Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Interspecific interactions affecting the foraging behavior of chum salmon fry (Oncorhynchus keta) Tompkins, Arlene Marie 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 T65.pdf [ 7.69MB ]
Metadata
JSON: 831-1.0101144.json
JSON-LD: 831-1.0101144-ld.json
RDF/XML (Pretty): 831-1.0101144-rdf.xml
RDF/JSON: 831-1.0101144-rdf.json
Turtle: 831-1.0101144-turtle.txt
N-Triples: 831-1.0101144-rdf-ntriples.txt
Original Record: 831-1.0101144-source.json
Full Text
831-1.0101144-fulltext.txt
Citation
831-1.0101144.ris

Full Text

Interspecif ic Interactions Af fec t ing T h e Foraging B e h a v i o r O f C h u m S a l m o n F r y (Oncorhynchus keta) Arlene Marie Tompkins B.Sc , St. Francis Xavier University, 1979 M . S c , University of Manitoba, 1982 A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of D O C T O R O F 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 April 1991 © Arlene Marie Tompkins, 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 The University of British Columbia Vancouver, Canada DE-6 (2/88) Abstract Interactions between fish utilizing nearshore habitats of the Fraser River estuary were investigated by field observations and laboratory experiments. Chum salmon fry (Oncorhynchus keta) were the most abundant salmonid captured between April and June. Non-salmonid species captured included: threespine stickleback (Gasterosteus aculeatus), prickly sculpin (Coitus asper), and peamouth chub (Mylocheilus caurinus). Potential predators included: prickly sculpin, and northern squawfish (Ptychocheilus oregonensis), but few had been feeding on fish. Chum fry fed predominantly on surface insects but the proportion of benthic prey in the diet increased over time. Stickleback shared the greatest diet overlap with chum fry. Interactions between two dissimilar prey, chum fry and threespine stickleback, and a predator, cutthroat trout (Oncorhynchus clarki) were investigated in the labora-tory. Prey response to hungry and satiated predators was related to the degree of risk. Although attack rates by trout on chum and stickleback were similar, trout captured more stickleback than chum, but consumed both prey at similar rates. I tested the hypotheses that prey foraging efficiency is reduced in the presence of a predator and increased in the presence of alternate prey. When alone, chum fed on surface Drosophila and mid-water Daphnia, while stickleback fed on benthic Tubifex and Daphnia. The feeding efficiency of chum increased in the presence of stickleback and decreased in the presence of trout. Hatchery and wild chum showed opposite dietary shifts in the presence of trout. Hatchery chum shifted from surface to mid-water feeding and the number of fish feeding significantly decreased. Wild chum fed at the surface, at significantly decreased feeding rates. In the presence of stickleback and trout the feeding behaviour of chum was similar to that when chum were alone. Stickleback feeding behaviour was not affected by presence of trout or ii chum. Chum and stickleback detected Daphnia faster than Drosophila or Tubifex, and chum responded to Daphnia significantly faster than stickleback. Foraging time per item was significantly less for chum than stickleback. Habitat use by fish prey was investigated in the presence and absence of trout and alternate prey. Wild chum shifted from mid-water to the surface in the presence of trout, but returned to mid-water when stickleback were present. Stickleback fed in bottom habitats regardless of the presence of trout or chum. When prey were confined to specific depths in the water column, trout attacked chum more frequently than stickleback in all locations and attacked both prey more frequently within 24 cm of the substrate. Movement by prey did not affect the attack rate. When given a choice between a food-rich open water habitat and a food-deficient vegetated habitat in the presence of trout and alternate prey, chum and stickleback used vegetated refugia significantly more in the presence of trout. Alternate prey presence decreased the proportion of chum but increased the proportion of stickleback using vegetation. Behavioural responses to avoid predation significantly reduced the foraging ef-ficiency of prey. Chum showed stronger responses to trout than stickleback. The presence of stickleback reduced the effect of predation on foraging efficiency. Possible explanations for the positive effect of stickleback on chum feeding efficiency were ex-perimentally examined including: social facilitation, reduced intraspecific competition, and the calming influence of stickleback on chum behaviour ("dither"). The results suggest that stickleback have a calming influence on chum behaviour and that mixed species feeding groups may reduce intraspecific competition. iii Table of Contents Abstract ii List of Tables viii List of Figures xi Acknowledgements xiv 1 General Introduction 1 Objectives and Organization of the Thesis 4 2 Prel iminary Survey of Species Interactions in the Fraser River Estuary 6 Introduction 6 Methods 8 Study Area 8 Sampling 10 Results 13 Species abundance 13 Evidence of predation . . . 13 Feeding behaviour 14 Discussion 31 Species abundance 31 Potential competitors 31 Potential predators 34 3 Antipredator Strategies of C h u m Salmon F r y and Stickleback 38 Introduction 38 iv Methods 40 General Experimental Methods 40 Collection of Fish 41 Experiment 1: Prey Evaluation of Risk 43 Experiment 2: Differential Prey Susceptibility 46 Results 48 Experiment 1: Prey Evaluation of Risk 48 Experiment 2: Differential Prey Susceptibility 48 Wild versus hatchery chum 48 Attack rates on each species 49 Prey response to predators 54 Capture and consumption rates 54 Discussion 56 Experiment 1: Prey Evaluation of Risk 56 Experiment 2: Differential Susceptibility to Predators . . . . 57 Antipredator strategies 58 4 Effects of Predation on the Foraging Behaviour of Chum Fry and Stickleback 62 Introduction 62 Methods 65 Experiment 3: Foraging Efficiency 65 Experiment 4: Foraging Time 67 Experiment 5: Foraging Site 70 Results 73 Experiment 3: Foraging Efficiency 73 Diet Composition 73 Percentage Feeding 76 Feeding Rates 78 Feeding Indices 79 v Experiment 4: Foraging Time 80 Experiment 5: Foraging Site 80 Discussion 84 Effects of trout on chum foraging behaviour 85 Possible reasons for the differences between hatchery and wild chum 89 Positive effects of stickleback presence on chum foraging behaviour 91 Effects of other species on stickleback foraging behaviour 93 Relationship between foraging behaviour and antipreda-tor strategies 93 5 Habitat Use 94 Introduction 94 Methods 97 Experiment 3: Effect of predation on foraging site within a habitat . 97 Experiment 6: Effects of predation on refugia use 98 Results 101 Experiment 3: Effect of predation on foraging site within a habitat 101 Experiment 6: Effects of predation on refugia use 101 Discussion 105 Experiment 3: Effect of predation on foraging site within a habitat 105 Experiment 6: Effects of predation on refugia use 106 vi 6 Mechanism of Enhanced Feeding Rates 108 Introduction 108 Methods 109 Results 112 Discussion 116 Stickleback as dither 116 Reduced intraspecific competition 117 Social facilitation 118 7 Conclusions 120 Species Interactions 122 Literature C i ted 122 vii List of Tables 2.1 Species of fish taken in beach seine samples from three sites in the lower Fraser River in 1982-1984 . 1 6 2.2 Frequency of occurrence and percent of catches of chum fry also containing stickleback, by sampling date and by seine haul, during February-June, 1982-1984, at three sites on the lower Fraser River 20 2.3 Number (n), mean fork length (FL), associated standard deviation (sd), and mean feeding indices (FI, food weight/body weight x 100) for four species sampled from the North Arm site at 4 h intervals over a 24 h period (April 30-May 1, 1984) 24 2.4 Estimates of diet overlap using Morisita's Measure for four species during April and May, 1984 28 3.1 Mean attack rates (mean number of attacks by cutthroat trout per individual prey per hour), standard errors (SE), number of replicates (n), total attacks, percent attacks resulting in cap-ture, and percent attacks ending in consumption in single and mixed species treatments (Experiment 2) 49 3.2 Mean attack frequency by cutthroat trout on chum and stickleback in three locations in the water column for eight replicate ex-periments (Experiment 2) 55 4.1 Mean, minimum and maximum fork length (mm), associated stan-dard error (se), and total number of each species recovered and examined in each treatment in Experiment 3 (1—alone, 2—with predator, 3—with alternate prey, 4—with alternate prey and predator) 74 4.2 Statistical probabilities from results of a two-way A N O V A under the null hypothesis of no effect of presence of trout and alter-nate prey on the percentage of fish feeding, feeding rate, and feeding indices of hatchery chum, wild chum, and stickleback (Experiment 3.) 78 viii 4.3 Mean response time (sec), mean foraging time (sec), and associated standard deviation (sd) for three food types consumed by " n " individual hatchery chum and stickleback (Experiment 4) 81 4.4 Results of a Mann-Whitney U test (MWU) comparing frequency of attacks by cutthroat trout on wild chum and stickleback and results of Kruskal-Wallace H tests comparing mean number of attacks by cutthroat trout on wild chum and stickleback confined to four locations in the water column (Experiment 5). . . 82 4.5 Nonparametric multiple comparisons of attack frequency by cut-throat trout on wild chum confined to four locations in the water column (b-bottom, 1-lower, u-upper, and s-surface, Experiment 5) 82 4.6 Mean and median number of attacks by cutthroat trout on stickle-back confined in the two upper locations versus the two lower locations of the water column (Experiment 5) 83 4.7 Comparison of the frequency chum and stickleback in each location were attacked first in n replicate experiments of foraging site vulnerability (Experiment 5) 83 4.8 Kruskal-Wallace H test (nonparametric A N O V A by ranks) on mean number of attacks on wild chum confined to four locations in the water column, when all prey were swimming (Experiment 5) 83 4.9 Nonparametric multiple comparisons of attack frequency on swim-ming chum in four locations in the water column (b-bottom, 1 1-lower, u-upper, and s-surface, Experiment 5) 84 5.1a Habitat use by wild chum in the presence and absence of stickleback and trout (Experiment 3) 102 5.1b Three-way factorial A N O V A of stickleback presence, trout presence and position on habitat use by wild chum (Experiment 3). . . . 102 5.2a Habitat use by stickleback in the presence and absence of chum and trout (Experiment 3) 103 5.2b Three-way factorial A N O V A of chum presence, trout presence and position on habitat use by stickleback (Experiment 3) 103 ix 5.3a Mean percentage of chum and stickleback using open water habitat, versus vegetated refugia, and associated standard error (se) in the presence and absence of alternate prey and predators (Experiment 6) 104 5.3b Results of an A N O V A on the effects of species, alternate prey presence, and predator presence on the percentage of fish in open water versus vegetation (Experiment 6) 104 6.1 Comparison of mean feeding rates (number of food items per hour) of chum fry in four treatments (1-with trout, 2-with trout and dither, 3-alone, and 4-with trout and stickleback) 114 x List of Figures 2.1 Daily counts of juvenile chum salmon migrating past Mission, 100 km upstream from the Fraser River estuary in 1982-84 7 2.2 Location of the three sampling sites on the North and South Arms of the Fraser River, British Columbia: 1—North Arm site, 2—Dow site, and 3—Green Slough site 9 2.3 Seasonal changes in the logarithm of catch per effort (log 1 0(CPE)), an index of fish abundance, between 1982-1984 in the North Arm of the Fraser River -,17 2.4 Seasonal changes in the logarithm of catch per effort (log 1 0(CPE)), an index of fish abundance, between 1982-1984 at the Dow site, South Arm of the Fraser River 18 2.5 Seasonal changes in the logarithm of catch per effort (log 1 0(CPE)), an index of fish abundance, between 1982-1984 at Green Slough on the South Arm of the Fraser River 19 2.6 Seasonal changes in mean lengths of chum fry, potential competitors (a), and potential predators (b) collected between 1982-1984 at the North Arm site, North Arm of the Fraser River 21 2.7 Seasonal changes in mean lengths of chum fry, potential competitors (a), and potential predators (b) collected between 1982-1984 at the Dow site, South Arm of the Fraser River 22 2.8 Seasonal changes in mean lengths of chum fry, potential competitors (a), and potential predators (b) collected between 1982-1984 at the Green Slough site, South Arm of the Fraser River 23 2.9 Comparison of the relative percentage (by numbers) of prey in the diets of juvenile chum salmon sampled during February-May 1983 at the North Arm site, North Arm of the Fraser River 25 2.10 Comparison of the relative percentage (by numbers) of prey in the diets of juvenile chum salmon sampled during February-May 1983 at the Dow site, South Arm of the Fraser River 26 2.11 Comparison of the relative percentage (by numbers) of prey in the diets of juvenile chum salmon sampled during March-May 1983 at the Green Slough site, South Arm of the Fraser River. . . . 27 xi 2.12 Comparison of the relative percentages (by numbers) of prey items in the diets of juvenile chum salmon, threespine stickleback, peamouth chub, and prickly sculpin in Apri l 1984 29 2.13 Comparison of the relative percentages (by numbers) of prey items in the diets of juvenile chum salmon, threespine stickleback, peamouth chub, and prickly sculpin in May 1984 30 3.1 The apparatus used in Experiments 1-3 45 3.2 Comparison of wild chum and stickleback feeding rates when at-tacked by hungry and by satiated trout (Experiment 1) 50 3.3 Comparison of attack rates on hatchery and wild chum in single and mixed species treatments (Experiment 2) 51 3.4 Comparison of attack rates by cutthroat trout on stickleback when with hatchery versus wild chum and in single versus mixed species treatments (Experiment 2) 52 3.5 Comparison of attack rates by cutthroat trout on chum fry and stickleback in single versus mixed species treatments (Exper-iment 2) 53 4.1 The apparatus used in Experiment 4 to measure the foraging time of chum and stickleback feeding on three types of invertebrates. . . . 69 4.2 The apparatus used in Experiment 5 to measure the relative at-tack rate by trout on chum and stickleback confined to four locations in the water column: bottom, lower water column, upper water column, and surface 71 4.3 Composition of the diet of a) hatchery chum, b) wild chum, and c) stickleback under four treatments: 1—prey alone; 2—prey with trout; 3—prey with alternate prey; and 4—prey with trout and alternate prey (Experiment 3) 75 4.4 The percentage of fish feeding, feeding rate, and feeding index (weight of stomach contents/body weight x 100) of hatchery chum, wild chum, and stickleback under four treatments: 1— alone; 2—with trout; 3—with alternate prey; and 4—with trout and alternate prey (Experiment 3) 77 5.1 The apparatus used in Experiment 6 to determine the effect of predation on habitat use by chum fry and stickleback 99 xii 6.1 The apparatus used in Experiment 7 to quantify changes in chum foraging behaviour when stickleback were provided as dither. . . I l l 6.2 Comparison of attack rates by cutthroat trout on hatchery chum (mean number per individual per hour) in three treatments: chum with trout, chum with stickleback and trout (Experi-ment 2), and chum with trout and stickleback as dither fish (Experiment 7) 113 6.3 Comparison of mean diet composition of hatchery chum in four treatments: 1—with predator; 2—with predator and stickle-back as dither; 3—alone; and 4—with predator and stickle-back (Experiments 3 and 7) 115 xiii Acknowledgements I thank my supervisor at U B C , Dr. C C . Lindsey, for the opportunity to conduct this study and for his generous support for longer than either of us ever anticipated. Dr. C D . Levings, my supervisor at the West Vancouver D F O , introduced me to the Fraser River estuary and the ecology of juvenile salmon and I am grateful for his support and advice along the way. I thank other members of my advisory committee, Drs. Liley, McPhail, and Northcote for constructive criticisms of the thesis manuscript. Drs. Levings, McPhail, and Northcote provided generous loans of equipment. Expert field assistance was provided by: Diana 'Mud Woman' Campbell, John Kinaham, Cindy Movold, Pat Murray, Grant Thompson, and an endless list of friends and fellow students from 'the Institute' particularly: Gordon Haas, Don Hall, and Ric Taylor. I am grateful to the staff of the Department of Fisheries and Oceans who pro-vided information and alternate sources of chum fry, especially: Morley Farwell, Kim West, Bob Brown and Murray MacDonald who operated the live trap at the Mamquam spawning channel, and the staff of the Chehalis Hatchery. I thank Don Hall and the staff of the Biological Data Center: Alistair Blachford, Susan Ertis, and Joerg Messer for assistance in data manipulation and for answering my endless computer questions. Dory Kufflick prepared the figures of the experimental apparatus. I am indebted to Carl Walters for the loan of his Compaq during the write-up phase. I thank my support team: Linda Berg, Gayle Brown, Simon Courtenay, Chris Foote, Gordon Haas, Walt Klenner (yet another thesis), Mike Lapointe, Lucila Lares, Molly Nevin-Haas, and Eva Samuelsson. Your friendship, support, and encouragement xiv will not be forgotten. My husband, Don Hall, shared this thesis with me, provided unceasing support and encouragement, and kept the ogres away. This work was supported by an NSERC operating grant to Dr. C C . Lindsey, by the Department of Fisheries and Oceans through Dr. C D . Levings, and by an NSERC post-graduate scholarship. xv 1. General Introduction Traditionally, ecologists have been interested in understanding patterns of species abundance and diversity, and the factors that structure populations and com-munities. The relative importance of predation and of competition as factors struc-turing communities has been debated for decades but recently this debate has become recognized as overly simplistic (Mittlebach 1986). The direct effects of predators on prey and on competing prey have been documented. Predation can structure com-munities through selective removal of particular species or sizes of prey (Mittlebach 1984; Werner 1986; Sih 1987). Predation may reduce interspecific competition, either by lowering demand for resources or by removing the dominant competitor (Paine 1966; Connell 1975; Mittlebach 1986). Predators also can have profound non-lethal effects on prey and prey interactions. Competitive interactions may be intensified if vulnerable size classes or species are concentrated into a common refuge (McCabe et al. 1983, Werner et al. 1983a). Werner et al. (1983a) and Sih (1987) have documented the effects of predators on prey feeding rates and have demonstrated that intra- and interspecific variation in spatial and temporal patterns are due to differential suscep-tibility to predators. More attention needs to be given to predator mediated changes in diet, habitat and lifestyles that affect interacting species (Sih 1987). Optimality theory has been used widely in the study of animal behaviour and has proven valuable as an explanatory and inferential tool for interpreting foraging strategies. Classical optimal foraging theory (Schoener 1971; Pyke et al. 1977; see Stephens and Krebs 1986 for review) assumes that selection favors animals that maxi-mize their average rate of energy intake, which is presumed to be positively correlated with fitness (Krebs et al. 1983; Stephens and Krebs 1986). The main justification for using optimality models is the notion that natural selection acts as a maximizing process; animals that do not behave optimally have reduced fitness and are selected against. 1 One method of increasing fitness is to increase the rate of food intake, a short term correlate of fitness (Krebs et al. 1983). Costs associated with foraging are time and energy expended, and frequently risk of mortality due to predation (Mittlebach 1986; Sih 1987). Early optimal foraging models assumed that risk of predation was a function only of foraging time and was independent of the resource utilized (Krebs et al. 1983; see Stephens and Krebs 1986 for review) . However, if risk varies with food type, the decision of when and where to forage should depend on the degree of risk. Behavioural ecologists have approached the trade-off problem in recent foraging models either by: (1) maximizing energy gains subject to time constraints imposed by competing activities or by (2) using survival as a common currency. It appears that animals make decisions based on trade-offs between costs in predator avoidance and benefits (maximizing energy gain as expressed in growth rate, Werner and Hall 1988, Gilliam 1982; feeding rate, Godin 1986; or habitat use, Werner and Hall 1988). Recent dynamic optimization models have considered alternative behaviours such as foraging, predator avoidance and reproduction in the sequential decision making process (Man-gel and Clark 1986; McNamara and Houston 1986). Although average rate maximization models have been used extensively in the study of prey selection and patch selection (see Stephens and Krebs 1986 for review), these models have been heavily criticized for their shortcomings (Krebs et al. 1983; Dill 1983). Firstly, the basic models are static and do not take into account the state of the forager (e.g. hunger level). Secondly, they assume that the forager has perfect knowledge of food types, food patches and profitabilities. Thirdly, the optimality criterion is overly simplistic. The models attempt to combine many factors into a single currency and express the consequences of alternative strategies in terms of long term rate of energy maximization. Since all animals operate under constraints when acquiring energy, animals may rarely be able to achieve optimal feeding rates in the wild. Constraint has been defined 2 by Stephens and Krebs (1986) as all factors that limit the animal's feasible choices and the pay-offs that may be obtained. Recent evidence on behavioural responses of fish to their predators and on how individuals balance predation risk and foraging gain, suggest that risk of predation and other constraints (e.g. territory defense, mate acquisition) may conflict with energy maximization in importance in shaping foraging behaviour (Milinski and Heller 1978; Dill et al. 1981; Gilliam 1982). One of the recurring themes in recent literature on behavioural ecology is the study of animal decision making (see Dill 1987 for review). It is assumed that animals are capable of assessing their environment and are sufficiently flexible to choose the al-ternative that maximizes their fitness (lifetime reproductive success). All behavioural alternatives have associated costs and benefits and animals should choose the alter-native with the greatest relative gain. Under certain circumstances the forager must trade-off foraging against some other activity at the cost of reduced energy intake. The optimum "trade-off" or "decision rule" will depend on the animal's state, e.g. hunger level, and the current environmental states, e.g. food abundance or predation risk (Godin 1990). Optimal foraging theory assumes animals have complete knowledge of their envi-ronment, but animals must sample to gather information regarding food and predators. There is a low cost associated with sampling food but a high cost associated with sam-pling predators. If the animal has a low probability of surviving an encounter with a predator the cost is death. To paraphrase the life dinner principle (Dawkins and Krebs 1979) the cost of missing dinner is less than the cost of being dinner. Recent evidence suggests that animals can assess variation in energetic profitability and risk of predation among available habitats, and can trade-off expected foraging gains and perceived risk of mortality in deciding where to feed (see Milinski 1986; Dill 1987; and Lima and Dill 1990 for reviews). Considerably less attention has been given to foraging decisions within a habitat under risk of predation. If the forager incurs different risks 3 of predation by feeding in different parts of the habitat or on different food types, then the risk of predation may be an important determinant of feeding rate and diet choice (Godin 1990). In this study I experimentally investigated the behavioural responses of dissim-ilar prey to risk of predation. In particular I examined changes in foraging behaviour and habitat use by prey to determine if responses to predation were consistent with trade-offs between energy intake and predator avoidance. Objectives and Organization of the Thesis In this study I have examined, through both field observations and labora-tory experiments, some of the interactions between a predator, cutthroat trout (On-corhynchus clarki), and two prey species, threespine stickleback (Gasterosteus aculea-tus) and chum salmon fry (Oncorhynchus keta). My main objectives were: (1) to compare the effects of predation on the foraging behaviour of dissimilar prey, (2) to identify antipredator defenses of the prey, and (3) to examine the effects of the availability of alternate prey on the outcome of predation. To determine if prey traded-off foraging efficiency and predator avoidance, I examined the response of prey to predation and the effects of the availability of alternate prey by manipulating species combinations and observing changes in antipredator behaviour, diet, foraging mode, foraging rate, and habitat use. In Chapter 2, I introduce the study animals, describe their habitat and feeding behaviour in the Fraser River estuary, and discuss the evidence for niche overlap and predation in the wild. Each of Chapters 3-6 deals with laboratory experiments designed to test a particular aspect of predator-prey interactions. Each presents a review of the pertinent literature, a description of the experiments and their results, and a discussion. 4 Chapter 3 examines prey selection by predators, prey evaluation of risk, and dif-ferential antipredator strategies of prey. In Experiment 1, I manipulated the intensity of predation and observed the feeding rates of dissimilar prey to test the hypothesis that prey response is proportional to the degree of risk of predation. In experiment 2, I examined the relative susceptibility of chum fry and stickleback to predation by trout and observed prey antipredator strategies. In Chapter 4, I focus on how the feeding efficiency of prey is affected in the presence and absence of a predator and alternate prey. I used laboratory Experiment 3 to compare changes in the composition of the diet, feeding rates, and feeding indices in the presence and absence of trout and of alternate prey. I tested the hypothesis that prey foraging efficiency is reduced under predation risk. To determine if changes in foraging behaviour were due to characteristics of food type that affected vulnerability, I measured the foraging time of the two prey species when feeding on different food taxa (Experiment 4) and the frequency of attacks by trout on prey at different locations in the water column (Experiment 5). In Chapter 5, I evaluate how habitat utilization by prey is affected by the presence of a predator and by the availability of alternate prey. I compared the position of prey within a habitat (Experiment 3) and use of vegetated refugia by dissimilar prey in the presence and absence of trout and alternate prey (Experiment 6) to determine if chum and stickleback trade-off habitat use and predator avoidance. In Chapter 6, I examine potential mechanisms for enhanced chum salmon feed-ing rates when stickleback are present, including: (1) facilitation of food detection, (2) facilitation of predator detection, (3) reduced intraspecific competition, and (4) reduced fright due to the dithering effect of stickleback (Experiment 7). Chapter 7 summarizes the results of this study and discusses the effects of predation on prey foraging behaviour and on the interactions of species in a complex community. 5 2. Preliminary Survey of Species Interactions in the Fraser River Estuary Introduction The Fraser River estuary is the largest estuary on the Pacific coast of Canada, receiving drainage from 233,100 km 2 of the interior of British Columbia (Hoos and Packman 1974). About 38 species of fish inhabit the lower Fraser system, of which 19 are non-migratory freshwater species, 14 are anadromous species, and 5 are semi-migratory (Hoos and Packman 1974). The importance of estuaries as rearing grounds for juvenile salmonids and non-salmonids has been documented by several researchers (McCabe et al. 1983; Levy and Levings 1978; Levy et al. 1979; Healey 1980; Simen-stad et al. 1982). Typically these shallow areas support rich populations of benthic organisms and receive substantial fallout of terrestrial insects from riparian vegetation. Many fish species may use estuaries simultaneously for feeding, resting, and possibly as sanctuary from predators. Chum salmon fry emergence and downstream migration from Fraser River rear-ing grounds begin in mid March. Chum are thought to move to the estuary immedi-ately, with peak numbers passing Mission (Federal Fisheries and Oceans enumeration station), 100 km upstream from the estuary, in April and May (Figure 2.1). Evidence reported for the Nanaimo, Squamish, and Fraser River estuaries suggests chum may linger for days or weeks in the lower rivers to feed (Healey et al. 1977; Levy and Levings 1978; and Levy et al. 1979). Mark-recapture studies of chum fry in the Nanaimo and Fraser estuaries (Healey 1979; Levy et al. 1979) have demonstrated that chum may spend up to three weeks in the inner estuary and are localized in their movements. 6 4000 5/31 Figure 2.1 Daily counts of juvenile chum salmon migrating past Mission, 100 k m up-stream from the Fraser River estuary in 1982-84 (data supplied by Canada Department of Fisheries and Oceans, New Westminster, British Columbia). 7 Dunford (1975) reports chum salmon fry present in Fraser marsh habitats several weeks after most migrants have passed Mission. Starvation and predation are recognized as the two factors responsible for the high mortality rates of juvenile salmonids (Suboski and Templeton 1989). Exposure to intense predation, competition, or both during estuarine residence could reduce fry growth and survival. Growth and survivorship are positively correlated in many fish (Ware 1975a, 1975b; see Werner 1986, for review). Any factor that slows individual growth prolongs the time spent in vulnerable size classes and increases the probability of death (Gilliam 1982; Mittlebach and Chesson 1987). Although much attention has been given to the importance of marsh habitats for juvenile salmonids (Levy and Northcote 1981, 1982; Dunford 1975) little interest has been focused on other estuarine habitats. The objectives of this field study were to: (1) document the relative abundance of salmonids and non-salmonids in riparian nearshore habitats, (2) determine the potential predators of juvenile salmonids, and (3) evaluate food consumption and potential interspecific competition between chum and non-salmonids. Methods Study Area Fish were collected from three nearshore sites along the North and South Arms of the lower Fraser River, British Columbia (Figure 2.2). These sites were character-ized by sedge (Carex lyngbyei) covered riverbanks and abundant riparian vegetation including willows (Salix sp.), cottonwood (Populus sp.), and alder (Alnus sp.). The Fraser River Estuary is a mixed semi-diurnal estuary (Thomson 1981). All study sites were largely freshwater but were subject to tidal fluctuations. During high tides and 8 FRASER RIVER ESTUARY USA Figure 2.2 Location of the three sampling sites on the North and South Arms of the Fraser River, British Columbia: 1—North A r m site, 2—Dow site, 3—Green Slough site. 9 peak freshet the riverbanks can be inundated, while at low tide or during reduced river discharge the vegetated banks may be completely exposed. A salt wedge intrudes into the North, Main, and Middle Arms as far as Deas Island during low discharge periods and only penetrates to the mouth of the river during freshet (Hoos and Packman 1974). Substrate composition ranges from fine silt to small gravel. All sites were accessible by boat and were sufficiently level to permit unobstructed seining. The North Arm site is located at the junction of River Road and No. 8 Road in Richmond, B .C. (123° 01' N, 49° 12' W). This site comprises the nearshore river habitat and a pond fed by access channels to the North Arm of the Fraser River. The pond may become completely inundated during high spring tides, but during periods of low tide and reduced river discharge it is often isolated from the river. The substrate is composed of fine silt, clay, and sand. The Green Slough site is near Ladner Marsh, B .C. (123° 04' N, 49° 07' W) bordering the Ladner exit of Highway 99 South. This site is fed by the waters of Deas Slough and was the most heavily vegetated site studied. The substrate is fine silt and detritus. The Dow site is near the Delta Bar Fishing Park off the Dow Chemical service road in Delta, B . C . (123° 00' N, 49° 09' W). This site has a shallow bay formed by Dow's offshore loading facility and the beach of the fishing bar. The substrate particle size ranges from small gravel to fine silt. The bay has a dense growth of sedges that alternately flood and drain with the tidal cycle. Sampling Fish were sampled monthly at most sites from June-October 1982, January-August 1983, and April-July 1984, using beach seines. Samples for June 1982-March 1983 were collected by employees of the Department of Fisheries and Oceans, West 10 Vancouver, B . C . The most extensive sampling was carried out in 1983 when collections were made at least once per month between January and August. A l l collections were made during high tide. The larger seine (15 m x 2 m, 1 cm mesh wings, 3 mm mesh bunt) was set with a 3.6 m inflatable boat with a 10 H P outboard. Seines were set from shore, towed into the river about 3 m, and returned to shore by making a semi-circle approximately 8 to 10 m upstream. The average area covered by each seine was 20 m 2 . A smaller seine (5 m x 1 m, 1 cm mesh) was used to sample the pond complex at the North A r m site. This seine was set by holding one end in place, towing the other end around the pond, and pursing both ends onto shore. The approximate area covered by the small seine was 3 m 2 . Fish captured by seining were anesthetized in MS-222 (methanesulfonate salt) or 2-phenoxyethanol (Bell 1964, Sigma Chemical Company, St. Louis, Missouri), iden-tified, enumerated and fork length measured to the nearest 1.0 mm. A minimum of ten chum were selected randomly and preserved in a 10% formalin solution for later diet analysis. Potential predators were dissected for evidence of piscivory. Once the remaining fish recovered from anaesthetic they were returned to the river. The relative abundance of the most common species at each site was assessed by comparing the mean number captured per standard seine haul at each sampling date. To identify spatial and temporal patterns in feeding behaviour, stomach contents were examined from chum collected at each site between February-May 1983. Stomach contents were removed, identified (to family where possible) and enumerated using a binocular dissecting microscope. Gut contents of the four most common species (chum fry, threespine stickleback, peamouth chub (Mylocheilus caurinus), and prickly sculpin (Cottus asper)) were ex-amined in Apr i l 1984 to determine diel feeding activity. Fish were collected by beach seining, every four hours over a 24 h period, and immediately preserved for later diet 11 analysis. Fish were weighed (to the nearest 0.001 g) and measured (to the nearest 1.0 m m fork length). Stomach contents were removed, blotted dry, and weighed to the nearest 0.0001 g (wet weight). A feeding index (food weight/body weight x 100) was calculated as an indicator of food consumption. Stomach contents of chum fry, stickleback, peamouth chub and prickly sculpin collected in Apri l and May 1984 were removed, identified and enumerated. Diet overlap between species was compared using Morisita's Measure (Morisita 1959). Morisita's Measure is calculated by the formula: c _ 2 X ) Pij Pik  E w t e ) + E » ( % f r ) where C\ = Morisita's index of overlap between species j and k, pij = proportion resource i is of the total resources used by species j, Pik = proportion resource i is of the total resources used by species k, riij = number of prey i found in the stomachs of species j, riik = number of prey i found in the stomachs of species k, Nj,Nk = total number of prey items consumed by species j, k (JVj = re, j ) . Values of C\ range from an upper limit of 1 when the proportion of food items in the diet are the same for both species to 0 when the two species have no food items in common. One disadvantage with overlap measures is that there is no universally accepted level of statistical significance. However for my purposes overlap values are used as a qualitative method of comparing relative overlap between species. Smith and Zaret (1982) compared seven commonly used measures of overlap and recommended 12 Morisita's Measure as the overlap measure with the least bias under changing numbers of resources and sample sizes. Results Species abundance Both salmonids and non-salmonids used nearshore habitats of the lower Fraser River (Table 2.1). Relative species abundances at each site are shown in Figures 2.3-2.5. Chum salmon fry, threespine stickleback, chinook salmon fry, redside shiner, northern squawfish, peamouth chub, and prickly sculpin were the most common species captured in the seines. Seasonal variations in occurrence and size of fish species were evident. Chum fry were the most abundant salmonid and a numerically important component of the catch from March to July. Chum first appeared in the beach seines in February at the North Arm and Dow sites (Figures 2.3, 2.4), and in March at the Green Slough site (Figure 2.5). A few chum were captured as late as August at the North Arm and Dow sites. Stickleback, peamouth chub, and prickly sculpin were consistently associated with chum at all sites. Stickleback were also taken on 86-100% of the dates that chum were collected at the three sampling sites (Table 2.2). The coincidence of chum fry and stickleback in individual seine hauls was greatest at the North Arm site; stickleback occurred in 62% of the seines with chum. Stickleback occurred less frequently with chum at the Dow and Green Slough sites-, 21% and 41% of the seines, respectively. Evidence of predation Figures 2.6-2.8 present mean lengths for each species by sample date and site. Chum fry increased in length as much as 20 mm during the sampling period (February-July). Chum fry and stickleback were the smallest prey available to potential predators 13 until May when other juveniles (peamouth chub, flounder) began to appear in the samples. Sculpins, capable of preying on chum fry, were captured as early as February. Northern squawfish were not taken until May and June. Evidence of predation on chum or on non-salmonids was scarce. Although north-ern squawfish and prickly sculpin large enough to handle chum fry and similar sized non-salmonids were captured in nearshore beach seines, few had been feeding on fish. Three incidents of piscivory were observed during the sampling program: the stom-ach of one northern squawfish contained an unidentified fish, the stomach contents of a Dolly Varden comprised 11 eulachon and four stickleback, and one prickly sculpin contained a juvenile stickleback. Feeding behaviour The feeding behaviour of chum fry collected between February and May 1983 varied between sites and over time within sites. In February, chum collected at the North Arm (Figure 2.9) and Dow (Figure 2.10) sites were feeding predominantly on surface insects (dipterans and collembolans). The proportion of benthic prey (e.g. dipteran larvae and harpacticoids) in the diet increased through March and May at all sites. Chum collected at Green Slough in May (Figure 2.11) fed almost entirely (81% composition by number) on benthic amphipods and harpacticoids (Figure 2.11). Species occurrence and feeding behaviour varied daily as well as seasonally. -Chum fry, stickleback, and peamouth chub were present in all seine hauls but fed predominantly during daylight hours (Table 2.3). Sculpins were absent from samples taken between 0800 and 1600 h. The diet of chum was compared to that of other species in April and May 1984. Unlike chum, which fed predominantly on surface prey in April, stickleback, peamouth chub and sculpin consumed mostly benthic prey and minor amounts of mid-water prey 14 (Figure 2.12). Harpacticoids and chironomid larvae were common in the diets of all species. In May (Figure 2.13), chum included proportionately more mid-water prey types in their diet as did stickleback. Peamouth chub and prickly sculpins fed entirely on benthic food types. Measures of dietary overlap between species are presented in Table 2.4. During peak chum migration (April and May), chum fry shared the greatest overlap with stickleback. Overlap increased over time; the measure of overlap in May (0.490) was greater than twice the value measured in April (0.225). Except for comparisons with sculpins, measures of overlap between all species increased from April to May. The greatest diet overlap was exhibited by stickleback and sculpins in April (0.821) and stickleback and peamouth chub in May (0.532). 15 Table 2.1 Species of fish taken in beach seine samples from three sites in the lower Fraser River in 1982-84. Species Name Oncorhynchus gorbuscha Oncorhynchus keta Oncorhynchus kisutch Oncorhynchus nerka Oncorhynchus tshawytscha Oncorhynchus tshawytscha Oncorhynchus clarki Salvelinus malma Prosopium williamsoni Thaleichthys pacificus Spirinchus thaleichthys Hypomesus pretiosus Catostomus macrocheilus Cyprinus carpio Richardsonius balteatus Ptychocheilus oregonensis Mylocheilus caurinus Ictalurus nebulosus Gasterosteus aculeatus • Pomoxis nigromaculatus Leptocottus armatus Cottus asper Platichthys stellatus Entosphenus tridentatus Common Name pink salmon fry chum salmon fry coho salmon smolts sockeye salmon smolts chinook salmon fry chinook salmon smolts cutthroat trout Dolly Varden mountain whitefish eulachon longfin smelt surf smelt largescale sucker carp redside shiner northern squawfish peamouth chub brown bullhead threespine stickleback black crappie staghorn sculpin prickly sculpin starry flounder lamprey ammocoetes 16 2.0 1.5 1.0 0.5 0.0 2.0 1.5 1.0 0.5 0.0 2.0 1.5 1.0 0.5 0.0 2.0 a -S 1-° O 0.5 0.0 2.0 1.5 1.0 0.5 0.0 2.0 1.5 1.0 0.5 0.0 2.0 1.5 1.0 Q5 0.0 a) chun b) threespine stickleback c) Chinook d) peamouth chub A , e) redside shiner f) northern squawfish g) prickly sculpin M J J A S O N D 1982 J F M A M J J A S O N D 1983 J F M A M J J A S O 1984 Figure 2.3 Seasonal changes in the logarithm of catch per effort ( log 1 0 (CPE) ) , an index of fish abundance, between 1982-1984 in the North A r m of the Fraser River. Data for June 1982 through Apr i l 1983 were supplied by the Department of Fisheries and Oceans, West Vancouver, British Columbia. 17 1.0 0.5 0.0 2.0 1.5 1.0 0.5 0.0 2.0 1.5 1.0 0.5 0.0 2.0 1.5 CL 1.0 • 0.5 0.0 2.0 1.5 1.0 0.5 0.0 2.0 1.5 1.0 0.5 0.0 2.0 1.5 1.0 0.5 Q0 a) chum b) threespine stickleback c) chinook d) peamouth chub e) redside shiner f) northern squawfish g) prickly sculpin S O 1982 M M J 1983 Figure 2.4 Seasonal changes in the logarithm of catch per effort ( log 1 0 (CPE) ) , an index of fish abundance, between 1982-1984 at the Dow site, South A r m of the Fraser River. Data for June 1982 through A p r i l 1983 were supplied by the Department of Fisheries and Oceans, West Vancouver, British Columbia. 18 a) chum b) threespine stickleback c) chinook d) peamouth chub e) redside shiner f) northern squawfish g) prickly sculpin M A S 1982 M A 1983 M Figure 2.5 Seasonal changes in the logarithm of catch per effort ( log 1 0 (CPE)) , an index of fish abundance, between 1982-1984 at Green Slough on the South A r m of the Fraser River. Data for June 1982 through A p r i l 1983 were supplied by the Department of Fisheries and Oceans, West Vancouver, British Columbia. 19 Table 2.2 Frequency of occurrence and percent of catches of chum fry also containing stickleback, by sampling date and by seine haul, during February-June, 1982-1984, at three sites on the lower Fraser River. The statistical probabilities were generated by a x2 test under the null hypothesis of no association between chum and stickleback presence. by sampling date site number of dates, chum with stickleback % number of dates, chum without stickleback X2 P value North Arm Dow Green Slough pooled sites 17 6 6 29 94 100 86 91 1 0 1 2 P < 0.05 by seine haul site number of hauls, chum with stickleback number of hauls, chum without stickleback X2 P value North Arm Dow Green Slough 65 12 7 62 21 41 40 45 10 0.05 > P > 0.01 P < 0.005 0.10 > P > 0.05 20 200 150 • chum * threespine stickleback • chinook • peamouth chub • redside shiner * starry flounder E E ¥ 100 IB c CD CD 50 • • : A - - f - A ft*'-.I-* .*' * A a) • 1^ • northern squawfish x prickly sculpin 150 c IB c co CD I - x-50 x b) M J J A S O N D 1982 J F M A M J J A S O N D 1983 J F M A M J J A S O 1984 Figure 2.6 Seasonal changes in mean lengths of chum fry, potential competitors (a), and potential predators (b) collected between 1982-1984 at the North A r m site, North A r m of the Fraser River. When more than one distinct size class appears in the same year classes are plotted separately. 21 200 150 * chum * threespine stickleback * chinook • peamouth chub * redside shiner * starry flounder a) E E P 100 .S? c CD CD E 50 *' . 4> • ...#- * -• • •A - • -A * • • •-• northern squawfish x prickly sculpin b) 150 c c ro CD 50 S O 1982 M M 1983 Figure 2.7 Seasonal changes in mean lengths of chum fry, potential competitors (a), and potential predators (b) collected between 1982-1984 at the Dow site, South A r m of the Fraser River. When more than one distinct size class appears in the same year classes are plotted separately. 22 200 150 P 100 IB c co CD 50 • chum A threespine stickleback • chinook • peamouth chub • redside shiner •*• starry flounder a) . • •* • • .A --•• .A * - A . - . l : ! : : - - * - * • northern squawfish x prickly sculpin b) 150 E E x • g> 100 IB c co CD • X. 50 X M A S 1982 A 1983 M Figure 2.8 Seasonal changes in mean lengths of chum fry, potential competitors (a), and potential predators (b) collected between 1982-1984 at the Green Slough site, South A r m of the Fraser River. When more than one distinct size class appears in the same year classes are plotted separately. 23 Table 2.3 Number (n), mean fork length (FL) , associated standard deviation (sd), and mean feeding indices (FI, food weight/body weight x 100) for four species sampled from the North A r m site at 4 h intervals over a 24 h period (Apri l 30-May 1, 1984). The insert figure shows tidal height at time of sampling. species 0400 0800 1200 1600 2000 2400 chum n 5 1 10 10 8 5 F L 40.0 45.0 . 42.1 42.0 40.1 41.4 sd 1.9 2.1 4.1 3.9 1.3 F I 0.029 0.038 0.051 0.036 0.030 0.018 stickleback n 5 1 7 6 6 3 F L 36.0 26.0 46.4 39.7 40.0 42.0 sd 4.2 8.6 4.2 4.6 4.6 FI 0.028 0.024 0.035 0.034 0.035 0.017 peamouth chub n 5 2 7 7 6 5 F L 108.5 27.5 50.1 69.3 128.7 61.4 sd 40.4 0.7 8.0 26.9 15.5 17.6 F I 0.000 0.021 0.024 0.012 0.009 0.000 prickly sculpin n 5 0 0 0 7 5 F L 103.8 67.1 95 sd 16.3 28.4 13.1 FI 0.029 0.006 0.016 24 O 3 CO i o "c CD JD Coltembola Chironomldae adults CNronomidae pupae % Dipteran adults Dpteran pioae other insect adults other Insect remains Arachnida Cyclopoida Calanoida Chydorid sp. Daphnid sp. other Cladocera Harpacticotda Ostracoda Oligocfiaeta Isopoda Amphipoda Neomysid sp. Gastropoda Chironomidae larvae other Dpteran larvae Hemipteran nymphs Trjchoptera Ephemeroptera Lepidoptera nymphs other insect larvae unidentified eggs February (n-4 fish, t-50) March (n=78 fish, t-614) M a y (n-18 fish, t-343) I I I I J I L J L J I I I 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 60 diet composition Figure 2.9 Comparison of the relative percentage (by numbers) of prey in the diets of juvenile chum salmon sampled during February-May 1983 at the North A r m site, North A r m of the Fraser River. Percentages are based on n individuals feeding on t total food items. 25 February (n-6 fish, t-20) March (n=39 fish, t-324) May (n-18 fish, t-263) CD o a 3 CO J5 "co 5 i E o CD Collembola Crironomidae adults CNronomidae pupae Dipteran adults Dpteran pupae other insect adutts other insect remains Arachrtda Cyclopoida Calanoida Chydorid sp. Daphnid sp. other Cladocera Harpacticoida Ostracoda Oligochaeta Isopoda Amphipoda Neomysid sp. Gastropoda Crironomidae larvae other Dpteran larvae Hemipteran nymphs Trichoptera Ephemeroptera Lepidoptera nymphs other insect larvae unidentified eggs 10 20 30 40 50 60 diet composition Figure 2.10 Comparison of the relative percentage (by numbers) of prey in the diets of juvenile chum salmon sampled during February-May 1983 at the Dow site, South A r m of the Fraser River. Percentages are based on n individuals feeding on t total food items. 26 March (n=4. t-57) May (n=7. t-174) CD o CO _CD "co I "D "E CollemDola CHronomldae adults CNronomidae pupae Dpteran adults Dipteran pupae other insect adults other insect remains Arachnida Cyclopoida Calanoida Chydorid.sp. Daphnid sp. other Cladocera c CD -O Harpactcoida Ostracoda Oligochaeta isopoda Amphipoda Neomysid sp. Gastropoda Chironomidae larvae other Dpteran larvae Hemipteran nymphs Trichoptera Ephemeroptera Lepidoptera nymphs other insect larvae unidentified eggs 80 0 20 diet composition Figure 2.11 Comparison of the relative percentage (by numbers) of prey in the diets of juvenile chum salmon sampled during March-May 1983 at the Green Slough site, South A r m of the Fraser River. Percentages are based on n individuals feeding on t total food items. v. 27 Table 2.4 Estimates of diet overlap using Morisita's Measure for four species during A p r i l and May, 1984. Ten fish of each species were examined. A p r i l species stickleback peamouth sculpin chum 0.225 0.068 0.141 stickleback 0.220 0.821 peamouth 0.081 May species stickleback peamouth sculpin chum 0.490 0.346 0.092 stickleback 0.532 0.220 peamouth 0.299 28 CD O 05 CO _CD "CD $ I "E c CD Collembola Crironomidae addts Crironomidae ptpae Dpteran adults Dpteran pupae other insect adults other insect remains Arachrida Cyciopoida Calanoida Chydorid sp. Daphnid sp. other Cladocera Harpacticoida Ostracoda Oligochaeta Isopoda Ampripoda Neomysid sp. Gastropoda Chironomidae larvae other Dpteran larvae Hemipteran nymphs Trichoptera Ephemeroptera Lepidoptera nymphs other insect larvae unidentified eggs chum (n=42. t-1066) stickleback (n=26. t-1736) peamouth chub (n-13. t-317) prickly sculpin (n-15. t-192) J L J I J I I L J I I I L J I I I L 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 60 diet composition Figure 2.12 Comparison of the relative percentages (by numbers) of prey items in the diets of juvenile chum salmon, threespine stickleback, peamouth chub, and prickly sculpin in A p r i l 1984. Percentages are based on n individuals feeding on t total food items. 29 CD O b _CD "co 5 o jc "c CD JD Collembola Crironomidae adults Crironomidae pupae Dpteran adults Dpteran pupae other insect adults other insect remains Arachrida Cyclopoida Calanoida Chydprid sp. Daphnid sp. other, Cladocera Harpacticoida Ostracoda Oligochaeta Isopoda AmpNpoda Neomysid sp. Gastropoda Crironomidae larvae other Dpteran larvae Hemipteran nymphs Trichoptera Ephemeroptera Lepidoptera nymphs other insect larvae unidentified eggs chum (n=10. t=223) stickleback (n=10. t-369) peamouth chub (n=10. t-179) prickly sculpin (n-10. t-151) j L J L J I I i i i j L J L 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 60 diet composition Figure 2.13 Comparison of the relative percentages (by numbers) of prey items in the diets of juvenile chum salmon, threespine stickleback, peamouth chub, and prickly sculpin in May 1984. Percentages are based on n individuals feeding on t total food items. 30 Discussion Both size and diets will affect competition and predation among fish in estuarine habitats (Murphy et al. 1988). A fish's size will influence its vulnerability to predators while its diet will determine potential competitors. However, size and diet are not independent. Survival will depend on size (Parker 1971; Hargreaves and LeBrasseur 1986) and size depends on diet and the ability to forage efficiently. Competition for food may decrease growth rates and prolong time vulnerable to predators. Species abundance The abundance and diets of fish species captured by beach seines in nearshore habitats of the lower Fraser River provide evidence that salmonids and non-salmonids share this habitat. Species specific patterns of seasonal variation in abundance were consistent at all sites. During the period of chum migration stickleback, peamouth chub, and sculpins were the most abundant non-salmonids in the samples. Stickleback frequently occurred in seine hauls with chum fry, particularly at the North arm site. Potential competitors Reduced feeding indices of chum, stickleback, and peamouth chub at night (Ta-ble 2.3) suggest these species are visual feeders that actively forage during the day. This diurnal pattern of feeding by chum agrees with the findings that salmonid fry and fingerlings remain close to shore during daylight (away from predators) and move into the current at night to migrate downstream (Neave 1955; Bams 1969; Reimers 1971). Manzer (1976) and Worgan and FitzGerald (1981) also report a diurnal pattern in the feeding behaviour of threespine stickleback. The absence of prickly sculpin in seine sets between 0800-1600 h suggests they are nocturnal foragers that are not active during the day. 31 Chum displayed a seasonal shift in diet composition. Although surface insects comprised a large part of chum diets during all months, chum gradually included more benthic prey types in the diet over time. Amphipods only formed a major component of the diet of chum sampled at Green Slough in May, which may have been due to differences in prey abundance, prey availability, or in chum feeding preferences between sites. Amphipods generally inhabit areas with abundant vegetative detritus and debris characteristic of the Green Slough substrate. Similar food types were observed in the diets of chum from the Fraser River estuary (Levy et al. 1979; Levy and Northcote 1981) and Washington State estuaries (Simenstad et al. 1982). Seasonal variation in Fraser estuary chum diets is consistent with that reported for chum in the Squamish River and the Nanaimo River estuaries (Levy and Levings 1978; Healey 1979). Possible explanations for changes in the composition of chum diets over time include: (1) ontogenetic shifts in diet preference, (2) interspecific or intraspecific competition, and (3) seasonal changes in resource abundance (Wiens 1977). The ability to switch food items depending on prey availability has important implications for juvenile fish inhabiting estuaries. The abundance and availability of prey in estuaries likely undergo rapid changes in response to changes in abiotic factors (i.e. turbidity, temperature, salinity, runoff, and discharge). The ability to switch from one prey to another provides an advantage in terms of growth and survival in unpredictable environments (Wiens 1977). If resource limitation is occurring in the Fraser River estuary it is presumably temporary since the period of spatial (and consequently dietary) overlap between chum and non-salmonids is brief. During this period exploitation of novel prey items might be limited by previously learned foraging behaviour. Chum fry tend to develop training biases for food types they have previously consumed (LeBrasseur 1969; Levy 1977). 32 Consequently, when resources are limited chum might suffer a short term reduction in food supply and possibly growth rates. Threespine stickleback are reputed to feed on a wide range of food types (Manzer 1976; Thorman 1983; Walsh and FitzGerald 1984) and I observed a high level of diet overlap with all species. Sambrook (1990) found that stickleback in Ladner Marsh, in the Fraser River estuary, fed on food items in proportion to their relative abundance in the environment. Stickleback were most similar to chum in size and diet and conse-quently represent their most likely competitor if resources become limited. Obrebski and Sibert (1976) also report recurring high indices of diet overlap between chum fry and threespine stickleback in the Nanaimo River estuary. They suggest that although stickleback may be an important competitor for food during periods of resource limi-tation they may actually increase chum feeding efficiency when resources are abundant by disturbing inaccessible prey on the substrate surface. Although the observed dietary overlap between species suggests the existence of potential competition for preferred food types if resources become limited, it also may represent the common use of an abundant resource. The data needed to assess competition quantitatively in this study are lacking. Resource limitation and its effects on consumer fitness in this environment would be difficult to measure, since changes in resource abundance and availability as well as consumer abundance are occurring over time. However, evidence of food limitation has been demonstrated in several other estuaries. Healey (1979) found that the productivity of chum fry in the Nanaimo Estu-ary was limited by food supply and food intake by chum in Hood Canal, Washington was lowest during the period that fry were most abundant (Simenstad et al. 1980; Simenstad et al. 1982). Simenstad and Salo (1980) suggested that food limitation in estuarine and nearshore environments of Hood Canal, Washington may be a major determinant of the production of chum salmon. They attributed the lower survival 33 of early outmigrants compared to late outmigrants to competition for resources since early migrating fry precede spring increases in zooplankton. Potential predators Potential predators of juvenile fish captured in the samples included: chinook smolts, and adult Dolly Varden, cutthroat trout, prickly sculpin, and northern squaw-fish. One possible reason so few predators were sampled is that the seining technique may have been less efficient at capturing faster adult fish. Alternatively, juveniles may be using nearshore areas because they serve as a refuge from predation. Preliminary sampling indicated larger fish were captured more frequently in deeper sets and were associated with particular areas within sites. Similarly, McCabe et al. (1983) studied the interrelationships between juvenile salmonids and non-salmonids in the Columbia River estuary and concluded that the estuary represented a sanctuary from predators. They attributed the low incidence of predation to sampling at low tide and suggested that predation might be greater if predators move into intertidal areas at high tide. I did not sample at low tide and consequently I could not compare the relative abun-dances of predators at different tidal phases. It is widely recognized that predation by birds, chars, trout, sculpins and north-ern squawfish plays a significant role in juvenile salmonid mortality in freshwater and estuaries (see Mace 1983 for review). However, since the impact of predation tends to be time and location specific, it is generally inappropriate to conclude that results or effects from one site are representative of other areas. Slaney et al. (1985) suggest there is potential for predatory interactions be-tween salmonids wherever small juveniles or fry coexist with larger fish. Chum salmon populations in southern Puget Sound declined to 10% of historic levels in 1960 and 1970 probably due to increased releases of hatchery coho stocks in spring during chum migration (Johnson 1974). Northcote et al. (1979) and Levy et al. (1979) reported 34 juvenile salmonids in the diets of wild chinook smolts (>50 mm) in the lower Fraser River. Resident and anadromous cutthroat trout have been considered major predators on chum fry (Pritchard 1936; Hunter 1959; Armstrong 1971; Sumner 1972). Lannan (cited in Pearcy et al. 1989) noted that most predation on chum salmon by fish during the downstream migration to Netarts Bay, Oregon, was by cutthroat trout. Simenstad et al. (1982) found that cutthroat trout in Washington state estuaries had a higher occurrence of juvenile salmonids in their diets than had other predators such as cod, sculpins, and coho or chinook smolts. The importance of sculpins as predators on juvenile salmonids in freshwater has received much attention and appears dependent on the system studied. Most investigations conclude that sculpin infrequently prey on salmon fry (see Mace 1983 for review). But other studies estimate that mortality due to sculpin predation may range from 1.3-3.9% (Patten 1971a) to 10-28% (Hunter 1959; Moyle 1977) of the total population. Hunter (1959) claims that sculpins greater than 5 cm in length are sufficiently large to take salmon fry as food. Sculpins were one of the commonest potential predators in the habitats sampled and as nocturnal foragers (Table 2.3) could have a large impact on the mass downstream migrations of chum at night. Patten (1971b) suggests that predation rates on fry may be higher at night, especially during moonlight nights and that certain salmonid species are more susceptible to predation with chum and pink fry being most vulnerable (Patten 1975). Young northern squawfish often compete with juvenile salmon and trout for food, and become increasingly important as predators aus they grow larger (see Brown and Moyle 1981 for review). Squawfish can consume salmonid fry when they reach a length of 100 mm (Ricker 1941). Thompson (1959) studied the food habits of northern squawfish in the lower Columbia River after hatchery smolt releases. Thompson found that out of 1272 countable fish in the stomachs of squawfish, 87% were salmon smolts. Foerster and Ricker (1942) attributed a total increase of 3.8 million sockeye smolts to 35 a predator (squawfish, char and coho) removal program at Cultus Lake, B .C. , between 1936-38. However, Ricker (1941) considered the average consumption of sockeye by individual squawfish to be less than that by the various salmonid predators: Dolly Varden, cutthroat trout and coho salmon. The vulnerability of chum salmon fry to predation has been related to their size (Parker 1971; Healey 1982a; Hargreaves and LeBrasseur 1986). Hiyama et al. (1972) and Fresh et al. (1980) found that losses during freshwater migration could be reduced by over 50% if fry were released when their mean size was equal to, or greater than, 50 mm. Pearcy et al. (1989) report results from experimental releases of different sized (1.0-6.5 g) chum fry into Netarts Bay in 1986 (for comparison, the maximum sizes of fry used in this study were 1.0 g and 0.5 g for hatchery and wild chum, respectively). They found that large (6.5 g) fry did not use the estuary but immediately migrated to the ocean. Adult returns from these large fry were 3-4 times higher than those of smaller (1.0 g) fry. These studies suggest that larger fry will enjoy higher survival. Larger fish may have a greater ability to avoid predation, and if predators select smaller fish, delaying migration from a refuge habitat until a critical size is attained may enhance survival. Growth of juvenile salmon in estuaries is usually inferred from changes in mean size although these estimates are subject to bias (LeBrasseur and Parker 1964; Healey 1979; Murphy et al. 1988). The average length of chum fry increased over the sampling period (February-June) suggesting fry are feeding and growing during their estuarine residence. My data are insufficient to conclude whether the increase in chum length was due to growth in the estuary, or selective removal of smaller fry, or due to larger fish migrating later. However the greatest abundance of fry captured in the estuary occurred in June (Figure 2.3-2.5), one to two months after the peak migration of chum past Mission (Figure 2.1), suggesting chum were spending considerable time in the estuary. Healey (1982a) reported that seaward migration was size related; larger fry 36 move seaward first and most chum in southern British Columbian estuarine habitats are gone by mid-July. Chum are thought to move from shallow estuarine waters into open neritic waters at a length of 45-55 mm (Simenstad and Salo 1980; Healey 1980, 1982b; Myers and Horton 1982). Myers (1980) also noticed a distinction in habitat use by salmonid fry related to size; larger fry occupied deeper, faster channels while smaller fry were found inshore. She attributed these differences in behaviour to small fry rearing along the beaches, and larger fry in channels migrating actively to the ocean. If large chum emigrate more quickly than small fry then my observed increases in chum size may be underestimated. Results from this study and others listed above suggest the importance of re-source limitations and species interactions in estuarine fish communities. Given the number and variety of fish species using the lower Fraser River for feeding and shelter from predators, there is a high probability of interactions between juvenile salmonids and non-salmonids. Competitive effects are generally size dependent (Larkin and Smith 1954) and mediated by size-specific growth rates that in turn strongly affect size-specific predation rates (Mittlebach 1984, 1986). Not only does competition re-duce growth rates but it also prolongs the time that prey are vulnerable to predation (Gilliam 1982; Werner 1986; Mittlebach and Chesson 1987). It is apparent that inter-actions between competition and predation are complex and will depend on charac-teristics of predators, prey and their environment. The complexity and uncontrollable nature of the environment in the Fraser River estuary made it necessary to resort to laboratory observations under controlled conditions, in order to isolate and quantify behavioural responses to interspecific interactions. In the following chapters I examine possible interactions involving competition and predation, using laboratory experi-ments with chum salmon fry and threespine stickleback as potentially competing prey species, and cutthroat trout as predators. 37 3. Antipredator Strategies of Chum Salmon Fry and Stickleback Introduction Predation represents a powerful selective force for the evolution in prey of an-tipredator adaptations that reduce mortality. Antipredator adaptations co-evolve with searching strategies by predators; the prey become more difficult to catch and eat, and the predators become more efficient in searching and obtaining prey (Harvey and Greenwood 1978; Hobson 1979; Sih 1984). At the same time, most predators are prey for other predators, so that their foraging behaviour also should reflect trade-offs between feeding and predator avoidance. This idea of trade-offs is central to the argu-ment that behavioural traits are a product of natural selection (Davies and Houston 1984; Lucas 1987); those animals that make the appropriate behavioural responses survive. Some behavioural traits are less fixed and may be responses to current envi-ronmental conditions but in either case the optimal behaviour for animals is to choose the alternative that maximizes their fitness. Lima and Dill (1990) review evidence that animals under risk of predation make decisions based on trade-offs between costs due to predation and benefits to be gained by engaging in certain activities. This implies that the process of natural selection has resulted in animals that possess the ability to assess the risk of being preyed on and incorporate this information into decision making. Antipredator defenses can be defined as any trait that reduces predation risk (Sih 1987). Predatory acts may be subdivided into discreet units: encounter, attack, capture and consumption. Antipredator strategies (see Edmunds 1974; Keenleyside 1979; Morse 1980 for reviews) can be grouped into two main categories depending on whether they occur before or after a predator-prey encounter: 38 1) avoidance occurs before an encounter and reduces the chance of encounter (e.g., an animal changes habitat or remains in a refuge) 2) escape occurs after an encounter and decreases the chance prey will be attacked, captured and consumed. Predator avoidance can involve more than the simple act of escaping (Charnov et al. 1976; Jeffries and Lawton 1984). Avoidance may involve long-term shifts in habitat use (Sih 1979, 1982, 1984; Werner et al. 1983a), time of activity, or movement (Stein and Magnuson 1976; Sih 1982, 1984; Clark and Levy 1988; Levy 1987). Examples of avoidance include vertical migrations by zooplankton, horizontal migrations by fish, crypsis, and countershading. The costs of remaining in the vicinity of a predator are less for cryptic and morphologically protected animals because they are less likely to be detected or attacked, and they might actually increase their vulnerability by moving. When attacked, prey can use defenses to reduce the probability of capture such as flight to shelter or protean escape. Protean behaviour has been denned by Humphries and Driver (1970) as behaviour that is sufficiently unsystematic to prevent a predator from predicting in detail the positions or actions of the prey. Some fish react to approaching predators by schooling (reviewed by Keenleyside 1979); groups under attack reduce interindividual distances and become more integrated in their group movements. Schooling results in earlier detection of predators through increased vig-ilance and reduced probability of capture through predator confusion or dilution of risk. Prey can reduce the success of predators by making handling difficult through such morphological traits as size or shape. Ydenberg and Dill (1986) suggest animals make decisions when confronted by predators based on the relative costs of fleeing versus staying. Natural selection should favor precise antipredatory responses because mistakes can be costly (Sih 1986). For prey that ignore dangerous predators the cost is death. But unnecessary avoidance of predators that pose little risk entails costs in lost energy and lost foraging opportunity 39 (Stein and Magnuson 1976; Sih 1980,1981,1982). Therefore, the behavioural response of prey should be a function of their vulnerability to a particular predator and the de-gree of risk (the threat-sensitivity hypothesis Helfman 1989) . Evidence supporting this threat-sensitivity hypothesis can be found for many animals. Hennessy and Ow-ings (1978) report that squirrels differentiate between different species of snakes, with greater response to more dangerous predators. Mayfly nymphs distinguish between predators and superficially similar nonpredators (Peckarsky 1980). More vulnerable age classes of prey show stronger responses to predators (Werner et al. 1983a sunfish; Sih 1980, 1982 notonectids). Prey may vary their responses to a given predator de-pending on factors that mediate risk, such as distance to a predator or distance to refuge (Lima 1985; Ydenberg and Dill 1986). When prey species differ in vulnerability the more susceptible should show stronger responses. Few studies have examined the influence of predators on the ac-tivity and behavioural repertoire of differentially susceptible prey (but see Neill 1970, fish; Schmitt 1982, marine snails; Huang and Sih 1990, isopods and salamander lar-vae). In the present study I experimentally tested the threat-sensitivity hypothesis by comparing the interaction between attack rate by cutthroat trout and types of defense employed by two differentially vulnerable prey: chum salmon fry and threespine stick-leback. I predicted that prey response to predators should be directly proportional to the degree of risk, and inversely proportional to the vulnerability of prey. Methods General Experimental Methods The results presented in Chapter 2 suggest that the number of possible species interactions occurring in the Fraser River estuary is large. Since it was impossible 40 to consider all possible interactions I restricted my study to interspecific interactions affecting chum fry feeding behaviour, and limited the complexity of this study to three species; chum fry and stickleback as prey ("prey" and "forager" will be used inter-changeably), and cutthroat trout as predators. Laboratory experiments were used to observe behavioural interactions between species. Although laboratory conditions are obviously artificial in their representation of the natural environment they provide a simplified environment in which to isolate and observe specific behavioural responses. Stickleback were chosen as the most likely competitor of chum salmon fry. Cutthroat trout were chosen as predators because of their ability to prey on both chum and stickleback. I assumed that cutthroat trout represented a level of predation risk that would have been experienced with any large predatory salmonid. Coastal cutthroat trout have been sampled in the arms and mainstem of the Fraser River by North-cote et al. (1978), and Nilsson and Northcote (1981) report fish, especially threespine stickleback, as important items in the diet of cutthroat trout in Mike and Ruby lakes, B .C . Collection of Fish Chum fry and stickleback were collected during March-May, 1986, 1988, and 1989 from two sites in the lower Fraser River, the North Arm and Dow fishing bar sites (Figure 2.2). Fish were captured with beach seines as described in Chapter 2 and stickleback were also collected with minnow traps. Due to difficulties in maintaining fish under laboratory conditions, in 1988 and 1989 it was necessary to obtain chum from additional sources. The use of fry from different sources must be considered in analysis of results as fry from different sources may differ in behaviour due to differences in inherited traits, rearing environments, and past learning. Chum fry were collected using an inclined plane trap in the Mamquam Creek spawning channel (49° 41' N, 123° 10' W), near Squamish, B .C. , and hatchery-reared fry were obtained 41 from the Chehalis Hatchery, near Harrison Mills, B .C. , operated by the Canadian Department of Fisheries and Oceans. Fish were transported to the laboratory at the University of British Columbia in 100 1 plastic containers. A continuous flow of oxygen and ice was supplied during transit. Coastal cutthroat trout, Oncorhynchus clarki clarki (Richardson), were collected by hook and line from Gwendoline Lake, in the University of British Columbia Research Forest near Haney, B .C. (49° 19' N, 122° 34' W). Trout were collected from this source because of their abundance and ease of capture. Although this trout population was not experienced at feeding on chum fry or stickleback in the wild and may not have been adapted to feeding on fish, trout were sufficiently large to handle these prey and quickly became conditioned to feeding on both species in the laboratory. At least 60 trout, exceeding 200 mm total length, were collected each year to be used as predators. Trout were transported to the laboratory by the same procedure used for prey. Stickleback and chum of different origins were transferred to separate holding tanks. Chum were held in 165 1 fiberglass tanks with a waterflow of 2 1 per min while stickleback were kept in 90 1 wooden tanks with a waterflow of 1.5 1 per min. Trout were held in 365 1 fiberglass tanks with a waterflow of 4.5 1 per min. Water temperature was maintained between 8 and 12° C. The laboratory was serviced by the municipal water system which was periodically treated with chlorine. In 1988 and 1989 a sodium thiosulphate solution was injected into the laboratory water reservoir (10 mg per 1) as a precautionary measure to neutralize unpredictable high chlorine levels. Light was supplied by overhead fluorescent lighting controlled by a photoelectric cell to approximate a natural photoperiod. All holding tanks were vigorously aerated. Pilot experiments revealed differences in feeding and antipredator behaviour between hatchery-reared and wild chum fry collected from natural habitats. The behaviour of chum obtained from the Mamquam spawning channel was similar to that 42 of chum sampled from the Fraser River so hereafter they will be collectively referred to as wild chum and their results will be combined. Hatchery chum were generally more surface oriented than wild fry. When alarmed, hatchery chum dropped to the bottom of the tank while wild chum tended to school at all depths in the water column. Trout were fed once a day with chopped beef liver supplemented with live chum fry and stickleback. W i l d chum fry were initially fed twice a day with live bloodworms (Tubifex sp., Oligochaeta) and water fleas (Daphnia sp., Cladocera) and later with Tetramine fish flakes and Oregon Moist Pellets ( O M P ) . Hatchery chum were fed O M P supplemented with live prey. Stickleback were fed daily with live Tubifex and Daphnia. Two weeks before being used in experiments, prey were moved to smaller holding tanks (90 1) and conditioned to the experimental food items. A l l experimental fish were deprived of food for 48 h before testing. Three food types (Daphnia, Tubifex,a.nd adult fruit flies, Drosophila, Diptera) were used in experiments on prey foraging behaviour. Daphnia were collected from three natural sources: Jericho Pond and Lost Lagoon in Vancouver, and Deer Lake in Burnaby. Daphnia were transported to the lab in 20 1 plastic containers and survived up to 14 days in aerated aquaria. Daphnia were sorted using nested metal sieves (mesh sizes; 1.5 mm, 1.0 mm). Only Daphnia greater than 1.0 mm were used in experiments. Tubifex were supplied by an aquarium retailer and maintained in laboratory aquaria with continuous water flow. Drosophila were cultured in 0.5 and 1 1 glass bottles on a commercial agar medium (formula 4-24 Drosophila medium, Carolina Biological Supply Co.). Adult flies were fresh frozen and kept until needed. Experiment 1: Prey Evaluation of Risk To determine if prey could assess the degree of risk and adjust their behaviour accordingly, feeding rates by wild chum and stickleback on a single food source were 43 compared at two levels of risk: (1) with hungry and (2) with satiated predators. A l l observations on foraging behaviour were made during A p r i l - M a y , 1986 in an exper-imental tank (180 x 120 x 60 cm, approximately 1100 1) made of brown painted plywood with a plexiglass front (Figure 3.1). The tank bottom was covered with a 2-4 cm layer of sand and fine gravel. The tank was divided into two unequal portions by a clear plexiglass partition, 60 cm from the left side. The smaller compartment was used to hold predators prior to testing. The partition could be raised remotely by the observer using a pulley system at the onset of a feeding trial. The tank was illuminated by two cool, white, 40 watt fluorescent overhead lights suspended 100 cm above the tank and controlled by a photoelectric cell to approximate a natural photoperiod (12 h daylight, 12 h dark). The entire experimental arena was surrounded by a black plastic blind with viewing windows through which the fish could be observed without being disturbed. Trout used in hungry predator treatments were deprived of food for 48 h before testing. In trials requiring a satiated predator, trout were fed 5-10 fry, two hours before tests were initiated. The fry used to satiate trout were anesthetized in 2-phenoxyethanol to facilitate capture by trout and added individually to the predator section of the experimental tank. The feeding period ended when the trout stopped eating fry. After two hours the tank divider was raised and the trout was allowed access to the prey. A single trout was tested with 10 chum and 10 stickleback simultaneously. Food for the prey was added after the hungry trout made five attacks or after 30 min in treatments with a satiated trout. Daphnia (1200) were mixed in one liter of water and flushed through a 2.5 cm dia plastic tube into the center of the experimental tank. The number of attacks by the trout on each fish species was recorded for one hour (replicated at least five times). A n attack was defined as a rapid movement by the trout toward an individual or group of prey. Trials were terminated after one hour and 44 ' - • • V ; FLUORESCENT LIGHTS 100 cm-FOOD DELIVERY OUTLET 10 cm 60 cm 180 cm-Figure 3.1 The apparatus used in Experiments 1-3. The clear divider separated trout (left compartment) from chum and stickleback during conditioning. 45 the trout returned to the holding tank. Individual trout were not used in more than one trial. The tank was drained and refilled between trials. Chum and stickleback were anesthetized in 2-phenoxyethanol, preserved in a 10% formalin solution for at least one week, and transferred to 70% ethanol. Stomach contents were removed and enumerated. Feeding rates were expressed as the total number of Daphnia consumed per hour. Mann-Whitney U (MWU, Mann and Whitney 1947) tests were used to test for differences in attack rates of hungry and satiated trout and for the effect of predation intensity on prey feeding rates. Experiment 2: Differential Prey Susceptibility In order to determine: (a) differential prey susceptibility to predation while foraging, (b) antipredator strategies, and (c) predator preference, I observed predation on single and mixed species groups (stickleback and wild chum or hatchery chum) while feeding in the laboratory. These experiments were run on separate groups of hatchery and wild chum during May-June, 1986 and April-June, 1988, respectively. A single trout was offered 20 fish, either a single prey species or equal numbers of both species (10 chum and 10 stickleback) replicated at least five times. The experimental apparatus was similar to that described above (Experiment 1). All fish were deprived of food for 48 hours and placed in their respective compartments at least 10 h before testing to allow them to adjust to the tank and to recover from handling. Trout were introduced to the experimental tank by raising the divider and the resulting interactions were recorded. After five attacks by the predator were observed, food items were added. All foraging periods (1 h) commenced at the time of food addition. In each trial, chum and stickleback were given a choice of food types repre-senting three feeding habitats (Tubifex, benthic; Daphnia, mid-water; and Drosophila, surface). Foragers had been preconditioned to these food types for at least two weeks 46 before testing and readily consumed all food types in the holding tanks. Daphnia and Tubifex (400 of each) were mixed in a liter of water and flushed through a removable 2.5 cm diameter plastic tube into the center of the tank. Drosophila (400) were scat-tered onto the water surface by an observer standing behind the blind. Disturbance created by the food additions was minimal. If trout did not attack the prey within the first 15 min after the food addition, the trial was terminated. Predator attacks were recorded for a total of 27 one-hour observation periods. Attack location was determined based on the position of the prey attacked. Position was recorded as bottom or surface if the prey was within 12 cm of the substrate or water surface respectively. The remaining 36 cm between surface and bottom was defined as mid-water. Standard observations recorded were: the species attacked, the location of each prey attacked, attack success (capture), and handling success (consumption). Antipredator responses employed by chum and stickleback and the sequence of events during an attack were noted qualitatively. Because of the number of fish used during each observation period I was unable to determine if individual prey were selected by the trout. The same trout was not used more than once in a two week period. The experimental tank was drained and refilled between trials. Mann-Whitney U tests (MWU) were used to test for differences in attack rates on chum and stickleback and between single and mixed species treatments. Data comparing the proportion of successful attacks and captures were normalized by arcsine square root transformations and analyzed using two tailed t-tests. The numbers of attacks in each position were compared for chum and stickleback using a two-factor ANOVA. Statistical probabilities exceeding 0.05 were considered nonsignificant. 47 Results Experiment 1: Prey Evaluation of Risk Satiated trout were less active than hungry trout. Satiated trout generally rested near the bottom and occasionally moved short distances (50-100 cm) along the side of the tank. Satiated trout infrequently chased prey but a few prey that came within 20 cm of the trout were attacked. By comparison, hungry trout swam continuously about the tank, chasing and attacking prey. Attack rates by hungry and satiated predators and feeding rates of chum (mean length 38.9 mm ± 4.4 standard deviation) and stickleback (mean length 42.2 m m ± 2.5 standard deviation) are presented in Fig-ure 3.2. Hungry trout made significantly more attacks on prey than satiated trout (attacks on chum, P=0.03; attacks on stickleback, P=0.01). Chum feeding rates were significantly lower in the presence of hungry trout than satiated trout (P=0.02). Hun-gry trout had relatively less effect on stickleback; feeding rates declined in treatments with hungry trout but not significantly (P=0.06). Experiment 2: Differential Prey Susceptibility Wild versus hatchery chum Results of the experiment on differential susceptibility of chum and stickleback are summarized in Table 3.1. During 27 one hour observation periods, a total of 678 predatory attacks were recorded. Results of Mann-Whitney U tests show attack rates on wild or hatchery chum were the same in single (P=0.67) or mixed species (P=0.36) treatments (Figure 3.3), therefore data for these treatments were combined and will be referred to as "chum" in subsequent analyses. Attack rates on stickleback in mixed species treatments were independent of chum type (Figure 3.4a, P=0.58) and these data have also been pooled. 48 Table 3.1 Mean attack rates (mean number of attacks by cutthroat trout per indi-vidual prey per hour), standard errors (SE), number of replicates (n), total attacks, percent attacks resulting in capture, and percent attacks ending in consumption in single and mixed species treatments (Experiment 2). attack total percent percent group n rates SE attacks captured consumed single species wild chum 5 1.6 .64 163 1.8 1.8 hatchery chum 5 1.5 .28 147 6.1 6.1 pooled chum 10 1.5 .33 310 3.9 3.9 stickleback 6 0.6 .12 82 17.1 0.0 mixed species group with added... wild chum stickleback 6 1.1 .40 66 1.5 1.5 hatchery chum stickleback 5 1.7 .40 83 0.0 0.0 pooled chum stickleback 11 1.4 .28 149 0.7 0.7 stickleback wild chum 6 1.4 .42 81 46.9 2.5 stickleback hatchery chum 5 1.1 .21 56 33.9 0.0 stickleback pooled chum 11 1.2 .24 137 41.6 1.5 Attack rates on each species Attack rate per individual was similar for both prey types in single species treat-ments, 0.6 for stickleback and 1.5 for chum (Table 3.1; Figure 3.5a, M W U , P=0.10). In mixed species treatments, predators showed no preference for either chum or stick-leback (mean 1.3 attacks, Figure 3.5b; M W U , P=0.79). Although stickleback tended to suffer more attacks when chum were present, there were no statistically significant differences of the presence of alternate prey (Figure 3.4b, stickleback M W U , P=0.06; Figure 3.3, chum M W U , P=0.96). 49 feeding rate of wild chum with hungry trout with satiated trout feeding rate of stickleback with hungry trout with satiated trout attack rate on wild chum by hungry trout by satiated trout attack rate on stickleback by hungry trout by satiated trout 10 2 0 3 0 4 0 5 0 number of attacks per hour number of Daphnia consumed per hour Figure 3.2 Comparison of wild chum and stickleback feeding rates when attacked by hungry and by satiated trout (Experiment 1). The data in each treatment are summarized by boxplots. The 5 vertical lines on each boxplot portray 5 percentiles whose P-values, from left to right, are 10, 25, 50 (median), 75, and 90. All values above the 90th percentile and below the 10th percentile are indicated by a circle (o). The experiment was replicated seven times for hungry trout and five times for satiated trout. With the exception of stickleback feeding rates, all paired comparisons of attack rates and feeding rates for each species were significantly different ( M W U test). 50 single prey treatments hatchery chum n = 5 wild chum n = 5 mixed prey treatments hatchery chum n = 5 wild chum combined hatchery and wild chum single prey treatments n = 10 mixed prey treatments n = 11 I c -J 1 1 I L 0 1 2 3 4 number of attacks per individual Figure 3.3 Comparison of attack rates on hatchery and wild chum in single and mixed species treatments (Experiment 2). Single species treatments used 20 chum and mixed species treatments used 10 chum and 10 stickleback. The data in each treatment are summarized by boxplots. The 5 vertical lines on each boxplot portray 5 percentiles whose P-values, from left to right, are 10, 25, 50 (median), 75, and 90. None of the paired comparisons (MWU) was significantly different. 51 a) stickleback with . . , hatchery chum n = 5 wild chum n = 6 b) stickleback single prey treatments n = 6 mixed prey treatments n = 11 1 2 3 number of attacks per individual Figure 3.4 Comparison of attack rates by cutthroat trout on stickleback when with hatchery versus wild chum and in single versus mixed species treatments (Experiment 2). Single species treatments used 20 stickleback and mixed species treatments used 10 stickleback and 10 chum. The data in each treatment are summarized by boxplots. The 5 vertical lines on each boxplot portray 5 percentiles whose P-values, from left to right, are 10, 25, 50 (median), 75, and 90. Extreme values above the 90th percentile and below the 10th percentile are indicated by an asterisk (*). None of the paired comparisons (MWU) was significantly different. 52 a) single prey treatments chum n = 1 0 stickleback n = 6 b) mixed prey treatments chum n = 11 stickleback n = 11 j i i i i_ 0 1 2 3 4 number of attacks per individual Figure 3.5 Comparison of attack rates by cutthroat trout on chum and stickleback in single versus mixed species treatments (Experiment 2). Single species treatments used 20 individuals of one species and mixed species treatments used 10 stickleback and 10 chum. The data in each treatment are summarized by boxplots. The 5 vertical lines on each boxplot portray 5 percentiles whose P-values, from left to right, are 10, 25, 50 (median), 75, and 90. None of the paired comparisons (MWU) was significantly different. 53 Prey response to predators Prey response to trout commonly took the form of a series of recognizable be-haviors used for predator avoidance, escape, or both. Stickleback appeared to rely on cryptic coloration and cessation of movement to avoid detection or recognition by trout. Stickleback generally fed close to the bottom or in mid-water. When attacked by trout, stickleback generally did not flee but became motionless, hovering just above or resting on the substrate. When motionless on the substrate they were often indistin-guishable to the observer. The stickleback's main protection against predator attack was its bony anterior lateral plates and dorsal and pelvic spines, which protected the fish from injury during manipulation by trout. Stickleback frequently escape unharmed after capture. Only four of 57 (7%) stickleback captured suffered noticeable injuries. Chum usually responded to an approaching predator by fleeing. Single fish or individuals within a school were used protean escape tactics, an erratic zig-zag dash away from the predator. Chum, and occasionally stickleback, responded to predators by joining single or mixed species groups. If the predator suddenly attacked the school it immediately scattered the prey in all directions. Once the predator had passed, the prey would reform a polarized school or segregate into species specific schools; stickleback remaining close to the bottom and chum forming tight groups higher in the water column. Occasionally I observed individual chum, intensely pursued by the trout, that fled to the water surface and "froze" near the edge or corner of the tank. Capture and consumption rates Significant differences between chum and stickleback were revealed in the events following attack. Results from 11 mixed species trials show that predators were more successful at capturing stickleback than chum (Table 3.1, t-test, P<001). Trout pur-sued chum for greater distances than stickleback but were generally unsuccessful at 54 capturing chum. Stickleback generally did not flee from an approaching trout but remained immobile where they had been feeding and were easily captured. Only one chum (0.7% of attacks) was caught compared with 57 stickleback (41.6% of attacks). But due to difficulty in handling, only 3.5% of stickleback captured were consumed. Trout manipulated stickleback in their mouths but generally were unable to swallow them. Consequently, the percentage of attacks resulting in successful consumption of prey was the same for both species (t-test, P=0.64). The percentage of attacks on chum ending in capture was significantly greater in single compared to mixed species treatments (t-test, P=0.04). The frequency of attacks on chum and stickleback in three locations in the water column is presented in Table 3.2. A two-factor A N O V A determined that the frequency of attack by trout was the same for each species at each location in the water column. Table 3.2 Mean attack frequency by cutthroat trout on chum and stickleback in three locations in the water column for eight replicate experiments (Experiment 2). location species surface se midwater se bottom se chum 4.3 2.0 5.9 1.8 5.6 1.8 stickleback 4.3 1.5 4.4 1.1 4.8 1.5 A N O V A (species x location) source SS df MS F P species 7.52 1 7.52 0.35 0.56 location 8.79 2 4.39 0.20 0.82 species x location 4.54 2 2.27 0.11 0.90 error 905.1 42 21.55 55 Discussion Experiment 1: Prey Evaluation of Risk Results from this experiment agree with the threat-sensitivity hypothesis; the response by chum and stickleback was greatly influenced by the activity of the trout. As predicted, the behavioural response by prey increased with increased risk. Chum and stickleback were capable of assessing the degree of risk of predation and modify-ing their feeding rates accordingly. Satiated trout displayed little predatory activity compared with hungry trout; they remained motionless near the bottom of the tank for most of the feeding trials. Compared to a hungry predator, the presence of a non-threatening predator had relatively little effect on the foraging activity of chum and stickleback. Chum and stickleback made "inspection visits" toward the trout and only exhibited a fright response (e.g. rapid flight) if there was sudden movement by the trout. When more active, hungry trout were present, chum feeding rates were reduced probably due to increased time spent monitoring and fleeing from approaching preda-tors. The insignificant reduction in stickleback feeding rates when faced with hungry trout suggested that the relative increase in perceived risk was less for stickleback than chum. The decision whether to flee from a predator or to remain feeding is under prey control and will depend on the prey's assessment of risk (Sih 1987). From an energetics perspective, prey should exhibit flight behaviour only if the potential for attack is high, because such behaviour is costly in terms of lost energy and lost feeding opportunity. Several researchers have reported a correspondence between intensity of prey response and degree of danger that are consistent with my findings. Dill (1972, 1974) docu-mented differential prey responses of zebra danios, Brachydanio rerio, to active versus inactive predators. As predator size and approach velocity increased, the antipredator 56 response of danios intensified. Laboratory studies indicate that bluegill sunfish pay little attention to largemouth bass in a tank until the bass shows inclinations to begin feeding (Stein and O'Brien quoted in Werner et al. 1983a). Similarly, the response of the cyprinid Leucaspivj delineatu3 varied depending on the type and activity of the predator (Ruppell and Gosswein 1972). When exposed to a free swimming pike, the integrity of the cyprinid school increased dramatically. But antipredator response was reduced if a predatory pike was held behind glass and no pike odor was evident, or if a nonpredatory tench (Tinea tinea) was presented. Contrary to my results, Magnhagen (1988) found that the presence of predatory chinook smolts had no effect on the food intake rate of chum fry in laboratory exper-iments. The failure of chum to respond to increased predation risk in Magnhagen's study may have been a result of the experimental protocol. Smolts were held in an adjacent aquaria and were visible to chum through glass. In the absence of actual at-tack chum may not have perceived the level of risk imposed by the isolated chinook to be sufficient to reduce their feeding rates, similar to the problems reported by Ruppell and Gosswein (1972). These results suggest that the effect of predation risk on feeding rates of prey in the wild will depend not only on the encounter rate with predators but also on the intensity and duration of the predatory act. Experiment 2: Differential Susceptibility to Predators Trout did not show a preference for prey type; attack rates were the same on stickleback and on hatchery and wild chum in all locations in the water column. Although attack rates were similar, risk of predation did not affect the prey species to the same degree. The relative costs of continuing to feed in the presence of a trout are 57 not equal for both chum and stickleback as they display different morphological and behavioural adaptations for defense. Antipredator strategies The defense used by stickleback includes physical features that reduce detection by predators. Crypsis, when animals are camouflaged to resemble part of their envi-ronment (Edmunds 1974), is an important primary defense of stickleback. Stickleback are colored drab olive or brown with patches of darker brown making them relatively inconspicuous against the vegetation or sediment. Stickleback deferred flight and con-tinued feeding or became motionless when a trout approached, probably due to the decreased likelihood of being detected near the substrate (McPhail 1969; Endler 1986). When motionless on the substrate they were often indistinguishable to the observer. Crypsis is effective only if prey remain motionless, but costs in lost energy intake as animals cannot feed while remaining motionless (Sih 1987). The stickleback's main protection against being eaten is morphological defenses (e.g. bony anterior lateral plates and dorsal and pelvic spines). These morphological adaptations decrease the probability of attack once detected, and the probability that they will be consumed once captured (Moodie et al. 1973; Giles and Huntingford 1984). The lack of these features may account for the greater evasive behaviour of chum which are more susceptible to consumption if captured. The dorsal and pelvic spines of stickleback can be locked in place and can increase the cross-sectional area as much as 75 percent (Gross 1978). As the spines will not collapse, the predator must break them or handle the fish with spines erect. The bony plates protect the fish from injury during manipulation by predators and stickleback frequently escaped unharmed after capture. Only four of 57 (7%) stickleback captured in my experiments suffered noticeable injuries. 58 Evidence from experimental studies and field investigations suggest that spines are an effective defense against fish predation (Hoogland et al. 1957; Moodie et al. 1973; Wootton 1984). Moodie (1972) noticed a positive correlation between the size of G. ac-uleatu3 eaten and the size of cutthroat trout eating them in the Queen Charlotte Is-lands. Trout had to reach 200 mm standard length to consume 30-40 mm stickleback and 400 mm to take sticklebacks 90 mm in length. It is probable that after repeated trials, trout would learn to handle or avoid sticklebacks. There was no evidence of predator learning in this study; the same trout was used in at most three trials. How-ever, spines may actually be a disadvantage when encountering invertebrate predators (e.g. giant water bugs, Belostomatidae) that use the spines to maintain a grip on the prey (Reist 1980). Once captured, chum were more vulnerable than stickleback and as predicted showed the most intense antipredator response, relying on behavioural adaptations to compensate for increased vulnerability. As a consequence of their more vigorous evasive behaviour fewer chum than stickleback were captured. The initial response of chum to the introduction of a trout was the formation of a loosely aggregated shoal with other chum or with stickleback. Chum reacted to direct attack by fleeing, individually or in tight schools. Flight was usually by protean escape, an erratic zig-zag dash away from the predator, followed by a regrouping of the school after the trout had passed. This disorganized movement apparently confused the trout, which could not anticipate the changes in direction. The resulting delay in response time of the predator increased the probability of the chum escaping. Flight would perhaps be an even more effective defense in vegetated habitats where refugia are nearby. When predation became intense, prey frequently segregated into single-species schools, stickleback remaining close to the bottom and chum near the water surface. School formation is common in fishes and experimental evidence reviewed by Pitcher 59 (1986) suggests predation and food resources are the two factors responsible for the evolution of schooling behaviour. Powell (1974) provided experimental evidence that, individual starlings reduce their surveillance of predators as group size gets larger, and that their foraging rate goes up simultaneously. Several antipredator mechanisms have been credited to schooling. The probability of attack on an individual is reduced (the dilution effect, Bertram 1978), the level of vigilance and ability to detect predators increases (Bertram 1978; Magurran et al. 1985), and the efficiency of the predator at capturing prey is reduced as group size increases (confusion effect, Hobson 1968; Major 1978; Milinski 1977). The confusion effect is an effective antipredator mechanism in situations where the predators hunt visually and take only one prey at a time (Keenleyside 1979). In a school, many stimuli elicit conflicting orienting responses (Humphries and Driver 1970) and may cause the predator to hesitate before attacking, allowing the prey to escape. One advantage attributed to schooling, the reduced probability of detection due to reduced encounter frequency important in the wild, is likely not relevant in this experimental study due to limitations imposed by the size of the aquaria. The tendency for chum and stickleback to segregate into species-specific schools when pursued by a predator implies that schools of conspecifics offer greater protection than mixed schools. Pitcher (1980) and Seghers (1974) suggest the value of schooling lies in the behaviour of the group after it has been detected and approached by the predator. According to Allan and Pitcher (1986), mixed schools segregate under threat of predation because single species groups are more coordinated and therefore more effective in evading predators. As well, the increased uniformity of appearance of the school increases the confusion of the predator. An individual that appeared different from the rest of the school would be more conspicuous to a predator (the oddity effect, Curio 1976). 60 The difference in capture rates observed for chum in single versus mixed species groups appears contradictory to the explanations given by Allan and Pitcher (1986). A greater proportion of chum were caught when they were the only prey type. As trout did not selectively attack either prey type and the rate of attack per individual prey was the same in single and mixed species treatments, the greater capture rates of chum in single species groups may be the result of the greater absolute number of attacks on chum. Furthermore, the higher capture rates for chum in single species groups appears due to the higher capture rates for hatchery chum but not for wild chum. The presence of stickleback appears to reduce the rate of capture of hatchery chum by trout. 61 4. Effects of P r e d a t i o n o n the Foraging B e h a v i o u r of C h u m F r y and Stickleback Introduction The theory of optimal foraging predicts that animals should forage in a way that maximizes their average rate of net energy intake (Schoener 1971; Pyke et al. 1977). Energy maximization is an optimal strategy only if feeding behaviour does not conflict with other needs, and constraints such as risk of predation may overshadow the importance of energy maximization in shaping foraging behaviour. Behavioral patterns should optimize processes involving costs and benefits measured in a currency of fitness. When prey are in motion they are more visible and more vulnerable to a predator (Zaret 1980; Ware 1973). Because foraging involves prey activity, and activity can elicit predator attacks (Chapman 1966; Ware 1973), the presence of a predator may reduce the time spent feeding. Metcalfe et al. (1987a) found that Atlantic salmon parr exposed to a model predator were less likely to attack passing food items and delayed attack until the food had reached its closest point. Experimental work by Dill (1983) and Dill and Fraser (1984) showed that juvenile coho detect and attack large prey items from a greater distance than small items. In the presence of a model predator, time spent moving and attack distances of coho decreased, and the degree of reduction was correlated with the frequency with which the model was presented. They concluded that the decreased attack distances resulted in decreased encounter rates with prey and consequently reduced food intake. Early optimal foraging theories assumed risk of predation was a function only of foraging time and was independent of the resource used (Milinski 1986; Lima and Dill 62 1990). However, studies on fish have shown that within a habitat the risk of predation can be different for feeding on different types of food, and this may alter the choice of food type by a forager (Dill and Fraser 1984; Milinski 1986). Animals are more vulnerable while handling prey items and the handling time or level of activity while handling is dependent on the type of prey (Godin 1990). When risk varies with prey type, animals that are vulnerable while feeding should change their diet to prey types that require reduced handling times or reduced activity for acquisition (Dunbrack and Dill 1983; Lima et al. 1985). Animals are limited in their ability to process visual information simultaneously from several sources (Milinski 1984). Milinski observed that the ability of ten-spine stickleback to detect approaching predators was inversely proportional to the atten-tion required for the feeding task. Adaptively, with increasing predation risk animals should pay more attention to monitoring approaching predators and less to feeding. Alternatively foragers may switch to food types that require less visual attention or handling time (Milinski and Heller 1978; Godin 1986), or slow down the rate of feeding (Sih 1982; Lendrem 1984) or the rate of responding to food (Metcalfe et al. 1987a). Milinski and Heller (1978) found that after exposure to a model predator, stickleback attacked lower density swarms of Daphnia, which provided a lower feeding rate but increased their ability to detect an approaching predator. Juvenile Atlantic salmon were less likely to orient to or attack food after sighting a predator, and Metcalfe et al. (1987a) accounted for this reduced rate of responding to food as the inability of salmon to attend simultaneously to two visual tasks, food assessment and vigilance for predators. Vigilance for predators may also reduce the efficiency of prey selection. With increased predation risk Atlantic salmon lost the ability to discriminate between foods 63 of different sizes (Metcalfe et al. 1987b). By attacking food that was too large to swal-low, salmon gained little and incurred the costs of energy expenditure and increased vulnerability to predators. There are many ways individuals can alter their foraging behaviour to reduce the risk of predation: by habitat shifts, reduced movement, increased vigilance, altering prey attack behaviour, changing feeding rates, or joining groups. The costs of altering behaviour may be expressed in reduced feeding rates, as shown by Sih (1982) with notonectids and Werner etal. (1983a) with sunfish, or lower food consumption as noted by Stein (1979) and Milinski (1986) studying crayfish and stickleback, respectively. Behavioral responses to avoid predators that concentrate vulnerable fish in protected habitats may result in increased competition for food or space (Mittlebach 1984), lower food quality (Lima 1988), and reduced survival (Larkin and Smith 1954). Competition for food also may play a role in the decision making of foragers (where, when, and how to feed) and influence the trade-off between foraging gain and predator avoidance. Dill and Fraser (1984) observed that when apparent competition was increased, coho took greater risk and travelled farther to intercept prey. Coho appeared to trade-off predator avoidance against the cost of lost foraging opportunity. I tested the hypotheses that the foraging efficiency of chum and stickleback decreases with increasing risk, and increases in the presence of alternate prey. How prey alter their feeding behaviour in the presence of predators has been well documented (studies listed above) but fewer studies have considered how predation-induced changes in the feeding behaviour of one species are influenced by the presence of alternate prey with different vulnerability and defenses. Chum salmon response to predation in the presence and absence of alternate prey (stickleback) was studied by observing changes 64 in foraging behaviour as indexed by: diet composition, feeding mode (surface, mid-water, and benthic), proportion of fish feeding, and feeding rate. I predicted, that under increasing predation risk: 1) the proportion of fish feeding and feeding rate of fish should decrease due to conflicts between feeding and monitoring for or avoiding predators, 2) fish should choose food that requires less visual attention or handling time, 3) fish should forage in habitats where they are less exposed to predation. Methods Methods are described for three experiments. Experiment 3 assessed the effects of predators on the foraging efficiency of chum salmon fry and threespine stickleback, two prey with dissimilar defenses, by observing changes in (1) diet composition, (2) the percentage offish feeding, (3) feeding rates, and (4) feeding indices. Experiments 4 and 5 determined if changes in foraging behaviour of chum and stickleback in the presence of a predator were due to two characteristics of food choice that affect vulnerability: (1) foraging time for a given item, and (2) foraging site. Experiment 3: Foraging Efficiency The effect of trout predation on chum and stickleback foraging efficiency was investigated in a series of experiments manipulating six different combinations of three prey (hatchery chum, wild chum, and stickleback) and one predator (trout): chum alone, stickleback alone, chum with stickleback, chum with trout, stickleback with trout, and chum with stickleback and trout. The numbers of prey (20 per trial) and food items (1200) were constant in each treatment. Single species treatments used 20 fish of one type while mixed species treatments used 10 chum and 10 stickleback. Stickleback and chum of similar sizes were selected from the experimental stock to 65 minimize any effect of size selective predation. Six replicates of each treatment were performed in random order for hatchery chum, wild chum, and stickleback. Trials were run between May and June 1986 for hatchery chum and stickleback, and between April and June 1988 for wild chum and stickleback. The experimental apparatus and protocol was similar to that used in Experi-ment 2 (Figure 3.2). During a predation trial, a single trout (mean length 263.3 mm ± 4.9 standard deviation) was introduced by raising the partition and the resulting interactions were monitored. An attack was defined as a rapid movement by the trout directed toward an individual or group of prey. After five attacks by the trout had been observed, food was added. For nonpredator trials food was added after the divider was raised. Three food types were used to represent three feeding modes: tubificid worms (benthic), Daphnia sp. (mid-water) and adult Drosophila (surface). Only Daphnia greater than 1.0 mm in length were used. Disturbance created by the food additions initially startled the chum and stickleback but did not appear to have lasting effects on their behaviour as they began feeding within seconds (mean response time 20 s). The disturbance had a greater effect on trout and mean time to the first attack (5 min 10 s). Some trout became alarmed and sought refuge in a bottom corner of the tank. If the trout did not pursue chum or stickleback in the first 15 min after the food addition the trial was terminated and not used in the analysis. The trout were never observed feeding on the invertebrates. All experiments began at the time of food addition. After 1 h foragers were removed, anesthetized in 2-phenoxyethanol and preserved in 10% formalin for later gut analysis. Samples were transferred to 70% ethanol after one week. Trout were returned to a holding tank separate from unused trout. To avoid learned responses by the trout, the same trout was used in no more than three trials. The experimental tank was drained and refilled between each trial. 66 Each fish was weighed to the nearest 0.001 g and fork length measured to the nearest mm. Stomachs were removed, blotted dry and weighed (wet weight) to the nearest 0.0001 g. Food items were identified and counted. Recovered Tubifex worms were frequently broken, and only recognizable head portions were counted. To compare the relative abundances of different food types in the diet between treatments, the number of food items of a given type consumed was expressed as a percentage of total food items for each species and treatment (percentage composi-tion by number, Windell and Bowen 1978). As data from numerical analysis provides little information on the value of food consumed, a feeding index (food weight/body weight x 100) was calculated as a rough approximation of food consumption between treatments. The percentage of fish feeding and the feeding rate (number of food items consumed per hour) in each treatment were determined. All percentage data were converted to proportions, transformed by arcsine square root to normalize distribu-tions, and analyzed using a two-way A N O V A on the main factors, predator presence and alternate prey presence. Significance values resulting from a two-way Anova take into consideration the effects of each factor across all treatments. As the percentages of the diet composed of different food types within a treatment are not independent, simultaneous comparisons of all food types would violate the assumptions of a mul-tiway A N O V A . Consequently, the contribution of each food type to the diet in each treatment was analyzed separately by a two-way A N O V A . Sample means and 95% confidence intervals were back-transformed to percentages for graphical presentation. Experiment 4: Foraging Time Under natural conditions, chum and stickleback consume many food types of varying micro-distributions, availability, handling difficulty, and caloric value. Assum-ing chum and stickleback that are foraging are more vulnerable to predation by trout, 67 and risk increases with time required to search and handle prey, the feeding time re-quired for different food types should be a good indicator of the relative degree of risk. In this experiment the foraging time for each food type, which includes the time for the components: searching, pursuing, capturing and handling prey, was examined in the laboratory. Each test consisted of an individual hatchery chum or stickleback feeding on a single type of food, in the absence of trout. Individual components of feeding were not measured because the time required for some activities was so short it could not be accurately recorded using a stopwatch. Experiments were conducted in two plywood tanks (Figure 4.1, 60 x 120 x 40 cm), painted brown, with plexiglass fronts. Each tank was partitioned into three observation chambers (60 x 40 x 40 cm, 70 1) by opaque plexiglass dividers. The dividers could be raised remotely using a pulley system to allow the individual chum or stickleback to move from the central holding area to the side test compartment. The tank bottom was covered with a 2-4 cm layer of sand and fine gravel. Illumination was provided by a single white fluorescent light, 1 m above the tank, controlled by a photoelectric cell to approximate a natural photoperiod. The tanks were enclosed by black plastic screens with viewing windows from behind which the fish could be observed without disturbance. Foraging time was measured for at least 30 individual hatchery chum and stick-leback, feeding on 70 of each of the three food types: Daphnia (>1.5 mm), Drosophila and Tubifex. Hatchery chum and stickleback were conditioned to each food type, twice daily for one week before testing. Fish were deprived of food for 24 h, and placed in the central chamber at least two hours before testing. Seventy individuals of a single food type were added to one side chamber and allowed to mix for five minutes. All food types were visible to the observer except for Tubifex once they had burrowed into 68 100 cm BLACK DIVIDER Figure 4.1 The apparatus used in Experiment 4 to measure the foraging time of chum and stickleback feeding on three types of invertebrates. Fish were conditioned to the tank in the center compartment and individuals were tested in either side compartment. 69 the substrate. The divider was raised briefly to let the subject enter the test chamber. The initial response time, beginning when the fish entered the test compartment until the first food item was consumed, was noted. As individual food items were consumed very quickly, the mean foraging time was averaged over the time for five consecutive bites (after swallowing the third item to after swallowing the eighth item) recorded us-ing a stopwatch. If the fish did not begin to feed within the first 15 min after entering the test compartment the trial was ended. Mean response times and mean foraging times for each food type were compared within and between chum and stickleback by using two tailed t-tests. Experiment 5: Foraging Site The object of this experiment was to determine if the degree of risk to detection while foraging by chum and stickleback is related to the location of the forager within the water column when feeding. Wild chum and stickleback were confined to specific depths in the water column and the number of attacks at each location by a single trout was recorded. The experimental apparatus and protocol were the same as that described in Experiment 2 (Chapter 2), with the following modifications. The food-deprived trout was placed in the holding compartment at least six hours before testing. The predator compartment was separated from the experimental arena by an opaque divider. Four chum or stickleback were placed in a vertical clear plexiglass tube (8 cm dia) in the center of the experimental arena (Figure 4.2). Each fish was confined to one of four separate compartments, 12 cm apart, representing bottom, lower water column, up-per water column and surface locations. Chum and stickleback were prevented from moving between compartments by fine mesh metal screens. The bottom of the tube was closed and covered with a layer of fine gravel. The top of the tube extended 9 cm 70 •0 FLUORESCENT LIGHTS 100 cm" PREDATOR COMPARTMENT OUTLET 10 cm 60 cm 180 cm-Figure 4.2 T h e apparatus used in Experiment 5 to measure the relative attack rate by trout on chum and stickleback confined to four locations in the water column: bottom, lower water column, upper water column, and surface. Trout were conditioned to the tank i n the left compartment and the opaque divider raised during testing. 71 above the water surface, preventing the fish in the surface compartment from jumping out of the tube. At the beginning of each trial, the opaque divider was raised, allowing the predator access to the experimental arena. Observations of predator activity were recorded for 30 min. An attack was defined as a rapid movement toward an individual prey, ending within 10 cm of the tube. Attacks frequently ended with the trout ramming or biting the tube surface. Charges at chum or stickleback that ended more than 10 cm from the tube were considered aborted attacks and were not included in the analysis. The target position of the first attack and the frequency of subsequent attacks at chum or stickleback in each position were noted. Replicate trials were made in random order on each prey type. Attack rates on chum and stickleback were compared by Mann-Whitney U tests ( M W U , nonparametric two-sample rank test, Mann and Whitney 1947). The effect of foraging site on the frequency that prey were attacked was analyzed by comparing the number of attacks by trout on each species, at each location, using Kruskal-Wallis H tests ( K W H , nonparametric analysis of variance by ranks, Kruskal and Wallis 1952). Locations with significantly different attack rates were identified by nonparametric multiple comparisons of rank sums (Zar 1984). Although attack frequencies on stick-leback in different locations were not significantly different, the data suggested the frequency of attack was greater in lower locations. Attack frequencies for stickleback were pooled for the two upper and two lower locations and compared using Mann-Whitney U tests (MWU). During these experiments it was observed that those chum attacked were usually swimming rather than resting on the bottom of the enclosures. To remove any bias resulting from prey movement, the data for chum were reanalyzed (KWH) using only those observations when all chum were swimming. 72 Results The results for each prey type are presented separately, as the most useful com-parisons are for chum or stickleback across treatments. Behavioral responses of wild and hatchery chum were not always the same so their results are analyzed separately when necessary. Stickleback did not show any difference in feeding behaviour whether with wild or hatchery chum and therefore data for stickleback in these treatments were combined for analysis. Mean lengths for each prey type were consistent among treat-ments (Table 4.1). However, there were significant differences in mean length among prey types ( A N O V A , P<0.001); listed in descending order of mean length, hatchery chum (52.3 mm), stickleback (46.3 mm) and wild chum (38.3 mm). Experiment 3: Foraging Efficiency Diet Composition The results of experiments on diet composition are presented in Figure 4.3. Hatchery chum alone (Figure 4.3a, treatment 1) consumed roughly equal numbers of Daphnia and Drosophila. The presence of either stickleback or trout had significant effects on the diet composition of hatchery chum (P=0.01 and P=0.002 respectively). When stickleback were present (treatment 3) Drosophila accounted for a greater per-centage (60%) of the chum's diet. When hatchery chum were exposed to trout they switched dramatically to a diet of 82% Daphnia (treatment 2). When hatchery chum were combined with sticklebacks and trout in treatment 4, the composition of the diet was similar to that of hatchery chum alone (treatment 1). Tubifex never accounted for more than 10% of the food consumed by hatchery chum. In contrast to hatchery fish, wild chum (Figure 4.3b) were less surface oriented; Daphnia consistently accounted for the greatest percentage (63-89%) of the diet in 73 Table 4.1 Mean, minimum and maximum fork length (mm), associated standard error (se), and total number of each species recovered and examined in each treatment (1— alone, 2—with predator, 3—with alternate prey, 4—with alternate prey and predator), " n " is the number of replicate experiments. total number mean minimum maximum treatment n of fish length se length length hatchery chum 1 6 119 53.3 3.1 43.6 60.7 2 6 113 50.9 2.7 43.4 61.8 3 6 59 53.9 3.2 41.9 62.6 4 6 59 51.1 2.9 41.8 61.6 overall 52.3 0.75 wild chum 1 6 114 37.3 1.3 33.9 43.1 2 6 104 37.5 1.7 34.0 44.9 3 6 59 39.2 1.7 35.0 44.8 4 6 56 39.4 1.8 33.9 45.6 overall 38.3 0.54 stickleback 1 6 119 48.6 0.9 45.7 51.9 2 6 109 46.7 2.1 37.1 51.5 3 12 119 45.7 1.4 32.4 50.3 4 12 109 45.6 0.6 40.0 47.9 overall 46.3 0.68 74 100 8 4 ° CD CL 20 0 80 • Drosophila Daphnia Tubifex 80 h 60 40 20 0 80 60 a) hatchery chum b) wild chum c) stickleback • WtL. m r 1 ! alternate prey * predator * alternate prey NS predator NS alternate prey NS predator NS treatment Figure 4.3 Composition of the diet of a), hatchery chum, b) wild chum, and c) stickle-back under four treatments: 1—prey alone; 2—prey with trout; 3—prey with alternate prey; and 4—prey with trout and alternate prey (Experiment 3). The number of filled circles (•) denote the number of prey types and the open circles ( O ) denote trout present. A n asterisk (*) indicates a significant effect and NS indicates a non signifi-cant effect of the presence of trout or alternate prey on diet composition. The top of each bar is the mean percentage. The vertical line is the 95% confidence interval for the mean. 75 all treatments. Although the composition of their diet did not change significantly, there was a tendency for wild chum to consume more Drosophila in treatments with stickleback or trout. In treatments with trout, the response of wild chum was in the opposite direction to that of hatchery chum; wild chum included proportionately more Drosophila in their diet (non significant increase) compared to hatchery fish that avoided feeding at the surface. Tubifex were rarely taken by wild chum (1-4% of the diet). Stickleback fed primarily on Daphnia (56-70% of the diet) in all treatments (Figure 4.3c). Benthic Tubifex accounted for 29-44% of the diet. Although stickleback consumed fewer Daphnia (14%) when with chum, the presence of chum or trout did not significantly affect the composition of the stickleback diet. Percentage Feeding The percentage of fish feeding in each treatment is presented in Figure 4.4a and the statistical probabilities for effects of each factor, presence of trout or alternate prey, are listed in Table 4.2. During single species trials nearly all hatchery chum were observed to feed. Results of an A N O V A revealed several significant effects of the presence of trout, stickleback and the interaction of both factors on the percentage of hatchery chum feeding (Figure 4.4a, Table 4.2). Increasing predation risk significantly reduced the number of hatchery chum feeding (38% reduction, P< 0.001). However when stickleback were also available as prey the percentage of hatchery chum feeding was the same as in the predator free condition. Wild chum did not show a significant difference in the percentage of fish feeding in the presence of trout or stickleback (range 85-99%, Table 4.2, P> 0.21). The percentage of stickleback feeding did not change when chum were present. However a significant decrease was noted in the presence of trout (P=0.01). 76 hatchery chum wild chum stickleback Figure 4.4 The percentage of fish feeding, feeding rate, and feeding index (weight of stomach contents/body weight x 100) of hatchery chum (•), wild chum (•), and stickleback (•) under four treatments: 1—prey alone; 2—prey with trout; 3—prey with alternate prey; and 4—prey with trout and alternate prey (Experiment 3). The number of filled symbols denotes the number of prey types and the open circles (O) denote trout present. Vertical lines are 95% confidence intervals for the mean. 77 Table 4.2 Statistical probabilities from results of a two-way A N O V A under the null hypothesis of no effect of presence of trout and alternate prey on the percentage of fish feeding, feeding rate, and feeding indices of hatchery chum, wild chum, and stickleback (Experiment 3). P values less than or equal to 0.05 are considered significant. Data are summarized in Figure 4.4. hatchery wild factor chum chum stickleback percentage predator <0.001 0.215 0.010 feeding alternate prey <0.001 0.277 0.572 pred. x alt. prey <0.001 0.392 0.727 feeding predator 0.270 0.036 0.070 rate alternate prey 0.001 0.003 0.111 pred. x alt. prey 0.032 0.430 0.799 feeding predator 0.318 0.427 0.297 index alternate prey 0.028 0.136 0.420 pred. x alt. prey 0.183 0.026 0.827 Feeding Rates Comparisons of chum and stickleback feeding rates in the presence or absence of trout and alternate prey are presented graphically in Figure 4.4b. Statistical probabil-ities for each factor resulting from an A N O V A are listed in Table 4.2. Hatchery chum showed a tendency for reduced feeding rates in the presence of trout (non significant, P=0.27) and a significant effect of the presence of stickleback (P=0.001). The two factors clearly interact (P=0.032); the feeding rate of hatchery chum depends on the presence of both trout and of stickleback. Both factors, trout and stickleback presence, had significant effects on the feed-ing rates of wild chum (P=0.036 and P=0.003, respectively); feeding rates increased with stickleback and decreased with trout. In treatments with stickleback and trout, 78 wild chum feeding rates were similar to feeding rates when wild chum were alone (Figure 4.4b). There was no significant interaction. Stickleback feeding rates were slightly reduced by the presence of trout and chum but the change was not statistically significant (Figure 4.4b, Table 4.2, P=0.111 and P=0.07 respectively). The feeding rate of stickleback was apparently independent of the presence of other species. Feeding Indices Feeding indices (food weight as a percent of body weight) for all prey types are displayed in Figure 4.4c. Results from a two-way A N O V A of feeding indices for hatchery chum (Table 4.2) show a main effect of the presence of stickleback (P=0.028), indicating the amount of food consumed (by weight) increased in the presence of stickleback. Neither main factor had any effect on the weight of food consumed by wild chum. There was a significant interaction of the presence of trout and stickleback (P=0.026), which implies that the weight of food consumed in the presence of trout depends on whether stickleback are present. Stickleback feeding indices were not affected by the presence of chum or stick-leback (P>0.2). Summarizing the four sets of data from Experiment 3, stickleback did not show any significant changes in feeding behaviour in response to the presence of either trout or chum. On the other hand, chum responded positively to the presence of sticklebacks, with hatchery chum showing a stronger response than wild chum. Chum response to trout was more variable; hatchery and wild chum showed opposite diet shifts in the presence of a predator. The addition of stickleback as well as trout tended to restore all feeding indices to the same level shown by chums alone. 79 Experiment 4: Foraging Time Daphnia were detected sooner than Dro3ophila and Tubifex, by hatchery chum (P<0.001) and stickleback (P=0.034, Table 4.3). Chum responded to Daphnia faster than did sticklebacks, in 92 versus 121 sec respectively. Chum and stickleback required at least twice as long as this to detect Drosophila or Tubifex, and each species did not take one food type. Only three of 30 chum were observed to feed on Tubifex and they consumed only those Tubifex that had not yet burrowed into the substrate. Only two stickleback fed on Drosophila. Although foraging times for each food type (Table 4.3) suggest certain food types were consumed faster than others, once food was detected, the foraging time per item was not statistically different for any food consumed by that species. Foraging time per item was significantly less for chum than for stickleback (two-tailed t-test, P=0.008). Experiment 5: Foraging Site In tests of foraging-site-specific vulnerability, trout attacked wild chum fry more frequently than stickleback in all locations in the water column (Table 4.4, M W U , P=0.02). Trout did not attack chum with equal frequency in all locations ( K W H , P=0.007). Nonparametric multiple comparisons confirmed that chum in the bottom and lower water column locations were attacked more frequently (Table 4.5). The frequency of attacks on stickleback were independent of the four locations in the water column ( K W H , P=0.079) although attack rates appear slightly higher on stickleback in the two lower locations. Comparison of pooled data for attacks on stickleback in the two upper versus the two lower locations agree with the findings for chum; trout attacked stickleback more frequently in the lower half than the upper half of the water column ( M W U , 0.05>P> 0.02, Table 4.6). Similarly, although the data are insufficient 80 Table 4.3 Mean response time (sec), mean foraging time (sec), and associated stan-dard deviation (sd) for three food types consumed by " n " individual hatchery chum and stickleback (Experiment 4). For missing values (—) the food type was not con-sumed, nor included in the analysis (two-tailed t-tests). P values less than or equal to 0.05 are considered significant. fork response foraging food type n length time sd time sd hatchery chum Daphnia 35 55.6 91.9 103.9 8.8 12.6 Drosophila 32 56.0 223.6 180.3 5.6 4.7 Tubifex 30 52.9 P < 0.001 P = 0.186 stickleback Daphnia 30 49.2 121.2 159.5 10.4 15.2 Drosophila 30 50.5 — — — — Tubifex 30 49.6 258.8 307.7 18.2 21.1 P = 0.034 P = 0.109 for statistical analysis, the first fish attacked in each trial was usually in the bottom half of the water column (Table 4.7). Movement did not affect the choice of prey attacked by trout. Analysis of only those attacks performed when all chum were swimming (Table 4.8, K W H , P=0.002) are consistent with results for total attacks on all prey; chum in the bottom and lower water column locations were attacked most frequently (Table 4.9). 81 Table 4.4 Results of a Mann-Whitney U test (MWU) comparing frequency of attacks by cutthroat trout on wild chum and stickleback and results of Kruskal-Wallace H tests ( K W H , nonparametric A N O V A by ranks) comparing mean number of attacks by cutthroat trout on wild chum and stickleback confined to four locations in the water column (Experiment 5). P values less than or equal to 0.05 are considered significant. location depth (cm) chum stickleback surface 0-12 5.0 1.1 upper 12-24 4.3 0.9 lower 24-36 8.3 1.9 bottom 36-48 7.3 2.4 replicates (n) 4 4 mean 6.2 1.6 sd 1.9 0.7 M W U = 16 P = 0.02 K W H 10.42 5.34 df 3 3 P 0.007 0.079 Table 4.5 Nonparametric multiple comparisons of attack frequency by cutthroat trout on wild chum confined to four locations (b-bottom, 1-lower, u-upper, and s-surface of the water column). Significant differences between locations are indicated by an asterisk (*). The experiment was replicated 12 times (Experiment 5). Locations with similar attack frequencies are underlined (SE—standard error, p—range of means, q—difference of rank sums/SE). rank 1 2 3 4 rank sums 215.5 216.5 363 381 location s u b 1 Rank Sum Comparison Difference SE q P <1.05,co,p 4-1 165.5 48.5 3.41 4 1.98* 4-2 164.5 36.5 4.50 3 1.59* 4-3 18.0 24.5 0.73 2 0.95 3-1 147.5 36.5 4.04 3 1.59* 3-2 146.5 24.5 5.98 2 0.95* 2-1 10.0 24.5 0.41 2 0.95 82 Table 4.6 Mean and median number of attacks by cutthroat trout on stickleback confined in the two upper locations versus the two lower locations of the water column. MWU—Mann-Whitney U test (Experiment 5). location n mean median surface + upper water column 8 0.88 0 lower water column + bottom 8 3.62 3.5 M W U = 52.5 tf0.05(2),8,8 = 51 0.05 > P > 0.02 Table 4.7 Comparison of the frequency chum and stickleback in each location were attacked first in n replicate experiments of foraging site vulnerability (Experiment 5). n bottom lower upper surface chum 12 7 4 1 0 stickleback 8 3 3 1 1 Table 4.8 Kruskal-Wallace H test (nonparametric A N O V A by ranks) on mean number of attacks on wild chum confined to four locations in the water column, when all prey were swimming (Experiment 5, n—number of observations, sd—standard deviation). location n mean sd bottom 11 4.7 5.8 lower 11 1.9 3.8 upper 11 0.5 1.2 surface 11 0.1 0.3 Kruskal-WaUis H = 14.64 df = 3 p = 0.002 83 Table 4.9 Nonparametric multiple comparisons of attack frequency on swimming chum in four locations (b-bottom, 1-lower, u-upper, and s-surface of the water col-umn). The experiment was replicated 11 times (Experiment 5). Significant differences between locations are indicated by an asterisk (*). Locations with similar attack fre-quencies are underlined (SE—standard error, p—range of means, q—difference of rank sums/SE). rank 1 2 3 4 rank sums 165.5 191 275 290 location s u b 1 Rank Sum Comparison Difference SE q P <1.05,oo,p 4-1 124.5 42.6 2.92 4 1.98* 4-2 99.0 32.1 3.08 3 1.59* 4-3 15.0 21.5 0.70 2 0.95 3-1 109.5 32.1 3.40 3 1.59* 3-2 84.0 21.5 3.91 2 0.95* 2-1 25.5 21.5 1.18 2 0.95* Discussion Three general conclusions can be made from these experiments. (1) Trout had significant effects on certain aspects of prey foraging behaviour. Behavioural responses by prey to reduce predation risk resulted in reduced foraging efficiency. In the presence of a predator hatchery chum switched from feeding on food at the surface to food in mid-water and the number of fish feeding decreased, wild chum feeding rates were reduced, and the percentage of stickleback feeding was reduced. (2) Risk of predation did not affect the three prey types to the same degree; much of the complexity of the interactions depended on the antipredator strategies of prey. A n d (3), the presence of alternate prey mediated the effect of predation risk on prey. The presence of alternate prey reduced the effect of predation on chum foraging efficieny but increased the effect of predation on stickleback foraging efficiency. 84 Chum and stickleback were rarely killed by trout but the need to avoid predators affected the feeding behaviour of the prey. The prey's relative fitness under the different treatments can be evaluated by comparing their feeding efficiency between treatments. Changes in feeding efficiency can have serious consequences on fitness. Dill and Fraser (1984) observed changes in diets of coho salmon caused by predation risk; coho fed on smaller items in the presence of a model rainbow trout, and Dill et al. (1981) correlated high levels of feeding and growth rates with greater fitness. Any effects on growth that result from behavioural responses of juvenile fish to predators are likely to have important consequences for survival. Reduced foraging may jeopardize survival because fish often experience disproportionately high predation rates when small (Forney 1974; Nilsson 1979; Mittlebach and Chesson 1987; Werner 1985) and survival may depend on rapid growth through a vulnerable size range (Larkin and Smith 1954). Gilliam (1982) developed a mathematical model that illustrates that during periods of the life history of fish where survivorship is already low, for example juvenile sunfish, a small reduction in growth can cause a large reduction in survivorship. Effects of trout on chum foraging behaviour In the absence of other species, hatchery chum tended to feed at the surface on Drosophila more than did wild chum (Figure 4.3). This difference in the importance of surface food types may reflect differences in past learning or conditioning of the two groups (see Suboski and Templeton 1989 for review). Training biases (Dill 1983) are common for different fish populations (LeBrasseur 1969; Bryan 1973; Milinski and Loewenstein 1980) and for individuals within a population (Allan 1981; Werner et al. 1981). In hatcheries, fish are raised on commercial food pellets, administered by throwing the pellets on the water surface. Fish in hatcheries do not have predators and fry must compete for food thrown on the surface (Johnsson and Abrahams 1991). This 85 feeding pattern would have been reinforced as chum were fed both live and commercial food in the holding tanks in the same manner. Behavioural responses by prey to trout predation varied depending on the prey type. Contrary to Johnsson and Abrahams (1991), who found that juvenile hybrid wild/domesticated trout were willing to risk greater exposure to predation when feed-ing than wild juveniles, hatchery chum displayed a greater response to predation risk than wild chum, resulting in fewer fish feeding (Figure 4.4) and a dietary shift from surface food items to mid-water Daphnia (Figure 4.3). The absence of a reduction in feeding rate, which might have explained the reduced feeding indices can also be ex-plained by the shift to feeding on greater numbers of small Daphnia. As well, Daphnia inhabit mid-water and swim constantly (O'Brien 1979) and are probably more readily detected by chum than immobile Drosophila on the water surface or Tubifex concealed in the sediment. Prey motion increases the reactive distance of foragers and as a result, the earlier detection and recognition of prey (Ware 1973). Irvine and North-cote (1983) found that juvenile rainbow trout selected live prey over dead prey due to prey movement. Experimental studies (Brooks 1968; Vineyard 1980) document that sunfish chose Daphnia over Diaptomus due to their different patterns of motion; daph-nids move continuously and copepods move intermittently. Preference for daphnids over copepods also might have resulted from the lower evasive capability of daphnids (Drenner et al. 1978). Drosophila were distributed on the surface of the water and were never observed to sink. When feeding on flies, hatchery chum must travel to the surface and orient their bodies such that their field of vision is directed toward the surface. It is unlikely the chum can simultaneously monitor for predators below or behind them. Switching to a diet of mid-water Daphnia should require less visual attention and allow chum to monitor predators in the water column while continuing to feed, and may decrease the 86 probability that chum will be captured even though trout attack prey more frequently in the lower half of the water column (Tables 4.4-4.6). The dietary shift to Daphnia by hatchery chum under risk of predation is also of significance in regard to food quality. Daphnia may be easier to detect because of movement, but provide the least caloric gain per unit of foraging activity. The caloric value per g dry weight for each food type is 5797, 5652, 5028 for Drosophila, Tubifex, and Daphnia respectively (Cummins and Wuycheck 1971). A n individual fruit fly is considerably larger than a single Daphnia and consequently provides significantly more food value per feeding act. Lima (1988) also noted that in the presence of a predator as the need for vigilance by juncos increased, the proportion of the diet composed of less profitable items increased. More profitable seeds required a head down position but less profitable seeds could be consumed with the head raised and without reducing vigilance. The absence of Tubifex in the chum diet does not appear to be a trade-off to avoid predation. When Tubifex were added they sank to the bottom and immediately burrowed into the substrate. Chum were observed to feed on Tubifex as they dropped through the water column and settled onto the sediment surface, but the worms never accounted for a significant portion of their diet probably due to the inability of chum to detect Tubifex once they had burrowed into the substrate. Although observations on foraging times revealed that the time required for the feeding task was the same for chum feeding on Daphnia or Drosophila (Table 4.3), the failure to detect a significant difference in foraging time per item for different food types may to be due to high variability and low sample sizes. As well, differences in foraging times do not consider variations in the time for the different components of foraging (i.e., search, pursuit, handling time), the visual attention required, or the success rate of feeding on different food types when under risk of predation (Milinski and Heller 87 1978; Dunbrack and Dill 1983; Godin 1986). Daphnia were detected significantly faster than other food types by chum and stickleback. Search time and therefore visual attention for Daphnia is less probably because they are constantly moving (O'Brien 1979; Vinyard 1980). Daphnia also dispersed throughout the water column unlike Drosophila and Tubifex, which are confined to the surface and substrate, respectively. However, predation rate is dependent on food type or foraging position (Table 4.5-4.6). In contrast to results presented in Table 3.2 (Experiment 2), when prey were confined to positions within the water column (Tables 4.4-4.6, Experiment 5), prey on the bottom or in the lower water column were subjected to a greater risk of attack by trout. In Experiment 2, prey may be avoiding positions in the water column where the probabiUty of attack is high and consequently differential attack rates may be obscured. Consistent with my predictions, hatchery chum shifted to feeding on prey that require less visual attention in the presence of a predator. At the same time, in contrast to my predictions, chum shifted to lower feeding positions with increased exposure to predators. In the absence of accurate measures of the time and energy benefits and the costs associated with scanning for predators and foraging for prey in the two positions, it appears maladaptive for hatchery chum to switch from surface feeding to mid-water feeding in the presence of cutthroat trout, unless the increased ability to monitor predators outweighs the increased vulnerability in this location. By comparison, the diet composition of wild chum remained the same in all treatments. The only significant change for wild chum concerned feeding rates (Figure 4.4), which as predicted, were reduced in the presence of trout and increased when stickleback were present. The reduction in feeding rates with trout implies that wild chum traded-off food intake for reduced movement, increased vigilance, or flight from predators. By reducing the time spent feeding, chum may also reduce the time spent vulnerable to attack, the major component of risk under the control of the prey. These 88 results suggest that wild chum can assess predatory risk, control their detection by trout and regulate the amount of time they are vulnerable to attack. Possible reasons for the differences between hatchery and wild chum The use of hatchery as well as wild chum was not planned but became necessary due to limitations on the availability of wild fish. The experiments were not designed to examine the differences but the results have practical as well as theoretical interest. The differences between the behavioural responses of hatchery and wild chum might be due to genetic, or environmental, or learned factors. To properly investigate differences between wild and hatchery fish, comparisons should be made between hatchery and wild fish from the same stock under similar rearing conditions. For the purposes of this study, it is sufficient to note that two groups of wild fish from distant locations were more similar in feeding and antipredator behaviours than were wild and hatchery chum that migrate through the Fraser River estuary. Behavioural differences shown by hatchery raised fish may be a product of their rearing environment. Fry in hatcheries are not exposed to natural stimuli that shape or reinforce normal feeding and antipredator responses (see Suboski and Templeton 1989 for review). Hatchery fish do not learn early to recognize wild food and studies have shown that chum develop training biases for food types that they first consumed (LeBrasseur 1969). Hatcheries condition fry to expect food following an overhead startle response and hatchery faculties do not provide refugia to train fry to seek shelter when startled. When hatchery fry are released they may be ill prepared to respond to hazards in the natural environment. Mortality of hatchery fry after release due to starvation and predation may occur because fish fail to recognize natural food and predators. Behaviour of fish consists of pre-organized behavioural sequences released by specific environmental stimuli (Barlow 1977) but some learning may be required 89 for responses to develop normally. Feeding responses and antipredatory responses are largely innate; fry "know" how to eat and how to evade predators. But naive fry must learn to recognize food and predators in order to respond. Even if hatchery fish learn to recognize and feed on live prey this does not ensure survival following release (Paszkowski and 011a 1985). Wild fish should out compete hatchery fry presumably because they are better adapted than hatchery fish to their natural habitat and have already been selected by predation and competition. Ex-perimental results presented here, show that hatchery chum were more responsive to predators than wild chum and shifted feeding positions and food types in the presence of a predator. However, behavioural responses by hatchery chum to avoid predators may actually increase risk of predation, and hatchery chum were captured more fre-quently than wild chum (Table 3.1). These results suggest that post-release mortality of hatchery chum could result from inefficient or suboptimal foraging, inappropri-ate social behaviour, competition from wild fish, or feeding behaviour that increases predation risk. Suboski and Templeton (1989) review studies reporting behavioural differences between wild and hatchery reared salmonid juveniles and suggest these dif-ferences may affect relative survival rates. Sosiak et al. (1979) showed that hatchery reared Atlantic salmon had a less varied diet after release than did wild juveniles, pre-sumably as a result of a reduced repertoire of search images which limited the extent and efficiency of feeding by released fish. The effect of lack of natural experience on the fright behaviour of hatchery reared rainbow trout also has been noted by Ritter and MacCrimmon (1973). Behavioural differences between wild and hatchery fish may be a result of size differences. The mean length of hatchery chum (52 mm) was significantly larger than that of wild chum (38 mm). If predation intensity is greater on larger prey, the dif-ference in size of hatchery and wild chum may account for the greater antipredator 90 responses shown by hatchery fish. This explanation seems unlikely as reports of preda-tion on chum fry indicate that predators selectively preyed on the smallest fry (Parker 1971; Fresh et al. 1980; Hargreaves and LeBrasseur 1986). Furthermore, results from Experiment 2 (Chapter 3, Figure 3.4) demonstrate that attack rates by trout were the same on wild and hatchery chum in single and mixed species treatments. Positive effects of stickleback presence on chum foraging behaviour The presence of stickleback increased the foraging efficiency of chum, particu-larly hatchery chum. In treatments with stickleback, chum feeding rates and feeding indices increased (Figure 4.4). When stickleback were added to predator trials, the diet composition, the proportion of fish feeding, and feeding rates of hatchery chum were similar to treatments of hatchery chum alone. Stickleback had a similar effect on wild chum feeding indices. There are several possible explanations for the positive effects of stickleback presence on chum foraging behaviour. Stickleback may provide an alternate food source for trout and since stickleback require greater handling time, time available to pursue chum might be reduced. If the behaviour of trout while handling stickleback makes them more conspicuous, chum might require less time to monitor approaching predators and have more time for feeding. Therefore stickleback may reduce predation pressure on hatchery chum by interfering in their detection or attack. However, none of these explanations seems likely, as attack rates per individual chum remained the same in single and mixed species treatments as shown in Experiment 2 (Chapter 3). A more likely interpretation is that stickleback may function as "dither" animals (Barlow 1968), i.e. the presence of stickleback could have a calming influence on chum behaviour. Barlow proposed the use of dither animals in certain types of experimental 91 procedures to reduce the fright response of experimental animals. As stickleback gen-erally did not flee from approaching trout and their feeding behaviour was relatively unaffected by predation it was as if stickleback had been introduced to have a calming effect on chum. An alternative explanation is that of social facilitation. Watching stickleback feeding behaviour may facilitate or enhance chum feeding, e.g. facilitate prey detection. Keenleyside (1955) and 011a and Samet (1974) reported that threespine stickleback and juvenile mullet (Mugil cephalus), respectively, initiated feeding sooner by observing conspecifics feeding than either by observing fish not feeding or by the presence of food alone. Obrebski and Sibert (1976) suggest that stickleback may actually increase food availability for chum by increasing disturbance of the substrate surface. This did not seem the case in my experiments. The increase in feeding rates cannot be attributed to increased Tubifex in the diet of chum. A third possible explanation is that replacing chum with an equivalent number of stickleback reduced intraspecific competition for food. In mixed prey trials without trout, the diets of both hatchery chum and stickleback tended to include more food types that were not shared (Figure 4.3). Studies on fish and birds, summarized in Lima and Dill (1990), have shown the importance of foraging groups; individuals spend less time being vigilant with increasing group size. However, a trade-off exists in groups between increased feeding efficiency and increased competition. Not only is group size important, but the position of the individual within the group and the group's composition is important when determining the degree of risk and resource use. In mixed species groups, overall competition may be reduced if each species has different preferences (Allan and Pitcher 1986). Van Havre and FitzGerald (1988) noted that a trade-off between safety and competition affects shoal size in stickleback; hungry fish form smaller shoals. These three alternatives will be further investigated in Chapter 6. 92 Effects of other species on stickleback foraging behaviour The foraging behaviour of stickleback was least affected by the presence of other species. Although feeding rates and feeding indices of sticklebacks tended to decrease, only the percentage of fish feeding showed a significant reduction when trout were present. The morphology of stickleback reduces the probability that an individual will be consumed if attacked, so that flight is infrequent and more time can be spent feeding in good habitats (McLean and Godin 1989). The presence of chum appeared to inhibit feeding by stickleback or increased competition for food items, as Tubifex formed a greater percentage of the stickleback diet (non significant 15% increase) when associated with chum. The lack of Drosophila in the diet appeared to be the result of preference by the stickleback, as they were readily consumed when presented as a single food type to fish in the conditioning tanks. However, water depths were much lower in conditioning tanks than in the experimental arena and surface items may have been more visible. Relationship between foraging behaviour and antipredator strategies Chum and stickleback displayed different behavioural responses to predation. Hatchery chum exhibited the greatest reduction in foraging efficiency. The results presented suggest there is a relationship between feeding behaviour and antipredator strategies. Chum avoided encounters with predators and fled when attacked. Predator avoidance reduced the time available for feeding. Those animals not feeding were likely those that have been attacked. When chum altered their feeding behaviour and activity levels to avoid trout, they also affected their encounter rates with other prey and with the common resources required by both. For stickleback, foraging versus predation risk trade-offs may not be as critical. The morphological defense used by stickleback decreased the probability they will be consumed if captured. Plates and spines are probably effective against all types of predators that stickleback would encounter in the wild. 93 5. Habitat Use Introduction Predation can have strong effects on the distribution and abundance of prey species (Cerri and Fraser 1983; Werner et al. 1983a; Gilliam and Fraser 1987). In the natural environment food is patchy, and patches vary in their abundance of food. Optimal foraging theory predicts that foragers should optimize net energy intake by feeding in patches with the greatest energetic return (see Stephens and Krebs 1986 for review). But patches of high food abundance are likely to be areas of high predation risk (Gilliam 1982; Milinski 1986; Dill 1987). Risk of predation, encounter rate with predators, and capture rate also vary with habitat (Lima and Dill 1990). Hence, the selection of a feeding habitat involves evaluating both costs and benefits and making trade-offs to maximize fitness. Habitat use patterns provide insight on how prey respond to variable predation risk. Individuals can reduce the risk of predation by habitat shifts or by reduced movement within a habitat. Juvenile salmon show marked vertical and diel habitat shifts in environments with many predators (Bams 1969; Reimers 1971; Baker 1978; Levy 1987). Chum fry delay migration downstream until dark (Neave 1955; Hunter 1959; Kobayashi 1958, 1964; Fresh et al. 1980), when predators are least efficient at catching prey (Reimers 1971; Godin 1980). Vertical migrations of juvenile sockeye salmon in Babine Lake, British Columbia, may be interpreted as responses to daily changes in predation risk and food availability characteristic of various depth strata (Baker 1978; Levy 1987; Clark and Levy 1988). Shifts in habitat use patterns may have associated costs and may be expressed in reduced feeding rates (Sih 1982, Notonectids; 94 Werner et al. 1983a, sunfish), lower food consumption (Stein 1979, crayfish), and increased competition for food or space (Mittlebach 1984, sunfish). Predator-prey interactions can be moderated by structurally complex environ-ments that act as refuges for the prey and reduce the efficiency of the predator. If prey actively avoid predators, structured environments will facilitate predator avoidance. The success of predators feeding on fish and invertebrates in freshwater systems is re-duced as vegetation density increases (Helfman 1986; Werner et al. 1983a; Stein 1977; Savino and Stein 1982, 1989; Gilinsky 1984). In laboratory experiments conducted by Glass (1971) the capture rate of bass on guppies decreased with increasing density of vertical wooden dowels, demonstrating that guppies were less vulnerable to bass in more complex habitats. Savino and Stein (1982) proposed three mechanisms to explain reduced prey vulnerability in complex habitats: (1) increased structural complexity may create visual barriers and reduce visual encounters with predators (Cooper and Crowder 1979) (2) vegetation may modify prey behaviour and thereby alter the probability of detection and capture, and (3) predator behaviour may be modified by vegetation. As vegetation density increases predator maneuverability may be reduced. In experiments using simulated submerged vegetation, Savino and Stein (1989) observed that at high plant densities, largemouth bass switched from searching for prey to ambushing prey. Minello and Zimmerman (1983) found that the foraging success of Atlantic croaker (Micropogonias undulatus), an inefficient predator that required more than one strike per capture, was reduced in vegetation. Ontogenetic shifts in habitat use are common in fishes (Werner and Gilliam 1984) and have been associated with size-specific predation risks and foraging gains (Werner et al. 1983a, 1983b; Mittlebach 1986; Mittlebach and Chesson 1987). Small in-dividuals of different species are often restricted to the same protective habitat (Werner 95 et al. 1977; Mittlebach 1984) resulting in increased overlap of resource use and reduced growth. Stickleback and juvenile chum may be restricted to vegetated habitats in the lower Fraser River due to predation risk. Studies of estuary utilization by juvenile salmonids suggest these habitats are important as feeding and rearing areas for chum, where chum may be able to outgrow the risk of predation to some predators before migration into the open ocean. Costs and benefits change with the size of the an-imal, such that the optimal habitat changes as the animal grows. At each size an animal should choose the habitat characterized by the minimum ratio of mortality rate/growth rate (Gilliam 1982; Werner and Gilliam 1984) or the minimum ratio of mortality rate/feeding rate (Gilliam and Fraser 1987). Not only does intimidation by predators shape patterns of habitat use and there-fore resource exploitation, but it can also influence species interactions (Gilliam and Fraser 1987). Differences in the vulnerability of taxa to predation have substantial implications for species interactions in predator refugia. Therefore, the relative vul-nerabilities of species should also be considered when assessing habitat use by prey. Body form, behaviour, and plant density (i.e. cover) influence relative vulnerability (Mauck and Coble 1971; Vinyard 1980; Laur and Ebeling 1983). Savino and Stein (1989) found that largemouth bass (Micropierus salmoides) and northern pike (Esox lucius), search and ambush predators respectively, were more efficient at capturing fathead minnows (Pimephales promelas) than bluegill sunfish (Lepomis macrochirus) because sunfish modified their behaviour more than minnows in response to predation. In this study I tested the hypothesis that prey assess predation hazard and modify their habitat use under risk of predation. Chum fry were more vulnerable than stickleback to capture by trout (Chapter 3) and therefore I predicted that the presence of trout should alter habitat use by chum fry more than stickleback. The first part of this investigation examined the effect of predation risk on foraging site use 96 within a habitat and the role of alternate prey in determining habitat use. The second part examined how the presence of predators influenced the use of refugia, specifically simulated vegetation, and compared the antipredator responses of two different prey types. I considered two questions. (1) Do prey assess risks and benefits within a habitat and trade-off energy gain and predator avoidance and (2) do habitat use patterns of differentially vulnerable prey differ in response to predation in structurally complex habitats? These questions were investigated by observing habitat use in the laboratory. Methods s Experiment 3: Effect of predation on foraging site within a habitat The effect of predation risk on habitat use was examined by observing the po-sition of feeding prey in the water column in the presence and absence of trout and alternate prey. Since these observations were made simultaneously with the study of effects of predation on foraging efficiency (Experiment 3), the experimental protocol was the same as that outlined in Chapter 4. Observations were made on habitat use by wild chum and stickleback in six species combinations including: (1) chum, (2) stickleback, (3) chum and stickleback, (4) chum and trout, (5) stickleback and trout, and (6) chum, stickleback and trout. Prey position was recorded as the percentage of prey in each of three positions defined by water depth: surface (0-12 cm) , mid-water (12-36 cm), and bottom (36-50 cm). Prey position was recorded every ten minutes in five one hour trials for each species combination. Changes in habitat use of wild chum and stickleback due to treatment were tested by a three-way A N O V A of position, alternate prey presence, and predator pres-ence. All percentage data were transformed (arcsine square root) to normalize dis-97 tributions for analysis. Resulting statistics were back transformed to percentages for presentation. Statistical probabilities exceeding 0.05 were considered non significant. Experiment 6: Effects of predation on refugia use To determine if the presence of refugia affects habitat use by prey under risk of predation (Experiment 6), wild chum and stickleback were given a choice of habitats in the presence and absence of trout and alternate prey. The habitat alternatives were: open water with high food density but high risk of predation, or vegetation with low predation risk but low food density. The number of chum and stickleback (20 per trial) was kept constant in each treatment. Single species treatments used 20 fish of one species, while mixed species treatments used 10 fish of each species. Four replicates of each treatment were randomly performed with stickleback (mean length 45.8 ± 3.7 standard deviation) and wild chum (mean length 44.3 ± 6.3 standard deviation) of similar sizes to avoid prey size selection. The experimental tank (180 x 120 x 60 cm) was constructed of plywood with a single plexiglass side providing a viewing window. All observations were made from behind a black plastic screen so fish would not be disturbed. The tank bottom was covered with a 2-4 cm layer of sand and fine gravel and a 40 x 120 cm zone of vegetation at one end provided refuge from predators (Figure 5.1). Vegetation was simulated by tapered green plastic strips (2 cm basal width, 50 cm long, and 250 strips per m) attached to the tank bottom. The base of the vegetation was covered by substrate and strips floated just below the water surface. Fish were deprived of food for 48 h prior to testing to standardize hunger. Chum and stickleback were placed in the experimental tank 24 h before the trial began and 40 of each food type (Tubifex, Daphnia sp. and adult Drosophila) added to the center of the open water habitat. Chum and stickleback had been preconditioned to these 98 V E G E T A T I O N Z O N E 40 c m •** F O O D DEL IVERY O U T L E T \ \ \ \ \ S S S \ 10 cm * * * * * * * * * * . C I J R C T R A T F * * * ' / * * * * * * * * , * * * * * * * * * * * 180 c m -Figure 5.1 The apparatus used in Experiment 6 to determine the effect of predation on habitat use by chum fry and stickleback. The area of simulated vegetation was defined as the vegetated habitat and the remainder of the tank was defined as the open water habitat. 99 food types for at least two weeks before testing. In trials with a predator a trout was added 5 h before the trial began to allow it to recover from handling and to increase prey awareness of predation hazard. Any fish consumed by trout were replaced before testing. At the start of each trial, 400 items of each food taxa were added to the center of the open water habitat by the methods described in Chapter 4. Although Daphnia tended to disperse throughout the water column, I assumed all three food types were more concentrated in the open area than in the vegetation during the experiment. Disturbance created by the food additions sometimes startled the trout; if the trout did not attack within the first 15 min the trial was discontinued. Recordings of habitat use in trials containing trout began after the first attack. The number of chum and stickleback in the open water habitat and the vegetation area was recorded every 10 min for a total of one hour. To assess the effects of the three factors - prey species, presence or absence of alternate prey and presence or absence of predators - a three-way factorial A N O V A was performed on the mean percentage of prey using the open water habitat in each treatment. Percentage data were normalized by arcsine square root transformations. Statistical probabilities exceeding 0.05 were considered non significant. Data were back-transformed to percentages for presentation. 100 Results Experiment 3: Effect of predation on foraging site within a habitat Wild chum alone generally used mid-water habitats, unless trout were present, when most fish switched to positions at the surface (Table 5.1a). When stickleback were present, chum used mid-water habitats. Stickleback used bottom habitats in all treatments (Table 5.2a), particularly when other species were present. Results of analyses of variance on prey position, in the presence and absence of alternate prey and predators, are presented for chum in Table 5.1b and for stickleback in Table 5.2b. The most significant result, for both species, was a second order interaction between position, prey presence and predator presence. My interpretation of this interaction is that the habitat use by chum and stickleback depends not only on the characteristics of the foraging position, but whether alternate prey and predators are present. Experiment 6: Effects of predation on refugia use Habitat use patterns for chum and stickleback in the refugia experiment are presented in Table 5.3a. In the absence of other species, 69% and 82% of chum and stickleback, respectively, occupied the open water habitat. The presence of trout significantly altered habitat use by prey (Table 5.3b, three-way A N O V A , P<0.001). When trout were present the percentage of prey in the open water decreased by 39% for chum and 24% for stickleback. A significant interaction between species and alter-nate prey indicated that the percentage of each species using the open water depended on whether alternate prey were present. Alternate prey presence increased the per-centage of chum but decreased the percentage of stickleback in the open water. It is notable that the greatest percentage of chum (78%) used the open water habitat when stickleback were present. 101 Table 5.1a Habitat use by wild chum in the presence and absence of stickleback and trout (Experiment 3). The results are mean percentages (back-transformed from A N O V A results) from five replicate experiments. trout absent present stickleback surface mid-water bottom surface mid-water bottom absent 20 49 17 72 0 28 present 16 43 35 21 49 16 Table 5.1b Three-way factorial A N O V A of stickleback presence, trout presence and position on habitat use by wild chum (Experiment 3). Significant effects are indicated by an asterisk (*). source SS df MS F P stickleback 0.0134 1 0.0134 0.123 0.728 predator 0.0109 1 0.0109 0.100 0.754 position 0.0724 2 0.0362 0.332 0.719 stickleback x trout 0.0001 1 0.0001 0.001 0.975 stickleback x position 1.0217 2 0.5108 4.686 0.014* trout x position 1.0888 2 0.5444 4.994 0.011* iticklebackx trout x position 1.2811 2 0.6405 5.875 0.005* error 5.2330 48 0.1090 102 Table 5.2a Habitat use by stickleback in the presence and absence of chum and trout (Experiment 3). The results are mean percentages (back-transformed from A N O V A results) from five replicate experiments. trout absent present chum surface mid-water bottom surface mid-water bottom absent 9 33 56 8 7 84 present 4 9 86 18 13 64 Table 5.2b Three-way factorial A N O V A of chum presence, trout presence and posi-tion on habitat use by stickleback (Experiment 3). Significant effects are indicated by an asterisk (*). source SS df MS F P chum 0.0012 1 0.0012 0.029 0.866 trout 0.0002 1 0.0002 0.004 0.951 position 6.3244 2 3.1622 78.249 0.000*** chum x trout 0.0037 1 0.0037 0.091 0.764 chum x position 0.0631 2 0.0315 0.780 0.464 predator x position 0.1669 2 0.0834 2.065 0.138 chum x trout x position 0.7050 2 0.3525 8.723 0.001*** error 1.9398 48 0.0404 103 Table 5.3a Mean percentage of chum and stickleback using open water habitat, versus vegetated refugia, and associated standard error (se) in the presence and absence of alternate prey and predators (Experiment 6). The data are percentages back-calculated from transformed data, from four replicate experiments. alternate prey trout mean percentage species present present in open se chum no no 69 0.3 no yes 30 0.4 yes no 78 0.2 yes yes 51 1.4 stickleback no no 82 0.1 no yes 58 1.6 yes no 43 0.3 yes yes 38 0.4 Table 5.3b Results of an A N O V A on the effects of species, alternate prey presence, and predator presence on the percentage of fish in open water versus vegetation (Ex-periment 6). Significant effects are indicated by an asterisk (*). source SS df MS F P species 0.0028 1 0.0028 0.114 0.739 alternate prey 0.0478 1 0.0478 1.975 0.173 trout 0.4958 1 0.4958 20.474 0.000 species x alternate prey 0.4304 1 0.4304 17.773 0.000 species x trout 0.0652 1 0.0652 2.694 0.114 alternate prey x trout 0.0565 1 0.0565 2.334 0.140 species x alt. prey x trout 0.0056 1 0.0056 0.229 0.636 error 0.5812 24 0.0242 104 Discussion Experiment 3: Effect of predation on foraging site within a habitat These results demonstrate that prey assess predation risk and modify habitat use when under risk of predation. Chum and stickleback displayed species-specific shifts in habitat use in response to the presence of predators and alternate prey. Single species groups of both chum and stickleback avoided the mid-water habitat when a trout was present. Chum responded to trout by shifting from mid-water to the surface. When stickleback were present the effect of trout on chum habitat use was negligible. Stickleback responded to the presence of trout or chum by shifting even farther toward the bottom. The habitat shift by wild chum to avoid predation in mid-water agrees with the results of Experiment 5 on position-dependent vulnerability (Chapter 4). Ob-served attack frequencies were greatest for prey in the lower half of the water column. The shift by chum to the surface may be an effective method of avoiding predators. Habitat shifts to avoid predators have been reported for other species, both fish and invertebrates. When exposed to aerial predators, guppies tend to stay in deeper wa-ter and become less active (Seghers 1974). Charnov et al. (1976) found that mayfly nymphs (Baetidae) responded to sockeye salmon (Oncorhynchus nerka) by moving to the corners of experimental tanks. Although shifts by stickleback to bottom habitats appear inappropriate and increased predation risk, they agree with predictions related to species-specific vulner-ability. Compared to chum, stickleback are much more cryptic against the substrate and may gain greater benefit by remaining motionless on the bottom. Advantages to remaining motionless in response to a predator have been reported for stickleback 105 (McPhail 1969) and other fish species (Pfeiffer 1962; Neill 1970). This tactic is effective if in ceasing to move prey become invisible to the predator (Neill 1970). Experiment 6: Effects of predation on refugia use The increased use of vegetated habitats with little or no food by both chum and stickleback in the presence of trout is evidence that prey were willing to trade-off energy gain for predator avoidance. Based on laboratory observations of predator-prey interactions in the vegetation it appears that vegetation was an effective visual barrier to predators. Chum and stickleback used the vegetation as cover; individuals hid motionless behind clumped strands. Although schooling was a common antipreda-tor response by chum in unstructured environments, when vegetation was available individuals left the school for the safety of cover. Schools did not reform in the vegeta-tion. Schooling in vegetation may actually increase risk as it may provide groups large enough for the trout to find. Movement to protective habitats to avoid predation is a common response of vulnerable species (crayfish, Stein and Magnusen 1976; notonec-tids, Sih 1980, 1982; catfish, Power 1984; and squirrels, Lima et al. 1985). Fraser and Cerri (1982) observed the response of blacknose dace Rhinichthys atratulus to preda-tors in artificial streams and noted the presence of structure was the most important determinant of habitat choice by prey. Chum and stickleback have adapted to predation risk by species-specific shifts in habitat use. The extent that the two species shifted spatial distributions within the water column and from open water to vegetation in the presence of predatory trout can be related to their relative probability of damage if captured. Compared to stickleback that rely on body armour, chum rely on evasive behaviour to avoid predators and thus show greater shifts in habitat use patterns than stickleback. All chum and stickleback did not respond in the same manner. Given the increased protection from trout in the vegetation, some prey continued to feed in the 106 open water habitat. I was unable to determine if these prey were specific individuals or if all prey periodically left the refuge to feed. Werner et al. (1983a) suggested that individual variation in the use of vegetative refugia by bluegills was due to the small size of the experimental ponds. Bluegills could feed in the more profitable open habitat and flee to cover when approached by a largemouth bass. Given the size of my experimental aquaria, chum and stickleback may have used the refuge in a similar way. Alternatively, changes in the level of trout activity may have changed the perceived level of risk and cost of feeding. Chum and stickleback may have associated periods of low predator activity with reduced risk and reduced costs of feeding. Dill and Fraser (1984) and Metcalfe et al. (1987a) reported that the response of salmon to predators is ongoing; salmon constantly update the trade-off between feeding and predator avoidance. Metcalfe et al. (1987a) suggested that salmon assess risk according to the time elapsed since the predator was last seen. A behavioural approach to studying habitat use in the laboratory is useful for interpreting species interactions in natural communities. In the wild, vegetated refuges also may be areas of high food abundance. Two possible reasons nearshore vegetated habitats may be used by chum and stickleback are; availability of food or avoidance of predators. If vulnerability decreases with complex vegetated habitats, selection of these habitats even if they yield poor energy returns would increase survival. This study may partially explain the presence of chum and stickleback in vegetated habitats in natural communities. Prey are less likely to shift away from habitats with predators if the habitats are structurally complex. The effects of shifting habitat use on fish growth and mortality rates in response to predators in natural communities would be of interest. Although much attention has been given to the importance of Fraser estuary marshes for juvenile salmon rearing (Levy et al. 1979; Levy and Northcote 1981, 1982), little attention has been given to other habitat types. My results suggest the importance of maintaining complex estuarine habitats, i.e. marshes, sloughs, riparian areas, for the benefit of juvenile salmonids. 107 6. Mechanism of Enhanced Feeding Rates by Chum Fry Introduction Results of experiments on prey foraging behaviour (Chapter 4) demonstrated that the presence of stickleback enhanced the feeding rate of chum when trout were present. Several possible explanations were proposed to account for the increased feeding rates. One possible explanation is that stickleback may reduce the fright response of chum by acting as "dither" animals. Barlow (1968) recognized the ability of zebra and pearl danios (genus Brachydanio) to reduce the occurrence of fright behaviour or hyperactivity in isolated cichlids. The danios were suitable dither fish because their behaviour did not resemble, that of the experimental fish. An alternative hypothesis is that stickleback may facilitate food detection by chum. Mixed-species social groups have been reported frequently for birds, and to a lesser extent, mammals and fish. Individual birds may enhance their feeding rates by observing the foraging behaviour of others and joining them or by seeking similar patches (Krebs et al. 1972; Krebs 1973; Morse 1980). Great tits show a strong ten-dency to join others who have found a food item and then to forage in that area (Krebs et al. 1972). Observations of Jamaican reef fishes (Itzkowitz 1974) and coral reef fishes in the Red Sea (Fishelson 1977) showed that members of groups respond to individu-als foraging, although Itzkowitz cautions against equating mixed-species schools with flocks of birds. He suggests that minority individuals in mixed-species schools are sub-jected to disproportionately high predation. Mixed-species flocks of birds tend to be less dense than mixed-species schools and oddity may not be an important cue used by avian predators (Morse 1980). 108 Lima and Dill (1990) reviewed studies demonstrating the importance of foraging group size in facilitating predator detection. Visual and auditory alarm responses com-municate a warning and animals in groups spend less time in surveillance than when alone. Chum may incur reduced predation risk in mixed-species groups if stickleback are superior to chum at detecting approaching predators. However, a trade-off exists between increased predator detection, increased feed-ing efficiency in groups, and increased competition. This trade-off suggests a third explanation for the enhanced feeding rates of chum with stickleback; competition for food may be reduced in mixed-species schools. For a given group size, diet overlap in mixed-species schools should be less than in single-species schools if each species has different food preferences (Allan and Pitcher 1986). This portion of my work investigated the mechanism by which stickleback en-hanced the foraging efficiency of chum under risk of predation. Comparisons of chum feeding behaviour were made to determine if enhanced feeding efficiency in the pres-ence of stickleback was due to social facilitation of food or predator detection, reduced intraspecific competition, or reduced fright in the presence of dither animals. Methods Experiment 7 was conducted to quantify changes in chum foraging efficiency when stickleback were provided as dither animals. The experimental apparatus and protocol were the same as in Experiment 3 (Chapter 4) with the addition of dither fish. A dither compartment was formed by enclosing a section of the experimental tank (120 x 10 cm) opposite the predator compartment using a clear plexiglass divider (Figure 6.1). The sticklebacks were visible to the chum but the two prey types could not interact directly. 109 During the conditioning period, hatchery chum were placed in the center ex-perimental arena, stickleback in the dither compartment, and trout in the predator holding area. Waterflow and hence odors were able to circulate freely between com-partments. In order to make results comparable in this experiment with observations of chum feeding in Experiment 3 (Chapter 4) chum density in the experimental arena was maintained at 20 fish per trial. Ten stickleback were used as dither animals in each trial corresponding to the number used as alternate prey in mixed prey trials in Ex-periment 3. Food was provided for chum but not stickleback. Under these conditions, I assumed stickleback were unable to facilitate the detection of food by chum. Trials were run for one hour, and the frequency of attack and attack success recorded. The attack rates on chum were compared by a Kruskal-Wallis H test (KWH, nonparametric ANOVA, Kruskal and Wallis 1952) to results obtained in Experiment 2 (Chapter 3) for attack rates on chum alone and chum with stickleback. Stomach contents of chum were removed and enumerated. The diet composition (percent com-position by number) and the feeding rate (number of food items per hour) was deter-mined. All percentage data were normalized by arcsine square root transformations before analyses. The feeding rates and diets of chum were compared by a one-way ANOVA to results obtained in Experiment 3 for chum alone, chum with a predator, and chum with a predator and stickleback. Significant differences between treatments were compared by multiple comparisons using the Student-Newman-Keuls test (SNK, Zar 1984). Results were back transformed to percentages for presentation. 110 100 cm ~ T T \ F L U O R E S C E N T L I G H T S F O O D Figure 6.1 The apparatus used in Experiment 7 to quantify changes i n chum foraging behaviour when stickleback were provided as dither. Stickleback were confined to a separate compartment on the right side of the tank. The divider between trout and chum (in the center compartment) was raised at the onset of a trial . Water circulated freely between all compartments. I l l Results The attack rates by trout on chum, chum with stickleback (Experiment 2), and chum with dither animals (Experiment 7) are compared in Figure 6.2. The attack rate per individual chum was similar in all treatments (median 1.4, KWH, P=0.32); the at-tack rate on chum by cutthroat trout was independent of the presence of alternate prey or dither animals. Capture success was the same in all treatments (median 0, KWH, P=0.13). Attacks were occasionally directed at stickleback in the dither compartment. Mean feeding rates for chum alone, chum with trout, chum with trout and stick-leback (Experiment 3), and chum with trout and dither animals (Experiment 7) are presented in Table 6.1. Mean feeding rates were significantly different among treat-ments (ANOVA, P<0.001). Multiple comparisons between treatment means (SNK test, Table 6.1) confirm that the feeding rate of chum was highest when with both stickleback and a trout. The feeding rates of chum alone and chum with a trout and dither animals were similar. The lowest feeding rate was observed for chum with a trout, without stickleback or dither animals. The composition of the chum diet during each treatment is presented in Figure 6.3. The diet of chum with trout was significantly different from all other treatments (Drosophila, ANOVA, P=0.002, SNK; also Daphnia, ANOVA, P=0.003, but an SNK test was not sensitive enough to distinguish differences among treatments). In the absence of alternate prey or dither animals, chum fed primarily on Daphnia when trout were present. The diet of chum in the dither trials was similar to that of chum alone or chum with stickleback and predators. 112 hatchery chum with . . . predator predator and stickleback predator and dither attacks per individual Figure 6.2 Comparison of attacks rates by cutthroat trout on hatchery chum (mean number per individual per hour) in three treatments: chum with trout, chum with stickleback and trout (Experiment 2), and chum with trout and stickleback as dither fish (Experiment 7). Data in each treatment are summarized by boxplots. The 5 vertical lines on each boxplot portray 5 percentiles whose P-values, from left to right, are 10, 25, 50 (median), 75, and 90. Treatments were replicated at least six times. The mean number of attacks per individual did not differ between any treatments ( K W H p=0.32). 113 Table 6.1 Comparison of mean feeding rates (number of food items per hour) of chum fry in four treatments (1-with trout, 2-with trout and dither, 3-alone, and 4-with trout and stickleback). Significant differences between treatments were determined by a Student-Newman-Keuls test and are indicated by an asterisk (*). Treatments in which chum have similar feeding rates are underlined. (SE)—standard error, p—range of means, q—difference of ranked means/SE). treatment species combination n mean feeding rate 95% CI 1 chum with trout 7 12.1 3.9 2 chum with trout and dither 7 22.6 5.7 3 chum alone 6 30.9 9.0 4 chum with stickleback and trout 6 41.8 11.7 A N O V A source SS df MS F P treatments 3065.4 3 1021.8 17.1 0.000 error 1311.2 22 59.6 Student-Newman-Keuls Test 1 2 3 4 ranked sample means 12.1 22.6 30.9 41.8 Ranked Mean Comparison Difference SE Q P <Z0.05,22,JJ 4-1 29.7 3.04 9.77 4 3.958* 4-2 19.2 3.04 6.32 3 3.578* 4-3 10.9 3.15 3.46 2 2.950* 3-1 18.8 3.04 6.18 3 3.578* 3-2 8.3 3.04 2.73 2 2.950 2-1 10.3 2.92 3.53 2 2.950* 114 Q Drosophila B Daphnia B Tubifex 1 2 3 4 ® ® • @ treatment Figure 6.3 Comparison of mean diet composition of hatchery chum in four treatments: 1—chum with trout; 2—chum with trout and stickleback as dither; 3—chum alone; and 4—chum with trout and stickleback (Experiments 3 and 7). The number of filled circles (•) denote the number of prey types, (D) denotes dither animals, and open circles ( 0 ) denote trout present. The vertical lines are 95% confidence intervals for the mean. Significant differences in diet composition are indicated by an asterisk (*). Treatments were replicated at least six times. 115 Discussion Attack rates on chum in the various treatments were not statistically different; therefore, changes in foraging rates and diet composition resulted from factors other than rate of attack by predators. The only variables that differed among treatments were the presence of stickleback and the relative density of chum. Chum feeding rates decreased significantly when under risk of predation. When stickleback were visible in an adjacent compartment, chum with trout fed at rates comparable to chum in treatments without trout. Thus, in the presence of dither fish, the effect of predation was reduced and chum resumed feeding at near normal rates. Stickleback as dither One explanation suggested for the enhanced feeding rates of chum is that stickle-back act as dither animals and have a calming influence on chum behaviour. Generally dither animals are used as part of the experimental protocol; however, the use of stick-leback in my experiments was as if stickleback had been been introduced as dither. Stickleback were suitable dither fish because they did not exhibit strong fright re-sponses and usually "froze" in response to approaching predators. It is unlikely that observations of stickleback facilitated predator detection by chum. Stickleback did not flee from trout and therefore, compared to chum, stickleback are less likely to provide recognizable visual stimuli signaling alarm or predator detection. Because chum may not detect an approaching predator in the presence of stickleback they may delay flight and feed for a longer period. Suboski (1988) reviewed evidence, in several fishes, of predator recognition from conspecific behaviour based on social communication. Predator evasion behaviour functions as a releasing stimulus for similar behaviour in naive conspecifics who have been observing the evasion (011a and Samet 1974; Godin and Morgan 1985). Visual 116 species-specific alarm displays have been reported for the X-ray fish, (Pristella riddlei, Keenleyside 1955) and suggested for the minnow (Phoxinus phoxinus, Magurran et al. 1985; Pitcher 1986). Barlow (1968) recommends the use of dither animals as visual stimulation in experiments to reduce fright responses of experimental animals. The practise of pro-viding visual dither has been used in the study of many species in different animal groups (cichlids, Dupuis and Keenleyside 1982; crabs, Molenock 1975, 1976). Barlow (1968) also recommends that a different species than the experimental animal be used as the dither animal. Species with different behavioural responses are most useful as dither. However, Milinski (1984) successfully used stickleback that had been habitu-ated to the experimental apparatus, as dither fish when studying feeding strategies of parasite infected stickleback. It is posssible, but unlikely, that the increased feeding efficiency of chum in the dither trials was simply a function of increased fish density. The greatest increase in chum feeding efficiency was observed in trials in which stickleback were also present as alternate prey and in which the total fish density was less than that in the dither trials. To determine whether the response by chum was a result of a true dither effect of stickleback on chum or a function of increasing fish density, an additional treatment with 20 chum in the experimental compartment and 10 chum in the dither compartment would have to be performed. Reduced intraspecific competition A second possible explanation for the greater feeding rates of chum in mixed-species groups is that intraspecific competition for food is reduced. It is important to note that the number of chum used in the dither trials was equal to the number used in the chum alone or the chum with predator trials. Therefore the greater feeding rates 117 in dither trials could not be attributed to decreased intraspecific competition because the number of chum remained the same. Chum feeding rates in the dither trials were significantly higher than in the chum with predator trials but not significantly different from the trials with chum alone. As stickleback were restricted to a separate compartment without food, it is unlikely their behaviour facilitated food detection. The significant increase in chum feeding rates in treatments with unconfined stickleback and trout, greater than in the dither trials, suggests reduced intraspecific competition may also be an important factor. In treatments with trout and stickleback, chum benefit by reduced fright (as measured by feeding rates) due to the dithering effect of stickleback as well as a reduction in the number of conspecifics and therefore intraspecific competition. Since differences in resource use between species are gen-erally greater than those within species, demands for any particular resource should be less among individuals of different species than of the same species. Morse (1980) states that, with few exceptions, interspecific competition is much less intense than intraspecific competition in mixed groups, as judged by frequencies of interaction. Social facilitation A third possible explanation for increased feeding rates with stickleback is that chum feeding may be facilitated through observational learning. By watching stick-leback feed, chum learn to detect prey and increase the range of prey types in their diet. If species overlap in their resource exploitation, some advantage may be gained in mixed species groups since different species search for food differently or in different places. Consequently, foraging in mixed-species groups may result in a greater range of foods discovered during common exploitation. Naive fish may acquire recognition of a novel foodstuff by observing an experienced forager (Suboski 1988). This expla-nation is not consistent with my data. Stickleback may facilitate the detection of food 118 types typically consumed by chum, but the range of food types in the diet was not increased (Experiment 3). My results agree with the assertion of Morse (1980), that the advantage of observational learning should be lower in mixed groups than in single species groups, as resources used by one species are less likely to be used by another species. These results suggest that the increased feeding rates of chum when with stick-leback can be attributed to the calming effect of dither fish and reduced intraspecific competition for food. These results have important implications for interspecific in-teractions affecting the survival of chum salmon fry in estuaries. In the wild, if chum school with other prey species, e.g. stickleback, they may reduce the intensity of in-traspecific competition for preferred food types while maintaining the advantages, other than predator detection, associated with groups.. Sambrook (1990) reported stickleback and chinook salmon fry feeding together in tidal channels of Ladner Marsh, in the lower Fraser River. Data presented in Chapter 2 for diet overlap, and the co-incidence of chum fry and stickleback in the seine hauls, suggests the potential for mixed-species feeding groups. 119 7. Conclusions 1) Many fish species utilized nearshore habitats of the Fraser River estuary for feeding and resting. Chum fry shared these habitats with many other species, both potential competitors and potential predators. Dietary, spatial, and temporal overlap between chum and stickleback suggests the potential for interspecific competition if resources become limited (Chapter 2). 2) Chum and stickleback each assessed the intensity of predation risk and mod-ified their behaviour accordingly, and the level of the response was a function of the degree of risk. Behavioural responses to hungry predators were greater than responses to satiated predators. The relative degree of risk and of the cost of avoiding predators was not the same for chum and stickleback. Relative susceptibility and the strength of avoidance behaviours are related to the type of defense employed by chum and stick-leback. Chum avoided predators to a greater extent than did stickleback (Chapter 3). 3) Trout had considerable effects on chum and stickleback foraging efficiency, as measured by diet composition, foraging location, the number of fish feeding, feeding indices, and feeding rates. Prey traded-off foraging efficiency for predator avoidance. The feeding behaviour of hatchery chum was affected more severely by trout than that of wild chum or stickleback. Hatchery chum exhibited a dietary shift in food items and a reduction in the number of fish feeding in the presence of trout. W i l d chum responded to predation risk by reduced feeding rates; stickleback responded with fewer fish feeding (Chapter 4). 4) The intensity of the effects of predation on chum foraging behaviour depended on whether stickleback were present. Changes in chum foraging behaviour caused by predation risk were suppressed when stickleback were present (Chapter 4). 120 5) The degree of risk of predation was dependent on the foraging site within a habitat. Chum and stickleback were attacked more frequently within 24 cm of the substrate and may be more vulnerable to attack when feeding on food items in this location (Chapter 4). 6) Trout affected patterns of habitat use, resource exploitation by prey, and species interactions when chum and stickleback responded to predation risk by avoid-ance behaviours. Chum assessed predation risk and traded-off energy gain for predator avoidance. When given a choice between habitats differing in food quality and risk of predation, chum traded-off a food-rich habitat for a safer vegetated refuge. The presence of both predators and chum increased refuge use by stickleback (Chapter 5). 7) The effects of predation risk on habitat use by chum depended on whether alternate prey were present. Chum avoided feeding in mid-water habitats in the pres-ence of a predator unless stickleback were present. A greater percentage of chum in the presence of a predator left the vegetated refuge to feed in the open water when stickleback were also present (Chapter 5). 8) The reduction in the fright response of chum in the presence of stickleback is partly attributable to differences between the two species in the perceived level of danger. Stickleback did not respond to predators in the same manner as chum, and consequently did not provide visual stimuli of predator detection or evasion recogniz-able to chum (Chapter 6). 121 Species Interactions The study of species interactions is essential in understanding factors shaping community structure, life history strategies and species co-existence. The distribution and abundance of a species is controlled by the physical environment and interactions with other species. Species interactions can be understood by observing patterns of behaviour and resource utilization. Interactions between predation and competition should be considered when studying prey response to predation. Predation can structure communities through selective predation on particular species or sizes of prey. Predation can reduce interspe-cific competition by lowering the demand for resources or by removing the dominant competitor. Predators can also affect diet and habitat use of prey through antipredator responses and thus affect intra- and interspecific interactions. Results from this study confirm that complex behavioural factors need to be considered when interpreting fish foraging behaviour. The Fraser River estuary is a dynamic system; the diversity and abundance of fish is continuously changing. Many species share common nearshore habitats and likely overlap in resource use. Juvenile fish inhabiting the estuary are exposed to a gauntlet of different types of predators, e.g. fish (both search and ambush predators), diving and wading birds, and mammals, and it is unlikely they are ever free from the threat of predation. Prey must continuously monitor for predators and must be sufficiently flexible to respond appropriately to the particular type of predator. As well, predators may develop feeding preferences for different types of prey or predators may learn to handle armoured prey, e.g. stickleback, and as a result the predation rate on chum may depend on other prey types are available. Hence, changes in the behaviour of one species to avoid predation may have potential impacts on all species that share common resources. 122 In laboratory experiments, behavioural responses by chum fry to avoid preda-tors resulted in reduced feeding efficiency. Chum avoided encounters with predators by altering their own niche use, activity levels, and foraging modes — the main factors that determine encounter rates with predators, other prey, and prey resources. Given species complexity in the wild, the numbers of possible interspecific interactions result-ing from niche shifts by prey to avoid predators are immense. Behavioural avoidance of predators in estuaries may severely reduce the habitat available for prey use, and competition for food may be intensified if competing prey must share a common habi-tat. However, chum may incur some advantage of the presence of other potential prey species in estuarine habitats. Chum fry were more willing to use habitats with preda-tors and the feeding efficiency of chum was unaffected by predators when stickleback were also present. Laboratory experiments indicated that risk of predation was dependent on the resource used. Chum fry and stickleback were more vulnerable to attack when feeding on prey near the substrate and in open water habitats. If risk varies with food type in the wild, risk will be an important determinant of diet choice, feeding rate, and habitat use. Risk of predation may prevent chum from feeding on optimal food types, may confine prey to vegetated habitats, and consequently may reduce feeding rates and prolong the time spent in vulnerable size classes. Evidence suggests that chum need to reach a certain size before migrating to sea and that larger fry have greater survival. Although laboratory experiments which expose prey to predators in close prox-imity on a continuous basis may exaggerate the relative importance of predators on prey behaviour, they are necessary to gain information on the behavioural responses of prey during predation. Future work in this system should examine the effects of sporadic contact of predators and competitors. 123 According to optimal foraging theories, natural selection favors animals that maximize their average rate of energy intake, or simply animals that forage efficiently. Antipredator responses may have important effects on prey foraging efficiency, and consequently growth, survival, and ultimately fitness. Since growth rates, competitive ability, and predation risk vary with prey size, competition and predation are linked through prey size. Because of constraints such as predation, animals rarely are able to maximize feeding rates in the wild. Therefore foraging prey should strike an adaptive balance between energy return and risk. As antipredator strategies usually involve trade-offs between energy intake, growth, survival, and reproduction, prey responses to predation should not be considered separately from other important aspects of prey life history. The indirect effects of predation presented here are due to behavioural responses by the prey. Indirect effects of predation take place much more rapidly than reduction in prey numbers and may be more significant than the actual losses due to predation. 124 Literature Cited Allan , J .D . 1981. Determinants of diet of brook trout (Salvelinus fontinalis) in a mountain stream. Canadian Journal of Fisheries and Aquatic Sciences 38: 184-192. Al lan , J .R., and T . J . Pitcher. 1986. Species segregation during predator evasion in cyprinid fish shoals. Freshwater Biology 16: 653-659. Armstrong, R . H . 1971. Age, food, and migration of sea-run cutthroat trout, Salmo clarki, at Eva Lake, Southeastern Alaska. Transactions of the American Fish-eries Society 100: 302-306. Baker, R .R. 1978. The evolutionary ecology of animal migration. Holmes and Meier Publishers, New York. Bams, R . A . 1969. Adaptations of sockeye salmon associated with incubation in stream channels. Pages 71-87. In T . G . Northcote (ed). Symposium on salmon and trout in streams. H.R. MacMillan Lectures in Fisheries, University of British Columbia, Vancouver. Barlow, G . W . 1968. Dither - A way to reduce undesirable fright behavior in ethological studies. Zeitschrift fur Tierpsychologie 25: 315-318. Barlow, G . W . 1977. Modal action patterns. Pages 98-134. In T . A . Sebeok (ed). How animals communicate. Indiana University Press, Bloomington, Indiana. Bell , G.R. 1964. A guide to the properties, characteristics, and uses of some general anaesthetics for fish. Bulletin of the Fisheries Research Board of Canada 148: 4p. Bertram, B . C . R . 1978. Living in groups: predators and prey. Pages 64-96. In J .R. Krebs, and N . B . Davies (ed). Behavioural ecology. Blackwell, Oxford. Brooks, J .L . 1968. The effects of prey size selection by lake planktivores. Systematic Zoology 17: 272-292. Brown, L . R . , and P .B . Moyle. 1981. The impact of squawfish on salmonid populations: a review. North American Journal of Fisheries Management 1: 104-111. Bryan, J .E . 1973. Feeding history, parental stock, and food selection in rainbow trout. Behaviour 45: 123-153. 125 Cerri, R .D. , and D.F . Fraser. 1983. Predation and risk in foraging minnows: balancing conflicting demands. The American Naturalist 121: 552-561. Chapman, D . W . 1966. Food and space as regulators of salmonid populations in streams. The American Naturalist 100: 345-357. Charnov, E .L . 1976. Optimal foraging: attack strategy of a mantid. The American Naturalist 110: 141-151. Charnov, E .L . , G . H . Orians, and K . Hyatt. 1976. Ecological implications of resource depression. The American Naturalist 110: 247-259. Clark, C.W., and D . A . Levy. 1988. Diel vertical migrations by juvenile sockeye salmon and the antipredation window. The American Naturalist 131: 271-290. Connell, J .H. 1975. Some mechanisms producing structure in natural communities. Pages 460-490. In M . L . Cody and J . M . Diamond (ed). Ecology and evolution of communities. Belknap Press, Cambridge, Massachusetts. Cooper, W . E . , and L .B. Crowder. 1979. Patterns of predation in simple and complex environments. Pages 257-267. In R . H . Stroud and H . Clepper (ed). Predator-prey systems in fisheries management. Sport Fishing Institute, Washington, D . C . Cummins, K . W . , and J .C. Wuycheck. 1971. Caloric equivalents for investigations in ecological energetics. Internationale Vereingung fur Theoretische und Ange-wandte Limnologie 18: 1-158. Curio, E . 1976. The ethology of predation. Zoophysiology and Ecology 7, Springer-Verlag, Berlin Heidelberg New York. 250 p. Davies, N.B. , and A . L . Houston. 1984. Territory economics. Pages 148-169. In J.R. Krebs, and N.B. Davies (ed). Behavioural ecology. Sinauer, Sunderland, Massachusetts. Dawkins, R., and J.R. Krebs. 1979. Arms races between and within species. Proceed-ings of the Royal Society of London B 205: 489-511. Dill, L . M . 1972. Visual mechanism determining flight distance in zebra danios (Brachy-danio Pisces). Nature 236: 30-32. Dill, L . M . 1974. The escape response of the zebra danio (Brachydanio rerio), II. The effect of experience. Animal Behaviour 27: 723-730. 126 Dill, L . M . 1983. Adaptive flexibility in the foraging behavior of fishes. Canadian Journal of Fisheries and Aquatic Sciences 40: 398-408. Dill, L . M . 1987. Animal decision making and its ecological consequences: the future of aquatic ecology and behaviour. Canadian Journal of Zoology 65: 803-811. Dill, L . M . , and A . H . G . Fraser. 1984. Risk of predation and the feeding behavior of juvenile coho salmon (Oncorhynchus kisutch). Behavioral Ecology and Sociobi-ology 16: 65-71. Dill, L . M . , R . C . Ydenberg, and A . H . G . Fraser. 1981. Food abundance and terri-tory size in juvenile coho salmon (Oncorhynchus kisutch). Canadian Journal of Zoology 59: 1801-1809. Drenner, R.W., J.R. Strickler, and W.J . O'Brien. 1978. Capture probability: the role of zooplankter escape in the selective feeding of planktivorous fish. Journal of the Fisheries Research Board of Canada 35: 1370-1373. Dunbrack, R.L. , and L . M . Dill. 1983. A model of size dependent surface feeding in a stream dwelling salmonid. Environmental Biology of Fishes 8: 203-216. Dunford, W . E . 1975. Space and food utilization by salmonids in marsh habitats of the Fraser River estuary. MSc. Thesis, University of British Columbia, Vancouver. 81 p. Dupuis, H . M . C . , and M . H . A . Keenleyside. 1982. Egg-care behaviour of Aequidens •paraquayensis (Pisces, Cichlidae) in relation to predation pressure and spawning substrate. Canadian Journal of Zoology 60: 1794-1800. Edmunds, M . 1974. Defense in animals. Longman, Essex. 357 p. Endler, J .A. 1986. Defense against predators. Pages 109-134. In M . E . Feder and G . V . Lauder (ed). Predator-prey relationships: perspectives and approaches from the study of lower vertebrates. University of Chicago Press, Chicago. Fishelson, L. 1977. Sociobiology of feeding behavior of coral fish along the coral reef of the Gulf of Elat (Gulf of Aqaba), Red Sea. Israel Journal of Zoology 26:114-134. Foerster, R .E . , and W . E . Ricker. 1942. The effect of the reduction of predaceous fish on survival of young sockeye salmon at Cultus Lake. Journal of the Fisheries Research Board of Canada 5: 315-336. 127 Forney, J .L. 1974. Interactions between yellow perch abundance, walleye predation, and survival of alternate prey in Oneida Lake, New York. Transactions of the American Fisheries Society 103: 15-24. Fraser, D.F . , and R.D. Cerri. 1982. Experimental evaluation of predator-prey rela-tionships in a patchy environment: consequences for habitat use patterns in stream minnows. Ecology 63: 307-313. Fresh, K . L . , R . D . Cardwell, B.P. Snyder, and E . O . Salo. 1980. Some hatchery strate-gies for reducing predation upon juvenile salmon (Oncorhynchus keta) in fresh-water. In B.R. Melteff and R.A. Neve (ed). Proceedings of the North Pacific Aquaculture Symposium, 1980. University of Alaska, Fairbanks. Giles, N . , and F . A . Huntingford. 1984. Predation risk and inter-population variation in antipredator behaviour in the three-spined stickleback Gasterosteus aculeatus L. Animal Behaviour 32: 264-275. Gilinsky, E . 1984. The role of fish predation and spatial heterogeneity in determining benthic community structure. Ecology 65: 455-468. Gilliam, J.F. 1982. Habitat use and competitive bottlenecks in size-structured fish populations. Ph.D. thesis, Michigan State University, East Lansing: Gilliam, J.F., and D.F . Fraser. 1987. Habitat selection when foraging under predation hazard: a model and a test with stream dwelling minnows. Ecology 68: 1856-1862. Glass, N.R. 1971. Computer analysis of predation energetics in the largemouth bass. Pages 325-363. In B .C . Patten (ed). Systems analysis and simulation in ecology, Volume 1. Academic Press, New York. Godin, J . -G.J . 1980. Temporal aspects of juvenile pink salmon (Oncorhynchus gor-buscha Walbaum) emergence from a simulated redd. Canadian Journal of Zo-ology 58: 735-744. Godin, J . -G.J . 1986. Antipredator function of shoaling in teleost fishes: a selective review. Le Naturaliste Canadien (Quebec) 113: 241-250. Godin, J . -G.J . 1990. Diet selection under the risk of predation. Pages 739-769. In R . N . Hughes (ed). Behavioural mechanisms of food selection. N A T O Advanced Science Institute Series Volume 20, Springer-Verlag, London Paris Tokyo Hong Kong. 128 Godin, J . -G.J . , and M . J . Morgan. 1985. Predator avoidance and school size in a cyprinodontid fish, the banded killifish (Fundulus diaphanus Lesueur). Behav-ioral Ecology and Sociobiology 16: 105-110. Gross, H . P . 1978. Natural selection by predators on the defensive apparatus of the three-spined stickleback, Gasierosteus aculeatus L . Canadian Journal of Zoology 56: 398-413. Hargreaves, N . B . , and R. J . LeBrasseur. 1986. Size selectivity of coho (Oncorhynchus kisutch) preying on juvenile chum salmon (0. keta). Canadian Journal of Fish-eries and Aquatic Sciences 43: 581-586. Harvey, P . H . , and P.J . Greenwood. 1978. Anti-predator defence strategies: some evolutionary problems. Pages 129-154. In J .R. Krebs, and N . B . Davies (ed). Behavioural ecology. Blackwell, Oxford. Healey, M . C . 1979. Detritus and juvenile salmon production in the Nanaimo River estuary: I. Production and feeding rates of juvenile chum salmon. Journal of the Fisheries Research Board of Canada 36: 488-496. Healey, M . C . 1980. Utilization of the Nanaimo River estuary by juvenile chinook salmon, Oncorhynchus tshawytscha. Fisheries Bulletin 77: 653-668. Healey, M . C . 1982a. Juvenile pacific salmon in estuaries: the life support system. Pages 315-341. In V .S . Kennedy (ed). Estuarine comparisons. Academic Press, New York. Healey, M . C . 1982b. Timing and relative intensity of size-selective mortality of juvenile chum salmon (Oncorhynchus keta) during early sea life. Canadian Journal of Fisheries and Aquatic Sciences 39: 952-957. Healey, M . C , R . V . Schmidt, F .P . Jordon, and R . M . Hungar. 1977. Juvenile salmon in the Nanaimo area. II. Length, weight, and growth. Canadian Manuscript Report Fisheries Marine Service 1438: 1-147. Helfman, G.S. 1986. Behavioural responses of prey fishes during predator-prey inter-actions. Pages 135-156. In M . E . Feder and G . V . Lauder (ed). Predator-prey relationships: perspectives and approaches from the study of lower vertebrates. University of Chicago Press, Chicago. Helfman, G.S. 1989. Threat-sensitive predator avoidance in damselfish-trumpetfish interactions. Behavioral Ecology and Sociobiology 24: 47-58. 129 Hennessy, D.F . , and D . H . Owings. 1978. Snake species discrimination and the role of olfactory cues in the snake-directed behaviour of the California ground squirrel. Behaviour 65: 115-124. Hiyama, Y. , Y . Nose, M . Shimizu, T . Ishihara, H . Abe, R. Sato, T . Maiwa, and T . Kajihara. 1972. Predation of chum salmon fry during the course of its seaward migration-II. Otsuchi River Investigation 1964 and 1965. Bulletin of the Japanese Society of Scientific Fisheries 38: 223-229. Hobson, E.S. 1968. Predatory behaviour of some shore fishes in the Gulf of California. U.S. Bureau of Sport Fisheries and Wildlife Research Report 73: 1-92. Hobson, E.S. 1979. Interactions between piscivorous fishes and their prey. Pages 231-242. In R . H . Stroud and H . Clepper (ed). Predator-prey systems in fisheries management. Sport Fishing Institute, Washington, D C . Hoogland, R., D. Morris, and N. Tinbergen. 1957. The spines of sticklebacks (Gas-terosteus and Pygosteus) as a means of defense against predators (Perca and Esox). Behaviour 10: 205-237. Hoos, L . M . , and G.L . Packman. 1974. The Fraser River Estuary status of environ-mental knowledge to 1974. Report of the estuary working group. Environment Canada, Special Estuarine Series No. 1: 518 p. Huang, C , and A. Sih. 1990. Experimental studies on behaviorally mediated, indirect interactions through a shared predator. Ecology 71: 1515-1522. Humphries, D . A . , and P . M . Driver. 1970. Protean defence by prey animals. Oecologia 5: 285-302. Hunter, J . G . 1959. Survival and reproduction of pink and chum salmon in a coastal stream. Journal of the Fisheries Research Board of Canada 16: 835-886. Irvine, J.R., and T . G . Northcote. 1983. Selection by young rainbow trout (Salvelinus gairdneri) in simulated stream environments for live and dead prey of different sizes. Canadian Journal of Fisheries and Aquatic Sciences 40: 1745-1749. Itzkowitz, M . 1974. A behavioural reconnaissance of some Jamaican reef fishes. Zoo-logical Journal of the Linnaean Society 55: 87-118. Jeffries, M.J . , and J .H . Lawton. 1984. Enemy free space and the structure of ecological communities. Biological Journal of the Linnean Society 23: 269-286. 130 Johnson, R. 1974. Effects of hatchery coho on native Puget Sound stocks of chum salmon fry. Pages 102-109. In D.R. Harding (ed). Proceedings of the 1974 Northeast Pacific pink and chum salmon workshop. Richmond, British Columbia. Johnsson, J .L, and M . V . Abrahams. 1991. Interbreeding with domestic strains increases foraging under threat of predation in juvenile steelhead trout (On-corhynchus mykiss): an experimental study. Canadian Journal of Fisheries and Aquatic Sciences 48: 243-247. Keenleyside, M . H . A . 1955. Some aspects of the schooling behaviour of fishes. Be-haviour 8: 183-248. Keenleyside, M . H . A . 1979. Diversity and adaptation in fish behaviour. Zoophysiology 11. Springer, Berlin Heidelberg New York. 208 p. Kobayashi, T . 1958. [An ecological study of the salmon fry, Oncorhynchus keta. [V.] The behaviour of chum salmon fry on their seaward migration.] Hokkaido Sake, Masu, Fukajo Kenkyu Hokoku 12: 2-30. (In Japanese, English abstract). Kobayashi, T . 1964. [An ecological study of the salmon fry, Oncorhynchus keta Walbaum. VII. Note on the behaviour of the fry during seaward migration.] Hokkaido Sake, Masu, Fukajo Kenkyu Hokoku 18: 1-6. (In Japanese, English abstract). Krebs, J.R. 1973. Social learning and the significance of mixed-species flocks of chick-adees (Parus sp.). Canadian Journal of Zoology 51: 1275-1288. Krebs, J.R., M . H . MacRoberts, and J . M . Cullen. 1972. Flocking and feeding in the great tit Parus major. An experimental study. Ibis 114: 507-530. Krebs, J.R., D . W . Stephens, and W.S. Sutherland. 1983. Perspectives in optimal foraging. Pages 165-216. In A . H . Brush and G . A . Clark (ed). Perspectives in ornithology. Cambridge University Press, London and New York. Kruskal, W . H . , and W . A . Wallis. 1952. Uses of ranks in one-criterion analysis of variance. Journal of the American Statistical Association 47: 583-621. Larkin, P.A. , and S.B. Smith. 1954. Some effects of the introduction of the redside shiner (Richardsonius balteatus) on the Kamloops trout (Salmo gairdneri kam-loops) in Paul Lake, British Columbia. Transactions of the American Fisheries Society 83: 161-175. 131 Laur, D.R., and A . W . Ebeling. 1983. Predator-prey relationships in surfperches. Environmental Biology of Fishes 8: 217-229. LeBrasseur, R.J. 1969. Growth of juvenile chum salmon (Oncorhynchus keta) under different feeding regimes. Journal of the Fisheries Research Board of Canada 26: 1631-1645. LeBrasseur, R.J. , and R.R. Parker. 1964. Growth rate of central British Columbia pink salmon (Oncorhynchus gorbuscha). Journal of the Fisheries Research Board of Canada 21: 1101-1128. Lendrem, D.W. 1984. Sleeping and vigilance in birds. II. An experimental study of the Barbary dove (Streptopelia risoria). Animal Behaviour 32: 243-248. Levy, D . A . 1977. The effects of experience on the acquisition of food by juvenile chum salmon, Oncorhynchus keta, in a tidal creek of the Squamish River estuary, British Columbia. MSc. thesis, University of British Columbia, Vancouver. 61p. Levy, D . A . 1987. Review of the ecological significance of diel vertical migrations by juvenile sockeye salmon (Oncorhynchus nerka). Pages 44-52. In H .D. Smith, L. Margolis, and C C . Wood (ed). Sockeye salmon (Oncorhynchus nerka) popula-tion biology and future management. Canadian Special Publication of Fisheries and Aquatic Sciences 96. Levy, D . A . , and C D . Levings. 1978. A description of the fish community of the Squamish River estuary, British Columbia: relative abundance, seasonal changes, and feeding habitats of salmonids. Canada Fisheries Marine Service Manuscript Report 1475: 63 p. Levy, D . A . , and T . G . Northcote. 1981. The distribution and abundance of juvenile salmonids in marsh habitats of the Fraser River estuary. University of British Columbia, Westwater Research Center Technical Report 25: 120 p. Levy, D . A . , and T . G . Northcote. 1982. Juvenile salmon residency in a marsh area of the Fraser River estuary. Canadian Journal of Fisheries and Aquatic Sciences 39: 270-276. Levy, D . A . , T . G . Northcote, and G.J. Birch. 1979. Juvenile salmonid, utilization of tidal channels in the Fraser River estuary, British Columbia. University of British Columbia, Westwater Research Center Technical Report 23: 70 p. Lima, S.L. 1985. Maximizing feeding efficiency and minimizing time exposed to preda-tors: a trade-off in the black-capped chickadee. Oecologia 66: 60-67. 132 Lima, S.L. 1988. Vigilance and diet selection: a simple example in the dark eyed junco. Canadian Journal of Zoology 66: 593-596. Lima, L .L . , and L . M . Dill. 1990. Behavioral decisions made under the risk of predation: a review and prospectus. Canadian Journal of Zoology 68: 619-640. Lima, S.L., T . J . Valone, and T . Caraco. 1985. Foraging efficiency-predation risk tradeoff in the grey squirrel. Animal Behaviour 33: 155-165. Lucas, J.R. 1987. Foraging time constraints and diet choice, p. 239-269. In A . C . Kamil, J.R. Krebs, and H.R. Pulliam (ed). Foraging behavior. Plenum Press, New York. Mace, P . M . 1983. Predator-prey functional responses and predation by staghorn sculpins (Leptocottus armatus) on chum salmon fry (Oncorhynchus keta). Ph.D. thesis, University of British Columbia, Vancouver. 526 p. Magnhagen, C. 1988. Predation risk and foraging in juvenile pink (Oncorhynchus gorbuschd) and chum salmon (0. keta). Canadian Journal of Fisheries and Aquatic Sciences 45: 592-596. Magurran, A . E . , W.J . Oulton, and T . J . Pitcher. 1985. Vigilant behaviour and shoal size in minnows. Zeitschrift fur Tierpsychologie 67: 167-178. Major, P.F. 1978. Predatory-prey interactions in two schooling fishes, Caranx ignobilis and Stolephorus purpureus. Animal Behaviour 26: 760-777. Mangel, M . , and C.W. Clark. 1986. Towards a unified foraging theory. Ecology 67: 1127-1138. Mann, H.B. , and D.R. Whitney. 1947. On a test of whether one or two random vari-ables is stochastically larger than the other. Annals of Mathematical Statistics 18: 50-60. Manzer, J.I. 1976. Distribution, food, and feeding of the threespine stickleback, Gas-terosteus aculeatus, in Great Central Lake, Vancouver Island, with comments on competition for food with juvenile sockeye salmon, Oncorhynchus nerka. Fish-ery Bulletin 74: 647-668. Mauck, W . L . , and'D.W. Coble. 1971. Vulnerability of some fishes to northern pike (Esox lucius) predation. Journal of the Fisheries Research Board of Canada 28: 957-969. 133 [ McCabe, G . T . Jr., W . D . Muir, R.L. Emmett, and J .T. Durkin. 1983. Interrelation-ships between juvenile salmonids and nonsalmonid fish in the Columbia River estuary. Fishery Bulletin 81: 815-826. McLean, E.B. , and J.-G.J. Godin. 1989. Distance to cover and fleeing from predators in fish with different amounts of defensive armour. Oikos 55: 281-290. McNamara, J . M . , and A.I. Houston. 1986. The common currency for behavioral decisions. The American Naturalist 127: 358-378. McPhail, J .D. 1969. Predation and the evolution of a stickleback. Journal of the Fisheries Research Board of Canada 26: 3183-3208. Metcalfe, N.B. , F . A . Huntingford, and J .E. Thorpe. 1987a. The influence of predation risk on the feeding motivation and foraging strategy of juvenile Atlantic salmon. Animal Behaviour 35: 901-911. Metcalfe, N.B. , F . A . Huntingford, and J .E. Thorpe. 1987b. Predation risk impairs diet selection in juvenile salmon. Animal Behaviour 35: 931-933. Milinski, M . 1977. Experiments on the selection by predators against spatial oddity of their prey. Zeitschrift fur Tierpsychologie 43: 311-325. Milinski, M . 1984. A predator's costs of overcoming the confusion effect of swarming prey. Animal Behaviour 32: 1157-1162. Milinski, M . 1986. Constraints placed by predators on feeding behaviour. Pages 236-252. In T . Pitcher (ed). The behaviour of teleost fishes. Croom Helm Limited, London. Milinski, M . , and R. Heller. 1978. Influence of a predator on the optimal forag-ing behaviour of sticklebacks (Gaaterosteus aculeatus L.). Nature (London) 275: 642-644. Milinski, M . , and C. Loewenstein. 1980. On predator selection against abnormalities of movement: a test of an hypothesis. Zeitschrift fur Tierpsychologie 53: 325-340. Minello, T . J . , and R.J. Zimmerman. 1983. Fish predation on juvenile brown shrimp, (Penaeus aztecus Ives): the effect of simulated Spartina structure on predation rates. Journal of Experimental Marine Biology and Ecology 72: 211-231. Mittlebach, G . G . 1984. Predation and resource partitioning in two sunfishes (Centrar-chidae). Ecology 65: 499-513. 134 Mittlebach, G . G . 1986. Predator-mediated habitat use: some consequences for species interactions. Environmental Biology of Fishes 16: 159-169. Mittlebach, G . G . , and P.L. Chesson. 1987. Predation risk: indirect effects of fish populations. Pages 315-332. In W . C . Kerfoot and A. Sih (ed). Predation: direct and indirect impacts on aquatic communities. University Press of New England, Hanover and London. Molenock, J . 1975. Evolutionary aspects of communication in the courtship behavior of four species of Anomuran crabs (Petrolisthes). Behaviour 53: 1-30. Molenock, J. 1976. Agonistic interactions of the crab Petrolisthes (Crustacea, Anomura). Zeitschrift fur Tierpsychologie 41: 277-294. Moodie, G . E . E . 1972. Morphology, life history and ecology of an unusual stickleback (Gasterosteus aculeatus) in the Queen Charlotte Islands, Canada. Canadian Journal of Zoology 50: 721-732. Moodie, G . E . E . , J .D. McPhail, and D.W. Hagen. 1973. Experimental demonstration of selective predation on Gasterosteus aculeatus. Behaviour 47: 95-105. Morisita, M . 1959. Measuring of interspecific association and similarity between com-munities. Memoirs of the Faculty of Science of Kyushu University Series E Biology 3: 65-80. Morse, D . H . 1980. Behavioral mechanisms in ecology. Harvard University Press, Cambridge. 383p. Moyle, P.B. 1977. In defense of sculpins. Fisheries 2: 20-23. Murphy, M . L . , J .F. Thedinga, and K . V . Koski. 1988. Size and diet of juvenile Pa-cific salmon during seaward migration through a small estuary in Southeastern Alaska. Fishery Bulletin 86: 213-222. Myers, K . W . W . 1980. A n investigation of the utilization of four study areas in Yaquina Bay, Oregon, by hatchery and wild juvenile salmonids. MSc. thesis, Oregon State University, Corvallis. 234 p. Myers, K . W . , and H.F . Horton. 1982. Temporal use of an Oregon estuary by hatchery and wild juvenile salmon. Pages 377-392. In V.S. Kennedy (ed). Estuarine comparisons. Academic Press, New York. Neave, F . 1955. Notes on the seaward migration of pink and chum salmon fry. Journal of the Fisheries Research Board of Canada 12: 369-374. 135 Neill, S.R.St.J. 1970. A study of antipredator adaptations in fish with special reference to silvery camouflage and shoaling. Ph.D. thesis, Oxford University. 159 p. Nilsson, N. -A. 1979. Food and habitat of the fish community of the region of Lake Vanern, Sweden. Report of the Institute of Freshwater Research, Drottningholm 58: 126-139. Nilsson, N. -A. , and T . G . Northcote. 1981. Rainbow trout (Salmo gairdneri) and cut-throat trout (5. clarki) interactions in coastal British Columbia lakes. Canadian Journal of Fisheries and Aquatic Sciences 38: 1228-1246. Northcote, T . G . , N .T . Johnston, and K. Tsmura. 1978. A regional comparison of species distribution, abundance, size and other characteristics of lower Fraser River fishes. University of British Columbia, Westwater Research Center Tech-nical Report 14: 38 p. Northcote, T . G . , N .T . Johnston, and K. Tsmura. 1979. Feeding relationships and food web structure of lower Fraser River fishes. University of British Columbia, Westwater Research Center Technical Report 16: 73 p. O'Brien, W.J . 1979. The predator prey interaction of planktivorous fish and zooplank-ton. American Scientist 67: 572-581. Obrebski, S., and J. Sibert. 1976. Diet overlaps in competing fish populations in the Nanaimo River estuary. Pages 139-146. In C A . Simenstad, and S.J. Lipovsky (ed). Fish food habits studies. 1st Pacific Northwest Technical Workshop. University of Washington, Seattle. Olla, B.L. , and C Samet. 1974. Fish-to-fish attraction and the facilitation of feed-ing behaviour as mediated by visual stimuli in striped mullet, Mugil cephalus. Journal of the Fisheries Research Board of Canada 31: 1621-1630. Paine, R . T . 1966. Food web complexity and species diversity. American Naturalist 100: 65-75. Parker, R.R. 1971. Size selective predation among juvenile salmonid fishes in a British Columbia inlet. Journal of Fisheries Research Board of Canada 28: 1503-1510. Paszkowski, C A . , and B.L. Olla. 1985. Foraging behavior of hatchery-produced coho salmon (Oncorhynchus kisutch) smolts on live prey. Canadian Journal of Fish-eries and Aquatic Sciences 42: 1915-1921. 136 Patten, B . G . 1971a. Predation on fall chinook salmon, (Oncorhynchus ishawytscha) fry of hatchery origin. U.S. National Marine Fisheries Service Special Scientific Report 621: 14p. Patten, B . G . 1971b. Increased predation by the torrent sculpin, Coitus rhotheus, on coho salmon fry, Oncorhynchus kisutch, during moonlight nights. Journal of the Fisheries Research Board of Canada 28: 1352-1354. Patten, B . G . 1975. Comparative vulnerability of Pacific salmon and steelhead trout to predation by torrent sculpin in stream aquaria. Fishery Bulletin 73: 931-934. Pearcy, W . G . , C D . Wilson, A . W . Chung, and J.W. Chapman. 1989. Residence times, distribution, and production of juvenile chum salmon, Oncorhynchus keta, in Netarts Bay, Oregon. Fishery Bulletin 87: 553-568. Peckarsky, B .L . 1980. Predator-prey interactions between stoneflies and mayflies: be-havioral observations. Ecology 61: 932-943. Pfeiffer, W. 1962. The fright reaction offish. Biological Review 37: 495-511. Pitcher, T . J . 1980. Some ecological consequences of fish school volumes. Freshwater Biology 10: 539-544. Pitcher, T . J . 1986. Functions of shoaling behaviour in teleosts. Pages 294-337. In T . J . Pitcher (ed). The behaviour of teleost fishes. Croom Helm Limited, London. Powell, G . V . N . 1974. Experimental analysis of the social value of flocking by starlings (Sturnu3 vulgaris) in relation to predation and foraging. Animal Behaviour 23: 504-508. Power, M . E . 1984. Depth distributions of armored catfish: predator-induced resource abundance. Ecology 65: 523-528. Pritchard, A . L . 1936. Stomach content analysis of fishes preying upon the young of Pacific Salmon during the fry migration at McClinton Creek, Massett Inlet, British Columbia. Canadian Field. Naturalist 1: 104-105. Pyke, G . H . , H.R. Pulliam, and E .L . Charnov. 1977. Optimal foraging: a selective review of theory and tests. The Quarterly Review of Biology 52: 137-154. Reimers, P.E. 1971. The length of residence of juvenile fall chinook salmon in Sixes River, Oregon. Ph.D. thesis, Oregon State University, Corvallis. 99 p. 137 Reist, J .D . 1980. Predation upon pelvic phenotypes of brook stickleback, Culaea inconstant, by selected invertebrates. Canadian Journal of Zoology 58: 1253-1258. Ricker, W . E . 1941. The consumption of young sockeye salmon by predaceous fish. Journal of the Fisheries Research Board of Canada 5: 293-313. Ritter, J .A . , and H.R. MacCrimmon. 1973. Influence of environmental experience on response of yearling rainbow trout (Salmo gairdneri) to a black and white substrate. Journal of the Fisheries Research Board of Canada 30: 1740-1742. Ruppell , G . , and E . Gosswein. 1972. Die Schwarme von Leucaspius delineatus (Cyprinidae, Teleostei) bei Gehahr in Hellen und im Dunkeln. Zeitschrift fur Vergleichende Physiologie 76: 333-340. Sambrook, R. 1990. Interactions between threespine stickleback (Gasterosteus aculea-tus) and juvenile chinook salmon (Oncorhynchus tshawytscha) in an estuarine marsh. M.Sc. thesis, University of British Columbia, Vancouver. 88 p. Savino, J .F. , and R . A . Stein. 1982. Predator-prey interaction between largemouth bass and bluegills by simulated, submerged vegetation. Transactions of the American Fisheries Society 111: 255-266. Savino, J .F. , and R . A . Stein. 1989. Behavioral interactions between fish predators and their prey: effects of plant density. Animal Behaviour 37: 311-321. Schmitt, R . J . 1982. Consequences of dissimilar defenses against predation in a subtidal marine community. Ecology 63: 1588-1601. Schoener, T . W . 1971. Theory of feeding strategies. Annual Review of Ecology and Systematics 2: 369-404. Seghers, B . H . 1974. Schooling behaviour in the guppy (Poecilia reticulata): an evolu-tionary response to predation. Evolution 28: 486-489. Sih, A . 1979. Stability and prey behavioural responses to predator density. Journal of Animal Ecology 48: 79-89. Sih, A . 1980. Optimal behavior: can foragers balance two conflicting demands? Sci-ence (Washington, D.C.) 210: 1041-1043. Sih, A . 1981. Stability, prey density and age dependent interference in an aquatic insect predator, Notonecta hoffmanni. Journal of Animal Ecology 50: 625-636. 138 Sih, A. 1982. Foraging strategies and the avoidance of predation by an aquatic insect Notonecta hoffmanni. Ecology 63: 786-796. Sih, A . 1984. The behavioral response race between predator and prey. The American Naturalist 123: 143-150. Sih, A . 1986. Antipredator responses and the perception of danger by mosquito larvae. Ecology 67: 434-441. Sih, A. 1987. Predators and prey lifestyles: an evolutionary and ecological overview. Pages 203-224. In W . C . Kerfoot and A. Sih (ed). Predation: direct and indirect impacts on aquatic communities. University Press of New England, Hanover and London. Simenstad, C . A . , and E . O . Salo. 1980. Foraging success as a determinant of estuarine and nearshore carrying capacity of juvenile chum salmon (Oncorhynchus keta) in Hood Canal, Washington. Pages 21-37. In B.R. Melteff and R .A. Neve (ed). Proceedings of the North Pacific Aquaculture Symposium, 1980. University of Alaska, Fairbanks. Simenstad, C . A . , K . L . Fresh, and E . O . Salo. 1982. The role of Puget Sound and Washington coastal estuaries in the life history of Pacific salmon: an unappre-ciated function. Pages 343-364. In V.S. Kennedy (ed). Estuarine comparisons. Academic Press. Simenstad, C . A . , W.J . Kinney, S.S. Parker, E . O . Salo, J.R. Cordell, and H . Buechner. 1980. Prey community structure and trophic ecology of outmigrating juvenile chum and pink salmon in Hood Canal, Washington: a synthesis of three years' studies, 1977-1979. Final Report, Fisheries Research Institute, University of Washington, Seattle, WA. FRI-UW-8026 113p. Slaney, T . L . , J .D. McPhail, D . Radford, and G.J. Birch. 1985. Review of the effects of enhancement strategies on interactions among juvenile salmonids. Canadian Manuscript Report of Fisheries and Aquatic Sciences 1852: 72 p. Smith, E.P., and T . M . Zaret. 1982. Bias in estimating niche overlap. Ecology 63:1248-1253. Sosiak, A.J . , R . G . Randall, J .A. McKenzie. 1979. Feeding by hatchery-reared and wild Atlantic salmon (Salrno salar) parr in streams. Journal of the Fisheries Research Board of Canada 36: 1408-1412. Stein, R . A . 1977. Selective predation, optimal foraging, and the predator-prey inter-action between fish and crayfish. Ecology 58: 1237-1253. 139 Stein, R . A . 1979. Behavioral response of prey to fish predators. Pages 343-353. In R . H . Stroud and H . Clepper (ed). Predator-prey systems in fisheries manage-ment. Sport Fishing Institute, Washington, D C . Stein, R . A . , and J.J. Magnuson. 1976. Behavioral response of crayfish to a fish predator. Ecology 57: 751-761. Stephens, D . W . , and J.R. Krebs. 1986. Foraging theory. Princeton University Press, Princeton, N J . 247 p. Suboski, M . D . 1988. Acquisition and social communication of stimulus recognition by fish. Behavioural Processes 16: 213-244. Suboski, M . D . , and J.J . Templeton. 1989. Life skills training for hatchery fish: social learning and survival. Fisheries Research 7: 343-352. Sumner, F . H . 1972. A contribution to the life history of the cutthroat trout in Oregon (with an emphasis on the coastal sub-species Salmo clarki clarki Richardson). Oregon State Game Commission Manuscript 180 p. Thompson, R . B . 1959. Food of the squawfish Ptychocheilus oregonensis (Richardson) of the lower Columbia River. Fisheries Bulletin 158: 43-58. Thomson, R . E . 1981. Oceanography of the British Columbia coast. Canadian Special Publication of Fisheries and Aquatic Sciences 56: 291 p. Thorman, S. 1983. Food and habitat resource partitioning between three estuarine fish species on the Swedish west coast. Estuarine, Coastal and Shelf Science 17: 681-692. Van Havre, N . , and G. J . FitzGerald. 1988. Shoaling and kin recognition in the threespine stickleback (Gasterosteu3 aculeatus L . ) . Behavioural Biology 13:190-201. Vinyard, G . L . 1980. Differential prey vulnerability and predator selectivity: effects of evasive prey on bluegill (Lepomis macrochirus) and pumpkinseed (L. gibbo3U3). Canadian Journal of Fisheries and Aquatic Sciences 37: 2294-2299. Walsh, G . , and G. J . FitzGerald. 1984. Resource utilization and coexistence of three species of sticklebacks (Gasterosteidae) in tidal salt-marsh pools. Journal of Fish Biology 25: 405-420. Ware, D . M . 1973. Risk of epibenthic prey to predation by rainbow trout (Salmo gairdneri). Journal of the Fisheries Research Board of Canada 30: 787-797. 140 Ware, D . M . 1975a. Relation between egg size, growth, and natural mortality of larval fish. Journal of the Fisheries Research Board of Canada 32: 2503-2512. Ware, D . M . 1975b. Growth, metabolism, and optimal swimming speed of a pelagic fish. Journal of the Fisheries Research Board of Canada 32: 33-41. Werner, E . E . 1985. The mechanisms of species interactions and community organi-zation in fish. Pages 360-382. In D.R. Strong, D . Simberloff, L . G . Abele, and A . B . Thistle (ed). Ecological communities: conceptual issues and the evidence. Princeton University Press, Princeton, N . J . Werner, E . E . 1986. Species interactions in freshwater fish communities. Pages 344-358. In J . M . Diamond and T. Case (ed). Ecological communities. Harper and Row, Publishers, New York. Werner, E . E . , and J .F . Gill iam. 1984. The ontogenetic niche and species interac-tions in size-structured populations. Annual Review of Ecology and Systematics 15: 393-425. Werner, E . E . , and D. J . Hal l . 1988. Ontogenetic habitat shifts in bluegill: the foraging rate-predation risk trade-off. Ecology 69: 1352-1366. Werner, E .E . , D . J . Hal l , D .R . Laughlin, D. J . Wagner, L . A . Wilsmann, and F . C . Funk. 1977. Habitat partitioning in a freshwater fish community. Journal of the Fisheries Research Board of Canada 34: 360-370. Werner, E . E . , G . G . Mittlebach, and J .D. Hall . 1981. The role of foraging profitability and experience in habitat use by the bluegill sunfish. Ecology 62: 116-125. Werner, E . E . , J .F . Gil l iam, D. J . Hal l , and G . G . Mittlebach. 1983a. A n experimental test of the effects of predation risk on habitat use in fish. Ecology 64:1540-1548. Werner, E . E . , G . G . Mittlebach, D. J . Hal l , and J .F. Gill iam. 1983b. Experimental tests of optimal habitat use in fish: the role of relative habitat profitability. Ecology 64: 1525-1539. Wiens, J . A . 1977. On competition and variable environments. American Scientist 65: 590-597. Windell , J .T., and S.H. Bowen. 1978. Methods for study of fish diets based on analysis of stomach contents. Pages 219-226. In T . B . Bagenal (ed). Methods for assessment offish production in fresh waters. I.B.P. Handbook No. 3, Blackwell Scientific Publications, Oxford and Edinburgh. 141 Wootton, R . J . 1984. A functional biology of sticklebacks. University of California Press, Berkeley. 265 p. Worgan, J .P. , and G. J . FitzGerald. 1981. Diel activity and diet of three sympatric sticklebacks in tidal salt marsh pools. Canadian Journal of Zoology 59: 2375-2379. Ydenberg, R . C , and L . M . Di l l . 1986. The economics of fleeing from predators. A d -vances in the Study of Behaviour 16: 229-249. Zar, J . H . 1984. Biostatistical analysis. Prentice-Hall, New Jersey. 718 p. Zaret, T . M . 1980. Predation and freshwater communities. Yale University Press, New Haven and London. 187 p. 142 

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.831.1-0101144/manifest

Comment

Related Items