Columbia River Project Water Use Plan

Transcription

1 Columbia River Project Water Use Plan Lower Columbia River (Year 1) Reference: CLBMON#45 Columbia River Water Use Plan Monitoring Program: Lower Columbia River Fish Indexing Survey Study Period: 27 Dustin Ford, Robyn Irvine, Joseph Thorley, Dana Schmidt, Larry Hildebrand April 28

2 LARGE RIVER FISH INDEXING PROGRAM LOWER COLUMBIA RIVER 27 PHASE 7 INVESTIGATIONS

3 LARGE RIVER FISH INDEXING PROGRAM LOWER COLUMBIA RIVER 27 PHASE 7 INVESTIGATIONS - FINAL REPORT - - CLBMON-45 - Prepared for BC HYDRO Power Supply Environmental Services 6911 Southpoint Drive (E16) Burnaby, British Columbia V3N 4X8 by GOLDER ASSOCIATES LTD. 21 Columbia Avenue Castlegar, B.C. V1N 1A8 Phone: (25) Fax: (25) April 28

4 Cover photo: David DeRosa captures an index species fish while boat electroshocking in Site ES21 (near Hugh L. Keenleyside Dam), 5 November 27. Suggested Citation: Golder Associates Ltd. 28. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations. Report prepared for BC Hydro, Burnaby, B.C. Golder Report No F: 78 p. + 6 app

5 EXECUTIVE SUMMARY In 27, BC Hydro commissioned Phase 7 of the Large River Fish Indexing Program. The primary purpose of the 27 Phase 7 study was to monitor the life history characteristics, distribution, and population abundance of selected index species in the Lower Columbia River, and compare these results to earlier phases (i.e., 21 to 26) of the program. The Lower Columbia River study area included the Columbia River from the outlet of Hugh L. Keenleyside Dam (HLK) downstream to the Canada-U.S. border. Within the study area, sampling was conducted in four main areas: the upper section from HLK (Km.) downstream to Blueberry Creek (Km 23.); the middle section from Blueberry Creek downstream to West Trail (Km 4.); the lower section from East Trail downstream to the Canada-U.S. border (Km 56.5); and, the Kootenay section from Brilliant Dam downstream to the mouth of the Kootenay River (Km. to Km 2.8). The main field data collection objectives involved: assessments of fish abundance and distribution; the collection of life history data including ageing structures; and, a mark-recapture program to develop fish population estimates. Fish were sampled by boat electroshocking at night within nearshore habitats (less than 3.5 m depth). The three fish species selected as index species for detailed study were mountain whitefish (Prosopium williamsoni), rainbow trout (Oncorhynchus mykiss), and walleye (Sander vitreus). Length, weight, and ageing structures were collected from rainbow trout and mountain whitefish; walleye were only sampled for length and weight (i.e., ageing structures were not collected). All healthy fish caught were marked with a Plastic Infusion Process (PIP) Passive Integrated Transponder (PIT) tag. Data for each index species were analyzed using species percent composition, catch-per-unit-effort, length-frequency, age-frequency, growth rate, survival, age-cohort analysis, Age-Structured Mark-Recapture (ASMR) analysis (conducted by Poisson Consulting Ltd.), and population abundance estimation techniques to discern trends in index species populations. These data also were compared to results from Phases 1 through 6, and where appropriate, to results from studies conducted in the study area in the early to mid-199s. The key findings of the Phase 7 study are summarized as follows: There appeared to be lower age- mountain whitefish recruitment in 23, 24, and 25 relative to 21, 22, 26, and 27. This difference is possibly related to increased predation by walleye on the age- cohort over that time period. The ASMR analysis suggests a substantial decline in the adult mountain whitefish population since 23. The uncertainty surrounding the ASMR estimates of the number of mountain whitefish recruits precludes any clear conclusions about changes in abundance. The Program MARK analysis on fish greater than the mean length-at-age for age-2+ noted the same decline in the adult population. Rainbow trout age- recruitment was lowest from 23 to 25 and was also low in 27. These reductions may be partially due to predation pressure or to other environmental factors. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page i

6 Walleye populations showed one mode between 351 and 45 mm FL, composed of four separate age-cohorts (the 21 to 24 brood years). The oldest and most numerous of these four cohorts was the 21 brood year, first observed in 23. Ageing rainbow trout and mountain whitefish scales continues to be problematic due to the presence of false annuli or lack of annuli. Population estimates for walleye based on the modified Schnabel, sequential Bayes algorithm, and Program MARK estimation techniques suggest an increase in abundance from 22 to 23, a decrease in abundance from 23 to 26, and an increase in abundance from 26 to 27. Based on catch-rates, relative abundance, and three population abundance estimate models, mountain whitefish abundance increased between 26 and 27, likely due to higher recruitment from the 26 brood year. All estimates exhibited wide confidence intervals due to the low number of recaptured fish. For mountain whitefish, average annual growth rates, inter-year recapture rates, and inter-year survival rates were higher for fish marked with PIT tags than for fish marked with T-bar anchor tags. Based on catch-rates, relative abundance, and the population abundance estimate models, rainbow trout abundance increased from 25 to 27. The ASMR analysis supports a moderate overall increase in adults and a more substantial increase in recruits. One of the objectives of the Phase 7 program was to provide recommendations for the Phase 8 (28) component of the program. In consideration of the findings provided above and the overall objectives of the Large River Fish Indexing Program in the Lower Columbia River, the following recommendations are provided: The nearshore sample program should be repeated using methods similar to the 27 study. The nearshore sample program should continue to be conducted in September and October when water temperatures are below 15 C. Mountain whitefish, rainbow trout, and walleye should continue to be the index species used for detailed population parameter analyses. Scales should be collected from all mountain whitefish and rainbow trout; however due to problems associated with ageing rainbow trout scales, rainbow trout ageing should be limited to inter-year recaptured individuals. There should be an increased focus on scales collected from older fish to reduce uncertainty in fitting mixed distributions to the sparsely sampled older age-classes. Compressed oxygen, pumped into the livewell through an airstone, should continue to be used to maintain dissolved oxygen levels similar to levels recorded in the river. A commercially available livewell additive (Rejuvenade TM HRT) also should be used. The methods used in 24 to 27 to reduce the frequency of electroshocking-induced injuries on large (>3 mm FL) rainbow trout should be maintained. PIT tags should continue to be used to mark index species fish and 8.5 mm PIT tags should be tested to mark smaller (i.e., <12 mm FL) fish. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page ii

7 Population age structures for rainbow trout and mountain whitefish should be developed using a combination of length-frequency data, capture efficiency data from the mark-recapture program, and size-at-age data. Analyses using the sequential Bayes Algorithm and modified Schnabel methods should be reduced, and increased effort should be allocated to ASMR and Program Mark modelling as these two methods show encouraging results. In future years, ASMR model fits should be examined with residuals and the possibility of using random effects to reduce the parameter set should be explored. If emigration/immigration vary through time in the Program MARK analysis, the temporal aspect should be incorporated into ASMR modelling. The ASMR models could be modified to include multiple capture sessions within a year. Different recapture rates between adult and immature fish should be incorporated into Program Mark rather than testing recapture rate differences with chi-squared tests. This should ensure that any weighting that estimates survival differently from adult to immature fish will be reflected in the averaged parameters regardless of whether chi-squared values are statistically significant. The ASMR methods show major differences in recruitment rates occurring over time; consequently these methods should be explored for applicability to earlier data sets to further expand the time series and improve the ability to relate recruitment rates to flow changes. The increase in rainbow abundance estimates corresponds with an increased abundance of spawners. Additional analyses with an expanded data set should examine the relationship between flow and other environmental parameters to recruitment changes adjusted for spawner-recruit relationships. To better characterize the variability in index species population abundance for the Lower Columbia River, a sampling program should be continued that will allow estimation of cohort strength and survival for index species. Substantial differences in some population parameters have occurred in mountain whitefish and rainbow trout populations between the early to mid-199s studies and the present indexing program and within the seven year span of the present program. The four year gap in data between the 199s and the relatively short duration of the time series from the present studies confound accurate identification of the reasons for these changes. Therefore, continuation of the program is strongly recommended. The ASMR approach has provided some indication about rainbow trout recruitment during the late 199s. Using a similar ASMR approach on the 199s dataset may help us understand the relationship between low recruitment rates, HLK operations, and other environmental variables. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page iii

8 ACKNOWLEDGEMENTS Phase 7 (27) of the Large River Fish Indexing Program for the Lower Columbia River was funded by BC Hydro in Castlegar, B.C. Golder Associates Ltd. would like to thank the following individuals for their contributions to the program: BC Hydro Paul Higgins Brent Mossop David DeRosa Bob Westcott Burnaby, B.C. Burnaby, B.C. Castlegar, B.C. Castlegar, B.C. B.C. Ministry of Environment Jeff Burrows Colin Spence Jolene Raggett Nelson, B.C. Nelson, B.C. Nelson, B.C. Golder Associates Ltd. would especially like to thank Dr. Joseph Thorley and Poisson Consulting Ltd. for conducting the Age-Structured Mark-Recapture (ASMR) analysis portion of the program. We also thank Lew Coggins for providing us with important information about ASMR and Lew Coggins and Steve Martell for providing us with code. The following employees of GOLDER ASSOCIATES LTD. contributed to the collection of data and preparation of this report. Dustin Ford, R.P. Bio. Larry Hildebrand, R.P. Bio. Dana Schmidt, R.P. Bio. Ph.D. Robyn Irvine, R.P. Bio. Ph.D. Lynn Westcott Paul Grutter, R.P. Bio. Demitria Burgoon Carissa Canning Chris King Edward Lem Steve Whitehead Teal Moffat Ron Giles Fred Salekin Carrie McAllister Project Biologist/Author Senior Fisheries Biologist/Project Leader/Editor Senior Fisheries Biologist/Project Manager Statistical Ecologist/Contributing Author Biologist/Editor Fisheries Biologist Biological Technician Biological Technician Biological Technician Biological Technician Biological Technician Biological Technician Warehouse Technician Warehouse Technician Administrative Assistant Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page iv

9 TABLE OF CONTENTS Page # EXECUTIVE SUMMARY... i ACKNOWLEDGEMENTS... iv LIST OF TABLES... vii LIST OF FIGURES... viii 1. INTRODUCTION Background Objective and Scope Study Area METHODS Physical Parameters Discharge Temperature Habitat Conditions Study Period Fish Inventory Boat Electroshocking Fish Processing Ageing Data Analyses Data Compilation Catch and Life History Data Population Estimates Comparison of Estimated Populations Age-Cohort Analysis PIT and T-Bar Anchor Tag Growth Comparisons Age Structured Mark-Recapture Analysis Background ASMR Analysis of rainbow trout and mountain whitefish PHYSICAL CONDITIONS Columbia River Discharge Water Temperature Kootenay River Discharge Water Temperature Habitat Conditions FISH RESOURCES Fish Species Composition and Distribution Life History Characteristics Mountain Whitefish Rainbow Trout Walleye PIT and T-bar Anchor Tag Growth Comparison Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page v

10 Mountain Whitefish Rainbow Trout Walleye Other Species Population Estimates Mountain Whitefish Rainbow Trout Walleye Comparison of Estimated Populations Age-Cohort Analysis Age Structured Mark-Recapture Analysis Mountain Whitefish Rainbow Trout DISCUSSION Mountain whitefish Percent Composition, Catch-rates, and Population Abundance Rainbow trout Percent Composition, Catch-rates, and Population Abundance Walleye Percent Composition, Catch-rates, and Population Abundance PIT Vs. T-Bar Anchor Tag Comparison Program MARK ASMR Analysis CONCLUSION AND RECOMMENDATIONS CLOSURE LITERATURE CITED APPENDICES APPENDIX A APPENDIX B APPENDIX C APPENDIX D APPENDIX E APPENDIX F HABITAT SUMMARY INFORMATION MAPS AND UTM COORDINATES DISCHARGE AND TEMPERATURE DATA CATCH AND EFFORT DATA SUMMARIES LIFE HISTORY DATA PROGRAM MARK OUTPUT ATTACHMENTS (LOCATED AT BC HYDRO, CASTLEGAR) ATTACHMENT A ATTACHMENT B DIGITAL DATA CD-ROM ALL AGEING STRUCTURES COLLECTED Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page vi

11 LIST OF TABLES Page # Table 2.1 Sampling periods within the Lower Columbia River, Table 4.1 Table 4.2 Table 4.3 Table 4.4 Table 4.5 Table 4.6 Table 4.7 Table 4.8 Table 4.9 Table 4.1 Number of fish caught and observed by boat electroshocking and their frequency of occurrence in sampled sections of the Lower Columbia River, 27 September to 6 November Number of fish caught and observed by boat electroshocking and their frequency of occurrence in sampled sections of the Lower Columbia River, 21 to Average fork length and weight-at-age for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November Average fork length and average weight-at-age for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November 27. Only rainbow trout initially captured in 22 to 26 and recaptured in 27 were aged during the 27 study Summary of the total number of each index species marked and recaptured by year and tag type in sampled sections of the Lower Columbia River, 25 to 26, 26 to Average annual growth rate (mm/year) of fish captured by boat electroshocking in sampled sections of the Lower Columbia River. Data from fish captured in 25 or 26 and recaptured one year later Total number of each index species caught, marked, and recaptured, and the number of mortalities in sampled sections of the Lower Columbia River, 22 to Total number of each index species fish caught, marked, and recaptured in sampled sections of the Lower Columbia River, 27 September to 6 November Age-Structured Mark-Recapture (ASMR) model evaluation results for mountain whitefish captured in the Lower Columia River, 21 to 27. AIC: Akaike Information Criterion, which measures the goodness of fit of the model. Lower values are better Age-Structured Mark-Recapture (ASMR) model evaluation results for rainbow trout captured in the Lower Columia River, 21 to 27. AIC: Akaike Information Criterion, which measures the goodness of fit of the model. Lower values are better Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page vii

12 LIST OF FIGURES Page # Figure 1.1 Overview of the Lower Columbia River study area, Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure 4.1 Figure 4.2 Figure 4.3 Figure 4.4 Figure 4.5 Figure 4.6 Figure 4.7 Figure 4.8 Mean daily discharge (m 3 /s) for the Columbia River at the Birchbank water gauging station, 27. The shaded area represents minimum and maximum mean daily discharge values recorded at Birchbank from 21 to 26. The white line represents average mean daily discharge values over the same time period Mean daily water temperature ( C) for the Columbia River at Hugh L. Keenleyside Dam (HLK; black line), 1 August to 14 November 27. The shaded area represents minimum and maximum mean daily water temperature values recorded at HLK from 21 to 26. The white line represents average mean daily water temperature values over the same time period Mean daily water temperature ( C) for the Columbia River at Birchbank water gauging station (black line), 1 August to 14 November 27. The shaded area represents minimum and maximum mean daily water temperature values recorded at Birchbank from 21 to 26. The white line represents average mean daily water temperature values over the same time period Mean daily discharge (m 3 /s) for the Kootenay River at Brilliant Dam (black line), 1 August to 14 November 27. The shaded area represents minimum and maximum mean daily discharge values recorded at Brilliant Dam from 21 to 26. The white line represents average mean daily discharge values over the same time period Percent composition (by year and sample section) of small (<25 mm FL; black bars) and large ( 25 mm FL; grey bars) mountain whitefish captured by boat electroshocking in the Lower Columbia River, 21 to Percent composition (by sample section and year) of large ( 25 mm FL) and small (<25 mm FL) mountain whitefish captured by boat electroshocking in the Lower Columbia River, 21 to Mean catch-rate (by sample section) for mountain whitefish captured by boat electroshocking in the Lower Columbia River, 21 to 27. The dotted lines denote 95% confidence intervals. CPUE data includes captured and observed fish identified to species; all size-cohorts combined Length-frequency distributions for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November Length-frequency distributions for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 21 to Age-frequency distributions for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 21 to von Bertalanffy growth curve for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 22 to Percent composition (by year and sample section) of small (<25 mm FL; black bars) and large ( 25 mm FL; grey bars) rainbow trout captured by boat electroshocking in the Lower Columbia River, 21 to Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page viii

13 Figure 4.9 Figure 4.1 Figure 4.11 Figure 4.12 Figure 4.13 Figure 4.14 Figure 4.15 Figure 4.16 Figure 4.17 Figure 4.18 Figure 4.19 Figure 4.2 Figure 4.21 Figure 4.22 Percent composition (by sample section and year) of large ( 25 mm FL) and small (<25 mm FL) rainbow trout captured by boat electroshocking in the Lower Columbia River, 21 to Mean catch-rate (by sample section) for rainbow trout captured by boat electroshocking in the Lower Columbia River, 21 to 27. The dotted lines denote 95% confidence intervals. CPUE data includes captured and observed fish identified to species; all size-cohorts combined Length-frequency distributions for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November Length-frequency distributions for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 21 to Age-frequency distributions for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 22 to 27. Only rainbow trout initially captured in 22 to 26 and recaptured in 27 were aged during the 27 study... 4 von Bertalanffy growth curve for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 22 to 27 combined. Only rainbow trout initially captured in 22 to 26 and recaptured one year later were aged Annual growth rates for rainbow trout caught and then recaptured after one year in the Lower Columbia River, (R.L.&L. 1995b.), 22-25, and Percent composition (by year and sample section) of walleye captured by boat electroshocking in the Lower Columbia River, 21 to Percent composition (by sample section and year) of walleye captured by boat electroshocking in the Lower Columbia River, 21 to Mean catch-rate (by sample section) for walleye captured by boat electroshocking in the Lower Columbia River, 21 to 27. The dotted lines denote 95% confidence intervals. CPUE data includes captured and observed fish identified to species Length-frequency distributions for walleye captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November Length-frequency distributions for walleye captured by boat electroshocking in sampled sections of the Lower Columbia River, 21 to Summary of survival estimates with 95% confidence intervals based on Program MARK robust model estimates for fish marked with T bar anchor tags (21 to 26 combined) and fish marked with PIT tags (25 to 27 combined) Population estimates with 95% confidence intervals using the modified Schnabel, sequential Bayes algorithm, and Program MARK for index species in the Lower Columbia River, 27. The red dots denote small (i.e., <25 mm FL) fish; the blue dots denote large (i.e., 25 mm FL) fish; the black dots denote both size cohorts combined Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page ix

14 Figure 4.23 Population estimates for mountain whitefish in the Lower Columbia River study area, 22 to 27. Estimates were generated using the modified Schnabel (blue lines), sequential Bayes algorithm (red lines), and Program MARK (black lines) estimation techniques. The solid lines represent the estimates; the dotted lines represent upper and lower 95% confidence intervals Figure 4.24 Population estimates for rainbow trout in the Lower Columbia River study area, 22 to 27. Estimates were generated using the modified Schnabel (blue lines), sequential Bayes algorithm (red lines), and Program MARK (black lines) estimation techniques. The solid lines represent the estimates; the dotted lines represent upper and lower 95% confidence intervals. In 24, differences in recapture rates between the two size-cohorts of rainbow trout (i.e., fish 25 mm FL and fish <25 mm FL) prevented the generation of an all size-classes combined population estimate using the modified Schnabel and sequential Bayes Algorithm estimation techniques. Data presented represents the sum total of the two size-class population estimates Figure 4.25 Population estimates for walleye in the Lower Columbia River study area, 22 to 27. Estimates were generated using the modified Schnabel (blue lines), sequential Bayes algorithm (red lines), and Program MARK (black lines) estimation techniques. The solid lines represent the estimates; the dotted lines represent upper and lower 95% confidence intervals Figure 4.26 Figure 4.27 Figure 4.28 Figure 4.29 Figure 4.3 Figure 4.31 Figure 4.32 Figure 4.33 Figure 4.34 Changes in Bayesian population estimates for large ( 25 mm FL), small (<25 mm FL), and combined (all fish) mountain whitefish in the Lower Columbia River, 22 to 27. The solid lines denote the percent change in population abundance between years and the dotted lines denote the probability of a population decrease between years Changes in Bayesian population estimates for large ( 25 mm FL) and small (<25 mm FL) rainbow trout in the Lower Columbia River, 22 to 27. The solid lines denote the percent change in population abundance between years and the dotted lines denote the probability of a population decrease between years Changes in Bayesian population estimates for walleye (all size-cohorts combined) in the Lower Columbia River, 22 to 27. The solid lines denote the percent change in population abundance between years and the dotted lines denote the probability of a population decrease between years Cumulative probability curves for a possible change in population size for mountain whitefish in the Lower Columbia River (based on changes in Bayesian population estimates) Cumulative probability curves for a possible change in population size for rainbow trout in the Lower Columbia River (based on changes in Bayesian population estimates) Cumulative probability curves for a possible change in population size for walleye in the Lower Columbia River (based on changes in Bayesian population estimates) Predicted length-at-age distribution curve for mountain whitefish captured by boat electroshocking in the Lower Columbia River, 27 September to 6 November 27. Graph includes age- to age-9+ (i.e., age-9 and older) cohorts...59 Predicted length-at-age distribution curve for rainbow trout captured by boat electroshocking in the Lower Columbia River, 27 September to 6 November 27. Graph includes age- to age-7+ (i.e., age-7 and older) cohorts Adult (age-3+) mountain whitefish abundance (black line) as estimated by ASMR 1 with a migration parameter of.4. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page x

15 Figure 4.35 Figure 4.36 Figure 4.37 Figure 4.38 Figure 4.39 Figure 5.1 Figure 5.2 Figure 5.3 Mountain whitefish recruits (age-1; black line) as estimated by ASMR 1 with a migration parameter of.4. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing Adult (age-2+) rainbow trout abundance (black line) estimated by ASMR 2 with a migration parameter of.. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing Adult (age-2+) rainbow trout abundance (black line) estimated by ASMR 2 with a migration parameter of.48. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing Rainbow trout recruits (age-1; black line) as estimated by ASMR 2 with a migration parameter of.. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing Rainbow trout recruits (age-1; black line) as estimated by ASMR 2 with a migration parameter of.48. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing Population estimates, catch-rates, abundance, and ASMR age-3+ abundance of mountain whitefish captured in the Lower Columbia River, 22 to Population estimates, catch-rates, abundance, and ASMR age-2+ abundance of rainbow trout captured in the Lower Columbia River, 22 to Summary comparison of population estimates, catch-rates, and abundance of walleye captured in the Lower Columbia River, 22 to Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page xi

16 1. INTRODUCTION BC Hydro recognizes the importance of defining impacts of the operation of its dams and reservoirs on fish populations in flow regulated watersheds to ensure that operations are both environmentally and economically sustainable. In recognition of this need and of the challenges associated with monitoring fish populations in large rivers, BC Hydro initiated the Large River Fish Indexing Program in 21 in the Peace and Columbia river watersheds. The primary components of the indexing program are: the development of reliable and cost-effective methods for indexing the fish community; investigations to identify and fill data gaps associated with life history and habitat use of species of special concern; and, the implementation of specialized monitoring activities in relation to identified species of concern. The objective of the program is to establish a fish monitoring protocol that can be used reliably across the Peace and Columbia river watersheds to provide an index of the general status (e.g., trends in population abundance, health, growth) of the fish community. The scope of the program is to: develop a standardized fish sampling methodology for indexing trends in the fish community in terms of abundance, distribution, and biological characteristics of the fish populations; advance an analytical framework for using monitoring data to assess how habitat changes due to dam or reservoir operations influence abundance, distribution, and biological characteristics of fish populations; assess the costs and benefits of implementing an index monitoring program in each watershed; and, provide recommendations and plans for subsequent implementation of long-term fish community monitoring programs in the Peace and Columbia River watersheds. 1.1 Background During the past three decades, BC Hydro has conducted numerous studies related to the effects of flow regulation on fish populations in the Columbia River Basin. Many of these studies were conducted on the Lower Columbia River (LCR), defined as the section of river between Hugh L. Keenleyside Dam (HLK) and the Canada-U.S. border (see R.L. & L. 21 for a complete summary). These studies were designed to obtain information either on the status of the general fish population or on the abundance of specific species with the objective of using the information to identify, and in some instances quantify, the effects of dam and reservoir operations. Many of these studies were reactive, in that they generally occurred after some type of major perturbation of the system due to flow regulation. The results of these studies, while often suggesting that impacts may have occurred, were of limited value in quantifying the biological significance of these impacts. This was mainly due to the lack of accurate, long-term baseline data, and because the studies were not systematically designed to quantify effects on fish at the population level. In 199, BC Hydro initiated a five-year study program that involved an intensive inventory program to collect information on fish populations in the Lower Columbia Development Area (Hildebrand and English 1991; R.L. & L. 1993a, 1993b, 1995a, 1995b, 1995c, 1995d). This information was to be used to predict the impacts of Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 1

17 proposed potential hydroelectric projects, and to assist in developing mitigation and compensation programs to reduce or offset any impacts of these projects on the aquatic resources in the LCR. These early studies focussed mainly on basic life history characteristics, distribution, population abundance estimates, and identification of important habitats. In 1995, BC Hydro initiated stabilized water releases from HLK during the mountain whitefish (Prosopium williamsoni) and rainbow trout (Oncorhynchus mykiss) spawning seasons to minimize egg losses in the LCR. During the mountain whitefish spawning season (December to February), BC Hydro reduces flow from HLK during the peak spawning period to encourage spawning at lower water level elevations. Subsequently, flows are managed (i.e., within the constraints of the Columbia River Treaty and food protection considerations) to provide stable or increasing water levels after this period, protecting the eggs until emergence. For the same reason, flows from HLK are reduced during the middle of the rainbow trout spawning season (April and May) to encourage the bulk of rainbow trout spawners to spawn at lower water level elevations. In 21, BC Hydro initiated Phase 1 of the Large River Fish Indexing Program to gather baseline information on fish distribution, abundance, life history characteristics, and population abundance of selected index species (Golder 22). Phase 2 (conducted in 22; Golder 23), Phase 3 (conducted in 23; Golder 24), Phase 4 (conducted in 24; Golder 25), Phase 5 (conducted in 25; Golder 26), and Phase 6 (conducted in 26; Golder 27), represented the continued development of a systematic, repetitive index sampling program for the LCR. The current 27 study (Phase 7) represents a continuation of the study plan designed and established during earlier phases of the program. Ideally, the data collected by this program will allow the calculation of fish population parameters at a level of resolution that can be used to identify changes to fish populations and assist in the determination of the biological and statistical significance of these changes in relation to mountain whitefish and rainbow trout spawning protection flows. In 27, BC Hydro completed the Water Use Planning process for its hydroelectric and storage facilities on the Columbia River. The Water Use Plan (WUP) may result in changes to the operation of HLK, which in turn will require monitoring to identify and assess the effects of altered operational parameters on the downstream aquatic ecosystem. The results of the Large River Fish Indexing Program for the LCR will assist in the development of monitoring tools to determine the effects of operational changes on downstream fish communities. 1.2 Objective and Scope The continuation of the Large River Fish Indexing Program will provide data to document the relative abundance, distribution, and biological characteristics of fish populations below hydroelectric facilities in the Columbia and Peace river drainages. The key management questions to be addressed by Phase 7 (current), and Phases 8 and 9 (scheduled for implementation in 28 and 29, respectively) for the Lower Columbia River study area are: Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 2

18 1. What is the abundance, growth rate, survival rate, body condition, age distribution and spatial distribution of sub-adult and adult mountain whitefish, rainbow trout, and walleye (Sander vitreus) in the LCR? 2. What is the effect of inter-annual variability in the mountain whitefish and rainbow trout flow regimes on the abundance, growth rate, survival rate, body condition, and spatial distribution, of sub-adult and adult mountain whitefish, rainbow trout, and walleye in the LCR? Specific objectives of Phase 7 (27) of the Lower Columbia River Large River Fish Indexing Program (as summarized from the Terms of Reference) were: to extend time series data on the abundance, distribution, and biological characteristics of nearshore fish populations in the Lower Columbia River; to examine long term trends in key index fish populations (mountain whitefish, rainbow trout, and walleye) during the continued implementation of mountain whitefish and rainbow trout protection flows in the Lower Columbia River; to build upon previous investigations for the further refinement of sampling strategy, sampling methodology, and analytical procedures required to establish a long term monitoring program for fish populations in the Lower Columbia River; to update the existing electronic storage and retrieval system for fish population and habitat monitoring data for the Lower Columbia River; to establish linkages between other biological monitoring programs being undertaken in the Lower Columbia River, in particular, the Physical Habitat and Ecological Productivity Monitoring Program; and, to identify data gaps regarding fish populations and procedures for sampling them and to provide recommendations for future monitoring and fisheries investigations. 1.3 Study Area The study area for the Phase 7 (27) program encompassed the approximately 56 km section of the riverine habitat between HLK and the Canada-U.S. border (Figure 1.1). This study area (collectively called the Lower Columbia River) included the Kootenay River below the Brilliant Dam and the Columbia-Pend d Oreille rivers confluence below Waneta Dam. For the purpose of this study, the mainstem Columbia River was divided into three relatively equal sections; these were termed the upper section (Km. at HLK - to Km 23.), the middle section (Km 23.1 to Km 4.), and the lower section (Km 4.1 to the Canada-U.S. border at Km 56.5). The Kootenay section was established as a separate sample section that extended from the Kootenay-Columbia rivers confluence upstream to Brilliant Dam (Km. to Km 2.8). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 3

19 Arrow Lakes Generating Station Hugh L. Keenleyside Dam Norns Creek Castlegar Brilliant Dam R i v e r K o o t e n a y N UPPER SECTION Blueberry Creek KOOTENAY SECTION MIDDLE SECTION China Creek C O L U M B I A Genelle R I V E R Champion Creek Birchbank Water Gauge Station Trail Rock Island reek Beaver C LOWER SECTION LEGEND Scale 1 : 2, 5 5Kilometres Fort Shepherd Eddy Waneta Dam Railway Island Bridge Swamp / Marsh Main Roads Sand or Gravel Bar River / Stream Lake REFERENCE Digital Data Sets 82F.2, 82F.3, 82F.12, 82F.13, 82F.22, 82F.23, 82F.32 and 82F.33 from BC Hydro. Datum: NAD 83 Projection: UTM Zone 11 PROJECT TITLE Seven Mile Dam P e n d Large River Fish Indexing Program Lower Columbia River Overview of Study Area Castlegar, British Columbia d ' O r e i l l e R i v e r BRITISH COLUMBIA, CANADA WASHINGTON, U.S.A. PROJECT No DESIGN EL 15 Jan. 28 GIS EL 15 Jan. 28 CHECK DF 15 Jan. 28 REVIEW LH 15 Jan. 28 SCALE AS SHOWN FIGURE 1.1 REV.

20 2. METHODS 2.1 Physical Parameters Discharge Discharge data for the mainstem Columbia River were obtained from BC Hydro (discharge through HLK and Arrow Lakes Generating Station) and from the Water Survey of Canada gauging station (No. 8NE49) at Birchbank (Figure 1.1). Discharge data for the Kootenay River were obtained from FortisBC (discharge through Brilliant Dam and Brilliant Expansion). Unless indicated otherwise, discharges throughout this report are daily averages presented as cubic metres per second (m 3 /s) Temperature Water temperatures for the Columbia River were obtained at hourly intervals using paired Vemco Minilog12 temperature data loggers (accuracy ±.1 C) installed in the tailrace of HLK, and paired Onset Tidbit temperature loggers (accuracy ±.2 C) at Birchbank. Vemco Minilog12 temperature data loggers were deployed in the tailrace of Brilliant Dam to monitor changes in Kootenay River water temperatures. When field crews retrieved the data loggers on 22 November 27 the station had been vandalized. Therefore, water temperature data for the Kootenay River between 1 August and 22 November 27 is not available for presentation in this report. All available temperature data were summarized to provide daily average temperatures. Spot measurements of water temperatures were obtained at all sample sites at the time of sampling using a hull-mounted Airmar digital thermometer (accuracy ±.2ºC) Habitat Conditions During the 27 study, several habitat variables were qualitatively assessed at all sample sites. Variables selected were limited to those for which information had been obtained in previous studies and were intended as a means to detect if gross changes in habitat availability or suitability had occurred in the sample sites between studies. The data collected were not intended to quantify habitat availability or imply habitat preferences. The type and amount of instream cover for fish were qualitatively estimated at all sites. Water velocities were visually estimated and categorized at each site as low (less than.5 m/s), medium (.5 to 1. m/s), or high (greater than 1. m/s). Water clarity was visually estimated and categorized at each site as low (less than 1. m depth), medium (1. to 3. m depth), or high (greater than 3. m depth). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 5

21 Each site was categorized into various habitat types using the Bank Habitat Types Classification System (Appendix A, Table A1), similar to the bank habitat types developed for the Peace and Lower Columbia River systems (R.L.&L. 1995b). Bank types for each site were established during the first sample session of the Phase 2 investigations. Bank type length within each site was calculated using ArcView GIS software (Appendix A, Table A2). Netters estimated the number of observed fish by species by bank habitat type. If the length of a bank habitat type was less than approximately 1 m, it was combined with adjacent bank habitat types to facilitate the netters ability to remember observed fish counts. 2.2 Study Period Seasonal differences in abundance, distribution, age and size-class composition, and habitat use occur for most of the major resident fish species in the study area (Hildebrand and English 1991, R.L. & L. 1993a, 1994, 1995b, 1996, 1998). These studies also indicated that the greatest number and diversity of species and species life stages are vulnerable to boat electroshocking at night during the summer and fall periods. In addition, these periods have been the most consistently sampled since 199 and, therefore, represent the periods for which substantial data sets are available for comparison to present study results. For these reasons, the late summer to early fall period was selected for the Phase 7 sample period. As stress on fish associated with capture and processing is greater at warmer water temperatures (Golder 22) sampling did not start until water temperatures were below 15 C. The dates of each session conducted in the present study are provided in Table 2.1. Table 2.1 Sampling periods within the Lower Columbia River, 27. Session Sample Date 1 27 September to 4 October 2 6 to 11 October 3 13 to 22 October 4 24 to 29 October 5 1 to 6 November Phase 7 sampling was initially scheduled to begin on 17 September 27; however, sampling was postponed until 27 September due to a delay in receiving a Species At Risk (SARA) permit for white sturgeon (Acipenser transmontanus). On 27 September, sampling was cancelled after the first site due to boat mechanical problems, and recommenced on 29 September. Sampling was cancelled after three sites on 3 October due to poor weather (rain). On 15 October field crews noted a large decrease in catch-rates while sampling sites ES7 and ES8, later associated with a cathode connection problem on the electroshocking boat. These two sites were re-sampled on 17 October. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 6

22 2.3 Fish Inventory Sample site locations and designations are provided in Appendix B, Figures B1 to B3. Site descriptions and UTM locations are listed in Appendix B, Table B1. Three index species (mountain whitefish, rainbow trout, and walleye) were selected for population abundance estimates and age and length-cohort analyses. Mountain whitefish and rainbow trout were selected because of their importance as sportfish and the large amount of historical data already available for these species from the study area. Walleye were selected because of their importance in the regional recreational fishery and their status as the dominant piscivore in the area. Overall, fish were captured and sampled using methods similar to previous phases of the project (Golder 22, 23, 24, 25, 26, 27). However, based on results of these phases, some alterations in methodologies have occurred between phases Boat Electroshocking Boat electroshocking was conducted in each of the four sections of the study area to capture fish within nearshore habitats along the channel margin. To provide sequential mark-recapture opportunities for the index species abundance study, five consecutive sampling sessions were conducted in each of the Kootenay (n = 2 sites), upper (n = 9 sites), middle (n = 5 sites), and lower (n = 7 sites) sections. The sites within each section were the same as those sampled during the 21 to 26 studies (Golder 22, 23, 24, 25, 26, and 27). Boat electroshocking employed a Smith-Root Inc. high-output electroshocker operated out of a jet-drive riverboat by a three-member crew. The electroshocking procedure consisted of manoeuvring the boat downstream along the shoreline of the sample sites. Two crew members positioned on a netting platform at the bow of the boat netted stunned fish, while a third individual operated the boat. The two netters attempted to capture all index species fish stunned by the electrical field. Captured fish were immediately placed into a 175 L onboard live-well. Any index species that avoided capture and all other species that were positively identified were enumerated and recorded as observed. The number of fish observed along each bank habitat type also was recorded. Time sampled (seconds of electroshocker operation) and length of shoreline sampled (kilometres) were recorded for each sample site. Electroshocking sites varied between 58 and 44 m in length. If, due to logistical reasons, the site was not fully completed, the difference in distance between what was sampled and the established site length was estimated and recorded on the site form, and then used as the sampled length in the subsequent analyses. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 7

23 To reduce fish mortalities and stress on the fish associated with capturing and handling, compressed oxygen was pumped into the livewell to maintain dissolved oxygen at levels similar to levels recorded in the river. In addition, a livewell additive (Rejuvenade TM HRT) was added to the livewell to reduce recovery time of the fish. Amperage output and frequency settings developed during the 24 study were proven to reduce electroshockinginduced injuries on large rainbow trout (Golder 25); as such, these settings were maintained during the current project. Although electrical output is variable (i.e., depending on water conductivity, water depth, and water temperature), field crews attempted to maintain electrical output at similar levels for all sites over all sessions Fish Processing A site form was completed at the end of each sampled site. Site habitat conditions and observed fish were recorded before the start of fish processing for life history data. Life history information [i.e., fork length (FL; recorded to the nearest 1 mm) and wet weight (recorded to the nearest 1 g)] was collected from all index species fish, and recorded directly into the Lower Columbia River Fish Indexing Database (Attachment A) using a laptop computer. All fish sampled were automatically assigned a unique identifying number by the database; this provided a method of cataloguing associated ageing structures. All index species fish 12 mm FL that were in good condition following processing were marked with a Plastic Infusion Process (PIP) Passive Integrated Transponder (PIT) tag (tag model ENSID Fusion 11 mm FDX-B). All tags and tag injectors were immersed in an antiseptic (Super Germiphene ) and rinsed with distilled water prior to insertion. Tags were inserted using a Simcro Tech Ltd. single shot or multi-shot applicator. All tags were inserted into the dorsal musculature on the left side below the dorsal fin and between the pterygiophores. Tags were then checked to ensure they were inserted securely and the tag number was recorded in the Lower Columbia River Fish Indexing Database. Information obtained from each marked fish included: species, fork length, weight, tag type or colour, tag number, date of marking, release location, and sex and maturity (determined, where possible, through external examination). For mountain whitefish and rainbow trout, ageing structures (scales) were collected from all individuals captured in accordance with the methods outlined in Mackay et al. (199). All scales were stored in appropriately labelled coin envelopes and air-dried before processing. Ageing structures were not collected from walleye because this species is a seasonal resident and uses the study area principally for feeding by adult and sub-adult cohorts. As a result, sensitive early life stages are unlikely to be affected by river regulation in the study area. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 8

24 2.3.3 Ageing Scales were processed in accordance to procedures described in Mackay et al. (199). Samples were temporarily mounted between two slides and examined using a microfiche reader. Where possible, several scales were examined and the highest quality scale was digitally scanned and saved as a JPEG-type picture file in the Lower Columbia River Fish Indexing Database. All mountain whitefish scale samples were examined independently by two experienced individuals and ages assigned. If assigned ages differed between the two examiners the sample was re-examined jointly by both examiners to establish a final age. Mountain whitefish that were captured during an earlier phase of the project (21 to 26) and recaptured during the current phase were aged using scale samples from both years. Rainbow trout that were captured during an earlier phase of the project (21 to 26) and recaptured during the current phase were aged using scale samples from both years in an effort to determine accuracy and consistency of ageing. Both scale samples were examined simultaneously by two experienced individuals and ages assigned. If assigned ages differed between the two examiners, or if a scale sample from 27 was not the appropriate number of years older than the corresponding 21 to 26 scale sample, both samples were re-examined jointly by both examiners to establish a final age. A copy of all digital scale images is provided in Attachment A. All scale samples collected from mountain whitefish and rainbow trout during Phase 7 are provided as Attachment B (archived at BC Hydro, Castlegar). 2.4 Data Analyses Data Compilation Raw data were entered directly into the Lower Columbia River Fish Indexing Database (Attachment A) using Microsoft Access 23 software. The database has several features to ensure that data are entered correctly and in a consistent manner. Validation rules, combo boxes, and input masks were used wherever possible to restrict data entry to a limited list of choices and to decrease data entry time. Several event procedures were used to ensure data were entered correctly, e.g., the database automatically calculated the condition factor of the fish based on its fork length and weight. If the condition factor was outside a previously determined range for that species, a message box appeared on the screen informing the crew of a possible data entry error before the fish was released, allowing crews to re-measure the fish. Basic sorts and summary statistics were performed prior to analysis to identify possible data entry errors and outliers. Additional queries and assessment of the accuracy of the database were performed as part of the mark-recapture and age assignment (Program MIX) analyses. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 9

25 2.4.2 Catch and Life History Data In order to provide data comparable with previous study results, catch-rates from the present study were calculated based on the number of fish captured plus observed fish (i.e., fish that were observed and positively identified to species but avoided the netter). Catch-rates for each site were expressed as the number of fish captured and observed per kilometre of shoreline sampled per hour of electroshocker operation (CPUE = no. fish/km/hr). The CPUE for each session was the sum of the number of fish captured and observed per kilometre of shoreline sampled per hour of electroshocker operation for all sites within a section. The average CPUE was estimated for each section by averaging the CPUE from all sites for all sessions within the section. Standard life history summaries were generated for each index species. These included length and age-frequency distributions and length and weight-at-age relationships. Length-at-age growth relationships were generated by averaging the length for each age-class and by using the von Bertalanffy growth model (Ricker 1975). The addition of.75 to the age category of mountain whitefish (i.e., age- fish are presented as age-.75) and.5 to the age category of rainbow trout allowed the length-at-age regression curve and the Y-intercept to more accurately reflect the actual age of the fish, given an assumed hatch date of 1 March for mountain whitefish and 1 June for rainbow trout. Growth curves for length-weight relationships were generated using a two parameter power equation in SigmaPlot. In some instances, changes in sampling methodologies (described in detail in Section 5.) between phases reduced the validity of direct comparisons of some metrics between phases Population Estimates Population abundance estimation techniques used during the current program assumed equal recapture probabilities over all size-classes; as such, homogeneity in recapture rates between size groups was tested by means of a chi-square goodness-of-fit test prior to generating estimates. This test was performed on mountain whitefish and rainbow trout mark-recapture data from 27 for individuals greater than or equal to 25 mm FL (hereafter referred to as large fish) and individuals less than 25 mm FL (hereafter referred to as small fish). Population estimates of index species during the present study were made using the modified Schnabel (Ricker 1975) and the sequential Bayes algorithm (Gazey and Staley 1986) estimation techniques, as well as a Pollock s Robust Design model (Pollock 1982) using Program MARK (White and Burnham 1999). These methods allowed comparisons between population estimates developed in previous studies (Golder 22, 23, 24, 25, 26, 27). For the modified Schnabel and sequential Bayes algorithm techniques, fish that were marked during an earlier phase of the program were not recorded as a recapture, unless the fish was captured two or more times during the current phase of Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 1

26 the program. The following assumptions were used for the modified Schnabel and the sequential Bayes algorithm population estimates: the population was closed (i.e., the population size remained the same over the period of the experiment); all individuals had the same capture probability, regardless of whether they were marked or unmarked; fish did not lose their marks during the study period; and, all marks were reported on recovery. The modified Schnabel population estimate (multiple census) involved fish that were marked and added to the population over time. During this same time, samples were taken and examined for recaptures. These mark-recapture data were then used to calculate population estimates (with 95% confidence intervals based on a Poisson distribution) for the following groupings: large mountain whitefish, small mountain whitefish, all mountain whitefish combined, large rainbow trout, small rainbow trout, all rainbow trout combined, and all walleye. The sequential Bayes algorithm is usually used when the number of animals marked and recaptured is low. During the present study, this population estimation method was used to provide a second estimate to compare to the modified Schnabel estimate. The advantages of using the Bayesian approach as listed by Gazey and Staley (1986) are: the probability of observing the data at all feasible population sizes is calculated exactly; it is applicable for all cases regardless of sample size or procedure; a plot of successive posterior distributions can be used as a visual diagnostic of conformity with basic assumptions; and, inferences can be made directly because the end product completely describes the uncertainty of the population size given the data. Population estimates using the sequential Bayes algorithm were dependent upon selection of the maximum prior values (Nk). The modified Schnabel method mean estimate was increased three-fold for initially setting the Nk parameter (as described in Gazey and Staley 1986) as the maximum probable population size in deriving the population parameters for each species and size-cohort. The minimum prior was set at the total number of fish marked during the 27 study. The index species populations also were estimated using Program MARK. At a very basic level, Program MARK uses information about the encounter history of a sample of fish from a population to estimate the most important demographic parameter of survival. In addition to survival, other parameters can be estimated depending upon the structure of the data and the questions of interest. Individual covariates and group covariates also can be incorporated into the model structure as well. Program MARK computes the parameter estimates for the model structure of choice using maximum likelihood techniques and various formulations of a particular model structure can be compared using information criterion. An example of model formulations that could be compared in the Program MARK model comparison framework would be a model that varies survival through time and with the covariate of fork length as compared to a model that does not vary survival at all. The main idea behind this approach is that a likely candidate set Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 11

27 of models for relative comparison is built based on a priori and the known biology of the population of interest, fitted to the data and then ranked using the Akaike Information Criterion (AIC) to determine which model is best suited to explain the data. Using data from 21 to 27, candidate model sets were constructed and assessed independently for the three index species. All size-classes were grouped together for mountain whitefish and walleye. Rainbow trout were assessed separately for large and small fish since the recapture rates for those two size-classes were shown to differ significantly with a chi-squared test. Pollock s Robust Design model was accessed through an R interface (R 2.6.1, 27) to Program MARK. The Program MARK model fitting provided estimates of survival, immigration and emigration, population abundance, and probabilities of capture and recapture through time. The RMark package (Version 1.7.3) provided the functions and interface (RMark, 28). The robust design model is a combination of the Cormack-Jolly- Seber (CJS; Cormack 1964, Jolly 1965, Seber 1965) live recapture model and the closed capture models. The model is described in detail by Kendall et al. (1997, 1995), and Kendall and Nichols (1995). The robust design model applies when multiple closely spaced capture occasions (i.e., sessions) are available between longer periods (i.e., phases). The assumption of a closed population is considered to be met (no mortality or migration) within each phase, allowing abundance to be estimated using a closed capture model. The longer intervals between phases allow an estimation of survival, temporary emigration from the study area, and immigration of marked fish back into the study area. A detailed description of Pollock s Robust Design model is provided at (White and Doherty 27) and a description of the alternative formulations of the closed capture models available within the robust model are available in the Program MARK help file. The standard robust design with closed captures model was used. The model uses all years of data from the beginning of the indexing program and with each year s additional data, the estimates of the demographic parameters generally improve as a result of increasing numbers of recaptured fish and particularly increased numbers of inter-year recaptured fish. These estimates have narrower confidence intervals and more precise values as the program proceeds. Consequently, earlier year s reports will have different estimates of the population abundance and confidence intervals than noted here. A candidate model set was constructed based on the current understanding of the biology of the three index species and the structure of the sampling program over the past seven years. The candidate model set for each index species consisted of many reasonable combinations for each parameter (e.g., constant, time varying, tag type varying, etc.), and also included the impact of covariates of moon phase, flow, and effort on the capture-recapture probabilities. The analysis fitted each of the candidate models then ranked them using the second-order Akaike Information Criteria (AIC c ) to determine the most likely model within each candidate set. Any models that were found to have either: 1) non-positive values for the beta variance parameters; or, 2) confidence intervals that were greater than.9, were dropped from the final model set prior to the calculation of the AIC c values or the parameter values. The AIC c value and the Akaike weight were calculated for each model in the final model set and the weights were used in model averaging Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 12

28 in order to obtain the parameter estimates for each index species. This information-theoretic approach to model selection and model averaging is detailed by Burnham and Anderson (22). In brief, model averaging allows each model in the set that has garnered any weight to be used when estimating parameter values. The Akaike weights determine how much influence each model s estimates have on the averaged parameters. The extreme example of this would be if the winning model had a weight of 1 and all other models had weights of. The model averaged parameters would be entirely from the top model. Tag loss rates can strongly affect survival estimates since an assumption of mark-recapture techniques is that once a tag is added to the population, it is available for recapture throughout the duration of the study. In 27, all captured fish were carefully examined for evidence of scars that may have resulted from a lost tag, either from the current study or from a previous year. This issue will be less of a concern in future phases of the program, as T-bar anchor tags have been replaced with PIT tags and the major issue with tag loss was with the T-Bar anchor tagged fish (which will gradually die and disappear from the population) Comparison of Estimated Populations Mark-recapture programs conducted during two or more sample periods (i.e., phases) can be used to test the hypothesis that there has been a change in the size of a population between periods. Gazey and Staley (1986) outlined how to assess population differences between samples, which involved finding the compound distribution of the difference in population size. The most probable difference is equal to the difference between the modes for each year, which could be found arithmetically from the corresponding estimates of the modes for each year. However, by estimating the compound distribution of the differences, the probability that there has been either an increase or decrease between years also can be estimated. Comparisons of estimated population size and cumulative probability distribution curves were created between the 26 and 27 studies for all index species and compared to cumulative probability distribution curves between the 21 and 22 studies, the 22 and 23 studies, the 23 and 24 studies, the 24 and 25 studies, and the 25 and 26 studies Age-Cohort Analysis Rainbow trout were aged using scale samples, but due to problems associated with consistent annuli production and due to the fact that only every third fish sampled was actually aged, a modelling approach was used to assign ages to the unaged fish. Length-frequency data was combined with correctly aged rainbow trout data to assign a conditional probability for each age-class to each fish. Those rainbow trout known to be correctly aged were fish that had been recaptured in multiple years and had their ages confirmed. If a fish was part of the aged sub-sample, the conditional probability was 1 (or 1% certainty) for the age-class in which it fell, and the age of that fish tracked forward each year if it was recaptured. If a fish wasn t aged, a conditional probability for each age-class was assigned using a mixed Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 13

29 distribution approach. This method was also used for mountain whitefish. The length-frequency distribution was combined with the age data to estimate the parameters for each age-class for mountain whitefish and rainbow trout. The fitting procedure used a mixed distribution approach which outputs the mean length-at-age for a particular age-class, the proportion of fish that fall within a given age-class, and the associated standard deviations. This method was developed by Macdonald and Pitcher (1979) and the fitting functions are available within Program MIX, which was run in R (Version 2.6.1; R 27). The R library was developed to give open access to Program MIX (MacDonald and Du 24). The assignment of conditional probabilities to each fish not only gave a more complete picture of the age structure and how it varied through the years, but also allowed the use of Age-Structured-Mark-Recapture (ASMR; see Section 2.4.7) modelling on the entire sampled population since ASMR requires each fish to be assigned to an age. Program MIX can fit component distributions to each age-class represented in the length-frequency histogram based only on length-frequency data. However, poor fits result when fish of the same fork length can have different ages or when sample sizes for a particular cohort are small. These situations occur most commonly after sexual maturity, when both growth rates and survival decrease, but also can occur prior to sexual maturity in a mixed stock fishery. For these reasons, initial parameters were estimated and the maximum length and age were incorporated into the model with assistance from the sub-sample of correctly aged fish for each species. This restriction of the maximum age and length allowed the fitting to more easily converge on distributions for older aged fish since it pooled bins with limited data. These initial estimates were used as a starting point to fit mixture models for each year of data with both aged and unaged fish combined. Three plausible distributions (normal, lognormal, and Poisson) were assessed for their fit to each year s data and various logical assumptions of constraints on the parameters were tested to see if they improved the fit significantly over an unconstrained model. The following biological constraints were used: 1) a constant coefficient of variation among the standard deviations for those models fitted with the normal and the lognormal distributions; and, 2) a coefficient of variation constrained to be constant by modifying the relationship between the means and standard deviations to meet the assumptions of the Poisson distribution. Ten iterations of each model were done to allow the starting parameters sufficient flexibility to fit each set of data. The model with the lowest chi-square statistic was selected as the best model and the output of this model allowed the calculation and assignment of conditional probabilities for each fish and each age-class PIT and T-Bar Anchor Tag Growth Comparisons In 25 and 26, index species fish were implanted with both PIT and T-Bar anchor tags. Based on previous markrecapture analyses, the a priori assumption was that fish tagged with PIT tags would have a higher survival than those tagged with T-bar anchor tags. To allow the combination of 25 and 26 data, homogeneity in inter-year recapture rates between the two tag types was tested using a one-tailed chi-square goodness-of-fit test for all index species. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 14

30 Annual growth rates for the two tag types were analyzed for each of the three index species using a von Bertalanffy growth model. The von Bertalanffy growth model was run for mountain whitefish and rainbow trout with three different groups; fish marked with 'T-bar' anchor tags, fish marked with PIT tags, and all marked fish (both tag types combined). The resulting three models were analyzed using a paired comparison maximum likelihood ratio test and a chi-square goodness-of-fit test to determine if a stronger model resulted from either separate or grouped tag types. Based on previous field observations, the a priori assumption was that fish tagged with PIT tags would grow faster than those tagged with T-bar anchor tags. Therefore a one-sided chi-squared test was used. For walleye, similar growth rates between the two tag types prevented model convergence. In addition, Program MARK was used to generate interannual survival of all index species by tag type Age Structured Mark-Recapture Analysis Background Estimates of recruit and adult abundance for both rainbow trout and mountain whitefish were produced using ASMR analysis (Coggins et al. 26a). The method is so named because it combines traditional mark-recapture methods with an age-structured virtual population analysis (VPA) framework. If its assumptions are met, ASMR can extract more information from sparse datasets than traditional mark-recapture methods since it uses information from both marked and unmarked fish. Furthermore, due to its structural assumptions, the method is able to estimate the abundance of unmarked fish in the first year of data collection (i.e., 21), as well as the number of recruits in prior years. Consequently, the ASMR method could potentially help fill in the missing years of sampling that occurred in the late 199s. Definition of ASMR Coggins et al. (26a) defined the ASMR method as follows. The expected numbers of unmarked ( U ˆ a, t ) and marked ( M ˆ a, t ) fish by age (a = 1,..., A) and year (t = 1,..., T ) at risk of capture and recapture are estimated by: 1) Uˆ ˆ ( ˆ a+ 1, t+ 1 = S a U a, t ma, t ) 2) M ˆ ˆ ( ˆ a+ 1, t+ 1 = S a M a, t + ma, t ) where Ŝ a is the annual survival of an age-a fish, m a, t is the number of age-a fish marked in year t, A is the oldest age-class and T the last year of data collection (i.e., 27). The equations treat multiple recaptures of the same individual within each year as a single recapture and assume that tagging has no effect on mortality or growth. The age-dependent annual survival ( Ŝ a ) is calculated from an independent estimate of the von Bertalanffy k growth parameter and the instantaneous annual mortality rate of fish of asymptotic length Mˆ, as: ˆ k ( a+ 1) M k e adult / 1 3) S = [ ka ] a e 1 ˆ adult Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 15

31 Note that the formula is corrected from a typographical error in Coggins et al. (26a, equation (7)) that omits the negation of Mˆ. adult In order to predict Uˆ and Mˆ the method uses a back-propagation approach in which the number of unmarked fish of the oldest age in each year ( U ˆ A, t ) is assumed to be, the number of marked fish in the initial year, M ˆ a, 1, prior to sampling is and all ages in the terminal year ( U ˆ a, T ) are treated as parameters to be estimated. The remaining U ˆ a, t are calculated with the following equation: U ˆ = Uˆ Sˆ + + m, 4) a, t ( a+ t, t 1 a ) a t Given the predictions of the numbers of unmarked and marked fish at risk of capture, the expected number of unmarked ˆ ) and marked ( r ˆ a, t ) fish captured by age and year are then estimated as follows: ( m a, t 5) m ˆ a, t = Uˆ a, t pˆ a, t 6) r ˆ a, t = Mˆ a, t pˆ a, t where ˆ is the estimated capture probability by age and year. p a, t The resultant log-likelihood function is: A T A T 7) log e L( m, r θ ) = [ mˆ a, t + ma, t log e ( mˆ a, t )] + [ rˆ a, t + ra, t log e ( rˆ a, t )] a= 1 t= 1 where θ is the parameter set to be estimated and m a, t and a= 1 t= 2 r a, t are independent samples from Poisson distributions. The ASMR method entails the determination of the number of unmarked fish in the terminal year ( U ˆ a, T ), the instantaneous annual mortality rate ( Mˆ adult ), and the capture probabilities ( p ˆ a, t ). Coggins et al. (26a) defined three alternative formulations of the method which make different assumptions. Each formulation is outlined below. ASMR 1 In this model, the capture probability is calculated using p ˆ a, t = pˆ tvˆ a, t, where v ˆ a, t is the age- and time-specific vulnerability to capture and pˆ is the conditional maximum likelihood estimate of the annual capture probability of a fish with a vulnerability of 1 which is calculated as: A 8) pˆ t = ( ma t + ra, t ) vˆ a= 1 t A, a, t ( Uˆ a t + Mˆ, a, t ) a= 1 The vulnerability to capture is a measure of the relative vulnerability. As such, the only constraints on its values are that they are greater than or equal to zero. The number of unmarked fish in the terminal year is then calculated as ˆ =. To simplify the equation, it is possible to assume that fish older than a specified age are equally U a T ma T pˆ,, a, T Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 16

32 vulnerable to capture and that the vulnerability to capture is constant across years. ASMR 1 maximizes equation 7 by varying θ = ( v, Mˆ, and pˆ T ). ˆ a, t adult ASMR 2 ASMR 2 differs from AMSR 1 in that it directly estimates the abundance of unmarked fish in the terminal year ( U ˆ a, T ). Consequently, ASMR 2 maximizes equation 6 by varying θ = ( v, ˆ a t M adult, and ˆ a T ). ˆ, U, ASMR 3 ASMR 3 differs from the first two formulations in that more flexibility is allowed in the estimation of age- and timespecific capture probabilities ( ˆ p a, t ). This flexibility is achieved at the expense of a much larger parameter set. The p ˆ a, t matrix is estimated using the conditional maximum likelihood estimator: ma, t + r a, t 9) pˆ a, t = Uˆ + Mˆ a, t a, t This removes the need to specify the vulnerability to capture ( v ˆ a, t ) and the annual capture probabilities ( pˆ t ). As in ASMR 2, ˆ is estimated directly. U a, T Extensions of ASMR Coggins and colleagues (Coggins et al. 26a; Coggins et al. 26b, Coggins 27) have generalized or extended the definition of ASMR above to include: 1) the stratification of recaptures by tag cohort; 2) inter-annual variation in the instantaneous annual mortality ( Mˆ adult ); 3) the estimation of the number of unmarked fish in the terminal year ( U ˆ a, T ) above a specified age by application of age-specific survivorship ( Ŝ a ); 4) the construction of credible intervals by sampling the posterior distribution using Markov Chain-Monte Carlo (MCMC) techniques (Gelman et al. 24); 5) the construction of confidence intervals representing the uncertainty in age assignment associated with not ageing all fish; 6) the use of AIC as an approximate model selection criterion; and, 7) model evaluation using Pearson residuals. All but the seventh extension were utilized in the ASMR analysis of the LCR rainbow trout and mountain whitefish data. The difference between frequentist or classical confidence intervals (which is what biologists typically use) and Bayesian credible intervals is based on philosophical differences in how the two approaches view the world (Gelman et al. 24). In short, a 95% confidence interval is expected to include the true value of the parameter being estimated 95% Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 17

33 of the time. In contrast there is a 95% probability that the value of the parameter falls within a 95% credible interval. For the purposes of interpreting the results of the ASMR analysis it is sufficient to treat the two types of interval as equivalent ASMR Analysis of rainbow trout and mountain whitefish The models were fitted using R (R Development Core Team, 27) and AD Model Builder (Otter Research Ltd. 2). The AD Model Builder code was based on code provided by Steve Martell and Lew Coggins. The number of captures ( m a, t ) and recaptures ( r a, t ) by age and year were produced by only considering the first encounter (capture or recapture) of an individual within each year. Fish that were released untagged were treated as if they had not been encountered. Fish that had not been aged were assigned an age by drawing at random from their length-dependent multinomial age probability distribution produced by the Program MIX analysis (see Section 2.4.5). In this way 5 datasets representing possible age assignments from the underlying conditional probability distributions were produced. Fifty datasets was considered sufficient to estimate the uncertainty due to not ageing all fish while still allowing the models to be fitted in a reasonable length of time. For each model a separate analysis was performed on each of the datasets and the results were averaged. Uncertainty due to ageing only a third of the sampled fish was then incorporated into the results by producing approximate point-wise 95% confidence intervals representing ± 2 standard deviations of the variation between the 5 analyses. If a particular analysis failed to converge or produced an output error, the analysis was not considered. If less than half of the analyses failed to fit then the model was considered to have failed to converge. The uncertainty inherent in the analyses (i.e., the uncertainty that cannot be resolved by ageing more fish) was also incorporated into the results by adding 95% credible intervals constructed from every 1 th MCMC trial in the second half of a chain of length 1. The 95% credible intervals for a model were calculated from the first of the 5 analyses to successfully fit the data. The ASMR method was further generalized to allow inter-annual migration into and out of the study area. The new formulation assumed that: 1) emigration and immigration were age- and time-constant; 2) emigration was permanent (i.e., once leaving the study area a fish could not return in a later year); and, 3) emigration and immigration were identical. Consequently, equations 1 and 2 were modified as follows: ˆ 1) ˆ ˆ ˆ ( γ. M a+ 1, t+ 1) U a+ 1, t+ 1 = S a ( U a, t ma, t ) + (1 γ ) 2) M ˆ = Sˆ ( Mˆ + m )(1 ) a+ 1, t+ 1 a a, t a, t γ where, γ is the probability of migrating (emigrating or immigrating). If γ = then there is no migration and the model is equivalent to the earlier formulation. Initially the method was programmed so that γ was estimated as one of the model parameters. However the models examined would not converge, perhaps because emigration and mortality were confounded. Similarly, when Coggins et al. (26b) attempted to incorporate movement between two areas into Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 18

34 the ASMR method they concluded that the ASMR model is over parameterised and unable to provide reliable estimates of movement rates. Consequently, the migration parameter was fixed at the mean estimate from the corresponding Program MARK analyses (.48 for rainbow trout and.4 for mountain whitefish). The average growth trajectory for each species was estimated by fitting a von Bertalanffy growth curve to the individual lengths for all years. The resultant von Bertalanffy k growth parameter was then used in the estimation of age-dependent annual survival ( Ŝ ). The k parameter was.8 for rainbow trout and.33 for mountain whitefish. a For the purposes of summarising adult abundance, rainbow trout were assumed to become adult at age-2 and mountain whitefish at age-3. To account for the fact that rainbow trout hatch around 1 June and mountain whitefish around 1 March but fish were not caught until mid October,.38 and.71 were added to the rainbow trout and mountain whitefish ages, respectively, when calculating the age-specific survival. Due to concerns over tagging effects, fish below a specified minimum length were not tagged. As ASMR assumes that all captured fish are tagged, fish below the length threshold were considered not to have been encountered and were excluded from the analyses. In effect, this assumption lowers the capture probabilities of the earliest age-classes. Due to changes in tagging experience and tag type, the minimum length dropped from 25 mm FL in to 16 mm FL in 24 to 12 mm FL in Consequently, the vulnerability to capture was allowed to differ between 21 to 23, 24, and 25 to 27. At the same time the vulnerability was allowed to vary with age until age-5 for rainbow trout and age-6 for mountain whitefish and the capture probability was allowed to vary by year. The number of unmarked fish in the terminal year ( ˆ ) was estimated for all but the last age which was calculated from the age-specific survivorship ( Ŝ ). a U a, T Preliminary analyses indicated that as expected and documented by Coggins (27), stratifying recaptures by tag cohort had little effect on the predicted abundances but slightly increased the precision. Therefore, only models in which recaptures were stratified by tag cohort were considered. Furthermore, preliminary analyses indicated that in the majority of cases, models with inter-annual variation in the instantaneous mortality rate ( Mˆ ) outperformed those with a fixed mortality rate. To limit the number of models, models with a fixed annual instantaneous mortality rate were not considered. adult For each species, six models were fitted, corresponding to the three formulations of ASMR (i.e., models 1 to 3) by two migration parameters ( and the fixed value from the Program MARK analyses, i.e.,.48 for rainbow trout and.4 for mountain whitefish). Each of the six models was fitted to 5 random age assignments from the multinomial probability distributions and an MCMC chain generated for the first successfully fitted age assignment. The six models were compared using the mean AIC (Burnham and Anderson 22). Although AIC is not strictly appropriate for selection of ASMR models, it is nevertheless useful for arbitration (Coggins 27). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 19

35 3. PHYSICAL CONDITIONS 3.1 Columbia River Discharge Peak daily average discharge recorded for the Lower Columbia River at Birchbank in 27 was 3851 m 3 /s on 6 June, and the lowest daily average discharge was 199 m 3 /s on 11 March. The bimodal pattern observed in 27 was typical of a normal (post-regulation) flow pattern; however, discharge from late March to mid-may was approximately 5 m 3 /s greater than the highest discharge recorded during the same time period from 21 to 26 (Figure 3.1). Daily discharges from 21 to 27 (each year individually) are presented in Appendix C, Figure C1. In 27, Columbia River discharge exhibited a bimodal pattern similar to the pattern recorded in 22, 23, 25, and 26, i.e., higher flows in early summer and late fall, and lower flows in early spring and early fall Discharge (m 3 /s) Sample Period Jan 1 Feb 1 Mar 1 Apr 1 May 1 Jun 1 Jul 1 Aug 1 Sep 1 Oct 1 Nov 1 Dec Date Figure 3.1 Mean daily discharge (m 3 /s) for the Columbia River at the Birchbank water gauging station, 27. The shaded area represents minimum and maximum mean daily discharge values recorded at Birchbank from 21 to 26. The white line represents average mean daily discharge values over the same time period. Overall, Columbia River discharge during the 27 surveys was lower than the average discharge recorded during the 21 to 26 surveys (Figure 3.1). During the 27 sample period, flow patterns throughout the mainstem Columbia River were unstable, but generally declined during session 1, remained relatively stable for sessions 2 through 4, and increased during session 5 (Figure 3.1; Appendix C, Figures C2 and C3). Discharges from HLK ranged from a high of Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 2

36 117 m 3 /s on 28 September to a low of 841 m 3 /s on 16 October (Appendix C, Figure C2). At Birchbank, discharges ranged from a high of 1646 m 3 /s on 28 September to a low of 128 m 3 /s on 18 October (Appendix C, Figure C3) Water Temperature Water temperature data from HLK tailrace (Figure 3.2) represents temperatures of water released from Arrow Lakes Reservoir, while the temperatures at the Birchbank and Cominco stations (Figure 3.3) reflect the mixing of Columbia and Kootenay river flows. The highest temperature recorded at HLK during the sample period was 14.3 C on 27 September while the highest temperature recorded at Birchbank was 14.5 C on 27 September. The lowest temperature recorded at HLK during the sample period was 7.5 C on 24 October while the lowest temperature recorded at Birchbank was 8.7 C on 25 October. Spot temperature readings for the Columbia River taken at the time of sampling ranged between 7.8 C and 14.1 C Water Temperature ( C) Sample Period Aug 8 Aug 15 Aug 22 Aug 29 Aug 5 Sep 12 Sep 19 Sep 26 Sep 3 Oct 1 Oct 17 Oct 24 Oct 31 Oct 7 Nov 14 Nov Date Figure 3.2 Mean daily water temperature ( C) for the Columbia River at Hugh L. Keenleyside Dam (HLK; black line), 1 August to 14 November 27. The shaded area represents minimum and maximum mean daily water temperature values recorded at HLK from 21 to 26. The white line represents average mean daily water temperature values over the same time period. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 21

37 2 18 Water Temperature ( C) Sample Period Aug 8 Aug 15 Aug 22 Aug 29 Aug 5 Sep 12 Sep 19 Sep 26 Sep 3 Oct 1 Oct 17 Oct 24 Oct 31 Oct 7 Nov 14 Nov Date Figure 3.3 Mean daily water temperature ( C) for the Columbia River at Birchbank water gauging station (black line), 1 August to 14 November 27. The shaded area represents minimum and maximum mean daily water temperature values recorded at Birchbank from 21 to 26. The white line represents average mean daily water temperature values over the same time period. Water temperature data from the late summer/early fall period of 27 exhibited a normal post-impoundment pattern (i.e., decreasing water temperatures from early August to early November). Overall, water temperatures for the Columbia River during the 27 survey were lower than the average water temperature recorded for the Columbia River during the 21 to 26 surveys (Figures 3.2 and 3.3). 3.2 Kootenay River Discharge During sampling, discharges of the Kootenay River were lower in 27 than during previous study years (Figure 3.4). In 27, Kootenay River discharge decreased in August and September but thereafter remained relatively stable in October. During the 27 study period, discharge ranged from a high of 472 m 3 /s on 31 October to a low of 333 m 3 /s on 3 November (Figure 3.4). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 22

38 16 14 Discharge (m 3 /s) Sample Period Aug 8 Aug 15 Aug 22 Aug 29 Aug 5 Sep 12 Sep 19 Sep 26 Sep 3 Oct 1 Oct 17 Oct 24 Oct 31 Oct 7 Nov 14 Nov Figure 3.4 Date Mean daily discharge (m 3 /s) for the Kootenay River at Brilliant Dam (black line), 1 August to 14 November 27. The shaded area represents minimum and maximum mean daily discharge values recorded at Brilliant Dam from 21 to 26. The white line represents average mean daily discharge values over the same time period Water Temperature Water temperature data for the Kootenay River during the 27 sample period is not available (see Section for a full description); however, based on spot temperature readings for the Kootenay River taken at the time of sampling, water temperatures decreased over the study period, ranging from a high of 12.5 C during session 1 to a low of 7.7 C during session 5 (Appendix C, Figure C4). 3.3 Habitat Conditions Reach habitat descriptions for the Lower Columbia River are provided in the 21 Phase 1 report (Golder 22). Habitat variables collected at each site over the course of the program have not indicated any gross changes in habitat conditions (Appendix A, Table A3, Attachment A). Fish counts by species for each bank habitat also have not suggested any gross changes in habitat preference (Appendix A, Table A4, Attachment A). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 23

39 4. FISH RESOURCES 4.1 Fish Species Composition and Distribution In the present study, fish were recorded from the four river sections within the Lower Columbia River. This value includes captured fish and fish observed and identified to species. Overall, more fish were recorded in 27 than in 21 (n = ), 22 (n = ), 23 (n = ), 24 (n = ), and 25 (n = 44 2), and less than in 26 (n = ). Of the 21 species recorded (including observed species combined into three non-sportfish genera), 13 were classed as sportfish, and five species and three genera [sculpins (Cottus spp.), suckers (Catostomus spp.), and dace (Rhinichthus spp.), were not identified to species] were classed as non-sportfish (Table 4.1). Table 4.1 Number of fish caught and observed by boat electroshocking and their frequency of occurrence in sampled sections of the Lower Columbia River, 27 September to 6 November 27. Sportfish Species Kootenay Section Upper Section Middle Section Lower Section All Sections n a % b n a % b n a % b n a % b n a % b Mountain whitefish ( 25 mm FL) Mountain whitefish (<25 mm FL) , Rainbow trout ( 25 mm FL) Rainbow trout (<25 mm FL) Walleye Brook trout 7 <1 8 <1 15 <1 Brown trout 1 <1 6 <1 7 <1 Bull trout 25 <1 2 <1 3 <1 3 <1 Burbot 5 <1 18 < Cutthroat trout 1 <1 1 <1 6 <1 8 <1 Kokanee 1 < <1 12 < Lake whitefish 7 <1 1 <1 25 < Smallmouth bass 16 <1 16 <1 White sturgeon 3 <1 4 <1 1 <1 3 <1 11 <1 Yellow perch 1 <1 1 <1 Sportfish Subtotal Non-sportfish Common carp 1 <1 1 <1 Dace spp. c 1 <1 1 <1 Northern pikeminnow 53 < <1 11 < Peamouth 1 <1 88 <1 2 <1 2 <1 93 <1 Redside shiner < Sculpin spp. c Sucker spp. c Tench 4 <1 1 <1 5 <1 Non-Sportfish Subtotal All Species a Includes fish observed and identified to species; does not include inter-year recaptured fish. b Percent composition of sportfish or non-sportfish catch. c Not identified to species. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 24

40 Mountain whitefish (5% increase) and kokanee (Oncorhynchus nerka; 3% decrease) showed the greatest change in percent composition between 26 and 27 among sportfish species, (Table 4.2). Walleye have decreased in absolute abundance each year since 23. Non-sportfish species showing the greatest changes in percent composition between 26 and 27 were redside shiner (Richardsonius balteatus; 12% decrease), sculpins (8% increase), and sucker spp. (5% increase). The 27 study was the first study in which sculpin spp. did not increase in absolute abundance, decreasing from fish in 26 to fish in 27 (12% decrease). Tench (Tinca tinca), not recorded from 21 to 25, were recorded in both 26 and 27 (Table 4.2). Table 4.2 Number of fish caught and observed by boat electroshocking and their frequency of occurrence in sampled sections of the Lower Columbia River, 21 to 27. a b c Species n a % b n a % b n a % b n a % b n a % b n a % b n a % b Sportfish Mountain whitefish Rainbow trout Walleye Brook trout 5 <1 8 <1 7 <1 3 <1 3 <1 4 <1 15 <1 Brown trout 1 <1 2 <1 1 <1 1 <1 2 <1 7 <1 Bull trout 16 <1 3 <1 18 <1 8 <1 8 <1 11 <1 3 <1 Burbot 3 <1 1 <1 59 < Cutthroat trout 1 <1 4 <1 2 <1 1 <1 5 <1 8 <1 Kokanee < <1 32 < Lake trout 1 <1 Lake whitefish 61 <1 14 <1 23 < Smallmouth bass 4 <1 3 <1 4 <1 53 <1 16 <1 White sturgeon 14 <1 6 <1 18 <1 5 <1 11 <1 14 <1 11 <1 Yellow perch 1 <1 4 <1 1 <1 24 <1 1 <1 Sportfish Subtotal Non-sportfish Common carp 2 <1 1 <1 1 <1 3 <1 1 <1 Dace spp. c 2 <1 3 <1 15 <1 17 <1 1 <1 Northern pikeminnow Peamouth 8 <1 25 <1 45 <1 51 <1 33 <1 52 <1 93 <1 Redside shiner Sculpin spp. c Sucker spp. c Tench 1 <1 5 <1 Non-sportfish Subtotal All species Includes fish observed and identified to species; does not include inter-year recaptured fish. Percent composition of sportfish or non-sportfish catch. Species combined for table or not identified to species. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 25

41 Mountain whitefish, the most abundant sportfish recorded during the present study, accounted for 7.8% of the total catch (sportfish and non-sportfish combined) and 45% of the sportfish catch (Table 4.2). The percent composition of this species in the sportfish catch was the highest in the Kootenay section (68%) and lowest in the middle section (26%; Table 4.1). Rainbow trout, the second most abundant sportfish recorded during the present study, accounted for 7% of the total catch and 39% of the sportfish catch (Table 4.2). The percent composition of this species in the sportfish catch was highest in the middle section (66%) and lowest in the Kootenay section (13%; Table 4.1). Walleye, the third most abundant sportfish recorded in the present study, accounted for 1.5% of the total catch and 9% of the sportfish catch (Table 4.2). The percent composition of this species in the sportfish catch was highest in the Kootenay section (18%) and lowest in the middle section (6%; Table 4.1). Catch-rates for sportfish increased over the sample period in the Kootenay section (Appendix D, Table D1). Catch-rates for sportfish in the upper and middle section decreased from session 1 to session 3 and remained constant from session 3 to session 5. Catch-rates for sportfish in the lower section were relatively constant for all sessions. Catch-rates for non-sportfish varied in all sections for all sessions (Appendix D, Table D2). They were highest for sportfish and non-sportfish in the upper section and lowest for sportfish and non-sportfish in the lower and middle sections, respectively. In some instances, localized shifts in habitat suitability/availability (i.e., caused by differences in discharge levels) reduced the validity of direct comparisons of species composition or abundance between sample sections or sessions. 4.2 Life History Characteristics Mountain Whitefish Mountain whitefish have consistently been most abundant in the upper section each year from 21 to 27 (Figure 4.1). In 27, the percent composition of large and small mountain whitefish within the total mountain whitefish catch (includes fish captured and observed) varied between sections. Similar to 21, 22, 23, and 26, small mountain whitefish were more abundant than large mountain whitefish in most sections. During the 24 and 25 studies, large mountain whitefish were more abundant than small mountain whitefish in all sections (Figure 4.2). In 27, the highest catch-rate for mountain whitefish (all size-cohorts combined) occurred in the upper section (16 fish/km/hr) and the lowest catch-rate occurred in the lower section (24 fish/km/hr; Figure 4.3). The same pattern was recorded in all other study years (i.e., 21 to 26). For all sections combined, the catch-rate of this species were highest in 21 to 23 and remained relatively low for 24 to 27 (Figure 4.3). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 26

42 5 4 small (<25 mm FL) large ( 25 mm FL) Percent Frequency Percent Frequency Kootenay Upper Middle Lower Section Kootenay Upper Middle Lower Section Figure 4.1 Percent composition (by year and sample section) of small (<25 mm FL; black bars) and large ( 25 mm FL; grey bars) mountain whitefish captured by boat electroshocking in the Lower Columbia River, 21 to 27. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 27

43 Percent Composition mm FL < 25 mm FL Kootenay Section Percent Composition Upper Section Year Year Percent Composition Middle Section Percent Composition Lower Section Year Year Figure 4.2 Percent composition (by sample section and year) of large ( 25 mm FL) and small (<25 mm FL) mountain whitefish captured by boat electroshocking in the Lower Columbia River, 21 to 27. In the present study, 1712 mountain whitefish (from all sections and sessions combined) were measured for fork length. Mountain whitefish lengths ranged from 99 to 514 mm FL with a modal length in the 211 to 22 mm FL size-range (Figure 4.4). Mountain whitefish were evenly distributed between the two size ranges (51% <25 mm FL, 49% 25 mm FL). Overall (all sections combined), the length-frequency histogram indicated two modes: 91 to 14 mm FL and 141 to 25 mm FL. These two modes were most evident in the lower, middle, and upper sections and represent age- and age-1 cohorts. In the Kootenay section, the catch was dominated by fish larger than 3 mm FL (i.e., age-2 or older) and the age- and age-1 cohorts were virtually absent (Figure 4.4). Length data from the upper, lower, and middle sections were pooled and this summary distribution was compared to the length distribution from the Kootenay section. The null hypothesis that the two length distributions were drawn from the same population was tested using the two-sample Kolmogorov-Smirnov test and found to be false (D=.5971; p < 2.2e-16). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 28

44 1 Kootenay Section 1 Upper Section CPUE (fish/km/hr) CPUE (fish/km/hr) Year Year 1 Middle Section 1 Lower Section CPUE (fish/km/hr) CPUE (fish/km/hr) Year Year 1 All Sections CPUE (fish/km/hr) Figure Year Mean catch-rate (by sample section) for mountain whitefish captured by boat electroshocking in the Lower Columbia River, 21 to 27. The dotted lines denote 95% confidence intervals. CPUE data includes captured and observed fish identified to species; all size-cohorts combined. From 21 to 25, length-frequency data for mountain whitefish suggested that the population structure was gradually shifting from a population dominated by smaller individuals to a population dominated by larger individuals (similar to the structure observed in the 199s; see R.L.&L and Golder 27; Figure 4.5). Data from the 26 and 27 studies suggest that the population structure may be reverting back to a structure similar to that observed during the 21 to 23 studies (a population dominated by smaller individuals; Figure 4.5). A higher proportion of age- and age-1 mountain whitefish were observed in 26 and in 27 than in 24 and 25 (Figure 4.6). Overall, the age-frequency histogram suggests decreased age- recruitment in 23, 24, and 25, relative to all other sample years (Figure 4.6). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 29

45 Percent Frequency Kootenay Section n = 297 Upper Section n = 798 Middle Section n = 259 Lower Section n = 358 All Sections n = Fork Length (mm) Figure 4.4 Length-frequency distributions for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November 27. Mountain whitefish in the Lower Columbia River exhibit rapid growth until age-4; thereafter growth slows considerably (Figure 4.7; Golder 22, 23, 24, 25, 26, 27). Mountain whitefish of larger sizes appeared to grow slower in 26 compared to 21 to 25 and 27 (Figure 4.7; Table 4.3). Annual growth rates for mountain whitefish captured in 21 to 26 and recaptured in 27 were not compared with historical data because of low numbers of inter-year recaptures. Although not statistically tested, length-weight regression analyses of mountain whitefish from 27 show a similar relationship to the 21 through 26 study results for all sections (Appendix E, Figure E1). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 3

46 Percent Frequency n = n = n = n = n = n = n = Figure 4.5 Fork Length (mm) Length-frequency distributions for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 21 to 27. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 31

47 (n = 38) (n = 77) (n = 699) 4 2 Percent Frequency (n = 74) 25 (n = 87) (n = 625) (n = 695) 4 2 Figure Age (years) Age-frequency distributions for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 21 to 27. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 32

48 Fork Length (mm) Lt = 47(1-e -.29(t+.28) ) 23 Lt = 487(1-e -.29(t+.25) ) 24 Lt = 467(1-e -.3(t+.3) ) 25 Lt = 438(1-e -.35(t+.31) ) 26 Lt = 421(1-e -.32(t+.38) ) 27 Lt = 445(1-e -.32(t+.43) ) Length-at-age 27 (n = 73) 22 predicted length-at-age 23 predicted length-at-age 24 predicted length-at-age 25 predicted length-at-age 26 predicted length-at-age 27 predicted length-at-age Age (years) Figure 4.7 von Bertalanffy growth curve for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 22 to 27. Table 4.3 Average fork length and weight-at-age for mountain whitefish captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November 27. Age-Class Average (mm) Fork Length Standard Deviation Range Average (g) Weight Standard Deviation Range a n = number of individuals sampled. n a Rainbow Trout The distribution of rainbow trout within sampled sections was similar in all study years, with greater numbers captured in the middle and lower sections and lower numbers captured in the upper and Kootenay sections (Figure 4.8). In 27, large rainbow trout were more abundant than small rainbow trout in catches from all sections. The highest percentage (47%) of the total rainbow trout catch was recorded in the middle section and the lowest percentage (4%) was recorded in the Kootenay section (Figure 4.9). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 33

49 4 small (<25 mm FL) 21 4 large ( 25 mm FL) Percent Frequency Percent Frequency Kootenay Upper Middle Lower Section Kootenay Upper Middle Lower Section Figure 4.8 Percent composition (by year and sample section) of small (<25 mm FL; black bars) and large ( 25 mm FL; grey bars) rainbow trout captured by boat electroshocking in the Lower Columbia River, 21 to 27. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 34

50 Percent Composition Kootenay Section 25 mm FL < 25 mm FL Percent Composition Upper Section Year Year Percent Composition Middle Section Percent Composition Lower Section Year Year Figure 4.9 Percent composition (by sample section and year) of large ( 25 mm FL) and small (<25 mm FL) rainbow trout captured by boat electroshocking in the Lower Columbia River, 21 to 27. In 27, the highest catch-rate for rainbow trout occurred in the middle section (116 fish/km/hr) and the lowest catchrate occurred in the Kootenay section (21 fish/km/hr; Figure 4.1). For each year from 21 to 26, the highest catchrates for rainbow trout also were recorded in the middle section. In this same time period, the lowest catch-rates were recorded in the Kootenay, upper, Kootenay, upper, upper, and Kootenay sections, respectively. For all sections combined, rainbow trout catch-rates decreased each year from 21 to 25, and increased from 25 to 27 (Figure 4.1). In the present study, 1569 rainbow trout (from all sections and sessions combined) were measured for fork length. Rainbow trout ranged from 88 to 69 mm FL with a modal length in the 291 to 3 mm FL size-range (Figure 4.11). Length-frequency histograms for rainbow trout indicated two distinct modes; 81 to 16 mm FL (age- cohort) and 171 to 37 mm FL (age-1 cohort). These two modes were most evident in catches from the lower and middle sections and less evident in catches from the upper and Kootenay sections (Figure 4.11). Length data from the lower and middle sections were pooled and this summary distribution was compared to the summary distribution of length data from the upper and Kootenay sections. The null hypothesis that the two length distributions were drawn from the same population was tested using a two-sample Kolmogorov-Smirnov test and found to be false (D=.1719; p < 8.2e-8). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 35

51 CPUE (fish/km/hr) Kootenay Section CPUE (fish/km/hr) Upper Section CPUE (fish/km/hr) Year Middle Section CPUE (fish/km/hr) Year Lower Section Year Year CPUE (fish/km/hr) All Sections Year Figure 4.1 Mean catch-rate (by sample section) for rainbow trout captured by boat electroshocking in the Lower Columbia River, 21 to 27. The dotted lines denote 95% confidence intervals. CPUE data includes captured and observed fish identified to species; all size-cohorts combined. Length-frequency data from 27 suggests a lower abundance of the age- cohort when compared to 21, 22, 23, and 26 (Figure 4.12). With the exception of 26, the age- cohort mode has generally decreased in abundance each year from 21 to 27. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 36

52 Percent Frequency Kootenay Section n = 83 Upper Section n = 297 Middle Section n = 661 Lower Section n = 528 All Sections n = Fork Length (mm) Figure 4.11 Length-frequency distributions for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November 27. Due to inconsistencies in annuli production on rainbow trout scales, first documented during the 23 study (see Golder 24 for a complete summary), rainbow trout age-cohort analysis for the current study was restricted to fish that were initially captured during the 22 to 26 studies and recaptured during the current study. This provided scale samples for the same fish over multiple years. Age-frequency and length-at-age data were compared only to the 22 to 26 data due to questionable age-class assignments in earlier studies. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 37

53 Percent Frequency n = n = n = n = n = n = n = Fork Length (mm) Figure 4.12 Length-frequency distributions for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 21 to 27. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 38

54 In 27, rainbow trout under age-2 were not available for analysis due to a lack of multiple scale samples from those age-classes. During all study years, most (i.e., >9% of all years) of the aged rainbow trout were captured in the middle and lower sections, due mainly to higher rainbow trout recapture rates in these sections. The 21 brood year appears over represented (Figure 4.13). Reasons for the increased recruitment in 21 are unknown but could be related to discharge, as flows were more stable in 21 than in 22 to 27 (Appendix C, Figure C1). Rainbow trout in the Lower Columbia River exhibit rapid growth until age-3; thereafter growth slows considerably (Figure 4.14). Similar to 26 (Golder 27), in 27 rainbow trout decreased in length and weight after age-3 (Table 4.4). This result is likely an artefact of the small sample size (n = 28) of fish older than age-3 and is not considered an accurate representation of the overall rainbow trout population. Table 4.4 Average fork length and average weight-at-age for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November 27. Only rainbow trout initially captured in 22 to 26 and recaptured in 27 were aged during the 27 study. Age-Class Average (mm) Fork Length Standard Deviation Range Average (g) Weight Standard Deviation Range a n = number of individuals sampled. n a Although not statistically tested, length-weight regression analyses of rainbow trout from 27 show a similar relationship to the 21 through 26 study results for all sections (Appendix E, Figure E2). A comparison of annual growth rates for rainbow trout initially captured in the late summer to early fall period and recaptured one year after tagging were compared using an alternate method of estimating the von Bertalanffy growth equation parameters (Fabens 1965 cited in Hilborn and Walters 1992) between the following groupings: 199 to 1994, 22 to 25, and 25 to 27. Annual growth rates for rainbow trout from 22 to 25 appear to be higher than during the 199 to 1994 studies, while annual growth rates from 25 to 27 appear to be higher than during the 22 to 25 studies (Figure 4.15). Reasons for the increase in growth rates are not known. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 39

55 (n = 83) (n = 16) (n = 112) Percent Frequency (n = 138) 26 (n = 18) 27 (n = 77) Age (years) Figure 4.13 Age-frequency distributions for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 22 to 27. Only rainbow trout initially captured in 22 to 26 and recaptured in 27 were aged during the 27 study. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 4

56 7 Fork Length (mm) L t = 478(1-e -.8(t-.436) ) Length-at-age (n = 624) Predicted length-at-age Age (years) Figure 4.14 von Bertalanffy growth curve for rainbow trout captured by boat electroshocking in sampled sections of the Lower Columbia River, 22 to 27 combined. Only rainbow trout initially captured in 22 to 26 and recaptured one year later were aged Growth rate at length (n = 85) Growth rate at length (n = 149) Growth rate at length (n = 11) Annual Growth (mm/year) Ci=(513-Li)(1-e -1.3 ) Ci=(478-Li)(1-e -.78 ) Ci=(559-Li)(1-e -.26 ) Initial Fork Length (mm) Figure 4.15 Annual growth rates for rainbow trout caught and then recaptured after one year in the Lower Columbia River, (R.L.&L. 1995b.), 22-25, and Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 41

57 4.2.3 Walleye During the present study, the greatest percentage (3%) of the total walleye catch was recorded in the lower section and the lowest percentage (21%) was recorded in the middle section (Figure 4.16). During the 21 study, the greatest percentage of walleye catch was recorded in the middle section, while in 22, 23, 24, 25, and 26, the greatest percentage of walleye catch was recorded in the lower section. During the 21 to 25 studies, the lowest percentage was recorded in the Kootenay section, while in 26, the lowest percentage was recorded in the upper section. Differences in composition of maturity classes among sections were not analyzed for walleye for the reasons provided in Section The percent composition of walleye in the Kootenay section has increased each year since 21 (Figure 4.17). In 27, the highest catch-rate for walleye occurred in the Kootenay section (27 fish/km/hr) and the lowest catch-rate occurred in the lower section (1 fish/km/hr; Figure 4.18). In 21, 22, 23, 24, 25, and 26, the highest catch-rates for walleye were recorded in the middle, Kootenay, upper, middle, upper, and Kootenay sections, respectively, and the lowest catch-rates were recorded in the lower section in all years except 22 (middle section). Between 24 and 27, catch-rates for walleye decreased in all sections. Overall (all sections combined), catch-rates for walleye have decreased each year since 23 (Figure 4.18). In the present study, 53 walleye (from all sections and all sessions combined) were measured for fork length. Walleye lengths ranged from 315 to 81 mm FL with a modal length in the 381 to 39 mm FL size-range (Figure 4.19). Most (73%) of the measured catch was within the 351 to 45 mm FL size interval. Overall (all sections combined), the length-frequency histogram indicates one strong mode between the 311 and 45 mm FL size-range. Length-frequency and growth rate data collected from 23 to 27 suggests that this mode corresponds to five age-cohorts. Based on Lake Roosevelt walleye length-at-age data collected between 1997 and 1999 by McLellan et. al. (1999) this mode likely corresponds to the 21 to 25 brood years. Out of these 5 brood years, 21 was the largest (first detected in the LCR in 23; Figure 4.2). Although not statistically tested, length-weight regression analyses of walleye from 27 show a similar relationship to the 21 through 26 study results for all sections (Appendix E, Figure E3). Annual growth rates for walleye captured in 21 to 26 and recaptured in 27 were not compared with historical data because of low numbers of inter-year recaptures. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 42

58 Percent Frequency Percent Frequency Kootenay Upper Middle Lower Section Kootenay Upper Middle Lower Section Percent Frequency Percent Frequency Kootenay Upper Middle Lower Section Kootenay Upper Middle Lower Section Percent Frequency Percent Frequency Kootenay Upper Middle Lower Section Kootenay Upper Middle Lower Section 5 27 Percent Frequency Figure 4.16 Kootenay Upper Middle Lower Section Percent composition (by year and sample section) of walleye captured by boat electroshocking in the Lower Columbia River, 21 to 27. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 43

59 Percent Composition Percent Composition Figure 4.17 CPUE (fish/km/hr) CPUE (fish/km/hr) Figure 4.18 Kootenay Section Middle Section Year Percent Composition Percent Composition Upper Section GOLDER ASSOCIATES LTD Lower Section Year Percent composition (by sample section and year) of walleye captured by boat electroshocking in the Lower Columbia River, 21 to 27. Kootenay Section Middle Section CPUE (fish/km/hr) CPUE (fish/km/hr) CPUE (fish/km/hr) Upper Section Lower Section All Sections Year Mean catch-rate (by sample section) for walleye captured by boat electroshocking in the Lower Columbia River, 21 to 27. The dotted lines denote 95% confidence intervals. CPUE data includes captured and observed fish identified to species. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 44

60 Percent Frequency Kootenay Section n = 115 Upper Section n = 158 Middle Section n = 74 Lower Section n = 156 All Sections n = Figure 4.19 Fork Length (mm) Length-frequency distributions for walleye captured by boat electroshocking in sampled sections of the Lower Columbia River, 27 September to 6 November 27. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 45

61 Percent Frequency n = n = n = n = n = n = n = Fork Length (mm) Figure 4.2 Length-frequency distributions for walleye captured by boat electroshocking in sampled sections of the Lower Columbia River, 21 to 27. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 46

62 4.2.4 PIT and T-bar Anchor Tag Growth Comparison Mountain Whitefish During the 25 and 26 studies, 1126 mountain whitefish were marked with T-bar anchor tags and 1398 fish were marked with PIT tags (Table 4.5). Of those 2524 fish, 5 T-bar anchor tags and 23 PIT tags were recaptured one year later. Recapture rates for mountain whitefish marked with T-bar anchor tags and mountain whitefish marked with PIT tags were different for the two tag types (t =8.894, p =.3; Appendix E, Table E1). Table 4.5 Summary of the total number of each index species marked and recaptured by year and tag type in sampled sections of the Lower Columbia River, 25 to 26, 26 to 27. Parameter Tag Type Year Total number marked Total number recaptured 'T-bar' anchor tags PIT tags 'T-bar' anchor tags PIT tags Mountain whitefish Rainbow trout Walleye 25 to to Total to to Total to to Total to to Total Annual growth (FL) of mountain whitefish marked with T-bar anchor tags from 25 to 27 ranged from 1 to 65 mm with an average annual growth rate of 24 mm (Table 4.6). Annual growth of mountain whitefish marked with PIT tags over the same time period ranged from 2 to 11 mm with an average growth rate of 29 mm. Differences in growth rates between the two tag types were significant using a maximum likelihood ratio test with a chi-squared approximation (p = <.5; Appendix E, Table E2). Survival of mountain whitefish marked with T-bar anchor tags was lower than for fish marked with PIT tags (survival =.375 and.5541, respectively), though confidence intervals for the two estimates overlapped (Figure 4.21). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 47

63 Table 4.6 Average annual growth rate (mm/year) of fish captured by boat electroshocking in sampled sections of the Lower Columbia River. Data from fish captured in 25 or 26 and recaptured one year later. Parameter Year Mountain whitefish Rainbow trout Walleye T-Bar PIT a T-Bar PIT T-Bar PIT Average Growth (mm) 25 to to to Standard Deviation 25 to to to Range 25 to to to n b 25 to to to a b Two PIT tagged mountain whitefish were removed from the analysis because of measurement error (i.e., the fork length recorded at recapture was less than the fork length recorded at initial capture.). n = number of individuals sampled. 1. T-bar anchor tags PIT Tags.8.6 Survival.4.2. Mountain whitefish Rainbow Trout <25 mm FL Rainbow Trout 25 mm FL Walleye Figure 4.21 Summary of survival estimates with 95% confidence intervals based on Program MARK robust model estimates for fish marked with T bar anchor tags (21 to 26 combined) and fish marked with PIT tags (25 to 27 combined). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 48

64 Rainbow Trout During the 25 and 26 studies, 1139 rainbow trout were marked with T-bar anchor tags and 12 fish were marked with PIT tags (Table 4.5). Of those 2339 fish, 55 T-bar anchor tags and 56 PIT tags were recaptured one year later. Recapture rates for rainbow trout marked with T-bar anchor tags and rainbow trout marked with PIT tags were not different for the two tag types (t =.9, p =.926; Appendix E, Tables E3). Annual growth (FL) of rainbow trout marked with T-bar anchor tags from 25 to 27 ranged from to 22 mm with an average annual growth rate of 125 mm (Table 4.6). Annual growth of rainbow trout marked with PIT tags over the same time period ranged from 4 to 218 mm with an average growth rate of 129 mm. Differences in growth rates between the two tag types were not significant using a maximum likelihood ratio test with a chi-squared approximation (p = >.99; Appendix E, Table E4). Due to the high growth rate and high variability in growth rates for rainbow trout in the study area, it is difficult to discern whether differences in growth are attributable to tag type of some other unknown environmental factor. Survival for both large and small rainbow trout marked with T-bar anchor tags (survival =.35 and.185, respectively) were similar to survival for large and small rainbow trout marked with PIT tags (survival =.429 and.35, respectively; Figure 4.21) Walleye During the 25 and 26 studies, 791 walleye were marked with T-bar anchor tags and 813 fish were marked with PIT tags (Table 4.5). Of those 164 fish, 18 T-bar anchor tags and 16 PIT tags were recaptured one year later. Recapture rates for walleye marked with T-bar anchor tags and walleye marked with PIT tags were not different for the two tag types (t =.64, p =.8; Appendix E, Table E5). Annual growth (FL) of walleye marked with T-bar anchor tags from 25 to 27 ranged from 9 to 66 mm with an average annual growth rate of 36 mm (Table 4.6). Annual growth of walleye marked with PIT tags over the same time period ranged from 16 to 65 mm with an average growth rate of 38 mm. Survival of walleye was similar for fish marked with T-bar anchor tags and fish marked with PIT tags (survival =.479 and.495, respectively; Figure 4.21) Other Species Information on all other fish species recorded during the present program has been included in the Lower Columbia River Fish Indexing Database (Attachment A). As index species were selectively captured, the information on other species is mainly anecdotal. Raw data are provided in Attachment A. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 49

65 4.3 Population Estimates The percent of mountain whitefish marked was higher in 27 (84.6%) than in all other years (range from 68% in 23 to 84% in 25; Table 4.7). This increase is likely due to the delayed sampling schedule in 27, as in earlier phases of the program (21 to 26) most of the mountain whitefish that were not taggable (due to warmer water temperatures) were captured during early sessions. In 27, 9% of captured rainbow trout were tagged, second only to 25 (92%). The percent of walleye marked in 24, 25, 26, and 27, was higher than in 22 and 23. The percent of mountain whitefish mortalities in 24, 25, 26, and 27 was lower than in 22 and 23. The percent of mortalities within the total catch for rainbow trout and walleye has been less than 1% each year since 22 (Table 4.7). Table 4.7 Total number of each index species caught, marked, and recaptured, and the number of mortalities in sampled sections of the Lower Columbia River, 22 to 27. Parameter Year a Mountain whitefish Total number of fish captured Total number (and percent) marked Total number (and percent) of mortalities Total number (and percent) of recaptures Rainbow trout Walleye b (7.6) 2219 (81.6) 497 (88.9) (68) 1546 (8.6) 1137 (89.7) 24 b 1282 (8.2) 1346 (85.5) 1576 (93.9) (83.8) 121 (91.5) 189 (93.6) (79.1) 128 (86) 63 (94.1) (84.6) 1512 (9.3) 476 (93.2) (9.3) 27 (1.) (.) (7.5) 9 (.5) 1 (.1) 24 b 37 (2.3) 1 (.1) () (1.4) 1 (.1) () (1.4) 4 (.3) () (2.) 1 (.1) () 22 6 (2.8) 2 (9) 11 (2.2) (2.3) 153 (11.1) 33 (3.2) 24 b 26 (2.3) 133 (1.5) 76 (5.3) (2.3) 12 (1.4) 5 (5.2) (2.6) 19 (1.1) 21 (3.7) (1.9) 16 (9.) 8 (1.9) a The 21 study was not included because sampling effort and marking effort were not comparable to the other studies. b The 24 study consisted of 7 sessions, while the 22, 23, 25, and 26 studies consisted of six sessions. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 5

66 During the first four sessions of the present study (fish marked during the final session are not used to generate population abundance estimates), 2734 fish of the three index species combined were marked (Table 4.8; Appendix D, Table D3). From 26 to 27, the number of marks released increased 17% for mountain whitefish and 1% for rainbow trout, but decreased 26% for walleye, and the percent of marks recaptured decreased for all three index species (Golder 27; Table 4.8). Table 4.8 Total number of each index species fish caught, marked, and recaptured in sampled sections of the Lower Columbia River, 27 September to 6 November 27. Species Size-Cohort Number Caught Number Marked Number of Recaptures Percent Recapture Mountain whitefish 25 mm FL <25 mm FL All Rainbow trout 25 mm FL <25 mm FL All Walleye All Parameter estimates from the AIC c selected model for all mountain whitefish, large rainbow trout, small rainbow trout, and walleye are presented in Appendix F, Tables F1, F2, F3, and F4, respectively. A list of the candidate models assessed and their associated AIC c scores from Program MARK for mountain whitefish, large rainbow trout, small rainbow trout, and walleye, are provided in Appendix F, Tables F5, F6, F7, and F8, respectively Mountain Whitefish The variances of recapture rates for the two size groups of mountain whitefish (i.e., large and small fish) were homogenous, so these data were combined for analysis (Appendix D, Table D4). Most (82%) of the large mountain whitefish caught during the first four sessions were marked and released; 1 fish (1.9% of tags released) were recaptured (Table 4.8). Estimates of population abundance from the modified Schnabel ( fish) and the sequential Bayes algorithm ( fish) were similar (Figure 4.22). Confidence intervals for both methods were large due to the small numbers of recaptures. Program MARK generated population estimates for the combined mountain whitefish catch (i.e., all size-cohorts combined) only. Most (86%) of the small mountain whitefish caught during the first four sessions were marked and released; 12 fish were recaptured (2.%; Table 4.8). Estimates of population abundance from the modified Schnabel ( fish) and the sequential Bayes algorithm (2 997 fish) were similar (Figure 4.22). Confidence intervals for both methods were large due to the small number of recaptures. Program MARK generated population estimates for the combined mountain whitefish catch (i.e., all size-cohorts combined) only. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 51

67 Population Abundance Modified Schnabel Sequential Bayes Algorithm Program MARK 1 Figure 4.22 Mountain whitefish Rainbow trout Walleye Population estimates with 95% confidence intervals using the modified Schnabel, sequential Bayes algorithm, and Program MARK for index species in the Lower Columbia River, 27. The red dots denote small (i.e., <25 mm FL) fish; the blue dots denote large (i.e., 25 mm FL) fish; the black dots denote both size cohorts combined. Estimates of abundance for the combined mountain whitefish catch based on the modified Schnabel ( fish), sequential Bayes algorithm ( fish), and Program MARK ( fish) were similar. All three methods showed an increase in mountain whitefish population abundance estimates from 26 to 27 although confidence intervals overlapped for all estimates (Figure 4.23). Population Abundance Sequential Bayes Algorithm Modified Schnabel Program MARK Year Figure 4.23 Population estimates for mountain whitefish in the Lower Columbia River study area, 22 to 27. Estimates were generated using the modified Schnabel (blue lines), sequential Bayes algorithm (red lines), and Program MARK (black lines) estimation techniques. The solid lines represent the estimates; the dotted lines represent upper and lower 95% confidence intervals. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 52

68 4.3.2 Rainbow Trout Recapture rates for the two size-cohorts of rainbow trout (i.e., large and small fish) were homogenous, so these data were combined for analysis (Appendix D, Table D5). In total, 186 (94%) large rainbow trout caught during the first four sessions were marked and released; 96 fish (8.8%) were recaptured (Table 4.8). Estimates of large rainbow trout abundance based on the modified Schnabel (8964 fish), the sequential Bayes algorithm (918 fish), and Program MARK (7668 fish) were similar (Figure 4.22). Most (76%) of the small rainbow trout caught during the first four sessions were marked and released; 1 fish (1.3%) were recaptured (Table 4.8). Estimates of small rainbow trout abundance based the modified Schnabel (68 fish), the sequential Bayes algorithm (752 fish), and Program MARK (393 fish) were similar (Figure 4.22). Estimates of combined rainbow trout abundance from Program MARK (7598 fish; Figure 4.22) were lower than those based on the modified Schnabel (9767 fish) and sequential Bayes algorithm (9917 fish). Rainbow trout population estimates (all size-cohorts combined) increased from 25 to 27 (though confidence intervals overlap). The 27 estimates were higher than in 23 to 26, for all methods (Figure 4.24) Sequential Bayes Algorithm Modified Schnabel Program MARK Population Abundance Year Figure 4.24 Population estimates for rainbow trout in the Lower Columbia River study area, 22 to 27. Estimates were generated using the modified Schnabel (blue lines), sequential Bayes algorithm (red lines), and Program MARK (black lines) estimation techniques. The solid lines represent the estimates; the dotted lines represent upper and lower 95% confidence intervals. In 24, differences in recapture rates between the two size-cohorts of rainbow trout (i.e., fish 25 mm FL and fish <25 mm FL) prevented the generation of an all size-classes combined population estimate using the modified Schnabel and sequential Bayes Algorithm estimation techniques. Data presented represents the sum total of the two size-class population estimates. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 53

69 4.3.3 Walleye Most (96%) of all walleye caught during the first four sessions were marked and released; 8 fish (1.9%) were recaptured (Table 4.8). Estimates of walleye abundance based on the modified Schnabel (11 18 fish), the sequential Bayes algorithm ( fish), and Program MARK (1 526 fish) were similar (Figure 4.22). Confidence intervals for the modified Schnabel and sequential Bayes algorithm methods were large due to the small numbers of recaptures. The 27 walleye population estimates increased from 26 to 27 (for all three estimation techniques; Figure 4.25) although confidence intervals overlapped among all years for all methods Sequential Bayes Algorithm Modified Schnabel Program MARK 3 Population Abundance Figure 4.25 Population estimates for walleye in the Lower Columbia River study area, 22 to 27. Estimates were generated using the modified Schnabel (blue lines), sequential Bayes algorithm (red lines), and Program MARK (black lines) estimation techniques. The solid lines represent the estimates; the dotted lines represent upper and lower 95% confidence intervals. Year 4.4 Comparison of Estimated Populations Population estimates for large mountain whitefish increased by 83% between 26 and 27, with an associated probability of increase in abundance estimated at 94%. Estimates for small mountain whitefish estimates increased by 35% between 26 and 27, with an associated probability of increase in abundance estimated at 71%. Estimates for Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 54

70 all mountain whitefish combined increased by 62% between 26 and 27, with an associated probability of increase in abundance estimated at 95% (Figure 4.26). Percent Change (Between Years) mm FL < 25 mm FL All Fish Year Figure 4.26 Changes in Bayesian population estimates for large ( 25 mm FL), small (<25 mm FL), and combined (all fish) mountain whitefish in the Lower Columbia River, 22 to 27. The solid lines denote the percent change in population abundance between years and the dotted lines denote the probability of a population decrease between years Probability of a Population Decrease (Between Years) Large rainbow trout population estimates increased by 59% between 26 and 27, with an associated probability of increase in abundance estimated at >99%. Small rainbow trout estimates decreased by 66% between 26 and 27, with an associated probability of decrease in abundance estimated at 99%. All rainbow trout combined population estimates increased by 31% between 26 and 27, with an associated probability of increase in abundance estimated at 98%. Overall (all size cohorts combined), rainbow trout population abundance has increased by 5% since 25 (Figure 4.27). Walleye population estimates increased by 59% between 26 and 27, with an associated probability of increase in abundance estimated at 92% (Figure 4.28). Phase 7 (27) is first year since 23 that walleye population estimates have exhibited this level of probable increase. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 55

71 2 1. Population Change (Between Years) mm FL < 25 mm FL Probability of a Population Decrease (Between Years) Year Figure 4.27 Changes in Bayesian population estimates for large ( 25 mm FL) and small (<25 mm FL) rainbow trout in the Lower Columbia River, 22 to 27. The solid lines denote the percent change in population abundance between years and the dotted lines denote the probability of a population decrease between years Percent Change (Between Years) Probability of a Population Decrease (Between Years) Year Figure 4.28 Changes in Bayesian population estimates for walleye (all size-cohorts combined) in the Lower Columbia River, 22 to 27. The solid lines denote the percent change in population abundance between years and the dotted lines denote the probability of a population decrease between years. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 56

72 The estimated probability of any specified change in population size from 26 to 27 was calculated for mountain whitefish, rainbow trout, and walleye. The cumulative probability distributions for these species are presented in Figures 4.29 to 4.31, respectively. Probability 1. Mountain.8 Whitefish 25 mm FL Population Change Population Change Population Change Population Change Population Change. Mountain.8 Whitefish <25 mm FL Population Change Population Change Population Change Population Change Population Change. Mountain.8 Whitefish All Population Change Population Change Population Change Population Change Population Change Figure 4.29 Change in Population Size Cumulative probability curves for a possible change in population size for mountain whitefish in the Lower Columbia River (based on changes in Bayesian population estimates). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 57

73 Probability Rainbow Trout 25 mm FL Population Change Population Change Population Change Population Change Population Change Rainbow Trout <25 mm FL Population Change Population Change Population Change Population Change Population Change Rainbow Trout All Population Change Population Change Figure 4.3 Change in Population Size Cumulative probability curves for a possible change in population size for rainbow trout in the Lower Columbia River (based on changes in Bayesian population estimates) All Walleye Probability Population Change Population Change Population Change Population Change Population Change Population Change Change in Population Size Figure 4.31 Cumulative probability curves for a possible change in population size for walleye in the Lower Columbia River (based on changes in Bayesian population estimates). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 58

74 4.5 Age-Cohort Analysis For rainbow trout, the maximum length was fixed to 6 mm FL and the maximum age was fixed at age-9. The best fit for rainbow trout had 8 age-classes (age-, age- 1 to age-6, and age-7+, where age-7+ included all fish age-7 and older). For mountain whitefish the maximum length was fixed at 45 mm FL and the maximum age was fixed at age-1. The best fit for mountain whitefish had 1 age-classes (age-, age-1 to age-8, and age-9+), with the oldest fish pooled into the age-9+ age-class. Estimates of the proportion of mountain whitefish in each of these age-classes and their associated mean length for 27 are presented in Figure Estimates for mountain whitefish for 21 to 26 are presented in Appendix E, Figures E4 to E9, respectively. Results of the age-cohort analysis for mountain whitefish suggest an increase in abundance of the age- and age-1 cohorts in 27 (Figure 4.32) and 26 (Appendix E Figure E9), approaching levels similar to 21 (Appendix E Figure E4). During the 23, 24, and 25 (Appendix E Figures E6 to E8), proportions of these age cohorts were lower. Predicted lengths-at-age for mountain whitefish generated using Program MIX were similar to results of the mountain whitefish von Bertalanffy analysis (Figure 4.7) over all ages and sample years Length-frequency data Sum of individual age distributions Estimated length-at-age distribution Predicted length-at-age Frequency of Occurrence Fork Length (mm) Figure 4.32 Predicted length-at-age distribution curve for mountain whitefish captured by boat electroshocking in the Lower Columbia River, 27 September to 6 November 27. Graph includes age- to age-9+ (i.e., age-9 and older) cohorts. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 59

75 Estimates of the proportion of rainbow trout in each of these age-classes and their associated mean length for 27 are presented in Figure 4.33.Estimates for rainbow trout for 21 to 26 are presented in Appendix E, Figures E1 to E15, respectively. The age-cohort analysis for rainbow trout suggests variability in abundance of the age- and age-1 cohorts between years (Figure 4.33, Appendix E, Figures E1 to E15). Reasons for the variability are unknown, but could be related to flow variations, predation, and other environmental and life history variables. Predicted lengths-at-age for rainbow trout generated using Program MIX suggest a slower growth rate than those from the von Bertalanffy analysis (Figure 4.14) over all ages Length-frequency data Sum of individual age distributions Estimated length-at-age distribution Predicted length-at-age Frequency of Occurrence Fork Length (mm) Figure 4.33 Predicted length-at-age distribution curve for rainbow trout captured by boat electroshocking in the Lower Columbia River, 27 September to 6 November 27. Graph includes age- to age-7+ (i.e., age-7 and older) cohorts. 4.6 Age Structured Mark-Recapture Analysis Mountain Whitefish ASMR analyses on mountain whitefish data provide improved estimates of recruitment to the population by incorporating all captured fish age and length information into the model. Initial testing of the models involved all age-classes of mountain whitefish. Since all ASMR models with age- mountain whitefish failed to fit the data, i.e., failed to converge or resulted in a numerical error during the fitting process, only models with captures and recaptures of age-1+ (age-1 and older) mountain whitefish were considered. After eliminating age- fish, four of the six Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 6

76 ASMR models successfully fitted the data. ASMR 1, with a migration parameter of.4, was the top model (Table 4.9). It predicted that the number of adult mountain whitefish fell from over 15 individuals in 23 to under 5 individuals in 27 (Figure 4.34). ASMR 3 predicted that in 21 the number of adult mountain whitefish was anywhere from to approaching 1 million individuals indicating that the data are incapable of supporting a model that estimates 61 parameters. Table 4.9 Age-Structured Mark-Recapture (ASMR) model evaluation results for mountain whitefish captured in the Lower Columia River, 21 to 27. AIC: Akaike Information Criterion, which measures the goodness of fit of the model. Lower values are better. ASMR Model Migration Number of Parameters AIC AIC An important advantage of the ASMR approach is that it allows the abundance of recruits to be estimated for years prior to the start of data collection based on the age-specific annual mortality rate ( Ŝ a ) and the abundance of unmarked fish in the first year of data collection ( ˆ ). However, it is important to realize that the abundance of recruits is the U a, t abundance of the earliest age-class included in the analysis that recruits to the gear. ASMR 1 with a migration parameter of.4 suggests that recruitment was highest from 1996 to 23. However, there is a great deal of uncertainty surrounding the exact number of recruits during this period (Figure 4.35) Number of Adults Figure Year Adult (age-3+) mountain whitefish abundance (black line) as estimated by ASMR 1 with a migration parameter of.4. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 61

77 5e+5 4e+5 Number of Recruits 3e+5 2e+5 1e+5-1e Figure 4.35 Brood Year Mountain whitefish recruits (age-1; black line) as estimated by ASMR 1 with a migration parameter of.4. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing Rainbow Trout All six ASMR models were successfully fitted to the rainbow trout data. Although ASMR 3, with a migration parameter of., was the top model (AIC:-79 65), the capture probabilities, which varied widely between ages and years, suggested that the model was over-parameterised. This conclusion was further supported by the relatively high levels of uncertainty surrounding the abundance estimates, by an apparently spurious spike in the number of recruits (in brood years 1999 and 2), and by the fact that ASMR 3 estimated 69 parameters. The failure of the information theoretic criteria to avoid selection of an over-parameterised model (Burnham and Anderson 22) may reflect that information theoretic criteria are not strictly valid when used with ASMR, as ASMR s equations are not based on pure likelihoods (Coggins 26a), and that AIC was used instead of AIC c which corrects for small sample sizes. When only ASMR 1 and 2 were considered, ASMR 2, with a migration parameter of., provided the best fit to the data and did not appear to be over-parameterised (Table 4.1). This model suggests that the number of adult (age-2+) rainbow trout increased from about 7 in 21 to about 1 in 27 (Figure 4.36). It is noteworthy that the uncertainty inherent to the model was greater than the uncertainty due to ageing only a sub-sample of fish. When migration was fixed at.48, the abundance of adults as predicted by ASMR 2 was approximately halved throughout the time series with the trend remaining relatively constant (Figure 4.37). In addition, the uncertainty due to ageing a subsample of fish increased to levels comparable with the uncertainty inherent in the model. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 62

78 Table 4.1 Age-Structured Mark-Recapture (ASMR) model evaluation results for rainbow trout captured in the Lower Columia River, 21 to 27. AIC: Akaike Information Criterion, which measures the goodness of fit of the model. Lower values are better. ASMR Model Migration Number of Parameters AIC AIC Number of Age-2+ Rainbow Trout Figure Year Adult (age-2+) rainbow trout abundance (black line) estimated by ASMR 2 with a migration parameter of.. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing. ASMR 2, with no migration, indicated that the number of rainbow trout recruits (age-1) rose from nearly in brood year 1993 to 25 in 21 before falling slightly (Figure 4.38). Interestingly, at the beginning of the time series the uncertainty was predominantly due to ageing a sub-sample, whereas from 2 onwards most of the uncertainty was inherent in the model. Fixing migration at.48 produced a model with a similar trend, but lower abundances and much greater uncertainty due to ageing (Figure 4.39). The number of recruits in years prior to data collection is calculated by iteratively dividing the number of unmarked fish of each age by the age-specific survival until they reach the age at which they recruit to the gear. In addition, the age-specific survival prior to data collection is assumed not to show any inter-annual variation. Consequently, the back-calculations of the number of recruits should be considered increasingly unreliable as the number of years prior to data collection increases. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 63

79 8 7 Number of Age-2+ Rainbow Trout Figure Year Adult (age-2+) rainbow trout abundance (black line) estimated by ASMR 2 with a migration parameter of.48. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing Number of Recruits Brood Year Figure 4.38 Rainbow trout recruits (age-1; black line) as estimated by ASMR 2 with a migration parameter of.. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 64

80 Perhaps not surprising, given the similarities between them, the abundance estimates of ASMR 1, with and without migrations, were close to those of ASMR 2 for both adults and recruits (not shown) Number of Recruits Brood Year Figure 4.39 Rainbow trout recruits (age-1; black line) as estimated by ASMR 2 with a migration parameter of.48. The blue lines bracket the Bayesian 95% credible interval and the red lines approximate the 95% confidence interval due to incomplete ageing. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 65

81 5. DISCUSSION The primary purpose of the Large River Fish Indexing Program is to monitor populations of resident fish stocks in the Lower Columbia River. These populations are viewed as indices of river conditions as affected by dam operations. To accomplish this objective, changes in methodologies were made during the 27 study based on results of preceding studies (21 to 26). These changes included: the number of boat electroshocking sample sessions was reduced to five (one less than in 26); and, fish were marked exclusively with PIT tags (in 26 fish were marked in an alternating pattern with either a T-bar anchor tag or a PIT tag). Overall, mortalities of index species fish captured in 27 were similar to 24 to 26 results. The proportion of the index species catch that was healthy enough to tag was higher or similar to all other years. These improvements (i.e., decreased mortality rates and higher proportion of tagged fish) are believed to result from several changes in methodologies made over the years, including altered electroshocker settings, the addition of compressed oxygen and Rejuvenade TM HRT to the livewell water, and lower catch-rates (which decreased the number of fish in the livewell, thereby decreasing stress and oxygen demands). The present program was designed to develop population estimates for index species and obtain sufficient numbers of fish to quantify population parameter statistics. 5.1 Mountain whitefish Percent Composition, Catch-rates, and Population Abundance The percent composition of mountain whitefish within the sportfish catch ranged from a high of 5% in 22 to a low of 35% in 23 (Figure 5.1). Population abundance estimates, based on the modified Schnabel and sequential Bayes algorithm, and catch-rates suggest a decrease in mountain whitefish abundance from 22 to 26 and an increase from 26 to 27. Program MARK and relative abundance suggest that population abundance of mountain whitefish has been increasing since 22. All mountain whitefish population estimates had wide confidence intervals due to the low number of recaptured individuals. All metrics suggest an overall increase in mountain whitefish abundance from 26 to 27, except the ASMR done on the large mountain whitefish, and the Program MARK analysis done on fish greater than age-2 (Figure 5.1). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 66

82 ASMR Age-3 Abundance Population Abundance Program MARK Modified Schnabel Sequential Bayes Algorithm Relative Abundance Catch-rate ASMR Age-3+ Abundance Catch-rate (fish/km/hr) Relative Abundance (in the sportfish catch) 2 Figure Year Population estimates, catch-rates, abundance, and ASMR age-3+ abundance of mountain whitefish captured in the Lower Columbia River, 22 to 27. Inconsistencies between metrics are partially due to differing assumptions between techniques. The decline in the ASMR values for fish older than age-3, and Program MARK estimates on fish greater than ~27 mm FL suggest that the adult population has likely declined substantially from 23 to 27. The increase in the population as a whole from 26 appears to be mainly due to large recruit cohorts in 26 and 27 since the analyses on the adults show a decreasing trend. The ASMR analysis for mountain whitefish recruits is inconclusive due to its wide confidence intervals. The Program MARK results include both recruits and adults and show a population increase from 26 to 27. As ASMR and Program MARK analyses make the most realistic assumptions, their estimates should be considered the most reliable. It is reassuring, therefore, that the trends predicted by the two methods are similar. The hypothesis that predation may be a partial cause of the decline in the mountain whitefish population is supported by the trend in walleye abundance. From 23 to 25, walleye abundance estimates were elevated compared to other study years and age- mountain whitefish are a major food source for walleye. A higher predation rate on the age- cohort would affect absolute and relative abundance and catch-rates, but would have less of an effect on the three population abundance estimates models (as a portion of this cohort was too small to mark). The theory of walleye predation is further supported by the response of the age- and age-1 mountain whitefish abundance estimates from 25 to 27 (that followed a decline in walleye abundance; see Figures 4.23 and 4.25). Population estimates for small Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 67

83 mountain whitefish have increased by as much as 399% (based on differences in the Bayesian mode) between 25 and 27 (Figure 4.26; Golder 26, 27) and also showed a potential increase in the ASMR model. However, this possible relationship should be viewed cautiously as it is based on very few data points. A possible reason for the increase in mountain whitefish abundance estimates from 26 to 27 (other than an actual increase in abundance) could be related to sample period. In 27, sampling was delayed by approximately one week (see Section 1.4). This delay meant that sampling in 27 was conducted closer to the mountain whitefish spawning season than during previous years. Mountain whitefish typically congregate in spawning areas prior to the spawning season, where shallow water and lack of available cover makes them easier to capture. In 27, catch-rates for mountain whitefish in the Kootenay section (a major spawning area for mountain whitefish) were higher than in most other study years and increased during each successive session. 5.2 Rainbow trout Percent Composition, Catch-rates, and Population Abundance The percent composition of rainbow trout within the sportfish catch ranged from 3% (in 23) to 42% (in 22) but did not exhibit any definite trends between 22 and 27 (Figure 5.2). The changes recorded likely reflected seasonal fluctuations in relative abundance levels rather than absolute changes in rainbow trout population abundance. However, population estimates based on the modified Schnabel and sequential Bayes algorithm, and catch-rates, suggest a decline in rainbow trout abundance from 22 to 25 and an increase from 25 to 27. Program MARK suggests a gradual increase in rainbow trout abundance since 23 as do the results from the ASMR analyses (Figure 5.2). The ASMR population estimate for rainbow trout adults increased from approximately 7 to 1 fish from 21 to 27. The ASMR predictions are considered less reliable earlier in the time series. Rainbow trout redd and spawner data collected by BC Hydro during annual aerial spawning surveys suggest a steady increase from 199 to 27 in both the number of redds and the number of spawners enumerated throughout the Lower Columbia River study area (Baxter 27). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 68

84 Population Abundance Figure 5.2 Population estimates, catch-rates, abundance, and ASMR age-2+ abundance of rainbow trout captured in the Lower Columbia River, 22 to Walleye Program MARK Modified Schnabel Sequential Bayes Algorithm Relative Abundance Catch-rate ASMR Age-2+ Abundance Year Percent Composition, Catch-rates, and Population Abundance The three population abundance models and the catch-rate data suggest a large increase in walleye abundance from 22 to 23 followed by a gradual decrease in abundance from 23 to 26 and another increase from 26 to 27. This pattern of abundance is the result of unusually high recruitment for the 21 brood year (see Section 4.2.3), which was first recorded migrating into the study area in 23. As the 21 brood year matured from 24 to 26 it decreased in abundance, which in turn decreased overall walleye population abundance. Reasons for the abundant 21 walleye brood year are unknown, but as walleye are seasonal residents and do not spawn in the study area, it is not believed to be related to flows in the Columbia River within the study area Catch-rate (fish/km/hr) Relative Abundance (in the sportfish catch) Population abundance estimates suggest an increase in walleye abundance from 26 to 27, while catch-rates and percent composition show a decrease. The abundance estimates should be interpreted with some caution since there were low numbers of recaptured individuals (n = 8) in 27. Walleye recapture rates in 27 (1.9%) were lower than in all other years (22 to 26; Table 4.7). Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 69

85 Population Abundance Program MARK Modified Schnabel Sequential Bayes Algorithm Relative Abundance Catch-rate Catch-rate (fish/km/hr) Relative Abundance (in the sportfish catch) Year 1 4 Figure 5.3 Summary comparison of population estimates, catch-rates, and abundance of walleye captured in the Lower Columbia River, 22 to 27. Based on previous studies (McLellan et. al.1999), the bulk of the walleye population recorded during the 23 to 25 studies was likely composed of individuals from the 21 brood year. The metrics examined suggest an inverse relationship between walleye abundance and redside shiner abundance (Table 4.2). Redside shiner are a major prey fish for walleye and the observed abundance of redside shiner within the non-sportfish catch decreases in years when walleye abundance increases, and increases in years when walleye abundance decreases (Table 4.2). Estimates of redside shiner abundance are based solely on netter observations and can be strongly influenced by water surface visibility, water clarity, and estimation errors. Additional years of data may be able to demonstrate if the trend between walleye and redside shiner abundance persists. 5.4 PIT Vs. T-Bar Anchor Tag Comparison For mountain whitefish tagged in 25 or 26 and recaptured one year later, average growth rates and inter-year recapture rates were higher for fish marked with PIT tags than for fish marked with T-bar anchor tags. In addition, Program MARK results suggested that mountain whitefish marked with PIT tags had higher inter-year survival than Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 7

86 mountain whitefish marked with T-bar anchor tags. These results indicate that T-bar anchors tags had a negative effect on growth and survival of mountain whitefish between 21 and 26, and may have influenced metrics that rely on inter-year recapture. There were no significant differences in average growth rates, inter-year recapture rates, or inter-year survival rates among rainbow trout and walleye marked with PIT or T-bar anchor tags. 5.5 Program MARK Program MARK provides the tools to combine Cormack-Jolly-Seber estimates of survival with closed capture models to create a robust model for survival and population abundance estimates based on tags recovered in the following year. This method has been used in the Lower Columbia River indexing program since 23. This year a number of modifications were made to the method, among which the most notable were: 1) conducting three separate analyses for each of the three index species; and, 2) separately estimating the difference in survival and other parameters for fish marked with PIT tags and fish marked with T-bar anchor tags. For mountain whitefish, eight models were given some Akaike weighting, but the majority of the model weight was allotted to the top two models (Appendix F, Table F5). The first ranked model (Akaike weight of.75) varied survival through time, had constant immigration and emigration levels, varied probabilities of capture and recapture by capture session and by year, and assessed abundance by tag type and through time. The second ranked model (Akaike weight of.15) was identical except that survival varied by tag type. Estimated survival ranged from 3.1 to 44.4% (95% CI of 16.1 to 64.6%). Immigration/emigration was estimated at 39.7% (95% CI of 14.6 to 71.7%). The probability of capture was equal to the probability of recapture and ranged from.4 to 1.3%. Rainbow trout were separated into small and large size-cohorts. For small rainbow trout, the model set generated three models that garnered all of the Akaike weight (Appendix F, Table F7). The only difference between the top three ranked models was in how the survival parameter was estimated. The top model estimated a constant survival between tag type and through time (Akaike weight of.54), while the second ranked model estimated survival to vary through time (Akaike weight of.27), and the third ranked model estimated survival to vary between tag type (Akaike weight of.2). The estimated survival after model averaging across the three top models ranged from 16. to 2.2% (95% CI of 3. to 55.5%). Immigration/emigration for small rainbow trout was estimated at 52.7% (95% CI of 7.3 to 94.%)and considered random. The probability of capture was equal to the probability of recapture (5.1%) and also was modelled to be constant through time. The fact that the third ranked model included differential survival by tag type is not contradictory to the result that was discussed in Section 5.4. The test reported in Section 5.4 uses a traditional hypothesis testing paradigm, while Program MARK is set within an information-theoretic context. These paradigms Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 71

87 differ considerably in how they assign importance to a model. Whereas the frequentist hypothesis testing paradigm is a yes or no at a set p-value level, the information-theoretic context ranks models relatively. For large rainbow trout, the model set also had three models with Akaike weight, though the majority (.97 of the weighting) was allotted to the top model. The top model estimated that survival differed between tag type and through time. For PIT-tagged, large rainbow trout the estimate of survival was 5.8% (95% CI from 36.1 to 65.5%) while for T-bar tagged fish, the survival estimates ranged from 21.4 to 6.5% (95% CI from 13.2 to 81.5%). Immigration/emigration was mainly considered constant (i.e., due to the high weighting of the top model that defined migration in this way), but did vary slightly through time (43% with a range from 31 to 58%). The probability of capture varied by phase and by capture session within each phase and ranged from.39 to 4.96%. The top model for walleye received 1% of the Akaike weighting (Appendix F, Table F8) and estimated that survival varied by tag type, immigration/emigration varied through time, and the probability of capture was equal to the probability of recapture and varied through capture session and among phases. The survival of walleye marked with PIT tags was estimated at 49.5% (95%CI from 24.8 to 74.5%), higher than walleye marked with T-bar anchor tags who had an estimated survival of 47.9% (95% CI from 37.6 to 58.4%). Immigration/emigration for walleye was estimated to range from 3.9 to 58.5% and the probability of capture ranged from.3% to 2.44%. The low recapture rates for this species likely affect the complexity of model that can be fitted to the data. 5.6 ASMR Analysis The ASMR analyses used mark-recapture within a virtual population analysis framework to estimate the abundance of rainbow trout and mountain whitefish. The results of the analyses were consistent with our current understanding of the changes in abundance derived from the other methods. However, ASMR has a number of advantages over the other methods. Firstly, it provides abundance estimates by age. This is particularly important when trying to understand the influence of environmental changes on a population since it provides key information about the population s long-term dynamics. Secondly, it allows the abundance of recruits to be estimated for each year of the study as well as prior years, although the abundance estimates for prior years become increasingly less reliable as time recedes. Thirdly, the analyses are performed using computer scripts, which allows the models to be extended or modified. Future modifications could include multiple capture sessions within a year and the modelling of inter-annual variation in survival as a randomeffect. However, like all methods, ASMR has its limitations. In particular, the assumption that survival increases with age in an allometric manner could compromise the abundance estimates. This assumption may have lead to the failure of the models to fit the mountain whitefish data when age- fish were included. Nevertheless, when ageing is reliable, ASMR is a valuable method that has already provided important insights into index species populations structure in the LCR. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 72

88 6. CONCLUSION AND RECOMMENDATIONS The key findings of the Phase 7 study are summarized as follows: There appeared to be lower age- mountain whitefish recruitment in 23, 24, and 25 relative to 21, 22, 26, and 27. This difference is possibly related to increased predation by walleye on the age- cohort over that time period. The ASMR analysis suggests a substantial decline in the adult mountain whitefish population since 23. The uncertainty surrounding the ASMR estimates of the number of mountain whitefish recruits precludes any clear conclusions about changes in abundance. The Program MARK analysis on fish greater than the mean length-at-age for age-2+ noted the same decline in the adult population. Rainbow trout age- recruitment was lowest from 23 to 25 and was also low in 27. These reductions may be partially due to predation pressure or to other environmental factors. Walleye populations showed one mode between 351 and 45 mm FL, composed of four separate age-cohorts (the 21 to 24 brood years). The oldest and most numerous of these four cohorts was the 21 brood year, first observed in 23. Ageing rainbow trout and mountain whitefish scales continues to be problematic due to the presence of false annuli or lack of annuli. Population estimates for walleye based on the modified Schnabel, sequential Bayes algorithm, and Program MARK estimation techniques suggest an increase in abundance from 22 to 23, a decrease in abundance from 23 to 26, and an increase in abundance from 26 to 27. Based on catch-rates, relative abundance, and three population abundance estimate models, mountain whitefish abundance increased between 26 and 27, likely due to higher recruitment from the 26 brood year. All estimates exhibited wide confidence intervals due to the low number of recaptured fish. For mountain whitefish, average annual growth rates, inter-year recapture rates, and inter-year survival rates were higher for fish marked with PIT tags than for fish marked with T-bar anchor tags. Based on catch-rates, relative abundance, and the population abundance estimate models, rainbow trout abundance increased from 25 to 27. The ASMR analysis supports a moderate overall increase in adults and a more substantial increase in recruits. One of the objectives of the Phase 7 program was to provide recommendations for the Phase 8 and Phase 9 components of the program. In consideration of the findings above and the overall objectives of the Large River Fish Indexing Program, the following recommendations are provided: The nearshore sample program should be repeated using methods similar to the 27 study. The nearshore sample program should continue to be conducted in September and October when water temperatures are below 15 C. Mountain whitefish, rainbow trout, and walleye should continue to be the index species used for detailed population parameter analyses. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 73

89 Scales should be collected from all mountain whitefish and rainbow trout; however due to problems associated with ageing rainbow trout scales, rainbow trout ageing should be limited to inter-year recaptured individuals. There should be an increased focus on scales collected from older fish to reduce uncertainty in fitting mixed distributions to the sparsely sampled older age-classes. Compressed oxygen, pumped into the livewell through an airstone, should continue to be used to maintain dissolved oxygen levels similar to levels recorded in the river. A commercially available livewell additive (Rejuvenade TM HRT) also should be used. The methods used in 24 to 27 to reduce the frequency of electroshocking-induced injuries on large (>3 mm FL) rainbow trout should be maintained. PIT tags should continue to be used to mark index species fish and 8.5 mm PIT tags should be tested to mark smaller (i.e., <12 mm FL) fish. Population age structures for rainbow trout and mountain whitefish should be developed using a combination of length-frequency data, capture efficiency data from the mark-recapture program, and size-at-age data. Analyses using the sequential Bayes Algorithm and modified Schnabel methods should be reduced, and increased effort should be allocated to ASMR and Program Mark modelling as these two methods show encouraging results. In future years, ASMR model fits should be examined with residuals and the possibility of using random effects to reduce the parameter set should be explored. If emigration/immigration vary through time in the Program MARK analysis, the temporal aspect should be incorporated into ASMR modelling. The ASMR models could be modified to include multiple capture sessions within a year. Different recapture rates between adult and immature fish should be incorporated into Program Mark rather than testing recapture rate differences with chi-squared tests. This should ensure that any weighting that estimates survival differently from adult to immature fish will be reflected in the averaged parameters regardless of whether chi-squared values are statistically significant. The ASMR methods show major differences in recruitment rates occurring over time; consequently these methods should be explored for applicability to earlier data sets to further expand the time series and improve the ability to relate recruitment rates to flow changes. The increase in rainbow abundance estimates corresponds with an increased abundance of spawners. Additional analyses with an expanded data set should examine the relationship between flow and other environmental parameters to recruitment changes adjusted for spawner-recruit relationships. To better characterize the variability in index species population abundance for the Lower Columbia River, a sampling program should be continued that will allow estimation of cohort strength and survival for index species. Substantial changes have occurred in mountain whitefish and rainbow trout populations between the early to mid-199s studies and the present indexing program and even within the seven year span of the present program. The four year gap in data between the 199s and the present studies confounds accurate identification of the reasons for these changes. Therefore, continuation of the program is strongly recommended. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 74

90 7. CLOSURE We trust this report provides the information required at this time. Should there be any questions regarding the contents of this report, or should further information be required please do not hesitate to contact the undersigned. Dustin Ford, B.Sc., R.P.Bio. Golder Associates Ltd. Fisheries Biologist Larry Hildebrand, B.Sc., R.P.Bio. Golder Associates Ltd. Senior Biologist, Principal Robyn Irvine, Ph.D., R.P.Bio. Golder Associates Ltd. Statistical Ecologist Joseph Thorley, Ph.D., R.P.Bio. Poisson Consulting Ltd. Fish Population Biologist Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 75

91 8. LITERATURE CITED Burnham, K. P., and D. R. Anderson. 22. Model selection and multimodel inference: a practical information-theoretic approach. Second Edition. Springer, New York. Coggins, L. G. Jr. 27. Abundance trends and status of the Little Colorado River population of humpback chub: an update considering data. U.S. Geological Survey Open-File Report , 53p. Coggins, L. G. Jr., W. E. Pine III, C. J. Walters and S. J. D. Martell. 26a. Age-structured mark-recapture analysis: a virtual-population-analysis-based model for analyzing age-structured capture-recapture data. North American Journal of Fisheries and Management 26: Coggins, L. G. Jr., W. E. Pine III, C. J. Walters, D. R. V. Haverbeke, D. Ward and H. C. Johnstone. 26b. Abundance trends and status of the Little Colorado River Population of Humpback Chub. North American Journal of Fisheries and Management 26: Cormack, R. M Estimates of survival from the sighting of marked animals. Biometrika 51: Fabens, A.J Properties and fitting of the von Bertalanffy growth curve. Growth Sep;29: Gazey, W.J. and M.J. Staley Population estimates from mark-recapture experiments using a sequential Bayes algorithm. Ecology 67: Gelman, A., J. B. Carlin, H. S. Stern and D. B. Rubin. 24. Bayesian data analysis. Second Edition. Chapman and Hall/CRC, Boca Raton, Florida. Golder Associates Ltd. 22. Lower Columbia River fish community indexing program. 21 Phase 1 investigations. Report prepared for BC Hydro, Burnaby, B.C. Golder Report No F: 52p + 6 app. Golder Associates Ltd. 23. Large River Fish Indexing Program Lower Columbia River 22 Phase 2 Investigations. Report prepared for B.C. Hydro, Burnaby, B.C. Golder Report No F: 47 p. + 5 app Golder Associates Ltd. 24. Large River Fish Indexing Program Lower Columbia River 23 Phase 3 Investigations. Report prepared for B.C. Hydro, Burnaby, B.C. Golder Report No F: 54 p. + 6 app Golder Associates Ltd. 25. Large River Fish Indexing Program Lower Columbia River 24 Phase 4 Investigations. Report prepared for B.C. Hydro, Burnaby, B.C. Golder Report No F: 57 p. + 6 app Golder Associates Ltd. 26. Large River Fish Indexing Program Lower Columbia River 25 Phase 5 Investigations. Report prepared for B.C. Hydro, Burnaby, B.C. Golder Report No F: 56 p. + 6 app Hildebrand, L., and K. English Lower Columbia development - Lower Columbia River fisheries inventory, 199 studies. Volume I. Main Report Prepared for BC Hydro, Environmental Resources by R.L. & L. Environmental Services Ltd. and LGL Limited. 17 p. + 7 app. Hilborn, R., and C.J. Walters Quantitative fisheries stock assessment: choice, dynamics and uncertainty. Routledge, Chapman & Hall, Inc. New York. 57 p. Jolly, G. M Explicit estimates from capture-recapture data with both death and immigration stochastic model. Biometrika 52: Kendall, W. L., and J. D. Nichols On the use of secondary capture-recapture samples to estimate temporary emigration and breeding proportions. Journal of Applied Statistics 22: Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 76

92 Kendall, W. L., K. H. Pollock, and C. Brownie A likelihood-based approach to capture-recapture estimation of demographic parameters under the robust design. Biometrics 51: Kendall, W. L., J. D. Nichols, and J. E. Hines Estimating temporary emigration using capture-recapture data with Pollock's robust design. Ecology 78: Laake, J.26. RMark: R Code for MARK Analysis. R package version Macdonald, P. and J. Du. 24. Mixdist: Mixture Distribution Models. R package version Macdonald, P.D.M. and T.J. Pitcher (1979). Age-groups form size-frequency data: a versatile and efficient method of analysing distribution mixtures. Journal of the Fisheries Research Board of Canada 36, Mackay, W.C., G.R. Ash and H.J. Norris Fish ageing methods for Alberta. R.L. & L. Environmental Services Ltd. in association with Alberta and Wildlife Division and University of Alberta, Edmonton. 133p. McLellan, J.G., H.J. McLellan, and A.T. Scholz Assessment of the lake Roosevelt walleye population: A compilation of data from Annual Report. Contributions to Fisheries Management in Eastern Washington State Number 3, March 22. Eastern Washington University. Fisheries Research Center. Department of Biology. Cheney, Washington 994. Otter Research Ltd. 2. An introduction to AD Model Builder version 4: for use in nonlinear modeling and statistics. Pollock, K.H A capture-recapture design robust to unequal probability of capture. Journal of Wildlife Management 46: R Development Core Team. 27. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN , URL Ricker, W.E Computation and interpretation of biological statistics of fish populations. Fisheries Research Board of Canada, Bulletin 191: 382. R.L. & L. Environmental Services Ltd. 1993a. Lower Columbia development - Lower Columbia River fisheries inventory, 1991 and 1992 studies. Data Report Prepared for BC Hydro, Environmental Resources. R.L. & L. Report No. 33D: 23 p. + 4 app. R.L. & L. Environmental Services Ltd. 1993b. Waneta Expansion Feasibility Study: 1992 fisheries study results and preliminary impact assessment. Report prepared for BC Hydro, Environmental Resources. R.L. & L. Report No. 35D: 51 p. + 1 app. R.L. & L. Environmental Services Ltd Waneta Expansion Feasibility Study: fisheries investigations data report. Report prepared for BC Hydro. R.L. & L. Report No D: 66 p. + 5 app. R.L. & L. Environmental Services Ltd. 1995a. Columbia Development Review Lower Columbia River fisheries inventory, 1993 study data report. Report prepared for BC Hydro, Environmental Resources. R.L. & L. Report No F: 33p. + 4 app. R.L. & L. Environmental Services Ltd. 1995b. Columbia Basin Developments Lower Columbia River. Fisheries Inventory Program 199 to Report Prepared for BC Hydro, Environmental Affairs, Vancouver, B.C., by R.L. & L. Environmental Services Ltd., Castlegar, B.C. R.L. & L. Report No D: 156 p. + 7 app. R.L. & L. Environmental Services Ltd. 1995c. Shallow-water habitat use by dace spp. and sculpin spp. in the Lower Columbia River. Report Prepared for BC Hydro, Burnaby, B.C. R.L. & L. Report No. 398D2: 62 p. + 4 app. Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 77

93 R.L. & L. Environmental Services Ltd. 1995d. Lower Columbia River whitefish monitoring program data report. Report prepared for BC Hydro, Environmental Affairs. R.L. & L. Report No. 44F: 5 p. + 6 app. R.L. & L. Environmental Services Ltd Fisheries inventory in the vicinity of the proposed Indian Eddy boat launch development, 1995 data report. Report Prepared for BC Hydro, Kootenay Generation, Castlegar, B.C. R.L. & L. Report No. 45F: 14 p. + 2 app. R.L. & L. Environmental Services Ltd Lower Columbia River mountain whitefish monitoring program, investigations. Draft Report Prepared for BC Hydro, Kootenay Power Supply/Power Facilities. R.L. & L. Report No. 514D: 11p. + 8 app. R.L. & L. Environmental Services Ltd Analysis of historical mountain whitefish data from the lower Columbia and Kootenay rivers, B.C. Report Prepared for BC Hydro. R.L. & L. Report No. 63D-F: Report A: 3 p. + 3 app.; Report B: 36 p. + 1 app. R.L. & L. Environmental Services Ltd Brilliant Expansion Project: Summary of aquatic inventory data of the Kootenay system near Brilliant Dam. Report prepared for Columbia Power Corporation, Castlegar, B.C. R.L. & L. Report No. 552F: 72 p. + 3 app. R.L. &. L. Environmental Services Ltd. 21. Water Use Plans Environmental information review and data gap analysis. Volume 2: Lower Columbia Keenleyside Project. Prepared for BC Hydro, Burnaby, B.C. by R.L. & L. Environmental Services Ltd. in association with Robertson Environmental Services Ltd., Pandion Ecological Research Ltd., Bruce Haggerstone Landscape Architect, Pomeray & Neil Consulting Inc, and DVH Consulting. R.L. & L. Report No. 858V2-F: 455p. Seber, G. A. F A note on the multiple recapture census. Biometrika 52: White, G.C., and K.P. Burnham Program MARK: Survival estimation from populations of marked animals. Bird Study 46 Supplement, Large River Fish Indexing Program - Lower Columbia River - 27 Phase 7 Investigations Page 78

94 APPENDIX A HABITAT SUMMARY INFORMATION

95 Table A1 Descriptions of categories used in the Lower Columbia River Bank Habitat Types Classification System. Category Code Description Armoured/Stable A1 Banks generally stable and at repose with cobble/small boulder/gravel substrates predominating; uniform shoreline configuration with few/minor bank irregularities; velocities adjacent to bank generally lowmoderate, instream cover limited to substrate roughness (i.e., cobble/small boulder interstices). A2 A3 A4 A5 A6 Banks generally stable and at repose with cobble/small boulder and large boulder substrates predominating; irregular shoreline configuration generally consisting of a series of armoured cobble/boulder outcrops that produce Backwater habitats; velocities adjacent to bank generally moderate with low velocities provided in BW habitats: instream cover provided by BW areas and substrate roughness; overhead cover provided by depth and woody debris; occasionally associated with C2, E4, and E5 banks. Similar to A2 in terms of bank configuration and composition although generally with higher composition of large boulders/bedrock fractures; very irregular shoreline produced by large boulders and bed rock outcrops; velocities adjacent to bank generally moderate to high; instream cover provided by numerous small BW areas, eddy pools behind submerged boulders, and substrate interstices; overhead cover provided by depth; exhibits greater depths offshore than found in A1 or A2 banks; often associated with C1 banks. Gently sloping banks with predominantly small and large boulders (boulder garden) often embedded in finer materials; shallow depths offshore, generally exhibits moderate to high velocities; instream cover provided by pocket eddies behind boulders; overhead cover provided by surface turbulence. Bedrock banks, generally steep in profile resulting in deep water immediately offshore; often with large bedrock fractures in channel that provide instream cover; usually associated with moderate to high current velocities; overhead cover provided by depth. Man-made banks usually armoured with large boulder or concrete rip-rap; depths offshore generally deep and usually found in areas with moderate to high velocities; instream cover provided by rip-rap interstices; overhead cover provided by depth and turbulence. Depositional D1 Low relief, gently sloping bank type with shallow water depths offshore; substrate consists predominantly of fines (i.e., sand/silt); low current velocities offshore; instream cover generally absent or, if present, consisting of shallow depressions produced by dune formation (i.e., in sand substrates) or embedded cobble/boulders and vegetative debris; this bank type was generally associated with bar formations or large backwater areas. D2 D3 Low relief, gently sloping bank type with shallow water depths offshore; substrate consists of coarse materials (i.e., gravels/cobbles); low-moderate current velocities offshore; areas with higher velocities usually producing riffle areas; overhead cover provided by surface turbulence in riffle areas; instream cover provided by substrate roughness; often associated with bar formations and shoal habitat. Similar to D2 but with coarser substrates (i.e., large cobble/small boulder) more dominant; boulders often embedded in cobble/gravel matrix; generally found in areas with higher average flow velocities than D1 or D2 banks; instream cover abundantly available in form of substrate roughness; overhead cover provided by surface turbulence; often associated with fast riffle transitional bank type that exhibits characteristics of both Armoured and Depositional bank types. SPECIAL HABITAT FEATURES BACKWATER POOLS - These areas represent discrete areas along the channel margin where backwater irregularities produce localized areas of counter-current flows or areas with reduced flow velocities relative to the mainstem; can be quite variable in size and are often an integral component of Armoured and erosional bank types. The availability and suitability of Backwater pools are determined by flow level. To warrant separate identification as a discrete unit, must be a minimum of 1 m in length; widths highly variable depending on bank irregularity that produces the pool. Three classes are identified: BW-P1 BW-P2 Highest quality pool habitat type for adult and subadult cohorts for feeding/holding functions. Maximum depth exceeding 2.5 m, average depth 2. m or greater; high availability of instream cover types (e.g., submerged boulders, bedrock fractures, depth, woody debris); usually with Moderate to High countercurrent flows that provide overhead cover in the form of surface turbulence. Moderate quality pool type for adult and subadult cohorts for feeding/holding; also provides moderate quality habitat for smaller juveniles for rearing. Maximum depths between 2. to 2.5 m, average depths generally in order of 1.5 m. Moderate availability of instream cover types; usually with Low to Moderate countercurrent flow velocities that provide limited overhead cover. Continued.

96 Table A1 Concluded. BW-P3 Low quality pool type for adult/subadult classes; moderate-high quality habitat for y-o-y and small juveniles for rearing. Maximum depth <1. m. Low availability of instream cover types; usually with Low-Nil current velocities. EDDY POOL EDDY Represent large (<3 m in diameter) areas of counter current flows with depths generally >5 m; produced by major bank irregularities and are available at all flow stages although current velocities within eddy are dependent on flow levels. High quality areas for adult and subadult life stages. High availability of instream cover. SNYE SN A side channel area that is separated from the mainstem at the upstream end but retains a connection at the lower end. SN habitats generally present only at lower flow stages since area is a flowing side channel at higher flows: characterized by low-nil velocity, variable depths (generally <3 m) and predominantly depositional substrates (i.e., sand/silt/gravel); often supports growths of aquatic vegetation; very important areas for rearing and feeding. Velocity Classifications: Low: <.5 m/s Moderate:.5 to 1. m/s High: >1. m/s

97 Table A2 Length of each bank habitat type by boat electroshocking site within the Lower Columbia River, 27. Length (m) of Bank Habitat Type b Total Section Site a Length A1 A2 A3 A4 A5 A6 A1/A2 A2/A3 D1 D2 D3 D1/D2 BW Eddy (m) Upper ES ES ES ES ES ES ES ES ES Upper Section Total Middle ES ES ES ES ES1A Middle Section Total Lower ES ES ES1A ES ES ES2A ES2B Lower Section Total Kootenay ESK ESK Kootenay Section Total Grand Total a See Appendix B, Figures B1 to B3 for sample site locations. b See Appendix A, Table A1 for bank habitat type descriptions.

98 Table A3 Summary of habitat variables recorded at boat electroshocking sites in the Lower Columbia River, 27 September to 6 November 27. Air Water Water Instream Section Site a Conductivity Session Temperature Temperature Cloud Cover b Surface (µs) Velocity c ( C) ( C) Visibility Water Clarity d Substrate Interstices Woody Debris Turbulence Cover Types (%) Aquatic Vegetation Terrestrial Vegetation Shallow Water Deep Water Kootenay ESK Overcast High High Low Kootenay ESK Clear High High High Kootenay ESK Mostly Cloudy Medium High High Kootenay ESK Mostly Cloudy High High High Kootenay ESK Mostly Cloudy High High High Kootenay ESK Partly Cloudy High High High Kootenay ESK Clear High High High Kootenay ESK Mostly Cloudy High High High 8 2 Kootenay ESK Mostly Cloudy High High High Kootenay ESK Mostly Cloudy High High High 9 1 Lower ES Mostly Cloudy Medium High High Lower ES Mostly Cloudy High High High Lower ES Clear High High High Lower ES Partly Cloudy High High High Lower ES Clear High High High Lower ES Mostly Cloudy Medium High High Lower ES Mostly Cloudy High High High Lower ES Clear High High High Lower ES Partly Cloudy High High High Lower ES Clear High High High Lower ES1A Mostly Cloudy Medium High High Lower ES1A Mostly Cloudy High High High Lower ES1A Clear High High High Lower ES1A Clear Medium High High Lower ES1A Clear High High High Lower ES Clear High High High Lower ES Mostly Cloudy High High High Lower ES Clear High High High Lower ES Clear High High High Lower ES Mostly Cloudy High High High Lower ES Clear High High High Lower ES Mostly Cloudy High High High Lower ES Clear High High High Lower ES Clear High High High a See Appendix B, Figures B1 to B3 for sample site locations. b Clear = <1%; Partly Cloudy = 1-5%; Mostly Cloudy = 5-9%; Overcast = >9%. continued c High = >1. m/s; Medium =.5-1. m/s; Low = <.5 m/s. d High = >3. m; Medium = m; Low = <1. m.

99 Table A3 Continued. Section Site a Session Temperature Air ( C) Water Temperature ( C) Conductivity (µs) Cloud Cover b Water Surface Visibility Instream Velocity c Water Clarity d Substrate Interstices Woody Debris Turbulence Cover Types (%) Aquatic Vegetation Terrestrial Vegetation Shallow Water Deep Water Lower ES Mostly Cloudy High High High Lower ES2A Clear High High High Lower ES2A Mostly Cloudy High High High Lower ES2A Clear High High High Lower ES2A Clear High High High Lower ES2A Mostly Cloudy High High High Lower ES2B Clear High High High Lower ES2B Mostly Cloudy High High High Lower ES2B Clear High High High Lower ES2B Clear High High High Lower ES2B Mostly Cloudy High High High Middle ES Mostly Cloudy High High High Middle ES Clear High High High Middle ES Partly Cloudy High High High Middle ES Overcast High High High Middle ES Clear High High High 8 2 Middle ES1A Mostly Cloudy High High High 5 5 Middle ES Clear High High High Middle ES Partly Cloudy High High High Middle ES Mostly Cloudy High High High Middle ES Clear High High High Middle ES Clear High High High Middle ES Overcast Medium High Medium Middle ES Partly Cloudy High High High Middle ES Partly Cloudy High High High Middle ES Clear High High High Middle ES Mostly Cloudy High High High Middle ES Clear High High High Middle ES Overcast Medium High Medium Middle ES Mostly Cloudy High High High Middle ES Clear Medium High High 9 1 Middle ES Partly Cloudy High High High 8 2 Middle ES Overcast High High High Middle ES Clear High High High 8 2 a See Appendix B, Figures B1 to B3 for sample site locations. b Clear = <1%; Partly Cloudy = 1-5%; Mostly Cloudy = 5-9%; Overcast = >9%. continued c High = >1. m/s; Medium =.5-1. m/s; Low = <.5 m/s. d High = >3. m; Medium = m; Low = <1. m.

100 Table A3 Continued. Section Site a Session Temperature Air ( C) Water Temperature ( C) Conductivity (µs) Cloud Cover b Water Surface Visibility Instream Velocity c Water Clarity d Substrate Interstices Woody Debris Turbulence Cover Types (%) Aquatic Vegetation Terrestrial Vegetation Shallow Water Deep Water Upper ES Partly Cloudy High High High Upper ES Clear High High High Upper ES Mostly Cloudy High Medium High Upper ES Mostly Cloudy High High High Upper ES Mostly Cloudy High High High 9 1 Upper ES Partly Cloudy High Low High Upper ES Clear High Low High Upper ES Mostly Cloudy High Low High Upper ES Clear High Low High Upper ES Clear High Low High Upper ES Partly Cloudy High Low High Upper ES Mostly Cloudy High Low High 8 2 Upper ES Overcast Medium Low High Upper ES Partly Cloudy High Low High Upper ES Clear High Low High Upper ES Partly Cloudy High Low High 1 9 Upper ES Mostly Cloudy High Low High Upper ES Overcast Medium Low High Upper ES Overcast High Low High 2 8 Upper ES Clear High Low High 2 8 Upper ES Mostly Cloudy High Low High Upper ES Mostly Cloudy High Low High Upper ES Overcast High Low High Upper ES Overcast High Low High Upper ES Clear High Low High Upper ES Partly Cloudy High Low High Upper ES Mostly Cloudy High Low High Upper ES Overcast High Low High Upper ES Overcast High Low High Upper ES Clear High Low High Upper ES Partly Cloudy High High High Upper ES Clear High High High Upper ES Mostly Cloudy High High High 8 2 Upper ES Clear Low High High a See Appendix B, Figures B1 to B3 for sample site locations. b Clear = <1%; Partly Cloudy = 1-5%; Mostly Cloudy = 5-9%; Overcast = >9%. continued c High = >1. m/s; Medium =.5-1. m/s; Low = <.5 m/s. d High = >3. m; Medium = m; Low = <1. m.

101 Table A3 Concluded. Section Site a Session Temperature Air ( C) Water Temperature ( C) Conductivity (µs) Cloud Cover b Water Surface Visibility Instream Velocity c Water Clarity d Substrate Interstices Woody Debris Turbulence Cover Types (%) Aquatic Vegetation Terrestrial Vegetation Shallow Water Deep Water Upper ES Mostly Cloudy High High High 9 1 Upper ES Partly Cloudy High Low High Upper ES Mostly Cloudy High Low High 1 Upper ES Overcast Medium Low High 5 95 Upper ES Clear High Low High Upper ES Clear High Low High Upper ES Partly Cloudy High Low High Upper ES Mostly Cloudy Low Low High 8 2 Upper ES Overcast High Low High Upper ES Overcast High Low High Upper ES Clear High Low High a See Appendix B, Figures B1 to B3 for sample site locations. b Clear = <1%; Partly Cloudy = 1-5%; Mostly Cloudy = 5-9%; Overcast = >9%. c High = >1. m/s; Medium =.5-1. m/s; Low = <.5 m/s. d High = >3. m; Medium = m; Low = <1. m.

102 Table A4 Summary of species counts for observed fish adjacent to bank habitat types in the Lower Columbia River, 27 September to 6 November 27. Section Site Species A1 A2 A3 A4 A5 A6 A1+A2 A2+A3 D1 D2 D3 D1+D2 BW Eddy Unknown Upper ES17 Kokanee 32 Lake whitefish 2 Mountain whitefish 512 Northern pikeminnow 6 Rainbow trout 31 Sculpin spp. 295 Sucker spp. 453 Walleye 9 White sturgeon 3 Bull trout 1 ES17 Total 1,344 ES18 Sculpin spp. 1,673 3 Sucker spp Redside shiner Rainbow trout 7 4 Northern pikeminnow Kokanee 11 3 Mountain whitefish 43 3 Walleye 1 Peamouth 1 ES18 Total 2, ES19 Northern pikeminnow Walleye Sucker spp. 1, Sculpin spp Rainbow trout Mountain whitefish Kokanee Burbot 1 Bull trout 1 Redside shiner 21 1 ES19 Total 2, ,523 ES2 Redside shiner 335 Northern pikeminnow 1 Walleye 2 Sucker spp. 1,589 Sculpin spp. 835 Peamouth 1 Kokanee 5 Mountain whitefish 371 Rainbow trout 19 ES2 Total 3,167 ES21 Rainbow trout 13 9 Walleye 4 16 Tench 4 Sucker spp Redside shiner White sturgeon 1 Peamouth 29 Northern pikeminnow 6 27 Mountain whitefish Kokanee Sculpin spp. 13 1,375 ES21 Total 38 2,87 2 ES22 Redside shiner Walleye 4 3 Sucker spp. 28 Sculpin spp Rainbow trout Peamouth 2 19 Northern pikeminnow 2 2 Kokanee Mountain whitefish ES22 Total ES25 Rainbow trout 136 Lake whitefish 8 Walleye 9 Sucker spp. 68 Sculpin spp. 3,99 Redside shiner 2 Peamouth 1 Burbot 1 Kokanee 18 Mountain whitefish 142 Northern pikeminnow 3 ES25 Total 4,396 Total ,344 1, , ,841 1, , , , ,55 3, , , ,396

103 Table A4 Continued. Section Site Species A1 A2 A3 A4 A5 A6 A1+A2 A2+A3 D1 D2 D3 D1+D2 BW Eddy Unknown ES28 Mountain whitefish 63 Kokanee 22 Carp spp. 1 Bull trout 1 Rainbow trout 4 Redside shiner 12 Sculpin spp. 15 Northern pikeminnow 2 Sucker spp. 2,216 Walleye 3 ES28 Total 2,447 ES29 Mountain whitefish 6 56 Northern pikeminnow 4 1 Peamouth Rainbow trout 9 1 Redside shiner Sculpin spp. 1, Sucker spp Walleye 12 2 Bull trout 1 Kokanee 9 8 ES29 Total 2,331 1,27 Upper Section Total 4,958 2,661 4,829 11,583 1,344 1,27 12 Kootenay ESK1 Lake whitefish 1 1 Sucker spp Walleye White sturgeon 1 Sculpin spp Redside shiner Rainbow trout Peamouth 1 Mountain whitefish Kokanee 1 Northern pikeminnow ESK1 Total 229 1, ESK2 Walleye Lake whitefish 3 2 Mountain whitefish Northern pikeminnow Rainbow trout Redside shiner Sculpin spp ,2 13 Sucker spp White sturgeon 1 1 ESK2 Total , Kootenay Section Total ,37 3, Middle ES1 Rainbow trout Peamouth 1 Redside shiner Sculpin spp , Sucker spp Walleye Mountain whitefish Lake whitefish 1 2 Burbot 2 4 Northern pikeminnow 1 5 ES1 Total , ES1A Mountain whitefish 11 Sucker spp. 2 Sculpin spp. 15 Rainbow trout 12 ES1A Total 4 ES7 Redside shiner Dace spp. 1 Walleye Sculpin spp Rainbow trout Peamouth 1 Northern pikeminnow Mountain whitefish Kokanee Burbot 4 Sucker spp ES7 Total Total , , , ,358 26, , , , ,687 6, , , , ,36

104 Table A4 Continued. Section Site Species A1 A2 A3 A4 A5 A6 A1+A2 A2+A3 D1 D2 D3 D1+D2 BW Eddy Unknown ES8 Burbot 5 Sculpin spp. 1,66 99 White sturgeon 1 Walleye 18 7 Sucker spp Redside shiner 9 5 Rainbow trout Northern pikeminnow 3 1 Lake whitefish 3 Mountain whitefish 92 1 ES8 Total 1, ES9 Redside shiner 3 Rainbow trout 19 Walleye 11 Sucker spp. 114 Northern pikeminnow 1 Burbot 3 Mountain whitefish 46 Lake whitefish 1 Sculpin spp. 2,37 ES9 Total 2,73 Middle Section Total 957 4, , Lower ES1-1 Rainbow trout Walleye Sucker spp Redside shiner White sturgeon 1 Mountain whitefish Lake whitefish 1 7 Burbot 1 Sculpin spp ES1-1 Total ES1-2 Rainbow trout Sucker spp Sculpin spp. 3,46 1, Redside shiner 5 Northern pikeminnow Mountain whitefish Burbot Walleye Lake whitefish 6 1 ES1-2 Total 3,579 1, ES1A Rainbow trout Sucker spp Walleye Sculpin spp. 2,1 49 Redside shiner 2 Mountain whitefish 98 2 Lake whitefish 71 Burbot 72 Northern pikeminnow 2 Kokanee 6 ES1A Total 2, ES2-1 Mountain whitefish Walleye Sucker spp Sculpin spp. 1,275 1, ,117 Rainbow trout Northern pikeminnow 1 Lake whitefish Kokanee Burbot 5 1 Peamouth 2 ES2-1 Total 1,324 2, ,216 1 ES2-2 Rainbow trout 197 Redside shiner 21 Sculpin spp. 5,24 Walleye 2 Mountain whitefish 127 Lake whitefish 6 Burbot 15 Sucker spp. 114 Northern pikeminnow 3 ES2-2 Total 5,743 Total 5 1, , ,37 2,73 1, ,81 2, , , , , , , , ,743

105 Table A4 Continued. Section Site Species A1 A2 A3 A4 A5 A6 A1+A2 A2+A3 D1 D2 D3 D1+D2 BW Eddy Unknown ES2A Sculpin spp Sucker spp Rainbow trout Northern pikeminnow 2 Mountain whitefish Lake whitefish Kokanee 1 Burbot Walleye ES2A Total ES2B White sturgeon 2 Burbot 2 4 Cutthroat trout 1 Lake whitefish 5 5 Mountain whitefish 27 2 Rainbow trout Sculpin spp Smallmouth bass 9 3 Sucker spp. 4 4 Walleye ES2B Total Lower Section Total 13, , , , All Sections Total 5,915 21, ,218 1, ,981 4,565 12, 7, ,615 2,479 1, Total , ,47 24,786 68,271

106 APPENDIX B MAPS AND UTM COORDINATES

107 d d Arrow Lakes Generating Station Ó S ES29 * Hugh L. Keenleyside Dam Columbia River * ES22 ) Pope and Talbot Sawmill Kootenay River ES21 Zellstoff Celgar N ) ES2 ) ES19 Ó ES28 ES18 * S5 ) Robson Ó Ó * ES17 N o r n s C r e e k ES25 Zuckerberg Island Ó Castlegar * Ó * C o l u m b i a R i v e r S1 ESK2 ESK1 Tin Cup Rapids K o o t e n a Ó y R i v e r * Brilliant Dam Ó ES1* S PROJECT TITLE LEGEND Boat Electroshocking Site River kilometre Paved Road 2 Lane River / Stream Lake Reservoir Dam Top Island Marsh or Swamp Sand or Gravel Bar Communities Bank Type Habitat A1 - Armoured Cobble/Gravel A1/A2 - Armoured Cobble/Gravel /Small Boulder A2 - Armoured Cobble/Small Boulder A2/A3 - Armoured Cobble/Small /Large Boulder A3 - Armoured Small/Large Boulder A4 - Armoured Large Boulder A5 - Bedrock Banks A6 - Man-made rip-rap BW - Backwater D1 - Depositional Sand/Silt D1/D2 - Depositional Sand/Silt/Gravel /Cobble D2 - Depositional Gravel/Cobble D3 - Depositional Large Cobble Eddy - Eddy N Scale 1 : 35, Metres FIGURE B1 Large River Fish Indexing Program Lower Columbia River Upper Section of Study Area : 27 Sample Site Locations PROJECT No DESIGN GIS CHECK REVIEW EL 15 Jan. 28 EL 15 Jan. 28 DF 15 Jan. 28 LH 15 Jan. 28 Inset Map Pen Castlegar, British Columbia 'Oreille River