Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Growth, fecundity, and recruitment responses of stunted brook trout populations to density reduction Hall, Donald Lincoln 1991

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

Item Metadata

Download

Media
831-UBC_1991_A1 H35.pdf [ 14.4MB ]
Metadata
JSON: 831-1.0100445.json
JSON-LD: 831-1.0100445-ld.json
RDF/XML (Pretty): 831-1.0100445-rdf.xml
RDF/JSON: 831-1.0100445-rdf.json
Turtle: 831-1.0100445-turtle.txt
N-Triples: 831-1.0100445-rdf-ntriples.txt
Original Record: 831-1.0100445-source.json
Full Text
831-1.0100445-fulltext.txt
Citation
831-1.0100445.ris

Full Text

Growth, Fecundity, And Recruitment Responses Of Stunted Brook Trout Populations To Density Reduction Donald Lincoln Hall B . S c , Humboldt State University, 1981 M . S c , University of British Columbia, 1986 A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of D O C T O R OF P H I L O S O P H Y in The Faculty of Graduate Studies Department of Zoology We accept this thesis as conforming to the required standard The University of British Columbia Apr i l 1991 © Donald Lincoln Hall , 1991 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Zoology  The University of British Columbia Vancouver, Canada Date April 4, 1991 DE-6 (2/88) Abstract Stunting is widespread among brook trout Salvelinus fontinalis populations in high alpine lakes in the eastern Sierra Nevada, California. Due to their small size and poor condition, stunted brook trout are undesirable as sport fish. In the same area, a few lakes contain large brook trout. Population density was the primary difference between lakes with different sized fish. I hypothesized that in lakes with large fish the food ration per individual was sufficient and that in lakes with stunted fish the food ration was the limiting factor. I carried out removal experiments on eight brook trout populations to test the hypothesis (1) that fish size is inversely related to population density, and by that evaluate density reduction as a means of improving growth in stunted brook trout. I considered seven additional hypotheses regarding the relationships between brook trout population density and growth, fecundity, and recruitment: (2) growth response is proportional to density reduction; (3) growth response is inversely proportional to pre-reduction density; (4) growth responses of juvenile and senescent fish are less affected by density reductions than mature, reproductively active fish; (5) growth response to density reduction is inversely proportional to lake elevation; (6) fish size is proportional to angling pressure; (7) fecundity response is proportional to the reduction in population density; and (8) recruitment response is inversely related to density. I used gillnets to simultaneously remove part of the population and estimate population size through catch depletion methods that allow variable catchability. Catchability varied with lake size and with abundance, increasing as population abundance declined. Increased catchability can be explained by behavioral responses. I measured and aged 16000+ brook trout from 71 lakes, 9800+ from the eight experimental lakes. I validated annual structures on otoliths using a fluorochrome mark. For the experimental lakes, I back-calculated previous population sizes using estimates of number at age in 1989, catch at age in 1987-1988, and survival rates at age estimated from catch data collected in 1987-1989. I converted population estimates into density estimates of fish and biomass per lake surface area and volume. ii I tested hypothesis 1 by using survey data from 61 populations and by experimentally manipulating density in eight populations. The survey data suggested that size differences be-tween populations of brook trout are a function of population density. Results from the eight removal experiments showed that fish size was inversely related to population density, though the increases in fish size were minor. The relationship between change in length and weight was roughly proportional to the change in density (hypothesis 2). Hypothesis 3 suggested differences in the severity of stunting in alpine lakes, and that the growth response of severely stunted populations would be more pronounced than the response of less stunted fish in lower density populations. The result was opposite; the growth response in lower density populations was greater than the response in higher density populations, suggesting that the growth re-sponse may have been proportional to the pre-reduction density. Hypothesis 4 suggested that the growth response for juvenile brook trout would be less than that for the pre-senescent adult population. The results refuted the juvenile portion of hypothesis 4: response for juveniles was greater than the response of the adults, perhaps because of greater recuperative abilities in young fish. The data supported the hypothesis that the growth response would be diminished in older fish. There was no relationship between elevation and growth response (hypothesis 5). Sport fishing had little effect on the growth of brook trout populations (hypothesis 6). Heavily fished populations were also stunted. Stunted brook trout had fecundities similar to non-stunted brook trout of the same size (hypothesis 7). Individual fecundity did increase in response to density reduction, but no more than would be expected from the increase in size. In several populations mean absolute fecundity decreased with age. Ovary weight was maintained by an apparent increase in mean egg size in older fish. The recruitment response varied between lakes (hypothesis 8). Recruitment did increase, likely in response to reduced cannibalism or competition, but I also found recruitment failure at the highest levels of density reduction. Strong cohorts were produced by increased juvenile survival rather than increased population fecundity, since population fecundity had decreased due to removal of most of the adult population. In one lake with almost no recruitment, densities remained low and fish weight doubled. For density reduction to be an effective means of increasing fish size, recruitment must be inhibited. i i i Table of Contents Abstract i i L i s t of Tables v i i i L i s t of Figures x Acknowledgements x i v 1 Introduct ion 1 Li tera ture Review 2 Genet ic Basis of S tunt ing 2 S tun t ing i n Salmonids 4 Longevi ty , Reproduc t ion , and Sexual M a t u r i t y 6 The A l l e v i a t i o n of S tunt ing S S tun t ing i n Sierra Nevada Brook Trou t 12 In i t i a l Hypotheses 15 Research His tory and A n Out l ine of Subsequent Chapters 18 2 Background Information and General Methods 21 Study A r e a 21 Sampl ing Overview 24 Removals Due to Sport F i sh ing 27 S tocking His tory 28 Gi l lne t Select ivi ty 28 Gi l lne t Saturat ion 38 Gi l l ne t t i ng Methods 39 F i e l d Sampl ing Methods 40 3 A g e Val idat ion and A g i n g Methods for Stunted B r o o k Trout 42 Methods 43 Age Va l ida t ion 43 O t o l i t h Prepara t ion and A g i n g Procedures 44 Sagi t ta l Section versus Cross Section 46 iv Results 46 Age Validation 46 Sagittal Section versus Cross Section 47 Discussion 50 Age Validation 50 Sagittal Section versus Cross Section 50 4 Estimating Population Size, Survival, and Density 53 Methods 54 Removal Experiments 54 Estimation of N and q 56 Estimates of Survival at Age 58 Estimation of Pre-Study Population Sizes 59 Estimation of Fish Density 60 Results 60 Estimates of N and q 60 Estimated Survival 61 Estimated Density and Abundance at Age 61 Comparisons of Abundance Estimates 61 Discussion 74 Estimates of N and q 74 Survival Estimation 75 Estimated Density 77 5 Spatial and Temporal Changes in Catchability 78 Methods . . 79 Differences in Catchability Among Lakes 79 Temporal Changes in Catch Per Effort 80 v Results 82 Differences in Catchability Among Lakes 82 Temporal Changes in Catch Per Effort 87 Change in Catchability Between Years 88 Change in Catchability Within Years 87 Discussion 97 Differences in Catchability Among Lakes 97 Temporal Changes in Catch Per Effort 99 Implications for Removal Estimates of Population Abundance . 99 Behavioral Hypotheses Regarding Changes in Catchability . . . 100 6 Fecundity Response to Density Reduction 103 Methods 103 Results 106 Interannual Changes Between and Within Lakes 106 Change in Fecundity with Size 115 Comparison of the Fecundity of Stunted and Non-Stunted Brook Trout 116 The Decline in Fecundity with Age 119 The Decline in Population Fecundity 120 The Relationship Between Fecundity, Abundance, and Density . . . . 123 7 Growth Responses to Density Reduction 126 Results 126 Changes in Length At Age and Weight At Age 126 The Relationship Between Fish Size and Density 138 Asymptotic Length and Weight Relationships 138 Changes In Fish Size Within Populations Due To Changes In Density 141 8 Recruitment Response to Density Reduction 143 Results and Discussion 143 The Effect of Recruitment on the Removal Experiments 143 The Response of Recruitment to Density Reduction 144 vi 9 Summary and Conclusions 151 Stunting in Brook Trout Populations 151 Models of Energy Shunting 153 Can Stunting Be Alleviated Through Density Reduction? 155 Bunny Lake Revisited 156 Suggestions for Further Study 157 Major Findings and Conclusions 159 Literature Cited 161 Appendix A 173 Appendix B 185 Appendix C 199 vii List of Tables 1.1 Selected examples of stunting in fish 3 1.2 Number offish sampled and mean age by weight group for 1986-1989 13 2.1 Lakes and number of brook trout sampled, 1985-1989 25 2.2 Selected measurements for ten of the original twelve experimental lakes 27 2.3 The random arrangement of 12 different mesh size panels in the ten gillnets used in this study 29 2.4 Modal age from the catch data for twelve lakes from 1987-1989 31 2.5 Mean length and mean weight of age 1 brook trout for twelve lakes from 1987-1989 31 2.6 Catchability coefficients (q) and relative vulnerability for ages 1-3 in seven of the experimental lakes, 1987-1989 32 2.7 Catchability coefficients (q) and relative vulnerability for four size classes (fork lengths in mm) in seven of the experimental lakes, 1987-1989. . . . 33 3.1 The number of stunted brook injected with oxytetracycline and released, the dates of release and recovery, the number of winters at large during which time annuli are formed, and the number fish recovered with O T C marks and fin-clips 47 3.2 The number of brook trout with valid O T C marks for recovery ages 2-10. . . . 49 4.1 Catch by netting period and total catch for the eight experimental lakes in 1989 56 4.2 Catch at age for the eight experimental lakes, 1986 or 1987 through 1989. . . . 62 4.3 Estimated initial population size (N), estimated capture probabilities (if), and minimum function values (F1) for three removal models and eight experimental lakes 63 4.4 Approximate 95% confidence intervals for the estimated initial population size (jV) for three removal models and eight experimental lakes 63 4.5 Brook trout survival at age for the eight experimental populations and 12 additional lake populations sampled between 1986-1989 67 4.6 Estimated number of fish at age for the eight experimental lakes, 1986 or 1987 through 1989 69 vii i 4.7 Estimated biomass offish at age for the eight experimental lakes, 1986 or 1987 through 1989 70 4.8 Brook trout density in the eight experimental lakes calculated as number per surface hectare, per surface hectare above the 3 m depth contour (Ab3), and per 10000 m 3 71 4.9 Brook trout density in the eight experimental lakes calculated as kilograms per surface hectare, per surface hectare above the 3 m depth contour (Ab3), and per 10000 m 3 72 4.10 Comparison of abundance estimates for age 1, age 2, and total popula-tion by three methods; Leslie, Schnute, and the back-calculation of numbers at age using equation 4.4 (VPA) 73 6.1 Sample sizes for fecundity samples, 1986-1989 104 6.2 Sampling dates for fecundity samples, 1986-1989 105 6.3 Estimated population fecundity at the beginning and end of each removal experiment for the eight experimental populations 122 6.4 Sample size, mean fecundity and standard error, and estimates of population abundance and density for the eight experimental lakes in 1988 and 1989 124 7.1 Sample sizes for mean length at age and mean weight at age data for the eight experimental lakes as presented in Figures 7.1-7.8 137 7.2 Parameters of von Bertalanffy growth equations for length and weight (Loo, Woo, K, io, t'o), estimated population size, estimated fish per 100 m 2 ( D A ) , estimated fish per 100 m 3 (Dv), estimated fish per 100 m 2 above 3 m (DAb3), and estimate fish per meter of shoreline (Dp) for the eight experimental lakes 139 8.1 Density in fish per hectare by age group and for all ages for the eight experimental lakes 147 ix List of Figures 1.1 Weight versus length for 16729 brook trout sampled from 71 lakes in 1986-1989 14 1.2 Mean weight versus catch per net hour, an index of population density, for 61 brook trout populations sampled in 1986-1988 14 2.1 Location of all lakes from which brook trout were sampled, with the eight experimental lakes emphasized and labelled 22 2.2 Catch frequency by mesh size for 1352 brook trout sampled between 1986-1989 34 2.3 Relative catch frequency versus length in mm by mesh size for 1352 brook trout sampled between 1986-1989 35 2.4 Relative catch frequency versus weight in g by mesh size for 1351 brook trout sampled between 1986-1989 36 2.5 Relative catch frequency versus age by mesh size for 1323 brook trout sampled between 1986-1989 37 2.6 Catch versus set duration in hours for 249 gillnet sets made between 1986-1989 39 3.1 Photographs of the anterior tip of an age 6 brook trout otolith showing a) the oxytetracycline (OTC) mark with reflected ultraviolet light and b) two annuli after the OTC mark with transmitted light. • 48 3.2 Age determinations made from one sagitta ground to the transverse mid-plane (cross section) versus age determinations made from the other sagitta ground to the sagittal midplane (sagittal section) based on 90 paired sagitta from 45 stunted brook trout 49 4.1 Relative size, depth and shape for eight experimental lakes in the Sierra Nevada, California 55 4.2 Catch by netting period for eight experimental brook trout populations 64 4.3 Separation statistic F(N) for model 1 {F\), model 2 (F 2 ) , and model 3 (F3). . . 65 4.4 The logarithm of catch frequency at age plotted against age for 20 lake populations of brook trout 66 4.5 Mean survival rate at age plotted against age for 20 lake populations of brook trout 68 x 5.1 Estimated capture probability (q) versus estimated population size (N) 83 5.2 Estimated capture probability (q) versus four measures of area occupied and four indices of fish density 85 5.3 Catchability per net (?///) versus four estimates of lake area occupied by brook trout in the eight experimental lakes 86 5.4 Catch per net hour (CPE) versus cumulative catch for four experimental lakes 91-92 5.5 Catch frequency at age for four of the experimental lakes 93-94 5.6 Catch per net hour (CPE) versus cumulative catch for four experimental lakes 95-96 6.1 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean length (mm), and mean weight (g) versus age for an experimental lake (Flower) and a nearby non-experimental lake (Matlock), in 1988 and 1989 j 107 6.2 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean length (mm), and mean weight (g) versus age for an experimental lake (Wonder 3) and an adjacent non-experimental lake (Wonder 2), in 1988 and 1989 108 6.3 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean length (mm), and mean weight (g) versus age for an experimental lake (Fishgut 1) and a nearby non-experimental lake (Fishgut 3), in 1988 and 1989. 109 6.4 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean length (mm), and mean weight (g) versus age for an experimental lake (Dingleberry) and a nearby non-experimental lake (Midnight), in 1988 and 1989 110 6.5 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean length (mm), and mean weight (g) versus age for an experimental lake (Hell Diver 3) and a nearby non-experimental lake (Hell Diver 1), in 1986, 1987, 1988, and 1989 I l l 6.6 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean length (mm), and mean weight (g) versus age for an experimental lake (Hell Diver 2) and a nearby non-experimental lake (Hell Diver 1), in 1986, 1988, and 1989 112 6.7 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean length (mm), and mean weight (g) versus age for an experimental lake (Par Value) in 1986, 1987, 1988, and 1989 113 xi 6.8 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean length (mm), and mean weight (g) versus age for an experimental lake (Gem 2) and an adjacent non-experimental lake (Gem 3), in 1986, 1987, 1988, and 1989 114 6.9 Mean fecundity versus mean fork length (mm) at age for 24 lakes from 1986-1989 (see Table 6.1) 117 6.10 Mean fecundity versus mean fork length (mm) at age for 24 lakes from 1986-1989 (see Table 6.1), and fecundity—length relationships from other studies of non-stunted brook trout populations 118 6.11 Mean fecundity versus estimated population size (panel a), estimated pop-ulation density per 100 m 2 surface area (panel b), estimated density per 100 m 2 above the 3 m depth contour (panel c), estimated den-sity per 100 m 3 volume (panel d), and estimated density per meter of shoreline (panel e) 125 7.1 Mean length at age and mean weight at age for Flower (experimental) and adjacent, non-experimental Matlock 128 7.2 Mean length at age and mean weight at age for Wonder 3 (experimental) and adjacent, non-experimental Wonder 2 129 7.3 Mean length at age and mean weight at age for Fishgut 1 (experimental) and nearby, non-experimental Fishgut 3 130 7.4 Mean length at age and mean weight at age for Dingleberry (experimental) and nearby, non-experimental Midnight 131 7.5 Mean length at age and mean weight at age for Hell Diver 3 (experimental) and nearby, non-experimental Hell Diver 1 132 7.6 Mean length at age and mean weight at age for Hell Diver 2 (experimental) and adjacent, non-experimental Hell Diver 1 133 7.7 Mean length at age and mean weight at age for Par Value (experimental). . . . 134 7.8 Mean length at age and mean weight at age for Gem 3 (experimental) and adjacent, non-experimental Gem 2 135 7.9 Mean length at age and mean weight at age for Fishgut pond (experimental) and adjacent, non-experimental Fishgut 3 136 7.10 Estimates of mean asymptotic size versus estimates of population abun-dance and population density for the eight experimental lakes in 1987-1989 (and 1986 for Hell Diver 2) 140 xii 7.11 Percent change in a) and b) Woo plotted against the percent change in density due to removals 142 8.1 Density of age 1 recruits in year t + 1 versus percent change in density in year t 148 8.2 Density of age 1 recruits in year t + 1 versus density of age 3+ fish in year t. . . 149 8.3 Estimated number of age 1 recruits in 1989 versus estimated population fecundity in 1987; a stock—recruitment relationship for stunted brook trout populations after the first season of removal experiments 150 xiii Acknowledgements Carl Walters got this study going and kept me on-track along the way. I especially appreciated his help in the Sierra Nevada; had he not seen some of data as it was collected, he probably would not have believed it. Phil Pister, Darrell Wong, Curtis Milliron, and Bob Brown, fisheries biologists with the California Department of Fish and Game in Bishop, provided continual assistance during the five summers that I worked out of their office and additional help in the field that went beyond their regular duties. Tom Jenkins provided a wealth of information on brook trout behavior. My supervisory committee, Drs. J.D. McPhail, T.G. Northcote, A .F . Tautz, and N.J. Wilimovsky, provided initial direction, generous loans of equipment, and helpful criticisms of the dissertation. Dr. W. Briggs of the Carnegie Institute of Washington (Stanford University) permitted my sampling in the Harvey Monroe Hall Natural Area. In the lab, Joel Sawada helped grind thousands of otoliths. Forest Stokstad counted over a million eggs. Dave DeRosa and Jeff Burrows also helped with the egg counting. Many people answered my aging questions: among those I remember are Shane MacLellan, Bruce Leaman, Doris Chilton, Karen Charles, Steve Campana, Glen Geen, Rod Cook, Jay Borseth, Barbara Rokeby, William Ovalle, and Bernie Cox. Not surprisingly, more people were willing to help catch fish in the Sierra Nevada than count eggs or grind otoliths in the lab. Arlene Tompkins, Gordon Haas, and Joel Sawada were outstanding field assistants. Others who helped out in the backcountry included Molly Nevin; Dave Derosa; Ulrike, Stephan, Ann, and Ray Hilborn; Daniel, William, and Sandy Walters; Chuck and Kyle Knutson; and Bill Fong. Scores of seasonal aides with the California Department of Fish and Game helped out over the five field seasons. Special thanks to Jan Goldberg, Patty Young, and Curtis Milliron, whose standing offers of accommodations and showers were appreciated more than they can imagine. Walt Schober provided free pack service into the middle fork of Bishop Creek lake basin, and Bart Cranney packed us and our gear in and out of the Sisters Lakes basin. Away from the mountains, Tammy and Scott Watson and Donna and Don Raub provided much needed homes away from home. xiv Personal support was provided by a University Graduate Fellowship, a Hugo E. Meilicke Graduate Fellowship, and an NSERC grant to Carl Walters. Research funding was provided by the Inyo-Mono County Board of Supervisors, upon recommendation by the Inyo-Mono County Fish and Game Commission, by contract to D.L. Hall from Dingell-Johnson-Wallop-Breaux funds administered by California Fish and Game, and by an NSERC grant to Carl Walters. It would be a human oversight not to acknowledge my faithful companion Cody. She packed in food and packed out nets and samples, warned us of bears, and reminded us that she was having a good time even when we weren't. My special thanks to Arlene Tompkins. We "forced marched" to over 60 lakes in 1986, sampled over 4000 fish in 1987, and she was still willing to marry me and put up with the last three years while finishing her own thesis. Its almost over. xv 1. Introduction The growth of fish may be limited by both density dependent and density independent factors. Density dependent factors include inadequate food (Comfort 1960) and limited space (lies). Density independent factors include interrelated factors such as water temperature (Dwyer and Smith 1983) and growing season (Van Oosten 1944), and other physical or chemical factors (Power 1980; Weatherly and Gill 1987). When limited resources or environmental conditions severely retard the growth of fish, the condition is called stunting. Stunting is a subjective measure of severely reduced growth when compared to the same species growing under more favorable conditions. I distinguish stunting from dwarfism, the latter indicating a genetic effect (Bailey and Lagler 1937). In the Sierra Nevada mountains of California there exist several hundred alpine lake populations of stunted brook trout, Salvelinus fontinalis (Mitchill). Brook trout growth in these lakes is severely limited by inadequate food, and perhaps by limited space, cold water temperatures, and a short ice-free growing season (Reimers 1958, 1979). In the same region there are a few lakes which contain comparatively large brook trout. One difference between lakes with stunted fish and lakes with large fish is the population density. Differences in water temperature, growing season, and physical and chemical lake characteristics are negligible (Reimers et al. 1955; Stoddard 1986, 1987a, 1987b; Landers et al. 1987). I hypothesized that in lakes with large fish the food ration per individual was sufficient and that in lakes with stunted fish the food ration was the limiting factor. If this hypothesis is valid, the growth of stunted brook trout could be improved by supplemental feeding or by reducing the population density. Supplemental feeding studies have shown successful growth recovery of stunted fish (Aim 1946; Tiemier and Elder 1960; Heath and Roff 1987). For alpine lakes of the Sierra Nevada, I considered supplemental feeding too expensive and logistically difficult. Instead, I examined the reduction of population density as a means of alleviating the stunted condition of brook trout. Previous density reduction studies have dealt with stunted fish in multi-species, eutrophic lakes (Dahl 1-917; Beckman 1941, 1943; Parker 1958; Warnick 1966), two-species salmonid communities (Fagerstrom 1972), and lately 1 single species salmonid populations (Pechlaner and Zaderer 1985; Langeland 1986; Donald and Alger 1989). Previous density reduction studies have experimented with only one population and usu-ally the only comparison has been with pre-reduction conditions. Because of these limitations, results from these studies have been inconclusive. In this study I conducted density reduction experiments on eight lakes, most with adjacent, non-experimental populations for comparisons of growth, fecundity, and recruitment over the four year study period. While I predicted that the immediate response to density reduction would be increased size, I hypothesized three alternative possibilities for longer term response of reproductive and recruitment rates: 1) decreased recruitment due to decreased egg production; 2) no recruitment change, due to limiting factors in spawning habitat and juvenile rearing areas; or 3) increased recruitment due to reduction in competition or cannibalism effects of older fish. Literature Review Stunting has been reported in many fish species (Table 1.1). The research has concen-trated on species that are commercially important or favored for sport. Stunting undoubtedly occurs in less studied fish. Genetic Basis of Stunting Bailey and Lagler (1937) distinguished stunting from dwarfism, the latter indicating a genetic effect. Though this distinction has been lost in the recent literature, there is little doubt that stunting is usually environmentally controlled. Stunted fish can generally resume normal growth when food is increased (Comfort 1960; Weatherly and Gill 1981), when transplanted to more favorable environments (Aim 1946, Tiemier and Elder 1960; Rabe 1967a), and when population densities are reduced (Dahl 1917; Beckman 1941, 1943; Warnick 1966; Pechlaner and Zaderer 1985; Langeland 1986; Donald and Alger 1989). Aim (1946) and Heath and Roff (1987) raised offspring from the gametes of stunted fish to normal size, and concluded that in perch and sunfish stunting was not an inherited trait. Studies that have attributed a 2 Table 1.1 Selected examples of stunting in fish. Alosa pseudoharengus Walton (1983) Ambloplites rupestris Beckman (1941, 1943) various centrarchids Hubbs and Cooper (1935); Bailey and Lagler (1937) Coregonus sardinella Mann and McCart (1981) Esox lucius Diana (1987) Ictalurus punctatus Tiemier and Elder (1960) Lebistes reticulatus Comfort (1960) Lepomis macrochirus Murnyak, Murnyak and Wolgast (1984) Micropterus dolomieui Emery (1975) Micropterus salmoides Parker (1958) Oncorhynchus clarki Robertson (1947) Oncorhynchus mykiss Rabe (1967b); Weatherly and Gill (1981) Oncorhynchus nerka Ricker (1938); Foerster (1968); McCart (1970) Perca flavescens Eschmeyer (1937, 1938); Heath and Roff (1987) Perca fluviatilis Aim (1946, 1959); Deelder (1951) Rutilis rutilis Linfield (1974); Burrough and Kennedy (1979) Salmo salar Dahl (1927) Salmo trutta Dahl (1917); Aim (1959); Jonsson (1977); Pechlaner and Zaderer (1985) Salvelinus alpinus Pechlaner (1984); Langeland (1986) Salvelinus fontinalis Reimers (1958, 1979); Rabe (1967a) Donald and Alger (1989) Salvelinus namaycush Donald and Alger (1986) Tilapia sp. lies (1973) genetic component to stunting have been speculative (Walton 1983) or have drawn conclusions from inconclusive data (Murnyak, Murnyak and Wolgast 1984). Special consideration is warranted for the literature that addresses the sympatric oc-currence of two or more forms of a species (Ricker 1938; Aim 1959; Mann and McCart 1981; Jonsson and Hindar 1982; Nordeng 1983, and references therein). The major difference be-tween forms is body size, and the smaller form can be considered stunted. For migratory and non-migratory sockeye salmon (Oncorhynchus nerka: Ricker 1938; McCart 1970), the size difference can be explained by feeding opportunities in the ocean versus lakes. The situation of lifelong sympatry is more complex. Mann and McCart (1981) explored this situation for least cisco Coregonus sardinella in an Arctic lake. Though the existence of two forms was clearly shown, the mechanism for maintaining the forms was unresolved. Mann and McCart 3 (1981) speculated that the forms either developed allopatrically and dispersed to the same lake, or the forms developed in sympatry because of interspecific competition for a limited food resource. They doubted that the forms were genetically fixed, but "may instead be an expression of differing environmental conditions during early developmental stages" (Mann and McCart 1981). Explanations for and the systematics of coexisting forms of Arctic char Salvelinus alpinus have been debated for decades. Jonsson and Hindar (1982) described three forms of char (anadromous, large and small freshwater resident) that spawn assortatively but are unlikely to be reproductively isolated. Nordeng (1983) showed that the parr of each form segregate into all three forms, and a single individual may manifest all three forms during a lifetime. Nordeng (1983) interpreted these results to mean that the forms came from the same gene pool, but electrophoretic results from Hindar et al. (1986) suggest that the forms may be genetically different. It is sufficient for my purposes to consider sympatric forms a special case of stunting that is more complex than the simple food limiting situation that I hypothesized for alpine brook trout. Size differences resulting from anadromy are known for brook trout (Scott and Crossman 1973), but the coexistence of different resident size forms has not been reported, and I did not observe this situation in any of the lake populations that I sampled. Stunting in Salmonids Stunting has been described in other Salvelinus species. Donald and Alger (1986) studied a stunted lake trout Salvelinus namaycush population in the Canadian Rocky mountains. They found lake trout with mean weights of 125 grams at age 10, and 281 grams at age 20. Typical mean weights for slow growing, northern lake trout populations are >500 grams at age 10 and >2000 grams at age 20. Absence of suitable food was the reason given for the extremely slow growth (Donald and Alger 1986). Donald and Alger (1989) manipulated density in a population of stunted brook trout in eastern British Columbia. I review their study in the section on the alleviation of stunting below. Reimers (1958, 1979) described a long term study of a brook trout population in Bunny Lake, California. Bunny Lake is in the same region as my study lakes. A brief review of 4 Reimers' study will aid in understanding the dynamics of brook trout populations in the Sierra Nevada. In 1951 approximately 1790 yearling brook trout were stocked in previously Ashless Bunny Lake (elevation 3322 m, area 1 hectare). In the first year, zooplankton and benthic in-vertebrate populations declined drastically, while trout growth was only slightly below normal. In subsequent years, "with the extreme reduction of food, fish growth approached a standstill" (Reimers 1979). The population was monitored sporadically for the next 23 years. A single spawning event (four fish) was observed in 1966, "years after all thoughts of natural reproduc-tion had been forgotten" (Reimers 1979). In 1974 the last fish from the original planting was removed at age 24. The Bunny Lake study was important for several reasons. 1) The maximum reported age for brook trout was tripled. 2) Brook trout were shown capable of delaying reproduction until conditions were suitable. In this case, a decrease in population caused by natural deaths and sampling removals seems to have increased trout nutrition so that energy was available to develop viable eggs and sperm. 3) Histological examinations showed that the physiological age of the fish was somewhat less than the chronological age. The increased longevity was attributed to "unremitting shortage of dietary calories in a temperature regime that was more often conducive to torpor than to activity" (Reimers 1979). Stunting has been described in other salmonids, but it does not seem as prevalent in On-corhynchus as in Salvelinus. Robertson (1947) reported on two connected lakes in the Wind River Range, Wyoming, that contained small cutthroat trout Oncorhynchus clarki at high density in the lower lake and larger fish at a lower density in the upper lake. He concluded that the size difference was due to increased food per individual in the lower density lake. The different densities were thought to be caused by differences in available spawning habitat; the high density population had access to good spawning habitat in the lower course of the interconnecting stream. Robertson's conclusions highlight a life history difference between On-corhynchus and Salvelinus. In general, Oncorhynchus prefer stream spawning and reproduce poorly when restricted to spawning in lakes. Salvelinus tolerate a wider range of spawning 5 habitats and will reproduce successfully in lakes. An exception to this is Salvelinus malma, which seems to prefer stream spawning. Like the stream spawning Oncorhynchus, stunting is rarely reported in Dolly Varden. Pechlaner and Zaderer (1985) reported a stunted popula-tion of brown trout Salmo trutta. I review their study in the section below on alleviation of stunting. In other studies that do report slow growing Oncorhynchus populations, it seems that populations and high densities are maintained by stocking (Rabe 1967b; Nelson 1987). Rabe (1967b) compared the growth of rainbow trout Oncorhynchus mykiss in four alpine lakes in Utah stocked at the same density but planted with existing populations of brook trout and cutthroat trout that varied in density. Rabe (1967b) found that rainbow trout grew well in low density populations, but grew slowly when planted with stunted (high density) brook trout populations. The observations extended only over the life span of the stocked rainbow trout, and subsequent reproductive success was not monitored. Nelson (1987) describes Colorado's extensive high alpine lake stocking program. Though there is limited reproduction by rainbow trout in some Colorado alpine lakes, most of the populations are maintained by the stocking program. The results of both Rabe (1967b) and Nelson (1987) support my observations on rain-bow trout, golden trout Oncorhynchus aguabonita, and brown trout populations in the Sierra Nevada. Unless maintained by repeated stocking or there is an abundance of suitable spawn-ing habitat, these species eventually die out. In contrast, brook trout populations reproduce successfully in almost all mountain lakes, and can quickly "overpopulate" a lake. The influ-ence of periodic stocking shows up in the varied size distributions of Oncorhynchus or Salmo populations, when compared to the uniform size distributions I observed repeatedly in stunted brook trout lakes. Longevity, Reproduction, and Sexual Maturity Woodhead (1978) and Craig (1985) reviewed the contributions of gerontology to the study of aging in fish. Most gerontologic studies find useful aging information in the extreme conditions brought about by experimentally induced stunting. Comfort (1960, 1961, 1963) 6 published a series of aging experiments using laboratory raised guppies Lebistes reticulatus. Comfort (1960) found that underfed guppies would resume normal growth rates when the food ration was increased. Mortality increased with age, suggesting that aging takes place in the absence of the ability to grow (Comfort 1961; also Gerking 1957). Stunted guppies showed increased longevity when compared to fish fed full rations (Comfort 1963). Other studies have shown increased longevity of stunted fish, usually in cold water (Craig 1985; Reimers 1979). The relationship between stunting and reproductive success is less clear than the rela-tionship between stunting and longevity. Roff (1986) defined reproductive success in a stable population as the ability of an individual female to produce one female offspring that her-self survives to reproduce. Three components of reproductive success are longevity, fecundity, and age at first reproduction. Within a species, fecundity is usually proportional to body size (Bagenal 1978), but stunted fish may make up for decreased fecundity with increased longevity, or by increasing the frequency of spawning events (lies 1973). Usually fast growing fish mature earlier than slow growing fish (Aim 1959), but individuals in stunted populations may show a younger age at maturity than in populations where growth is faster. Aim (1959) explained this apparent inconsistency by speculating that age at maturity may be set by juvenile growth rates. This explanation appears consistent with many studies that show that the onset of stunting may not occur until the second or third year of life (Beckman 1943; Aim 1946; Deelder 1951; Heath and Roff 1987), probably because of feeding differences between juveniles and adults (Hazzard 1933). Thus the age at sexual maturity in stunted fish may be set before stunting occurs. Food limitation and severe climate in high alpine lakes may represent the extreme case of limited reproductive success in salmonids. It took 16 years for the Bunny Lake brook trout to successfully reproduce. Reimers (1979) described deformed testes and ovaries, probably resulting from severe malnutrition, that further limited the chance of reproductive success for the Bunny Lake trout population. Though the Bunny Lake situation was extreme, I sampled other lakes with weak or missing year classes showing that reproductive success does vary from year to year. 7 The Alleviation of Stunting The stunting literature concentrates on commercial and sport fish because stunting is a management concern. Pond culturalists and sport fishermen prefer large fish. There are not as many studies on the theoretical aspects of stunted fish growth as there are studies on methods of improving growth. I have already referred to studies that investigated supplemental feeding and dismissed the technique as a method for use in the Sierra Nevada. Here I discuss studies that examine population reduction as a means of alleviating stunting. Dahl (1917) removed brown trout by seine from overpopulated lakes to improve fish growth in the remaining populations. The trout responded with greatly improved growth rates. An unfortunate side effect of the annual netting was the complete removal of old and large fish. To the fishermen, the situation had not changed; small, slow growing old fish had been replaced with small, fast growing young fish (Lindstrom et al. 1970). The study by Dahl (1917) was pioneering in demonstrating the relative ease with which the growth of fish in natural waters could be altered. Beckman (1941, 1943) reduced a population of stunted rock bass Ambloplites rupestris along with populations of perch, various forage fish, and suckers, by poisoning half a lake with rotenone and then letting the untreated populations disperse. Only 42 of over 6000 perch and rockbass collected after poisoning were of legal size. Beckman (1941, 1943) reported significant growth increases in the rock bass over the next several years, "too great to be accounted for by any normal growth fluctuation". Even without experimental controls the evidence that Beckman presented was convincing for rock bass, but no mention was made of the effects of population reduction on the other species. His experiment was confounded further by the introduction of several other gamefish both as adults and fingerlings after the 1937 poisoning. Particularly interesting was the addition of 5000 fmgerling bluegills Lepomis macrochirus in 1938 that did not show up in subsequent sampling. I suspect that the additions of juvenile gamefish acted as supplemental prey to further enhance the growth response caused by the population reduction. 8 Mixed results were also obtained by Parker (1958). Parker attempted to improve the growth of largemouth bass Micropterus salmoides by fyke netting substantial proportions of the bass and sunfish populations. Though he did not find a significant increase in growth of bass, some of the sunfish responded to the decreased densities, either by increasing growth or increasing population numbers to levels much higher than before the removals (Parker 1958). Warnick (1966) reduced two lake populations of perch Perca flavescens using a diluted piscicide. Surviving perch showed increased growth after one full growing season. Like Beck-man's experiment with rock bass (Beckman 1941, 1943) Warnick did not use experimental controls, and he confounded results from the depletion by introducing a prey species to sup-plement the food of the surviving perch. Fagerstrom (1972) attempted to increase the growth of brown trout and improve angling quality by removing a large proportion of a sympatric Arctic char population. As predicted, initially the average size of the fish declined because of removals of the larger fish by gillnet. Contrary to expectations, the average size of the brown trout and char did not increase signif-icantly over the course of the 13 year experiment. Growth rates did increase, but the increase did not improve the average weight of fish captured by sport fishermen. Instead, the sport catch consisted of young, fast growing fish that were about the same average weight as fish before the experiment began. Fagerstrom speculated that had the lake been unexploited for a few years after the population removal, the remaining fish would have had an opportunity to grow to the desired larger size. In a study to determine the productivity of an alpine lake in Colorado, Walters and Vincent (1973) removed brook trout, monitored the invertebrate population, then restocked the lake with cutthroat trout. One brook trout managed to survive the poisoning, and in the absence of competition grew to the probable maximum size for brook trout in Emmaline Lake (Walters 1969). Nyman (1984) and Pechlaner (1984) recommended the reduction of population density as a means of improving the growth of stunted fish but the technique has not been widely employed or reported. Since my study began, three reports of density reduction as a means 9 to alleviate stunting in salmonids have been published (brown trout, Pechlaner and Zaderer 1985; Arctic char, Langeland 1986; brook trout, Donald and Alger 1989). Because of the direct relevance of these studies, here I discuss each in detail. Pechlaner and Zaderer (1985) used gillnets to reduce the density of brown trout in Gossenkollesee, a 1.7 hectare lake in the Eastern Alps of Austria. They estimated that the number of fish above 144 mm in length was reduced from 250 to 92 over 4 years. Based on the surface area of Gossenkollesee, density was reduced from 147 to 54 fish per hectare (—63%), for fish above 144 mm. They evaluated growth changes by comparing the size distribution of trout in 1979-1980 with the size distribution in 1983, and by comparing condition factors with and without the weight of the gonads for the same years. They stated that "In 1979/80 no trout was bigger than 221 mm, but one third of the 1983 catch belonged to the size class above 224 mm. The biggest trout was 253 mm long." They found that the condition factor increased from 0.88 in 1979 and 0.92 in 1980 to 1.04 in 1983, and that the gonads made up 2.82% of body weight in July 1983 versus 0.73% in 1979. From these results they concluded that " . . . trout in Gossenkollesee respond to the decrease of adult population density not only with enhanced growth and a proportionally higher fertility, but with a fertility increase compensating for the artificial regulation of population numbers." As I discuss in chapter 6, their conclusions regarding fertility are suspect and contrary to more detailed results from my study. Langeland (1986) measured the effects of heavy exploitation of Arctic char on age com-position, length-frequency distributions, growth, and recruitment in Lake 0vre Stavatj0nn, Norway, a 4 hectare lake at 824 m above sea level. Langeland estimated that the density of age 2+ fish was reduced from 1100 per hectare in 1979 to 97 per hectare in 1983 (—91%). Mean weight increased from 67 g in 1979 to 91 g in 1983. In 1984 the mean weight dropped to 74 g with the recruitment of a large age 2 cohort. The weight of the largest fish caught increased from 160 g in 1979 to 645 g in 1984, a remarkable quadrupling in maximum weight. The fraction of larger fish (over 125 g) in the total yield increased from 1-3% in 1979-1980 to 73% in 1984, another indication of substantial growth in weight. The length frequency 10 distributions shifted oddly toward larger fish after two years, then shifted back toward a pre-dictable distribution of many small fish after 3-4 years. The choice of small mesh sizes may have been the cause of the initial shift toward larger fish in the catch. Recruitment appeared to be reduced because of the intense fishing, but a high proportion of age 1 fish in the 1984 catch may have signalled a strong 1983 year class recruiting at the end of the experiment. Donald and Alger (1989) evaluated exploitation as a means of improving growth in a stunted population of brook trout in Olive Lake, a 2.0 hectare subalpine lake at 1631 m above sea level in eastern British Columbia. They removed approximately 20% of the estimated population size annually. The decline in density was modest, from 797 fish per hectare in 1982 to 711 in 1985. The decrease due to exploitation was mostly offset by increasing recruitment. Maximum weight of the brook trout increased from 61 g to 158 g, though mean weight only increased from 38 g to 42 g. The age distribution shifted toward younger age classes over the four years of sampling. By including results from Pechlaner and Zaderer (1985) and Langeland (1986), Donald and Alger (1989) concluded that density reductions of 12, 63, and 91% caused similar growth responses among lakes. In contrast, results from my study from lakes within the same area and between populations of the same species show that population responses do vary in proportion to density reduction. There are several studies that describe the improved growth response of stunted fish transplanted into environments with more favorable growth conditions (Aim 1946; Greene 1955; Tiemier and Elder 1960; Saunders and Smith 1962; Rabe 1967a). Rabe (1967a) dealt specifically with the transplant of stunted alpine brook trout into barren lakes. I conducted a transfer experiment in 1986-1987 and obtained results nearly identical with Rabe (1967a). Transplanting small numbers of stunted fish into barren lakes demonstrates the maximum growth potential for stunted trout, but it does not address the management problem of stunted brook trout populations in California for two reasons. 1) There are few barren lakes in the Sierra Nevada that are suitable for trout introduction. 2) It is not a goal of the California Department of Fish and Game to stock barren waters with brook trout and continue to spread 11 the problem. Barren lakes are either left barren or stocked with golden trout, a fish native to the southern Sierra Nevada. The problem is how to deal effectively with the hundreds of stunted brook trout popula-tions that are already present. Unlike populations of rainbow, golden, and brown trout, brook trout populations are successful at reproducing in the harsh conditions of high alpine lakes. Many populations that I have sampled have persisted after only a single stocking forty to fifty years ago. There is no indication that the brook trout will die out if left unattended. I designed this study to approach the problem of stunted brook trout populations in a simple and practical manner. At the same time, extensive sampling of lake populations provided rare information on the population dynamics of exploited and unexploited alpine trout populations. Stunting in Sierra Nevada Brook Trout As a preliminary investigation to the experimental studies that were the main source of my results, I conducted a broad survey of growth and relative abundance (indexed by gillnet catch rates) in southern Sierra Nevada brook trout populations by visiting 71 lakes. Sampling methods for this survey used techniques described in chapters 2 and 3. Here I present a summary of the results as illustration of the stunting problem. Stunting is severe and widespread among populations of brook trout in the eastern Sierra Nevada. The extent and severity of stunting is summarized in Figure 1.1 and Table 1.2. I added reference lines in units of interest to sport anglers in Figure 1.1 to show that only 70 of 16729 (0.4%) fish that I sampled from 71 lakes were above 10 inches and one-half pound. The mean age data in Table 1.2 show that fish in the small weight classes were not just young fish. The data in Figure 1.1 may not provide a clear indication on the widespread nature of stunting, since in 1987-1989 I concentrated sampling effort on stunted populations. The survey data for 51 lakes in 1986 show the same pattern of few large brook trout (Table 1.2). 12 Table 1.2 Number offish sampled and mean age by weight group for 1986-1989. weight range (g) 0-100 101-200 201-300 301-400 401+ total 1986 (n) 842 383 33 4 4 1266 mean age 5.0 6.8 7.6 8.5 7.0 (51 lakes) 1987 (n) 4059 690 15 6 1 4771 mean age 4.1 7.3 7.1 6.7 11.0 (19 lakes) 1988 (n) 4774 722 12 4 1 5513 mean age 3.4 6.9 8.4 9.5 13.0 (20 lakes) 1989 (n) 4399 421 16 3 1 4840 mean age 2.3 6.0 6.2 7.0 13.0 (20 lakes) total 14074 2216 76 17 7 16390 mean age 3.4 6.8 7.3 7.8 9.3 The 100 fish over 200 g sampled in 1986-1989 (Table 1.2) are from 22 of the 71 lakes sampled. Oyer half of these 100 fish are from three lakes, suggesting that large fish are concen-trated in a few lakes. Assuming that catch per net hour is a rough index of population density, Figure 1.2 shows that large fish are found only in a few lakes with low population density. The relationship between mean weight and catch per net hour was almost constant at catch rates above three fish per net hour. Mean weight increased substantially for populations with catch rates below two fish per net hour. The data in Figure 1.2 show that many populations with low catch rates also have low mean weight. I do not know whether these points accurately represent low lake productivity, or are misleading due to variation in the relationship between catch per net hour and true density. The pattern of a measurement being almost constant over a wide range of densities and increasing sharply at very low densities is a result common to most of the responses measured in this study. The relationship between density and size as measured in many populations (Figure 1.2) seems to be maintained within populations; that is, as density decreased, size increased. The data in Figure 1.2 accurately suggest that it requires a massive reduction in population density to produce a measurable response. 13 8 0 0 6 0 0 Figure 1.1 Weight versus length for 16729 brook trout sampled from 71 lakes in 1986-1989. The dashed lines are at 254 mm (10 inches) on the abscissa and 227 g (one-half pound) on the ordinate, showing the few fish caught (70, 0.4%) that are above these sizes. CD O) CD ? c CO CD E 6 0 0 5 0 0 4 0 0 3 0 0 2 0 0 -100 -10 2 0 catch per net hour 3 0 Figure 1.2 Mean weight versus catch per net hour, an index of population density, for 61 brook trout populations sampled in 1986-1988. 14 Initial Hypotheses I initially proposed to investigate six hypotheses regarding stunted brook trout growth and population density. I added two additional hypotheses during the study regarding fecun-dity and recruitment responses to density reduction. Here I review these hypotheses with a brief statement of explanation and finding. HI. The size of alpine brook trout is inversely related to population density. Hypothesis 1 was the crux of the study. The null hypothesis was that fish size is independent of population density. I tested this hypothesis using data among populations (Figure 1.2) and by experimentally manipulating population density in eight populations. The data in Figure 1.2 suggest that size differences between populations of brook trout are a function of population density. Results from the removal experiments showed that fish size was inversely related to population density (Figure 7.10). H2. The growth response of stunted alpine brook trout is proportional to the reduction in population density. Hypothesis 2 was an extension of hypothesis 1. Hypothesis 1 suggested that a change in growth could be accomplished by reducing the population density. Hypothesis 2 was erected to test if the growth response is proportional to the density reduction. The null hypothesis was that the growth response is independent of the proportion of the population removed. The relationship between change in length and weight was roughly proportional to the change in density (Figure 7.11). H3. The growth response of stunted alpine brook trout is inversely propor-tional to the pre-reduction fish density. Hypothesis 3 suggested that there was a difference in the severity of stunting in alpine lakes, and that the growth response of severely stunted populations would be more pronounced than the response of less stunted fish in lower density populations. The null hypothesis was 15 that the growth response is independent of pre-reduction fish density. The result was exactly opposite, but for unexpected reasons. Fish in lower density populations were aided further by the additional reduction in density. It was also easier to extensively deplete populations that were already at low density. Lacking the situation of a severely stunted population that was reduced to a density as low as populations that were initially at lower density, I could not test this hypothesis. H4. The growth responses of juvenile brook trout (age 1 and 2) and older fish (ages 11+) are less affected by population density reductions than are growth rates for mature, reproductively active fish (ages 3-10). Previous studies showed that reduced growth due to stunting may not occur until after the first few years of life, probably because of dietary differences between juveniles and adults (Hazzard 1933; Beckman 1943; Aim 1946; Deelder 1951; Heath and Roff 1987). This hypothesis suggested that the growth response for juvenile brook trout would be less than that for the pre-senescent adult population. The results refuted the juvenile portion of hypothesis 4. The response for juveniles was greater than the response of the adults, perhaps because of greater recuperative abilities in young trout. Dahl (1917), Comfort (1960), and Reimers (1979) showed that there was a reduced capacity for growth recovery in old fish. My results concur with the previous findings, and support the latter half of hypothesis 4. H5. The growth response of stunted alpine brook trout to density reduction is inversely proportional to lake elevation. Lake elevation combines several environmental factors: temperature, growing season, ice-free period, and nutrient availability. Hypothesis 5 suggested that growth response depends on these environmental factors. In general, as elevation increases, environmental conditions become less conducive to growth. I did not find any relationship between elevation and growth response, probably because environmental differences were not great enough over the range of elevations where populations were available for experimental manipulations. 16 H6. Brook trout populations that are heavily fished for sport will show better growth than populations with little or no exploitation. Hypothesis 6 was the original impetus for this study. The hypothesis was based on the idea that the additional source of mortality caused by sport fishing would reduce population density and increase the food per individual, compared to lakes with no fishing mortality. Sport fishing had little effect on the growth of brook trout populations. Heavily fished populations are often stunted, though the age distributions may differ from unexploited populations. H7. The fecundity response of stunted brook trout is proportional to the reduction in population density. I expected to find that stunted fish also had reduced fecundities, which would increase after an initial increase in weight and length. Contrary to my expectations, stunted brook trout have fecundities similar to non-stunted brook trout of the same size. Fecundity did increase in response to density reduction, but no more than would be expected from the increase in size. H8. The recruitment response of stunted brook trout is inversely related to density reduction. Previous studies have found that strong recruitment may be one result of density reduction experiments (Dahl 1917, Fagerstrom 1972, Langeland 1986, Donald and Alger 1989). I expected to find increased recruitment due to reduction in competition or cannibalism effects of older fish, but I also considered the alternatives of either decreased recruitment due to decreased egg production, or no recruitment change, due to limiting factors in spawning habitat and juvenile rearing areas. The response varied between lakes. I did find increased recruitment that I attribute to reduced cannibalism or competition, but I also found recruitment failure and reduced recruitment at the highest levels of density reduction. 17 Research History and An Outline of Subsequent Chapters The study began with the hypothesis that stunted brook trout in lakes that are subject to sport fishing should grow more rapidly than for populations that are unfished. A two week trip to 12 lakes in the Sierra Nevada in 1985 was enough to refute hypothesis 6. The selection of the 12 lakes was fortunate; they produced the largest and some of the smallest fish that I sampled in five field seasons. Growth differences existed between lakes, but exactly contrary to hypothesis 6. The largest, fastest growing brook trout were located in the least fished lakes. Fish size within a population was quite uniform. In lakes with stunted brook trout, all the fish were small. In lakes with larger brook trout, nearly all mature fish were big. One obvious difference between the populations of different sized fish was density. The largest fish were in remote lakes with very low density (hypothesis 1). I confirmed this hy-pothesis with the survey sampling in 1986. The next step seemed logical: if density could be reduced in lakes with stunted populations, then these fish should also grow. I began two exper-iments in 1986. In the first experiment I transferred stunted brook trout into a Ashless pond that had an obvious abundance of invertebrate prey. In the second experiment I removed what I believed to be a large proportion of a stunted population to promote growth by increasing the food ration per individual. In hindsight, the hypothesis of removal and growth was naive, as it did not give adequate consideration to the reproductive capabilities of the brook trout populations that had led to stunting in the first place. I describe the study area and sampling program, including field techniques and methods in chapter 2. The bulk of chapter 2 justifies the use of gillnets as my only sampling method. My observations suggested that age 1 brook trout were not fully vulnerable to the gillnets. I used differences in catchability coefficients to estimate relative vulnerabilities for age 1-3 brook trout. I recognized at the outset that age determinations would be an important part of this study. Nearby Bunny Lake produced a 24 year old brook trout (Reimers 1979), and I expected to find old brook trout in other lakes. I planned to use length and weight at age as my pri-mary measures of growth differences between and within populations. I started age validation 18 experiments in 1986 and 1987 that continued through 1989. For aging, I had planned to use a semi-automated system (camera, microscope, and computer) to assist with the determination of annuli. The technological solution failed for several reasons and I had to modify traditional techniques to age about 5000 fish per field season. I describe the aging methods and age validation experiments in chapter 3. I was surprised by the number of fish that I removed from the lakes in 1987, but I thought I had done a good job in most of the lakes of removing the bulk of the populations. That thought disappeared with the second removals in 1988. No one could have convinced me before the 1988 field season that I would be catching as many or more fish in the lakes that I had "depleted" the year before. The unexpected number of fish has led to continuing speculation on how the rest of the fish avoided capture (chapter 5). I present and discuss the data from the removal experiments in chapter 4. A logistic problem in 1987 led to a fortunate result. I had to leave an experimental lake (Wonder 3) before I had done an adequate job of depleting the population. When I returned to the lake two months later, the catch per effort was higher on my return than during the initial netting periods. This finding led to a set of ancillary experiments that I describe in chapter 5, along with further discussion on brook trout behavior in relation to capture with passive gillnets. When it became apparent from preliminary results in 1987 that improved growth was not meeting my expectations, I formulated hypothesis 7, that the increased energy available to survivors of the removal experiments was being put into reproduction. In 1988 I increased my sampling effort for ovaries to monitor an expected increase in fecundity. I present these results in chapter 6. The growth response to density reduction was mixed. Two populations did show sub-stantially increased size at age, but in six other experimental lakes the increases in length and weight were slight. The magnitude of the growth response was inversely related to density. Increased survival of juvenile brook trout severely limited my ability to control density. The 19 experimental results were aided by the lack of recruitment in the lake with the lowest density (chapter 7). The bulk of the 1989 catches were 1 year old brook trout. The enormous recruitments of age 1 fish into most of the experimental lakes was apparently due to increased juvenile survival, presumably because of reduced cannibalism or reduced competition due to removals of most of the adult population. I present my findings on the recruitment responses of the populations in chapter 8. I conclude the dissertation by reviewing new and old ideas regarding the causes of stunting, offering suggestions for further research, and summarizing my major findings and conclusions. 20 2. Background Information and General Methods Study Area Reimers et al. (1955) provided a thorough description of the Sierra Nevada mountain range in California, including summaries of vegetation, climate, and geology. The following description is extracted from their study. Reimers et al. (1955) studied the limnology of ten lakes in the Convict Creek Basin, located midway in the range of latitude of lakes that I sampled in this study. The Sierra Nevada is approximately 430 miles long and occupies a large part of eastern California, extending from Tehachapi Pass in the south to the Feather River Basin in the north. It is from 40 to 80 miles wide and consists essentially of a single, tilted granitic block of the earth's crust, formed by uplift during the end of the Pliocene and the beginning of the Pleistocene epochs. The range is strongly asymmetrical in profile. The western side slopes gradually, and the eastern side is an abrupt fault escarpment with many peaks in excess of 13000 feet. The main axis of the Sierra Nevada is north and south, but in the vicinity of Bishop, Calif., it bears westward bypassing an extensive volcanic area dominated by Mammoth Mountain. . . . The climate of the eastern slope of the Sierra Nevada is similar to that of other western mountain areas which lie to leeward of high ranges, with respect to prevailing westerly winds and Pacific storm systems. Summer rains fall on the western slopes and the high country about Convict Creek Basin is dry from May to October except for scattered thunderstorms. Summer daytime tempera-tures in the basin reach 70° to 75° F . , dropping into the thirties at night. Winters are characterized by heavy snows, severe windstorms, and frequent subzero tem-peratures. . . . Four glacial cycles occurred in the million-year duration of the Pleistocene epoch, and as a result of this tremendous glaciation several thousand lakes now exist throughout the range. These lakes are all young geologically, as none are older than the most recent glaciation cycle. Some Sierra lakes are more recent, as evidenced by the existence of glaciers which are still in the process of recession. As the earth warmed and the last glacial cycle came to an end, lakes of two principal types were created: those occupying bedrock basins and those held by morainal deposits. My study area was concentrated north and south of the westerly bypass around Mam-moth Mountain as described by Reimers et al. (1955). The influence of the Mammoth Crest is seen as the break in the 3000 m contour to the west of Mammoth Mountain in Figure 2.1. The area to the south of the contour break is the true crest of the Sierra Nevada, which culminates with Mt. Whitney (elevation 4416 m), about 22 km south of Flower Lake (Figure 2.1). 21 120° 119° 118° 36° Figure 2.1 Location of all lakes from which brook trout were sampled, with the eight exper-imental lakes emphasized and labelled. The contour is 3000 m (after Stoddard 1987b). The inset map shows the location of the study area within the state of California. 22 There are about 500 lakes above 3000 m in the area covered by Figure 2.1 that have been catalogued by California Fish and Game (CFG). Starting in the early 1900's, fish have been stocked into almost every lake and stream in the region to provide for recreational angling. At present, about 300 lakes contain populations of brook trout. I captured brook trout in 71 lakes from 1985-1989 (Figure 2.1, Table 2.1). Sixty-nine of the seventy-one lakes were on the eastern side of the Sierra Nevada crest. I sampled additional lakes that were either barren of fish or did not contain brook trout. I am aware of four previous limnological studies of Sierra Nevada lakes that were con-ducted or that included lakes within my study area (Reimers et al. 1955; Reimers 1958; Landers et al. 1987, Eilers et al. 1987; and Stoddard 1986, 1987a, 1987b). Stoddard (1986) measured pH (mean and standard error, 7.1 ±0.09; range, 5.7-9.4) and conductivity (mean and standard error, 14.9 ± 2.06; range, 3.8-111.4; in micromhos/cm at 25°C) in 67 lakes in the same region as my study area. Landers et al. (1987) concluded that "The total ionic concentration of Sierra Nevada lakes, as reflected in conductance values, shows that these lakes as a group are the most dilute sampled to date in the United States." Tonnessen (1983) and Cooper et al. (1988) provide limnological data for alpine lakes on the western slope of the Sierra Nevada. I did not collect chemical data for the lakes that I sampled for fish, and the physical data are limited to morphometric measurements given in Table 2.2 and Appendix A, and the bathymetric maps presented in Appendix A. The duration of ice cover is an important measurement in alpine lakes, since much of the annual food intake occurs during the ice-free period (Elliot and Jenkins 1972). Data on the duration of ice cover are limited. Elliot and Jenkins (1972) reported 6-7 months of complete ice cover in three high alpine Sierra Nevada lakes. They found that brook trout fed actively throughout the winter, even at temperatures as low as 1°C, though the quantity and variety of prey consumed was reduced compared to the ice-free period. Stoddard (1987a; his Figure 2) shows diagrams of ice cover in Gem Lake (referred to as Gem 3 in this study) lasting into July in 1982 and 1983, and from December 1983 (the start of the measurement) through June 1984 (at least 6 months of ice cover). In general, high alpine lakes in the Sierra Nevada are assumed 23 to become ice covered in October-December, and remain frozen until May-July, providing an ice-free period of 3-6 months. During 1987-1989, most of the lakes that I sampled were ice-free by mid-June, about 2-4 weeks earlier than in years with "normal" snowfall (E.P. Pister, personal communication). Snowfall in the Sierra Nevada was below the long-term average during my study. S a m p l i n g O v e r v i e w I began sampling in 1985 by capturing 187 brook trout from 12 lakes (Table 2.1). I explored different aging procedures with the fish caught in 1985. While developing aging procedures I destroyed most of the otoliths from the 1985 samples. Lacking age data, I excluded these samples from subsequent analyses. I sampled 1316 brook trout from 51 lakes in 1986 (Table 2.1). The purpose of the sampling in 1986 was to provide a broad overview of many brook trout populations in lakes that varied in size (0.4-4.2 surface area in hectares), latitude (36° 45'-38° 16'), and elevation (2955-3580 m above sea level). In addition to the survey sampling, I began two experiments in 1986. In the first experiment, I transferred 16 stunted brook trout to a Ashless lake in which I observed a substantial number and variety of invertebrate organisms. In the second experiment, I removed 137 brook trout from Hell Diver 2 to reduce the population density. Prior to the 1987 field season I decided to further test the hypothesis that growth stunting in brook trout could be alleviated by reducing population density. The concept of increasing fish size by increasing the food ration per individual by reducing the number of individuals was straightforward and appealing. Similar experiments had been conducted in the past with mixed results (Dahl 1917; Beckman 1941, 1943; Parker 1958; Warnick 1966; Fagerstrom 1972; Pechlaner and Zaderer 1985; Langeland 1986; Donald and Alger 1989). The situation of severely stunted, single species brook trout populations in hundreds of isolated small lakes seemed like the ideal location in which to test the hypothesis of stunting alleviation. I began eleven removal experiments in 1987, and continued the removal experiment that began in 1986. I sampled 4849 brook trout in 1987 from 20 lakes (Table 2.1); 4649 fish from the 12 experimental lakes, 71 fish from four non-experimental comparison lakes, and 129 fish 24 Table 2.1 Lakes and number of brook trout sampled, 1985-1989. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17. 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 Lake 1985 1986 1987 1988 1989 total Alpine 2 52 2 Barney 52 Bench 17 17 Bighorn (Convict Cr.) 29 29 Bighorn (Tioga Pass) 20 20 Bonnie 5 5 Bottleneck 22 632 748 120 1522 Cinko 32 32 Cloverleaf 41 41 Constance 29 29 Cora 18 18 Dingleberry 32 654 904 676 2266 East 25 25 Edith 51 51 Emerald 4 29 29 Finger 13 13 Fishgut Pond 8 8 Fishgut 1 250 383 463 1096 Fishgut 2 46 177 43 266 Fishgut 3 21 19 153 150 343 Flower 15 557 714 604 1890 31 376 31 376 Gardisky 8 8 118 Gem 1 22 14 49 33 Gem 2 133 66 11 210 Gem 3 26 21 89 136 Genevieve 50 50 Gilbert 15 15 Gilman 23 23 Golden Trout 2 48 48 Golden Trout 4 35 35 Granite 38 38 Green 51 51 Green Treble 16 16 Harriett 32 32 Heart (Rock Cr.) 91 91 Hell Diver 1 9 19 180 107 315 Hell Diver 2 137 54 97 233 521 Hell Diver 3 27 81 68 68 244 Hoover (lower) 24 24 Hoover (upper) 25 25 Inconsolable 309 27 336 Leavitt 10 10 Marsh 9 9 Matlock 18 150 39 207 Maul 9 2 23 8 51 93 Midnight 26 220 246 Mildred 43 43 Muriel 23 23 Nurserv 13 13 Par Value 23 636 764 877 2300 Pass 438 551 267 1256 Ruth 19 19 Schober 1 21 8 7 11 47 Schober 3 1 3 4 Slim 23 23 Spuller 16 16 Stella . 13 13 Summit 28 28 Sunset 8 8 Thunder & Lightning 6 6 Topsy Turvy 33 33 Treasure 1 7 7 Treasure 2 8 8 Wahoo 1 17 17 West 13 13 Witsonahpah 39 39 Wonder 2 193 145 338 Wonder 3 529 429 616 1574 Z 19 19 total 187 1316 4849 5693 4864 16909 25 from four lakes that I sampled for other reasons. I chose the twelve experimental lakes based on the 1986 surveys. The lakes represented typical small alpine lakes of the Sierra Nevada. I designed the sampling program so that the twelve experimental lakes would be divided into three replicates at four levels of density reduction (25, 50, 75, and 90 percent). This design proved impossible to carry out because an accurate estimate of population size during the removal experiment was not possible for reasons discussed in chapters 4 and 5. Instead, the percent reduction in density depended on factors specific to each lake, such as lake size, weather, time available for the experiment, and other factors. I redepleted ten of the twelve experimental lakes in 1988 to again estimate population sizes and to reduce population density further. Selected measures of lake morphometry are presented in Table 2.2 for the ten lakes redepleted in 1988. I did not redeplete two of the original twelve lakes (Gable 3 and Inconsolable). I intended to leave Gable 3 and Inconsolable unfished for two years and sample the lakes in 1989, but time constraints prevented the 1989 visits. In total, I sampled 5693 brook trout in 1988; 4724 fish from ten experimental lakes, 923 fish from seven non-experimental comparison lakes, and 46 fish from three lakes of continuing interest. I redepleted eight of the original twelve experimental lakes in 1989 to again estimate population size. I did not attempt to redeplete Bottleneck in 1989. Bottleneck was too large for the gear and help available. The attempt to redeplete Pass in 1989 failed because of adverse weather conditions; I carried out only two netting periods. I sampled 4864 brook trout in 1989; 3548 fish from eight experimental lakes, 878 fish from nine comparison lakes, and 438 fish from three lakes that I sampled for continuing interest, including Bottleneck and Pass. I based most of the subsequent analyses on data from the eight experimental lakes that were redepleted in 1989 (Flower, Wonder 3, Fishgut 1, Dingleberry, Hell Diver 3, Hell Diver 2, Par Value, and Gem 2). The removal experiment for Hell Diver 2 began in 1986; the experiments for the other seven lakes began in 1987. The order of presentation (Flower, Wonder 3, Fishgut 1, Dingleberry, Hell Diver 3, Hell Diver 2, Par Value, and Gem 2) was the order in which I sampled the lakes in 1987. I used data from other lakes for comparison and to support or refute suppositions regarding the eight experimental lakes. 26 Table 2.2 Selected measurements for ten of the original twelve experimental lakes. Column headings are as follows: elev—elevation above sea level in meters; vol—lake volume in m 3 x 104; area—surface area in m 2 x 104 (hectares); Ab3—surface area in m 2 x 104 (hectares) that is above the 3 m depth contour; z—mean depth in meters; zm—maximum depth in meters; lm—maximum length in meters; wm—maximum width in meters; L—shoreline length in meters; DL—shoreline development, the ratio of the shoreline length (L) to the length of the circumference of a circle of area equal to that of the lake, D L = L/2y/,7r(area), (Wetzel 1975); stock—last recorded year that brook trout were introduced into the lake (CFG records). lake elev vol area Ab3 z Zm lm w m L D L stock Flower 3200 2.3 1.9 1.9 1.2 2.4 251 110 611 1.26 1962 Wonder 3 3375 4.6 1.3 0.5 3.5 7.0 152 112 492 1.22 1947* Fishgut 1 3315 1.1 0.6 0.5 1.7 3.7 177 52 411 1.45 1956 Dingleberry 3195 3.8 2.1 1.7 1.8 6.7 217 130 606 1.19 1940 Hell Diver 3 3580 5.7 0.9 0.2 6.5 13.1 130 106 397 1.19 1950 Hell Diver 2 3480 1.1 0.4 0.2 2.8 5.2 100 63 261 1.15 1950 Par Value 3135 18.0 2.4 0.5 7.5 17.7 220 172 671 1.22 1955 Gem 2 3335 1.3 0.7 0.6 1.8 4.3 169 73 414 1.39 1949 Bottleneck 3390 38.5 4.2 0.5 9.1 18.3 365 247 996 1.36 1958 Pass 2955 2.5 0.8 0.4 3.4 7.9 149 76 376 1.22 1971 * CFG survey records indicate that brook trout were present in Wonder 3 in 1947, but the initial date of stocking is unknown (no stocking has occurred since 1947). Removals Due to Sport Fishing A l l the lakes sampled were open to sport fishing. The amount of fishing varied substan-tially, depending strongly on the proximity of the lake to the trailhead and existing trails. Of the eight experimental lakes, Dingleberry and Flower received the most use by anglers. Both lakes are two to three hour hikes from their trailheads and the main trails skirt the shorelines. The other six lakes received little or no angling use. The effect of sport fishing on the experimental results was negligible. Even in Dingle-berry and Flower lakes, the proportion of fish captured by anglers was small compared to the proportion of fish removed by gillnetting. Personal experience and reports from anglers indicated that during and after a removal experiment the probability of catching a brook trout using angling gear was low to zero. This is in contrast to angling for brook trout in undisturbed lakes, where even Carl Walters can occasionally catch one. 27 Stocking History I checked stocking records and survey reports maintained by California Fish and Game to determine the last year that brook trout were stocked into the lakes that I sampled (Table 2.2), and the frequency of previous introductions. The most recent brook trout stocking was in Pass in 1971. There was no recorded stocking date for Wonder 3. A survey report indicated that brook trout were already present in 1947. California Fish and Game has attempted to displace brook trout by stocking rainbow trout, golden trout, or brown trout on top of established brook trout populations. The results from my survey sampling in 1986 show that these attempts amount to little more than an annual or biennial feeding event for the resident brook trout (stocking is done by releasing fingerlings from an airplane, to fall free in a swath over the lake surface; most if not all of these fingerlings are probably consumed before they can swim to refuge areas near shoreline). Stocking in lakes without resident brook trout populations appears more successful, and is probably necessary to maintain rainbow trout and golden trout populations at fishable levels. Of the eight experimental lakes, Flower last received about 2000 rainbow trout fingerlings in 1982, Par Value was reported to have received 2250 golden trout fingerlings in 1974, and the other lakes have not been stocked with any fish since the early 1960's. Gillnet Selectivity I captured 16809 brook trout using gillnets and 100 by angling. I make no distinction between the two capture methods in subsequent analyses. The gillnets used in 1986 were provided by California Fish and Game. The nets were 37.5 m long by 1.8 m high, with five 7.5 m long panels of mesh sizes of 19, 25, 33, 38, and 50 mm (mesh sizes are in bar lengths; the stretched length or width of a mesh is twice the bar length). The 19 mm mesh was too large for small brook trout and the 50 mm mesh was too large for the largest brook trout captured in 1986. I obtained four gillnets in 1987 from Lundgrens (Sweden) that included the mesh sizes 4, 6.25, 8, 10, 12.5, 16.5, 18.5, 22, 25, 30, 33, and 38 bar mm. The nets measured 36 m long by 1.5 m high, with twelve randomly arranged panels of each mesh size. The netting material was light green monofilament, and the nets were weighted to fish on the bottom. I acquired Table 2.3 The random arrangement of 12 different mesh size panels in the ten gillnets used in this study. The year of purchase and first use is listed in the first column. panel year net 1 2 3 4 5 6 7 8 9 10 11 12 1987 1 10 4 16.5 38 30 25 6.25 12.5 33 18.5 22 8 1987 2 6.25 18.5 38 16.5 10 22 33 12.5 25 30 4 8 1987 3 10 12.5 8 30 22 38 16.5 4 18.5 33 25 6.25 1987 4 33 8 16.5 22 10 6.25 12.5 4 18.5 25 30 38 1988 5 38 10 25 12.5 16.5 33 6.25 30 8 18.5 4 22 1988 6 4 30 25 6.25 22 38 12.5 33 16.5 8 10 18.5 1989 7 6.25 12.5 22 16.5 25 4 30 10 8 38 33 18.5 1989 8 8 38 6.25 18.5 33 16.5 12.5 22 25 30 10 4 1989 9 22 6.25 38 30 4 12.5 8 25 33 16.5 10 18.5 1989 10 4 12.5 30 25 18.5 22 8 16.5 38 6.25 10 33 two additional nets of this design in 1988 and four additional nets in 1989. The order of the panels is listed in Table 2.3. Gillnets are notoriously selective and any analyses and conclusions based on gillnet sample data are suspect. The remote location of the lakes required that the sampling gear was lightweight and easy to carry. Each gillnet weighed less than 1.5 kg. Angling gear is also lightweight, but it was inefficient. I tested a large trapnet in Dingleberry lake in 1988. The 40 kg trapnet required a mule to carry it to the lake and two inflatable boats to set it. On the first night of fishing, six gillnets captured 248 fish while the trapnet captured 9 fish. I did not test other gear types because of obvious constraints: the lake bottoms were too rocky for beach seines; purse seines require sturdy rafts or boats; and portable electric fishing gear is ineffective in lakes, especially in low conductivity water typical of alpine lakes. Any gear is sufficient if it samples the population in proportion to actual abundance. Based on the following analyses, I considered brook trout age 2 and older to be fully vulnerable to the gillnets. Determining the age at which brook trout were fully vulnerable to the gillnets using catch data from the gillnets is circular and subjective, but it was my only practical option. Ricker (1975) suggested that the age of complete recruitment is at or near the modal 29 age from a catch sample. The modal age for most samples was age 2 in 1987 and age 1 in 1988 and 1989 (Table 2.4). The decrease in the modal age from age 2 to age 1 can be explained by the increase in the mean size of age 1 brook trout during the study (Table 2.5). The probability of a fish becoming entangled in a gillnet depends on size, shape, swimming speed (momentum), the chance of encountering a net, and characteristics of the gillnet (for example, material, color, and mesh sizes). Age 1 fish were not sampled in proportion to their abundance in 1987 because they were either too small or slow to become entangled, or they did not encounter the gillnets (for example, by avoiding predaceous adults in shallow water). The age 1 brook trout in 1989 were larger, and encountered the gillnets. Additional evidence of age 1 brook trout becoming vulnerable to gillnets is available from two sample periods in Wonder 3 in 1987. I caught 242 fish during June 27-30, but no age 1 fish. During August 28-30, I caught 287 fish, 23 age 1 (8%). The age 1 brook trout were either growing into a body size that was susceptible to entanglement, or were beginning to venture into parts of the lake where the gillnets were fishing. I calculated relative vulnerability for ages 1, 2, and 3 versus ages 4-10 and for three size classes (fork lengths in mm of < 120, 121-150, and 151-180, versus fish > 181 mm) as a more rigorous means of estimating the age and size of full vulnerability to the gillnets. I calculated estimates of catchability coefficients (q) from standard Leslie depletion methods (Ricker 1975; also described in chapter 4) for each age and size class (Tables 2.6 and 2.7). Catchability coefficients can be used to calculate relative vulnerabilities, even if the overall estimate of population size estimated by the Leslie method are biased (C.J. Walters, personal communication). I eliminated q values that were based on small sample sizes (10-20 fish over several netting periods) or that were based on regressions with r2 values less than 0.30. With the estimates that remained I calculated relative vulnerabilities by dividing the q for the fully vulnerable groups (ages 4-10, lengths > 181) into the q for the younger ages and smaller size classes (Tables 2.6 and 2.7). Because of the large number of missing estimates, I calculated mean relative vulnerabilities for each age and size class by year. 30 Table 2.4 Modal age from the catch data for twelve lakes from 1987-1989. 1987 1988 1989 Flower 2 1 1 Gable 3 2 Wonder 3 2 1 1 Inconsolable 2 Fishgut 1 3 3 1 Dingleberry 2 3 1 Hell Diver 3 7 2 1,2 Hell Diver 2 5 2 1 Bottleneck 2 1 1 Pass 2 1 1 Par Value 2,6 1 1 Gem 2 1,4 2 6 Table 2.5 Mean length and mean weight of age 1 brook trout for twelve lakes from 1987-1989. mean length (mm) mean weight (g) 1987 1988 1989 1987 1988 1989 Flower 97.0 105.4 121.5 10.5 12.9 19.8 Gable 3 78.0 5.0 Wonder 3 89.0 111.2 125.5 7.3 16.3 22.2 Inconsolable 116.0 19.0 Fishgut 1 91.0 95.7 128.9 7.8 10.6 24.3 Dingleberry 91.0 111.8 131.0 8.5 17.1 27.1 Hell Diver 3 149.0 123.7 136.5 42.0 22.2 30.0 Hell Diver 2 79.0 94.0 103.0 4.7 9.0 12.0 Bottleneck 89.0 105.7 133.8 8.5 13.8 28.4 Pass 117.0 122.4 143.8 18.1 22.1 33.8 Par Value 100.0 119.1 110.6 10.6 21.0 15.8 Gem 2 125.0 128.0 20.5 23.0 31 Table 2.6 Catchability coefficients (q) and relative vulnerability for ages 1-3 in seven of the experimental lakes, 1987-1989. Missing values are either unavailable or insufficient for the analysis. See text for further explanation. Flower age 1 2 3 4-10 catchability (q) 1987 1988 0.00286 0.00440 0.00523 0.00361 1989 0.00509 0.00621 0.00718 relative vulnerabilty 1987 1988 0.79 1.22 1.47 1989 0.71 0.87 Wonder 3 1 2 3 4-10 0.00242 0.00716 0.00750 0.00434 0.00340 0.00486 0.00484 0.00328 0.56 1.65 1.73 1.04 1.48 1.48 Fishgut 1 1 2 3 4-10 0.01582 0.02434 0.02165 0.00788 0.00673 0.00595 0.00672 0.00820 0.00689 0.73 1.12 1.33 1.13 0.98 1.19 Dingleberry 1 2 3 4-10 0.00666 0.00508 0.00562 0.00143 0.00434 0.00385 0.00299 0.00220 0.00304 0.00253 0.00524 1.18 0.90 0.48 1.48 1.28 0.42 0.58 0.64 Hell Diver 3 1 2 3 4-10 0.01361 0.00778 0.01060 0.02972 0.01461 1.74 Hell Diver 2 1 2 3 4-10 0.02510 0.01286 0.01078 0.00685 0.02432 0.01430 1.19 0.48 1.70 Par Value 1 2 3 4-10 0.00653 0.01195 0.01062 0.01165 0.00354 0.00244 0.00267 0.00353 0.00203 0.00267 0.00334 0.00266 0.56 1.03 0.91 1.00 0.69 0.76 0.76 1.00 1.26 mean relative 1 vulnerability 2 3 0.56 0.98 0.98 0.71 1.35 1.26 0.74 0.99 1.19 32 Table 2.7 Catchability coefficients (if) and relative vulnerability for four size classes (fork lengths in mm) in seven of the experimental lakes, 1987-1989. Missing values are either unavailable or insufficient for the analysis. See text for further explanation. length catchability (if) 1987 1988 1989 relative vulnerabilty 1987 1988 1989 Flower < 120 121-150 151-180 > 181 0.00297 0.00449 0.00385 0.00366 0.00171 0.00485 0.00704 0.81 1.23 1.06 0.24 0.69 Wonder 3 < 120 121-150 151-180 > 181 0.00127 0.00583 0.00576 0.00419 0.00223 0.00372 0.00516 0,00297 0.30 1.39 1.37 0.75 1.25 1.73 Fishgut 1 < 120 121-150 151-180 > 181 0.02834 0.03547 0.03968 0.00576 0.00656 0.00549 0.00548 0.00695 0.00839 0.00687 0.71 0.89 1.05 1.19 0.80 1.01 1.22 Dingleberry < 120 121-150 151-180 > 181 0.00808 0.00565 0.00553 0.00155 0.00505 0.00389 0.00314 0.00097 0.00276 0.00131 0.00392 0.49 1.46 1.61 1.02 1.24 0.25 0.70 0.33 Hell Diver 3 < 120 121-150 151-180 > 181 0.01166 0.01726 0.00729 0.02285 0.01859 2.36 1.23 Hell Diver 2 < 120 121-150 151-180 > 181 0.01733 0.02632 0.01585 0.00689 0.02640 0.02077 1.09 0.33 1.27 Par Value < 120 121-150 151-180 > 181 0.00606 0.01327 0.01283 0.01166 0.00347 0.00350 0.00252 0.00357 0.00194 0.00285 0.00230 0.00250 0.52 0.97 1.14 0.98 1.10 0.71 0.78 1.14 0.92 mean relative vulnerabilty < 120 121-150 151-180 0.52 1.10 1.00 0.64 1.22 1.32 0.52 1.02 1.06 33 300 mesh size (bar mm) Figure 2.2 Catch frequency by mesh size for 1352 brook trout sampled between 1986-1989. The relative vulnerability data show that fish age 2 and older, and fish above 120 mm, were fully vulnerable to the gillnets (Tables 2.6 and 2.7). The mean relative vulnerabilities for age 1 fish increased from 0.56 in 1987 to 0.74 in 1989. The mean relative vulnerabilities for fish < 120 mm were 0.52, 0.64, and 0.52 in 1987-1989. The increase in relative vulnerability for age 1 versus the lack of change in the smallest size group can be explained by the increase in the size of age 1 fish (Table 2.5). I use the relative vulnerability data to correct catch at age 1 data presented in chapter 4. Juvenile brook trout were not too small for the smallest mesh size. I recorded the mesh size of entanglement for 1352 brook trout captured in 1986-1989 (Figure 2.2). Relative frequencies of length, weight, and age by mesh size are presented in Figures 2.3, 2.4, and 2.5. The capture efficiency of the three smallest mesh sizes (4, 6.25, and 8 mm) was low and these meshes did not entangle the smallest fish. The few fish captured in the smallest meshes were usually larger fish that became entangled by their mouth parts (Figure 2.5). Age 1 brook trout were captured most effectively by the 10 mm mesh (Figure 2.5). 34 .2 .1 -.2 .1 38 mm n = 40 33 mm n = 48 30 mm n = 63 25 mm n = 173 22 mm n = 270 o c CD 3 cr CD CD > CD 18.5 mm n = 273 16.5 mm n = 220 12.5 mm n = 130 r - T ~ L 10 mm n = 78 n n o mm n = 35 n 6.25 mm n = 10 4 mm n = 12 100 150 200 250 fork length (mm) 300 350 Figure 2.3 Relative catch frequency versus length in mm by mesh size for 1352 brook trout sampled between 1986-1989. Mesh sizes in bar mm and sample sizes are listed in each panel. 35 .1 -38 mm n = 40 -I 1 33 mm n = 48 30 mm n = 63 22 mm n = 270 18,5 mm n = 273 16.5 mm n = 220 12.5 mm n = 130 10 mm n = 77 a mm n = 35 6.25 mm n = 10 JUL 4 mm n = 12 50 100 150 weight (g) 200 250 >300 Figure 2.4 Relative catch frequency versus weight in g by mesh size for 1351 brook trout sampled between 1986-1989. Mesh sizes in bar mm and sample sizes are listed in each panel. 36 38 mm n = 40 33 mm n = 47 1 30 mm n = 62 , 1 ""1 1 1 25 mm n = 169 22 mm n = 262 1 1 | 18.5 mm n = 269 I 1 16.5 mm n = 216 12.5 mm n = 126 10 mm n = 77 I l 8 mm n = 33 - 6.25 mm n = 10 -4 mm n = 12 10 11 12 13 14 15 age Figure 2.5 Relative catch frequency versus age by mesh size for 1323 brook trout sampled between 1986-1989. Mesh sizes in bar mm and sample sizes are listed in each panel. 37 The problems of gillnet selectivity are reduced if the size differences between ages are small. For stunted adult brook trout, growth in length slowed at approximately 200 mm and growth in weight slowed at approximately 80-100 g (Figures 2.3 and 2.4, and chapter 7). Based on size, the probability of capture for an age 12 brook trout would be the same as the probability of capture for an age 5 brook trout, since growth almost ceased by age 5 for most populations (chapter 7). The data presented in Figures 2.2-2.5 suggest that the largest mesh sizes (33 and 38 mm) were too large and inefficient for the sizes of brook trout encountered in stunted populations in the Sierra Nevada. The number of captures declined in these meshes, and the size of the fish captured was variable. The wide range of sizes that were captured in the largest meshes suggests capture by entanglement of body parts rather than wedging (Hamley 1975). In an experimental study of using gillnets for testfishing, Hammar and Filipsson (1985; in their Figure 6) showed that the mean length of Arctic char Salvelinus alpinus increased with increasing mesh size. They also used gillnets manufactured by Lundgrens. Their selection of mesh sizes (10, 12.5, 16.5, 22, 25, 30, 33, 38, and 50 bar mm) was identical with the larger mesh sizes used in my study with the addition of the 50 mm mesh. Their data show that when large fish are present, the larger mesh sizes are effective. Therefore it seems reasonable to assume that I did not catch large brook trout because few to no large brook trout were present in the lakes, not because of size selective bias due to the gillnets. Gillnet Saturation Another problem of gillnets is that the efficiency of the gear and therefore catch may decline with time during a set. Other fish trapped in the net may reduce the probability of additional fish becoming entangled. The data from individual nets set during the first netting period in each lake (from one to six nets per lake) are presented in Figure 2.6. The data are widely scattered because of density differences between lakes. The lack of linearity through the origin and the means of the two groups suggest that saturation may be a factor. Circumstantial evidence suggests that saturation was not a major problem. First, catch per unit of effort did decline over the course of a depletion experiment, and the highest catch rates were usually in 38 80 70 -60 -50 -0 6 12 18 24 set duration (hours) Figure 2.6 Catch versus set duration in hours for 249 gillnet sets made between 1986-1989. The data are from the first netting period in each lake. One to six nets were used during the first netting period. Each point represents the catch from one net. The large filled circles are the means of the two groups (below and above 9 h set durations). The change in slope suggests gillnet saturation, or total depletion. the first netting period (chapter 4; Figure 4.2). If net saturation had been a significant factor, catch from the first few netting periods would have been similar with no decline in catch rate. Second, I observed by location of entanglement that entangled fish attracted other fish. For example, trapped juvenile brook trout attracted larger predaceous adults. I also observed clusters of similar size fish, but clustering can be explained by schooling as well as attraction. Gillnetting Methods I usually set the gillnets perpendicular to shore. I tied the floatlines to shore at distances depending on the nearshore depth. In the small lakes (Fishgut 1, Hell Diver 2, and Gem 2) the nets were longer than the width of the lakes. In these lakes I set the nets at an angle to allow the full length to be fished. I set the nets from a one-person raft by paddling backward 39 and letting the net slide out over the bow of the raft. At the offshore end I tied a rock to the leadline. Before sinking the net I attached a long line and a marker buoy to the floatline. I pulled the nets starting at the offshore buoy into a 10 L bucket. An assistant on shore removed the fish and prepared the net for resetting. If few fish were in the net, I removed the fish as I pulled the net and then reset the net directly out of the bucket. I recorded approximate locations of each set during the experiment. I attempted to set each net in a new location to provide thorough coverage, but lake size limited the number of positions. With extra nets or few entangled fish, removing fish and resetting the gillnets took about 10-20 minutes per net, depending on distance between net locations. The time spent per net was important in maintaining consistent units of effort. For the survey samples in 1986, I usually set one or two nets overnight or for several hours during the day to catch a sample of 20-50 fish. For samples from control lakes and other lakes of interest in 1986-1989, the number of nets fished and the soak time depended on the number of samples required. For the removal experiments in 1987 and 1988, I usually used three netting periods per 24 hour period, with one period overnight and two periods during the day. I adjusted the number of nets to catch about 150-200 fish per day, the limit that two people could process for length, weight, sex, and otolith removal during daylight hours. In hindsight this approach was unfortunate because it limited analysis of the catch depletion data to methods that allow variable efforts over the depletion period. In 1989 I set and pulled either four or six nets (depending on the size of the lake) as close to dusk and dawn as possible, so as to standardize the unit of effort throughout each removal experiment. This approach allowed the use of better models for estimating population size from the removal data, but it was difficult to process all the fish captured at the beginning of each experiment. Field Sampling Methods I removed the fish from the gillnets soon after pulling each net and placed them into a nylon stuff sack associated with the net of capture. If mesh size data were to be recorded, I kept the fish separate by mesh and net. I processed the fish usually within 12 hours and always 40 within 24 hours following capture. Processing consisted of measuring length, head length, weight, identifying the sex, removing a subsample of gonads, and removing the otoliths. I measured fork length from the snout to the fork of the tail to the nearest mm using a custom built aluminum measuring board. The snout was placed against an endboard and the measurement read from a scale in place under the fish. I measured head length from the snout to the posterior edge of the operculum to the nearest mm by placing a rule against the endboard and on top of the fish. I measured weight to the nearest g using a spring scale (Pesola model number 140; Jennings 1989) by suspending each fish from the scale using a spring loaded clip attached to a fin or the mouth. I weighed fish heavier than 300 g (the capacity of the scale) in sections. I checked the scale with calibration weights periodically during the field season and it was always accurate. I inspected the gonads of each fish to determine sex. I classified each fish as male, female, or juvenile if I could not distinguish between testes and ovaries. In 1989 I added the classification of immature female. I determined from fecundity samples taken in 1988 that eggs from small females could not be separated for counting. The classification of immature female eliminated these fish from ovary sampling. I preserved the ovaries in 5% formalin contained in 6 ounce Whirlpaks. I describe ovary subsampling and measurements in chapter 6. I describe aging methods in chapter 3. 41 3. Age Val idat ion and Ag ing Methods for Stunted B r o o k Trout As part of this study of stunted brook trout Salvelinus fontinalis in alpine lakes in the Sierra Nevada, California, I required age data to assess population dynamics. Beamish and McFarlane (1983) reminded fisheries biologists of the necessity of validating ages for all studies that involve fish age data. I used oxytetracycline to produce a time mark in the otoliths of fish that were released in situ for a year or more after marking. The results showed that there are discernible annual features in otoliths ground to the sagittal midplane. Age determinations made using otolith cross sections were similar to ages by sagittal section, suggesting that deposition continues on most surfaces of the otolith throughout the life of a stunted brook trout. Here I describe the age validation experiment and aging methods for small otoliths from stunted brook trout. Previous studies have determined that the calcium carbonate and protein matrix that forms otoliths is not uniformly deposited on the surface of the otolith (Irie 1960; Mugiya 1974; Beamish 1979a). Matrix deposition may be limited to the proximal (sulcus) surface of the otolith, especially in older, slow-growing fish (Beamish and Chilton 1982). Several studies have found that fish age determinations made by cross sectioning the otolith are older than age determinations made by viewing the surface of the whole otolith (Beamish 1979a, b; Beamish and Chilton 1982; Chilton and Beamish 1982; Barber and McFarlane 1987; Chilton and Stocker 1987; Fargo and Chilton 1987). Cross sections are difficult to produce for small otoliths. Instead, I prepared otoliths from stunted brook trout by grinding away the calcareous overburden to the sagittal midplane (Pannella 1980b; Campana and Neilson 1985). I compared age determinations for the two techniques by grinding to the transverse midplane and to the sagittal midplane (from now on called cross sections and sagittal sections, though the term "section" is misleading) to test the hypothesis of unequal matrix deposition. Long-lived, slow-growing brook trout were an ideal fish for investigating this hypothesis in fresh-water salmonids. 42 Methods I obtained brook trout by angling or by gillnet in four lakes (Hell Diver 3, Fishgut 2 and 3, and Gem 2) on the eastern side of the Sierra Nevada crest. The lakes were chosen as typical of hundreds of high alpine lakes in the southern Sierra Nevada. The four lakes ranged in elevation from 3330-3580 m and in surface area from 0.7-3.7 hectares. Brook trout were the only fish species present. A l l brook trout sampled from the lakes were growth stunted. The largest fish marked with oxytetracycline was 224 mm in fork length and 154 g in weight. Age Validation I chose otoliths as the structure from which to age brook trout, based on information from previous studies in the Sierra Nevada (Reimers 1958, 1979). I diluted oxytetracycline hydrochloride (OTC) in solution (100 mg per mL; brand name Liquamycin 100) 1:1 with a modified Krebs' saline solution to reduce viscosity. The following compounds measured in grams were dissolved in 1 L of distilled water to make up the saline solution (Wolf 1963): The saline solution did not contain calcium, as the O T C will bind with the calcium in solution instead of binding with calcium in the otolith ( B . M . Leaman, Canada Department of Fisheries and Oceans, Pacific Biological Station, personal communication). I used a dosage of 50 mg of O T C to 1 kg of fish body weight. The fish chosen for O T C treatment showed no apparent injury due to capture. Angled fish were treated within 1 h of capture. Gillnetted fish were kept in an enclosure for at most 2 d, and treated and released when the netting was finished. The treatment consisted of anesthetizing the fish with MS222, measuring for length and weight, injecting the O T C solution, and removing the adipose fin to provide an external mark. The injection was made ventrally, midway between the pelvic and N a C l K C I N a 2 C 0 3 M g S 0 4 • 7 H 2 0 7.41 0.37 0.16 0.31 N a H 2 P 0 4 - H 2 0 N a 2 H P 0 4 - 2 H 2 0 K H 2 P 0 4 0.40 0.20 0.17 43 pectoral fins. A 1 mL syringe with a 27 G 1/2 needle was inserted at a shallow angle to avoid damage to internal organs. A l l recoveries were made with gillnets. I identified marked fish by the absence of the adipose fin. The sagittae were removed after sectioning the head along the medial ventral line. Membranes and surrounding tissues were removed from the otoliths. The otolith pair from recoveries made in 1987 were stored separately, one dry, and the other in a 1:1 solution of glycerin and distilled water, with a pinch of thymol added to 1 L of the solution to reduce the growth of bacteria, fungae, and algae (Chilton and Beamish 1982). Neither storage method had a degrading effect on the OTC mark, so otoliths sampled in 1988 and 1989 were stored in the glycerin solution contained in 0.5 ml microcentrifuge tubes, then placed in light-proof containers (since the OTC mark may degrade when exposed to light). The otoliths were ground to the sagittal plane (below) and examined for a tetracyline mark under a compound microscope at 100-200X with reflected ultraviolet light. I used Kodak Ektachrome 160 tungsten film for color slides and Kodak Tmax 100 film for black and white photographs. In 1986, seven brook trout were injected and held in an enclosure to determine the amount of time before the OTC mark appeared on the otoliths. Otoliths were removed from three of the fish after 32 h and the remaining four fish after 46 h. Otolith Preparation and Aging Procedures Otoliths from stunted brook trout are too small (1-3 mm in length) to age by "break and burn" techniques (Chilton and Beamish 1982). Sawed cross sections (Beamish 1979a) were too slow and difficult to process the 16000+ otoliths sampled during my study. Not all annuli were discernible by viewing the magnified lateral surface of the whole otolith under a variety of lighting and immersion conditions, especially for larger, more opaque otoliths, presumably from older brook trout. Campana and Neilson (1985) recommended grinding for removing the calcareous overburden to improve structural clarity. I ground otoliths to the sagittal plane (Pannella 1980a) using a simple and efficient method. One person can mount, grind, and age about 100 otoliths per normal working day. Otoliths from stunted brook trout are somewhat flat, making the sagittal grind technique especially well suited to their shape. I used the 44 used the mounting and grinding method described to prepare both OTC marked sagitta and unmarked sagitta for aging. The otoliths were removed from the glycerin solution and blotted dry. I used the right sagitta for aging if it was present and not aberrant (Mugiya 1972). A small drop of Crystalbond thermosetting plastic resin (Neilson and Geen 1981) was placed on a standard glass microscope slide and kept liquid on a hot plate. The sagitta was placed sulcus side-up on the resin drop. The resin set within 30 s after the slide was removed from the hot plate, allowing enough time to position the sagittal plane parallel to the surface of the slide. If necessary, the Crystalbond resin can be reheated to reposition the sagitta. Each otolith was rough ground with 1200 grit abrasive paper, then fine ground with aluminum oxide or silicon carbide lapping film at 15, 12, or 9 /im particle size. About 90% of the sagittae were aged at this stage with no further processing. I used two procedures for sagittae that did not show clear annuli at this stage. If clarity was poor near the edge, mechanical polishing with jewellers rouge often improved clarity. If translucence was poor toward the nucleus, I turned the sagitta over and ground part of the distal surface by the same two-stage grinding process, followed by polishing if necessary. Though clarity could be improved by wetting the ground otolith surface with oil or glycerine, my ability to differentiate annuli was not much improved, and most sagittae were aged dry. The number of annuli were counted for each sagitta. For stunted brook trout otoliths, annuli were usually distinct along at least one radius extending from the nucleus to the an-terodorsal, posterodorsal, and posteroventral edges, and along the anterior tip (terminology from Pannella 1980a). The only zone that was consistently "unreadable" was the area ventral to the nucleus extending to an arc bounded by the anteroventral and posteroventral edges. I aged most sagitta by illuminating with a fiber optic lamp and viewing with a stereo dissecting microscope at 25X. A compound microscope at 40X, 100X, and 400X was used for sagitta that required greater magnification or transmitted light to resolve the annuli. 45 Sagittal Section versus Cross Section I made cross-sections of brook trout otoliths by breaking off the anterior third of the sagitta, mounting vertically (posterior end down) on a microscope slide, grinding to a cross-section plane that passed through the nucleus, turning and remounting the sagitta on end, and grinding from the posterior edge toward the cross-section plane. I ground cross-sections until transmitted light uniformly illuminated the section from the nucleus toward the ventral, dorsal, and proximal edges. The section was fine ground and polished using the same methods described for the sagittal sections. Forty-five sagittae from brook trout that had already been aged by the sagittal section were selected at random, stratified by age-class 2-15. I placed emphasis on older fish (11-15) to detect the effect of unequal growth zone deposition. I counted annuli on each cross section without knowledge of the count from the corresponding sagittal section. Results Age Validation Eight of the 39 fish recovered with clipped adipose fins did not show an OTC mark (Table 3.1). Two of those eight had aberrant otoliths (Mugiya 1972) in which the uptake of OTC may have been inhibited. In the other six brook trout, OTC injections may have been faulty, perhaps by injection into the digestive tract or by the OTC solution seeping out of the point of injection. No fish older than age 10 or younger than age 2 were sampled (Table 3.2). Al l 31 fish that showed an OTC mark had the correct number of annuli after the mark, corresponding exactly to the number of winters at large (Table 3.1). Figure 3.1 shows an age 6 brook trout with 2 annuli after the OTC mark. Note the discontinuity in Figure 3.1b, apparently created by uptake of OTC into the otolith, stress from the sampling process, or both. The fish in Figure 3.1 was injected with the OTC solution on August 26, 1987 and recaptured on July 20, 1989. 46 Table 3.1 The number of stunted brook injected with oxytetracycline and released, the dates of release and recovery, the number of winters at large during which time annuli are formed, and the number fish recovered with OTC marks and fin-clips. date of date of winters OTC fin lake n release recovery at large marks clips Hell Diver 3 20 23 Aug 1986 26 Jul 1987 1 . 8 11 28 Jul 1988 2 4 6 14 Aug 1989 3 1 2 Fishgut 2 and 3 16 30 Aug 1986 19 Jul 1987 1 8 8 Gem 2 13 26 Aug 1987 10 Jun 1988 1 3 3 10 Aug 1988 1 6 8 20 Jul 1989 2 1 1 31 39 Results from the short-term treatments of OTC injected fish were inconclusive. One otolith from a 32 hour fish had a fluorescent mark when viewed whole under a dissecting microscope and a hand-held ultraviolet light source. None of the otoliths from the : six other fish showed a similar mark. Based on recovery data, the dosage of 50 mg OTC per kg of fish body weight did not appear to cause additional mortality for injected brook trout (Table 3.1). Of 20 injected fish in Hell Diver 3, 19 fin-clipped fish were recovered, 13 with visible OTC marks. Of 13 injected brook trout in Gem 2, 12 fin-clipped fish were recovered, 10 with visible OTC marks. The recovery rate was lower in brook trout marked from Fishgut 2 and 3 (50%), but the 16 marked fish were also part of a separate experiment that was likely fatal to some of the marked fish, which might explain why only eight fish were recovered. Sagittal Section versus Cross Section In 35 of 45 samples, I counted the same number of annuli in cross sections and sagittal sections (Figure 3.2). Eight of the ten differences involved cross section counts that were below sagittal section counts, whereas two of the ten differences were cross section counts that were greater than sagittal section counts. The deviations increased with age. 47 a) Figure 3.1 (follows) Photographs of the anterior tip of an age 6 brook trout otolith showing a) the oxytetracycline (OTC) mark with reflected ultraviolet light and b) two annuli after the OTC mark with transmitted light. Note the discontinuity associated with the OTC mark in b) . The fish was injected with an OTC solution on August 26, 1987 and recaptured on July 20, 1989. The first annulus is outside the frame of both photographs. 48 15 CD a> 03 c _o ~o CD CO co co O o 10 5 -10 15 sagittal section age Figure 3.2 Age determinations made from one sagitta ground to the transverse midplane (cross section) versus age determinations made from the other sagitta ground to the sagittal midplane (sagittal section) based on 90 paired sagitta from 45 stunted brook trout. The line at 45 degrees represents one to one correspondence between the age determination techniques. The plotting symbols represent from 1 to 5 occurrences at each point as denned in the legend. Table 3.2 The number of brook trout with valid OTC marks for recovery ages 2-10. age year 2 3 4 5 6 7 8 9 10 total 1987 3 4 2 1 3 1 2 16 1988 3 6 2 2 13 1989 1 1 2 total 3 3 0 10 3 1 4 3 4 31 49 Discussion Age Validation I validated ages 2, 3, and 5-10 in this study of stunted brook trout. I did not recapture any brook trout age 1, age 4, or older than age 10, so these ages were not validated. Validation of age 1 brook trout required capture of age 0 juveniles, which were rarely entangled by gillnets and never captured by angling. For the purposes of subsequent analyses that use the age data, I considered validation through age 10 sufficient, because 15982 of 16574 aged brook trout (96%) were younger than age 11. Beamish and McFarlane (1983) were critical of age validation studies that only validated ages from young fish in the fast-growth phase and then extrapolated the validations to older, slow-growing fish. For stunted brook trout in alpine lakes of the Sierra Nevada, growth in length and weight almost ceases by age 5. There were no indications of growth changes in non-validated ages 11-16 that were different from growth changes during the validated, slow-growth ages 5-10. Sagittal Section versus Cross Section For brook trout otoliths, sagittal sections were prepared in about one-tenth the time it took to make a cross section. Sagittal sections were easier to age than cross sections because of greater distances between annuli in the sagittal plane, especially toward the anterior tip in older aged brook trout. This explains why eight of the sagittal section ages in Figure 3.2 were older than the corresponding cross section ages in older brook trout. Previous studies have found that ages determined from cross sections (either sawed or "break and burn") were higher than corresponding ages determined from viewing the unground surface of otoliths (Beamish 1979a,b; Beamish and Chilton 1982; Chilton and Beamish 1982; Barber and McFarlane 1987; Chilton and Stocker 1987; Fargo and Chilton 1987). Beamish and Chilton (1982), Chilton and Beamish (1982), and Barber and McFarlane (1987) have suggested that this finding may be the result of allometric otolith growth as found in previous studies (Irie 1960; Mugiya 1974; Beamish 1979a). 50 Stunted brook trout otoliths do increase in size allometrically, but there is no indication that deposition of the calcium carbonate and protein matrix continues on the proximal (sulcus) surface while ceasing on other surfaces. Irie (1960) and Mugiya (1974) have been incorrectly cited as evidence that deposition is heaviest or restricted to the proximal surface of the otolith in older fish (Beamish and Chilton 1982; Chilton and Beamish 1982; Barber and McFarlane 1987). Neither Irie (1960) or Mugiya (1974) reported deposition patterns in old fish; the otoliths used in both studies were from fish younger than age 3. Neither study found that deposition was restricted to or accumulated more rapidly on the proximal surface. Irie (1960) found that calcium deposition in cross sections was most abundant on the dorsal and ventral edges, least on the distal surface, and medium on the proximal surface. Mugiya (1974) found that a calcium-45 isotope was deposited and accumulated abundantly on the proximal surface, the dorsal and ventral margins, and the anterior and posterior tips. Observations of stunted brook trout otoliths show that older fish have elongated anterior tips (though not always along a straight line), suggesting that matrix deposition continues along the anterior tip (and possibly on all otolith surfaces) throughout the life of the fish. It is also intuitive that the longest structural feature of stunted brook trout otoliths (the anterior tip) should be the location of the most discernible annuli, since the distances between annuli are greatest along this feature. The OTC data (Table 3.2) confirm that matrix deposition occurs on all otolith surfaces visible in sagittal sections up through age 10, since the OTC mark and subsequent annuli were visible in the sagittal sections. Reimers (1979) shows a photograph of an age 20 brook trout, with annuli most clearly visible along the anterior tip in the unground sagittal plane. The data confirm that calcium carbonate and protein matrix deposition continues on otolith surfaces visible in sagittal and cross sections throughout the life of stunted brook trout. Previous studies that have advocated cross sections based on unequal deposition may have been misled by annuli that were not visible in the whole surface (unground sagittal) view. Cross sections are more appropriate for many shapes and sizes of otoliths, but for reasons of technique, not necessarily because of the absence of annular growth zones. Before the simpler 51 technique of whole surface is used in aging studies, it should be compared with sagittal and cross section techniques to determine if all annuli are discernible in the whole surface view. If all annuli are not visible in the whole surface view, the otolith must be sectioned or ground to obtain accurate age estimates. Depending on the size and shape of otoliths from other species, the sagittal grind may be the most advantageous technique. 52 4. Estimating Population Size, Survival, and Density I required estimates of population size and fish density to assess changes in growth, fe-cundity, survival, and recruitment. Two tasks were necessary: an estimate of initial population size, and the removal of part of the population. I used the removal method to accomplish both tasks simultaneously. Most estimation procedures for removal data require the assumption that the probability of capture (catchability, q) be constant among individuals and between capture occasions (Leslie and Davis 1939; Delury 1947; Hayne 1949; Moran 1951; Zippin 1956; Seber and Le Cren 1967; Seber and Whale 1970; Ricker 1975; Carle and Strub 1978; Seber 1982; Crittenden and Thomas 1989; see Cowx 1983, and Gatz and Loar 1988, for reviews of removal methods). Previous studies have determined that the proportion of fish captured may decline between capture occasions, even with the same amount of effort (Cross and Stott 1975; Bohlin and Sundstrom 1977; Mahon 1980; Peterson and Cederholm 1984; Kelso and Shuter 1989), thus violating the assumption of constant catchability. Removal data from brook trout populations in high alpine lakes in the Sierra Nevada suggest that catchability is not constant during a removal experiment (Figure 4.2). The catch from the first capture occasion tended to be greater than catches from subsequent capture occasions, suggesting that the first capture occasion removed fish with an above average prob-ability of capture. Otis et al. (1978) and Schnute (1983) developed models for estimating initial population size and capture proportions for removal data that exhibit variable or con-stant catchability. I estimated population sizes at the start of fishing in 1989 for eight lake populations of brook trout using the methods of Otis et al. (1978) and Schnute (1983). I back-calculated pre-study population sizes for the eight lakes using the 1989 population estimates, catch data from 1986-1988, and estimates of survival at age for each lake population. 53 Methods Removal Experiments I carried out removal experiments on eight lake populations of brook trout in 1989 (Figure 4.1). The lakes were small, with four of the eight lakes under 1 hectare. None of the lakes had any macrophytic vegetation; the only refugia were rocks in all lakes and scarce fallen trees and branches in the six lakes below tree line (about 3400 m). Brook trout were the only fish species captured in six of the eight lakes, and dominated catches in the two lakes with other species present (in 1989, one rainbow trout was captured in Dingleberry and two brown trout were captured in Flower). Two of the lakes are isolated from any source of fish immigration (Par Value and Gem 2), and in five of the other six lakes immigration was likely minimal to none due to separation from adjacent lakes by lengthy, cascading streams. Only Dingleberry has substantial inlet and outlet streams that may have held enough fish to provide a significant source of immigration. Emigration was possible in seven of the eight lakes; Gem 2 has no stream outlet. The effects of recruitment, immigration, emigration, and natural mortality on the population estimates were reduced or eliminated by conducting short duration removal experiments that lasted no longer than seven days. I tried other methods of estimating population size and found them ineffective and difficult to conduct in remote, high alpine lakes (accessible only on foot). Direct counts from surface vantage points or underwater were not possible in the larger, deeper lakes. Mark-recapture methods require a means of capturing a large number of uninjured fish. The difficult access eliminated gear that might achieve this objective (for example, traps, seines, electro-fishing). I tested an Oneida type trap net in Dingleberry Lake, but it was ineffective. Angling was slow and injurious. Gillnets were efficient, but most fish were injured or killed by the gear and not suitable for release for a mark-recapture experiment. Gillnets were suitable for the removal experiments. I fished four or six nets (Table 4.1) for approximately 12 hours as one unit of effort. The nets were picked and reset at dawn and dusk during the duration of the experiments, except Flower Lake, where gillnets were set from 54 a) Flower b) Wonder 3 c) Fishgut 1 d) Dingleberry e) Hell Diver 3 f) Hell Diver 2 g) Par Value h) Gem 2 A N 1 net 100 meters Figure 4.1 Relative size, depth, and shape for the eight experimental lakes in the Sierra Nevada, California. In panel g (Par Value) the contour interval is 3 m. In all other panels the contour interval is 1 m. Scales for 100 m and the length of one gillnet (36 m) are provided in the legend. Bathymetric maps showing inlets and outlets and morphometric measurements for each of the lakes are also presented in Appendix A . 55 Table 4.1 C a t c h by net t ing per iod and total catch for the eight exper imental lakes i n 1989. In the last two columns, "nets" were the number of gillnets fished dur ing each experiment and "nights" was the number of consecutive nights that the nets were fished. net t ing per iod lake 1 2 3 4 5 6 7 8 9 10 11 to ta l nets nights F lower 270 192 198 137 124 921 6 5 Wonder 3 145 110 80 68 68 39 37 32 37 616 6 5 Fishgut 1 227 56 75 55 50 463 6 3 Dingleberry 225 143 129 71 101 47 64 65 68 913 6 5 H e l l D ive r 3 45 9 14 68 4 2 H e l l D ive r 2 103 46 35 18 23 8 233 4 3 P a r Value 236 68 113 83 117 61 51 51 78 26 60 944 6 6 G e m 2 9 1 1 0 0 11 4 3 dusk to dawn. The nets were selected from a pool of ten nets that were ident ica l except i n the order of the different mesh-size panels (Table 2.3). E s t i m a t i o n o f N a n d q Otis et a l . (1978) developed a generalized removal model that relaxed the assumption of constant ca tchabi l i ty that had confined previous methods. Us ing m a x i m u m l ike l ihood , the models estimate i n i t i a l popula t ion size (N) and capture probabil i t ies for each capture occasion (<?,•; "p j" i n Ot is et a l . 1978). The est imation scheme starts w i th the two parameter constant catchabi l i ty mode l (N, q) where q = qx - q2 = , . . . ,<&. Nex t , the scheme evaluates the three parameter model (N, q\, q), where q\ is estimated separately from subsequent q. Th i s scheme continues es t imat ing occasion-specific un t i l the number of parameters is two less than the number of capture occasions. Schnute (1983) refined the procedures of Otis et a l . (1978) and developed a model so that the m a x i m u m l ike l ihood estimate for a vector of parameters Z was the same as the value of Z that min imized a function F. Schnute (1983) considered three models of capture probabil i t ies . In mode l 1, Z = (N,q), the model of constant catchabil i ty. In mode l 2, Z = (N,q\,q), a mode l of variable catchabi l i ty where the in i t i a l qi may differ from subsequent q. In model 3, Z = ( iV, qi, q, a) , where a is a constant between 0 and 1 that describes a rate of decline i n q. 56 Schnute (1983) found that his model 3 did not provide a significantly better fit to test removal data than model 2. The same result was true for removal data from the eight experimental populations in this study. In place of Schnute's model 3 I extended Schnute's model 2 to the four parameter variable q model Z = (N,q\,q2,q), referred to as model 3 in this study. Schnute (1983) included a statistical test for determining the model that best fit removal data. The test is based on the property that if F' is the minimum F for one model and F" is the minimum F for an extended model with n additional parameters, then 2(F' — F") is approximately x2 distributed with n degrees of freedom (df) when the first model is true (Schnute 1983). With a probability of type I error of 0.05, the critical value of the x2 distribution is 3.841 with 1 df. Thus, H2(F{ - F2") > 3.841, model 1 can be rejected in favor of model 2, and if 2(F£ — F^') > 3.841, model 2 can be rejected in favor of model 3. Otis et al. estimated confidence intervals for N by N ± 1.96 SE (N). Routledge (1989) criticized this method since, "Neither is the estimator close to being normally distributed, nor is the estimated standard error independent of the value of the estimator itself." Otis et al. (1978) and Schnute (1983) warned that by using the standard error of N to generate confidence intervals the lower limit may be less than the catch. Both studies recommended that the lower limit should be constrained so that it is not less than the catch. Schnute (1983) found that another problem can occur when the standard error of N is used to estimate an upper confidence limit when the minimum for the likelihood function is poorly defined. In this situation the upper confidence limit generated by the standard error of N may be low. For example, for model 2 for Par Value (Figure 4.3g), the upper confidence limit would be 1805 by the method of Otis et al. (1978) and 1937 by the method of Schnute (1983). To generate confidence intervals for N, Schnute (1983) used the property that the statis-tic 2[F(N) — F'] is x2 distributed with 1 df. For 95% confidence limits, this value will be less than 3.84. This requirement is equivalent to the condition F(N) < F' + 1.92. Combined with the constraint that N must be greater than the observed catch, these conditions define 95% confidence intervals for N. Routledge (1989) determined through a simulation study that this method of generating confidence intervals is reliable for removal estimators. 57 Otis et al. (1978) developed the software package C A P T U R E for analysis of removal and capture-recapture data (see also White et al. 1978, and White et al. 1982). C.J. Walters programmed the methods described by Schnute (1983) and I modified the program to include the four parameter model (Z = N ,qi,qi,q). I estimated the parameters N and qi, and con-fidence intervals for N with both programs. The values of N and qi were identical. I used confidence intervals from Schnute (1983) for reasons given above. Estimates of Survival at Age I estimated survival at age using catch at age data (catch curves, Ricker 1975). Catch at age data were available for the eight experimental lakes (Table 4.2) and twelve additional lakes with sufficient sample sizes for analysis of catch curves. I considered over 100 aged brook trout per lake to be sufficient for the estimation of survival at age. Al l eight experimental lakes and 10 of the 12 additional lakes had more than one year of catch at age data. Ricker (1975) recommended combining catch at age samples from successive years to reduce recruitment irregularities. After combining catch at age data from successive years most catch curves still displayed variation between ages that prohibited the calculation of age-specific estimates of survival directly (i.e., the difference between log 1 0(C a+i) and log 1 0 (C a ) must be negative to calculate survival). Ricker (1949) used moving averages to produce a smoothly declining catch curve. Another procedure is to fit piece-wise linear regressions to the apparent linear sections of the catch curve. Given the convex appearance of most of the brook trout catch curves (Figure 4.4), I rejected Ricker's (1949) method and the piece-wise linear regression approach in favor of the non-linear function l o g 1 0 ( $ > ) „ y _ 4 ^ _ ^ (4.1) X 1986 7 \ i / The right hand side of equation 4.1 is a convenient form of the logistic growth equation (C.J. Walters, personal communication), subtracted from the scaling factor Y. The subtraction inverts the sigmoid curve. The function describes the logarithm of cumulative catch frequency at age (Ca) using age (a) and four parameters Y, b, k, and m. The four estimated parameters do not have biological relevance. I used the function to smooth variations in the catch curves, 58 not to estimate biologically relevant parameters. The function parameters were estimated from the catch curve data using non-linear modeling procedures in SYSTAT (Wilkinson 1988). I estimated population specific survival at age by evaluating equation 4.1 to determine the expected logarithm of catch at age (Ca) for each lake. I calculated survival rates at age for each population from ga — i o g 1 0 ( c a + i ) - iog l 0 (c a ) (4.2) Estimation of Pre-Study Population Sizes The estimation of N and q described above produced estimates of population size for each lake at the beginning of the removal experiments in 1989. Al l eight experimental lakes had been sampled before 1989 (Table 2.1). To estimate the change in population size over the course of the study, I required estimates of pre-study population size. I estimated the number of fish at age for each lake at the beginning of the removal experiments in 1989 by A>a = Npa , (4.3) where pa is estimate of the proportion of fish at age a from the catch data (corrected for low vulnerability of age 1 fish; Table 4.2). I estimated the number of fish at age in year t by Na,t = ^ F ^ 1 + Ca . (4.4) Equation 4.4 is used recursively through 1986 for Hell Diver 2 and 1987 for the seven other experimental lakes. I compared these number at age and total population estimates to Na and N from age-specific removal data. For 1987-1988, I calculated abundance estimates using standard Leslie depletion methods (Ricker 1975). For 1989, I calculated total and age-specific abundance estimates using both the Leslie method and the Otis-Schnute methods described above. 59 Estimation of Fish Density I estimated fish density by number by dividing the estimate of population size for each lake by measures of surface area and volume. I estimated fish density by weight by converting the population estimate into a biomass estimate by max a lake biomass = ^ N0}t WGyt , (4.5) where Wajt is the mean weight at age a in year t from the catch data, and Naj are from equation 4.4. I divided the biomass estimate for each lake by measures of lake surface area and lake volume to produce estimates of biomass density. I enlarged lake perimeters and map scales from 1:24000 United States Geological Survey topographical maps using a binocular dissecting microscope at 12X with an attached 1.5X drawing tube. Lake surface areas were measured from the enlargements using a digitizing tablet and commercially available measurement software (Jandel Sigmascan). To estimate lake volume, I made 58-157 soundings (Appendix A) per lake using a hand-held depth finder (Fish Ray FR-100). An onshore observer estimated and recorded the position of each sounding on enlarged maps of the lakes. Contour intervals were fit by eye to the soundings, aided by drawings and photographs of the lakes. I estimated total lake volume by digitizing each contour interval and summing the volumes of the series of truncated cones (Wetzel 1975). Based on the difficulty of estimating position on water, and the subjective fitting of the contour intervals, estimates of lake volume are rough at best. Results Estimates of N and q The removal data for 1989 are listed in Table 4.1 and displayed in Figure 4.2. Population estimates and capture probabilities for the 1989 data are given in Table 4.3, and confidence limits for N in Table 4.4. Function values, minimum function values, and confidence limits for models 1, 2, and 3 are displayed in Figure 4.3. 60 Hell Diver 3, with only three netting periods, can only be fit by the constant q model. Of the seven other catch series, three were best fit by the constant q model and four by a variable q model. The catch series for Fishgut 1 and Par Value were best fit by the four parameter model (N, qi, qi, q). I fit the five parameter model (iV, qi, qi, q^, q) to the catch series for Par Value but it was not significantly different from the four parameter model. Both the Fishgut 1 and Par Value catch series are characterized by the second catch being substantially lower than the first and third catches (Figure 4.2c and g). It appears that this was due to diurnal variation in q. The data in Figure 4.2 show that catchability was lower during the day. Estimated Survival The catch curves for the eight experimental lakes and twelve additional lakes are pre-sented in Figure 4.4. The points are log 1 0 (catch frequency) plotted against fish age. The curve is equation 4.1 fit to the data. Survival rates at age (Sa) are listed in Table 4.5. The mean survival rate at age Sa and 95% confidence limits for Sa are given in Table 4.5 and displayed in Figure 4.5. Estimated Density and Abundance At Age Estimated number of brook trout at age at the start of each removal experiment are listed in Table 4.6. Estimated biomass of brook trout at age for the eight experimental lakes are listed in Table 4.7. Densities in number of fish per hectare, per surface hectare above the 3 m depth contour, and per 10000 m 3 are listed in Table 4.8. Densities in biomass of fish per hectare, per surface hectare above the 3 m depth contour, and per 10000 m 3 are listed in Table 4.9. Comparison of Abundance Estimates I compared abundance estimates for age 1, age 2, and total population by three meth-ods; Leslie, Schnute, and the back-calculation of numbers at age using equation 4.4 (virtual population analysis (VPA) estimates in Table 4.6). The estimates for all three methods are given in Table 4.10. 61 Table 4.2 Catch at age for the eight experimental lakes, 1986 or 1987 through 1989. The second to last column is the number of age 1 fish corrected for differential vulnerability by dividing by the relative vulnerability for each year (Table 2.6). The total includes the number of corrected age 1 fish. age corrected 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 age 1 total Flower 1987 28 126 103 89 92 56 18 19 6 6 8 3 50 576 1988 155 112 113 55 37 34 47 14 12 3 1 2 1 218 649 1989 237 177 53 38 31 19 23 11 4 5 1 3 320 685 Wonder 3 1987 23 138 66 31 28 107 34 33 23 17 12 6 2 2 2 41 542 1988 1 106 39 28 85 46 8 39 25 19 14 10 4 1 149 468 1989 27 392 58 16 28 15 14 3 26 8 6 7 8 3 2 530 751 Fishgut 1 1987 6 48 98 26 38 16 9 1 3 1 1 0 11 252 1988 36 32 116 100 27 29 22 7 6 7 1 0 51 398 1989 13 215 81 12 33 46 15 20 9 3 8 5 1 0 291 537 Dingleberry 1987 23 225 120 37 83 53 25 32 23 20 1 2 1 41 663 1988 2 96 251 266 103 20 54 26 22 17 16 17 3 1 1 135 934 1989 8 409 56 81 55 17 6 15 7 7 4 2 5 1 553 817 Hell Diver 3 1987 8 2 2 9 3 5 15 7 12 10 2 1 14 82 . 1988 6 25 4 3 1 3 1 3 11 7 2 2 8 70 1989 22 22 6 2 2 2 3 1 2 4 1 30 75 Hell Diver 2 1986 4 1 1 3 28 13 13 9 23 17 13 8 3 7 139 1987 12 1 5 4 4 8 8 2 3 4 1 2 54 1988 1 40 27 2 3 6 4 1 4 5 2 2 1 97 1989 152 1 52 17 1 1 1 4 3 205 285 Par Value 1987 7 82 89 36 9 23 101 26 70 54 86 37 6 2 146 692 1988 10 369 82 35 25 16 7 57 7 23 34 50 23 17 2 1 520 909 1989 406 302 27 12 9 2 4 28 14 15 16 27 9 549 1014 Gem 2 1987 1 40 10 77 2 71 161 1988 1 38 27 1 66 1989 1 4 6 11 62 Table 4.3 Estimated initial population size (JV), estimated capture probabilities (g), and minimum function values (JF") for three removal models and eight experimental lakes. The parameter estimates for the best fit model are in bold type. Model selection is explained in the text. * — previous model rejected with a P < 0.05; ** — previous model rejected with P < 0.001. model 1 model 2 model 3 lake JV Q F' JV 9 F' JV §i q JF" Flower 1489 .175 6.80 1608 .168 .153 6.32 1331 .203 .181 .222 4.61 Wonder 3 735 .183 7.88 755 .192 .168 7.03 778 .186 .174 .154 6.50 Fishgut 1 529 .339 32.68 1132 .201 .073 6.07** 645 .352 .134 .204 4.94* Dingleberry 1152 .160 28.84 1285 .175 .122 20.33** 1388 .162 .123 .103 19.06 Hell Diver 3 75 .546 6.62 Hell Diver 2 247 .376 7.01 257 .401 .308 4.98* 254 .406 .304 .329 4.89 Par Value 1194 .132 68.29 1523 .155 .077 36.97** 1311 .180 .063 .106 28.51* Gem 2 11 .786 0.67 11 .818 .667 0.52 11 .818 .500 .999 0.00 Table 4.4 Approximate 95% confidence intervals for the estimated initial population size (JV) for three removal models and eight experimental lakes. The best fit is listed in bold type. model 1 model 2 model 3 lake JV lower upper N lower upper N lower upper Flower 1489 1310 1784 1608 1326 2287 1331 1146 1820 Wonder 3 735 694 794 755 702 840 778 708 915 Fishgut 1 529 502 569 1132 667 oo 645 534 1549 Dingleberry 1152 1083 1248 1285 1158 1502 1388 1193 1827 Hell Diver 3 75 68 92 Hell Diver 2 247 238 262 257 241 292 254 239 302 Par Value 1194 1123 1292 1523 1311 1937 1311 1177 1528 Gem 2 11 11 12 11 11 14 11 11 12 63 300 200 a) Flower o CO o 250 200 -O "cO O CO o 200 o "CO o c) Fishgut 1 0 1 2 3 e) Hell Diver 3 150 100 b) Wonder 3 d) Dingleberry 0 2 4 6 f) Hell Diver 2 netting period netting period Figure 4.2 Catch by netting period for eight experimental brook trout populations. The data are from removal experiments conducted in 1989. Each netting period represents a capture occasion, with four to six nets fishing for about 12 hours. Filled circles (•) represent removals made primarily at night; open circles (o) are daytime removals. The data are presented nu-merically in Table 4.1. 64 Figure 4.3 Separation statistic F(N) for model 1 (F\), model 2 (F2), and model 3 (F3). Most curves have three tick marks. The middle tick mark corresponds to the estimate JV, at the minimum value of F. The left and right tick marks correspond to the lower and upper confidence limits. The function is evaluated over a range of N from the observed total catch to an arbitrary value greater than the upper confidence limit. 65 Figure 4.4 The logarithm of catch frequency at age plotted against age for 20 lake popula-tions of brook trout. The first eight panels (a-h) are data from the experimental lakes. The remaining twelve panels are data from additional lakes sampled during 1986-1989. Al l eight of the experimental lakes and nine of the twelve additional lakes (exceptions: panels j , 1, and o) had catch from more than one year of sampling. Catch at age for two or more years of sampling was summed before taking logarithms. The curve fit to each data set is from the function described by equation 4.1. Each curve was used to estimate the annual survival rates at age listed in Table 4.5. 66 Table 4.5 Brook trout survival at age for the eight experimental populations and 12 additional lake populations sampled between 1986-1989. Methods for estimating survival at age are described in the text. age lake 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Flower .87 .81 .75 .71 .67 .64 .61 .59 .57 .55 .54 .53 .52 Wonder 3 .96 .92 .88 .84 .80 .76 .73 .69 .66 .64 .61 .59 .57 .56 Fishgut 1 .86 .80 .74 .70 .66 .63 .61 .59 .57 .56 .55 Dingleberry .90 .84 .79 .74 .70 .66 .63 .60 .57 .55 .54 .52 .51 .50 Hell Diver 3 1.00 .99 .99 .97 .95 .93 .90 .86 .82 .78 .73 Hell Diver 2 1.00 1.00 1.00 .99 .98 .96 .93 .88 .81 .72 .62 .50 .39 Par Value 1.00 1.00 1.00 1.00 .99 .98 .96 .92 .85 .74 .59 .42 .26 .14 Gem 2 .99 .97 .88 .70 .44 .20 Matlock .90 .84 .79 .74 .69 .65 .62 .59 .56 .54 Gable 3 .95 .92 .90 .87 .84 .81 .79 .76 .74 .72 .70 .68 .66 Wonder 2 .87 .82 .78 .75 .72 .69 .67 .66 .64 .63 .63 Inconsolable .95 .91 .85 .79 .74 .68 .63 .58 .54 .50 .46 Fishgut 2 .76 .74 .72 .71 .70 .70 .70 .70 .70 .70 .71 .71 Fishgut 3 .99 .98 .96 .92 .88 .82 .76 .69 .61 .54 .47 .41 Midnight .85 .81 .79 .77 .75 .74 .73 .73 .72 .72 .72 .72 Hell Diver 1 1.00 .99 .98 .97 .95 .93 .90 .86 .83 .78 .74 .69 .64 Bottleneck .95 .91 .86 .82 .77 .73 .69 .65 .61 .58 .55 .53 .51 .49 Pass .95 .91 .86 .82 .78 .73 .69 .66 .63 .60 .57 .55 Gem 1 .79 .75 .73 .71 .70 .69 .68 .68 .68 Gem 3 .84 .80 .77 .73 .71 .68 .67 .65 mean .92 .89 .85 .82 .77 .73 .73 .70 .67 .64 .61 .57 .51 .47 ower 95% CI .89 .85 .81 .77 .71 .66 .68 .65 .63 .59 .56 .50 .40 .22 pper 95% CI .96 .93 .90 .87 .84 .81 .79 .76 .72 .69 .66 .64 .62 .72 67 1.0 0.0 10 15 age Figure 4.5 Mean survival rate at age plotted against age for 20 lake populations of brook trout. The vertical bars represent 95% confidence limits for the mean. The data are from the last three rows of Table 4.5. The slight discontinuity that occurs between ages 6 and 7 is caused by below average survival rates at ages 4-6 from Gem 2 (Table 4.5, Figure 4.4h). 68 Table 4.6 Estimated number of fish at age for the eight experimental lakes, 1986 or 1987 through 1989. Estimation methods are explained in the text. age 1' 2 3 4 5 6 7 8 9 10 11 12 13 14 15 total Flower 1987 342 400 301 229 262 191 66 71 16 28 8 7 2 1923 1988 659 254 222 150 99 114 86 29 30 6 12 2 1 1664 1989 698 384 115 82 67 42 51 24 9 10 1 6 1489 Wonder 3 1987 99 202 182 106 43 204 84 73 60 53 27 14 2 2 2 1153 1988 208 56 59 102 63 12 74 37 28 24 23 9 5 700 1989 545 57 15 27 15 14 3 26 8 6 7 8 3 2 735 Fishgut 1 1987 69 260 339 103 138 79 31 40 33 7 1 1101 1988 164 50 169 179 54 67 40 14 23 17 3 780 1989 365 97 14 39 55 18 24 11 4 10 6 1 645 Dingleberry 1987 485 669 297 81 210 120 91 79 59 77 11 2 3 2 2186 1988 233 402 375 139 33 89 44 41 28 21 31 5 1 1 1444 1989 886 89 127 86 27 9 23 12 12 6 3 8 1 1285 Hell Diver 3 1987 45 8 6 10 6 10 20 20 26 12 5 1 168 1988 30 31 6 5 1 5 4 4 13 12 3 2 118 1989 30 22 6 2 2 2 3 1 2 4 1 75 Hell Diver 2 1986 59 4 2 12 39 21 22 30 41 25 27 10 11 303 1987 87 55 3 1 9 11 8 9 19 15 6 8 1 2 235 1988 2 87 42 2 1 4 7 4 1 9 10 2 2 173 1989 184 1 47 15 1 1 1 4 3 257 Par Value 1987 263 140 73 28 35 198 54 120 128 235 123 46 10 7 1460 1988 911 117 51 37 19 12 95 27 46 62 110 51 17 2 1 1558 1989 707 391 35 16 12 3 5 37 18 20 21 35 12 1311 Gem 2 1987 113 10 135 2 261 1988 2 42 41 85 1989 1 4 6 11 69 Table 4.7 Estimated biomass (kg) offish at age for the eight experimental lakes, 1986 or 1987 through 1989. Estimation methods are explained in the text. age 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 total Flower 1987 3.6 15.0 17.3 16.1 22.7 17.0 5.6 6.3 1.5 2.6 0.7 0.9 109.3 1988 8.5 9.9 13.5 10.4 7.9 8.9 7.0 3.2 2.3 0.4 1.4 0.2 0.1 73.6 1989 13.8 19.4 8.3 6.6 5.7 3.7 4.9 2.2 0.8 0.9 0.1 0.4 66.9 Wonder 3 1987 0.7 5.6 8.2 5.8 2.7 15.6 6.6 5.6 4.7 4.2 1.9 1.0 0. .1 0.1 0.1 62.9 1988 3.4 2.3 3.1 6.4 3.9 0.9 6.1 3.1 2.2 1.8 1.8 0.7 0. .3 36.0 1989 12.1 3.0 1.1 2.2 1.2 1.3 0.2 2.4 0.8 0.5 0.5 0.7 0. .2 0.2 26.3 Fishgut 1 1987 0.5 6.9 13.0 4.8 7.1 4.4 1.5 2.3 1.6 0.3 0.1 42.5 1988 1.7 1.5 7.7 9.0 2.8 3.5 2.3 0.8 1.2 0.8 0. .1 31.5 1989 8.9 4.8 0.8 2.9 3.9 1.2 1.8 0.7 0.3 0.7 0. ,4 0.1 26.4 ;leberry 1987 4.1 27.9 22.1 6.4 17.9 10.4 7.7 7.0 5.2 7.5 0. .9 0.2 0.3 0.0 117.5 1988 4.0 19.3 28.3 12.0 2.7 7.9 4.1 3.8 2.5 2.0 2 .8 0.5 0.1 0.1 90.0 1989 23.9 6.4 11.9 9.3 3.3 1.2 2.7 1.4 1.2 0.8 0, .3 0.8 0.1 63.2 Hell Diver 3 1987 1.9 0.2 0.4 0.7 0.9 1.0 2.7 2.6 3.4 1.6 0.7 0.1 16.3 1988 0.7 1.8 0.5 0.5 0.1 0.7 0.7 0.6 2.1 1.7 0.5 0.3 10.2 1989 0.9 1.9 0.7 0.2 0.3 0.4 0.5 0.2 0.3 0.5 0.1 6.2 Hell Diver 2 1986 0.3 0.0 0.1 0.6 2.9 1.8 2.0 2.7 3.8 2.3 2.7 0.9 1.1 21.2 1987 0.4 1.0 0.2 0.1 0.7 1.0 0.8 0.8 1.8 1.4 0.6 0.7 0.1 0.0 9.5 1988 0.0 3.4 3.2 0.2 0.0 0.5 0.9 0.5 0.1 1.0 1.2 0.3 0.3 11.6 1989 2.2 0.0 3.5 1.6 0.1 0.1 0.1 0.4 0.3 8.4 Par Value 1987 2.8 7.0 6.0 2.5 3.5 20.0 5.5 12.5 13.3 24.1 13. .1 5.2 1. .0 0.0 116.6 1988 19.1 6.5 3.8 3.5 1.9 1.2 9.8 2.6 4.9 6.4 11. .5 5.3 1. .8 0.2 0.1 78.6 1989 11.2 20.6 2.8 1.5 1.2 0.3 0.5 4.1 2.0 2.2 2 .2 3.7 1. .1 53.3 Gem 2 1987 2.3 0.2 10.2 0.2 13.0 1988 3.0 4.8 7.8 1989 0.0 0.7 1.4 2.2 70 Table 4.8 Brook trout density in the eight experimental lakes calculated as number per surface hectare, per surface hectare above the 3 m depth contour (Ab3), and per 10000 m 3 . The change in density over the course of the experimental manipulation of the populations is listed in column 6, and the percent decrease in density over the course of the experiments is listed in the last column. percent 1986 1987 1988 1989 decrease decrease per hectare . . . Flower 1028 890 796 232 22.6 Wonder 3 885 538 564 321 36.3 Fishgut 1 1713 1214 1004 710 41.4 Dingleberry 1052 695 619 434 41.2 Hell Diver 3 190 134 85 105 55.4 Hell Diver 2 746 579 426 633 133 15.2 Par Value 607 647 545 62 10.2 Gem 2 370 121 16 355 95.8 per Ab3 . . . Flower 1028 890 796 232 22.6 Wonder 3 2304 1399 1469 835 36.3 Fishgut 1 2082 1475 1220 862 41.4 Dingleberry 1284 848 755 529 41.2 Hell Diver 3 677 475 302 375 55.4 Hell Diver 2 1544 1198 882 1310 234 15.2 Par Value 2945 3142 2644 301 10.2 Gem 2 439 143 19 421 95.8 10000 m 3 . . . Flower 831 719 643 188 22.6 Wonder 3 250 152 160 91 36.3 Fishgut 1 984 697 576 407 41.4 Dingleberry 580 383 341 239 41.2 Hell Diver 3 29 21 13 16 55.4 Hell Diver 2 264 205 151 224 40 15.2 Par Value 81 86 73 8 10.2 Gem 2 202 66 9 194 95.8 71 Table 4.9 Brook trout density in the eight experimental lakes calculated as kilograms per surface hectare, per surface hectare above the 3 m depth contour (Ab3), and per 10000 m 3 . The change in density over the course of the experimental manipulation of the populations is listed in column 6, and the percent decrease in density over the course of the experiments is listed in the last column. 1986 1987 per hectare . . . Flower 58.4 Wonder 3 48.3 Fishgut 1 66.1 Dingleberry 56.6 Hell Diver 3 18.5 Hell Diver 2 52.2 23.4 Par Value 48.5 Gem 2 18.4 per Ab3 . . . Flower 58.4 Wonder 3 125.7 Fishgut 1 80.4 Dingleberry 69.0 Hell Diver 3 65.7 Hell Diver 2 108.1 48.4 Par Value 235.2 Gem 2 21.9 per 10000 m 3 . . . Flower 47.2 Wonder 3 13.7 Fishgut 1 38.0 Dingleberry 31.2 Hell Diver 3 . 2.9 Hell Diver 2 18.5 8.3 Par Value 6.5 Gem 2 10.1 percent 1988 1989 decrease decrease 39.4 35.8 22.7 38.8 27.6 20.2 28.1 58.2 49.0 41.1 25.1 37.9 43.3 30.4 26.1 46.2 11.6 7.0 11.4 62.0 28.6 20.7 31.5 60.4 32.7 22.1 26.3 54.3 11.1 3.1 15.3 83.1 39.4 35.8 22.7 38.8 71.9 52.5 73.1 58.2 59.6 49.9 30.4 37.9 52.9 37.1 31.9 46.2 41.1 25.0 40.7 62.0 59.1 42.8 65.2 60.4 158.5 107.5 127.7 54.3 13.1 3.7 18.2 83.1 31.8 28.9 18.3 38.8 7.8 5.7 7.9 58.2 28.1 23.6 14.4 37.9 23.9 16.8 14.4 46.2 1.8 1.1 1.8 62.0 10.1 7.3 11.2 60.4 4.4 3.0 3.5 54.3 6.0 1.7 8.4 83.1 Table 4.10 Comparison of abundance estimates for age 1, age 2, and total population by three methods; Leslie, Schnute, and the back-calculation of numbers at age using equation 4.4 (VPA). Because the VPA method starts with the estimates from the Schnute method, the estimates of total population are the same for 1989. Schnute estimates are not available for 1987 and 1988. Other blanks represent data insufficient for the analyses. age 1 age 2 total population Leslie Schnute VPA Leslie Schnute VPA Leslie Schnute VPA Flower 1987 342 400 655 1923 1988 263 659 152 254 871 1664 1989 1169 698 284 356 384 1340 1489 <= Wonder 3 1987 99 202 345 1153 1988 179 208 43 56 542 700 1989 446 452 545 61 62 57 720 735 <= Fishgut 1 1987 64 69 56 260 311 1101 1988 162 164 36 50 445 780 1989 230 409 365 83 121 97 488 645 <= Dingleberry 1987 105 485 220 669 738 2186 1988 142 233 231 402 1005 1444 1989 707 836 883 88 97 89 1086 1285 <= Hell Diver 3 1987 10 45 0 8 102 168 1988 6 30 23 31 81 118 1989 28 24 30 21 22 72 75 <= Hell Diver 2 1987 87 20 55 56 235 1988 2 87 192 173 1989 177 172 184 1 1 240 257 Par Value 1987 80 263 77 140 596 1460 1988 355 911 94 117 883 1558 1989 545 671 707 370 504 391 1120 1311 Gem 2 1987 60 113 10 10 157 261 1988 2 13 42 71 85 1989 0 1 11 11 « $ = 73 In general, Leslie estimates were lower than the two other methods. Leslie estimates are known to be negatively biased (Otis et al. 1978, Schnute 1983). For the 1989 estimates of total population, the Leslie estimates averaged 9.6% lower than the Schnute estimates. The Leslie estimates are much worse in comparison with VPA estimates in 1987 and 1988 (61.1% and 28.0%, respectively), suggesting either the VPA estimates are too high because of survival rates that are too low, or the Leslie estimates have improved during the study because of standardizing the effort in 1989. The Leslie estimates for 1987 and 1988 are suspect, since most estimates are not much greater than the catch (Table 2.1 and corrected catch in Table 4.2), and catches of fully vulnerable ages in subsequent years revealed that there were many more fish present than estimated by the Leslie method. Given the tendency for the Leslie method to underestimate population size and my own observations that always showed there were more fish than I expected, I risk overestimating population abundance by using the VPA estimates in subsequent analyses. Discussion Estimates of N and q The results of the procedures for model selection described by Schnute (1983) and used in this study conform with my subjective observations. For example, the estimate of N from model 1 in Fishgut 1 (529, Table 4.3) was rejected in favor of model 2 (1132). I doubt that there were more than 1000 brook trout in Fishgut 1 at the beginning of the removal experiment in 1989. The best estimate from model 3 (645) is reasonable. For most lakes the differences between the estimates of N from the three models are small (Table 4.3). Though Otis et al. (1978) and Schnute (1983) relaxed the assumption of constant prob-ability of capture for all fish during a removal experiment, the extent to which this assumption was relaxed is still limited. For example, their models do not allow for behavioral or periodic changes in individual catchability. I discuss the violation of these assumptions in chapter 5. 74 Survival Estimation Ricker (1975) recommended combining samples of successive years to reduce irregular-ities from unstable recruitment. This procedure is valid for random recruitment variation. If recruitment variation is not random, combining successive years may accentuate the irregu-larities. An inspection of the logarithms of the combined catch frequencies plotted against age (Figure 4.4) suggests that recruitment may not be constant or randomly varying in the 20 lakes. This is especially apparent in the Hell Diver lakes (Figure 4.4e, f, p) and Par Value (Figure 4.4g). The pronounced dip in the data is likely due to the weak recruitment of sev-eral successive age groups. For most lakes, combining successive years of catch at age data did reduce apparent recruitment variation, though serial correlation may still be apparent (for example, Dingleberry in Figure 4.4d). For this study, I considered the combined catch at age data sufficient for estimating survival at age. I required survival at age estimates for back-calculating population sizes prior to large scale fish removals that marked the beginning of the study (1986 for Hell Diver 2 and 1987 for the seven other experimental lakes). Any bias introduced by combining the catch at age data to estimate survival at age should not markedly influence pre-study estimates of population size, since the period of back-calculation (1989 to 1987 or 1986; two or three years) is short relative to the life span of brook trout. The problems of non-random recruitment variation can be alleviated by estimating sur-vival within a year class (Ricker 1975). The decline in abundance at successive ages can be estimated with age specific removal data, or using catch per unit of effort at age as an index of abundance. However, with catch data from removal experiments, the change in abundance or index of abundance from age, to age t + 1 between years will be due primarily to the removals, masking any measurable change in non-removal mortality, and causing this technique to be inappropriate for removal data. There are several weaknesses in the survival rate at age estimates given in Table 4.5 and the means presented in Figure 4.5. The estimates for age 1 fish may be too high because of increasing vulnerability. Though the estimates are based on the number of age 1 fish corrected for the relative vulnerabilities by year (Table 2.6), I suspect that the estimated 75 relative vulnerabilities are too high, especially for age 1 fish in 1987. If relative vulnerabilities were actually lower, more age 1 fish would have been estimated, lowering the survival rate from age 1 to age 2. The high survival rates for ages 1-3 are also a result of the curve fitting method that I used to estimate Sa- Equation 4.4 fit poorly those lakes that had a pronounced dip in the catch curves (Figure 4.5e, f, g, h, and p). Estimates of Sa for these lakes were near or at 1.00 for ages 1-3. The curves obviously do not fit the data for these lakes. I considered estimating survival directly from the catch data or from estimates of abundance at age, but both these methods are influenced by the strong recruitment of age 1 fish in some lakes in 1988, and in most lakes in 1989, due to the removals (chapter 8). It is possible that the decrease in survival rate at older ages depicted in Figures 4.4 and 4.5 is a result of non-random sampling, persistent above average recruitment, or aging error. Catch curves will be convex if older age classes are under-represented in the catch. It is possible that older brook trout are less active or more wary than young brook trout, and therefore do not get sampled in proportion to their true abundance." Catch curves will be convex if recruitment was above average for successive cohorts in the middle of the range of ages sampled. For example, in Figure 4.4d, the catch curve would be nearly straight if not for the positive residuals for ages 9-12 (suggesting above average recruitment for the 1977-81 cohorts). I do not have data that would reject the hypotheses of non-random sampling or persistent above average recruitment as the cause of the convex curves in Figures 4.4 and 4.5. A third alternative explanation for the convex catch curves is aging error. For example, annuli in older brook trout may be imperceptible, not annual, or simply much harder to dif-ferentiate as the true age of the fish increases. This problem is weakly evident in Figure 3.2, for ages greater than 10. These errors would result in fewer old brook trout and more mid-dle aged brook trout, similar to the outcome from the hypothesis of persistent above average recruitment. The data to refute this explanation are limited. The age validation study pre-sented in chapter 3 only validated annuli through age 10. The data presented in Table 4.5 and Figures 4.4 and 4.5 suggest that survival at age decreases from age 1 onwards, and that the rate of decrease accelerates from age 11-14. I am confident of the survival at age estimates 76 over ages 4-10, but I cannot reject the hypothesis that aging error was at least a factor in the appearance of increasing mortality with age for older brook trout outside the range of validated ages. After casting doubt on the conclusion that survival rates decrease with age and admitting the possibility of aging error, it is necessary to point out that the analyses and conclusions presented in this study are robust to aging error and estimation error for most of the age specific survival rates. Cohort analysis can be sensitive to both the abundance estimate of the last age class and survival rates, especially if the starting point for estimating the abundance of each cohort is the oldest age class and the analysis is extended back in time for many years. In contrast, the cohort analysis that I conducted starts with estimates of abundance at each age in 1989 and extends back in time only two years for seven lakes and three years for Hell Diver 2. Thus the effects of possible aging error or non-random sampling producing biased estimates of survival at age was limited to a short interval, and for the segments of populations that were younger than age 4 and older than age 10. Estimated Density Comparable estimates of brook trout density in high alpine lakes are scarce. Cooper et al. (1988) used mark-recapture data to estimate densities of 88-646 brook trout per hectare (9.8-44.8 kg/hectare) in four alpine lakes on the western slope of the Sierra Nevada. Donald and Alger (1989) estimated 750 brook trout per hectare (30 kg/hectare) in Olive Lake, British Columbia, using mark-recapture data. For stunted brown trout, Pechlaner and Zaderer (1985) reported densities of 147 fish per hectare for fish above 144 mm in length. Langeland (1986) reported density in number of 1100 fish per hectare and density in weight of 71 kg per hectare for Arctic char age 2 and older. The estimates in this study ranged from 187-1771 per hectare (18.8-69.1 kg/hectare; Tables 4.8 and 4.9) for brook trout age 1 and older. The eight experi-mental lakes were at higher altitude and were likely affected by more variable environmental conditions than the lakes studied in Cooper et al. (1988) or Donald and Alger (1989). I used estimates of brook trout density for comparing catchability among lakes in chapter 5, and for comparing fecundity, growth, and recruitment among and within lakes in chapters 6-8. 77 5. Spatial and Temporal Changes in Catchability Estimates of population size (JV) and catchability (if) calculated in chapter 4 provide a means of comparing q versus JV and indices of density in the eight experimental lakes. I compared changes in catchability spatially by analyzing differences between the eight popula-tions. Lacking comparable data to analyze q and JV directly for removal experiments conducted between 1986 and 1988, I relied on trends in catch per effort (CPE) data to infer changes in catchability and population size. I examined temporal changes in CPE at two time scales: among years and within years. Paloheimo and Dickie (1964) predicted that catchability would vary inversely with popu-lation abundance and with area. Many studies have found or suspected an inverse relationship between catchability and population size within commercial and sport fished populations (Gul-land 1964; Garrod 1964, 1977; Schaaf and Huntsman 1972; Clark 1974; Pope and Garrod 1975; Radovich 1976; MacCall 1976; Ulltang 1976, 1980; Clark and Mangel 1979; Pope 1980; Schaaf 1980; Peterman and Steer 1981; Condrey 1984; Peterman et al. 1985; Winters and Wheeler 1985; Crecco and Savoy 1985; Crecco and Overholtz 1990). Winters and Wheeler (1985) also reported an inverse relationship between catchability and the area occupied by several herring stocks. There are two differences between these studies and the brook trout data that I present. The cited studies have examined the relationship between catchability and population size (or density) over time. The brook trout data presented here are from eight populations during the 1989 field season. The second difference is that the cited studies dealt with fish that school or are confined to rivers or inlets during migration. The general hypothesis is that as stock size decreases, the area searched by each unit of nominal fishing effort should represent a larger portion of the area occupied by the clumped fish population (Peterman and Steer 1981; in reference to Paloheimo and Dickie 1964). This is not the situation for the brook trout data. First, there is no indication that brook trout are clumped, and second, the gillnets are passive and are not actively fishing more of the lake. For passive fishing gear, the probability of capture depends on the fish encountering the net, not the net encountering the fish. 78 If catchability is inversely proportional to the area occupied by the fish, then smaller lakes should have higher q values regardless of density or abundance. If brook trout activity increases at low densities due to improved foraging conditions or increased territory size, q should increase more than expected by the differences in lake area. I found q was inversely proportional to four estimates of area occupied, but that estimates of q for two lakes were higher than would be expected from differences in area occupied alone. I formulated alternative hypotheses that also explain increases in q with decreasing N and density. I found that catch per effort (CPE) increased between years and between sampling trips within lakes. Temporal changes in CPE within lakes may be due to changes in brook trout behavior. Three hypotheses of brook trout behavior are consistent with the observed changes in CPE. The first hypothesis is that activity (foraging, territorial behavior) increased as population abundance and density decreased. The second hypothesis is that individual activity was periodic. The third hypothesis is that the brook trout populations have a hierarchical behavior structure (for example, dominance) that was disturbed by the removal experiments. A l l three hypotheses violate the assumptions of methods for estimating population size using removal data. Methods Differences In Catchability Among Lakes Catchability can be defined as the probability that an individual fish will be captured by a unit of effort, and, when applied to a population, is the proportion of the population captured by that unit of effort (Paloheimo and Dickie 1964). Conceptually, catchability is the area "swept" by the gear divided by the area occupied by the fish. Paloheimo and Dickie (1964) expressed catchability as q = ca/A, where c was gear efficiency, a was the area swept by a unit of gear, and A was the area occupied by the total population. For my study, I assume gear efficiency (c) was constant because the nets were identical. For set gillnets, the area swept by a unit of gear depends on the effective area of a net (also a constant), the swimming speed 79 of the fish (u), and the number of nets fished (77). For this study, catchability can thus be expressed as « = ^ (5.1) where A is the area of the lake occupied by the fish. If the product cvr) is constant, then q should be inversely proportional to lake area. A constant swimming speed v implies constant behavior. If q is not inversely proportional to A, the variation should be due to (1) changes in swimming speed (i.e., activity), or (2) the number of nets, or (3) the area occupied relative to the total measured lake area. Variation in the number of nets fished can be eliminated by expressing catchability as q per net, ! = T <5-2> 77 A to isolate variation in v. To test the hypothesis that q is inversely proportional to the area occupied by the fish, I plotted q versus four measures of lake size (estimates of area occupied); surface area in hectares, surface area above the 3 m depth contour in hectares, lake volume in 10000 m 3 , and lake perimeter in meters. To test the hypothesis that swimming speed (i.e., fish activity) increases at lower densities or abundances, I compared plots of q versus N and q versus indices of density to the q versus area occupied plots. To isolate the variation in v, I plotted q/n versus four estimates of area occupied (A) and fit equation 5.2 using six of the experimental lakes (Flower, Wonder 3, Fishgut 1, Dingleberry, Hell Diver 2, and Par Value). I fit the model using estimates of q from Table 4.3 divided by the number of nets fished (77) from Table 4.1 versus the four measures of A. The parameter v was estimated using non-linear techniques in SYSTAT (Wilkinson 1988). I fit the model with only six of the experimental lakes to isolate suspected changes in v in Gem 2 and Hell Diver 3. Temporal Changes in Catch Per Effort I assumed catch per effort (CPE) was proportional to the available population by the relationship C P E t = qNt, where C P E t is the catch per unit effort during the interval t, and Nt is the mean population surviving during time interval t (Pucker 1975). If the catchability 80 coefficient q is constant throughout a removal experiment, both q and the original population size N can be estimated by the linear regression C P E t = qN — qKt, where Kt is the cumulative catch to the start of interval t plus half the catch taken during the interval (Leslie and Davis 1939; as described by Ricker 1975). Because the assumption of a constant q was usually violated for brook trout populations in this study, estimates of q and JV by this method are generally biased (q too high, N too low). In chapter 4 I presented estimates of N and q based on catch data with constant effort from removal experiments conducted in 1989. Effort in 1986-1988 varied during most of the removal experiments. Since comparable estimates of N and q are not available from the 1986-1988 removal experiments, I relied on trends in the CPE versus cumulative catch data presented in Figure 5.4 to illustrate changes in CPE over time. Cumulative catch in Figure 5.4 is at the end of each netting period. The definition of Kt used by Ricker (1975) is not used in Figure 5.4, because the data are not used for estimating N and q, and the size of the catch for each netting period is easier to see when presented as in Figure 5.4. I measured catch per effort as catch per net hour, with both catch and the number of hours summed for all nets during a netting period. Though q was not constant, the relationship C P E ( = qNt is still helpful in assessing changes in CPE between years. The assumption that CPE is positively related to abundance appears valid, since CPE did decrease as fish were removed from the population (Figure 5.4). By accepting the general form of the relationship without regard to the constancy of q, the expectation is that an increase in CPE requires an increase in q, an increase in iV, or both. 81 Results Differences In Catchability Among Lakes Estimates of catchability (if) are plotted against estimates of population size (iV) for the eight experimental lakes for both the constant q model and the variable q model in Figure 5.1a and b. The estimates of N and q are from Table 4.3. Figures 5.1c and d show a negative, linear relationship between if and log 1 0(jV) for both models. Estimates of catchability (if) are plotted against four measures of lake size (estimates of the area occupied by fish) and four indices of fish density in Figure 5.2. As predicted, if varied inversely with the different measures of area occupied by the fish (Figure 5.2a, c, e, and g). In general, the differences observed between q and density (Figure 5.2b, d, f, h) can be explained by differences in lake size as predicted by equation 5.1 with constant behavior (constant swimming speed v). In the plots of if versus measures of area occupied (Figures 5.2a, c, e, and g), Gem 2 and Hell Diver 3 stand out in three of the four panels as somewhat separate from the relationship between if and A for the other six lakes. This is especially apparent in Figure 5.2a, where the relationship between q and A appears linear for the other six lakes. From data presented in chapter 7, brook trout were larger in Gem 2 and Hell Diver 3 than fish in the other six lakes, suggesting that either large fish are more active (larger v) or more easily entangled than small fish. Some of the variation in the relationship between q and lake area occupied (Figure 5.2a, c, e, and g) was due to differences in the number of nets fished between lakes (rj in equation 5.1). Catchability per net (q/rj) versus the four estimates of area occupied is presented in Figure 5.3. The curve in each panel of Figure 5.3 represents the model qjr\ = cv/A (equation 5.2). If v is constant between lakes, all estimates of q/rj versus A should lie near the curve. The curve was fit using six of the experimental lakes (Flower, Wonder 3, Fishgut 1, Dingleberry, Hell Diver 2, and Par Value) to test the hypothesis that qjr\ was higher than expected for Gem 2 and Hell Diver 3. The result depended on the estimate of A. For two estimates of A (surface area in hectares, Figure 5.3a, and lake perimeter in meters, Figure 5.3d), estimates of q/rj are 82 constant q model variable q model 400 800 1200 1600 300 600 900 1200 1500 1800 N N 1.0 1.5 2.0 2.5 30 35 10 15 2.0 2.5 3.0 35 log10(N) log 1 0 (N) Figure 5.1 Estimated capture probability (q) versus estimated initial population size (JV) and log10(JV). Panels a and c are from the constant q model. Panels b and d are from the variable q model, where the initial q\ may vary from the subsequent q. The solid circles (•) represent qi and in panels b and d the open triangles (A) represent q. The data are from 1989 removal experiments. The sample size is eight brook trout populations for panels a and c and seven populations for panels b and d. 83 Figure 5.2 (follows) Estimated capture probability (q) versus four measures of area occupied and four indices offish density: a) surface area of lakes in hectares; b) fish per 100 m 2 surface area; c) surface area of lakes above the 3 m depth contour, in hectares; d) fish per 100 m 2 surface area, for surface waters above the 3 m depth contour; e) lake volume (10000 m 3 ); f) fish per 100 m 3 lake volume; g) lake perimeter in meters; and h) fish per meter of lake shoreline (perimeter). The data are from the eight removal experiments in 1989. Densities were calculated using the best estimates of population size (iV) from the Schnute method (Table 4.3). The estimates of catchability are the q from model 1 in Table 4.3. The points are labelled to show how the relationship between q, area occupied, and density can change depending on the measure of area occupied. Fl—Flower, W3—Wonder 3, Fg—Fishgut 1, Db—Dingleberry, H3—Hell Diver 3, H2—Hell Diver 2, PV—Par Value, and G2—Gem 2. 84 area occupied density 1.0 1.0 a) b) 0.8 G2 . 0.8 - . G2 0.6 K J . q 0.6 F C . 0.4 H2. Fg. 0.4 02 • " ° PV. 0 2 & . D b : B 0.0 I I I I I I 0.0 I I I I I I I 0.0 0.5 1.0 1.5 2.0 2.5 surface area (hectares) 0 2 4 6 8 1 0 12 fish per 1 0 0 m 2 surface area 1.0 1.0 c) d) 0 8 G2 . 0 8 - . G2 0.6 H3. q 0.6 M3. 0.4 H2. . Fg. 0 4 Fg. 0.2 w 3 - nh • Fl PV. • D b 0.2 L 0 - . PV 0.0 I I I I I 0 0 I l l l 0 0 0 5 1.0 1.5 2.0 0 10 2 0 3 0 surface area above 3 m (hectares) fish per 1 0 0 m 2 above 3 m 1.0 1.0 e) f) 0.8 - G 2 . 0 8 - . G2 0.6 F C . < cr 0 6 . KO 0 4 - H2. Fg. 0.4 H2. . F g 0 2 0 2 " W 3 ' Cb. -Fl 0.0 I I I I I 0 0 i i i i i i i i 0 5 10 15 2 0 lake volume ( 1 0 0 0 0 m 3 ) 0 1 2 3 4 5 6 7 fish per 1 0 0 m 3 volume 1.0 1.0 g) h) 0 8 G2 . 0 8 - . G2 0 6 H3. q 0.6 H3. 0.4 H2. Fg. 0.4 H2. . F g 0.2 • ^ . PV 0.2 • w 3 Db • Fl . P V 0 0 i . i . i i o.o I I I I I I 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 0 0 O S 1.0 1 5 2 0 2 S lake perimeter (m) fish per meter of shoreline Figure 5.2 Estimated capture probability (q) versus four measures of area occupied and four indices offish density. (Full caption on previous page.) S5 <B ai5 CD CL 0.10 - \ (32. a) - H3. - H2i, -1 1 1 i 0.0 0.5 1.0 1.5 2.0 surface area (hectares) 0.25 0.20 0.10 0.06 0.00 - ; G2. b) - H2; -0.0 0.5 1 0 1.5 2.0 surface area above 3 m (hectares) ® 0.15 CD CL - (32. c) • H3. - hd. _ '">•' Fi.'-D^3. • . P V 0.25 0.15 0.10 0.05 - G2. d) -H3. " - . .H2 . " - • - . . F g . lake volume (10000 rrr 200 300 400 500 600 lake perimeter (m) Figure 5.3 Catchability per net (5/77) versus four estimates of lake area occupied by brook trout in the eight experimental lakes. The line represents the model 17/77 = cv/A (equa-tion 5.2) fit to six of the experimental lakes (Fl—Flower, W3—Wonder 3, Fg—Fishgut 1, Db—Dingleberry, H2—Hell Diver 2, and PV—Par Value), where q is the estimate of catcha-bility from the Schnute model 1 (Table 4.3), A is the estimate of area occupied for the four different measures of lake size, 77 is the number of nets fished (Table 4.1), and v represents swimming speed (i.e., activity). Differences in catchability per net between the six lakes used to fit the model and the two other experimental lakes (H3—-Hell Diver 3 and G2—Gem 2) are discussed in the text. S6 much higher than expected due to the change in area occupied alone. However, where A was estimated as surface area above the 3 m depth contour (Figure 5.3b), the estimate of q/rj for Hell Diver 3 fits the model well, though Gem 2 is noticeably different. Where A was estimated as lake volume (Figure 5.3c), Gem 2 fits the model well and Hell Diver 3 is noticeably different. Gem 2 is shallow and small, whereas Hell Diver 3 is deep and small. The data in Figure 5.3 suggest that catchability may have been higher in Gem 2 and Hell Diver 3, but without actual measurements of the amount of area occupied by the fish the data are inconclusive. The plots of if versus estimates offish density (Figure 5.2b, d, f, h) reflect the inadequacy of standard measures of density for ultra-oligotrophic, high alpine lakes. For example, in the plot of q versus fish per 100 m 3 lake volume (Figure 5.2f), the point nearest the origin is from Par Value. Par Value was the largest and deepest of the eight experimental lakes (Table 2.2, Appendix Figure A7). The low density per volume for Par Value does not realistically reflect the area of the lake occupied by brook trout. Brook trout seemed more abundant near shore than in the middle of the large, deep lakes that I sampled. Other measures of lake area that reflect few fish in the pelagic zone may provide more accurate estimates of effective density. When compared using fish per 100 m 2 above the 3 m depth contour (Figure 5.2d), or fish per meter of shoreline (Figure 5.2h), Par Value had one of the most dense populations, reflected in the low q. Temporal Changes in Catch Per Effort Change In Catchability Between Years The changes in catch per effort between years show a consistent pattern of increase (Figure 5.4). In every case, except Gem 2 from 1988-89, CPE increased between the last netting period in year t and the first netting period in year t + 1. Most of the increase can be attributed to increasing abundance between years due to recruitment. Figure 5.5 shows the age distributions from the eight experimental lakes. The age distributions shifted markedly toward a predominance of age 1 and age 2 fish during the study (Figure 5.5). 87 The effects of recruitment can be removed from Figure 5.4 by following an age group throughout all removal experiments. Figure 5.6 shows the CPE data for age 1+ in 1986 (Hell Diver 2, Figure 5.6 f), age 2+ in 1987, age 3+ in 1988, and age 4+ in 1989. The data for 1988 and 1989 show that older brook trout were substantially depleted in all eight populations. Curiosities remain in the CPE data presented in Figure 5.6, even after the effect of recruitment has been removed. In 13 of the 17 between-year comparisons, CPE increased be-tween the last netting period in year t and the first netting period in year t +1. I expected CPE to decline between years, given the reduction in abundance of the age group due to removals and natural mortality. The increases in CPE indicate increases in q, N, or both, between years. Population abundance (N) may have increased in some lakes through immigration, or by "new" fish becoming vulnerable to the gillnets from within each lake. Catchability (q) may have increased due to increased activity between years. Change In Catchability Within Years Another way to reduce the effect of recruitment is to measure change in catchability within years. I conducted five delayed removal experiments: Wonder 3 in 1987 and 1988, and Dingleberry, Par Value, and Gem 2 in 1988. The procedure was to start a removal experiment with several netting periods during the first fishing, then stop netting for several days to several weeks, then recommence the removal experiment for one to several netting periods. The data are presented in Figures 5.4 and 5.6 (CPE data from the second fishing period are represented by open (A) or filled (A) triangles). The largest increase in CPE between fishing periods was in Wonder 3 in 1987 (Fig-ure 5.4b). During the first fishing (June 27-30), six netting periods reduced the CPE from 3.30 to 0.88. Returning to Wonder 3 two months later, the CPE was 5.59 for the first netting period of the second fishing (August 28-30), 60% higher than the initial CPE on June 27. Catch per effort for the second and third netting periods during the second fishing were as high or higher than any CPE for netting periods 2-5 during the first fishing. It is unlikely that the increase in CPE was due to immigration. There are no fish in the drainage above Wonder 3, and access from below is impeded by several cascades. Figure 5.6b indicates that 88 the increase was not due to age 1 brook trout becoming vulnerable to the gillnets, a possibility that I described in the section on gillnet selectivity in chapter 2. Since N decreased by 275 fish during the first fishing, the increase in CPE is either the result of increased activity or an increase in fish becoming vulnerable to the gillnets from within the lake. There was no increase in CPE in Wonder 3 in 1988 between fishing periods (Figures 5.4b, 5.6b). The net-free interval between fishing periods was only five days. The data look as if the removal experiment was not interrupted. One explanation for the lack of an increase in CPE is that there is a delay before fish activity increases. For Dingleberry in 1988 (Figure 5.4d), the CPE increased from 0.45 to 1.40 after a net-free interval of eight days. Removals during the first fishing totaled 827 brook trout. The CPE of 1.40 from the second fishing was equivalent to the CPE from the second and third netting periods during the first fishing (1.36 and 1.47). The cumulative catch at the end of the second and third netting periods was 321 and 431 brook trout. If CPE were proportional to N with constant q, then the CPE observed during the second fishing (1.40) would have required an additional 827 - 431 = 396 to 827 - 321 = 506 fish to match CPE levels observed during the first fishing. It is possible that 400-500 fish may have immigrated, as Dingleberry was the only lake with substantial stream resident populations in both the inlet and outlet streams. It is also possible that the increase may have been met by age 1 brook trout becoming vulnerable to the gillnets; not through growth, but by juveniles venturing further into a lake with fewer predaceous adults (Johnson 1976). Age composition data for the second fishing do not support the latter explanation; only 6 of the 77 fish captured were age 0 or age 1. The data in Figure 5.6d indicate that the increase in C PE was due to an increase in abundance offish age 34-. One possible explanation for the increased CPE is that individual catchability may have increased, such that apparent population abundance may have increased by "new" fish becoming vulnerable to the gillnets from within Dingleberry. Catch per effort increased in Par Value in 1988 from 0.54 to 1.22 after a net-free interval of 15 days (Figure 5.4g). The increase is similar to the increase in Dingleberry in 1988 described in the preceding paragraph, except that Par Value is isolated from any source of immigration. 89 If the increase in CPE between fishing periods was due solely to an increase in N, it would have required an additional 400-450 fish becoming vulnerable to the gillnets during the 15 days between fishing periods. Age composition data suggest that it may have been age 0 and age 1 fish becoming vulnerable to the gillnets during the last fishing period, as the age 0 and age 1 age classes made up 65% of the 79 fish caught in the second fishing, compared to 46% of the 75 fish captured in the last two netting periods of the first fishing. The data in Figure 5.6g show that CPE increased only slightly due to fish age 3+, supporting the explanation that either vulnerability or activity increased in brook trout age 2 and younger. The last delayed removal experiment was conducted in Gem 2 in 1988 (Figure 5.4h). Catch per effort was identical (0.70) between the last netting period of the first fishing and the only netting period of the second fishing, with a net-free interval of 58 days. Catch per effort should decrease for successive removals, not increase or remain the same. Gem 2 had no inlet or outlet during the net-free interval in 1988, so no immigration occurred. The age composition data show that only age 2 and age 5 brook trout were captured on the second fishing, the same two age classes that dominated catch (96%) during the first fishing. Either q increased or N increased from within Gem 2 to account for the stability in CPE between fishing periods. 90 a) Flower o o 1987 1988 1989 o o o O • • • o o ° • • • o ° ° Q> • "° "° . • • 0 100 200 300 400 500 600 700 0 100 200 300 "»00 500 300 700 900 0 200 400 600 800 1000 b) Wonder 3 -1987 1988 1989 • ° % o • • O 0 • O * o . o • o • 0 100 200 300 400 500 000 c) Fishgut 1 0 100 200 300 400 500 0 100 200 300 400 500 BOO 700 1987 1988 1989 • « • • • 0 0 • o °° . o o 9 0 100 200 300 0 100 200 300 400 0 100 200 300 400 500 d) Dingleberry 1987 • 1988 1989 o • • o o m Q • ° o o * o • 0 . o °* . 0 • • 0 . o • o 0 100 200 300 400 500 BOO 700 0 200 400 800 800 1000 0 200 400 800 900 1000 cumulative catch Figure 5.4 (continues) Catch per net hour (CPE) versus cumulative catch for four lakes. 91 e) Hell Diver 3 LU CL O ' 1987 1988 1989 • O • - • • o • • 1 o o • I I I 1 O I I I I I . • 0 20 40 60 80 100 0 .0 20 30 40 50 80 70 0 10 20 30 40 50 60 70 f) Hell Diver 2 0 50 100 150 0 10 20 30 40 60 60 70 0 20 40 60 90 100 0 60 100 160 200 250 g) Par Value U J * CL • o 3 • 1987 1988 1989 - O • • - • 0 o • 4 • • • - A O O * o o o * ° • —I— 1 1 1 1 1 1 i , i ° « o • o 0 100 200 300 400 500 600 700 0 100 200 300 400 600 600 700 800 0 200 400 600 800 1000 h) Gem 2 1987 1988 1989 L U C L O 0 50 100 150 0 10 20 30 40 50 60 70 0 5 10 15 cumulative catch Figure 5.4 (continued) Catch per net hour (CPE) versus cumulative catch for four exper-imental lakes. Triangles are netting periods from the second fishing of a delayed removal experiment. Filled symbols (•, A ) represent removals made primarily at night; open symbols (o, A) are daytime removals. 92 a) Flower o CD 3 CD JZ O O 0 1 2 3 4 6 6 7 8 9 10 11 12 13 14 16 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 6 6 7 8 9 10 11 12 13 14 15 b) Wonder 3 >-o c CD 3 cr CD o CO o 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 a 9 10 11 12 13 14 15 0 1 2 3 4 5 8 7 8 9 10 1  12 13 14 15 c) Fishgut 1 o c CD 3 C J CD JZ o CD O 0 1 2 3 4 5 3 7 8" 9 10 II 12 13 14 15 0 1 2 3 4 5 8 7 8 9 10 1  12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 1  12 13 14 15 d) Dingleberry o c CD 3 CJ CD CQ O 0 1 2 3 4 5 8 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 8 7 8 6 10 t l 12 13 14 15 0 1 2 3 4 5 8 7 a 9 10 11 12 13 14 15 age Figure 5.5 (continues) Catch frequency at age for four of the experimental lakes. 93 e) Hell Diver 3 o c CD CJ CD o to o 1987 1988 "h-n-r 1989 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 6 6 7 8 9 10 1» 12 13 14 15 0 1 2 3 4 5 8 7 a 9 10 11 12 13 14 15 f) Hell Diver 2 0 1 2 3 4 6 6 7 a 9 10 II 12 13 14 16 0 1 2 3 4 6 fl 7 a 9 10 1  12 13 14 16 0 1 2 3 4 6 9 7 a 8 10 1  12 13 14 15 0 1 2 3 4 6 6 7 8 9 101112 13 1416 g) Par Value 1987 >-o c CD n CJ CD JC O CO CJ 1988 1989 0 1 2 3 4 6 6 7 8 8 10 11 12 13 14 15 0 1 2 3 4 6 6 7 a 8 10 1  12 13 14 15 0 1 2 3 4 5 6 7 S 9 10 1 1 12 13 14 15 80 70 o 60 -CD D 50 -CJ CD 40 • r~ 30 -O ~C0 20 -O 10 -0 h) Gem 2 1987 1988 1989 _ Q Q. 0 1 2 3 4 5 8 7 8 9 10 1 1 12 13 14 15 0 1 2 3 4 5 8 7 B 8 10 1  12 13 14 15 0 1 2 3 4 5 8 7 8 8 10 1  12 13 14 15 age Figure 5.5 (continued) Catch frequency at age for four of the experimental lakes. The catch of age 1 and in some lakes age 2 brook trout increased considerably in 1989, presumably due to an increase in juvenile survival. The hatched portion of the age 1 bars is the corrected number of age 1 fish based on relative vulnerabilities (Table 2.6 and 4.2). The clear portion of the bars is the actual catch. 94 a) Flower U J CL 3 o 1987 2+ 1988 3+ 1989 4+ o a o • • o CO o • V o » • • • ° ° o * 9 • 0 100 200 300 400 600 600 0 100 200 300 400 0 60 100 150 200 b) Wonder 3 1987 2+ 1988 3+ 1989 4+ • 0 •„ o . A • °° . * O ; • o 0 100 200 300 400 500 600 0 100 200 300 0 60 100 150 c) Fishgut 1 1987 2+ • 1988 3+ 1989 4+ o O • o • ° oo ' • o • o • 0 60 100 150 200 260 0 100 200 300 400 0 50 100 160 d) Dingleberry 1987 2+ 1988 3+ 1989 4+ • o 0 • •o o • • • • * o o • . o * o . ° . • O 0 100 20O 300 400 600 600 700 0 100 200 300 400 500 600 0 50 100 150 200 cumulative catch Figure 5.6 (continues) Catch per net hour (CPE) versus cumulative catch for four lakes. 95 e) Hell Diver 3 LU CL O 1987 2+ 1988 3+ 1989 4+ OB 0.6 • O 0.4 02 o • • O • • • * 0.0 <? 0 10 20 30 40 50 60 ro 0 10 20 30 40 0 5 10 15 20 f) Hell Diver 2 • 1986 1+ 1987 2+ 1988 3+ 1989 4+ -o • • o • o • o * 0 50 100 150 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 g) Par Value 6 1987 2+ o 1988 3+ 1989 4+ 5 -4 * 3 -2 1 o • • o * 0 « • , ° o ,* o • o • * • 0 n # o • o •n. •, 0 100 200 300 400 500 600 0 100 200 300 0 60 100 150 h) Gem 2 1987 2+ 1988 3+ 1989 4+ 2 1 • o O 0 ° • - 8* • O A. • t 0 20 40 60 80 1O0 0 10 20 30 cumulative catch 0 1 2 3 4 5 6 7 Figure 5.6 (continued) Catch per net hour (CPE) versus cumulative catch for four exper-imental lakes. The data are adjusted to include only those fish that were age 1+ in 1986, age 24- in 1987, age 3+ in 1988, and age 4+ in 19S9. Triangles are netting periods from the second fishing of a delayed removal experiment. Filled symbols (•, A) represent removals made primarily at night; open symbols (o, A) are daytime removals. 96 Discussion Differences In Catchability Among Lakes The first impression from the relationships between catchability, population abundance and population density is that q increases at low N and density (Figures 5.1 and 5.2). The data of q versus area occupied in panels a, c, e, and f of Figure 5.2 show that much of the relationship between q, N, and density can be explained by the inverse proportionality between q and A. However, the data for Gem 2 and Hell Diver 3 in Figures 5.2a, e, g, and 5.3 are not entirely explained by differences in area occupied. Data in chapter 7 show that brook trout from both these lakes were larger than brook trout from the other six lakes, suggesting the influence of other factors to explain the relationship between catchability and area occupied. I present several alternative hypotheses that also explain increasing q at low abundance and low density. Though I cannot reject the null hypothesis that the increase in q was due to the differences in area between the experimental lakes, it is still instructive to explore the alternative hypotheses. Increasing catchability can also be explained by changes in brook trout behavior at low population size or low density that lead to increased activity. Individual brook trout may be more active at low population size and by that encounter the gillnets more frequently than individuals living in high density populations. One explanation is that when density is low, food is more abundant, providing incentive to increase foraging activity, and hence increase the probability of encountering a gillnet. I did not test this hypothesis, though there was supportive circumstantial evidence. There was a significant relationship between mean size and density (Figure 7.10), suggesting that brook trout may feed more often in low density populations. This raises a confounding factor: large brook trout may have a higher probability of capture than small brook trout, independent of feeding activity. Encounter rates with gillnets may be similar between small and large fish, but large fish may be more likely to become entangled upon encounter, perhaps due to body shape, swimming speed, momentum, and that it is easier for a larger fish to become entangled in small mesh than a small fish to become entangled in large mesh. I do not have independent size data that are sufficient to test this hypothesis over 97 as wide a range of fish size as encountered between populations represented by the extremes in N and density in Figures 5.1 and 5.2. Another explanation for increased activity is that when density is low, territory size increases, increasing the mobility of the fish and thereby increasing the probability of encountering a gillnet. I do not have any data on brook trout territory size. There are other explanations of the data in Figures 5.1-5.3 that are independent of the increased activity hypothesis. The alternative explanation of large fish having a higher probability of capture has already been stated, though this explanation in not necessarily independent of the increased activity hypothesis. A second alternative explanation is gillnet saturation: in lakes with few fish there is less chance that gillnet saturation will influence catch rates, and hence ultimately N. I reject this alternative for the same three reasons that I stated in the section on gillnet saturation in chapter 2 (catch did decline rapidly during the first few netting occasions, the relationship between catch per set and set duration was slightly positive (Figure 2.6), and observations of entanglement patterns are contrary to a repulsion effect). A fourth reason for rejecting this explanation is that if net saturation were a major factor in the relationship between q and N, the expectation would be for q to increase as decreased during a removal experiment. The data do not support this expectation. Catchability tends to decline during a removal experiment, presumably because fish with high probabilities of capture were removed first (Otis et al. 1978; White et al. 1982; Schnute 1983). A third alternative explanation is that brook trout learn to avoid nets in lakes with high density. This explanation is similar to the gillnet saturation hypothesis in that fish avoid entanglement by being repulsed by fish already entangled. It differs from the saturation hypothesis in that the fish that are repulsed learn to avoid the gillnets during subsequent capture occasions. The distinction is important, because this learning hypothesis is consistent with the patterns of catch during a removal experiment (Figure 4.2), whereas the saturation hypothesis is not consistent with these catch patterns. In lakes with low population size, there would not be enough fish to have both an entangled group and a learning group. As implausible as this explanation may sound, I do not have data that refutes this hypothesis directly. The 98 data on temporal changes in C P E show that when brook trout are fished for several days, then left unfished for one week to two months, and then fished again, that upon the second exposure to fishing catch per unit effort is as high or higher than the first fishing. Simply put, if the brook trout do learn to avoid nets during the first few days of netting, then they forget the lesson in about a week. In contrast, a behavior hypothesis is consistent with both the data presented in Figure 5.1-5.3 and temporal changes in C P E . There are doubtless other explanations for the observed inverse relationship between q, N, and density. After rejecting three explanations, I am left with explaining the observed differences in q for Gem 2 and Hell Diver 3 by either sampling variation in the q = CVTJ/A relationship or real differences in v because of increased activity or increased probability of entanglement. Size data suggest that brook trout in lakes with decreased fish abundance are larger on average than brook trout in lakes with high abundance (chapter 7). This result provides some support to the hypothesis that brook trout in lakes of low fish abundance forage more actively than brook trout in lakes of high fish abundance. Temporal Changes in Catch Per Effort Implications for Removal Estimates of Population Abundance Both of the explanations for the observed increase in C P E during the delayed removal experiments have serious consequences for methods of estimating population abundance from removal data. Estimation methods that allow catchability to vary (Otis et al. 1978; Schnute 1983) accommodate variation in q by assuming that catchability can vary between individu-als in the population, but that probability of capture for any individual is constant (White et al. 1982). The explanation of individual fish activity increasing (i.e., "immigration" from within a lake) violates assumptions of the removal estimation methods. The explanation vio-lates the assumption of constant individual catchability, and the concept of immigration from within a lake violates the fundamental assumption of population closure. I am not aware of any method that can estimate population size from removal data when individual behavior fluctuates periodically. 99 The estimate of population abundance (N) is valid only for that portion of the population that is vulnerable to the gear. White et al. (1982; their Figure 1.3) provided an example of a closed population separated into catchable and non-catchable groups. In their example, the non-catchable group is permanently non-catchable (very large fish in a lake, old and wary coyotes). If the non-catchable group is temporary and periodically catchable, as may happen with the brook trout populations, the total population size will be unknown, unless additional information on temporal changes in catchability is collected. Without additional information, estimates of population abundance from removal estimation methods such as Otis et al. (1978) and Schnute (1983) are biased, and the total population size will be greater than the estimate ofN. Behavioral Hypotheses Regarding Changes in Catchability Other than the data presented in Figures 5.4 and 5.6, I did not collect observations on temporal changes in catchability. The data presented in Figures 5.4 and 5.6 have led me to speculate on population processes that are consistent with the observed catch data. I present three hypotheses of brook trout behavior. The first hypothesis is that activity increases as abundance and density are reduced. I base this hypothesis on data presented in Figures 5.1-5.4 and 5.6, and on discussion pertaining to those figures. To reiterate one explanation for this hypothesis: as density is reduced, feeding opportunities may increase and provide incentive for increased foraging, increasing the prob-ability of capture. The data presented in Figure 5.4 indicate that activity does not increase immediately, else CPE would remain high throughout a removal experiment as increased ac-tivity compensated for decreased density. A time lag is consistent with the explanation of increased feeding opportunity, for example if there is a delay in increased prey abundance. The second hypothesis is that brook trout feed periodically, and during the interval between feeding periods the fish are stationary. I base this hypothesis on the data in Figure 5.4, on results in Shepard (1970), and on discussions with C.J. Walters and T .M. Jenkins. The data in Figure 5.4 and explanations developed in this section suggest that temporal increases in catchability may be explained by part of the population being temporarily invulnerable to 100 the fishing gear. Shepard (1970) found that cutthroat trout Oncorhynchus clarki in a small lake moved little and maintained home ranges for up to five months. Independent observations of brook trout feeding behavior by C.J. Walters and T .M. Jenkins (personal communications) suggest that individual fish may occupy an area of less than 1 m 2 on the lake bottom for several days or weeks before actively foraging in other parts of a lake. The idea of sedentary brook trout with periodic bursts of feeding activity was contrary to my notion of brook trout that are constantly foraging. The third hypothesis is that a feeding or territorial hierarchy exists within a lake popula-tion of brook trout. The first netting period removes those fish at the top of the hierarchy with the highest probability of capture (for example, brook trout actively foraging). Subsequent netting occasions remove fish lower in the hierarchy as they gradually fill the top feeding or territorial positions. If the removal experiment is delayed during a field season or between field seasons, the delay allows the hierarchy to fill. This hypothesis is distinct from the hypothesis of periodic feeding, in which individual fish alternate periods of activity and inactivity. For the hierarchical hypothesis, individual fish are active only when the opportunity for moving up in the hierarchy arises. This hypothesis is based on speculation and one observation of structured behavior. On August 27, 1986, I observed a sedentary group of 14 brook trout arranged two or three abreast in a well ordered column, presumably positioned over an underwater spring in Emerald 4. As I retrieved a spinning lure over the group, one fish would rise to the lure, then return to its position in the column. I repeated the cast four or five times, and each time a different fish would rise and then return to its position in the column. After witnessing this display of apparent social order, I would not be surprised if a feeding or territorial hierarchy existed within the entire population. A l l three hypotheses are consistent with personal observations and reports from sport fishermen that show angling success decreases to nil during and immediately after a removal experiment. Actively feeding brook trout seem completely absent from the lake. Al l three hypotheses would predict a time lag before brook trout activity increased after a removal experiment. 101 I can think of two experimental designs that would test the proposed hypotheses. A basic problem is that the removal data are dependent on the behavior of the brook trout to become entangled in the gillnets. The first study would supplement removal data using passive gillnets with an active fishing gear such as seining or electrofishing. This would test the hypotheses of periodic feeding and hierarchical population structure by actively removing the dormant segment of the population. To test all three hypotheses, the second design would require behavior data on individual fish. Data could be collected either through extensive visual observations or tagging methods such as radio or sonic tags that allow remote monitoring of fish position. 102 6. Fecundity Response to Density Reduction To test the hypothesis that fecundity should increase with decreasing population density, I collected ovary samples during the removal experiments in the experimental lakes, and during survey sampling in non-experimental lakes. Fecundity did increase because of decreased pop-ulation density, but the increase can be explained by increased fish size. I found that stunted brook trout populations in the Sierra Nevada have similar fecundities to brook trout of simi-lar size from non-stunted populations measured in other studies, suggesting that fecundity is primarily size limited even where extreme food shortage has resulted in obvious stunting of growth. For several lake populations, mean fecundity tended to decrease with increasing age after a peak at age 2-5. One possible explanation for this pattern would be selective mortality of more fecund fish. However, I found that as age increased and fecundity declined, egg size appeared to increase to maintain constant ovary weight at age; thus, selective mortality need not be considered as an explanation. Methods I collected about 25 stomach and gonad samples per lake in 1986 and 1987. The number of fecundity samples was the number of mature females in the sample (Table 6.1). The two exceptions were Hell Diver 2 in 1986 and Wonder 3 in 1987. These lakes were part of separate experiments in which I collected additional ovary samples. I increased the sampling effort for ovaries in 1988 and 1989 to measure changes in fecundity due to the removal experiments that began in 1987. Ovary sampling in 1988 and 1989 was opportunistic. The number of samples taken from each lake was limited by the logistics of transporting the samples out of the mountains; besides other gear, about 100-200 ovary samples could be carried by two people. The number of ovaries sampled also depended on the amount of time available for ovary sampling and the number of mature females in the catch. Ovaries were fixed in 5% formalin and transported to the lab in Whirlpaks identified by lake and sample number. In the laboratory I transferred the ovaries to 70% ethanol and glass 103 Table 6.1 Sample sizes for fecundity samples, 1986-1989. The eight experimental lakes are listed in bold type. The experimental lakes are grouped with nearby non-experimental lakes that were sampled for comparative purposes. Other lakes from which fecundity samples were taken are listed in the last seven rows of the table. 1986 1987 1988 1989 total Flower 134 157 291 Matlock 57 15 72 Wonder 3 62 183 175 420 Wonder 2 76 68 144 Fishgut 1 185 124 309 Fishgut 2 13 75 22 110 Fishgut 3 4 78 64 146 Fishgut pond 6 6 Dingleberry 17 11 335 155 518 Midnight 71 71 Hell Diver 3 6 33 29 68 Hell Diver 2 41 5 34 51 131 Hell Diver 1 6 69 31 106 Par Value 10 6 97 196 309 Gem 2 4 34 3 45 Gem 1 7 13 20 40 Gem 3 18 11 - 48 77 Bottleneck 13 4 80 24 121 Pass 9 14 107 130 Barney 22 22 Inconsolable 16 11 27 Maul 3 4 24 31 Schober 1 6 5 11 Schober 3 1 1 total 136 132 1528 1406 3202 104 Table 6.2 Sampling dates for fecundity samples, 1986-1989. The eight experimental lakes are listed in bold type. The experimental lakes are grouped with nearby non-experimental lakes that were sampled for comparitive purposes. Other lakes from which fecundity samples were taken are listed in the last seven rows of the table. 1986 1987 1988 1989 Flower Jun 13-19 Jul 11-16 Matlock Jun 19-20 Jul 15 Wonder 3 Aug 28-30 Aug 11-19 Aug 23-28 Wonder 2 Aug 16-17 Aug 25-27 Fishgut 1 Jul 2-6 Aug 19-22 Fishgut 2 Aug 30 Jul 6-7 Aug 21 Fishgut 3 Aug 29 Jul 7-8 Aug 18-19 Fishgut pond Jul 18 Dingleberry Aug 27 Jul 20-25 Jul 16-22 Aug 6-11 Midnight Aug 11-12 Hell Diver 3 Jul 25-28 Jul 25-28 Aug 13-15 Hell Diver 2 Aug 23-28 Jul 26-28 Jul 25-29 Aug 12-15 Hell Diver 1 Aug 28 Jul 26-29 Aug 13-15 Par Value Jul 30 Aug 17-22 Aug 1-6 Jul 23-29 Gem 2 Aug 24-27 Jun 10-12 Jul 19-22 Gem 1 Jul 14-15 Aug 9-10 Jul 21 Gem 3 Jul 12-13 Aug 9-10 Jul 20-21 Bottleneck Aug 29 Jul 31-Aug 5 Jul 10-14 Aug 16 Pass Aug 13-16 Jun 26-30 Jul 30-Aug 1 Barney Aug 1 Inconsolable Jul 3-6 Aug 15 Maul Jul 7-8 Jun 22 Jul 9-10 Schober 1 Aug 31 Jul 13 Schober 3 Aug 31 105 vials. Each ovary was blotted and weighed individually to the nearest 0.01 g. I calculated ovary weight as the sum of the ovary pair. Maturing eggs and atretic eggs (Vladykov 1956) were separated by teasing the ovary apart and counting individual eggs. I denned absolute fecundity as the number of maturing eggs. Recruitment stock eggs (Vladykov 1956) were not counted. Large unattached eggs, apparently unshed from the previous spawning, were more prevalent in females from lakes with the worst crowding and stunting. These eggs were in various stages of resorption and I included them in the count of atretic eggs. I calculated mean egg weight in mg by dividing the ovary weight in g by the number of eggs and multiplying by 1000. This measure was biased by including the weight of recruitment stock eggs and ovarian tissue, but it was sufficient for my purposes. Results Interannual Changes Between and Within Lakes Mean ovary weight, mean egg weight, mean fecundity, mean length, and mean weight are plotted against age for the eight experimental lakes and nearby non-experimental lakes in Figures 6.1-6.8. The data are presented numerically in Appendix B. The fecundity data are highly variable, both within and among populations. Large variation in fecundity is typical of fish populations (Bagenal 1978). The annual differences in ovary weight and egg weight (panels a, b, c, and d in Fig-ures 6.1-6.8) may be a result of differences in sampling dates between years. For exam-ple, in Fishgut 1 the increase in ovary weight and egg weight between 1988 and 1989 (Fig-ure 6.3a and c) was likely caused by further egg development at the later sampling date in 1989 (Table 6.2). Ideally, fecundity sampling should take place just before spawning in October-December, but this was not possible for the lakes in this study. The differences in fecundity between years may also be due to seasonal changes in egg development and atresia. Vladykov (1956) reasoned that there was not enough space in the body cavity for all maturing eggs present early in the summer to enlarge as spawning time 106 1.5 -1.0 CO > o 0.5 c co CD E 0.0 5 CD , 4 jrz co CD 3 CD 2 CD CD C 1 CO CD 0 1000 >^  800 ""a c 600 o _CD c 400 CO CD e 200 0 "§ 250 E JH 200 CD c 150 © 100 c~ 50 CD CD g 0 a) c) e) Flower (experimental) '1988 • «1989 Matlock •1988 • -1989 b) d) 5 10 age 15 o 10 15 age Figure 6.1 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean fork length (mm), and mean weight (g) versus age for an experimental lake (Flower) and a nearby non-experimental lake (Matlock), in 1988 and 1989. The vertical bars are ± one standard error. Points without error bars have a sample size of one at that age. 107 6 5 4 3 2 1 0 12 10 8 6 4 2 0 6 0 0 5 0 0 4 0 0 3 0 0 2 0 0 100 0 2 5 0 2 0 0 150 100 5 0 0 Wonder 3 (experimental) 1987 * ' 1 9 8 8 • - 1 9 8 9 Wonder 2 •1988 • - 1 9 8 9 g) b) h) 5 10 age 15 o 10 age Figure 6.2 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean fork length (mm), and mean weight (g) versus age for an experimental lake (Wonder 3) and an adjacent non-experimental lake (Wonder 2), in 1988 and 1989. The vertical bars are ± one standard error. Points without error bars have a sample size of one at that age. Mean fecun-dity at age tended to decrease for both lakes and all years of the study (panels e and f), while mean length at age was constant (panels g and h). 108 Fishgut 1 (experimental) ' 1 9 8 8 • - 1 9 8 9 Fishgut 3 1988 • « 1 9 8 9 O) JZ co CD <= CO > o c co CD 15 CD E JZ 10 CD 'CD CD CD 5 CD C CO CD E 0 1 0 0 0 >> 8 0 0 c z> 6 0 0 o ci> c 4 0 0 CO CD E 2 0 0 0 'e 2 5 0 E 2 0 0 JZ CD c 1 5 0 1 0 0 JZ 5 0 CD CD 5 0 a) e) 9) b) d) h) 5 10 age 15 0 10 age Figure 6.3 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean fork length (mm), and mean weight (g) versus age for an experimental lake (Fishgut 1) and a nearby non-experimental lake (Fishgut 3), in 1988 and 1989. The vertical bars are ± one stan-dard error. Points without error bars have a sample size of one at that age. Mean fecundity at age tended to decrease slightly for both lakes in all years of the study (panels e and f), while mean length at age was constant (panels g and h). 109 Dingleberry (experimental) ' 1 9 8 8 • « 1 9 8 9 Midnight • ' 1 9 8 9 CD CD CD CO > o c co CD c CD E CD "CD CD CD CD C co 0 E c o CD CO CD c CD C CD CD CD 5 7 6 5 4 3 2 1 0 2 5 2 0 15 10 5 0 6 0 0 5 0 0 4 0 0 3 0 0 2 0 0 100 0 2 5 0 2 0 0 150 100 5 0 0 a) c) e) 9) •••--•••-f....,...+....t.>-b) d) h) 5 10 age 15 0 5 10 age Figure 6.4 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean fork length (mm), and mean weight (g) versus age for an experimental lake (Dingleberry) and a nearby non-experimental lake (Midnight), in 1988 and 1989. The vertical bars are ± one standard error. Points without error bars have a sample size of one at that age. Mean fecundity at age tended to decrease for Dingleberry in both years (panel e), while mean length at age was constant (panel g). 110 Hell Diver 3 (experimental) 1987 * ' 1 9 8 8 • "1989 Hell Diver 1 1986 ' ' 1 9 8 8 • ' 1 9 8 9 10 -4 -0 15 10 -0 1 5 0 0 1 0 0 0 5 0 0 0 E 2 5 0 E 2 0 0 J Z O) c 1 5 0 w @ 1 0 0 JZ 5 0 CD 'CD <= 0 c) g) b) d) h) 10 15 0 10 15 age age Figure 6.5 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean fork length (mm), and mean weight (g) versus age for an experimental lake (Hell Diver 3) and a nearby non-experimental lake (Hell Diver 1), in 19S6, 1987, 1988, and 1989. The vertical bars are ± one standard error. Points without error bars have a sample size of one at that age. Mean fecundity at age tended to decrease for Hell Diver 1 in 1988 and 1989 (panel f), while mean length at age was relatively constant (panel h). I l l c 3 o CD c CO CD E E co c CD CO JZ CD Hell Diver 2 (experimental) 1986 * ' 1 9 8 8 • « 1 9 8 9 Hell Diver 1 1986 ' ' 1 9 8 8 -1989 3) 10 -ght 8 -wei 6 ->> ovar 4 -mean 2 -0 -E^ 15 -ight 10 -wei CO CO CD 5 -mean 0 -1000 300 600 400 200 0 250 200 150 100 50 0 C) e) g) b) ^ • d) h) 5 10 age 15 o 5 10 age 15 Figure 6.6 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean fork length (mm), and mean weight (g) versus age for an experimental lake (Hell Diver 2) and a nearby non-experimental lake (Hell Diver 1), in 1986, 1988, and 1989. The vertical bars are ± one standard error. Points without error bars have a sample size of one at that age. Mean fecundity at age tended to decrease for Hell Diver 1 in 1988 and 1989 (panel f), while mean length at age was relatively constant (panel h). 112 Par Value (experimental) ' 1 9 8 6 • ' 1 9 8 7 5 -20 -~ 15 10 1000 >- 800 5 600 ^ 400 CO CD E 200 0 •c 250 E ^ 200 "5 g 1 5 0 @ 1 0 0 6 50 CD 5 o a) c) e) g) Par Value (experimental) ' 1 9 8 8 • - 1 9 8 9 b) d) 10 15 0 10 1 5 age age Figure 6.7 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean fork length (mm), and mean weight (g) versus age for an experimental lake (Par Value) in 1986, 1987, 1988, and 1989. The vertical bars are ± one standard error. Points without error bars have a sample size of one at that age. Par Value does not have a comparable lake nearby. 113 Gem 2 (experimental) Gem 3 • - - 1987 • '1988 • ' 1 9 8 9 ' - - 1 9 8 6 - ' 1 9 8 8 • ' 1 9 8 9 age . age Figure 6.8 Mean ovary weight (g), mean egg weight (mg), mean fecundity, mean fork length (mm), and mean weight (g) versus age for an experimental lake (Gem 2) and an adjacent non-experimental lake (Gem 3), in 1986, 1987, 1988, and 1989. The vertical bars are ± one standard error. Points without error bars have a sample size of one at that age. Gem 2 showed a significant increase in fecundity at age and length at age as population density was reduced between 1987 and 1989, while Gem 3 showed little change in either measurement during the study. 114 approached, and therefore some eggs must be absorbed. Vladykov (1956) estimated that " . . . only 56-61% of maturing eggs (Class b) eventually ripen. This reduction of nearly half of the maturing eggs occurs during 3 or 4 summer months." Henderson (1963) disputed Vladykov's interpretations, suggesting that Vladykov had ignored growth over the summer in reasoning that some eggs must be absorbed. Henderson (1963) found only 3.1-4.7% atresia when she included an adjustment for length. Wydoski and Cooper (1966) found a rate of atresia higher than 50% in natural stream populations of brook trout, and suggested that the low rates of atresia that Henderson (1963) reported were due to conducting experiments with hatchery reared brook trout. Scott (1962) experimented with rainbow trout under different feeding regimes and found that the extent of atresia was related to the degree of starvation during the maturation period. The fecundity data that I collected are not sufficient to comment on rates of atresia. The interannual changes in fecundity may be due to the amount of atresia that had occurred by the sampling date, or to real differences in the number of eggs maturing in that year. The data are insufficient to make the necessary interannual comparisons. Lacking fecundity data collected at the same stage of egg development, questions of interannual differences cannot be answered. Change in Fecundity with Size The data are sufficient to analyze fecundity changes given increases in the size of brook trout between years. The greatest increases in size due to density reduction are from a transfer experiment that I conducted in 1986-1987, and from the removal experiment in Gem 2 in 1987-1989. I described the transfer experiment briefly in chapter 2. To reiterate, 16 stunted brook trout were captured and moved from Fishgut 2 and Fishgut 3 into nearby 0.4 hectare Fishgut pond that was Ashless and literally crawling with invertebrate prey. In 1987, eleven months after the transfer, I recaptured 8 of the 16 brook trout. Mean fecundity versus mean length at age data are plotted in Figure 6.9. The data from Gem 2 in 1989 are identified with open circles (G) for the two ages at which fecundity data were available in 1989. The data from the 1986-1987 transfer experiment are identified 115 with open triangles (A) for the four ages at which mean fecundity and mean length data were available. Most of the data in Figure 6.9 are clustered about mean lengths of 180-220 mm and mean fecundities of 200-500. The points that are above this range are data from lakes with non-stunted brook trout populations in the Sierra Nevada. The data from the transfer experiment (A) and from the removal experiment in Gem 2 (0) fall within the range of data for non-stunted populations, suggesting that fecundity is primarily a function of size. When I reduced the population density in Gem 2 by removals, and transferred fish into a low density environment, fecundity did increase but no more than would be expected by the increase in size. Comparison of the Fecundity of Stunted and Non-Stunted Brook Trout Additional information regarding the relationship between fecundity and size is pro-vided by comparing stunted brook trout in the Sierra Nevada with non-stunted populations from other studies. Rounsefell (1957) in his Figure 1 and Table 1 presented data from four studies of brook trout fecundity (Hayford and Embody 1930, New Jersey hatchery fish; Vla-dykov and Legendre 1940, Quebec wild stocks; Cooper 1953, Michigan stream fish; and Allen 1956, Wyoming beaver ponds). I compared the data from Rounsefell (1957) to the fecundity data from this study (Figure 6.10). I have also included the fecundity versus length regression presented in Wydoski and Cooper (1966), with total length converted to fork length by multi-plying by 0.92 (Rounsefell 1957), and inches converted to mm. The data in Figure 6.10 show that stunted brook trout have similar fecundities to non-stunted brook trout, suggesting that fecundity is size-limited and not limited by scarce food resources. The predominance of points about a mean length of 200 mm in Figure 6.10 includes young and old fish, since growth for stunted trout ceases at about 200 mm (chapter 7). This is in contrast to data from the other studies, in which mean fecundity estimates were from small fish that were young. On closer examination the data do not support the hypothesis that fecundity is independent of age. 116 2000 1500 A A "6 c 3 © 1000 c CO CD E © © A A 500 - . -I I I I I I ' i l l l I I I I I I I L_ _L J I I I I L_ 100 200 300 mean length (mm) at age 400 Figure 6.9 Mean fecundity versus mean fork length (mm) at age for 24 lakes from 1986-1989 (see Table 6.1). The emphasized points are data from the removal experiment in Gem 2 (©) and the transfer experiment in Fishgut pond (A) The data from these two experiments showed that when stunted trout are provided with reduced density and presumably more food per individual, fecundity and size increased to levels similar to non-stunted populations. 117 2 0 0 0 100 200 3 0 0 400 mean length (mm) at age Figure 6.10 Mean fecundity versus mean fork length (mm) at age for 24 lakes from 1986-1989 (see Table 6.1), and fecundity—length relationships from other studies of non-stunted brook trout populations. The data for New Jersey, Quebec, Michigan, and Wyoming were extracted by Rounsefell (1957) from studies by Hayford and Embody (1930, New Jersey), Vladykov and Legendre (1940, Quebec), Cooper (1953, Michigan), and Allen (1956, Wyoming). I extracted the data for Pennsylvania from Wydoski and Cooper (1966). The comparison between the Sierra Nevada fecundity—size data and the data from other studies suggests that stunted brook trout have similar fecundities to non-stunted brook trout of similar size. 118 The Decline in Fecundity with Age Seven of the stunted brook trout populations showed mean fecundity that decreased with increasing age (Flower 1989, Figure 6.1e; Wonder 3 and Wonder 2, Figures 6.2e and f; Fishgut 1 and Fishgut 3, Figures 6.3e and f; Dingleberry, Figure 6.4e; and Hell Diver 1, Figure 6.5f). Mean length and weight at age were constant in the populations (same figures, panels g and h), eliminating the explanation that the decline in fecundity with age can be explained by an inverse relationship between size and age. Nikolskii (1969; in his Table 17) refers to two studies that showed a decline in abso-lute fecundity with length, and presumably age. In a study of fecundity in chinook salmon Oncorhynchus tshawytscha Healey and Heard (1984) stated "For two populations the average fecundity of the oldest spawners was less than the fecundity of the preceding age, and in one instance was less than the fecundity of the youngest spawners (Yukon River)." Their definition of a population was not as restrictive as in my study. Fecundity samples from the Yukon were taken from in-river fisheries representing a mixture of spawning populations (Healey and Heard 1984). It is difficult to detect a decline in absolute fecundity with increasing age in non-stunted fish populations, because of the confounding influence of increasing size. Previous studies that have shown a negative relationship between fecundity and age have used fecundity to fish weight ratios (relative fecundity; see Bagenal 1978 for a review) or regression techniques to account for the usual confounding positive relationship between age and size. Since stunted 4 fish do not grow, no adjustment for increasing size with age is necessary. Some data suggest that as fecundity declined with age, mean ovary weight was constant while mean egg weight increased over the same range of ages. The best example is the 1989 data for Dingleberry (Figure 6.4), though the relationship is also evident in several other years and lakes (Flower 1989; Wonder 3 1987-1989; Wonder 2 1988, 1989; Fishgut 1 1989; Fishgut 3 1989; Dingleberry 1988; Hell Diver 1 1988, 1989). Many studies have shown that egg size is positively correlated with fish size (see Bagenal 1978 for a review), but the female brook trout 119 that I sampled were not increasing in size as they aged. I am unaware of other data from fish suggesting that as fish age, they produce fewer, larger eggs in ovaries of constant weight. Lacking more thorough fecundity data, I can only speculate on the reasons for the constancy of ovary weight with age, the size-independent inverse relationship between fecundity and age, and the size-independent positive correlation between egg weight and age. The constancy of ovary weight with age was likely a direct function of body size. Previous studies have shown that ovary weight and body size are correlated (see Bagenal 1978 for a review). Simply put, there is no more space in old stunted brook trout for ovaries than in young stunted brook trout, since they are the same size. The decline in fecundity with age may be a direct result of aging. The decline may be accelerated by scarce food resources, i.e., the negative effects of stunting on fecundity may be cumulative. If roughly the same amount of energy is "allocated" to annual gonad development, space limitations for ovaries suggest that individual eggs can be larger if fewer are produced. I prefer this explanation of the decline in fecundity with age and the corresponding increase in egg size to any explanation that suggests reproductive strategies of an adaptive nature. Several studies have shown that egg size and subsequent survival are positively correlated (Bagenal 1969, 1978), suggesting an advantage to larger egg size. Selective mechanisms that would favor large eggs in older brook trout versus small eggs in young brook trout are difficult to imagine. The Decline in Population Fecundity I expected that the number of eggs laid by the experimental populations would decline during the study, due primarily to the removals of mature females. Alternatively, it was possible that a combination of strong recruitment, decreased age of maturity, and increased fecundity could have offset the decreased number of eggs laid by older females. Population fecundity, the number of eggs that could be laid by all females in one season (Bagenal 1978), can be estimated by 15 total eggs = ^ Na pa fa fha for unfished populations (6-1) a = l and 120 15 total eggs = (Na — Ca) pa fa rha for experimental populations (6.2) a-l where Na = estimated number at age a, pa = proportion female at age, fa = mean fecundity at age, fha = proportion of females mature at age, and Ca = catch at age. Equation 6.1 estimates population fecundity before the removal experiments, and equa-tion 6.2 after the removal experiments. The Na are from Table 4.6; /„ are listed in Appendix B; pa and Ca are from the catch data. The maturity data that I collected were very limited; it is impossible to predict which fish will spawn in the fall based on observations of the ovaries in the summer. The category of immature female that I used for sampling in 1989 showed that about 50% of the age 1 females were "immature" (i.e., the eggs were not developed enough to separate and count). In 1989, all the age 2 fish were "mature" by this definition. Using this information I assumed rha was 0.5 for age 1 brook trout and 1.0 for all ages 24-. My obser-vations suggest that mi and rn,2 increased during the study as a function of increased growth (Aim 1959, McCormick and Naiman 1984) in response to decreased population abundance, but I do not have alternate data to show this. Pre-removal and post-removal estimates of population fecundity are presented in Ta-ble 6.3. The pre-removal estimates for Hell Diver 2 in 1986 and the other lakes in 1987 show population fecundities before heavy exploitation. The pre-removal estimates for the second and subsequent years of the experiments show population response to heavy exploitation. In every lake and year except Hell Diver 3 and Hell Diver 2 in 1987-1988, the populations were unable to balance the loss of eggs due to removals in the previous year, and population fe-cundity declined. The post-removal data estimate the number of eggs that potentially might have been laid in the subsequent spawning period. The post-removal data in Table 6.3 show a substantial decline in population fecundity in every year and lake. The percent change in pop-ulation fecundities between pre-removal estimates at start of the study and the post-removal 121 Table 6.3 Estimated population fecundity at the beginning and end of each removal exper-iment for the eight experimental populations. The estimates are derived from equations 6.1 and 6.2 and are explained in the text. Post-removal population fecundity declined in every lake and year. The percent change is the decrease measured from the pre-removal estimate at the start of the study to the post-removal estimate at the end of the study. lake year pre-removal post-removal % decrease Flower 1987 418043 278063 1988 325728 202657 1989 187100 103267 -75 Wonder 3 1987 141846 77169 1988 115272 42307 1989 65271 10351 -93 Fishgut 1 1987 192559 147638 1988 132849 60157 1989 45863 10495 -95 Dingleberry 1987 368002 231562 1988 267313 91261 1989 115792 43148 -88 Hell Diver 3 1987 30281 13522 1988 31571 12488 1989 20514 175 -99 Hell Diver 2 1986 59468 27312 1987 21236 12770 1988 40744 17356 1989 23900 0 -100 Par Value 1987 309754 169336 1988 245530 117079 1989 134070 31496 -90 Gem 2 1987 39976 20128 1988 26396 6856 1989 3035 0 -100 122 estimates at the end of the study ranged from —75% in Flower to —100% in Gem 2 and Hell Diver 2 (Table 6.3). The Relationship Between Fecundity, Abundance, and Density In their study of the effects of density reduction on a population of brown trout, Pech-laner and Zaderer (1985) found that the mean condition factor (A' = 100 x weight/length3) was 0.89 in 1979 and 0.92 in 1980, and increased to 1.04 in 1983. K for soma only (weight of gonads subtracted) was 0.88 in 1979, 0.92 in 1980, and 1.01 in 1983. They concluded that brown trout compensated for the artificial regulation of population numbers by increasing fer-tility. They based this conclusion on two results: (1) the lack of differences between total K and somatic K in 1979 (0.89 - 0.88 = 0.01) and 1980 (0.92 - 0.92 = 0.00) versus "the distinct decrease" in 1983 (1.04- 1.01 = 0.03), and (2) that the average gonad weight was 2.82% of total body weight in July 1983 versus 0.73% in July 1979. Regarding the latter result, the same month of sampling does not guarantee that the ovary is at the same stage of development between years. Regarding the differences between measures of total K and somatic A", the difference in 1983 is small and an indirect method by which to measure an increase in fertility. In this study, as I reduced population abundance and density through the removal experiments, individual fecundity increased. The data in Figure 6.11 show that the increase was slight over most of the range of population sizes and densities sampled. Individual fecundity increased sharply as population abundance was reduced below 100 fish and to densities of less than 1 fish per 100 m 2 and 0.25 fish per 100 m 3 . The data in Figure 6.11 are for the eight experimental lakes in 1988 and 1989, the two years that most ovary samples were collected and population estimates are available. The fecundity data in Figure 6.11 are averaged over all age classes. Sample size and numeric values are listed in Table 6.4. As density was decreased through the removal experiments, population fecundity de-clined (Table 6.3). The influx of large cohorts produced in 1987 and 1988 were not enough to balance the loss in population fecundity caused by the removal experiments. I discuss the implications of these data for juvenile survival in chapter 7. 123 Table 6.4 Sample size, mean fecundity and standard error, and estimates of population abun-dance and density for the eight experimental lakes in 1988 and 1989. Population estimates are from Table 4.6. Density estimates are from Table 4.8, scaled to fish per 100 m 2 (area) and fish per 100 m 3 (volume). The data are plotted in Figure 6.11. density lake year n / SE N area volume Ab3 perim Flower 1988 134 462.7 14.60 1664 8.90 7.19 8.90 2.72 Wonder 3 1988 183 343.1 7.18 700 5.38 1.52 13.99 1.42 Fishgut 1 1988 185 377.1 7.19 780 12.14 6.97 14.75 1.90 Dingleberry 1988 335 427.1 7.20 1444 6.95 3.83 8.48 2.38 Hell Diver 3 1988 28 546.4 25.60 118 1.34 0.21 4.75 0.30 Hell Diver 2 1988 34 522.6 27.99 173 4.26 1.51 8.82 0.66 Par Value 1988 97 455.9 15.63 1558 6.47 0.86 31.42 2.32 Gem 2 1988 34 574.6 31.73 85 1.21 0.66 1.43 0.21 Flower 1989 157 369.5 8.79 1489 7.96 6.43 7.96 2.44 Wonder 3 1989 175 290.4 6.29 735 5.64 1.60 14.69 1.49 Fishgut 1 1989 124 232.5 6.73 645 10.04 5.76 12.20 1.57 Dingleberry 1989 155 442.1 9.52 1285 6.19 3.41 7.55 2.12 Hell Diver 3 1989 29 693.0 53.39 75 0.85 0.13 3.02 0.19 Hell Diver 2 1989 51 515.1 21.04 257 6.33 2.24 13.10 0.98 Par Value 1989 196 426.2 8.70 1311 . 5.45 0.73 26.44 1.95 Gem 2 1989 3 1012.3 54.61 11 0.16 0.09 0.19 0.03 124 c 3 o c CO CD 1200 1000 - T a) 800 1 600 400 t A * * *. » " • 200 - • 0 i . . i . . 1 . . 1 , i . . I . . i 0 300 600 900 1200 1500 1800 population estimate (N) 1200 1000 c 800 o 600 (— co 400 CD C 200 1200 1000 - t c) 800 J 600 400 • + « A A • • * 200 - • 0 1 t . 1 . . fish per 100 m surface area 0 8 16 24 32 fish per 100 m 2 above 3 m T3 c 3 O CD C CO CD E fish per 100 m 3 volume 0 1 2 3 fish per meter of shoreline • Figure 6.11 Mean fecundity versus estimated population size (panel a), estimated population density per 100 m 2 surface area (panel b), estimated density per 100 m 2 above the 3 m depth contour (panel c), estimated density per 100 m 3 volume (panel d), and estimated density per meter of shoreline (panel e). The data are from the eight experimental lakes in 1988 (A) and 1989 (•). Vertical bars are ±1 standard error of the mean fecundity. Where the vertical bars are not visible, the standard error is less than the half the vertical dimension of the plotting symbol. Mean fecundity increased markedly when population size and density were reduced to extremely low levels. 7. Growth Response to Density Reduction To test the hypothesis that size should increase through growth with decreasing pop-ulation density, I collected data on fish length, weight, and age. Most of the brook trout populations that I sampled had a strong tendency to cease growth in length at about 200 mm. The relationship between weight and age was also asymptotic but more variable than the length-age relationship. The mean asymptotic length parameter (Loo) from the von Berta-lanffy growth equation (Ricker 1975) increased slightly as densities decreased. The increase in the mean asymptotic weight parameter (Woo) was more pronounced as densities decreased due to removals. The growth response to density reduction varied between lake populations. In general, weight at age increased more than length at age within a population. Mean size at age increased for most ages in all eight experimental populations. Though increased size at age was the common response to density reduction, the increases were minor and hardly noticeable in most lakes. The three exceptions to this were Hell Diver 3 and Gem 2, where fish size increased considerably due to removals, and Fishgut pond, where transplanted fish grew remarkably. Results Changes in Length At Age and Weight At Age The size responses of the eight experimental populations to density reduction are pre-sented graphically in Figures 7.1-7.8 and numerically in Appendix C. Sample sizes at age are listed in Table 7.1. Comments specific to each experiment are given in the captions to Figures 7.1-7.8. In general, the size of stunted brook trout increased slightly during the study, presumably due to density reduction from the removal experiments. Most of the non-experimental populations showed no change in size at age over the same period, or if both lake populations showed increased size over the same period, the increase for the experimental lake was greater (Wonder 3 and Wonder 2, Figure 7.2). 126 In two experiments the increases in mean length at age and mean weight at age were greatest for younger age groups (Wonder 3, age 2-6, Figure 7.2a and c; Fishgut 1, age 1-5, Figure 7.3a and c). In these figures it appears that the growth curves are shifted to the left, and the asymptotic size is reached at younger ages. Data from other experimental lakes indicate that the asymptote increased over a wider range of ages, (Dingleberry, age 3-8, Figure 7.4a and c; Hell Diver 3, age 3-10, Figure 7.5a and c; Hell Diver 2, age 4-13, Figure 7.6a and c). The appearance of a stable asymptote in Figures 7.2a and 7.3a may be the result of differences in growth between younger and older fish. Younger brook trout from stunted populations may have better recuperative abilities for growth than older fish that have been growth stunted for years. Differences in diet may favor the accelerated growth of young brook trout. Either inter-pretation suggests that the growth response in Wonder 3 and Fishgut 1 (Figures 7.2a and 7.3a) may be an intermediate response to density reduction compared with other experimental lakes, in which the response occurred over a wider range of ages. In all eight experiments, the relative increase in weight at age was greater than the corresponding increase in length at age (Figures 7.1-7.8). I expected that an increase in tissue and fat would precede an increase in skeletal length. In 1986, I moved 16 fished from stunted populations in Fishgut 2 and Fishgut 3 into a lake barren of fish and with abundant invertebrate fauna. Mean length and weight from this experiment are plotted in Figure 7.9. Rabe (1967a) conducted a similar experiment and my results were nearly identical. The data from these transplant experiments strongly suggest that stunting in alpine brook trout populations is environmentally rather than genetically controlled. 127 Figure 7.1 Mean fork length at age and mean weight at age for Flower (experimental) and adjacent, non-experimental Matlock. The vertical bars are plus and minus one standard error of mean length or mean weight. Where the error bars are not visible, the standard error was less than half the height of the plotting symbol, or the sample size was one. Sample sizes at age are listed in Table 7.1. The length and weight data for Flower and Matlock showed little change during the study. Both mean length at age and mean weight at age did increase slightly in 1989 compared with 1987 and 1988, though not enough to make a noticeable difference in fish size. The removal experiments removed an estimated 23% of the fish by number and 39% by weight between 1987 and 1989 (Tables 4.6-4.9). Brook trout sampled from Matlock in 1988 and 1989 were slightly larger than the fish from Flower. Matlock is larger, deeper, and has less stream spawning area than Flower, and I suspect a lower fish density. The mean length data for Flower (panel a) show the characteristic asymptotic growth curve for stunted brook trout in the Sierra Nevada. Growth in length ceased at about 200 mm at age 3-5. Growth in weight was more variable but was also asymptotic at about 80-90 grams in Flower (panel c). 128 Figure 7.2 Mean fork length at age and mean weight at age for Wonder 3 (experimental) and adjacent, non-experimental Wonder 2. The vertical bars are plus and minus one standard error of mean length or mean weight. Where the error bars are not visible, the standard error was less than half the height of the plotting symbol, or the sample size was one. Sample sizes at age are listed in Table 7.1. The length and weight data for Wonder 3 and Wonder 2 showed minor change during the study. Mean length at age and mean weight at age increased modestly for ages 1-6 and slightly for ages 8-14 in 1989 compared with 1987 and 1988. The removal experiments removed an estimated 36% of the fish by number and 58% by weight between 1987 and 1989 (Tables 4.6-4.9). Brook trout sampled from Wonder 2 in 1988 and 1989 were slightly larger than the fish from Wonder 3. The mean length data for Wonder 3 (panel a) show the characteristic asymptotic growth curve. Growth in length ceased at about 200 mm. The age at which the asymptotic size was reached decreased in 1989. In 1987 and 1988, brook trout growth in length reached 200 mm at age 6-7; in 1989 at age 3-5. Growth in weight was more variable and appears dome shaped due to a decline in mean weight at ages 9-15. 129 Figure 7.3 Mean fork length at age and mean weight at age for Fishgut 1 (experimental) and nearby, non-experimental Fishgut 3. The vertical bars are plus and minus one standard error of mean length or mean weight. Where the error bars are not visible, the standard error was less than half the height of the plotting symbol, or the sample size was one. Sample sizes at age are listed in Table 7.1. The length and weight data for Fishgut 1 showed minor change during the study. Mean length at age increased modestly for ages 1-5 in 1989. Mean weight at age increased for all ages except age 8 and 12. The removal experiments removed an estimated 41% of the fish by number and 38% by weight between 1987 and 1989 (Tables 4.6-4.9). Brook trout sampled from Fishgut 3 in 1988 and 1989 were similar in size to the fish from Fishgut 1. The characteristic asymptotic growth curve is evident in the length data for Fishgut 1 and Fishgut 3. Growth in length ceased at slightly less than 200 mm. The age at which the asymptotic size was reached decreased in 1989. In 1987 and 1988, brook trout growth in length reached near 200 mm at age 6-7; in 1989 at age 4. Growth in weight was more variable and peaked at about 75 g for age 4 and 7 brook trout in 1989, a slight increase over the asymptotic weight of 50-55 g for Fishgut 1 in 1987 and 1988, and Fishgut 3 in 1988 and 1989. 130 Figure 7.4 Mean fork length at age and mean weight at age for Dingleberry (experimental) and nearby, non-experimental Midnight. The vertical bars are plus and minus one standard error of mean length or mean weight. Where the error bars are not visible, the standard error was less than half the height of the plotting symbol, or the sample size was one. Sample sizes at age are listed in Table 7.1. The length data for Dingleberry showed minor change during the study. Mean length at age increased for ages 1-8 in 1989. Weight at age increased modestly for all ages except age 13. The fish were noticeably larger in 1989 compared with previous years. Most of the fish sampled appeared to have filled out and had lost the emaciated look of stunted fish. The removal experiments removed an estimated 41% of the fish by number and 46% by weight between 1987 and 1989 (Tables 4.6-4.9). Brook trout sampled from Midnight in 1989 were smaller size at age than the fish sampled from Dingleberry in 1987-1989. The characteristic asymptotic growth curve is evident in the length data for Dingleberry. Growth in length ceased at about 200 mm at age 4-5 in 1987 and 1988 and at about 210-220 mm in 1989. In 1987 and 1988 the mean weight at age data for Dingleberry (panel c) indicates an asymptotic weight of 80-90 g. In 1989, mean weight at age for ages 4-12 ranged from 104-133 g. 131 Figure 7.5 Mean fork length at age and mean weight at age for Hell Diver 3 (experimental) and nearby, non-experimental Hell Diver 1. the vertical bars are plus and minus one standard error of mean length or mean weight. Where the error bars are not visible, the standard error was less than half the height of the plotting symbol, or the sample size was one. Sample sizes at age are listed in Table 7.1. Mean length at age and mean weight at age increased for most ages in 1989 compared with 1988 and 1987, and in 1988 compared with 1987. By 1989, the brook trout in Hell Diver 3 had reached a respectable size for sport fishing. The removal experiments removed an estimated 55% of the fish by number and 62% by weight between 1987 and 1989 (Tables 4.6-4.9). Brook trout sampled from Hell Diver 1 in 1988 and 1989 were smaller size at age than the fish sampled from Hell Diver 3 during the experiment. The length at age curves for Hell Diver 3 (panel a) indicate that the asymptotic length was about 225 mm when the fish were first sampled extensively in 1987. Length at age increased about 15 mm in 1988 and again by about 10-15 mm in 1989. The brook trout in Hell Diver 3 were larger at the start of the removal experiment and showed increased capacity for growth in length and weight over most ages. The latter finding is in contrast to other lakes (Wonder 3, Fishgut 1, Dingleberry) where increased growth was confined to younger age classes. Density in Hell Diver 3 was the lowest of all eight experimental lakes at the start of the removal experiments in 1987 (Tables 4.6-4.7), which likely explains why the fish were initially larger. 132 Hell Diver 2 (experimental) 1986 1987 1988 1989 Hell Diver 1 1988 • '1989 Figure 7.6 Mean fork length at age and mean weight at age for Hell Diver 2 (experimental) and adjacent, non-experimental Hell Diver 1. The vertical bars are plus and minus one standard error of mean length or mean weight. Where the error bars are not visible, the standard error was less than half the height of the plotting symbol, or the sample size was one. Sample sizes at age are listed in Table 7.1. Mean length at age and mean weight at age increased for all ages between 1986 and 1987. The fish were noticeably larger in 1987. Most of the fish sampled appeared to have filled out and had lost the emaciated look of stunted fish. The initial increase in length and weight appeared promising but growth did not continue. Mean length for ages 5+ varied about 210-220 mm for 1987-1989, and mean weight varied widely about 120-130 g. The removal experiments removed an estimated 15% of the fish by number and 60% by weight between 1986 and 1989 (Tables 4.6-4.9). Brook trout sampled from Hell Diver 1 in 1988 and 1989 were of similar size at age to the fish sampled from Hell Diver 2 at the beginning of the removal experiment in 1986. The length at age curves for Hell Diver 2 (panel a) indicate that the asymptotic length was about 200 mm when the fish were first sampled in 1986. The asymptotic length was about 220 mm in 1987-1989. Weight at age was asymptotic at about 90 g in 1986, and varied about 120 g in 1987-1989. Though the approximate 25% increase in mean weight is statistically significant, from a sport fishing point of view there is little difference between a fish that weighs 120 g or 90 g. 133 Figure 7.7 Mean fork length at age and mean weight at age for Par Value (experimental). There was no lake of similar size, elevation, and species assemblage in the same drainage as Par Value for comparative purposes. The vertical bars are plus and minus one standard error of mean length or mean weight. Where the error bars are not visible, the standard error was less than half the height of the plotting symbol, or the sample size was one. Sample sizes at age are listed in Table 7.1. The mean length at age and mean weight at age data show very little change during the study. The removal experiments removed an estimated 10% of the fish by number and 54% by weight between 1986 and 1989 (Tables 4.6-4.9). This was the lowest percent removal by number for the eight experimental lakes. The age composition data in Figure 5.2g and biomass at age estimates in Table 4.7 show that the bulk of the biomass of older fish was removed by the depletion experiment and replaced very strong age 1 and age 2 cohorts in 1989. The data in this figure suggest that the failure to reduce the population numerically maintained the constancy of the length and weight at age relationships between years. The mean length data for Par Value (panel a) show the characteristic asymptotic growth curve for stunted brook trout in the Sierra Nevada. Growth in length ceased at slightly above 200 mm at age 4-5. Growth in weight ceased at about 100 g at age 5. 134 Gem 2 (experimental) • - - 1987 ' ' 1 9 8 8 • - 1 9 8 9 Gem 3 •1988 • ' 1 9 8 9 3 0 0 h 2 0 0 1 0 0 0 2 5 0 a) b) 2 0 0 150 1 0 0 5 0 0 C) d) 10 15 0 age 5 10 15 age Figure 7.8 Mean fork length at age and mean weight at age for Gem 3 (experimental) and adjacent, non-experimental Gem 2. The vertical bars are plus and minus one standard error of mean length or mean weight. Where the error bars are not visible, the standard error was less than half the height of the plotting symbol, or the sample size was one. Sample sizes at age are listed in Table 7.1. For comparison with weight at age data in Figures 7.1-7.7, note that the maximum value for the ordinate scale is 250 g. Mean length at age and mean weight at age changed substantially during the experiment in Gem 2, while there was a minor increase in the adjacent, non-experimental Gem 3. The removal experiments removed an estimated 96% of the fish by number and 83% by weight between 1987 and 1989 (Tables 4.6-4.9), by far the largest rate of removal of the eight exper-imental lakes. The removals were aided by a lack of recruitment in Gem 2. As indicated by the age composition in Figure 5.2h, age classes were either abundant or absent from the pop-ulation. The reasons for the success or failure of a year class are unknown. Recruitment was probably inhibited by lack of suitable spawning habitat in many years or by juvenile cohorts are completely cannibalized due to lack adequate refugia. Gem 2 is peculiar in that it has no apparent inlet or outlet. The mean length data for the 1983 cohort increased from 194 mm in 1987 (age 4) to 219 mm as age 5 in 1988 to 268 mm in 1989 (age 6). The same cohort increased in mean weight from 77 g to 117 g to 232 g from 1987-1989. Not only was the increase in mean length and mean weight statistically significant between years, but by 1989 the fish in Gem 2 were approaching banker size for brook trout of the eastern Sierra Nevada. 135 Figure 7.9 Mean fork length at age and mean weight at age for Fishgut pond (experimental) and adjacent, non-experimental Fishgut 3. The vertical bars are plus and minus one standard error of mean length or mean weight. Where the error bars are not visible, the standard error was less than half the height of the plotting symbol, or the sample size was one. For comparison with Figures 7.1-7.8, note that the maximum value for panels a and b is 320 mm and for panels c and d is 500 g. Mean length at age increased 40-85% and mean weight at age increased by 5-8 times as a result of transplanting 16 fish from severely stunted populations in Fishgut 2 and Fishgut 3 into a Ashless lake with abundant food. The 16 transplanted fish were originally similar in size to the fish from Fishgut 3. Eight of the transplanted fish were recovered in 1987, eleven months after the transfer. 136 Table 7.1 Sample sizes for mean length at age and mean weight at age data for the eight experimental lakes as presented in Figures 7.1-7.8. Most sample sizes are less than the catches listed in Table 2.1; in a few lakes, not all of the catch was sampled for age, and in most lakes, a few ages could not be determined from the otoliths. age year 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 total Flower 1987 28 126 103 89 92 56 18 19 6 6 8 3 554 1988 155 112 113 55 37 34 47 14 12 3 1 2 1 586 1989 237 177 53 38 31 19 23 11 4 5 1 3 602 Wonder 3 1987 23 138 66 31 28 107 34 33 23 17 12 6 2 2 2 524 1988 1 106 39 28 85 46 8 39 25 19 14 10 4 1 425 1989 27 392 58 16 28 15 14 3 26 8 6 7 8 3 2 613 Fishgut 1 1987 6 48 98 26 38 16 9 1 3 1 1 247 1988 36 32 116 100 27 29 22 7 6 7 1 383 1989 13 215 81 12 33 46 15 20 9 3 8 5 1 461 Dingleberry 1987 23 225 120 37 83 53 25 32 23 20 1 2 1 645 1988 2 96 251 266 103 20 54 26 22 17 16 17 3 1 1 895 1989 8 409 56 81 55 17 6 15 7 7 4 2 5 1 673 Hell Diver 3 1987 8 2 2 9 3 5 15 7 12 10 2 1 76 1988 6 25 4 3 1 3 1 3 11 7 2 2 68 1989 22 22 6 2 2 2 3 1 2 4 1 66 Hell Diver 2 1986 1 1 3 28 13 13 9 23 17 13 8 3 132 1987 12 1 5 4 4 8 8 2 3 4 1 54 1988 1 40 27 2 3 6 4 1 4 5 2 2 97 1989 152 1 52 17 1 1 1 4 3 232 Par Value 1987 7 82 89 36 9 23 101 26 70 54 86 37 6 2 628 1988 10 369 82 35 25 16 7 57 7 23 34 50 23 17 2 1 758 1989 406 302 27 12 9 2 4 28 14 15 16 27 9 871 Gem 2 1987 1 40 10 77 2 130 1988 1 38 27 66 1989 1 4 6 11 137 The Relationship Between Fish Size and Density Asymptotic Length and Weight Relationships A striking feature of the graphs of length at age is the regularity of a 200 mm length asymptote (Figures 7.1-7.8). This apparent limit on length was measured repeatedly in most of the lakes that I sampled (Table 2.1). Cooper et al. (1988) presented length at age data for stunted brook trout in Emerald Lake on the western slope of the Sierra Nevada that suggest an asymptote of about 180 mm standard length, which would convert to about 200 mm fork length. Other studies of stunted brook trout have not reported length at age data. The parameters mean asymptotic length (Loo) and mean asymptotic weight (Woo) from the von Bertalanffy growth equations (Ricker 1975) / ^ L o o U - e - ' ^ - ' o ) ) , (7.1) and ^ = W c o ( l - e - K ( t - f o ) ) , (7.2) provide a convenient method for quantifying the asymptotes evident in Figures 7.1-7.8. In the above equations K is the Brody growth coefficient, and t0 and t'0 are the hypothetical ages that a fish would have been of zero length and zero weight had it always grown in the manner described by the equation (Ricker 1975). I estimated the parameters of both equations following iterative search methods described by Schnute (1982) using the non-linear module in SYSTAT (Wilkinson 1988). The parameter estimates are presented in Table 7.2. Mean asymptotic length (Loo) was negatively correlated with surface area density (Fig-ure 7.10c; r = —0.76, P < 0.001). In contrast, Loo appears constant over a wide range of population sizes and densities by volume (Figure 7.10a and e). For both these relationships, Loo increased at very low levels of abundance and density. Mean asymptotic weight Woo in-creased as population abundance and density were decreased (Figure 7.10b, d, f). The rate of increase accelerated at low population sizes and densities. In general, the growth coefficient K increased as population densities were reduced (Table 7.2), supporting the observation made above that some of the growth curves were shifted to the left. 138 Table 7.2 Parameters of von Bertalanffy growth equations for length and weight (Loo, Woo, K, *o, *o)> estimated population size (JV), estimated fish per 100 m 2 ( D A ) , estimated fish per 100 m 3 (Dy), estimated fish per 100 m 2 above 3 m (DAb3), and estimate fish per meter of shoreline (Dp) for the eight experimental lakes. Loo and Woo are plotted against N, D A , and Dv in Figure 7.10. Loo K to Woo K t'o N D A D v DAb3 D P Flower 1987 208.6 0.58 -0.09 95.6 0.43 0.76 1923 10.28 8.31 10.28 3.15 1988 203.7 0.67 -0.08 84.7 0.55 0.74 1664 8.90 7.19 8.90 2.72 1989 208.3 0.70 -0.26 92.7 0.61 0.62 1489 7.96 6.43 7.96 2.44 Wonder 3 1987 198.2 0.53 -0.12 70.3 0.56 0.90 1153 8.85 2.50 23.04 2.34 1988 200.3 0.48 -0.74 78.7 0.39 0.08 700 5.38 1.52 13.99 1.42 1989 204.1 0.67 -0.50 84.3 0.55 0.07 735 5.64 1.60 14.69 1.49 Fishgut 1 1987 195.0 0.45 -0.45 52.6 0.63 0.77 1101 17.13 9.84 20.82 2.68 1988 187.0 0.57 -0.27 52.0 0.83 0.76 780 12.14 6.97 14.75 1.90 1989 192.9 0.71 -0.56 69.7 0.61 0.01 645 10.04 5.76 12.20 1.57 Dingleberry 1987 211.3 0.74 0.22 89.8 0.67 0.88 2186 10.52 5.80 12.84 3.61 1988 210.9 0.56 -0.44 90.7 0.48 0.13 1444 6.95 3.83 8.48 2.38 1989 220.7 0.69 -0.39 116.0 0.55 0.10 1285 6.19 3.41 7.55 2.12 Hell Diver 3 1987 239.0 0.26 -2.13 159.1 0.21 0.26 168 1.90 0.29 6.77 0.42 1988 244.1 0.45 -0.57 156.2 0.37 0.69 118 1.34 0.21 4.75 0.30 1989 257.6 0.51 -0.53 183.7 0.39 0.48 75 0.85 0.13 3.02 0.19 Hell Diver 2 1986 209.1 0.45 -0.06 101.8 0.31 0.98 303 7.46 2.64 15.44 1.16 1987 217.9 0.81 0.89 119.9 0.79 1.70 235 5.79 2.05 11.98 0.90 1988 228.0 0.55 0.03 136.1 0.38 0.90 173 4.26 1.51 8.82 0.66 1989 225.6 0.66 0.09 118.1 0.57 0.92 257 6.33 2.24 13.10 0.98 Par Value 1987 216.2 0.51 -0.54 103.7 0.45 0.20 1460 6.07 0.81 29.51 2.18 1988 215.8 0.59 -0.47 106.1 0.42 0.11 1558 6.47 0.86 31.49 2.32 1989 219.2 0.72 0.03 106.4 0.62 0.77 1311 5.45 0.73 26.50 1.95 Gem 2 1987 209.7 0.36 -1.33 108.4 0.20 -0.06 261 3.70 2.02 4.39 0.63 1988 222.6 0.81 -0.05 125.7 0.63 0.68 85 1.21 0.66 1.43 0.21 1989 268.0 1.42 1.29 232.7 1.34 1.82 11 0.16 0.09 0.19 0.03 139 L 300 200 100 W 440 880 1320 1760 2200 population estimate (N) 440 880 1320 1760 2200 population estimate (N) 300 200 100 • • c) * • * * / A • 1 . , , . 1 , , . . 1 • , , , 250 0 5 10 15 .20 fish per 1 0 0 m 2 surface area W 0 5 10 15 20 fish per 1 0 0 m 2 surface area 300 200 -100 • • # • e) • • A * • i » I . I . I i i r 0 2 4 6 3 10 fish per 1 0 0 m 3 volume W fish per 1 0 0 m volume Figure 7.10 Estimates of mean asymptotic size versus estimates of population abundance and population density for the eight experimental lakes in 1987-1989 (and 1986 for Hell Diver 2). Twenty-five points are plotted in each panel, one for each lake and year (x—1986, •—1987, A—1988, and •—1989). and are the mean asymptotic length and weight parameters from von Bertalanffy growth equations for length and weight (Ricker 1975). 140 Changes In Fish Size Within Populations Due To Changes In Density Length and weight of stunted brook trout tended to increase as densities were reduced through removal experiments (Figures 7.1-7.10). It is difficult to quantify the changes with a single measurement, since the change varies with age along the size at age curve. Mean length or mean weight averaged over all ages is a poor measurement, since the substantial shift in age and size distribution (Figure 5.5) will mask any increase in adult size. Instead, I used the LQO and Woo parameters from the von Bertalanffy growth equations for length and weight as described above. These parameters provide a measure of the mean asymptotic size that an average fish would reach if it continued to live and grow indefinitely according to equations 7.1 and 7.2 (Ricker 1975). The parameter estimates for Loo and Woo listed in Table 7.2 seemed to reflect accurately the changes in the size at age relationships observed in each year. The percent change in LQO and W o^o is plotted against the percent change in density (from Table 4.8) in Figure 7.11. The data show a 25% increase in Loo after two years in Gem 2, from an estimated density reduction of 95%. For the same lake, Woo more than doubled during the experiment. More modest size increases were obtained at less substantial density reductions. Woo increased over 25% in Dingleberry and Fishgut 1 with over 40% reductions in density. The data for Par Value and Flower suggest that density reductions have to be greater than 25% of the population to produce a change in fish size. The inverse relationship between Woo and density depicted in Figure 7.11b contradicts one conclusion by Donald and Alger (1989). After evaluating exploitation as a means of improving growth of a stunted brook trout population and reviewing two studies that mea-sured the effect of exploitation on salmonid populations (Pechlaner and Zaderer 1985, Arctic char; Langeland 1986, brown trout), they concluded that "For stunted salmonid populations, relationships between weight gain and magnitude of density reduction appear to be subtle (Ta-ble 4). Reductions of 12 and 90% caused the same general population changes." A somewhat stronger relationship between density and growth in weight is suggested by the data collected in this study. 141 125 100 75 50 25 - 2 5 100 75 5 0 25 O o 6 a) -25 b) o • * x o o - O - ^ - - A-o -100 -50 0 percent change in density 5 0 Figure 7.11 Percent change in a) and b) W M plotted against the percent change in density due to removals. The open plotting symbols represent growth and density differences between field seasons (x, 1986-1987; o, 1987-1988; A , 1988-1989). The filled circles (•) represent growth and density differences between the beginning of the experiments (1986 in Hell Diver 2 and 1987 in the other lakes) and samples collected at the end of the experiments in 1989. 142 8. Recruitment Response to Density Reduction While I predicted that the immediate response to density reduction would be increased size, I hypothesized three alternative possibilities for longer term response of reproductive and recruitment rates: 1) decreased recruitment due to decreased egg production; 2) no recruitment change, due to limiting factors in spawning habitat and juvenile rearing areas; or 3) increased recruitment due to reduction in competition or cannibalism effects of older fish. To test these hypotheses, I estimated the number of age 1 fish as recruits in 1986-1989, and considered relationships between recruitment and change in density for different age groups. The recruitment response varied between lakes. I did find increased recruitment that I attribute to reduced cannibalism or competition, but I also found recruitment failure and reduced recruitment at the highest levels of density reduction. Results and Discussion The Effect of Recruitment on the Removal Experiments Except for Gem 2, I was disappointed with the growth response of the stunted brook trout. The apparent reason that brook trout grew considerably in Gem 2 and little in the other lakes was variable recruitment. For an unknown reason, the fish in Gem 2 did not produce a strong year class during the study (Figure 5.5g). In contrast, the populations in the seven other experimental lakes produced cohorts in 1988 (measured as age 1 fish in 1989) that were 2-5 times larger than estimates of age 1 abundance in 1988 or 1987 (1987 and 1986 cohorts, respectively), and three of the populations produced large cohorts in 1987 (Figure 5.5, Table 4.6). I can speculate using circumstantial evidence on reasons for the large cohorts. First, I do not believe that the large cohorts were a result of large scale favorable environmental conditions. The non-experimental lakes did not show comparable increases in juvenile abundance. Second, the large cohorts were not a function of increased population fecundity. Population fecundity 143 decreased as a result of the removals (Table 6.3). More conclusively, the large 1988 cohorts were produced from eggs contained in females in the summer of 1987 and spawned in the fall of 1987. There was no opportunity for the survivors of the 1987 depletion experiments to increase fecundity in the time between the experiments and the fall spawning. Likewise, the large 1987 cohorts were developing as eggs in 1986, before the removal experiments began. The data presented in Table 6.3 showed that population fecundity decreased markedly in all lakes during the study. Having ruled out favorable environmental conditions and increased population fecun-dity, the only possible explanation for the large cohorts in 1988 and 1989 was increased juve-nile survival due to reduced competition or reduced predation by cannibalistic adults. Both explanations are plausible and neither can be eliminated with the data that are available. The production of large cohorts counteracted the effect of the removal experiments. Table 8.1 shows that while I reduced the density of fish age 3+ in all eight lakes, the density of fish age 1 and 2 increased in most lakes. The increased recruitment was not enough to offset the decline in numbers and biomass of fish due to the removals (Tables 4.6 and 4.7). The strong recruitments that I observed support one conclusion by Donald and Alger (1989) when they suggested that "To increase brook trout weight in a lake such as Olive, it may be more efficient, and just as effective, to block access to spawning sites and thereby reduce population density through year-class failure in one or more years." The action of blocking spawning sites may not be feasible for lakes in the Sierra Nevada, but the idea of controlling recruitment as the primary means of controlling density is important and is the next logical step into the problem of improving growth of stunted fish. The Response of Recruitment to Density Reduction Studies that attempt to evaluate recruitment in response to controlled levels of exploita-tion are rare because of many difficulties, including problems of estimating recruitment and the time lag between exploitation and its effect on recruitment. Healey (1980) conducted a large scale study of the relationship for lake whitefish Coregonus clupeaformis, and found that 144 recruitment did increase as exploitation increased, but the increase was variable and not clearly associated with the level of exploitation. In this study, the percent change in density due to removals is analogous to a level of exploitation. To determine a relationship between recruitment and density reduction I plotted the density of age 1 recruits (numbers in Table 4.6 divided by lake surface areas in Table 2.2) in year t + 1 against percent change in density due to removals in year t. Decreasing density can affect recruitment in three general ways; predation, competition, and egg production. The decline in density due to removal experiments in 1987 could have affected the age 1 recruits in 1988 (A in Figure 8.1) only through predation or competition, because the age 1 recruits in 1988 hatched in the spring of 1987, before the removal experiments began. The data represented by filled circles (•) in Figure 8.1 are number of age 1 recruits in 1989 plotted against the percent change in density due to removals in 1988. The density of age 1 recruits in 1989 may have been affected by predation and competition from other fish in 1988 and, in addition, may have been affected by egg production since they were spawned in the fall of 1987, after the first removal experiments. Regardless of the specific factors controlling recruitment, the data in Figure 8.1 show that recruitment densities (age 1 fish) were high when densities were slightly reduced, and decreased to nil as densities were reduced below 40% of the previous year's value. The data in Figure 8.1 suggest that it should be easy to over-fish brook trout populations and induce recruitment failures. Such an interpretation should be made with caution. It may be that if total densities can be reduced 40-60%, depensatory factors such as cannibalism may help regulate population abundance. However, as I discovered in most of the removal experiments, it was beyond my control to maintain density at 40-60% of the original density. Strong recruitments offset density reductions in most of the lakes (Figure 5.5, Table 4.6). I suspect that the low recruitments in Hell Diver 3 and Gem 2 (Figure 8.1) were aided by the reduction in density, but since both lakes had displayed weak or missing year classes before (Figure 5.5), I am certain that other factors besides density reduction were acting to inhibit recruitment in these lakes. 145 In Figure 8.2 I have plotted density of age 1 recruits in year t + 1 versus the density of age 3+ fish in year t. Note that as the densities of age 3+ fish were reduced between 1987 (A) and 1988 (•) in individual lakes, recruitment of age 1 fish the following year tended to increase in most lakes. Again the suggestion is that intermediate reductions of adult population density result in increased recruitment. Low recruitment was correlated with, but not necessarily caused by, severely reduced densities of age 3+ fish in Gem 2 and Hell Diver 3. Figure 8.2 is not a plot of a stock and recruitment relationship, since the density of age 1 fish in year t would depend on the abundance of age 3+ fish in year t - 2. The relationship between post-removal population fecundity in 1987 (stock, as an estimate of the number of eggs laid) and the number of age 1 recruits in 1989 is presented in Figure 8.3. For the first year of the removals, recruitment clearly depended on the number of eggs laid. Lacking recruitment data for 1990, I can only speculate that the slope of the relationship would have increased in subsequent years. The hypothesis that depensatory factors may maintain low density would be consistent with population densities and size structures that I observed in lakes with few, large fish (for example, Schober 1, Schober 3, and Maul). The population characteristics of low density and large size in these lakes seemed to be "stable", at least during the five summers of sampling that I conducted. Some factor must control density in these lakes; either spawning is inhibited or juvenile mortality is high. The data presented in Figure 8.1 lead to the speculation that if density could be reduced enough, then the population would regulate itself to maintain few numbers and large size. This speculation is an extension of the naive hypothesis that led to this study. The fundamental conclusion from the data presented in this chapter, in Figure 5.5, and in Table 4.6, is that stunted brook trout populations are extremely resilient to exploitation, and the objective of improving growth through density reduction is not likely to succeed without some method of inhibiting recruitment. 146 Table 8.1 Density in fish, per hectare by age group and for all ages for the eight experimental lakes. The data are from estimates of numbers at age presented in Table 4.6, divided by the lake surface area in hectares. The data show that while the densities of age 3+ fish were greatly reduced, recruitment into age group 1-2 reduced the potential effects of the reductions in several populations. density percent change lake year age 1-2 age 3+ all ages years age 1-2 age 3+ all age Flower 1987 397 631 1028 87 88 4-22 -36 -13 1988' 488 402 890 88 89 +19 -45 -11 1989 579 218 796 87 =• 89 +46 -65 -22 Wonder 3 1987 231 654 885 87 -» 88 -12 -49 -39 1988 203 335 538 88 -» 89 + 128 -69 +5 1989 462 103 564 87 => 89 + 100 -84 -36 Fishgut 1 1987 512 1200 1713 87 -» 88 -35 -27 -29 1988 333 881 1214 88 - f 89 + 116 -68 -17 1989 719 283 1004 87 => 89 +40 -76 -41 Dingleberry 1987 555 497 1052 87 -» 88 -45 -22 -34 1988 306 389 695 88 -» 89 +53 -61 -11 1989 469 151 619 87 =• 89 -15 -70 -41 Hell Diver 3 1987 60 132 190 87 -* 88 + 15 -53 -29 1988 69 62 134 88 89 -14 -58 -39 1989 59 26 85 87 => 89 -2 -80 -55 Hell Diver 2 1986 155 591 746 86 87 + 126 -62 -22 1987 350 227 579 87 -» 88 -37 -8 -26 1988 219 207 426 88 - 89 + 108 -14 +49 1989 456 177 633 86 => 89 + 194 -70 -15 Par Value 1987 167 439 607 87 -> 88 + 156 -50 +7 1988 427 220 647 88 -» 89 +7 -60 -16 1989 456 89 545 87 =• 89 + 173 -80 -10 Gem 2 1987 175 194 370 87 — 88 -65 -70 -67 1988 62 58 121 88 -> 89 -98 -76 -87 1989 1 14 16 87 => 89 -99 -93 -96 147 CO 0 CO "3 o .0 CO co 600 -500 400 -£ 300 -200 -co S 100 -100 -50 50 percent change in density in year t Figure 8.1 Density of age 1 recruits in year t + 1 versus percent change in density in year t. Density is estimated as number per hectare. For example, the X symbol represents the density of age 1 recruits in Hell Diver 2 in 1987 versus the percent change in density for Hell Diver 2 from 1986-1987. The filled triangles ( A ) represent age 1 recruits in 1988 versus percent change in density for 1987-1988, and the filled circles (•) represent age 1 recruits in 1989 versus percent change in density for 1988-1989. Symbol labels are: Fl—Flower, W3—Wonder 3, Fg—Fishgut 1, Db—Dingleberry, H3—Hell Diver 3, H2—Hell Diver 2, PV—Par Value, and G2—Gem 2. 148 03 CD JZ ro '3 o CD CD C D co o >. 'ro c CD 600 -500 400 -300 -200 100 • F 9 H2 # - P V # - H 2 X A F Q -H3# -I G2tf2A I I I I I 250 500 750 1000 density of age 3+ fish in year t 1250 Figure 8.2 Density of age 1 recruits in year t -f 1 versus density of age 3+ fish in year t. Density is estimated as number per hectare. The filled triangles (A) represent age 1 recruits in 1988 versus age 3+ fish in 1987, and the filled circles (•) represent age 1 recruits in 1989 versus age 3+ fish in 1988. Symbol labels are: Fl—Flower, W3—Wonder 3, Fg—Fishgut 1, Db—Dingleberry, H3—Hell Diver 3, H2—Hell Diver 2, PV—Par Value, and G2—Gem 2. Note the shift in points between years: in general, 1988 points are to the left and higher than 1987 values, suggesting that intermediate reductions of adult population density result in increased recruitment. 149 CD CO CO c w "3 b CD CD co co CD E rs c 1 0 0 0 8 0 0 -6 0 0 4 0 0 -2 0 0 1 0 0 0 0 0 2 0 0 0 0 0 post-removal population fecundity in 1987 3 0 0 0 0 0 Figure 8.3 Estimated number of age 1 recruits in 1989 versus estimated population fecundity in 1987; a stock—recruitment relationship for stunted brook trout populations after the first season of removal experiments. Population fecundity is an estimate of the number of eggs laid during the 1987 spawning season that gave rise to the age 1 recruits in 1989. The symbol label H2-86 represents recruits in 1988 versus population fecundity in 1986 for Hell Diver 2. Other symbol labels are: Fl—Flower, W3—Wonder 3, Fg—Fishgut 1, Db—Dingleberry, H3—Hell Diver 3, H2—Hell Diver 2, PV—Par Value, and G2—Gem 2. 150 9. Summary and Conclusions Stunting in Brook Trout Populations Reasons for individual fish to grow slowly and possibly be growth stunted have been discussed; they include the density dependent factors of inadequate quantity or quality of food (Comfort 1960), and lack of sufficient space (Res 1973). Density independent factors that may lead to stunting include interrelated factors such as water temperature (Dwyer and Smith 1983) and growing season (Van Oosten 1944), and other physical or chemical factors (Power 1980; Weatherly and Gill 1987). Density independent factors do not control stunting in the Sierra Nevada, since some lakes have large brook trout, and these lakes are probably not environmentally, physically or chemically different from lakes with stunted populations (Reimers et al. 1955; Stoddard 1986, 1987a, 1987b). I assume that limited space is a minor factor in controlling stunting in Sierra Nevada brook trout populations. This assumption could be tested by food addition experiments, but I did not in this study. Aim (1946) increased the size of stunted perch by in situ feeding, indicating that food, not space, was limiting. Having ruled out other causes of stunting, having shown that stunted fish grow when they are transferred to a low density environment with abundant prey, and having shown that stunted fish grow when densities are reduced, I am certain that inadequate food is the main factor causing stunting in Sierra Nevada brook trout populations. Previous studies of stunted fish have also found that inadequate food caused by over-crowded conditions leads to stunting (Pechlaner 1984; Pechlaner and Zaderer 1985; Donald and Alger 1989). The important question is why do fish overpopulate lakes, thus leading to stunted populations. For brook trout, initial stocking density had no effect. I found no relationship between stocking history (i.e., single stocking versus multiple stocking) and present densities. The populations I sampled have been self-sustaining for 20-50 years (Table 2.2). The lack of interspecific predation, the possibility of cold water increasing longevity (Reimers 1979; Craig 1985), and the decreased mortality in lakes (compared with streams) create conditions that are ideal for overpopulation. 151 Roff (1986) stated that a fish population will remain stunted only if survival or fecundity at a given size increases, and he cited Res (1973) as an example of increased fecundity (by increasing the number of broods per season) maintaining stunted conditions in Tilapia. The fecundity data in Figure 6.10 showed that stunted brook trout in the Sierra Nevada produce the same number of eggs as similar size fish in non-stunted populations. I am unaware of survival rate data from unexploited, non-stunted brook trout populations that would allow a similar direct comparison to the survival rate data presented in Table 4.5. The increased longevity of stunted brook trout compared with non-stunted populations is a qualitative indication that survival rates are higher in stunted brook trout populations. The apparent increase in juvenile survival in response to density reduction (Figure 5.5 and discussed in chapter 7) is another indication that increased survival is maintaining stunted populations. If population survival is an indication of successful adaptation, then the persistence of brook trout populations in high alpine lakes suggests that brook trout are well adapted to this environment. I suggest that stunting in alpine populations of brook trout is an indication of intermediate adaptation to environmental conditions. Brook trout do well enough to survive, but from a size (and management) perspective they produce too many offspring, probably because as individuals they are unable to produce less. The ideal fish for high alpine lakes would grow large and produce few eggs, something brook trout are apparently unable to do. This argument presumes that stunting is an abnormal condition based on the assumption that "natural" brook trout populations in lakes grow larger under less dense conditions. Brook trout size data reported in other studies suggest that this assumption is probably valid (Ricker 1932a, 1932b; Vladykov 1956; Bridges and Mullan 1958). Other salmonids that have been introduced into high alpine lakes in the Sierra Nevada are not as well adapted for survival as are populations of brook trout. Rainbow trout, golden trout, and brown trout are apparently limited by a lack of suitable spawning habitat in most high alpine lakes and streams. In many Sierra Nevada lakes, these three species are maintained by periodic stocking. Of the four fish species present in alpine lakes in the Sierra Nevada, brown trout were the only species that I sampled that had grown large enough to feed on stunted 152 brook trout. In contrast, I did not sample a single brook trout from a stunted population that was large enough to consume a stunted adult. Unlike brown trout, stunted brook trout are unable to grow out of the 200 mm size trap. Carlander (1966) noted that uniform size may be a common characteristic of stunted populations. The resilient reproductive efforts of brook trout may help populations persist after catas-trophic events such as avalanches or winter-kill. I suspect the probability of a population sur-viving such an event increases with the number offish. Considering my removal experiments as similar catastrophic events, the only population that did not survive was the only population that did not successfully reproduce (I believe I completely fished out the population in Gem 2). Models of Energy Shunting A persistent concept concerning the growth of fish is that at the onset of sexual ma-turity, the rate of somatic growth decelerates because of energy being diverted from soma to gonad. The model underlying this concept is that there is an input of energy as food, and the excess of this input over metabolic and excretory losses is partitioned between somatic and gonadal growth. Before maturation, the energy is available for somatic growth. After sexual maturation, the nutritional pool is somehow allocated between soma and gonads. Aim (1959) first challenged this concept by showing that growth curves for most natural fish populations decline gradually, without the sharp decline at maturity as stated by Hubbs (1926). lies (1974) reviewed the subject and renewed the challenge, providing an alternative model of seasonal fish growth. His model divided somatic growth and gonadal growth into seasonal periods that might have little overlap. In support of his model, lies (1974) cites several studies that found seasonal differences in somatic and gonadal growth patterns. Growth histories of stunted brook trout support the nutritional pool model. After reaching the age of growth cessation, all energy appears to be diverted from somatic growth into maintaining gonadal production. In support of lies (1974) seasonal growth model, it could be argued that the period of food availability in high alpine lakes (as limited by ice cover and prey production cycles) is during the period of gonadal growth. This argument fails when 153 results from the transfer experiment in Fishgut pond and the removal experiments in Gem 2 and Hell Diver 3 are considered. Fish in these experiments increased soma and gonads while stunted populations were developing gonads at the expense of somatic development. However, brook trout do not show a marked decline in somatic growth rate with the onset of maturity. The near cessation of growth in stunted brook trout at about 200 mm in length occurs at ages 3-5 (Figures 7.1-7.7), after the age of sexual maturity. In Emerald lake on the western slope of the Sierra Nevada, Cooper et al. (1988) found that males first become mature at age 1+, females at age 2+, and most fish are mature by age 3+. Though I did not measure maturity directly, I found that about 50% age 1 and all age 2 females in 1989 had eggs that were developed enough to separate and count, and I observed that the incidence of age 1 males with developed testes increased between 1988 and 1989. Stunted brook trout appear to become sexually mature at sizes less than the stunting limit. This result does not rule out the possibility of a sharp decline in growth rate at the onset of maturity, but it does show that the near cessation of growth and the onset of maturity do not occur at the same age. In observations of 16000+ otoliths, I did not notice a distinctive decrease in annuli spacing that would correspond to the onset of maturity at ages 1-3. Interannular distance tended to decrease gradually at ages 3-5, corresponding to the age that the length asymptote was reached. Assuming a strong relationship between interannular spacing on otoliths and past growth patterns, these observations do not support the hypothesis that somatic growth rate decreases sharply at maturity due to a shunting of energy into gonadal development. The similarity in fecundity between stunted and non-stunted brook trout (Figure 6.10) suggests that if energy shunting does occur, gonad development has priority over somatic development; otherwise stunted brook trout would be expected to produce fewer eggs. As I stated above, it appears that stunted brook trout are unable to produce fewer eggs. Perhaps there is a lower limit on the minimum number of eggs that a brook trout of a given size will produce through ovulation, independent of the physiological resources available to develop these eggs once they are ovulated. 154 The ubiquity of the 200 mm length asymptote suggests inadequate food resources strongly limit the size of stunted brook trout. Fecundity depends on size and brook trout are unable to decrease egg production. Populations of brook trout do not have other compen-satory responses that would decrease population size and presumably increase growth. Simply put, brook trout in high alpine lakes are the wrong fish in the wrong place. Can Stunting be Alleviated Through Density Reduction? Given the persistence of brook trout populations in high alpine lakes, and the resilient response of the populations to density reduction determined in this study and Donald and Alger (1989), I conclude that density reduction alone will not improve the growth of stunted brook trout populations. Gem 2 was the only experimental lake in which growth improved considerably and the remaining fish became a desirable size for angling (by my subjective definition). The distinctive feature about Gem 2 was that recruitment was nearly non-existent during the study. The age composition data for Gem 2 (Figure 5.5h) revealed that recruitment had failed before, since there were only two abundant age classes during the study. I suspect that recruitment failure in Gem 2 was related to the lack of suitable spawning habitat and the lack of refugia for juveniles. In most of the other experimental lakes, density reduction was followed by very large juvenile cohorts, either one or two years after the removals began (Table 4.6, Figure 5.5). Without some mechanism to inhibit recruitment, removals will not have the desired effect of increasing the food ration per individual and improving growth. Donald and Alger (1989) reached the same conclusion and recommended blocking access to spawning sites. This specific method is impractical for lakes in the Sierra Nevada for several reasons. Brook trout spawn in the lakes (Cooper et al. 1988), creating difficulties in blocking specific sites. Most of the lakes are in wilderness areas with regulations that prohibit the alteration of the natural environment. Any field work during the brook trout spawning season (October-December) is limited by often severe weather. There are other ways of decreasing recruitment. Traps or nets might exist or could be developed that would fish continually and would select juvenile brook trout, providing an 155 additional source of mortality for young age groups. Any unattended trap or net would have to be out of sight and out of reach (for example, from fishing lures) from recreational users of the lakes. An invisible method of reducing recruitment would be to introduce sterile fish that would interbreed with resident brook trout (Thorgaard 1983). The introduction of predators to reduce the density of stunted brook trout populations has not been formally evaluated in the Sierra Nevada or elsewhere. My sample of 149 fish from Heart lake (Rock Creek drainage, Mono County) contained two large brown trout (472 mm, «1400 g and 345 mm, 422 g), 56 brown trout averaging 234 mm and 136 g, and 91 brook trout averaging 168 mm, and 54 g. Repeated introductions of brown trout had not increased the size of the stunted brook trout. I also captured 8 brown trout during three years of removing 1875 brook trout from Flower lake. One brown trout weighed over 4 kg and the others were in the 200-300 g range. The few brown trout present in Flower had no apparent effect on the size of the resident brook trout. California Fish and Game biologist Darrell Wong told me of other lakes in the region in which brown trout have had a positive effect on brook trout size, but I did not sample from those lakes. Bunny Lake Revisited In chapter 1 I described the extended study of brook trout in Bunny lake conducted by Reimers (1958, 1979). Reimers documented the growth and population history of 1790 introduced brook trout fingerlings and their effect on the food resources in an isolated, ultra-oligotrophic, high alpine lake. His often cited 1979 paper provided much of the impetus for this study. It is instructive to compare some of Reimers' findings, and my original expectations, with my results. The most dramatic finding by Reimers was a tripling of the previously reported max-imum age for brook trout. Reimers (1979) collected a known age 19 fish in 1969 and otolith aged 20, 23, and 24 year old fish in subsequent years. I expected to sample brook trout of similar ages, but the oldest fish I sampled were age 16, two females from Par Value and one male from Hell Diver 1. 156 Though spawning activity was observed intermittently throughout his study, Reimers found that the Bunny lake brook trout did not successfully reproduce until 1966, at age 16. Though I did not expect to find as extreme an example as Bunny lake, I did expect to find uneven age distributions that would indicate periodic and persistent year class failures. I was surprised that age distributions in most of the lakes that I sampled were typical of fish populations, with a peak at age 1-3 and a decline toward older age classes (Figure 5.5). Recruitment was much more constant than I expected. Hell Diver 2 did have two very weak year classes during my study, but the general pattern was completely different from the situation described in Bunny lake. Based on the differences between Reimers' study and my results I conclude that the Bunny lake situation represents an extreme, if not unique occurrence in the population dynamics of stunted brook trout. In 1986 I took the opportunity to visit Bunny lake and set one gillnet. I observed no signs of fish and I pulled the empty net after one hour. In 1988 a crew from California Fish and Game set a gillnet overnight in Bunny lake and collected three fish. These fish were large (326, 328, and 340 mm in length) and the two smaller females were very fecund (about 1400 eggs each). There was some speculation that the fish may have been from the 1966 spawning that Reimers reported, but I aged the samples as 10, 10, and 8 years old. The Bunny lake brook trout population had not "finally run its course" as Reimers concluded in 1979. At least two different spawning seasons had produced a third generation of Bunny lake brook trout. Suggestions for Further Study Stunted brook trout populations seem ideal for a variety of studies. Reimers (1979) began preliminary investigations into the effects of aging on brook trout. There is no doubt more information yet to be obtained on the processes of aging in undisturbed, wild populations of long-lived brook trout. The relationships between fecundity, egg weight, and age revealed in this study demand further investigation. Stunted trout seem ideal for an expanded examination of these results, since they lack the confounding factor of increasing size with age. Changes in maturity at age in response to decreased density should be quantified. Collection of fecundity and maturity 157 data is hindered by problems of sampling these lakes during the fall-winter spawning season. Cooper et al. (1988) reported that spawning occurred in October-December in the lakes that they studied, but I collected several post-yolk sac fry in late August in Hell Diver 2, suggesting that either spawning occurs much later or the time to hatching is longer than reported in other brook trout populations. Methods of inhibiting recruitment should be tested as a means of reducing population density and, if possible, for eradicating brook trout. Survival rates and independent abundance estimates for juvenile brook trout should be quantified. The age composition data hint that there may be environmental factors controlling recruitment in undisturbed populations, but I could not detect clear relationships with the data that I collected. Snow and ice cover may directly influence year class strength. The current drought in the Sierra Nevada that persisted throughout this study probably reduced the variation in ice cover duration between years from what might normally be expected in high alpine lakes. Further attention should be directed toward the effects of fish behavior on techniques for estimating population size, and the types of behavior that might cause an apparent increase in probability of capture as population abundance and density decline. I proposed various behavioral hypotheses -and suggested possible methods of study in chapter 5. Investigation of diurnal differences in q, as evident in Figure 4.2, should be part of a further study into behaviorally related changes in catchability. The lakes of the Sierra Nevada are ideal for studying the dynamics of stunted trout populations. There are hundreds of lakes to choose from, and most are small and easy to sample. In many lakes, brook trout are the only fish species present. California Fish and Game, the agency responsible for the management of the high alpine lakes in the eastern Sierra Nevada, allowed me unlimited flexibility in manipulating brook trout populations. The widespread problem of small fish was understood by most of the recreational users in the backcountry, and almost everyone that I met was sympathetic to research that would improve the size of stunted fish, even by massive removal. 158 M a j o r Findings and Conclusions 1. The size of stunted brook trout increased in response to density reduction experiments, presumably due to an increased food ration per individual. The size increases were minor in most of the experimental lakes. 2. The growth response was inversely proportional to the reduction in population density. The growth response was strongest in juvenile brook trout and generally declined with age. 3. Brook trout populations were extremely resilient to massive removals. Recruitment in-creased markedly in response to depletions, apparently through increased juvenile survival, suggesting a strong compensatory factor normally regulates brook trout population abun-dance. 4. Fecundity increased in response to reduced density, but no more than would be expected from the increased size. Fecundity at size in stunted brook trout populations was similar to non-stunted brook trout, suggesting that metabolic resources are made available first to gonadal development and second to somatic growth. 5. Absolute fecundity declined with age, while ovary weight was almost constant, apparently maintained by increased egg size. Population fecundity declined during the study in lakes where density reduction occurred. Increased abundance of mature young brook trout did not offset the loss of egg production due to removal of most of the adult population. 6. Density reduction by itself is an ineffective management technique for increasing the size of stunted brook trout. Combined with some method of inhibiting recruitment, density reduction might be an effective technique. Recruitment inhibition alone may be the best long-term solution to the problem of stunting. 7. Catchability increased as population abundance and density were decreased. In the two experimental lakes with the largest growth response to depletions, catchability increased more than expected by the differences in lake area alone, suggesting individual brook trout may behave differently at lower densities. Hypothesized behavioral patterns invalidate techniques for estimating population size. 159 8. Given the growing recognition that the assumption of constant catchability is usually vio-lated in catch depletion data, methods for estimating population abundance that rely on this assumption should be discarded in favor of other estimation methods that can accom-modate variable catchability. 9. I validated annual structures in stunted brook trout otoliths for ages 2-3 and 5-10. I showed that ages determined from sagittal sections were similar to ages determined from cross sections. The justification for cross sections given in previous studies is unproven and misleading. 10. Survival at age seemed to decrease in stunted brook trout populations. However, the apparent increase in mortality in older fish can be explained by other factors, such as non-random sampling, persistent above average recruitment, or aging errors. 11. Growth in length ceased at about 200 mm in most stunted populations. Stunted populations are characterized by uniform adult size. 12. Relatively few lakes on the eastern side of the Sierra Nevada crest contain large brook trout. These populations are usually in the most remote lakes, suggesting that periodic catastrophic events such as avalanches or partial winter-kill regulate population abundances. The few adult brook trout in these lakes were usually all large. 13. Mortality due to sport fishing had little or no effect on the size structure of brook trout populations. I doubt that any sport fishing regulations (e.g., bag or size limits) could improve the growth of stunted brook trout. 160 Literature Cited Allen, G.H. 1956. Age and growth of the brook trout in a Wyoming beaver pond. Copeia (1): 1-9. Aim, G. 1946. Reasons for the occurrence of stunted fish populations (with special regard to the perch). Institute of Freshwater Research, Drottningholm, Report Number 25: 1-146. Aim, G. 1959. Connection between maturity, size, and age in fishes. Institute of Freshwater Research, Drottningholm, Report Number 40: 5-145. Bagenal, T.B. 1969. Relationship between egg size and fry survival in brown trout Salmo trutta L. Journal of Fish Biology 1: 349-353. Bagenal, T.B. 1978. Aspects offish fecundity. Pages 75-101 in S.D. Gerking, editor. Ecology of freshwater fish production. Blackwell Scientific Publications, Oxford. Bailey, R .M. , and K . F . Lagler. 1937. An analysis of hybridization in a population of stunted sunfishes in New York. Papers of the Michigan Academy of Sciences, Arts, and Letters. 23: 577-606. Barber, W.E., and G.A. McFarlane. 1987. Evaluation of three techniques to age Arctic char from Alaskan and Canadian waters. Transactions of the American Fisheries Society 116: 874-881. Beamish, R.J. 1979a. Differences in the age of Pacific hake (Merluccius productus) using whole otoliths and sections of otoliths. Journal of the Fisheries Research Board of Canada 36: 141-151. Beamish, R.J. 1979b. New information on the longevity of Pacific ocean perch (Sebastes alutus). Journal of the Fisheries Research Board of Canada 36: 1395-1400. Beamish, R.J., and D.E. Chilton. 1982. Preliminary evaluation of a method to determine the age of sablefish (Anoplopoma fimbria). Canadian Journal of Fisheries and Aquatic Sciences 39: 277-287. Beamish, R.J., and G.A. McFarlane. 1983. The forgotten requirement for age validation in fisheries biology. Transactions of the American Fisheries Society 112: 725-743. Beamish, R.J., and G.A. McFarlane. 1987. Current trends in age determination methodology. Pages 15-42 in R.C. Summerfelt and G.E. Hall, editors. Age and growth offish. Iowa State University Press, Ames. Beckman, W.C. 1941. Increased growth rate of rockbass, Ambloplites rupestris (Rafinesque) following reduction in density of the population. Transactions of the American Fisheries Society 70: 143-148. 161 Beckman, W.C. 1943. Further studies on the increased growth rate of the rock bass Ambloplites rupestris (Rafinesque) following reduction in density of the population. Transactions of the American Fisheries Society 72: 72-78. Bohlin, T., and B. Sundstrom. 1977. Influence of unequal catchability on population esti-mates using the Lincoln index and the removal method applied to electro-fishing. Oikos 28: 123-129. Bridges, C.H., and J.W. Mullan. 1958. A compendium of the life history and ecology of the eastern brook trout Salvelinus fontinalis (Mitchill). Massachusetts Division of Fisheries and Game, Fisheries Bulletin No. 23. Burrough, R.J., and C.R. Kennedy. 1979. The occurrence and natural alleviation of stunting in a population of roach, Rutilis rutilis (L.). Journal of Fish Biology 15: 93-109. Campana, S.E., and J.D. Neilson. 1985. Microstructure of fish otoliths. Canadian Journal of Fisheries and Aquatic Sciences 42: 1014-1032. Carlander, K .D. 1966. Relationship of limnological features to growth of fishes in lakes. Verh. Inter. Verein. Limnol. 16: 1172-1175. Carle, F.L. , and M.R. Strub. 1978. A new method for estimating population size from removal data. Biometrics 34: 621-630. Chilton, D.E., and R.J. Beamish. 1982. Age determination methods for fishes studied by the Groundfish Program at the Pacific Biological Station. Canadian Special Publication of Fisheries and Aquatic Sciences 60. Chilton, D.E., and M . Stocker. 1987. A comparison of otolith and scale methods for aging Pacific herring. North American Journal of Fisheries Management 7: 202-206. Clark, C.W. 1974. Possible effects of schooling on the dynamics of exploited fish populations. -Journal du Conseil, Conseil International pour l'Exploration de la Mer 36: 7-14. Clark, C.W., and M . Mangel. 1979. Aggregation and fishery dynamics: a theoretical study of schooling and the purse seine tuna fisheries. United States National Marine Fisheries Service Fishery Bulletin 77: 317-337. Comfort, A . 1960. The effect of age on growth-resumption in fish (Lebistes) checked by food restriction. Gerontologia 4: 177-186. Comfort, A. 1961. The longevity and mortality of a fish (Lebistes reticulatus Peters) in cap-tivity. Gerontologia 5: 209-222. Comfort, A. 1963. Effect of delayed and resumed growth on the longevity of a fish (Lebistes reticulatus Peters) in captivity. Gerontologia 8: 150-155. 162 Condrey, R.E. 1984. Density-dependent searching time: implications in surplus production models. United States National Marine Fisheries Service, Fishery Bulletin 82: 449-456. Cooper, E.L. 1953. Mortality rates of brook trout and brown trout in the Pigeon River, Otsego County, Michigan. The Progressive Fish-Culturist 15: 163-169. Cooper, S.D., T .M. Jenkins, Jr., and C. Soiseth. 1988. Integrated watershed study: an investigation of fish and amphibian populations in the vicinity of Emerald lake basin, Sequoia National Park. Final Report, California Air Resources Board Contract A 4 -122-32. Cowx, I.G. 1983. Review of the methods for estimating fish population size from survey removal data. Fisheries Management 14: 67-82. Craig, J.F. 1985. Aging in fish. Canadian Journal of Zoology 63: 1-8. Crecco, V .A . , and T.F. Savoy. 1985. Density-dependent catchability and its potential causes and consequences on Connecticut River American shad, Alosa sapidissima. Canadian Journal of Fisheries and Aquatic Sciences 42: 1649-1657. Crecco, V .A . , and W.J. Overholtz. 1990. Causes of density-dependent catchability for Georges Bank haddock Melanogrammus aeglefinus. Canadian Journal of Fisheries and Aquatic Sciences 47: 385-394. Crittenden, R.N., and G.L. Thomas. 1989. A conditional generalized least squares method for estimating the size of a closed population. Canadian Journal of Fisheries and Aquatic Sciences 46: 818-823. Cross, D.G., and B. Stott. 1975. The effect of electric fishing on the subsequent capture of fish. Journal of Fish Biology 7: 349-357. Dahl, K. 1917. Studier og fors0k over 0rret og 0rretvand, cenraltrykkeriet, Kristiania, Norway. (English summary in Salmon and Trout magazine, numbers 17 and 18, December 1918 and April 1919. Title: Trout and trout-waters in Norway.) Dahl, K. 1927. The "bledge" or dwarf-salmon. Skrifter utgitt av Det Norske Videnskap-Akademi 9. DeLury, D.B. 1947. On the estimation of biological populations. Biometrics 3: 145-167. Deedler, C.L. 1951. A contribution to the knowledge of the stunted growth of perch (Perca fluviatilis L.) in Holland. Hydrobiologia 3: 357-378. Diana, J.S. 1987. Simulation of mechanisms causing stunting in northern pike populations. Transactions of the American Fisheries Society 116:612-617. Donald, D.B., and D.J. Alger. 1986. Stunted lake trout (Salvelinus namaycush) from the Rocky Mountains. Canadian Journal of Fisheries and Aquatic Sciences 43: 608-612. 163 Donald, D.B., and D.J. Alger. 1989. Evaluation of exploitation as a means of improving growth in a stunted population of brook trout. North American Journal of Fisheries Management 9: 177-183. Dwyer, W.P., and C E . Smith. 1983. Brook trout growth efficiency as affected by temperature. Progressive Fish-Culturalist 45: 161-163. Eilers, J .M. , P. Kanciruk, R.A. McCord, W.S. Overton, L. Hook, D.J. Blick, D.F. Brakke, P.E. Kellar, M.S. DeHaan, M.E. Silverstein, and D.H. Landers. 1987. Characteristics of lakes in the western United States. Volume II. Data compendium for selected physical and chemical variables. EPA-600/3-86/054b, U.S. Environmental Protection Agency, Washington, D.C. Elliot, G.V., and T .M. Jenkins. 1972. Winter food of trout in three high elevation Sierra Nevada lakes. California Fish and Game 58: 231-237. Emery, A.R. 1975. Stunted bass: a result of competing cisco and limited crayfish stock. Pages 154-164 in R.H. Stroud and H. Clepper, editors. National symposium on the biology and management of centrarchid basses. Sport Fishing Institute, Washington, D.C. Eschmeyer, R.W. 1937. Some characteristics of a population of stunted perch. Papers, Michi-gan Academy of Science, Arts, and Letters 22: 613-628. Eschmeyer, R.W. 1938. Further studies of a perch population. Papers, Michigan Academy of Science, Arts, and Letters 23: 611-631. Fagerstrom, A. 1972. Netting for better angling in a small mountain lake. Report from the Institute of Freshwater Research, Drottningholm, Report 52: 38-49. Fargo, J., and D.E. Chilton. 1987. Age validation for rock sole in Hecate Strait, British Columbia. Transactions of the American Fisheries Society 116: 776-778. Foerster, R.E. 1968. The sockeye salmon, Oncorhynchus nerka. Fisheries Research Board of Canada Bulletin 162. Garrod, D.J. 1964. Effective fishing effort and the catchability coefficient. Rapports et Proces-Verbaux des Reunions, Conseil International pour l'Exploration de la Mer 155: 66-70. Garrod, D.J. 1977. The North Atlantic cod. Pages 216-242 in J.A. Gulland, editor. Fish population dynamics. John Wiley and Sons, New York. Gatz, A.J . , and J .M. Loar. 1988. Petersen and removal population size estimates: combining methods to adjust and interpret results when assumptions are violated. Environmental Biology of Fishes 21: 293-307. Gerking, S.D. 1957. Evidence of aging in natural populations of fishes. Gerontologia 1: 287-305. 164 Greene, A.F .C . 1955. Will stunted brook trout grow? Progressive Fish-Culturalist 17: 91. Gulland, J.A. 1964. Catch per unit effort as a measure of abundance. Rapports et Proces-Verbaux des Reunions, Conseil International pour l'Exploration de la Mer 155: 8-14. Hamley, J .M. 1975. Review of gillnet selectivity. Journal of the Fisheries Research Board of Canada 32: 1943-1969. Hammar, J., and 0. Filipsson. 1985. Ecological testfishing with the Lundgren gillnets of mul-tiple mesh size: technique modified for Newfoundland Arctic char populations. Report from the Institute of Freshwater Research, Drottningholm, Number 62. Hayford, C.C., and G.C. Embody. 1930. Further progress in the selective breeding of brook trout at the New Jersey state hatchery. Transactions of the American Fisheries Society 60: 109-113. Hayne, D.W. 1949. Two methods for estimating population from trapping records. Journal of Mammalogy 30: 399-411. Hazzard, A.S. 1933. Low water temperature, a limiting factor in successful production of trout in natural waters. Transactions of the American Fisheries Society 63: 204-207. Healey, M.C. 1980. Growth and recruitment in experimentally exploited lake whitefish (Core-gonus clupeaformis) populations. Canadian Journal of Fisheries and Aquatic Sciences 37: 255-267. Healey, M . C , and W.R. Heard. 1984. Inter- and intra-population variation in the fecundity of chinook salmon (Oncorhynchus tshawytscha) and its relevance to life history theory. Canadian Journal of Fisheries and Aquatic Sciences 41: 476-483. Heath, D., and D.A. Roff. 1987. Test of genetic differentiation in growth of stunted and nonstunted populations of yellow perch and pumpkinseed. Transactions of the American Fisheries Society 116: 98-102. Henderson, N.E. 1963. Extent of atresia in maturing ovaries of the eastern brook trout, Salveli-nus fontinalis (Mitchill). Journal of the Fisheries Research Board of Canada 20: 899-908. Hindar, K. , N . Ryman, and G. Stahl. 1986. Genetic differentiation among local populations and morphotypes of Arctic charr, Salvelinus alpinus. Biological Journal of the Linnean Society 27: 269-285. Hubbs, C.L., and G.P. Cooper. 1935. Age and growth of the long-eared and green sunfishes in Michigan. Papers of the Michigan Academy of Science, Arts and Letters 20: 669-696. lies, T.D. 1973. Dwarfing or stunting in the genus Tilapia (Cichlidae) A possibly unique re-cruitment mechanism. Rapports et Proces-Verbaux des Reunions, Conseil International pour l'Exploration de la Mer 164: 247-254. 165 lies, T.D. 1974. The tactics and strategy of growth in fishes. Pages 331-345 in F.R. Harden Jones, editor. Sea fisheries research. Ebek Science, London. Irie, T. 1960. The growth of the fish otolith. Journal of the Faculty of Fisheries and Animal Husbandry, Hiroshima University 3: 203-221. Jennings, M.R. 1989. Use of spring scales for weighing live fish in the field. North American Journal of Fisheries Management 9: 509-511. Johnson, L. 1976. Ecology of Arctic populations of the lake trout, Salvelinus namaycush, lake whitefish, Coregonus clupeaformis, Arctic char, S. alpinus, and associated species in unexploited lakes of the Canadian Northwest Territories. Journal of the Fisheries Research Board of Canada 33: 2459-2488. Jonsson, B. 1977. Demographic strategy in a brown trout population in Western Norway. Zoologica Scripta 6: 255-266. Jonsson, B., and K. Hindar. 1982. Reproductive strategy of dwarf and normal Arctic charr (Salvelinus alpinus) from Vangsvatnet Lake, western Norway. Canadian Journal of Fisheries and Aquatic Sciences 39: 1404-1413. Kelso, J .R.M., and B.J. Shuter v 1989. Validity of the removal method for fish population estimation in a small lake. North American Journal of Fisheries Management 9: 471-476. Landers, D.H., J .M. Eilers, D.F. Brakke, W.S. Overton, P.E. Kellar, M.E . Silverstein, R.D. Schonburd, R.E. Crowe, R.A. Linhurst, J .M. Omernik, S.A. Teague, and E.P. Meier. 1987. Characteristics of lakes in the western United States. Volume I. Population descriptions and physico-chemical relationships. EPA-600/3-86/054a, U.S. Environmental Protection Agency, Washington, D.C. Langeland, A . 1986. Heavy exploitation of a dense resident population of Arctic char in a mountain lake in central Norway. North American Journal of Fisheries Management 6: 519-525. Leslie, P.H., and D.H.S. Davis. 1939. An attempt to determine the absolute number of rats on a given area. Journal of Animal Ecology 8: 93-113. Lindstrom, T., A. Fagerstrom, and K.-J . Gustafson. 1970. Fishing pressure, growth and recruitment in a small high mountain lake. Report from the Institute of Freshwater Research, Drottningholm 50:100-132. Linfield, R.S.J. 1974. The errors likely in ageing roach Rutilus rutilus (L.), with special refer-ence to stunted populations. Pages 167-172 in T.B. Bagenal, editor. The aging offish. Unwin Brothers Ltd. 166 MacCall, A .D. 1976. Density-dependence of catchability coefficient in the Californian Pacific sardine, Sardinops sagax caerula, purse-seine fishery. California Cooperative Oceanic Fisheries Investigations Reports 18: 136-148. Mahon, R. 1980. Accuracy of catch-effort methods for estimating fish density and biomass in streams. Environmental Biology of Fishes 5: 343-360. Mann, G.J., and P.J. McCart. 1981. Comparisons of sympatric dwarf and normal populations of least cisco (Coregonus sardinella) inhabiting Trout Lake, Yukon Territory. Canadian Journal of Fisheries and Aquatic Sciences 38: 240-244. McCart, P.J. 1970. A polymorphic population of Onchorynchus nerka in Babine Lake, B.C. Ph.D. dissertation, University of British Columbia, Vancouver, British Columbia. McCormick, S.D., and R.J. Naiman. 1984. Some determinants of maturation in brook trout, Salvelinus fontinalis. Aquaculture 43: 269-278. Moran, P.A.P. 1951. A mathematical theory of animal trapping. Biometrika 38: 307-311. Mugiya, Y . 1972. On aberrant sagittas of teleostean fishes. Japanese Journal of Ichthyology 19: 11-14. Mugiya, Y . 1974. Calcium-45 behavior at the level of otolithic organs of rainbow trout. Bulletin of the Japanese Society of Scientific Pisheries 40: 456-463. Murnyak, D.F., M.O. Murnyak, and L . J . Wolgast. 1984. Growth of stunted and nonstunted bluegill sunfish in ponds. Progressive Fish-Culturalist 46: 133-138. Neilson, J.D., and G.H. Geen. 1980. Method for preparing otoliths for microstructure exami-nation. The Progressive Fish-Culturist 43: 90-91. Nelson, W.C. 1987. Survival and growth of fingerling trout planted in high lakes of Colorado. Technical Publication Number 36, Colorado Division of Wildlife. Nikolskii, G.V. 1969. Theory of Fish Population Dynamics as the Biological Background for Rational Exploitation and Management of Fishery Resources. (Translated by J.E.S. Bradley). Oliver and Boyd, London. Nordeng, H. 1983. Solution to the "char problem" based on Arctic char (Salvelinus alpinus) in Norway. Canadian Journal of Fisheries and Aquatic Sciences 40: 1372-1387. Nyman, L. 1984. Management of allopatric and sympatric populations of landlocked Arctic charr in Sweden. Pages 23-34 in L. Johnson and B.L. Burns (editors). Biology of the Arctic charr, proceedings of the international symposium on Arctic charr, Winnipeg, Manitoba, May 1981. University of Manitoba Press, Winnipeg. Otis, D.L., K.P. Burnham, G.C. White, and D.R. Anderson. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monographs 62. 167 Paloheimo, J.E., and L . M . Dickie. 1964. Abundance and fishing success. Rapports et Proces-Verbaux des Reunions, Conseil International pour l'Exploration de la Mer 155: 152-163. Pannella, G. 1980a. Growth patterns in fish sagittae. Pages 519-560 in D.C. Rhoads and R.A. Lutz, editors. Skeletal growth of aquatic organisms. Plenum Press, New York and London. Pannella, G. 1980b. Methods of preparing fish sagittae for the study of growth patterns. Pages 619-624 in D.C. Rhoads and R.A. Lutz, editors. Skeletal growth of aquatic organisms. Plenum Press, New York and London. Parker, R.A. 1958. Some effects of thinning on a population of fishes. Ecology 39: .304-317. Pechlaner, R. 1984. Dwarf populations of Arctic charr in high-mountain lakes of the Alps resulting from under-exploitation. Pages 319-327 in L. Johnson and B.L. Burns, editors. Biology of the Arctic charr, Proceeding of the International Symposium on Arctic Charr, Winnipeg, Manitoba, May 1981. University of Manitoba Press, Winnepeg. Pechlaner, R., and P. Zaderer. 1985. Interrelations between brown trout and chironomids in the alpine lake Gossenkollesee (Tryol). Verh. Inter. Verein. Limnol. 22: 2620-2627. Peterman, R .M. , and G.J. Steer. 1981. Relation between sportfishing catchability coefficients and salmon abundance. Transactions of the American Fisheries Society 110: 585-593. Peterman, R.M. , G.J. Steer, and M.J . Bradford. 1985. Density-dependent catchability coeffi-cients. Transactions of the American Fisheries Society 114: 436-440. Peterson, N.P., and C.J. Cederholm. 1984. A comparison of the removal and mark-recapture methods of population estimation for juvenile coho salmon in a small stream. North American Journal of Fisheries Management 4: 99-102. Pope, J.G. 1980. Some consequences for fisheries management of the behavior of pelagic fish. Rapports et Proces-Verbaux des Reunions, Conseil International pour l'Exploration de la Mer 177: 466-476. Pope, J.G., and D.J. Garrod. 1975. Sources of error in catch and effort quota regulations with particular reference to variation in the catchability coefficient. International Commission for the Northwest Atlantic Fisheries Research Bulletin 11: 17-30. Power, G. 1980. The brook charr, Salvelinus fontinalis. Pages 141-203, it in E.K. Balon, editor. Charrs: salmonid fishes of the genus Salvelinus. Dr. W. Junk, The Hague. Rabe, F.W. 1967a. The transplantation of brook trout in an alpine lake. The Progressive Fish-Culturalist 29: 53-55. Rabe, F.W. 1967b. Age and growth of rainbow trout in four alpine lakes. Northwest Science 41: 12-22. 168 Radovich, J. 1976. Catch-per-effort: fact, fiction, or dogma. California Cooperative Oceanic Fisheries Investigations Reports 18: 31-33. Reimers, N . 1958. Conditions of existence, growth, and longevity of brook trout in a small, high-altitude lake of the eastern Sierra Nevada. California Fish and Game 44: 319-333. Reimers, N . 1979. A history of a stunted brook trout population in an alpine lake: a lifespan of 24 years. California Fish and Game 65: 196-215. Reimers, N . , J.A. Maciolek, and E.P. Pister. 1955. Limnological study of the lakes in Convict Creek basin, Mono County, California. Fishery Bulletin 103: 437-503. Ricker, W.E. 1932a. Studies of speckled trout (Salvelinus fontinalis) in Ontario. University of Toronto Studies, Biological Series No. 36: 69-110. Ricker, W.E. 1932b. Studies of trout producing lakes and ponds. University of Toronto Studies, Biological Series No. 36: 113-167. Ricker, W.E. 1938. "Residual" and kokanee salmon in Cultus Lake. Journal of the Fisheries Research Board of Canada 4:192-218. Ricker, W.E. 1949. Mortality rates in some little-exploited populations of fresh-water fishes. Transactions of the American Fisheries Society 77: 114-128. Ricker, W.E. 1975. Computation and interpretation of biological statistics offish populations. Bulletin of the Fisheries Research Board of Canada 191. Robertson, O.H. 1947. An ecological study of two high mountain trout lakes in the Wind River Range, Wyoming. Ecology 28: 87-112. Roff, D.A. 1986. Predicting body size with life history models. Bioscience 36: 316-323. Rounsefell, G.A. 1957. Fecundity of North American salmonidae. Fishery Bulletin 57: 451-468. Routledge, R.D. 1989. The removal method for estimating natural populations: incorporating auxiliary information. Biometrics 45: 111-121. Saunders, J.W., and M.W. Smith. 1962. Transplantation of brook trout (Salvelinus fontinalis (Mitchill)) within a small stream system. Transactions of the American Fisheries Society 91: 388-394. Schaaf, W.E. 1980. An analysis of the dynamic population response of Atlantic menhaden, Brevoortia tyrannus, to an intensive fishery. Rapports et Proces-Verbaux des Reunions, Conseil International pour l'Exploration de la Mer 177: 243-251. Schaaf, W.E., and G.R. Huntsman. 1972. Effects of fishing on the Atlantic menhaden stock: 1955-1969. Transactions of the American Fisheries Society 101: 290-297. 169 Schnute, J. 1982. A manual for easy nonlinear parameter estimation in fishery research with in-teractive microcomputer programs. Canadian Technical Report of Fisheries and Aquatic Sciences 1140. Schnute, J. 1983. A new approach to estimating populations by the removal method. Canadian Journal of Fisheries and Aquatic Sciences 40: 2153-2169. Scott, D.P. 1962. Effect of food quantity on fecundity of rainbow trout, Salmo gairdneri. Journal of the Fisheries Research Board of Canada 19: 715-731. Scott, W.B., and E.J. Crossman. 1973. Freshwater fishes of Canada. Fisheries Research Board of Canada Bulletin 184. Seber, G.A.F. 1982. Estimation of animal abundance and related parameters (2nd edition). Griffin, London. Seber, G.A.F. , and E.D. Le Cren. 1967. Estimating population parameters from catches large relative to the population. Journal of Animal Ecology 36: 631-643. Seber, G.A.F. , and J.F. Whale. 1970. The removal method for two and three samples. Bio-metrics 26: 393-400. Shepard, B.G. 1970. Temporal and spatial differences in movement of cutthroat trout in Placid Lake, British Columbia. M.Sc. dissertation. University of British Columbia, Vancouver, British Columbia. Stoddard, J.L 1986. Nutritional status, microcrustacean communities and susceptibility to acid precipitation of high elevation lakes in the Sierra Nevada, California. Ph.D. dissertation. University of California, Santa Barbara. Stoddard, J.L 1987a. Alkalinity dynamics in an unacidified alpine lake, SierTa Nevada, Cali-fornia. Limnology and Oceanography 32:825-839. Stoddard, J.L 1987b. Microcrustacean communities of high elevation lakes in the Sierra Nevada, California. Journal of Plankton Research 9: 631-650. Thorgaard, G.H. 1983. Chromosome set manipulation and sex control in fish. Pages 405-434 in W.S. Hoar, D.J. Randall, and E .M. Donaldson, editors. Fish Physiology, volume 9, part B. Academic Press, New York. Tiemier, O.W., and J.B. Elder. 1960. Growth of stunted channel catfish. Progressive Fish-Culturalist 22: 172-176. Tonnessen, K . A . 1984. Potential for aquatic ecosystem acidification in the Sierra Nevada, California. Pages 147-169 in G.R. Hendrey, editor. Early biotic responses to advancing lake acidification. Butterworth, Boston. 170 Ulltang, 0. 1976. Catch per unit of effort in the Norwegian purse seine fishery for Atlanto-Scandian (Norwegian spring spawning) herring. FAO Fisheries Technical Paper 155: 91-101. Ulltang, O. 1980. Factors affecting the reaction of pelagic fish stocks to exploitation and requiring a new approach to assessment and management. Rapports et Proces-Verbaux des Reunions, Conseil International pour l'Exploration de la Mer 177: 489-504. Van Oosten, J. 1944. Factors affecting the growth of fish. Transactions of the Ninth North American Wildlife Conference, pages 177-183. Vladykov, V.D. 1956. Fecundity of wild speckled trout (Salvelinus fontinalis) in Quebec lakes. Journal of the Fisheries Research Board of Canada 13: 799-841. Vladykov, V.D. , and V . Legendre. 1940. The determination of the number of eggs in ovaries of brook trout (Salvelinus fontinalis). Copeia (4): 218-220. Walters, C.J. 1969. Effects of fish introduction on the invertebrate fauna of an alpine lake. Ph.D. dissertation, Colorado State University. Walters, C.J., and R.E. Vincent. 1973. Potential productivity of an alpine lake as indicated by removal and reintroduction of fish. Transactions of the American Fisheries Society 102:675-697. Walton, C.J. 1983. Growth parameters for typical anadromous and dwarf stocks of alewives, Alosa pseudoharengus (Pisces, Clupeidae). Environmental Biology of Fishes 9: 277-287. Warnick, D.C. 1966. Growth rates of yellow perch in two North Dakota lakes after population reduction with toxaphene. U.S. Fish and Wildlife Investigations in Fish Control 5: 1-9. Weatherly, A .H . , and H.S. Gill. 1981. Recovery growth following periods of restricted rations and starvation in rainbow trout Salmo gairdneri Richardson. Journal of Fish Biology 18: 195-208. Weatherly, A .H. , and H.S. Gill . 1987. The Biology of Fish Growth. Academic Press, London. Wetzel, R.G. 1975. Limnology. Saunders, Philadelphia. White, G . C , K.P. Burnham, D.L. Otis, and D.R. Anderson. 1978. User's manual for program C A P T U R E . Utah State University Press, Logan, Utah. White, G . C , D.R. Anderson, K.P. Burnham, and D.L. Otis. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory, Los Alamos, New Mexico. Wilkinson, L. 1988. SYSTAT: The System For Statistics. Systat, Inc., Evanston, Illinois. 171 Winters, G.H., and J.P. Wheeler. 1985. Interaction between stock area, stock abundance, and catchability coefficient. Canadian Journal of Fisheries and Aquatic Sciences 42: 989-998. Wolf, K. 1963. Physiological salines for fresh-water teleosts. The Progressive Fish-Culturist 25: 135-140. Woodhead, A.D. 1978. Fish in studies of aging. Experimental Gerontology 13: 125-140. Wydoski, R.S., and E.L. Cooper. 1965. Maturation and fecundity of brook trout from infertile streams. Journal of the Fisheries Research Board of Canada 23: 623-649. Zippin, C. 1956. An evaluation of the removal method of estimating animal populations. Biometrics 12: 163-189. 172 Appendix A Bathymetry and morphometry of the eight experimental lakes and three additional lakes. The measurements and units are latitude, longitude, elevation in meters above sea level, the number of soundings made with a hand-held sonar device, volume (m 3), surface area (m 2), mean depth (m), maximum depth (m), maximum length (m), maximum width (m), shoreline length (m), and the shoreline development, the ratio of the shoreline length to the length of the circumference of a circle of area equal to that of the lake, D L = L/2^/^(area), (Wetzel 1 9 7 5 ) . Measurements were made using a computer digitizing tablet from enlarged maps of each lake (described in chapter 4 ) . Arrows represent inlet and outlet streams. 1 7 3 Figure A l . Flower lake. latitude 36° 46 ' A longitude 118° 21 ' elevation 3200 m soundings 58 N volume 23000 m 3 | area 18700 m 2 mean depth 1.2 m maximum depth 2.4 m 0 meters 100 maximum length 250 m | I I I I I I I I I | maximum width 110 m shoreline length 610 m contour interval 1 meter shoreline development 1.26 174 Figure A 2 . Wonder 3 lake. latitude 37° 13' A longitude 118° 39' elevation 3375 m soundings 124 volume 46000 m 3 area 13000 m 2 mean depth 3.5 m maximum depth 7.0 m meters 100 maximum length 150 m I I I I I I I I | maximum width 110 m shoreline length 490 m contour interval 1 meter shoreline development 1.22 175 Figure A 3 . Fishgut 1 lake. latitude 37° 12' A longitude 118° 39' elevation 3315 m soundings 66 N volume 11000 m 3 | area 6400 m 2 mean depth 1.7 m maximum depth 3.7 m 0 meters 100 maximum length 180 m j I I I I I I I I I | maximum width 50 m shoreline length 410 m contour interval 1 meter shoreline development 1.45 176 Figure A 4 . Dingleberry lake. latitude 37° 11' A longitude 118° 38' elevation 3195 m soundings 95 N volume 38000 m 3 | area 20800 m 2 mean depth 1.8 m maximum depth 6.7 m 0 meters 100 maximum length 220 m | I I I I I I I I I | maximum width 130 m shoreline length 610 m contour interval 1 meter shoreline development 1.19 177 Figure A 5 . Hell Diver 3 lake. N meters H 1 h 100 H h contour interval 1 meter latitude longitude elevation soundings volume area mean depth maximum depth maximum length maximum width shoreline length shoreline development 37° 10' 118° 39 ' 3580 m 79 57000 m 3 8800 m 2 6.5 m 13.1 m 130 m 110 m 400 m 1.19 178 Figure A 6 . Hell Diver 2 lake. latitude 37° 10' A longitude 118° 39' elevation 3480 m soundings 65 N volume 11500 m 3 | area 4060 m 2 mean depth 2.8 m maximum depth 5.2 m 0 meters 100 maximum length 100 m | I I I I I I j I I j maximum width 60 m shoreline length 260 m contour interval 1 meter shoreline development 1.15 179 Figure A 7 . Par Value lake. A N 0 meters 100 | 1 1 1 1 1 1 1 1 1 1 contour interval 3 meters latitude 38° 05' longitude 119° 20' elevation 3135 m soundings 157 volume 180000 m 3 area 24100 m 2 mean depth 7.5 m maximum depth 17.7 m maximum length 220 m maximum width 170 m shoreline length 670 m shoreline development 1.22 180 Figure A 8 . Gem 2 lake. N meters 100 H 1 h H 1 h latitude longitude elevation soundings volume area mean depth maximum depth maximum length maximum width shoreline length 37° 23 ' 118° 45 ' 3335 81 13000 7000 1.8 4.3 170 70 410 m m ' m m m m m contour interval 1 meter shoreline development 1.39 181 N Figure A 9 . Hell Diver 1 lake. meters I 1 h 100 contour interval 1 meter latitude longitude elevation soundings volume area mean depth maximum depth maximum length maximum width shoreline length 37° 10' 118° 39' 3480 m 89 57000 m 3 11700 m 2 4.9 m 9.8 m 190 m 90 m 560 m shoreline development 1.47 182 Figure A10 . Bottleneck lake. meters H 1 1 100 H 1 1 contour interval 3 meters latitude 37° 11' longitude 118° 39 ' elevation 3390 m soundings 101 volume 385000 m 3 area 42400 m 2 mean depth 9.1 m maximum depth 18.3 m maximum length 365 m maximum width 250 m shoreline length 1000 m shoreline development 1.36 183 Figure A l l . Pass lake. i latitude 38° 08' longitude 119° 27' elevation 2955 m 1^  soundings 80 . volume 25000 m 3 | area 7500 m 2 mean depth 3.4 m maximum depth 7.9 m 0 meters 100 maximum length 150 m I 1 1 1 1 1 1 1 1 1 1 maximum width 80 m contour interval 1 meter shoreline length 380 m shoreline development 1.22 184 Appendix B Sample size at age, mean fecundity at age, mean ovary weight at age in g, mean egg weight at age in mg, mean fork length at age of females sampled in mm, mean weight of females sampled in g, and associated standard errors for the eight experimental lakes and the comparison lakes during 1986 or 1987 through 1989, as presented in Figures 6.1-6.8. experimental Figure Table page comparison Figure Table page Flower 6.1 D l 186 Matlock 6.1 B9 194 Wonder 3 6.2 B2 187 Wonder 2 6.2 BIO 195 Fishgut 1 6.3 B3 188 Fishgut 3 6.3 B l l 196 Dingleberry 6.4 B4 189 Midnight 6.4 B9 194 Hell Diver 3 6.5 B5 190 Hell Diver 1 6.5 B12 197 Hell Diver 2 6.6 B6 191 Hell Diver 1 6.6 B12 197 Par Value 6.7 B7 192 none Gem 2 6.8 B8 193 Gem 3 6.8 B13 198 1S5 Table B l . Sample size at^ age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Flower in 1988-1989. year age n 1988 1 1 2 27 3 39 4 19 5 8 6 12 7 15 8 6 9 4 10 11 1 12 1989 1 17 2 68 3 17 4 19 5 9 6 7 7 11 8 5 9 10 3 11 12 1 / SE 300.0 382.6 19.19 458.6 18.07 478.1 35.12 448.8 41.41 592.0 71.96 451.6 29.95 539.2 181.25 455.5 38.41 916.0 215.1 12.86 390.7 12.16 408.6 23.44 399.6 19.85 395.2 43.76 328.1 24.75 422.4 25.36 351.0 48.55 226.7 45.29 311.0 0 SE 0.22 0.32 0.028 0.57 0.038 0.53 0.044 0.72 0.148 0.89 0.176 0.73 0.099 0.89 0.311 0.66 0.112 1.21 0.06 0.006 0.61 0.038 0.74 0.073 0.85 0.077 0.83 0.132 0.81 0.131 1.20 0.101 0.93 0.163 0.91 0.254 0.92 E SE 0.73 0.81 0.045 1.24 0.063 1.14 0.068 1.56 0.208 1.41 0.119 1.56 0.121 1.56 0.150 1.43 0.160 1.32 0.27 0.024 1.55 0.084 1.84 0.179 2.15 0.160 2.06 0.126 2.56 0.446 2.86 0.175 2.68 0.357 3.87 0.370 2.96 L SE 113.0 154.9 1.60 179.6 1.76 184.4 2.34 200.4 9.34 206.8 5.40 200.3 3.79 214.8 10.05 205.8 1.70 255.0 129.3 1.81 165.6 1.42 183.4 2.23 192.6 2.01 202.2 3.64 202.9 4.33 208.3 3.15 201.8 4.96 211.0 1.00 190.0 W SE 14.0 39.6 1.23 58.8 2.13 61.8 2.89 84.9 16.24 89.5 8.74 77.2 6.37 106.0 22.55 85.5 6.08 119.0 23.5 0.93 50.5 1.25 65.9 2.77 73.4 2.62 82.3 5.55 81.9 4.54 93.3 5.33 80.8 4.40 94.3 2.91 71.0 186 Table B2. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Wonder 3 in 1987-1989. year age n / SE 0 SE 1987 1 2 3 296.3 22.17 0.84 0.181 3 7 293.4 20.76 2.49 0.376 4 4 411.3 131.47 3.69 1.237 5 3 259.3 22.45 2.41 0.748 6 19 277.2 23.05 4.54 0.506 7 7 224.0 35.26 3.18 0.318 8 5 203.6 12.31 3.20 0.373 9 6 166.2 16.84 3.88 0.632 10 5 179.2 34.35 3.66 0.912 11 2 180.5 40.50 2.56 0.420 1 1 279.0 0.02 2 21 342.6 15.10 0.60 0.059 3 15 380.8 23.37 1.29 0.200 4 54 393.8 11.79 1.51 0.076 5 22 365.5 17.61 1.41 0.132 6 4 305.0 22.83 1.91 0.450 7 18 301.9 16.82 2.11 0.185 8 12 271.8 14.17 1.82 0.139 9 14 281.6 32.24 2.22 0.277 10 8 308.0 57.73 3.64 2.024 11 8 261.1 27.14 1.59 0.223 12 3 300.0 31.21 1.04 0.226 13 1 191.0 0.98 1 84 281.9 6.38 0.14 0.011 2 27 373.2 19.73 1.13 0.088 3 9 304.7 19.14 1.86 0.192 4 11 297.8 19.79 1.58 0.149 5 6 320.7 38.22 2.15 0.488 6 6 306.2 49.38 3.12 0.664 7 3 236.7 42.88 1.34 0.356 8 12 236.2 17.35 2.01 0.258 9 3 301.7 38.12 1.61 0.516 10 4 173.5 12.32 1.69 0.179 11 4 237.0 47.01 1.88 0.503 12 3 205.3 40.13 1.44 0.266 13 1 218.0 1,13 14 2 181.5 37.50 1.76 0.065 E SE L SE W SE 2.76 0.438 149.3 2.85 38.7 3.18 8.50 1.159 166.9 3.23 50.9 3.49 8.85 0.530 170.8 6.21 54.8 5.53 9.35 2.973 174.3 8.41 58.0 8.08 16.42 0.988 194.4 1.63 74.5 2.23 16.04 2.498 192.3 3.36 71.6 3.16 15.69 1.485 200.6 6.79 77.6 8.64 22.86 2.166 201.2 2.29 79.5 4.85 20.19 2.061 197.2 3.84 72.4 5.87 14.38 0.901 197.0 1.00 74.5 3.50 0.07 112.0 15.0 1.76 0.163 152.9 1.36 41.3 1.07 3.34 0.463 168.5 3.60 54.9 3.52 3.97 0.220 177.9 1.74 61.9 1.63 4.19 0.513 174.9 2.01 60.0 1.96 6.29 1.396 183.3 5.44 66.5 5.52 7.15 0.598 196.8 1.84 77.4 1.95 6.84 0.614 201.5 2.87 78.5 3.46 8.62 1.021 201.3 1.98 79.8 2.80 9.00 2.476 198.6 1.61 74.6 2.21 6.31 0.983 194.4 2.28 65.9 2.56 3.42 0.562 198.7 8.19 74.7 6.67 5.13 195.0 66.0 0.49 0.034 135.4 1.14 26.8 0.67 3.10 0.214 168.4 2.15 52.7 2.01 6.15 0.606 188.4 2.64 70.9 3.86 5.40 0.482 184.3 4.38 70.2 3.78 6.47 0.920 198.2 4.50 84.8 8.36 10.28 1.463 199.5 4.48 86.3 6.67 5.51 0.907 203.3 6.23 83.3 7.88 8.73 1.000 203.5 2.35 86.4 3.75 5.34 1.420 201.3 1.86 80.7 4.63 9.71 0.663 202.8 4.94 79.0 4.14 8.51 2.407 206.0 .3.29 76.8 2.69 7.92 2.353 196.3 7.42 79.3 2.40 5.18 222.0 90.0 10.08 1.725 203.5 5.50 74.0 9.00 187 Table B3. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Fishgut 1 in 1988-1989. year age n 1988 1 2 7 3 65 4 53 5 13 6 18 7 10 8 6 9 5 . 10 7 11 1 12 1989 1 27 2 35 3 8 4 13 5 16 6 8 7 5 8 6 9 2 10 1 11 2 12 1 / SE 252.0 19.81 393.0 10.35 379.6 14.21 412.7 33.44 360.4 21.89 392.1 37.69 329.0 16.48 354.8 49.82 362.1 37.00 274.0 212.8 8.70 297.9 12.29 248.8 14.29 199.7 13.85 217.8 15.96 180.9 18.13 173.2 14.43 239.2 42.99 173.5 6.50 136.0 140.0 34.00 79.0 0 SE 0.16 0.012 0.49 0.018 0.50 0.020 0.49 0.044 0.47 0.036 0.47 0.052 0.41 0.018 0.48 0.052 0.60 0.147 0.34 0.08 0.006 1.20 0.106 1.31 0.178 1.91 0.116 1.63 0.131 2.08 0.249 2.12 0.346 1.14 0.241 1.32 0.495 1.78 0.98 0.505 0.73 E SE 0.64 0.057 1.28 0.062 1.35 0.043 1.18 0.060 1.32 0.082 1.24 0.162 1.28 0.104 1.40 0.164 1.62 0.286 1.24 0.37 0.018 4.03 0.337 5.39 0.866 10.09 0.946 7.92 0.711 11.75 1.229 11.94 1.200 6.14 1.664 7.54 2.571 13.09 6.55 2.017 9.24 L SE 135.7 2.06 161.4 0.82 166.5 0.94 173.0 1.89 174.4 1.92 178.8 2.72 186.2 2.98 192.0 3.51 184.7 1.76 184.0 135.0 1.70 163.2 1.88 174.1 1.93 184.5 1.61 183.5 1.87 186.1 2.23 185.4 3.49 188.2 5.56 183.0 1.00 185.0 193.5 3.50 200.0 W SE 27.1 1.14 44.2 0.77 46.5 0.78 48.5 1.51 47.1 1.18 47.9 1.45 55.2 0.75 50.8 4.39 48.1 3.96 40.0 27.5 1.02 48.3 1.44 56.6 2.38 66.1 1.60 63.3 1.57 68.0 2.76 66.0 4.25 57.0 3.43 62.0 3.00 62.0 63.5 4.50 62.0 188 Table B4. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Dingleberry in 1988-1989. 1989 age n 1 2 58 327.3 3 105 510.7 4 53 458.2 5 11 448.3 6 27 390.5 7 18 372.0 8 13 393.7 9 14 370.0 10 13 439.2 11 14 345.1 12 3 378.7 13 14 1 175.0 15 1 552.0 1 2 23 401.4 3 47 501.3 4 39 452.5 5 8 479.6 6 5 440.2 7 10 400.1 8 6 419.8 9 5 300.4 10 4 392.0 11 2 339.0 12 5 304.4 13 1 258.0 14 SE 0 SE 11.15 0.46 0.036 12.07 1.57 0.082 15.27 1.61 0.100 26.88 1.66 0.186 18.77 1.69 0.157 25.32 1.74 0.196 30.99 1.58 0.268 23.41 1.35 0.155 50.77 1.60 0.191 28.62 1.31 0.176 40.91 1.84 0.198 1.20 1.48 20.62 2.10 0.250 17.71 3.52 0.223 13.57 4.53 0.359 25.42 5.67 0.838 45.90 4.47 0.426 42.62 4.68 0.652 18.42 4.34 0.277 26.02 4.07 0.154 120.96 5.46 1.548 47.00 4.41 2.615 51.18 4.22 1.218 4.98 E SE L 1.45 0.149 164.5 3.10 0.152 189.2 3.66 0.244 194.2 3.64 0.294 197.4 4.30 0.347 202.9 4.72 0.467 203.2 3.95 0.502 205.8 3.69 0.372 207.9 4.02 0.659 215.5 3.74 0.338 211.6 4.85 0.020 212.7 6.86 211.0 2.68 208.0 5.22 0.557 183.4 7.08 0.382 201.5 9.94 0.722 210.3 12.11 1.889 211.6 10.56 1.291 224.6 12.15 1.388 213.6 10.44 0.818 226.2 13.93 1.289 210.4 15.18 2.239 219.3 12.19 6.024 223.0 13.15 2.051 217.4 19.30 216.0 SE W SE 1.37 51.7 1.30 1.18 76.9 1.36 1.65 81.0 1.95 1.99 82.5 2.29 2.23 85.5 2.52 3.14 89.9 3.62 3.82 88.7 4.55 2.19 86.3 3.34 3.24 93.9 4.35 2.24 89.4 2.66 3.28 87.3 5.55 78.0 85.0 1.82 74.7 2.64 2.29 95.8 2.85 2.52 105.3 3.42 4.04 108.8 3.98 4.68 130.0 13.02 2.99 108.8 4.42 4.08 119.5 3.69 3.22 97.0 3.90 L1.21 125.5 27.06 3.00 111.0 5.00 5.84 104.2 12.17 97.0 189 Table B5. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Hell Diver 3 in 1987-1989. year age n / SE 0 SE E SE L SE W SE 1987 1 o 1 395.0 0.30 0.76 155.0 44.0 z 3 A 1 380.0 0.87 2.29 173.0 63.0 5 c 1 493.0 2.30 4.67 212.0 106.0 0 7 Q 2 483.0 48.00 3.07 0.350 6.49 1.370 215.0 1.00 115.5 7.50 O 9 1 524.0 3.34 6.37 217.0 112.0 10 11 12 1988 1 2 12 531.8 42.89 0.38 0.039 0.71 0.052 167.8 4.65 55.8 4.55 3 1 462.0 0.35 0.75 165.0 51.0 4 c 2 575.5 44.50 2.33 0.313 4.02 0.234 215.0 11.00 113.5 13.50 0 6 1 454.0 1.62 3.57 219.0 123.0 7 8 9 1 741.0 4.87 6.58 246.0 162.0 5 580.8 55.00 3.64 0.773 6.24 1.242 238.6 4.61 150.8 10.39 10 4 506.3 96.57 2.56 0.871 4.84 1.076 240.0 4.56 141.3 17.75 11 12 2 590.5 32.50 2.47 0.745 4.27 1.496 245.0 151.0 24.00 1989 1 1 240.0 0.09 0.37 150.0 40.0 2 13 731.8 71.33 1.69 0.158 2.44 0.250 196.2 3.91 93.7 5.25 3 3 929.3 57.21 5.76 1.302 6.17 1.321 217.7 9.40 125.3 11.57 4 1 1381.0 4.96 3.59 232.0 144.0 5 6 1 884.0 5.42 6.13 229.0 146.0 7 8 3 448.3 146.54 5.43 2.570 10.15 3.057 251.3 2.85 164.3 11.20 9 1 699.0 4.47 6.39 260.0 188.0 10 1 877.0 8.31 9.48 247.0 163.0 11 4 485.5 34.82 4.26 1.525 9.26 3.672 238.8 3.84 134.0 14.61 12 1 428.0 4.08 9.53 234.0 125.0 190 Table B6. Sample size at^ age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Hell Diver 2 in 1986 and 1988-1989. year 1986 1988 1989 age n / 5 12 404.8 6 2 420.5 7 5 466.0 8 2 464.5 9 9 378.2 10 3 337.7 11 5 414.6 12 13 3 441.0 1 2 6 362.7 3 14 536.3 4 1 589.0 5 6 1 346.0 7 3 681.3 8 3 642.0 9 10 3 575.7 11 3 483.7 1 2 3 34 520.5 4 10 571.8 5 6 1 835.0 7 8 1 337.0 9 10 11 2 305.0 12 3 358.7 SE 15.22 43.50 42.91 105.50 24.95 43.84 32.68 0 SE 3.56 3.52 3.72 2.54 3.98 3.65 4.16 0.309 0.355 0.307 0.039 0.551 0.349 0.461 8.71 8.55 8.27 5.78 10.58 11.06 9.97 SE 0.522 1.730 0.996 1.398 1.211 1.151 0.587 182.3 197.0 193.4 200.5 200.3 196.3 204.2 SE 2.34 4.00 2.32 3.50 3.87 2.03 5.08 W 70.5 83.0 83.6 84.0 91.1 80.3 92.8 31.85 36.03 169.55 98.06 0.22 1.39 1.97 2.95 5.08 4! 10 0.040 0.171 1.617 0.579 0.60 2.66 3.35 8.53 7.25 6.75 0.073 0.379 0.617 1.573 156.2 187.3 200.0 219.0 227.7 224.3 3.78 3.63 9.28 6.06 40.7 76.7 91.0 111.0 140.7 125.7 23.65 3.02 0.264 40.38 3.65 0.374 4.41 9.70 5.70 0.382 190.3 6.S9 1.172 213.6 5.28 28.78 237.0 214.0 2.12 75.5 5.28 104.7 145.0 105.0 SE 3.26 4.00 3.04 9.00 4.17 4.26 7.61 29.26 3.58 0.325 8.22 1.044 215.0 5.13 103.0 8.02 2.78 4.54 22.67 7.75   .  60.29 2.82 0.925 4.89 1.669 210.0 17.16 101.3 25.83 78.50 6.01 3.523 10.91 4.933 220.3 9.13 107.3 17.37 2.71 7.83   89.00 1.67 0.040 5.94 1.603 235.0 1.00 104.0 4.00 39.40 5.11 1.842 14.00 4.197 226.3 4.37 102.3 5.04 191 Table B7. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (O), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Par Value in 1988-1989. year age 1988 1989 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 n / SE 0 SE E SE L SE W SE 3 244.0 46.13 0.10 0.027 0.40 0.067 137.0 12.29 30.0 6.66 12 410.3 32.22 0.46 0.081 1.08 0.124 169.6 2.16 56.7 2.02 7 443.7 35.60 0.88 0.081 2.01 0.148 188.9 6.62 79.4 3.32 5 567.2 81.93 1.35 0.297 2.58 0.739 202.4 4.02 87.6 5.30 2 681.5 130.50 1.41 0.460 2.01 0.289 214.0 9.00 102.0 20.00 2 584.0 139.00 1.43 0.830 2.24 0.889 204.0 0.00 95.5 2.50 15 483.3 28.74 1.35 0.124 2.98 0.399 209.7 2.19 97.1 2.96 3 459.3 85.19 1.99 0.141 4.67 0.944 211.7 6.01 102.7 10.73 7 456.3 62.51 1.89 0.363 4.55 1.241 210.9 1.67 97.4 2.70 13 449.1 39.34 1.66 0.341 3.89 0.779 208.1 2.46 92.0 3.67 13 436.4 43.43 1.68 0.149 4.35 0.610 212.8 1.89 97.1 2.57 7 452.9 77.94 1.37 0.271 2.95 0.455 217.6 3.22 107.0 5.96 4 413.3 28.35 1.66 0.608 3.82 1.091 211.0 2.35 88.8 6.80 1 150.0 1.93 12.87 217.0 94.0 1 345.0 0.93 2.70 218.0 101.0 2 199.5 0.50 0.05 0.000 0.25 0.001 129.0 1.00 24.0 2.00 110 401.3 8.10 0.45 0.015 1.14 0.035 167.7 1.16 52.7 1.06 9 512.8 43.68 0.71 0.070 1.40 0.081 195.8 3.16 83.3 4.60 6 463.2 39.91 0.93 0.190 1.97 0.289 205.8 3.65 87.7 4.35 5 462.6 56.10 1.08 0.227 2.24 0.278 214.0 3.86 100.4 3.64 1 440.0 0.65 1.48 213.0 94.0 1 372.0 0.65 1.75 211.0 94.0 14 539.7 37.85 0.94 0.068 1.80 0.138 217.2 2.17 104.7 3.48 11 490.6 53.77 1.06 0.099 2.28 0.226 214.4 3.30 103.2 3.98 11 421.5 37.43 0.97 0.109 2.30 0.184 218.8 2.69 108.6 4.75 6 447.8 44.76 0.86 0.082 1.99 0.179 215.5 3.74 95.5 5.97 12 340.3 36.22 0.75 0.094 2.43 0.422 221.2 3.58 105.8 5.10 5 560.8 94.24 1.03 0.168 1.86 0.096 219.8 4.66 94.6 6.26 1 553.0 1.25 2.26 -21S.0 95.0 192 Table B8. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Gem 2 in 1987-1989. year age n 1987 1 1 2 3 4 3 5 6 1988 1 1 2 16 3 4 5 17 6 1989 1 2 3 1 4 5 / SE 317.0 298.7 7.13 204.0 510.3 44.09 656.8 35.61 1090.0 973.5 66.50 0 SE 0.35 3.55 0.369 0.10 1.33 0.208 2.25 0.404 3.22 4.12 0.030 E SE 1.10 11.95 1.513 0.49 2.47 0.345 3.36 0.563 2.95 4.25 0.259 L SE 134.0 197.3 2.33 128.0 185.1 4.70 217.6 2.79 245.0 277.0 13.00 W SE 23.0 74.7 1.76. 23.0 76.4 5.44 114.4 5.59 202.0 252.5 30.50 193 Table B9. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (O), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Matlock in 1988-1989 and Midnight in 1989. year age n / Matlock 1 2 6 410.5 3 14 426.0 4 19 393.6 5 4 412.0 6 6 380.7 7 6 398.3 8 1 262.0 9 1 266.0 1 4. 257.3 2 6 480.2 3 4 2 385.0 5 3 603.0 6 7 8 9 Midnight 1989 1 2 22 237.3 3 11 215.8 4 14 239.4 5 8 209.8 6 2 154.0 7 3 156.3 8 4 259.8 9 3 276.7 10 1 197.0 11 2 239.5 12 13 1 274.0 SE 0 SE 23.03 0.37 0.022 30.58 0.50 0.031 26.34 0.49 0.049 74.91 0.60 0.148 44.31 0.64 0.120 59.88 0.56 0.094 0.50 0.35 35.66 0.06 0.008 43.50 0.33 0.071 22.00 0.57 0.015 53.67 0.88 0.098 9.32 0.51 0. .034 13.05 0.59 0. .064 44.67 1.04 0. .257 24.32 0.88 0. .124 20.00 0.78 0. .235 17.33 0.67 0. .075 61.52 0.79 0. .054 106.07 0.81 0. .307 0.60 25.50 1.23 0, .745 2.28 E SE L 0.91 0.059 168.8 1.20 0.056 193.8 1.20 0.072 199.4 1.44 0.176 208.8 1.70 0.320 212.5 1.41 0.075 228.8 1.91 198.0 1.32 223.0 0.23 0.006 126.5 0.68 0.114 170.0 1.50 0.047 200.5 1.45 0.121 213.7 2.23 0.172 153.6 2.77 0.261 164.2 4.12 0.442 171.5 4.36 0.524 173.4 5.39 2.226 181.0 4.46 0.922 182.7 3.58 0.829 192.3 3.01 0.438 200.0 3.05 193.0 5.55 3.702 185.0 8.32 185.0 SE . W SE 3.75 50.3 3.24 2.69 74.3 3.15 2.81 78.3 3.64 8.47 94.0 10.66 3.33 93.3 7.89 4.95 109.0 8.52 70.0 91.0 3.97 23.8 2.29 2.99 55.3 3.08 5.50 85.0 11.00 9.24 98.7 10.09 2.03 39. .5 1. .44 1.72 42. .3 1 .67 4.15 50. .6 4. .96 2.36 47. .6 1. .74 1.00 52. .5 0. .50 1.20 48, .7 3. .84 4.40 55. .8 3. .20 8.50 60 .7 8. .82 59 .0 18.00 57. .0 19 .00 72.0 194 Table BIO. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Wonder 2 in 1988-1989. year age n / 1988 1989 1 1 148.0 2 11 406.0 3 33 380.3 4 13 319.0 5 5 376.6 6 3 291.0 7 3 284.7 8 1 327.0 9 2 160.5 10 3 243.0 11 1 13 297.8 2 6 526.8 3 7 416.4 4 12 331.6 5 9 329.8 6 6 319.2 7 1 306.0 8 ' 4 266.8 9 4 223.0 10 2 301.0 11 4 238.8 SE 0 SE 0.12 30.58 1.34 0.242 16.07 1.63 0.102 23.97 2.00 0.211 29.97 1.25 0.150 23.71 2.02 0.123 29.28 1.87 0.398 0.61 95.50 0.55 0.325 49.92 0.89 0.129 28.03 0.26 0.075 40.10 1.42 0.231 25.72 2.03 0.319 24.18 2.66 0.281 29.06 2.36 0.416 18.46 2.02 0.211 2.21 42.93 2.53 0.645 12.25 1.76 0.454 22.00 1.27 0.280 7.42 1.68 0.243 E SE L 0.81 130.0 3.71 0.811 170.0 4.56 0.332 180.1 6.92 0.988 191.8 3.53 0.674 196.4 6.95 0.133 198.7 6.65 1.596 193.0 1.87 202.0 3.49 0.050 193.5 3.74 0.227 211.3 0.81 0.165 150.3 2.76 0.469 179.7 4.95 0.748 193.3 8.83 1.469 194.9 8.14 1.751 196.6 6.41 0.626 195.8 7.22 204.0 9.10 1.493 206.3 7.74 1.706 208.5 4.17 0.625 208.0 7.15 1.231 205.3 SE W SE 27.0 3.24 59.7 3.43 1.67 68.4 1.79 2.07 75.8 2.14 2.48 82.2 1.59 8.09 80.3 5.81 4.04 70.7 2.91 58.0 0.50 78.0 12.84 86.7 20.17 3.18 40.1 2.66 3.24 65.5 2.33 2.36 78.1 2.05 1.96 80.8 2.87 3.66 83.7 3.07 2.41 80.0 4.20 94.0 4.52 88.3 7.35 1.26 87.3 5.94 5.00 95.0 6.00 4.27 79.5 2.50 195 Table B l l . Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Fishgut 3 in 1988-1989. year age n 1988 1 2 4 3 22 4 10 5 4 6 6 7 13 8 9 9 6 10 1 11 3 1989 1 1 2 15 3 9 4 13 5 9 6 2 7 4 8 5 9 2 10 4 11 / SE 372.5 17.15 412.2 23.87 432.4 29.60 400.3 51.60 383.2 36.00 380.7 30.10 302.2 25.14 254.2 26.73 131.0 363.0 41.28 178.0 235.1 18.48 228.8 13.32 155.5 7.24 140.3 11.97 105.5 0.50 149.8 29.11 133.8 10.90 94.5 23.50 185.5 37.85 0 SE 0.20 0.024 0.50 0.052 0.50 0.044 0.35 0.017 0.43 0.055 0.55 0.070 0.34 0.028 0.29 0.042 0.27 0.39 0.025 0.03 0.66 0.079 0.72 0.068 0.76 0.063 0.94 0.128 0.75 0.305 0.76 0.157 0.64 0.138 0.64 0.155 0.86 0.180 E SE 0.54 0.085 1.20 0.101 1.16 0.066 0.90 0.066 1.12 0.081 1.47 0.176 1.15 0.080 1.11 0.099 2.06 1.09 0.088 0.17 2.97 0.463 3.32 0.486 5.06 0.536 7.19 1.358 7.14 2.857 5.05 0.248 4.79 1.033 7.71 3.558 4.87 0.969 L SE 141.0 4.65 163.2 2.59 170.5 3.99 165.8 2.81 177.8 1.49 181.6 2.10 179.9 4.33 180.2 3.76 200.0 191.0 9.71 118.0 155.1 3.27 161.9 0.75 169.0 1.79 166.9 0.84 179.5 4.50 186.3 9.19 182.4 4.33 182.0 1.00 179.3 3.35 W SE 31.8 3.17 47.5 2.66 53.4 4.39 42.3 2.72 51.5 2.63 52.8 3.27 49.6 3.59 42.3 2.43 73.0 48.3 6.12 16.0 41.1 2.19 47.3 0.62 49.7 1.22 49.0 1.37 49.0 4.00 64.8 14.79 58.0 6.33 51.5 0.50 47.3 2.17 196 Table B12. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Hell Diver 1 in 1986 and 1988-1989. 1988 1989 age n / 5 2 257.0 6 1 239.0 7 1 246.0 8 1 308.0 9 1 188.0 1 2 5 392.0 3 12 436.9 4 9 471.1 5 1 380.0 6 2 401.0 7 3 434.3 8 9 372.8 9 6 377.3 10 10 365.2 11 3 335.0 12 4 353.5 13 14 3 266.0 1 2 16 366.1 3 5 546.8 4 1 454.0 5 6 1 435.0 7 o o 9 1 290.0 10 2 258.0 11 4 262.8 12 1 252.0 SE 0 SE 14.00 2.60 0.034 2.86 3.02 3.41 1.63 31.46 0.39 0.108 32.08 1.40 0.121 59.60 1.62 0.191 1.08 67.00 1.61 0.435 37.75 2.53 0.570 22.96 3.12 0.433 58.94 2.17 0.418 34.44 2.39 0.423 32.05 1.32 0.216 79.80 1.61 0.303 37.75 1.43 0.192 37.92 0.66 0.108 25.79 2.57 0.423 1.91 1.53 3.78 8.00 2.62 0.425 80.53 2.47 0.861 2.65 E SE L 10.14 0.686 172.5 11.96 173.0 12.28 195.0 11.09 184.0 8.70 186.0 0.95 0.210 151.2 3.33 0.321 190.8 3.59 0.292 197.9 2.85 209.0 3.94 0.426 204.5 5.74 1.096 220.7 8.21 0.822 206.2 6.23 1.499 209.8 6.29 0.814 208.3 4.06 0.922 197.0 4.82 0.716 208.0 5.68 1.336 209.3 1.75 0.205 157.5 4.60 0.599 190.6 4.21 200.0 3.52 208.0 13.03 227.0 10.13 1.333 210.5 8.63 1.526 215.0 10.52 210.0 SE W SE 4.50 60.0 6.00 57.0 86.0 66.0 70.0 4.19 38.8 2.89 3.59 80.0 4.60 2.47 86.9 4.14 92.0 0.50 90.5 9.50 10.71 120.3 16.18 2.04 100.9 3.90 6.37 98.0 10.34 2.85 97.6 5.55 5.77 82.0 6.00 4.71 97.8 3.84 4.06 84.3 3.93 3.75 41.3 2.79 4.55 74.4 4.25 83.0 89.0 111.0 4.50 88.5 1.50 2.12 90.3 3.01 86.0 197 Table B13. Sample size at age (n), mean fecundity at age (/), mean ovary weight at age (0), mean egg weight at age (E), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Gem 3 in 1986 and 1988-1989. 1986 1988 1989 age n / SE 0 SE E SE L SE W SE 1 2 4 261.8 33.22 0.21 0.074 0.84 0.348 143.3 7.20 29.0 4.34 3 3 264.0 31.24 0.44 0.110 1.61 0.245 159.0 13.50 40.3 8.84 4 1 329.0 0.39 1.19 177.0 51.0 5 2 331.0 40.00 0.49 0.220 1.58 0.856 161.5 19.50 42.5 13.50 6 5 286.2 39.91 0.67 0.079 2.40 0.240 182.8 6.37 61.2 7.18 7 1 323.0 0.47 1.46 190.0 72.0 8 1 413.0 0.92 2.23 194.0 65.0 9 1 286.0 0.51 1.78 192.0 57.0 10 11 12 1 2 3 1 260.0 0.78 3.00 171.0 47.0 4 1 275.0 1.17 4.25 184.0 62.0 5 1 294.0 0.81 2.76 190.0 59.0 6 1 346.0 0.97 2.80 185.0 54.0 7 3 270.0 22.85 1.43 0.318 5.40 1.251 196.0 8.62 63.7 8 2 214.0 40.00 2.55 0.570 11.83 0.452 211.0 7.00 82.0 9 1 228.0 1.12 4.91 199.0 61.0 10 11 12 1 483.0 0.92 1.90 200.0 52.0 5.70 1.00 1 2 19 245.4 14.40 0.20 0.028 0.77 0.075 135.1 2.82 25.0 1.43 3 7 297.1 31.49 0.43 0.052 1.44 0.093 157.9 3.45 37.9 2.74 4 9 328.8 57.88 0.54 0.106 1.85 0.262 169.6 3.78 44.0 2.13 5 6 248.0 20.45 0.51 0.133 2.10 0.497 163.5 6.66 38.7 4.49 6 7 4 259.3 39.31 0.46 0.066 1.81 0.122 176.5 6.17 45.0 4.85 1 8 Q 3 201.7 14.90 0.48 0.048 2.41 0.208 204.7 1.33 58.7 4.84 y 10 11 12 198 Appendix C Sample size at age, mean length in mm at age, mean weight in g at age, and associated standard errors for the eight experimental lakes and the comparison lakes during 1986 or 1987 through 1989, as presented in Figures 7.1-7.9. experimental Figure Table page comparison Figure Table page Flower 7.1 CI 200 Matlock 7.1 C9 208 Wonder 3 7.2 C2 201 Wonder 2 7.2 C9 208 Fishgut 1 7.3 C3 202 Fishgut 3 7.3 C10 209 Dingleberry 7.4 C4 203 Midnight 7.4 C4 203 Hell Diver 3 7.5 C5 204 Hell Diver 1 7.5 C l l 210 Hell Diver 2 7.6 C6 205 Hell Diver 1 7.6 C l l 210 Par Value 7.7 C7 206 none Gem 2 7.8 C8 207 Gem 3 7.8 C l l 210 Fishgut pond 7.9 CIO 209 Fishgut 3 7.3 C10 209 199 Table CI . Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Flower in 1987-1989. age n L SE W SE n L SE W SE 1986 1987 0 1 28 97.1 1.77 10.5 0.51 2 126 149.4 0.98 37.4 0.71 3 103 173.3 1.04 57.5 1.09 4 89 186.7 1.38 70.2 1.87 5 92 199.5 1.64 86.8 2.91 6 56 202.7 2.14 88.7 3.34 7 18 202.0 4.59 84.2 6.88 8 19 206.2 3.45 89.5 4.64 9 6 208.2 9.25 94.2 19.83 10 6 210.7 6.08 93.3 9.68 11 8 214.6 4.99 91.9 7.74 12 3 233.7 17.15 134.7 37.12 13 14 15 1988 1989 0 1 155 105.4 1.03 12.9 0.36 237 121.5 0.68 19.8 0.32 2 112 153.6 1.06 38.8 0.82 177 166.4 0.83 50.4 0.75 3 113 180.0 1.18 60.8 1.41 53 187.8 1.35 72.5 1.76 4 55 188.9 1.71 69.4 2.57 38 195.9 2.00 80.3 3.17 5 37 195.3 3.56 79.2 5.00 31 200.8 1.83 85.7 2.89 6 34 198.9 2.39 78.4 3.62 19 205.6 2.44 88.7 3.70 7 47 201.5 2.49 81.0 3.89 23 209.6 2.69 96.2 4.20 8 14 217.7 6.20 109.2 12.27 11 206.5 3.55 91.1 4.66 9 12 201.2 3.21 76.2 4.81 4 207.5 7.49 89.5 3.66 10 3 194.7 6.23 68.7 8.57 5 209.6 2.58 91.0 2.61 11 1 255.0 119.0 1 195.0 76.0 12 3 195.0 5.00 73.3 6.17 13 2 218.5 3.50 93.5 0.50 14 1 222.0 99.0 15 200 Table C2. Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Wonder 3 in 1987-1989. age n L SE 1986 W SE n L SE 1987 W SE 0 1 23 89.6 1.09 7.3 0.32 2 138 133.9 1.36 27.6 0.82 3 66 159.8 1.77 45.3 1.29 4 31 172.0 1.64 54.8 1.45 5 28 182.2 2.12 62.9 1.91 6 107 195.7 0.77 76.2 0.90 7 8 34 33 198.9 199.5 1.88 2.40 78.2 76.9 2.66 3.09 9 23 199.0 1.96 78.3 2.47 10 17 200.1 3.28 78.2 4.93 11 12 197.8 2.27 71.3 3.13 12 6 202.5 4.81 70.2 6.49 13 2 196.5 0.50 58.0 2.00 14 2 193.5 6.50 61.5 1.50 15 1988 2 186.5 9.50 1989 53.5 7.50 0 1 59.0 2.0 27 59.0 1.24 2.3 0.10 1 106 111.2 1.43 16.3 0.67 392 125.5 0.86 22.2 0.43 2 39 152.0 1.38 40.7 1.07 58 168.9 1.43 52.4 1.32 3 28 166.8. 2.26 52.6 2.30 16 187.6 2.01 71.4 2.70 4 85 178.4 1.26 62.4 1.20 28 193.0 3.10 79.9 3.99 5 46 177.1 1.62 62.0 1.65 15 196.7 2.18 82.3 3.58 6 8 188.0 4.34 73.1 5.05 14 200.7 2.65 90.9 3.78 7 39 200.2 1.26 81.7 1.51 3 203.3 6.23 83.3 7.88 8 25 204.7 1.79 85.1 2.55 26 207.1 1.91 93.4 3.17 9 19 200.2 1.62 80.3 2.24 8 206.0 3.93 93.5 8.59 10 14 200.4 2.49 75.4 3.42 6 207.8 4.83 86.7 5.57 11 10 202.0 5.99 78.6 9.57 7 203.0 2.88 76.9 2.90 12 4 198.0 5.83 72.8 5.09 8 201.4 4.71 83.3 6.06 13 1 195.0 66.0 3 199.7 11.20 66.7 12.35 14 2 203.5 5.50 74.0 9.00 201 Table C3. Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Fishgut 1 in 1987-1989. age n L SE W SE n L SE W SE 1986 1987 0 1 2 3 98 155.8 1.79 38.4 0.75 4 26 166.1 2.02 46.3 1.89 5 38 175.9 2.00 51.5 1.59 6 16 184.9 2.23 55.1 2.54 7 9 187.8 2.87 47.9 3.55 8 1 192.0 58.0 9 3 188.0 11.36 48.3 4.98 10 1 191.0 37.0 11 12 1 202.0 64.0 13 14 15 1988 1989 0 13 63.5 1.78 2.7 0.26 1 36 95.7 2.82 10.6 0.98 215 128.9 0.88 24.3 0.48 2 32 137.8 1.58 30.0 1.85 81 164.0 1.13 49.1 0.90 3 116 162.3 0.76 45.3 0.65 12 175.2 2.04 58.3 2.16 4 100 168.8 0.80 50.4 0.89 33 188.9 2.46 73.8 4.39 5 27 172.7 1.69 51.4 1.51 46 187.6 1.48 70.1 1.83 6 29 177.3 1.60 53.1 1.93 15 187.7 2.35 67.9 2.95 7 22 185.3 2.09 58.0 3.03 20 191.9 2.16 75.0 3.27 8 7 189.3 4.01 58.1 3.04 9 187.2 3.81 '59.4 2.59 9 6 192.8 2.98 52.3 3.90 3 188.7 5.70 67.3 5.61 10 7 184.7 1.76 48.1 3.96 8 194.5 2.88 69.1 5.43 11 1 184.0 40.0 5 195.8 6.64 74.2 12.59 12 1 200.0 62.0 13 14 15 202 Table C4. Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Dingleberry in 1987-1989 and Midnight in 1989. age n L SE W SE n L SE W SE 1987 1988 0 2 48.0 10.00 1.5 0.50 1 23 92.0 2.11 8.4 0.68 96 111.8 1.50 17.1 0.82 2 225 155.9 0.76 41.7 0.60 251 159.7 0.76 48.0 0.64 3 120 189.9 1.14 74.8 1.21 266 187.2 0.69 75.4 0.84 4 37 195.3 1.83 78.4 2.28 103 196.8 1.30 86.3 1.69 5 83 201.8 1.34 85.1 1.81 20 194.1 2.19 81.4 2.62 6 53 202.3 1.47 86.6 1.96 54 203.0 1.74 88.9 2.05 7 25 205.5 2.70 85.2 3.43 26 204.5 2.56 93.2 3.09 8 32 206.9 2.43 88.9 3.69 22 207.0 2.49 90.9 3.19 9 23 209.0 2.16 87.4 3.04 17 210.4 2.32 88.2 2.95 10 20 214.7 2.14 97.3 3.72 16 216.5 2.96 97.6 5.44 11 1 218.0 83.0 17 211.7 1.85 90.5 2.26 12 2 206.5 8.50 78.5 11.50 3 212.7 3.28 87.3 5.55 13 1 223.0 103.0 14 1 211.0 78.0 15 1 208.0 85.0 1989 Midnight 1989 0 8 54.4 0.82 2.0 1 409 131.0 0.69 27.1 0.43 28 114.1 2.16 17.0 0.95 2 56 181.4 1.68 72.0 1.90 52 155.1 1.29 40.2 0.86 3 81 200.4 1.79 93.6 2.33 19 167.1 1.75 45.5 1.47 4 55 210.5 2.24 107.5 3.06 57 177.5 1.59 55.6 1.71 5 17 217.7 4.94 122.5 10.54 24 177.5 2.57 53.4 2.51 6 6 227.5 4.79 133.5 11.19 7 188.1 4.29 62.7 6.07 7 15 217.2 3.15 115.1 5.02 8 191.9 2.97 58.5 3.41 8 7 224.9 3.69 118.4 3.29 8 195.9 3.56 64.1 5.65 9 7 212.6 2.93 103.9 5.25 4 200.0 6.01 65.3 7.74 10 4 219.3 11.21 125.5 27.06 6 199.3 5.57 68.2 9.21 11 2 223.0 3.00 111.0 5.00 6 203.3 10.09 91.5 26.85 12 5 217.4 5.84 104.2 12.17 13 1 216.0 97.0 1 185.0 72.0 14 203 Table C5. Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Hell Diver 3 in 1987-1989. age n L SE W 1986 SE n L SE 1987 W SE 0 1 8 149.5 9.82 42.0 6.03 2 2 125.0 15.00 23.0 8.00 3 2 178.5 5.50 67.5 4.50 4 9 181.3 5.62 73.2 5.98 5 3 218.7 4.41 127.7 10.84 6 5 207.6 8.99 105.6 17.48 7 15 224.1 3.15 136.5 8.15 8 7 226.3 2.87 130.4 5.69 9 12 223.3 3.34 134.8 5.32 10 10 223.3 3.94 134.0 6.80 11 2 231.5 5.50 151.5 4.50 12 1 226.0 133.0 13 14 15 1988 1989 0 1 6 123.7 7.80 22.2 4.11 22 136.5 3.11 30.0 2.07 2 25 169.8 2.53 57.6 2.62 22 194.5 3.60 88.3 3.75. 3 4 187.3 10.38 78.0 15.59 6 211.5 10.94 124.5 11.14 4 3 203.3 13.28 99.3 16.17 2 220.0 12.00 121.0 23.00 5 1 233.0 136.0 2 245.5 16.50 132.0 14.00 6 3 232.3 10.93 137.7 11.79 7 1 246.0 162.0 2 264.5 5.50 209.0 1.00 8 3 243.3 1.76 143.7 3.53 3 251.3 2.85 164.3 11.20 9 11 240.7 3.74 156.8 9.66 1 260.0 188.0 10 7 239.4 2.53 141.4 9.80 2 247.0 0.00 159.5 3.50 11 2 230.0 14.00 142.0 25.00 4 238.8 3.84 134.0 14.61 12 2 245.0 0.00 151.0 24.00 1 234.0 125.0 13 14 15 204 Table C6. Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Hell Diver 2 in 1986-1989. age n L SE 1986 W SE n L SE 1987 W SE 0 1 4 79.0 1.61 4.7 0.63 2 1 118.0 18.0 12 126.0 4.06 22.5 2.37 3 1 173.0 57.0 1 189.0 86.0 4 3 168.0 4.49 53.0 4.36 5 28 184.0 2.25 ' 73.9 2.86 5 194.6 4.77 90.4 8.10 6 13 197.0 3.27 87.5 5.15 4 209.0 7.74 118.8 7.23 7 13 197.0 2.46 90.5 3.74 4 220.5 6.89 124.7 14.36 8 9 200.0 2.35 89.9 4.91 8 211.4 4.31 120.4 7.54 9 23 201.0 2.30 92.0 3.18 8 216.6 4.02 128.0 10.55 10 17 204.0 3.15 91.3 4.45 2 226.5 6.50 125.0 28.00 11 13 212.0 3.45 100.8 6.03 3 230.3 5.24 145.0 17.21 12 8 209.0 4.33 92.4 11.46 4 209.3 3.47 102.8 9.50 13 3 215.0 5.13 103.0 8.02 1 223.0 122.0 14 15 1988 1989 0 1 1 94.0 9.0 152 103.0 0.92 12.0 0.34 2 40 152.0 1.66 38.6 1.19 1 161.0 41.0 3 27 185.7 2.20 74.8 2.71 52 188.2 1.89 74.0 2.27 4 2 199.0 1.00 90.5 0.50 17 213.5 3.55 104.8 5.21 o 6 3 226.0 4.36 124.3 7.06 1 237.0 145.0 7 6 221.8 6.09 131.3 11.66 1 212.0 116.0 8 4 221.5 5.14 118.3 9.22 1 214.0 105.0 9 1 223.0 143.0 10 4 215.5 13.32 112.8 21.54 11 5 229.0 7.30 129.8 17.62 4 230.3 4.13 115.8 7.05 12 2 239.0 23.00 143.5 36.50 3 226.3 4.37 102.3 5.04 13 2 230.5 14.50 131.0 34.00 14 205 Table C7. Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Par Value in 1987-1989. age n L SE W SE n L SE W SE 1986 1987 0 7 58.4 0.65 2.0 0.00 1 82 100.2 1.10 10.6 0.44 2 89 162.3 1.31 50.0 1.28 3 36 192.1 1.40 82.9 1.76 4 9 196.9 3.53 91.9 4.01 5 23 204.8 1.58 98.8 2.31 6 101 205.7 0.83 101.4 1.14 7 26 208.0 2.08 101.7 2.77 8 70 208.8 1.00 104.0 1.52 9 54 210.5 1.40 104.4 1.99 10 86 208.3 1.09 102.5 1.67 11 37 212.7 2.06 105.9 3.25 12 6 232.0 9.23 112.7 3.80 13 2 220.5 0.50 102.5 5.50 14 15 16 1 211.0 72.0 1988 1989 0 10 54.7 1.43 1.9 0.10 1 369 119.1 1.02 21.0 0.51 406 110.6 0.73 15.8 0.30 2 " 82 169.5 1.09 55.4 1.01 302 167.5 0.71 52.7 0.68 3 35 187.5 1.75 75.0 1.65 27 192.2 2.59 79.2 3.36 4 25 205.9 2.05 94.8 2.83 12 208.5 "2.78 95.5 4.02 5 16 211.8 2.71 101.8 3.86 9 212.9 2.61 100.4 4.30 6 7 206.6 1.86 95.0 2.14 2 214.0 1.00 100.0 6.00 7 57 211.5 1.10 102.9 1.66 4 219.3 3.42 104.8 4.31 8 7 207.6 4.01 96.4 6.35 28 218.9 2.13 110.9 4.12 9 23 215.3 2.40 107.0 4.16 14 217.6 3.54 107.1 4.74 10 34 212.7 1.77 102.4 2.57 15 218.9 2.18 110.5 4.69 11 50 214.1 1.37 104.4 2.41 16 217.8 2.63 102.9 4.85 12 23 214.3 1.96 103.1 3.06 27 219.6 2.19 104.4 3.69 13 17 219.1 2.60 106.5 4.67 9 221.3 3.28 97.2 5.43 14 2 219.5 2.50 107.0 13.00 15 1 218.0 101.0 16 1 218.0 95.0 206 Table C8. Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Gem 2 in 1987-1989. age n L SE W SE n L SE W SE 1986 1987 9 10 11 12 13 14 15 0 1 81.0 5.0 1 40 125.8 1.58 20.5 0.78 2 10 132.4 2.87 24.6 1.72 3 4 77 194.5 0.84 75.6 0.91 5 6 7 2 194.0 4.00 78.0 10.00 1988 1989 0 1 1 128.0 23.0 2 38 180.7 3.43 70.9 3.99 1 170.0 49.0 3 4 244.3 5.36 184.5 7.37 4 5 27 219.0 2.64 117.4 5.45 6 6 267.7 5.56 231.8 11.67 7 8 9 10 11 12 13 14 15 207 Table C9. Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Matlock and Wonder 2 in 1988-1989. age n L SE W SE Matlock 1988 0 1 17 106.5 3.79 14.0 1.34 2 21 158.1 3.12 42.0 2.30 3 23 195.2 2.10 75.8 2.59 4 46 203.4 1.76 85.0 2.24 5 18 214.2 3.85 96.8 5.38 6 9 217.3 3.61 103.6 8.64 7 9 231.7 4.67 120.2 10.60 8 2 210.0 12.00 85.0 15.00 9 4 236.8 5.04 119.8 13.55 10 11 1 233.0 79.0 12 13 14 15 Wonder 2 1988 0 1 20 134.6 3.02 30.8 2.28 2 31 167.5 . 1.77 56.4 1.76 3 67 182.8 1.20 69.6 1.26 4 30 190.7 1.69 76.5 1.82 5 11 197.8 2.82 83.0 2.57 6 8 197.6 3.59 83.6 3.43 7 10 203.7 2.86 88.7 4.78 8 5 206.6 4.09 83.4 9.57 9 3 196.0 2.52 75.7 2.33 10 6 212.0 6.87 87.0 9.18 11 12 13 14 15 n L SE W SE Matlock 1989 15 122.1 2.30 20.9 1.08 12 174.2 2.90 59.8 3.27 1 189.0 73.0 2 200.5 5.50 85.0 11.00 5 213.6 5.26 102.0 6.41 2 211.0 1.00 102.5 1.50 1 224.0 122.0 1 207.0 88.0 Wonder 2 1989 45 150.0 1.91 40.6 1.59 16 180.6 2.07 68.9 2.01 18 195.1 1.46 82.2 1.89 22 197.7 1.97 85.1 3.16 16 200.4 2.82 92.4 4.24 8 195.4 3.20 80.6 4.40 2 212.0 8.00 96.0 2.00 5 206.6 3.52 88.6 5.71 5 212.6 4.21 95.4 9.36 3 216.7 9.13 107.3 12.81 4 205.3 4.27 79.5 2.50 1 206.0 84.0 208 Table CIO. Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Fishgut pond in 1987 and Fishgut 3 in 1987-1989. age n L SE W SE Fishgut pond 1987 0 1 2 3 2 286.5 '9.50 327.5 25.50 4 5 2 308.0 4.00 433.5 3.50 6 2 312.5 1.50 459.0 2.00 7 Q 1 308.0 418.0 O 9 10 1 267.0 327.0 11 12 13 14 15 Fishgut 3 1988 0 1 4 94.3 3.35 7.7 2.66 2 12 139.3 1.85 29.9 1.23 3 38 164.0 1.94 47.7 1.83 4 21 174.0 2.70 57.1 3.20 5 5 171.2 5.87 52.0 9.97 6 15 180.1 2.50 57.5 2.64 7 26 182.7 1.43 55.7 2.23 8 13 182.3 3.29 54.5 3.91 9 12 182.9 2.46 48.8 3.01 10 3 194.3 2.96 70.3 2.19 11 4 190.8 6.87 49.0 4.38 12 13 14 15 n L SE W SE Fish gut 3 1987 2 94.5 6.50 8.5 2.50 1 155.0 38.0 2 172.5 3.50 47.5 0.50 4 177.5 4.73 50.5 2.06 3 195.7 9.94 68.0 7.37 4 199.5 8.77 87.5 18.66 2 186.0 9.00 49.0 9.00 1 190.0 64.0 Fishgut 3 1989 25 121.2 2.92 19.9 1.50 36 157.0 1.57 42.8 1.19 16 164.6 1.27 47.9 0.85 29 175.2 1.99 55.6 2.15 16 175.7 2.87 56.7 2.77 3 177.0 3.61 49.3 2.33 5 182.6 8.00 61.8 11.83 7 184.1 3.21 62.3 5.18 4 181.3 1.18 49.5 2.18 8 181.5 2.04 52.1 2.47 1 196.0 55.0 209 Table C l l . Sample size at age (n), mean length at age (L), mean weight at age (W), and associated standard errors (SE), for Hell Diver 1 and Gem 3 in 1988-1989. age n L SE W SE HeU Diver 1 1988 0 1 18 99.4 2.34 10.9 0.89 2 33 149.5 2.90 38.3 2.27 3 22 187.0 3.09 75.5 3.73 4 12 197.0 2.35 86.0 3.52 5 4 204.5 7.79 97.5 10.14 6 7 210.9 4.45 104.0 7.58 7 7 213.6 5.53 108.1 8.10 8 21 209.2 2.02 103.8 3.11 9 11 210.3 3.46 103.5 6.24 10 22 212.7 2.51 108.0 5.13 11 7 206.9 5.93 97.7 8.33 12 6 206.0 4.31 94.2 6.58 13 2 203.5 7.50 92.5 8.50 14 5 207.6 3.31 89.6 4.91 n L SE W SE Hell Diver 1 1989 38 103.3 2.08 12.2 0.80 35 158.2 2.42 42.7 1.78 13 190.0 2.43 75.1 2.60 4 200.5 3.66 85.0 5.96 2 215.0 0.00 99.5 0.50 1 208.0 89.0 1 227.0 111.0 3 215.3 5.49 100.3 11.86 7 220.0 5.80 102.0 8.33 2 231.5 21.50 144.0 58.00 Gem 3 1988 0 1 1 111.0 13.0 2 1 147.0 30.0 3 1 171.0 47.0 4 1 184.0 62.0 5 2 190.0 62.5 3.50 6 1 185.0 54.0 7 5 196.2 4.75 64.8 3.29 8 3 207.3 5.46 80.0 2.0S 9 3 203.3 4.84 63.3 2.85 10 1 214.0 80.0 11 12 1 200.0 52.0 13 14 Gem 3 1989 2 116.5 8.50 16.5 2.50 36 136.9 1.91 ' 26.0 0.98 16 153.3 2.23 35.3 1.52 13 166.5 3.41 42.4 2.16 11 170.5 5.46 44.4 3.89 5 177.8 4,95 46.4 4.01 4 201.8 3.07 60.5 3.88 1 210.0 68.0 210 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0100445/manifest

Comment

Related Items