1991. Habitat requirements of salmonids in streams. In: Meehan, W.R. [ed.] Influences of forest and rangeland management on salmonid fishes and their habitats. American Fisheries Society Special Publication 19, Bethesda, MD. Bohlin, T., C. Dellefors, U. Faremo, and A. Johlander. 1994. The energetic equivalence hypothesis and the relation between population density and body size in stream-living salmonids. Am. Nat. 143: 478-493. Bolby, J.N. and J.C. Roff. 1986. Trout biomass and habitat relationships in southern Ontario streams. Trans. Am. Fish. Soc. 115: 503-514. Bonnell, R.G. 1991. Construction, operation, and evaluation of groundwater-fed side channels for chum salmon in British Columbia. Amer. Fish. Soc. Symp. 10: 109-124. Bothwell, M.L., G. Derksen, R.N. Nordin, and J.M. Culp. 1992. Nutrient and grazer control of algal biomass in the Thompson River, British Columbia: a case history of water quality management. In: R.D. Roberts, and M.L. Bothwell (eds.). Aquatic ecosystems in semi-arid regions: implications for resource management. N.H.R.I. Symposium Series 7, Environment Canada, Saskatoon, Canada. 131 Bothwell, M.L. 1988. Growth rate responses of lotic periphytic diatoms to experimental phosphorus enrichment: the influence of temperature and light. Can. J. Fish. Aquat. Sci. 45: 261-270. Bradford, M.J. 1995. Comparative review of Pacific salmon survival rates. Can. J. Fish. Aquat. Sci. 52: 1327-1338. Bradford, M.J., G.C. Taylor, and J.A. Allan. 1996. Empirical review of coho salmon smolt abundance and the prediction of smolt production at the regional level. Trans. Amer. Fish. Soc. 126: 49-64. Brown, L.R., P.B. Boyle, and R.M. Yoshiyama. 1994. Historical decline and current status of coho salmon in California. N. Amer. J. Fish. Mgmt.14: 237-261. Brown, T.G. 1985. The role of abandoned stream channels as over-wintering habitat for juvenile salmonids. Master's thesis. University of British Columbia, Vancouver, Canada. Brown, T.G., and G.F. Hartman. 1988. Contributions of seasonally flooded lands and minor tributaries to the production of coho salmon {Oncorhynchus kisutch) in Carnation Creek, British Columbia. Trans. Amer. Fish, Society. 177: 546-551. Brylinsky, M. 1980. Estimating the productivity of lakes and reservoirs. In: E.D. Le Cren, and R.H. Lowe-McConnell [eds.]. The functioning of freshwater ecosystems. International Biological Programme 22. Cambridge University Press, Cambridge. Bugert, R.M., T.C. Bjornn, and W.R. Meehan. 1991. Summer habitat use by young salmonids and their responses to cover and predators in a small Southeast Alaska stream. Trans. Amer. Fish. Soc. 120 (4): 474-485. Bustard, D.R., and D.W. Narver. 1975. Preferences of juvenile coho salmon {Oncorhynchus kisutch) and cutthroat trout {Salmo clarki) relative to simulated alteration of winter habitat. J. Fish. Res. Board Can. 32: 681-687. Cederholm, C.J., and W.J. Scarlett. 1982. Seasonal immigrations of juvenile salmonids into four small tributaries of the Clearwater River, Washington, 1977-1981. In: E.L. Brannon and E. Salo, [eds.]. Proceedings of the salmon and trout migratory symposium. Univ. of Washington, School of Fisheries, Seattle: pp. 98-110. Chambers, J.M., W.S. Cleveland, B. Kleiner, P.A. Tukey. 1983. Graphical methods for data analysis. Wadsworth International Group, Belmont CA and Duxbury Press, Boston MA. 395 p. Chapman, D.W. 1962. Aggressive behaviour in coho salmon as a cause of emigration. J. Fish. Res. Bd. Can. 19(6): 1047-1080. 132 Church, M. 1992. Channel morphology and typology. In: C. Callows and G. Petts [eds.] The river handbook: hydrological and ecological principles. Oxford, Basil Blackwell. Cone, R.S. 1989. The need to reconsider the use of condition indices in fishery science. Trans. Amer. Fish. Soc. 118: 510-514. Cook, R.D. 1977. Detection of influential observations in linear regression, Technometrics. 19: 15-18. Coronado, C , and R. Hillborn. 1998. Spatial and temporal factors affecting survival in coho salmon {Oncorhynchus kisutch) in the Pacific Northwest. Can. J. Fish. Aquat. Sci. 55: 2067-2077. Crozier, W.W., and G.J. A. Kennedy. 1994. Application of semi-quantitative electrofishing to juvenile salmonid surveys. J. Fish Biology. 45:159-164. Culp, J.M. 1986. Experimental evidence that stream macroinvertebrate community structure is unaffected by different densities of coho salmon fry. J. N. Am. Benthol. Soc. 5: 140-149. Cunjak, R.A. 1996. Winter habitat of selected stream fishes and potential impacts form land-use activity. Can. J. Fish. Aquat. Sci. 53(Suppl. 1): 267-282. Cunjak, R.A. 1988a. Behaviour and habitat of young Atlantic salmon (Salmo salar) during winter. Can. J. Fish. Aquat. Sci. 45: 2156-2160. Cunjak, R.A. 19886. Physiological consequences of over-wintering: the cost of acclimatization? Can. J. Fish. Aquat. Sci. 45: 443-452. Cunjak, R.A., and G. Power. 1986. Winter habitat utilization by stream resident brook trout (Salvelinus fontinalis) and brown trout {Salmo trutta). Can. J. Fish. Aquat. Sci. 43: 1970-1981. Cunjak, R.A., and Randall RG. 1993. In-stream movements of young Atlantic salmon during winter and early spring. In: R.J. Gibson, R.E. Cutting [eds.] Production of juvenile Atlantic salmon (Salmo salar) in natural waters. Can. Spec. Publ. Fish. Aquat. Sci. 118: 43-51. Decker, S. 1998. Influence of Off-Channel Habitat Restoration and Other Enhancement on the Abundance and Distribution of Salmonids in the Coquitlam River. Unpublished report prepared for B.C. Hydro Power Facilities, Burnaby, B.C. 35 P-133 Deegan, L.A., B.J. Peterson, H. Golden, C.C. Mclvor, and M.C. Miller. 1997. Effects of fish density and river fertilization on algae standing stocks, invertebrate communities, and fish production in an arctic river. Can. J. Fish. Aquat. Sci. 54: 269-283. Deegan, L.A., and B.J. Peterson. 1992. Whole-river fertilization stimulates fish production in an Arctic tundra river. Can. J. Fish. Aquat. Sci. 49: 1890-1901. Dill, L.M., R.C. Ydenburg, and A.H.G. Fraser. 1981. Food abundance and territory size in juvenile coho salmon. Can. J. Zoo. 59: 1801-1809. Dolloff, CA. 1993. Predation by river otters (Lutra canadensis) on juvenile coho salmon (Oncorhynchus kisutch) and Dolly Varden (Salvelinus malma) in southeast Alaska. Can. J. Fish. Aquat. Sci. 50:312-315. Dolloff, CA. 1983. The relationships of wood debris to juvenile salmonid production and microhabitat selection in small Southeast Alaska streams. Ph.D. thesis, Montana State University, Bozeman, MT. 100 p. Elliot, J.M. 1993. The self-thinning rule applied to juvenile sea-trout, Salmo trutta. J. Anim. Ecol. 62: 371-379. Elson, P.F. 1962. Predator-prey relationships between fish-eating birds and Atlantic salmon. Bull. Fish. Res. Board Can. 133. Environment Canada. 1990. Historic streamflow summary: British Columbia to 1989. Inland Waters Directorate, Water Resources Branch, Ottawa. 861 p. Environment Canada. 1977. Water temperatures: British Columbia and Yukon Territory. Inland Waters Directorate, Water Resources Branch, Ottawa. 486 p. Everest, F.H., G.H. Reeves, J.R. Sedell, J. Wolfe, D. Hohler, and D.A. Heller. 1986. Abundance, behavior, and habitat utilization by coho salmon and steelhead trout in Fish Creek, Oregon, as influenced by habitat enhancement. 1985 Annual Report, Bonneville Power Administration, Division of Fish and Wildlife. Project 84-11, Portland, OR. Feltmate, B.W., and D.D. Williams. 1989. Influence of rainbow trout (Oncorhynchus mykiss) on density and feeding behaviour of perlid stonefly. Can. J. Fish. Aquat. Sci. 46: 1575-1580. Ford, T.E., and R.J. Naiman. 1989. Groundwater-surface water relationships in boreal forest watersheds: dissolved organic carbon and inorganic nutrients dynamics. Can. J. Fish. Aquat. Sci. 46: 41-49. 134 Frasen, B.R., P. A. Bisson, J.W. Ward, and R.E. Bilby. 1993. Physical and biological constraints on summer rearing of juvenile coho salmon (Oncorhynchus kisutch) in small western Washington streams. In: Proceedings of the 1992 coho workshop. Nanaimo, B.C. pp. 271-288. Frissell, C.A. 1993. Topology of extinction and endangerment of native fishes in the Pacific Northwest and California (USA). Conservation Biology 7: 342-354. Frissell, C.A., and R.K. Nawa. 1992. Incidence and causes of physical failure of artificial habitat structures in streams in western Oregon and Washington. N. Amer. J. Fish. Mgmt.12: 182-197. Gibson, R.J., and R.L. Haedrich. 1988. The exceptional growth of juvenile Atlantic salmon (Salmo salar) in the city waters of St. Johns, Newfoundland, Canada. Polskie Archiwum Hydrobiologii 35: 385-407. Gore, J.A. 1996. Discharge measurements and streamflow analysis. In: F.R. Hauer and G.A. Lamberti [eds.] Methods in stream ecology. Academic Press, pp. 53-74. Grant, J.W.A., and D.L. Noakes. 1987. A simple model of optimal territory size for drift-feeding fish. Can. J. Zoo. 65: 270-276. Grant, J.W.A., S.O. Steingrimsson, E.R. Keeley, and R.A. Cunjak. 1998. Implications of territory size for the measurement and prediction of salmonid abundance in streams. Can. J. Fish. Aquat. Sci. 55 (Suppl. 1): 181-190. Griffiths, J.S., and D.F. Alderdice. 1972. Effects of acclimation and acute temperature on the swimming speed of juvenile coho salmon. J. Fish. Res. Bd. Can. 29: 251-264. Hall, J.D., and M.S. Field-Dodgson. 1981. Improvement of spawning and rearing habitat for salmon. In: C L Hopkins [ed.] Proceedings of the salmon symposium. New Zealand Ministry of Agriculture and Fisheries, Fisheries Research Division. Oceas. Publ. No. 30. Hart, D.D., and CT. Robinson. 1990. Resource limitation in stream community: phosphorus enrichment effects on periphyton and grazers. Ecology 71: 1494-1502. Hartman, G.F. 1965. The role of behaviour in the ecology and interaction of underyearling coho salmon (Oncorhynchus kisutch) and steelhead trout (Salmo gairdneri). J. Fish. Res. Bd. Canada. 22: 1035-1081. Hartman, G.F., J.C Scrivener, and M.J. Miles. 1996. Impacts of logging in Carnation Creek, a high energy coastal stream in British Columbia, and their implication for restoring fish habitat. Can. J. Fish. Aquat. Sci. 53 (Suppl.1): 237-251. 135 Healey, M.C. 1982. Timing an relative intensity of size-selective mortality of juvenile chum salmon (Oncorhynchus keta) during early sea life. Can. J. Fish. Aquat. Sci. 39: 952-957. Heifetz, J, M.L. Murphy, and K.V. Koski. 1986. Effects of logging on winter habitat of juvenile salmonids in Alaskan streams. N. Amer. J. Fish. Mgmt. 6: 52-58. Heggenes, J., O.M.W. Krog, O.R. Lindas, J.G. Dokk, and T. Bremnas. 1993. Homeostatic behavioural responses in a changing environment: brown trout (Salmo trutta) become nocturnal during winter. J. Anim. Ecol. 62: 295-308. Hillborn, R. 1992. Hatcheries and the future of salmon in the Northwest. Fisheries 17(1): 5-8. Hillman, T.W., J.S. Griffith, and W.S. Platts. 1987. Summer and winter habitat selection by juvenile chinook salmon in a highly sedimented Idaho stream. Trans. Amer. Fish. Soc. 116: 185-195. Holomuzki, J.R., and R.J. Stevenson. 1992. Role of predatory fish in community dynamics of an ephemeral stream. Can. J. Fish. Aquat. Sci. 49: 2322-2330. Holtby, L.B. 1988. Effects of logging on stream temperatures in Carnation Creek, British Columbia and associated impacts on coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 45: 502-515. Holtby, L.B., and G.F. Hartman. 1982. The population dynamics of coho salmon (Oncorhynchus kisutch) in a west coast rain forest stream subject to logging. In: G.F. Hartman [ed.] Proceedings of the Carnation Creek work-shop, a ten year review. Pacific Biological Station, Nanaimo, B.C: pp. 308-347. Holtby, L.B., and R. Kadowaki. 1998. A risk assessment for north coastal coho fisheries in 1998, with commentary on risk in southern inside and outside fisheries. PSARC Working Paper S98-12: 63 p. Holtby, L.B., B.C. Andersen, and R.K. Kadowaki. 1990. Importance of smolt size and early ocean growth to interannual variability in marine survival of coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 47: 2181-2194. Hoyer, M.V., and D.E. Canfield, Jr. 1991. A phosphorus-fish standing crop relationship for streams? Lake Reservoir Management. Vol. 7, no. 1, July 1991. Hunt, R.L. 1969. Overwinter survival of wild fingerling brook trout in Lawrence Creek, Wisconsin. J. Fish. Res. D. Canada 26: 1473-1493. Hynes, H.B.N, 1970. The ecology of running waters. University of Toronto Press. 555 p. 136 Irvine, J.R., and B.R. Ward. 1989. Patterns of timing and size of wild coho salmon (Oncorhynchus kisutch) smolts migrating from the Keogh River watershed on northern Vancouver Island. Can. J. Aquat. Sci. 46: 1086-1094. Johnston, N.T., C.J. Perrin, P. A. Slaney, B.R. Ward. 1990. Increased juvenile salmonid growth by whole-river fertilization. Can. J. Fish. Aquat. Sci. 47: 862-872. Jones, M.L., and J.D. Stockwell. 1995. A rapid assessment procedure for the enumeration of salmonine populations in streams. N. Am. J. Fish. Mgmt. 15: 551-562. Kalleberg, H. 1958. Observations in a stream tank of territoriality and competition in juvenile salmon and trout (Salmo salar and S. trutta). Inst. Freshw. Res. Drottningholm 39: 55-98. Kawanabe, H., S. Mori, and N. Mizuno. 1959. On the food economy of \"Ayu\" fish with relation to the production of algae. Physiol, and Ecol. Japan. 8: 117-123. Keeley, E.R. 1998. Behavioural and demographic responses to food and space competition by juvenile steelhead trout (Oncorhynchus mykiss). Ph.D. thesis, the University of British Columbia, Vancouver, B.C. Keeley, E.R., and J.D. McPhail. 1998. Food, abundance, intruder pressure, and body size as determinants of territory size in juvenile steelhead trout (Oncorhynchus mykiss). Behaviour. 135: 65-82. Keeley, E.R., and C.J. Walters. 1994. The British Columbia Watershed Restoration Program: summary of the experimental design, monitoring and restoration techniques workshop. Province of British Columbia, Ministry of Environment, Lands and Parks and Ministry of Forests. Watershed Restoration Project Report No. 4: 34 p. Kennedy, H.D. 1967. Seasonal abundance of aquatic invertebrates and their utilization by hatchery reared rainbow trout. U.S. Fish, and Wildl. Serv. Tech. Paper No. 12. Wash. D.C. Kimble, L.A. and, T.A. Wescle. 1975. Relationships between selected physical parameters and benthic community structure in a small mountain stream. Water Resources Series No. 55. Laramie, Univ. Wyoming. Knight, N.J. 1980. Factors affecting the smolt yield of coho salmon (Oncorhynchus kisutch) in three Oregon streams. M.Sc. dissertation. Oregon State University, Corvallis. Konecki, J.T., C.A. Woody, and T.P. Quinn. 1995. Temperature preference in two populations of juvenile coho salmon, (Oncorhynchus kisutch). Environmental Biology Of Fishes 44: 417-421. 137 Koski. K.V., and D.A. Kirchhofer. 1984. A stream ecosystem in an old-growth forest in southeastern Alaska: part III: Food of juvenile salmonids in relation to relative abundance of drift and benthos. Pages 81-87 In: W.R. Meehan, T.R. Merrell, Jr., and T.A. Haney [eds.] Proc. Symposium on Fish and Wildlife Relationships in Old-Growth Forests, Juneau, Alaska, April 1982. Amer. Inst. Fish. Res. Biol: pp. 12-15. Kyle, G.B. 1994. Assessment of trophic level responses and coho salmon (Oncorhynchus kisutch) production following nutrient treatment (1981-1986) of Bear Lake Alaska. Fisheries Research 20: 243-261. Labaugh, J.W., D.O. Rosenberry, and T.C. Winter. 1995. Groundwater contribution to the water and chemical budgets of Williams Lake, Minnesota, 1980-1991. Can. J. Fish. Aquat. Sci. 52: 754-767. Labelle, M., C.J. Walters, and B. Riddell. 1997. Ocean survival and exploitation of coho salmon stocks from the east coast of Vancouver Island, British Columbia. Can. J. Fish. Aquat. Sci. 54: 1433-49. Lestelle, L.C, G.R. Blair, and S.A. Chitwood. 1993. Approaches to supplementing coho salmon in the Queets River, Washington. In: Proceedings of the 1992 coho workshop. Nanaimo, B.C. pp.104-119. Lewis, C.A., N.P. Lester, A.D. Bradshaw, J.E. Fitzgibbon, K. Fuller, L. Hakanson, and C. Richards. 1996. Considerations of scale in habitat conservation and restoration. Can. J. Fish. Aquat. Sci. 53(Suppl.): 440-445. Lister, D.B., R.J. Finnigan. 1997. Rehabilitating off-channel habitat. In: P.A. Slaney and D. Zaldokas. [eds.] Fish habitat rehabilitation procedures. Watershed Restoration Technical Circular No. 9. Lobon-Cervia, J., and CG. Utrilla. 1993. A simple model to determine stream trout (Salmo trutta L.) densities based on one removal with electrofishing. Fisheries Research 15: 369-378. Magnuson, J.J. 1962. An analysis of aggressive behaviour, growth and competition for food and space in medaka. Can. J. Zoo. 40: 313-363. Marschall, E.A., and L.B. Crowder. 1995. Density-dependent survival as a function of size in juvenile salmonids in streams. Can. J. Fish. Aquat. Sci. 52: 136-140. Marshall, D.E., and E.W. Britton. 1990. Carrying capacity of coho salmon streams. Can. Man. Rep. Fish. Aquat. Sci. 2058. Mason, J.C 1976. Response of underyearling coho salmon to supplemental feeding in a natural stream. J. Wildl. Manage. 40: 775-788. 138 Mason, J.C., and D.W. Chapman. 1965. Significance of early emergence, environmental rearing capacity, and behavioral ecology of juvenile coho salmon in stream channels. J. Fish. Res. Bd. Can. 22: 173-190. 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. Can. J. Fish. Aquat. Sci. 48:2184-2190. McFadden, J.T. and E.L. Cooper. 1962. An ecological comparison of six populations of brown trout (Salmo trutta). Trans. Amer. Fish. Soc. 91: 53-62. MacMahon, T.E., and G.F. Hartman. 1989. Influence of cover complexity and current velocity on winter habitat use by juvenile coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 46: 1551-1557. Metcalfe, N.B., and J.E. Thorpe. 1992. Anorexia and defended energy levels in over-wintering juvenile salmon. J. Anim. Ecol. 61: 175-181. Minns, C.K., J.R.M. Keslo, and R.G. Randall. 1996. Detecting the response of fish to habitat alterations in freshwater ecosystems. Can. J. Fish. Aquat. Sci. 53(Suppl.): 403-414. Mouldey, S., and K. Ashley. 1996. Development and testing of slow release fertilizer for restoring salmon habitat. Province of British Columbia, Ministry of Environment, Lands and Parks. Fisheries Project Report No. RD 54. Mundie, K.S. Simpson, and C.J. Perrin. 1991. Responses of stream periphtyon and benthic insects to increases in dissolved inorganic phosphorus in a mesocosm. Can. J. Fish. Aquat. Sci. 48: 2061-2072. Murphy, M.L., and W.R. Meehan. 1991. Stream ecosystems. In: W.R. Meehan [ed.] Influences of forest and rangeland management on salmonid fishes and their habitats. Amer. Fish. Soc. Spec. Publ. 19, Bethesda, MD. Murphy, M.L., J. Heifetz, S.W. Johnson, K.V. Koski, and J.F. Thedinga 1984. A stream ecosystem in an old-growth forest in southeastern Alaska: part V: Seasonal changes in habitat utilization by juvenile salmonids. Pages 89-98 In: W.R. Meehan, T.R. Merrell, Jr., and T.A. Haney [eds.] Proc. Symposium on Fish and Wildlife Relationships in Old-Growth Forests, Juneau, Alaska, April 1982. Amer. Inst. Fish. Res. Biol: pp. 12-15. Murphy, M.L., CP. Hawkins, and N.H. Anderson. 1981. Effects of canopy modification and accumulated sediment on stream communities. Trans. Amer. Fish. Soc. 110: 469-478. 139 Nehlsen, W., J.E. Williams, and J. Lichatowich. 1991. Pacific salmon at the crossroads: stocks at risk from California, Oregon, Idaho, and Washington. Fisheries 16: 4-21. Neilson, J.D., and G.H. Geen. 1986. First year growth rate of Sixes River chinook salmon as inferred from otoliths: effects on mortality and age at maturity. Trans. Amer. Fish. Soc. 115: 28-33. Nickelson, T.E., and P.W. Lawson. 1998. Population viability of coho salmon (Oncorhynchus kisutch), in Oregon coastal basins: application of a habitat-based life cycle model. Can. J. Fish. Aquat. Sci. 55: 2383-2392. Nickelson, T.E., M.F. Solazzi, S.L. Johnson, and J.D. Rodgers. 1993. An approach to determining stream carrying capacity and limiting habitat for coho salmon. In: Proceedings of the 1992 coho workshop. Nanaimo, B.C. pp. 251-260. Nickelson, T.E., J.D. Rodgers, S.L. Johnson, and M.F. Solazzi. 1992. Seasonal changes in habitat use by juvenile coho salmon (Oncorhynchus kisutch) in Oregon coastal streams. Can. J. Fish. Aquat. Sci. 49: 783-789. Nickelson, T.E., M.F. Solazzi, and S.L. Johnson. 1986. Use of hatchery coho salmon (Oncorhynchus kisutch) pre-smolts to rebuild wild populations in Oregon coastal streams. Can. J. Fish. Aquat. Sci. 43: 2443-2449. Oliver, J.D., G.F. Holeton, and K.E. Chua. 1979. Overwinter mortality of fingerling smallmouth bass in relation to size, relative energy stores, and environmental temperature. Trans. Amer. Fish. Soc. 108: 130-136. Organization for economic co-operation and development. 1982. Eutrophication of waters: monitoring, assessment and control. OECD Publications, Paris, p. 138-139. Paul, A.J., C.W. Koning, K.I. Ashley, P.A. Slaney, P.W. Davidson, and R.W. Land. 1996. Development of a premier northern fishery: Mesilinka River, the first year of fertilization (1994). B.C. Ministry of Environment, Lands and Parks, Fisheries Project Report No. RD 50. Perrin, C.J., M.L. Bothwell, and P.A. Slaney. 1987. Experimental enrichment of a coastal stream in British Columbia: effects of organic and inorganic nutrient addition on autotrophic periphyton production. Can. J. Fish. Aquat. Sci. 44: 1247-1256. Peterman, R.M. 1982. Model of salmon age structure and its use in preseason forecasting and studies of marine survival. Can. J. Fish. Aquat. Sci. 39: 1444-1452. Peterson, B.J., B. Fry, and L. Deegan. 1993. The trophic significance of epilithic algal production in a fertilized tundra river ecosystem. Limnol. Oceanogr. 38(4): 872-878. 140 Peterson B.J., J.E. Hobbie, T.L. Corliss, and K. Kriet. 1983. A continuous-flow periphyton bioassay: tests of nutrient limitation in a tundra stream. Limnol. Oceanogr. 28:583-591. Peterson, N. P. 1985. Riverine pond enhancement project: October 1982 - December 1983. Wash. Dep. Fish, and Wash. Dep. Nat. Resour. Prog. Rep. 233, Olympia, WA. 48 p. Peterson, N.P., and L.M. Reid. 1984. Wall-base channels: their evolution distribution and use by juvenile coho salmon in the Clearwater River, Washington. In: J.M. Walton, and D.B. Houston [eds.] Olympic wild fish conference. Peninsula College, Port Angeles, WA. Peterson, N.P. 1982. Immigration of juvenile coho salmon (Oncorhynchus kisutch) into riverine ponds. Can. J. Aquat. Sci. 39: 1308-1310. Post, J.R., and A.B. Prankevicius. 1987. Size-selective mortality in young-of-the-year yellow perch (Perca JIavescens): evidence from otolith microstructure. Can. J. Fish. Aquat. Sci. 44:1840-1847. Post, J.R., and D.O. Evans. 1989. Size-dependent over-winter mortality of young-of-the-year yellow perch (Perca flavescens): laboratory, in situ enclosure, and field experiments. Can. J. Fish. Aquat. Sci. 46:1958-1968. Power, M.E. 1990. Effects offish in river food webs. Science 250: 811-813. Ptolemy, R.A. 1993. Maximum salmonid densities in fluvial habitats in British Columbia. Pages 223-250 In L. Berg and P.W. Delaney, [eds.] Proceedings of the Coho Workshop, Nanaimo, B.C., May 26-28, 1992. Pucket, K.J., and L.M. Dill. 1985. The energetics of feeding territoriality in juvenile coho salmon (Oncorhynchus kisutch). Behaviour. 92:97-111. Quinn, T.P., and Peterson, N.P. 1996. The influence of habitat complexity and fish size on over-winter survival and growth of individually marked juvenile coho salmon (Oncorhynchus kisutch) in Big Beef Creek, Washington. Can. J. Fish. Aquat. Sci. 53: 1555-1564. Rabeni, C.F., and S.P. Sowa. 1996. Integrating biological realism into habitat restoration and conservation strategies for small streams. Can. J. Fish. Aquat. Sci. 53(Suppl.): 252-259. Rabeni, C.F., and G.W. Minshall. 1977. Factors influencing the microdistribution of stream benthos. Oikos. 29: 33-43. 141 Rand, P.S., C.A.S. Hall, W.H. McDowell, N.H. Ringler, and J.G. Kennen. 1992. Factors limiting primary productivity in Lake Ontario tributaries receiving salmon migrations. Can. J. Fish. Aquat. Sci. 49: 2377-2385. Reeves, G.H., J.D. Hall, T.D. Roehlofs, T.L. Hickman, and CO. Baker. 1991. Rehabilitating and modifying stream habitats. In: W.R. Meehan [ed.] Influences of forest and rangeland management on salmonid fishes and their habitats. Amer. Fish. Soc. Spec. Publ. 19, Bethesda, MD. Reiman, B.E., and D.L. Myers. 1992. Influence of fish density and relative productivity on growth of kokanee in ten oligotrophic lakes and reservoirs in Idaho. Trans. Amer. Fish. Soc. 121:178-191. Reimers, N. 1963. Body condition, temperature and over-winter survival of hatchery-reared trout in Convict Creek, California. Trans. Amer. Fish. Soc. 92: 39-46. Ricker, W.E. 1975. Computation and interpretation of biological statistics of fish populations. Bulletin Fish. Res. Bd. Can. 191: 382 p. Ricker, W.E. 1969. Effects of size-selective mortality and sampling bias on estimates of growth, mortality, production, and yield. J. Fish. Res. Bd. Canada 26: 479-541. Riley, S.C and J. Korman. 1995. Juvenile salmonid monitoring in the Cheakamus River: results of a pilot study and recommendations for monitoring design. Unpublished report prepared for B.C. Hydro (June 28, 1995). 15 pages + figures and tables. Riley, S.C, and K.D. Fausch. 1995. Trout population response to habitat enhancement in six northern Colorado streams. Can. J. Fish. Aquat. Sci. 52: 34-53. Rosemond, A.D., P.J. Mulholland, and J.W. Elwood. 1993. Top-down and bottom-up control of stream periphyton: effects of nutrients and herbivores. Ecology 74: 1264-1280. Rosenfeld J S. 1997. The influence of upstream predation on the expression offish effects in downstream patches. Fresh. Bio. 37: 535-543. Ruggerone, G.T., and D.E. Rogers. 1992. Predation on sockeye salmon fry by juvenile coho salmon in the Chignik Lakes, Alaska: implications for management. N. Amer. J. Fish. Mgmt. 12: 87-102. Ruggles, CP. 1966. Depth and velocity as a factor in stream rearing and production of juvenile coho salmon. Can. Fish. Cult. 38: 37-53. Sawada, J. O. 1993. An examination of differential survival in downstream migrating coho salmon smolts. MSc. thesis. University of British Columbia, Vancouver, B.C. 142 Scrivener, J.C., and B.C. Andersen. 1984. Logging impacts and some mechanisms that determine the size of spring and summer populations of coho salmon fry (Oncorhynchus kisutch) in Carnation Creek, British Columbia. Can. J. Fish. Aquat. Sci. 41:1097-1105. Sheng, M., M. Foy, and A.Y. Fedorenko. 1990. Coho salmon enhancement in British Columbia using improved groundwater-fed side channels. Can. Man. Rep. Fish. Aquat. Sci. 2071:45 p. Shirvell, C.S. 1990. Role of instream rootwads as juvenile coho salmon (Oncorhynchus kisutch) and steelhead trout (O. mykiss) cover habitat under varying streamflows. Can. J. Fish. Aquat. Sci. 47: 852-861. Shuter, B.J., J.A. MacLean, F.E. Fry, and H.A. Regier. 1980. Stochastic simulation of temperature effects on first-year survival of smallmoufh bass. Trans. Amer. Fish. Soc. 109: 1-34. Slaney, P.A. and B.R. Ward. 1993. Experimental fertilization of nutrient deficient streams in British Columbia. In: G. Shooner, et S. Asselin (es.) Le developpement du Saumon atlantique au Quebec: connaitre les regies du jeu pour reussir. Colloque international de la Federation quebecoise por le saumon atlantique. Quebec, Decembre 1992.. Collection Salmo salar 1: pp. 128-141. Slaney T.L., K.D. Hyatt, T.G. Northcote, and R.J. Fielden. 1996. Status of anadromous salmon and trout in British Columbia and Yukon. Fisheries. 21(10): 20-35. Smith, R.W., and J.S. Griffith. 1994. Survival of rainbow trout during their first winter in the Henry's Fork of the Snake River, Idaho. Trans. Amer. Fish. Soc. 123: 747-756. Smith, J.J., and H.W. Li. 1983. Energetic factors influencing foraging tactics of juvenile steelhead trout, (Salmo gairdneri). pp. 173-180 In: D.L.G. Noakes, D.G. Lindquist, G.S. Helfman, and J.A. Ward [eds.] Predators and prey in fishes. Dr. W.J. Junk, The Hague, Netherlands. Smoker, W.A. 1955. Effects of streamflow on silver salmon production in western Washington. Ph.D. dissertion, University of Washington, Seattle, Wash. Sokal, R.R., and F.J. Rohlf. 1981. Biometry, principles and practices of statistics in biological research, 2nd edition. W.H. Freeman Co., New York. 859 p. Stockner, J.G., and K.R.S. Shortreed. 1978. Enhancement of autotrophic production by nutrient addition in a coastal rainforest stream on Vancouver Island. J. Fish. Res. Board. Can. 35: 28-34. 143 Strickland, J.D.H. and T.R. Parsons. 1972. A practical handbook of seawater analysis. Bull. Fish. Res. Board Can. 67: 311 p. Swales, S., and CD. Levings. 1989. Role of off-channel ponds in the life cycle of coho salmon (Oncorhynchus kisutch) and other juvenile salmonids in the Coldwater River, British Columbia. Can. J. Fish. Aquat. Sci. 46: 232-242. Swales, S., R.B. Lauzier, and CD. Levings. 1986. Winter habitat preferences for juvenile salmonids in two interior rivers in British Columbia. Can. J. Fish. Aquat. Sci. 64: 1506-1514. Swales, S., F. Caron, J.R. Irvine, and CD. Levings. 1988. Population characteristics of coho salmon (Oncorhynchus kisutch) and other juvenile salmonids overwintering in lakes, tributary stream and main river habitats in the Keogh River system, Vancouver Island, British Columbia. Can. J. Zool. 66: 254-261. SYSTAT. 1997. SYSTAT Users Guide, version 7.0. SPSS Inc., Chicago, IL. Taylor, E.B. 1988. Water temperature and velocity as determinants of microhabitats of juvenile chinook and coho salmon in a laboratory stream channel. Trans. Amer. Fish. Soc. 117: 22-28. Thedinga, J.F., and K.V. Koski. 1984. A stream ecosystem in an old-growth forest in southeast Alaska. Part VI: The production of coho salmon (Oncorhynchus kisutch) smolts and adults from Porcupine Creek, pp. 99-108. In: W.R. Meehan, T.R. Merrell, Jr., and T.A. Hanly [eds.] Fish and wildlife relationships in old-growth forests, proceedings of a symposium held April 1982, Juneau, AK. Am. Inst. Fish. Res. Biol., Juneau, AK. Thedinga, J.F., M.L. Murphy, J. Heifetz, K.V. Koski, and S.W. Johnson. 1989. Effects of logging on size and age composition of juvenile coho salmon (Oncorhynchus kisutch) and density of pre-smolts in southeast Alaska streams. Can. J. Fish. Aquat. Sci. 46: 1383-1391. Toth, B., K. Ashley, G. Wilson, C.W. Koning, P. Slaney, and R.W. Land. 1996. Development of a premier river fishery: Big Silver Creek fertilization experiment, year two (1995) of low-level inorganic nutrient addition. B.C. Ministry of Environment, Lands and Parks, Fisheries Project Report No. RD 56. Tschaplinski, P.J. 1982. Aspects of the population biology of estuary-reared and stream-reared juvenile coho salmon in carnation creek: a summary of current research. In: G.F. Hartman [ed.] Proceedings of the Carnation Creek work-shop, a ten year review. Pacific Biological Station, Nanaimo, B.C: pp. 289-307. 144 Tschaplinski, P.J., and G.F. Hartman. 1983. Winter distribution of juvenile coho salmon (Oncorhynchus kisutch) before and after logging in Carnation Creek, British Columbia, and some implications for over-winter survival. Can. J. Aquat. Sci. 400:452-461. Van Nieuwenhuyse, E.E., and J.R. Jones. 1996. Phosphorus-chlorophyll relationship in temperate streams and its variation with stream catchment area. Can. J. Fish. Aquat. Sci. 53:99-105. Walters, C.J. 1993. Where have all the coho gone? In: Proceedings of the 1992 coho workshop. Nanaimo, B.C. Pages 1-8 In L. Berg and P.W. Delaney, [eds.] Proceedings of the Coho Workshop, Nanaimo, B.C., May 26-28, 1992. Walters, C.J. and P. Cahoon. 1985. Evidence of decreasing spatial diversity in British Columbia salmon stocks. Can. J. Fish. Aquat. Sci., 42(5): 1033-1037. Walters, C.J., and J.R. Post. 1993. Density-dependent growth and asymmetries in size-structured fish populations: a theoretical model and recommendations for field experiments. Trans. Amer. Fish. Soc. 112: 35-45. Ward, B.R., and P.A. Slaney. 1988. Life history and smolt-to-adult survival of Keogh River steelhead trout (Salmon gairdneri) and the relationship to smolt size. Can. J. Fish. Aquat. Sci. 45: 1110-1122. Warren, C.E., J. Wales, G. Davis, and P. Doudoroff. 1964. Trout production in an experimental stream enriched with sucrose. J. Wildl. Manage. 28:617-660. Warren, W.G. 1994. Removal estimates of Atlantic salmon parr: maximum likelihood and bayesian methods. ICES C.M.1994/D:8 Werner, E.E. 1986. Species interactions in freshwater fish communities. In: J.M. Diamond, T.J. Case [eds.] Community ecology. Harper and Row, New York, NY: pp. 344-357. White, R.J. 1996. Growth and development of North American stream habitat management for fish. Can. J. Fish. Aquat. Sci. 53(Suppl.): 342-363. Williams, D.D., and J.H. Mundie. 1978. Substrate size selection by stream invertebrates and the influence of sand. Limnol. Oceanogr. 23: 1030-1033. Wipfli, M.S., J. Hudson, and J. Caouette. 1998. Influence of salmon carcasses on stream productivity: response of biofilm and benthic macro invertebrates in southeastern Alaska, U.S.A. Can. J. Fish. Aquat. Sci. 55: 1503-1511. 145 Wood, C.C. 1997. Predation of juvenile Pacific salmon by the common merganser (Mergus merganser) on eastern Vancouver Island. Part 2: Predation of stream-resident juvenile salmon by merganser broods. Can. J. Fish. Aquat. Sci. 44: 950-959. Zar, J.H. 1974. Biostatistical analysis. Prentice-Hall, Inc. Englewood Cliffs, N.J. 620 p. 146 "@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "1999-11"@en ; edm:isShownAt "10.14288/1.0089052"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Forestry"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en ; ns0:scholarLevel "Graduate"@en ; dcterms:title "Effects of primary production and other factors on the size and abundance of juvenile coho salmon in artificial off-channel habitat"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/9267"@en .