FORAGING BEHAVIOUR OF COMMON MERGANSERS (Mergus merganser) DISPERSION IN RELATION TO THE AVAILABILITY OF JUVENILE PACIFIC SALMON B.Sc.(Hons.), Simon Fraser University, 1977 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Zoology) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA AND THEIR by Christopher Charles Wood March 1984 (c) Christopher Charles Wood, 1984 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f the requirements f o r an advanced degree a t the U n i v e r s i t y o f B r i t i s h Columbia, I agree t h a t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and study. I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e copying o f t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the head o f my department o r by h i s o r her r e p r e s e n t a t i v e s . I t i s understood t h a t copying o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be allowed without my w r i t t e n p e r m i s s i o n . Department o f " Z - ^ r o ( ^ p 7 The U n i v e r s i t y o f B r i t i s h Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 E-6 (3/81) i i ABSTRACT Common mergansers (Mergus merganser) can i n f l i c t serious mortality on salmonid populations when abundant on salmon and trout-rearing waters. I examined the foraging behaviour and dispersion of mergansers to determine when, and where, aggregations of these ducks are li k e l y to occur on salmonid streams on eastern Vancouver Island, B.C. F i r s t , I investigated the effect of f i s h availability on a merganser's hunting performance and i t s choice of feeding site by stocking enclosed sections of a natural stream with various densities of juvenile coho salmon (Oncorhynchus kisutch). Coho smolt with access to bank cover or with previous exposure to mergansers were less vulnerable to predation. Merganser foraging success was not significantly affected by flock size for flocks <^ 25. A type II functional response was observed under a l l experimental conditions. Although 40 g coho smolt were selected over 2 g coho fry, the discrepancy in capture frequency can be explained by differences in conspicuousness due to size without inferring preference. A merganser's daily food requirement (approx. 400 g) can be satisfied at smolt densities of 2 - 30/100 m^ depending on the availability of cover and previous exposure to attack. Mergansers spent more time searching in the more profitable enclosures. The frequency of visits to the enclosure site was influenced by previous foraging success and the size of flocks already present. The probability of departure from the site was generally independent of flock size but decreased with increasing pr o f i t a b i l i t y . i i i Second, I examined changes in the abundance of mergansers on 5 neighbouring streams during the spring and summer months of 1980 - 1982. Mergansers congregated on hatchery streams following fish releases and reciprocal trends in abundance were evident among streams. At least 9 of 13 mergansers resighted after being marked and released on the Big Qualicum R., visited other nearby streams. Overall, merganser abundance declined steadily from March through June but increased following recruitment of juvenile birds. Flock-size distributions predicted by an equilibrium arrival-departure model, were consistent with those observed during May to mid - June, but not those during late - June. I used a similar model incorporating observed relationships between fis h availability and frequencies of arrival and departure, to predict aggregation patterns on hatchery streams; i t , too, predicted trends more successfully during March through May than in later months. Social interactions appeared to influence dispersion to a greater extent during late - June to August so that assumptions of the model were violated. Merganser fli g h t activity also declined from May to September. I suggest that the nesting dispersion of mergansers is also influenced by the size of juvenile salmonid migrations early in the spring, because breeding pairs congregate at profitable feeding sites. The number of merganser broods reared on 8 coastal streams was positively correlated (r^=.90) with drainage area and total juvenile salmonid migration including production from hatcheries and spawning channels. Other data on breeding pair densities, time of brood emergence and survivorship of merganser ducklings support the 'committed aggregation' hypothesis, but the evidence is not conclusive. If true, however, breeding pairs are 'deceived' about i v the natural productivity of enhanced streams; because hatchery fish are unavailable to merganser broods, the intensity of predation on wild salmonids may be unusually severe. Mortality due to mergansers feeding on juvenile Pacific salmon migrating downstream in the Big Qualicum and L i t t l e Qualicum rivers was probably < 8%. Despite the aggregative response by mergansers, mortality was depensatory because individual fish were at risk for very short duration. Merganser predation w i l l be minimized i f releases are few, but large, and scheduled as late as possible in the spring or early summer. Mortality of stream-resident salmonids due to merganser broods probably exceeded 20% in the Big Qualicum R. However, i t is not clear whether this mortality, which occurs before autumn freshet mortality, limits wild smolt production. V TABLE OF CONTENTS ABSTRACT i i LIST OF TABLES x i LIST OF FIGURES x i i ACKNOWLEDGEMENTS xv Chapter 1 : GENERAL INTRODUCTION Background to the Problem 1 The common merganser - d i s t r i b u t i o n and l i f e h i s t o r y 2 Assessing predation by mergansers . 3 PART I : FORAGING BEHAVIOUR 9 Chapter 2 : CONSTRAINTS ON FEEDING SUCCESS In t r o d u c t i o n 10 V u l n e r a b i l i t y and the f u n c t i o n a l response to prey d e n s i t y 10 The f u n c t i o n a l response to predator d e n s i t y 12 Prey s e l e c t i o n 13 Methods 15 Study Area 15 Enclosure design 15 Experimental design 22 Observational procedure 27 v i Other observations 28 S t a t i s t i c a l analysis 28 Results 30 No Alte r n a t i v e Prey 30 Rate of e f f e c t i v e search - coho smolt 30 Rate of e f f e c t i v e search - coho f r y 31 Derivation of the functional response 42 Alt e r n a t i v e Prey Available 51 Size s e l e c t i o n - fry vs smolt 51 An apparent size hypothesis 56 E f f e c t of Flock Size 60 The Feeding Chain 67 Search 67 Pursuit and capture 75 Discussion 80 Factors influencing hunting performance 80 Prey s e l e c t i o n and appetite 84 Chapter 3 : PROFITABILITY AND PATCH CHOICE Introduction 86 Methods 90 Study s i t e and observation procedure 90 Interpretation of movements on enclosures 90 Decoy experiment 91 Results 9 3 Patch Assessment 93 Search pattern within an enclosure 93 v i i Search pattern among enclosures 93 A r r i v a l s at the Enclosure Site 106 E f f e c t of previous experience 106 E f f e c t of fl o c k s i z e 109 Departures from the Enclosure Site 112 E f f e c t of flock size 112 Ef f e c t of p r o f i t a b i l i t y 113 Discussion 122 Patch assessment 122 A r r i v a l and departure 125 PART II : DISPERSION IN RELATION TO THE AVAILABILITY OF JUVENILE SALMONIDS 128 Chapter 4 : THE AGGREGATIVE RESPONSE Introduction 129 Methods 133 Study area 133 Census me thod 133 Marking method 136 Index of juvenile salmonid density (JSl) 136 Floc k - s i z e d i s t r i b u t i o n s 138 Results 140 Trends in Dispersion Unrelated to Salmonid Density 140 Population density and composition 140 v i i i Ranging behaviour 143 Movements of marked mergansers 148 Dispersion in Relation to Salmonid Density 155 Reciprocal movements between r i v e r systems 155 Reciprocal movements between fresh and t i d a l waters 155 Pr e d i c t i n g flock size . 160 Predict i n g aggregations 170 Discussion 181 Evaluating the Assumptions 181 Conclusions 188 Chapter 5 : NESTING DISPERSION AND THE AGGREGATIVE RESPONSE Introduction 191 Methods 194 Study area 194 Breeding pair census " 194 Brood census 194 Estimates of brood production 195 Results 197 R e l i a b i l i t y of the Brood Production Estimates 197 Census coverage 197 Census frequency 198 Factors Limiting Brood Production 203 Hypothesis 1 - Breeding population l i m i t s brood density below carrying-capacity of streams 203 Hypothesis 2 - Brood carrying-capacity i s determined by the size of streams 203 Hypothesis 3 - Brood carrying-capacity i s related to salmonid production 213 Discussion 225 PART III : MERGANSER PREDATION AND SALMONID ENHANCEMENT 233 Chapter 6 : MERGANSER PREDATION ON JUVENILE SALMONIDS DURING THEIR SEAWARD. MIGRATION Introduction 234 Methods 236 Study area 236 Computations for predation 236 Results 239 Dispersion of foraging a c t i v i t y 239 Estimates of mortality 244 Ef f e c t of size and spacing of f i s h releases 247 Discussion 255 Chapter 7 : PREDATION BY MERGANSER BROODS ON STREAM-RESIDENT JUVENILE SALMON Introduction 261 Methods 264 Study area 264 Computations for predation by juvenile mergansers 264 X Results 267 Dispersion of foraging a c t i v i t y 267 Diet 276 Daily consumption 277 Estimates of mo r t a l i t y 277 Discussion 284 Chapter 8 : CONCLUDING REMARKS 291 LITERATURE CITED 295 LIST OF TABLES x i Table I. Target densities for coho smolt and f r y stocked during enclosure experiments at Rosewall Creek . 25 Table I I . Estimated rates of e f f e c t i v e search for coho smolt and f r y with respect to enclosure, previous exposure to attack and l e v e l of s a t i a t i o n 32 Table I I I . Duration and success rate of prey handling a c t i v i t e s for coho smolt and f r y 33 Table IV. Summary of aggressive in t e r a c t i o n s with respect to flock size 114 Table V. Comparison of v i s i t duration when s o l i t a r y and i n f l o c k with respect to p r o f i t a b i l i t y 115 Table VI. Summary of s t a t i s t i c s pertaining to the size and juvenile salmonid production of r i v e r s censused 134 Table VII. Summary of resightings of banded mergansers 153 Table VIII. Summary of merganser brood production and survivorship on the major streams censused 221 Table IX. Estimates of mortality by species due to adult mergansers during downstream migration i n the Big Qualicum R 247 Table X. Estimates of predation of coho parr by merganser broods on freshwater 283 x i i LIST OF FIGURES F i g . 1. Organization of thesis; the conceptual framework to assess predation by mergansers 6 F i g . 2. Map showing study areas 16 F i g . 3. Detailed maps of p r i n c i p a l study areas 18 F i g . 4. Layout of stream enclosures in Rosewall Creek 23 F i g . 5. P a r t i a l functional responses (excluding digestive pause) of mergansers to coho smolt density with respect to amount of cover available to smolt and t h e i r previous exposure to mergansers 34 F i g . 6. P a r t i a l functional responses (excluding digestive pause) of mergansers to coho smolt density with respect to v u l n e r a b i l i t y of prey and cumulative consumption by mergansers 36 F i g . 7. P a r t i a l functional responses with respect to search i n t e n s i t y ... 38 F i g . 8. P a r t i a l functional response to coho fr y in upper enclosure 40 F i g . 9. Relationship between apparent digestion time per coho smolt and time since a r r i v a l at enclosure s i t e 43 F i g . 10. Pattern of f i s h consumption by i n d i v i d u a l mergansers observed for > 4 h 47 F i g . 11. Foraging time (including digestive pauses) required to s a t i s f y d a i l y requirements (0.4 kg) with respect to f i s h density 49 F i g . 12. Capture rates of hungry mergansers feeding on mixtures of coho f r y and smolt 52 F i g . 13. Size s e l e c t i o n by mergansers foraging for coho smolt and f r y .... 54 x i i i F i g . 14. P a r t i a l functional responses to coho smolt with respect to number of mergansers foraging on enclosures 61 F i g . 15. Individual capture rates vs number of mergansers foraging for steelhead smolt at the mouth of Rosewall Creek 63 F i g . 16. Individual capture rates vs number of downy young mergansers in broods foraging for young-of-year sculpins at mouth of the L i t t l e Qualicum R 65 F i g . 17. The feeding chain 68 F i g . 18. Search i n t e n s i t y vs time since l a s t pursuit 70 F i g . 19. Relationships between search i n t e n s i t y and degree of s a t i a t i o n with respect to prey size and density 72 F i g . 20. Frequency of pursuit vs l e v e l of s a t i a t i o n and coho smolt density in lower enclosure 76 F i g . 21. Pursuit success with respect to cover available to coho smolt and previous exposure to merganser attack 78 F i g . 22. A conceptual model for patch choice 88 F i g . 23. Time spent foraging within quadrants of enclosure vs coho smolt density 94 F i g . 24. D i s t r i b u t i o n of Giving-Up-Times (GUT) with respect to coho smolt density and corresponding random (exponential) d i s t r i b u t i o n 97 F i g . 25. Comparison of duration of foraging bouts on lower, middle and upper enclosures 100 F i g . 26. Trend in choice of enclosure on which foraging bouts are i n i t i a t e d 102 x i v F i g . 27. D i s t r i b u t i o n s of GUT on upper enclosure with respect to experience 104 F i g . 28. Trends in a r r i v a l rate and abundance during enclosure experiments at Rosewall Creek 107 F i g . 29. Proportion of f l i g h t s overhead that landed at the study plot with respect to the size of an a r t i f i c i a l f lock already present. 110 F i g . 30. The r e l a t i o n s h i p between duration of v i s i t s to the enclosure s i t e and coho smolt density 117 F i g . 31. Relationship between the duration of v i s i t s to the enclosure s i t e and foraging success 119 F i g . 32. Trends in abundance of mergansers on 3 adjacent r i v e r systems ( t i d a l and freshwater zones) 141 F i g . 33. Seasonal trends in abundance, p a i r i n g , sex r a t i o and brood emergence of mergansers during 1981 144 F i g . 34. Seasonal trend in f l i g h t a c t i v i t y of mergansers in 1981 146 F i g . 35. Foraging a c t i v i t y i n r e l a t i o n to time of day and tide height at Rosewall Creek and Big Qualicum R 149 F i g . 36. F l i g h t a c t i v i t y in r e l a t i o n to time of day and tide height at Rosewall Creek and Big Qualicum R 151 F i g . 37. Reciprocal r e l a t i o n s h i p s in merganser abundance on t i d a l and freshwater zones of 3 neighbouring systems with respect to ju v e n i l e salmonid migration from Big Qualicum R 156 F i g . 38. Relationships between the proportion of mergansers on freshwater and j u v e n i l e salmonid migration by month, March - June 158 XV F i g . 39. Comparison of observed and expected d i s t r i b u t i o n s of equilibrium flock sizes with respect to coho smolt density stocked i n enclosures at Rosewall Creek 161 F i g . 40. Comparison of observed and expected d i s t r i b u t i o n s of equilibrium flock sizes at the mouth of the L i t t l e Qalicum R ... 164 F i g . 41. Comparison of merganser flocks on fresh and t i d a l waters 167 F i g . 42. Seasonal trend in abundance of mergansers in units of the equilibrium aggregation model 171 F i g . 43. Comparison of observed and predicted trends in abundance of mergansers at Rosewall Creek 177 F i g . 44. Comparison of observed and predicted trends in abundance of mergansers at Big Qualicum R 179 F i g . 45. S e n s i t i v i t y of the aggregation model to v a r i a t i o n s in parameter values with respect to predictions at Rosewall Creek, 1980 186 F i g . 46. Relationship between census coverage and estimated brood production 199 F i g . 47. Relationship between census frequency and estimated brood production 201 F i g . 48. D i s t r i b u t i o n of egg-laying/incubation a c t i v i t y back-calculated from hatching dates 204 F i g . 49. Comparison of breeding-pair density before and during peak egg-laying period 206 F i g . 50. Comparison of breeding-pair density on freshwater reaches of neighbouring streams during peak egg-laying period 209 F i g . 51. Dispersion of breeding-pairs during peak egg-laying period 211 F i g . 52. Relationship between merganser brood density, drainage area of r i v e r system and juvenile salmonid production -. 215 x v i F i g . 53. Comparison of hatching dates on neighbouring streams 219 F i g . 54. Survivorship of merganser downy young during 1980 and 1981 222 F i g . 55. The recruitment hypothesis 227 F i g . 56. Dispersion of mergansers on the Big Qualicum R 240 F i g . 57. The proportion of mergansers (of river-system count) foraging on the Big Qualicum estuary with respect to time of day and tide height 242 F i g . 58. Relationships between % mortality and salmonid density over periods when a single species i s very abundant r e l a t i v e to al t e r n a t i v e prey 245 F i g . 59. Graphic representation of s i z e - s e l e c t i o n assumptions for computations i n Table X 249 F i g . 60. Relationships between mortality expected due to merganser aggregation and the size and spacing of releases of hatchery f i s h 253 F i g . 61. Biomass and dispersion of merganser broods (< 50 d) observed on 3 neighbouring streams 268 F i g . 62. Age d i s t r i b u t i o n s of merganser brood biomass ( i . e . appetite) observed on freshwater 272 F i g . 63. Estimated biomass of merganser broods (<_ 50 d) on freshwater reaches of 3 neighbouring streams 274 F i g . 64. Foraging a c t i v i t y of merganser broods on the L i t t l e Qualicum estuary with respect to time of day and age of brood 278 F i g . 65. Daily consumption by merganser ducklings with respect to age ... 280 x v i i ACKNOWLEDGEMENTS I wish to thank both my supervisor, C.J. Walters, and CS. Rolling for introducing me to the topic of this study. Dr. Walters gave inspiration, valuable advice and generous financial support from his NSERC operating grants. I am grateful to M.C. Healey and R.J. Beamish for services and equipment provided by the Pacific Biological Station and, especially, I thank Dr. Healey for his advice and encouragement. I have also benefitted from comments by J.N.M. Smith, H.R. Pulliam, D.L. Kramer, T.G. Northcote, R. Hilborn and J.D. McPhail, and from discussions with C.L. Gass, P.M. Mace and M. Jones. Editorial comments by Dr. Smith were particularly helpful. I am grateful to staff at the Rosewall Creek and Big Qualicum hatcheries for their assistance and to the Big Qualicum Indian Band for permission to work within their reserve. Dr. G. Kaiser and Wolfgang Carolsfield provided advice and equipment for banding mergansers; Dave Zi t t i n and Marc Hamer assisted with computer programming. Meredith Brown and Maureen Palmer typed much of the manuscript. I received financial support through NSERC postgraduate scholarships and a UBC fellowship. Finally, I wish to express appreciation to fellow graduate students, especially Jacky Booth, Don Furnell, Tom Kessler, Gary Kingston, David Marmorek and Ross Tallman, and to my parents, George and Phyllis Wood, from whom I have received help and encouragement. Above a l l , I am indebted to Claudia Hand for her love, drafting s k i l l s and competent f i e l d assistance. 1 CHAPTER 1; GENERAL INTRODUCTION Background to the Problem: The common merganser (Mergus merganser) is widely recognized as a potential threat to salmon and trout populations because i t is large (by avian standards), feeds almost exclusively on fis h , and occurs throughout the temperate latitudes of the northern hemisphere. Most early studies have indicated that mergansers can impose serious mortality on populations of trout and juvenile salmon wherever they are concentrated on wintering grounds (Leonard and Shetter 1936, Beach 1937, Salyer and Lagler 1940) and nesting areas (White 1936,1937). Subsequent research, involving experimental removal of mergansers, has shown that mergansers can limit salmonid production in at least some situations (Huntsman 1941, Elson 1962, Shetter and Alexander 1970). Aside from surveys by Munro and Clemens (1936, 1937), the potential for predation by mergansers on juvenile Pacific salmon has received l i t t l e attention. This i s surprising since i t has often been suggested that freshwater predators limit the survival of Pacific salmon (e.g. Neave 1953, Meacham and Clark 1979). Predation may also maintain dominance among Pacific salmon runs by causing greater (percentage) mortality in small populations (Ricker 1962, Ward and Larkin 1964, Larkin 1971). In light of recent investment in large f a c i l i t i e s to increase the freshwater production of juvenile salmonids, i t has become s t i l l more important to assess the impact of mortality due to freshwater predators. 2 Because mergansers pose a threat to salmonid populations only when they are abundant i n salmonid-rearing habitat, i t i s desirable to understand when, and where, undesirable concentrations of mergansers are l i k e l y to occur. The present study focusses on the foraging behaviour and dispersion of common mergansers i n r e l a t i o n to salmonid enhancement programs on the east coast of Vancouver Island, B r i t i s h Columbia. This research was conducted concurrently with studies of other p o t e n t i a l predators of salmonid f r y - staghorn sculpins, Leptocottus armatus and Bonaparte's g u l l s , Larus Philadelphia (Mace 1983) and p r i c k l y sculpins, Cottus asper (Jones, i n preparation). The Common Merganser - D i s t r i b u t i o n and L i f e History; The common merganser, or goosander, i s one of the largest ducks (family Anatidae, t r i b e Mergini) and can weigh as much as 2 kg. I t i s larger and more widely-distributed than e i t h e r the red-breasted merganser (M. serrator) or the hooded merganser (M. c u c u l l a t u s ) . Both the common and red-breasted mergansers feed almost e x c l u s i v e l y on f i s h . Unless indicated, a l l further reference to mergansers w i l l p e r t a i n only to M^ merganser. Mergansers are common throughout the boreal and montane forest regions of North America. They breed across Canada south of the t r e e l i n e and i n much of northern USA. Wintering areas are r e s t r i c t e d to the P a c i f i c coast and Maritime provinces of Canada but include most of the contiguous states of USA. Migration through the P a c i f i c Northwest region peaks during November - December and February - March (Bellrose 1978). 3 Nesting takes place near freshwater in cavities within trees or eroded stream banks; the eggs are laid from late March to mid-June (White 1957). The broods emerge from late May to July and remain on or near their natal stream unt i l able to f l y at 10 wk of age (Erskine 1971). Male mergansers leave the nesting areas in June prior to the wing moult when they are unable to f l y . Breeding females begin to moult during July - August once the broods are well grown. It is presumed that moulting birds seek the safety of open water during this vulnerable period (Bellrose 1978). On Vancouver Island, mergansers are common throughout the year. However, they are most abundant during the early spring at the peak of the northerly migration and during late summer following recruitment of the juveniles. Merganser densities are lowest in this area during the late nesting season (June) after the males have dispersed. Assessing Predation by Mergansers: Early studies of predation by mergansers aimed to identify what proportion of the merganser's diet i s comprised by salmonids and how diet varies among different habitats (e.g. Leonard and Shetter 1936; White 1936,1937; Munro and Clemens 1936,1937; Salyer and Lagler 1940). Some investigators recorded daily consumption by captive birds and attempted to relate diet composition to the types of fishes available (e.g. White 1957; Latta and Sharkey 1966). These studies demonstrated the potential for predation by mergansers and provided incentive for further research. Diet studies, however, cannot reveal the extent nor the impact of predation; i t is also necessary to investigate the predator-prey interaction directly. 4 There are two basic problems i n assessing the impact of a predator: the f i r s t i s to measure the o v e r a l l consumption of prey and the second i s to determine the s i g n i f i c a n c e of these losses. For example, predators may remove prey that are surplus to the rearing capacity of the habitat or eliminate vulnerable i n d i v i d u a l s that would die anyway. Merganser-removal experiments were undertaken by the F i s h e r i e s Research Board of Canada to address the l a t t e r problem. These investigations showed that freshwater production of hatchery-reared A t l a n t i c salmon parr (Salmo s a l a r ) planted i n a natural stream could be increased by an average of 500% i f the parr were protected from predation by mergansers (Elson 1962). Unfortunately, i t i s d i f f i c u l t to r e l a t e these conclusions to other systems because the two basic questions - mortality due to predation vs subsequent s u r v i v a l of uneaten prey - are confounded. Mo r t a l i t y of parr due to mergansers was not evaluated, except i n d i r e c t l y by i t s influence on parr s u r v i v a l . Whereas estimates of mortality due to mergansers may be applicable to other r i v e r systems, s u r v i v a l of the uneaten parr i s probably very dependent on l o c a l conditions, the abundance of a l t e r n a t i v e predators and d e t a i l s of the hatchery-stocking program. For example, compensatory k i l l by other predators may obscure mortality caused by the controlled predator (Alexander 1979). Nevertheless, Elson's report provides compelling evidence that mergansers can l i m i t salmonid production i n freshwater. The dynamic nature of predator-prey i n t e r a c t i o n s has l a r g e l y been ignored i n studies of salmonid predators (but see Peterman and Gatto 1978, Mace 1983). The d e n s i t i e s of prey and predator are seldom constant - yet they are fundamental variables determining the rate of predation (Holling 1959a). Studies to assess predation that take no account of f u n c t i o n a l 5 r e l a t i o n s h i p s between the abundance of prey and predator w i l l , i n g e n e r a l , have l i m i t e d value i n other contexts. Moreover, such s t u d i e s w i l l not be u s e f u l f o r making d e c i s i o n s about enhancement and e x p l o i t a t i o n p o l i c i e s that may a l t e r prey d e n s i t y d r a s t i c a l l y . For these reasons, I have i n v e s t i g a t e d aspects of the merganser's behaviour that i n f l u e n c e feeding success, d i e t , choice of f o r a g i n g s i t e and f l o c k s i z e i n r e l a t i o n to the a v a i l a b i l i t y of j u v e n i l e P a c i f i c salmon. By understanding these r e l a t i o n s h i p s , i t i s p o s s i b l e to p r e d i c t the r e l a t i v e i n t e n s i t y of predation by mergansers ( F i g . 1). I have computed estimates of m o r t a l i t y due to mergansers f o r streams where there i s adequate i n f o r m a t i o n regarding f i s h d e n s i t y . However, I have not attempted to determine the impact of t h i s m o r t a l i t y i n terms of l o s t production by salmonid populations. This t h e s i s i s organized i n t o three parts as f o l l o w s : In Part I , I d i s c u s s the behaviour of i n d i v i d u a l mergansers and t h e i r i n t e r a c t i o n s w i t h c o n s p e c i f i c s , w h i l e f o r a g i n g on enclosed s e c t i o n s of a n a t u r a l stream stocked w i t h v a r i o u s d e n s i t i e s of j u v e n i l e coho salmon (Oncorhynchus k i s u t c h ) . Factors that i n f l u e n c e feeding success w i t h i n a patch of constant prey d e n s i t y ( i . e . an enclosure) are considered i n Chapter 2. In the f o l l o w i n g chapter, I examine the patch-assessment behaviour of mergansers and the r e l a t i v e importance of p r o f i t a b i l i t y and s o c i a l i n t e r a c t i o n s i n determining the d u r a t i o n of v i s i t s to the enclosure s i t e . In Part I I , my o b j e c t i v e has been to determine to what extent the l o c a l d i s p e r s i o n of mergansers on Vancouver I s l a n d can be explained by the a v a i l a b i l i t y of prey, and i n p a r t i c u l a r , the a v a i l a b i l i t y of j u v e n i l e 6 Fig. 1. Organization of thesis; the conceptual framework to assess predation by mergansers. PREDATION BY ADULT MERGANSERS CHAP. 2: INDIVIDUAL FEEDING RATE V S SALMONID AVAILAB IL ITY CHAP. 3: CHOICE OF FORAGING SITE V S SALMONID AVAILABILITY CHAP. U: SIZE OF F L O C K S AND AGGREGATIONS V S SALMONID AVAILABIL ITY CHAP. 6: MORTALITY OF DOWNSTREAM MIGRANTS PREDATION BY BROODS CHAP. 5: NUMBER O F BROODS V S SALMONID AVAILABILITY CHAP. 7: MORTAL ITY OF S T R E A M - R E S I D E N T J U V E N I L E S 8 salmonids. In Chapter 4, I extend conclusions from Chapters 2 and 3 regarding the behaviour of individual birds, to the population level and then to a larger spatial scale. Scheduled releases of hatchery fish which cause large perturbations in salmonid density, provide the means for testing these predictions about population responses to pr o f i t a b i l i t y . In Chapter 5, I consider how nesting dispersion is also influenced by the availability of juvenile salmonids. Finally, in Part III, I draw conclusions about the intensity of predation by mergansers on juvenile salmonids within the context of salmonid enhancement programs. I f i r s t examine predation of salmon fry and smolt during their seaward migration in Chapter 6. Then, in Chapter 7, I discuss the potential for predation of stream-resident juvenile salmonids by merganser broods. Chapter 8 contains some concluding remarks and suggestions for further research. FORAGING BEHAVIOUR 10 CHAPTER 2 : CONSTRAINTS ON FEEDING SUCCESS INTRODUCTION The h u n t i n g p e r f o r m a n c e o f a p r e d a t o r i s c o n s t r a i n e d by many f a c t o r s . N e v e r t h e l e s s , two r e l a t i o n s a r e f u n d a m e n t a l t o a l l f o r a g i n g s i t u a t i o n s - t he r e l a t i o n s h i p between a p r e d a t o r ' s f e e d i n g s u c c e s s and p r e y d e n s i t y ( t h e f u n c t i o n a l r e s p o n s e t o p r e y d e n s i t y , H o l l i n g 1959a) and t h a t between a p r e d a t o r ' s f e e d i n g r a t e and t he d e n s i t y o f p r e d a t o r s ( t h e f u n c t i o n a l r e s p o n s e t o p r e d a t o r d e n s i t y , H o l l i n g 1961 ) . I n t h e p r e s e n t s t u d y , the r a t e o f p r e d a t i o n on j u v e n i l e coho sa lmon by t he common merganse r was e v a l u a t e d w i t h r e s p e c t t o sa lmon d e n s i t y and t he s i z e o f mergan se r f l o c k s f e e d i n g on f i s h s t o c k e d w i t h i n e n c l o s e d s e c t i o n s o f a n a t u r a l s t r e a m . T h i s was a f i r s t s t e p toward s a s s e s s i n g changes i n t he i n t e n s i t y o f p r e d a t i o n t h a t m i g h t r e s u l t f r om m a n i p u l a t i o n o f j u v e n i l e s a l m o n i d d e n s i t i e s t h r o u g h h a t c h e r y enhancement p rog rams . V u l n e r a b i l i t y and t he F u n c t i o n a l Response t o P r e y D e n s i t y : The e f f i c i e n c y w i t h w h i c h a p r e d a t o r c a p t u r e s p r e y can be r e p r e s e n t e d by two p a r a m e t e r s - t he r a t e o f e f f e c t i v e s e a r c h , a , and t he t i m e , t h , r e q u i r e d t o consume and d i g e s t each p r e y ( H o l l i n g 1 9 5 9 a ) . These p a r a m e t e r s d e s c r i b e a b a s i c f u n c t i o n a l r e s p o n s e t o p r e y d e n s i t y t h a t i s known as 11 t h e ' d i s c e q u a t i o n ' ( R o l l i n g 1959b): N a = aNT / (1 + a t h N ) [2.1] where N a i s the ( i n s t a n t a n e o u s ) number o f s u c c e s s f u l a t t a c k s , N i s the number o f p r e y and T i s the t o t a l t ime p r e y a r e exposed. I f a i s c o n s t a n t , t h e a t t a c k r a t e i n c r e a s e s a s y m p t o t i c a l l y w i t h p r e y d e n s i t y ( a type I I r e s p o n s e ) because the model d e f i n e s a time budget w h e r e i n l e s s time becomes a v a i l a b l e f o r s e a r c h as more p r e y a r e h a n d l e d . T y p i c a l l y , hunger w i l l i n f l u e n c e a p r e d a t o r ' s r a t e o f s e a r c h but i f s e a r c h r a t e d e c l i n e s p r o g r e s s i v e l y w i t h d e c r e a s i n g h u n g e r , the t y p e I I r e s p o n s e w i l l be r e i n f o r c e d . Other t y p e s o f f u n c t i o n a l r e s p o n s e ( i . e . s i g m o i d and dome-shaped) may o c c u r whenever the v u l n e r a b i l i t y o f p r e y depends on p r e y d e n s i t y ( H o l l i n g 1959a; see Mace 1983 f o r a r e c e n t r e v i e w o f a l t e r n a t i v e m o d e l s ) . V u l n e r a b i l i t y i s d e f i n e d i n t h i s c o n t e x t as the r a t e at w h i c h p r e y i s e n c o u n t e r e d ; i t i s the p r e d a t o r ' s r a t e o f e f f e c t i v e s e a r c h from the p r e y ' s p o i n t o f v i e w . Among j u v e n i l e salmon, v u l n e r a b i l i t y i s l a r g e l y dependent on a b i o t i c f a c t o r s such as t e m p e r a t u r e (Hartman 1965), t u r b i d i t y o r l i g h t i n t e n s i t y ( P a t t e n 1971, G i n e t z and L a r k i n 1976), t i d e h e i g h t (Mace 1983 - A p p e n d i x ) and c o v e r ( H o l t b y and Hartman 1982). Y e t r e l a t i v e l y few a b i o t i c f a c t o r s can be e x p e c t e d t o produce d e n s i t y - d e p e n d e n t changes i n v u l n e r a b i l i t y t h a t a l t e r t h e form o f the p r e d a t o r ' s f u n c t i o n a l r e s p o n s e . Those t h a t do, f o r example, 12 the amount of cover affording refuge from predation, may be very important i n natural sit u a t i o n s (Errington 1946). Nevertheless, functional responses are l i k e l y to be most p l a s t i c with repect to behavioural a t t r i b u t e s of both predator and prey. Predator t r a i n i n g may increase prey v u l n e r a b i l i t y as encounters with prey become more frequent. The formation of a s p e c i f i c searching image whereby a predator learns to detect prey more e a s i l y following previous encounters (Tinbergen et a l . 1967) has been demonstrated in several vertebrate species ( f i s h , Ware 1971; b i r d s , Dawkins 1971, Mueller 1974 and Pietrewicz and Kamil 1981). A l t e r n a t i v e l y , experience gained by the prey following recent exposure to predators may decrease i t s v u l n e r a b i l i t y ( D i l l 1974, Ginetz and Larkin 1976). For these reasons, i t was necessary to investigate, and to control the e f f e c t s of cover a v a i l a b l e to j u v e n i l e salmon and t h e i r previous exposure to merganser attacks, p r i o r to evaluating the functional response. Merganser hunting behaviour i s also analyzed in terms of sequential components that determine the rate of ingestion; that i s , the e f f e c t s of f i s h density, s a t i a t i o n and previous rates of encounter are examined for each component of the 'feeding chain' (Holling 1959a, C o l l i e r and Rovee-Collier 1981) from search to capture and f i n a l l y , to ingestion. The Functional Response to Predator Density: The number of other predators foraging in close proximity i s l i k e l y to influence a predator's feeding success. Group foraging may be b e n e f i c i a l i f 13 activity by the group causes a favourable disturbance of the prey, increases the probability of capture or reduces handling time. Organized (co-operative) hunting behaviour is characteristic of some species (e.g. Bartholomew 1942, Kruuk 1972) but individuals foraging independently within a group may also experience greater success (Moynihan 1971; Hafner et a l . 1982). Alternatively, social interactions among predators may be unfavourable; mutual interference may arise directly through aggressive behaviour that wastes time (Rogers and Hassell 1974, Sutherland and Koene 1982) or indirectly through increased opportunities for prey training and avoidance behaviour to occur (e.g. Charnov et a l . 1978). The relative gain or loss in feeding efficiency associated with foraging in groups w i l l be important in determining social behaviour and the maximum size of foraging groups. This., in turn, has important consequences for the dispersion of predators and thus, the intensity of predation in specified areas. Several investigators (White 1957, Huntingdon and Roberts 1959, Des Lauriers and Brattstrom 1965, Miller 1973) have suggested that mergansers forage co-operatively, but these reports have not been confirmed by experiment. Accordingly, I have attempted to evaluate the effect of flock size on the foraging success of mergansers while controlling for the effect of prey av a i l a b i l i t y . The implications of these results for flock size and salmonid mortality are examined in subsequent chapters. Prey Selection: Potential prey is seldom of a single species or size class in natural situations and any attempt to understand predation must face up to the issue 14 of diet selection. Two general approaches for predicting diet exist: frequency-dependent models and optimal diet models. The former class of models is conveniently represented in terms of the multispecies functional response (Murdoch 1973). If the search and handling-time parameters are independent of density, prey of each type w i l l be taken in proportion to its abundance. Disproportionate selection of prey types is evidence that the functional response parameters are not independent of density, either because of differential availability or preference (Cock 1978). On the other hand, most optimal foraging models assume that prey w i l l be ranked (i.e. preferred) according to its pro f i t a b i l i t y in terms of energetic value and ease of acquisition. The decision to include a potential food item in the diet (subject to constraints of toxicity and specific nutrient requirements) should not depend on its abundance relative to other items, but only on the abundance of more profitable items (e.g. Pyke et a l . 1977). However, the ranking of prey types may change as a function of relative density due to changes in the rate of effective search (Ostfeld 1982, Visser 1982). Moreover, in practice, i t is d i f f i c u l t to distinguish between a preference for higher-ranking prey and biased selection due to differential v i s i b i l i t y among prey types (e.g. Werner and Hall 1974). In a second series of experiments with stream enclosures, predation rates by mergansers were compared for two size classes of juvenile coho salmon - smolt (average weight = 40 g) and fry (2 g). These results are discussed with reference to predictions from optimal diet theory and a frequency-dependent model based on apparent size. 15 METHODS Study S i t e : Rosewall Creek i s a small coastal stream draining an area of 45 km2 on the east coast of Vancouver Island ( F i g . 2 and 3A). Although i t originates i n mountainous t e r r a i n at 1500 m elevation, the lower 3.8 km i s navigable by salmon and has an average gradient of 6%. The r i v e r supports only a small wild salmonid population due to extreme annual fluctuations in water l e v e l . It was therefore an a t t r a c t i v e s i t e for experiments involving manipulation of f i s h density. Moreover, the Rosewall Creek experimental hatchery, maintained by the Canadian Department of F i s h e r i e s and Oceans, provided a convenient source of f i s h already acclimated to the r i v e r water. Stream flow declined and temperature increased from 6 to 10°C over the course of the experiments (May 1 - June 8, 1981). I presume that these changes did not have a major e f f e c t on my r e s u l t s . Enclosure Design: Enclosures were part i t i o n e d with 1.3 cm mesh seine-netting stretched across the stream. The foot of the netting was sealed against the riverbed with l e a d l i n e s and sandbags and the upper margin was secured as inconspicuously as possible to f l o a t i n g logs; the logs were chained together and anchored with s t e e l posts to permit adjustment to changes in r i v e r depth. Three adjacent enclosures were constructed, each including t y p i c a l stream habitat that varied in the amount of cover available from undercut banks and 16 F i g . 2 . Map showing study areas. Q-LC - Quinsam and lower Campbell R. ; TS -Tsable R.; WC - Wilfred Creek; RC - Rosewall Creek; NC - Nile Creek; BQ - Big Qualicum R.; LQ - L i t t l e Qualicum R.; ENG - Englishman R. 18 F i g . 3. Detailed maps of p r i n c i p a l study areas: A - Rosewall Creek; B -L i t t l e Qualicum R.; C - Big Qualicum R. I n t e r t i d a l zone i s s t i p p l e d . 19 20 21 22 overhanging roots ( F i g . 4). The upper and lower enclosures were nearly the same s i z e , with surface areas of 320 m2 vs 310 m2, average depths of 0.40 m vs 0.46 m and hence t o t a l volumes of 130 m3 vs 140 m3. The middle enclosure was smaller but deeper (195 m2 by 0.49 m deep), and included rapids and undercut tree stumps for much of i t s length. Experimental Design: Two age classes of juvenile coho salmon were stocked as prey: smolt hatched 16 mo e a r l i e r which averaged 42.9 ± 0.7 g (95% C.I.) and f r y , 4 mo old averaging 2.27 ± 0.02 g. Fish were transferred from the hatchery to the enclosure i n large p l a s t i c p a i l s just p r i o r to each t r i a l . A remote-controlled f i s h release system was constructed to compensate for depletion of f i s h due to predation within the enclosures; surplus f i s h were retained inside p l a s t i c mesh cages and released as required to maintain f i s h density within 10% of the desired l e v e l (± 1 f i s h or 20% at the very lowest density). The f i s h were allowed 15 min to explore t h e i r enclosures before observations began. With two exceptions, at least one hour elapsed before the f i r s t mergansers a r r i v e d . In the f i r s t series of experiments (May 1 - 28) coho smolt were stocked at d e n s i t i e s of 5, 10, 25, 50, 100 and 200 fish/enclosure (Table I ) . Paired comparisons (5 vs 10, 25 vs 100, 50 vs 200) for the upper and lower enclosures were chosen to provide contrast between enclosures and between the t o t a l quantity stocked on successive days. The order of presentation was random, except that each comparison was repeated with enclosures reversed in the following t r i a l . Because the middle enclosure d i f f e r e d from the others with Fig. 4. Layout of stream enclosures in Rosewall Creek. 24 bl ind D e p t h D i s t r i b u t i o n s U p p e r : a r e a - 3 2 0 m 2 vo l - 130 m 3 S d e p t h - 0 . 4 m 0.5 1.0 D E P T H L o w e r : a r e a - 3 1 0 m 2 v o l - U 0 m 3 x d e p t h - 0.46 m 0.5 1.0. m D E P T H M 2 5 Table I: Target den s i t i e s of coho smolt and f r y stocked during enclosure experiments at Rosewall Creek, 1981. Asterisks indicate that the majority of f i s h remained from the previous t r i a l . Coho smolt averaged 43 g, coho f r y , 2 g. ENCLOSURE LOWER MIDDLE UPPER Date Smolt Fry Smolt Fry Smolt Fry May 4 200 10 50 5 50 10 200 8 25 10 100 9 100 10 25 10 5 10 10 12 10 10 5 14 50 25 200 16 200 25 50 18 5 25 10 20 10 25 5 22 100 25 25 24 25 25 • 100 26 5 195 20 175 0 25 28 5 195 5 0 0 50 29 0 500 5 10* 25 175 30 25 175 5 45* 0 500 June 1 0 50 5 10* 5 195 2 0 200 5 40* 2 198 6 50 150 50 0 50 2* 7 10 190 50 20* 1* 200 8 10 190 50 25* 2* 200 Evening of June 8 40,000 steelhead smolt (avg. weight = 30 g) released from hatchery 26 respect to several important variables, direct comparisons were not attempted. Rather, fi s h density was held constant at 10 smolt for the f i r s t 6 t r i a l s and at 25 smolt for the next 6 replications, to provide a control for time effects. During the second set of experiments (May 29 - June 8) both fry and mixtures of smolt and fry were used. Pure fry were stocked at densities of 10, 50, 200 and 500 and the mixtures ranged from 1 to 25% smolt to make up an overall density of 200 f i s h . Each treatment was presented randomly and independently in both upper and lower enclosures, except that identical treatments were never scheduled for the same day; this procedure ensured that maximum information would be provided about patch assessment and movements of birds between the enclosures (Chap. 3). At the end of each t r i a l , the nets were raised and fis h were allowed to escape. A l l smolt l e f t the study site prior to the next t r i a l and presumably had emigrated to sea overnight. Unfortunately, during the later experiments, residual (experimental) fry and newly-hatched wild fry tended to remain throughout the study area. Attempts to remove these f i s h by electrofishing were not entirely successful due to the small size of the fry and the extremely low conductivity of the water; thus, i t proved convenient to flush fry out of the upper and lower enclosures and into the middle enclosure where their numbers could be estimated. The density of residual fry tended to remain f a i r l y constant over time and did not exceed 50. Electrofishing also revealed a large number of small (0+ age class) prickly sculpins (Cottus asper) hidden beneath stones throughout the study area. However, their presence had no obvious effect on the experiments - the 3 sculpins captured 27 during the entire observation period comprised < 0.5% of a l l prey taken by mergansers. Observational Procedure: Two observers watched mergansers forage from a camouflaged blind, situated upstream from the enclosures. The behaviour of focal individuals and their position within each enclosure was observed continuously with 12x binoculars, and recorded on a Digitorg event recorder (Gass 1977). Because foraging behaviour was of principal interest, focal individuals were chosen using a combination of 'fixed behaviour' and 'rotation' c r i t e r i a (Altmann 1974). The f i r s t bird to enter an enclosure was observed for 20 min or until i t had ceased foraging for at least 5 min; thereupon another individual was chosen arbitrarily from those remaining and observed for a similar period. The activity and position of a l l mergansers in view (scan samples) were also recorded on the Digitorg at 20 min intervals. The second observer kept track of a l l occurrences of arrival, departure, capture, and whenever possible, attack on prey; usually individual birds could be followed continuously or recognized by distinctive markings. All-occurrence and focal records were later cross-referenced to provide a detailed account of flock formation and feeding rates on each day. Otters and eagles sometimes visited the enclosures, eating the fish and frightening the mergansers; these observation periods were ignored. On two occasions, satiated immature mergansers continued to maim fis h without eating them and I disregarded further activity on this enclosure. 28 Other Observations: Additional information on the effect of flock size on feeding success was collected in two less-controlled f i e l d studies. Feeding rates were monitored for flocks of adult mergansers foraging at the mouth of Rosewall Creek following the release of approximately 16,000 steelhead smolt on May 15 and June 5, 1980 from the hatchery upstream. These smolt averaged 37.7 i 2.1 g (95% C.I.) in the May release and 47.0 ± 3.1 g during the June release. A l l observations were recorded from a nearby blind under very similar conditions of tide (> 3 m), weather (calm and clear) and time of day (17:00 - 22:00 h) and on both occasions, within 30 h of fi s h release. Steelhead smolt continued to emigrate downstream and out to sea in large numbers for 3 - 4 d; thus, fi s h density at the mouth of the river probably remained relatively constant during the period of observation. The effect of brood size on feeding rates of merganser ducklings was evaluated from observations made June 10 - 12, 17, 19 and 24 - 26 at the mouth of the L i t t l e Qualicum R., 30 km southeast of Rosewall Creek. (Fig. 3B). While feeding there, ducklings ate primarily small sculpins (especially Leptocottus armatus) which were very abundant in the estuary. Again, because of the high density of sculpins at this site, I ignored depletion by merganser broods over the period of observation. St a t i s t i c a l Analysis: Digitorg behaviour records were decoded by computer (Gass 1977) and summarized for approximately 30 s intervals. Few intervals were exactly 30 s 2 9 because each behavioural event has a f i n i t e duration; however, 92% of the i n t e r v a l s were between 30 - 35 s and a l l f e l l within 30 - 40 s. I selected the i n t e r v a l duration for three reasons. F i r s t , i t i s short enough to allow e f f i c i e n t use of data during b r i e f foraging bouts. Second, i t i s reasonable to assume that successive i n t e r v a l s are s t a t i s t i c a l l y independent with respect to pursuit and capture behaviour since the duration of these events << 30 s (Table I I I ) ; because changes in prey experience or in motivation to attack would v i o l a t e t h i s assumption over long periods, I grouped observations according to l e v e l of merganser s a t i a t i o n and exposure of prey to mergansers. Third, two captures were very r a r e l y made within a single i n t e r v a l , so that capture frequencies can be defined as the proportion of i n t e r v a l s during which capture occurred. Thus, each i n t e r v a l was considered to be an independent B e r n o u l l i t r i a l . Contingency tests based on a log - l i n e a r model (BMDP number 3F), were performed to determine whether capture frequency was affected by treatments ( i . e . s a t i a t i o n , prey exposure, prey density, cover, merganser flo c k s i z e , sex and maturity) and the i r i n t e r a c t i o n s . (Log transformation i s recommended for analysing frequencies that vary widely between 0 and 50%, Snedecor and Cochran, 1980.). P a r t i a l functional responses (not including the diges t i v e pause, e.g. F i g . 5) were f i t t e d by weighted non-linear regression ( l e a s t sum-of-squares) to capture rates using the disc-equation model, equation [2.1], where T and t h are measured i n 30 s u n i t s . 30 RESULTS No A l t e r n a t i v e Prey Rate of E f f e c t i v e Search - Coho Smolt: S t a t i s t i c a l analysis of smolt capture frequencies using a l o g - l i n e a r model indicated that capture rate increased with density (p < .0001, x2, Table I I , F i g . 5), and was higher in the enclosure with least cover (p < .0001). Moreover, p r o b a b i l i t y of capture following a single exposure to merganser attacks was s i g n i f i c a n t l y reduced for a l l subsequent attacks during that day (p < .0002). The number of smolt eaten during the previous 2 h was used as an index of s a t i a t i o n and i s referred to as the '2 h cumulative consumption' index (2 h c c ) ; new a r r i v a l s were presumed to have a 2 h cc = 0. Mergansers whose 2 h cc > 2 showed a s i g n i f i c a n t l y lower capture rate than those with a 2 h cc < 2 (p < .05). Neither the 'search i n t e n s i t y ' index (defined as the proportion of time spent with head submerged) nor the 'search rate' index (the number of head dips/min) were s i g n i f i c a n t predictors of capture rate; neither index was strongly correlated with the 2 h cc index (r = -.12 and -.24, r e s p e c t i v e l y ) . These observations are analyzed in more d e t a i l i n a l a t e r section. The sex and maturity of mergansers had no s i g n i f i c a n t e f f e c t on p r o b a b i l i t y of capture. Unfortunately, the data are i n s u f f i c i e n t to evaluate the influence of experience gained by i n d i v i d u a l mergansers on capture rates a f t e r c o n t r o l l i n g for density, enclosure, s a t i a t i o n and prey exposure. However, there i s no evidence that p r o b a b i l i t y of capture increased on the middle enclosure over the course of the smolt 31 experiments (May 1 - 26), either as a r e s u l t of long-term t r a i n i n g e f f e c t s or changes in other factors such as water l e v e l and temperature (p > .10, x2). Rates of e f f e c t i v e search (a) were estimated by f i t t i n g the disc-equation model [2.1] to observed capture rates, f i r s t using observed (independent) estimates of p a r t i a l handling time ( t p , excluding d i g e s t i o n ) , d second by estimating a and t p simultaneously (Table I I ) . The observed t p were independent of both smolt density and enclosure and are summarized in Table I I I . The corresponding ( p a r t i a l ) functional responses presented i n F i g . 5, i l l u s t r a t e the e f f e c t s of enclosure and prey exposure. Those in F i g . 6 indicate the r e l a t i v e s i g n i f i c a n c e of s a t i a t i o n under conditions of maximum and minimum prey v u l n e r a b i l i t y . The r e l a t i o n s h i p s with search i n t e n s i t y presented in F i g . 7 are d i r e c t l y comparable (except for ordinate scale) to the r e l a t i o n s h i p s with 2 h cc < 2 in the previous figure ( F i g ; 6B). Rate of E f f e c t i v e Search - Coho Fry: Fewer data were obtained during the experiments with f r y because mergansers v i s i t e d the enclosures less often. This appears to have been i n response to reduced p r o f i t a b i l i t y (Chap. 3 and 4). No reduction in rate of capture following previous exposure was apparent for f r y (sample size was too small for an adequate s t a t i s t i c a l comparison); - accordingly the p a r t i a l functional response in F i g . 8 i s derived from the combined data for the upper enclosure. Too few data were obtained from the lower enclosure to permit comparison. Table II: Estimated rates of e f f e c t i v e search (a) for coho smolt and f r y with respect to enclosure ( i . e . cover), previous exposure to mergansers and l e v e l of s a t i a t i o n . P a r t i a l handling time (tp) does not include digestive pause. Values are given for 30 s time u n i t s ; i . e . t p = 0.4 i s equivalent to 12 s. Prey Enclosure Consumption (2 h cc) Previous Exposure? t p KNOWN t p ESTIMATED _ l p -smolt lower <2 no yes .004 .002 0.40 .002 .002 2.28 1.43 >2 no yes .0004 ,0005 no convergence .0004 -2.67 upper <2 no yes .024 .011 0.40 .019 0.23 .011 0.38 >2 no yes ,001 .009 no convergence .025 1.46 fry upper <2 a l l .004 0.23 .005 0.34 33 Table I I I : Duration and success rate of prey handling a c t i v i t i e s for coho smolt and f r y . Time spent i n unsuccessful pursuit or handling i s included in duration estimates. DURATION (sec) PREY ACTIVITY Mean Std. Dev. Number % SUCCESS Smolt pursuit and capture ( t c ^ 2.7 2.5 20 1 35.8 subdue and eat ( t e ) 9.2 9.4 68 5 0 . 0 t o t a l ( t p ) 11.9 17.9 Fry 1.6 1.6 22 36.4 t e 5.3 5.1 124 3 3 . 9 6.9 12.3 34 Fig. 5. Partial functional responses (excluding digestive pause) of mergansers to coho smolt density with respect to amount of cover available to smolt and their previous exposure to mergansers. A - upper enclosure (minimal bank cover); B - lower enclosure (undercut banks). Dashed lines - smolt not previously exposed to merganser attacks; solid lines - smolts previously exposed; a 1 refers to upper bound on estimate of rate of effective search (a). Curves fitt e d to 'disc equation' (Holling 1959b) by least squares where tp = 0.398 (30 s units) from independent observations. Numbers indicate number of 30 s intervals of observation; 95% C.I. given for Poisson distribution. A UPPER ENCLOSURE (minimal bank cover) 1 o o no previous exposure a = .018 p<.005 • previous exposure a= .001 p<.025 a'(max) = .0295 B LOWER ENCLOSURE (undercut banks) a =.004 pOO a = .002 p<.05 9 ° 9 o „ — — ^92 50 100 150 200 NUMBER OF SMOLT 3 6 Fig. 6. Partial functional responses (excluding digestive pause) of mergansers to coho smolt density with respect to vulnerability of prey and cumulative consumption by mergansers. .A - maximum vulnerability (upper enclosure, no previous exposure to attacks); B- minimum vulnerability (lower enclosure, previously exposed). Solid line - hungry mergansers (2 h cc < 2); dashed line - mergansers near satiation (2 h cc > 2). Curves fitted to 'disc equation' (Holling 1959b) by least squares where t p = 0.398 (30 s units) from independent observations. Numbers refer to number of 30 s intervals of observation; 95% C.I. given for Poisson distribution. o LO O ro Q LU CC ID f— CL < O cc LU DO 2.0-1.5-1.0-0.5-A MOST VULNERABLE (upper enclosure, no previous exposure) • 2 h cc«=2; a=.02A o 2 h cc ^2; a = .001 p<.005 .05. 25 0.5-109 50 100 150 NUMBER OF SMOLT 200 38 F i g . 7. P a r t i a l functional responses with respect to search i n t e n s i t y . S o l i d l i n e - vigorous search (head submerged > 40% of time); dashed l i n e - casual search (head submerged •_< 40% but > 1% of time). Lower enclosure; a l l f i s h previously exposed; 2 h cc < 2. Curves f i t t e d to 'disc equation' ( H o l l i n g 1959b) by least squares where t p = 0.398 (30 s units) from independent observations. Numbers r e f e r to number of 30 s i n t e r v a l s of observation; 95% C.I. given for Poisson d i s t r i b u t i o n . 40 Fig. 8. Partial functional response to coho fry in upper enclosure. Prey exposure not controlled; mergansers hungry (2 h cc < 2). Curves fitted to 'disc equation' (Rolling- 1959b) by least squares where tp = 0.230 (30 s units) from independent observations; 95% C.I. given for Poisson distribution. 41 NUMBER OF FRY 42 Derivation of Functional Response The digestive pause (t .25, F i g . 9), which indicates that mergansers are not becoming gorged due to incomplete digestion between foraging bouts. No such control was possible in obtaining the f i r s t estimate; hence, I consider the l a t t e r to be more r e l i a b l e . 43 F i g . 9. Relationship between apparent digestion time per coho smolt ( t ^ 1 ) and time since a r r i v a l at enclosure s i t e . S o l i d figures - female and immature mergansers (x = 55 min); open figures - mature males (x = 64 min) combined mean = 62 ± 23 min (95% C.I.). Each symbol represents an i n d i v i d u a l b i r d ( t o t a l of 8). y = 66.4 - 3.36x, n = 18, r = -.14 F l , 1 6 = 0.31, p > .25 6 b o CO LU 2004 1504 r -CO LU S2 100-Q I-~ZL LU < Q_ Q_ < 504 0 • immature • AT f ema les O D A V males v 0 o ! x= 62 ± 2 3 min ( 9 5 % CI) regression NS n=18. p>.25 o 1 2 3 4 5 TIME SINCE ARRIVAL (h) 45 I t follows that t o t a l handling time ( t h = t p + t d ) i s approximately 1 h. This implies that the maximum number of smolt consumed per day, given unlimited prey density, i s equivalent to the length of time i n hours available for foraging - for example, 12 smolt (480 g) given 12 h daylight. Yet because td » tp, 95% of the maximum can be obtained at enclosure de n s i t i e s of only 10 inexperienced smolt with poor cover or 150 previously-exposed smolt with good cover. An alternate estimate of maximum d a i l y appetite was obtained by extrapolating the observed food intake rates for mergansers that remained i n sight continuously for more than 4 h ( F i g . 10). These are the same ind i v i d u a l s from which the estimate of digestive pause was computed. Thus, the extrapolated estimate of d a i l y appetite d i f f e r s from the functional response predictions only with respect to the rate of e f f e c t i v e seach; indeed the former estimate can be regarded as a check on the independence of a, derived from many b i r d s , and td, which i s derived from only a select few. In a 12 h day, female and immature birds are ex'pected to consume 450 g or 40% of body weight, whereas the larger, mature males would eat 290 g or 20% of body weight. These estimates are consistent with values predicted by the functional response at a density of 125 previously-exposed smolt with good cover or 5 inexperienced smolt with poor cover. Furthermore, the estimates agree well with average d a i l y appetites reported for captive mergansers ranging from 306 - 440 g (White 1957). It i s informative to represent the functional response in terms of time required to s a t i s f y the d a i l y appetite of, l e t us say, 400 g (10 smolt or 200 fry) for a range of prey v u l n e r a b i l i t y ( F i g . 11). For f r y , two a l t e r n a t i v e 46 assumptions have been made: the f i r s t , that digestion time is proportional to fish weight (i.e: 1/20 x 60 min = 3.0 min for fry) and the second, that digestion time is only half the former value because of the greater surface area/volume ratio of fry. These assumptions define an upper bound for predation of fry and indicate that fry densities must be on the order of 50 -400 fry/enclosure (without bank cover) to satisfy the daily appetite. This is an order of magnitude greater than required for smolt. 47 Fig. 10. Pattern of fish consumption by individual mergansers vis i t i n g the enclosure site for > 4 h. Weight of f i s h consumed is expressed as a percentage of body weight. A - 3 females and 1 immature bird; y = 8.40 + 2.65 (x-1), n = 20, r = 0.61, p < .005, y (12 h) = 37.6% = 450 g. B - 4 mature males; y = 4.96 + 1.18 (x-1), n = 21, r = 0.67, p < .001, y (12 h) = 17.9 = 290 g. Firs t hour on enclosures is ignored in regression because a l l mergansers are hungry upon arrival. Each symbol represents an individual bird. 48 40H A o immature 3»c females n = 20, p < .005 avg. weight = 1200 g 301 I o r o r < z •— _ ci) CO J o-o iZ O -Q 20-10-o-0_ Z> (/) z o CJ 30-(MO «t) B all ma le s , avg. weight = 1600 g n = 21 p < .001 20H 10-0 0 2 4 6 T IME S INCE ARRIVAL (h) 8 4 9 Fig. 11. Foraging time (including digestive pauses) required to satisfy daily requirements (0.4 kg) with respect to fish density. A - coho smolts (th = 60 min), B - coho fry (th = 3.0 min, lowest bound, th" = 1.5 min). Lower bounds computed from functional responses to most vulnerable fish (upper enclosure, no previous exposure), upper bounds, from least vulnerable fish (lower enclosure, previous exposure). F*8 A 20-wk-th = 60 min 15- ,minimum vulnerability 1 *^-"^« • * • * r *- *i f"»* — M - m V nW-- mm — - f m\\t *m\m. IT/ T' W - - m*, • »-* 1 10-; maximum vulnerability i i 5-i i i i i i i i 1— 1 1 1 1 1 1— 0 20 40 60 NUMBER OF SMOLT PER 100 m 2 B 20- t h = 3 min 15-minimum vulnerability '; ^f&azg&^'&i* 10- i i r. i I'. ' i •. i i *. i 1 i" l •.. i \ Nmaximum vulnerability 5- i I l l , i i i 1 i \ t h= 1.5 min, max. vulnerability n i j u C ) 200 400 600 800 1000 1 1 NUMBER OF FRY PER 100 m 2 51 Alternative Prey Available Size Selection - Fry vs Smolt: Feeding rates by unsatiated mergansers (2 h cc <^ 2) presented with mixtures of previously exposed smolt and fry are summarized in Fig. 12 for both upper and lower enclosures. Unfortunately, usable data were not obtained for several of the treatments due to low visitation rates and a learned preference for the upper enclosure (Chap. 3). No fry were captured in the lower enclosure even at high densities (195 fry, 5 smolt) which suggests that bank cover is especially important for the smaller f i s h . In the upper enclosure, fry were eaten frequently only when they comprised 99% (91% by weight) of the available prey; even then, they contributed only 35%, by weight, to the overall catch. The extent of selection in favour of smolt is illustrated in Fig. 13. Fig. 13A includes the data of the previous figure with additional observations of mergansers near satiation (2 h cc > 2). A second analysis of additional data is presented in Fig. 13B; the larger sample sizes were achieved at the expense of relaxed control over total f i s h density (range 15 to 200) and prey exposure. It should be noted, however, that the middle enclosure included a substantial proportion of fry remaining from earlier experiments (see Methods). If the residual fry had become less susceptible to predation due to selection or greater experience, the size-selection curve would be biased in favour of smolt. S t i l l , the curves in Fig. -13A and B are similar and i t is noteworthy that only a single data point, based on only two captures, f a l l s below the diagonal. The lumped data in Fig. 13B are consistent with, but offer l i t t l e support for an earlier conclusion that selection is more intense on smolt when adequate bank cover is 5 2 Fig. 12. Capture rates of hungry mergansers (2 h cc < 2) feeding on mixtures of coho fry and smolt. Total density = 200 fish, a l l previously exposed. A - upper enclosure (minimal bank cover); B - lower enclosure (undercut banks). Dots - overall capture rate (fry and smolt); circles - smolt only. Note change in scale of y-axis. Numbers refer to number of 30 s intervals of observation; 95% C.I. given for Poisson distribution. A U P P E R ENCLOSURE 0.8H 0.6-O ro OX 3 0.2 ZD O smolt • total (smolt + fry) ± 95 % C.I. 97 27 l 20, < cc LU CO O 0 B LOWER ENCLOSURE 0.3-0.24 0.14 0-no fry captured 32 99 r e © 0 >45 200 50 100 150 NUMBER OF SMOLT (fry comprise balance of 200 fish) 5 4 Fig. 13. Size selection by mergansers foraging for coho smolt and fry. A - total fish density = 200, a l l previously exposed; merganser hunger level not controlled. B - total fish density varies between 15 and 200; exposure and hunger level not controlled. Numbers refer to number of fish captured; curves fitted by eye. 55 o LU o r Z> r-D_ < O •c CO LL CD O < no selection — o UPPER — • LOWER 200 NUMBER OF SMOLT (fry comprise balance of 200 fish) o co ! o o i — or o a. o or CL 0.5-1 0 7 23 7 • • 4 3 • • .* • *^--no selection 3 23 38 .•'114 • • • • • O UPPER • MIDDLE • LOWER f i l l 0 T" T " " 0.5 1 l • 1 1 1.0 PROPORTION SMOLT IN ENCLOSURES 5 6 a v a i l a b l e for f r y . An Apparent-Size Hypothesis: At f i r s t glance, s e l e c t i o n i n favour of smolt appears to be consistent with predictions of optimal d i e t where rank i s defined in terms of weight/handling time ( t p ) ; this r a t i o i s on the order of 3.3 g/s for smolt but only 0.3 g/s for f r y . However, se l e c t i o n of smolt can be explained equally well i n terms of t h e i r greater conspicuousness, without postulating a preference for l a r g e r , more rewarding f i s h . Following Werner and H a l l (1974) and Confer and Blades (1975), I have assumed that a merganser perceives a f i s h only i f i t s size and distance are such that the r e s u l t i n g image ( i . e . apparent size) exceeds some c r i t i c a l angle at the merganser's eye. Thus, the zone of perception depends on the length of the f i s h and the geometry of the v i s u a l f i e l d . For a spherical v i s u a l f i e l d , the zone of perception w i l l be proportional to the length of the f i s h cubed and for a planar v i s u a l f i e l d , i t w i l l be proportional to the square of f i s h length. Because the enclosures used in t h i s study were shallow in comparison to the p o t e n t i a l size of the v i s u a l f i e l d , and because mergansers search from the surface down, the actual zone of perception w i l l be of intermediate s i z e . The p r o b a b i l i t y of encountering randomly-distributed prey w i l l be proportional to the size of the zone of perception. Thus, the d i f f e r e n t i a l v i s i b i l i t y of f r y and smolt w i l l be given by the r a t i o of t h e i r zones of perception. It i s assumed that prey i s s u f f i c i e n t l y dispersed so that f i s h of d i f f e r e n t s i z e do not occur within the zone of perception simultaneously. The rate of e f f e c t i v e search, a, depends on the p r o b a b i l i t y of 5 ? encountering and recognizing prey within the reactive f i e l d and on the probability of i n i t i a t i n g a successful attack given the encounter. Since the present comparison involves only different size classes of the same species, the importance of other attributes influencing recognition is reduced. Moreover, pursuit and k i l l success for smolt and fry are comparable (Table III). Therefore, differences in the rate of successful search under comparable conditions are attributed to either the probability of encounter (i.e. differential v i s i b i l i t y ) or willingness to i n i t i a t e an attack (i.e. preference). Accordingly, the ratio of the size of zones of perception for fry and smolt can be compared to the corresponding ratio of the rates of successful search to test the following predictions: (1) i f a f < (lf\ then \ Vs) the apparent-size hypothesis is not sufficient to account for differences in vulnerability; preference for smolt is implied. Subscripts f and s denote fry and smolt respectively; 1 denotes length. then selection of smolt can be explained on the basis of apparent size alone; preference for smolt is not indicated. 58 A l t e r n a t i v e l y , i f mergansers search for prey against a fixed area of background ( i . e . zone of perception independent of prey s i z e ) , i t i s l i k e l y that the p r o b a b i l i t y of detecting a prey item w i l l be proportional to i t s long-sectional area - that i s the proportion of the reactive f i e l d i t occupies. Because growth in juvenile salmonids i s nearly isometric, the girth/length r a t i o i s comparable for smolt and f r y and t h e i r long-sectional area, as presented to a merganser, i s given by the same formula. The long-sectional area of a cy l i n d e r with length, 1, and radius, r , i s given by 2 r l and i t s volume (or weight at unit density) by 7rr2l. The radius of an isometric c y l i n d e r with weight = Trr^l = c l 3 , where c = constant, i s given by ^c/ir 1, and the r a t i o of long-sectional areas of two isometric cylinders with lengths If and l s i s therefore ( l f / l s ) 2 . I f juvenile salmonids are adequately represented by c y l i n d e r s , then the n u l l hypothesis for size s e l e c t i o n i s again a f / a s = ( l f / l s ) 2 . Note that for the upper enclosure where both prey had been previously exposed, a f / a s = 0.37 whereas ( l f / l s ) 2 = 0.13. Thus, the observed rate of e f f e c t i v e search for coho smolt i s i n s u f f i c i e n t , in comparison with that of f r y , to r e j e c t e i t h e r formulation of the n u l l hypothesis that s e l e c t i o n of smolt r e s u l t s merely from t h e i r greater conspicuousness. If anything, f r y are taken more frequently than would be expected from t h e i r apparent size - at least i n the upper enclosure where cover was minimal. In f a c t , a f / a s l f / l s > which implies that prey were taken in proportion to t h e i r lengths rather than long-sectional areas or reactive volumes. However, the data are inadequate for a convincing t e s t . If the rate of e f f e c t i v e search for f r y that generates a curve passing through the lowest 95% C.I. bound in F i g . 8 i s taken as a minimum value (af' = 0.0022) and s i m i l a r l y , the 59 value generating a curve through the upper 95% C.I. bound for capture rate of smolt with no previous exposure (Fig. 5A) is taken as a maximum value (as 1 = 0.0295), the minimum ratio, af'/a s' = 0.075. Since af'/a s' will exceed the true ratio i f either af 1 is overestimated (p « .025), or as' i s underestimated (p ss.025), the ratio can be considered as an (approximate) lower 95% confidence bound. This lower bound includes ( l f / l s ) ^ D u t n o t ( l f / l s ) 3 ; accordingly, the null hypothesis of no preference should not be rejected. 60 E f f e c t of Flock Size Unsatiated mergansers (2 h cc < 2) foraging alone or i n pairs for smolt that had been previously exposed were neither more nor less successful than those foraging in groups of 3 - 7 ( F i g . 14). Observations of flocks of up to 25 mergansers foraging i n an area of comparable size at the mouth of Rosewall Creek, likewise, show no s i g n i f i c a n t evidence of interference or cooperation ( F i g . 15). These observations are less rigorous however; f i s h density i s unknown but assumed to have been constant (see Methods). Also, any birds experiencing reduced feeding success due to competition may have l e f t the s i t e before successful i n d i v i d u a l s and, as a r e s u l t , interference e f f e c t s may be obscured. Even so, i t i s c l e a r that feeding rates did not improve with increasing flock s i z e . The same conclusion follows from observations of merganser broods of up to 15 downy young foraging in very close proximity for small sculpins (Cottidae) at the mouth of the L i t t l e Qualicum River ( F i g . 16). Once again, sculpin density i s assumed to have been constant over the periods of observation (June 10 - 12, 17 - 19, and 24 - 26) but in t h i s case, downy young always remained together so that departure of unsuccessful birds cannot have biased the a n a l y s i s . The age of broods (estimated from 14 to 32 d) was not a s i g n i f i c a n t f actor. 61 Fig. 14. Partial functional responses to coho smolt with respect to number of mergansers foraging on enclosures. Fish least vulnerable (i.e. lower enclosure, previous exposure). Dots - <^ 2 birds foraging; circles - 3 to 7 birds foraging. Focal merganser was hungry (2 h cc < 2). Curve fitted to combined data by least squares using 'disc equation' (Rolling 1959b) where tp = 0.398 (30 s units) from independent observation; 95% C.I. given for Poisson distribution. NUMBER CAPTURED / 30 sec 2 9 6 3 Fig. 15. Individual capture rates vs number of mergansers foraging for steelhead smolt at the mouth of Rosewall Creek, May 16-18 and June 5-6, 1980. Numbers indicate number of flocks observed; 95% C.I. given for Poisson distribution. Total observation time = 9.5 h. Weighted linear regression of ungrouped data is not significant (p > .25): y = 0.077 - .002x, n = 12 r = -.042, F 2 > 1 0 = 0.035 total observation time = 9.5 h a — or c CD £ \ \ ± r o LU E o r L U o r CD -Q E .25 .20-.15-M - .10' .05-0 no flocks = 6 1-2 3-4 5-6 12-18 22-25 SIZE OF FORAGING GROUP 65 Fig. 16. Individual capture rates vs number of downy young mergansers in broods foraging for young-of-year sculpins at mouth of L i t t l e Qualicum R. , June 10-26, 1981. Numbers refer to weights in regression (number of young X minutes observed); 95% C.I. given for Poisson distribution. y = 0.19 + O.Olx, r = 0.76 F l , 3 = 4.1, p > 0.10 66 _1 NT o Q or LU Q_ LU < or LU or z> h-Q_ < c Z5 o in cu xi £ C NUMBER OF DUCKLINGS IN BROOD 6 ? The Feeding Chain Search: As demonstrated previously, search intensity was not a satisfactory predictor of capture rates even at constant smolt density (Fig. 7). This result warrants more detailed consideration of the behavioural sequence from search through ingestion as illustrated in Fig. 17. There is evidence that vigorous search activity is as much a consequence of sighting prey as a prerequisite for i t . Search intensity increased immediately following pursuit of prey and declines gradually over a period of minutes p < .001, F test, Fig. 18). As expected, search intensity declined as mergansers approached satiation while feeding on smolt (Fig. 19A). Moreover, search was always slightly more vigorous at high smolt density (50 - 200) than at low density (5 - 25) presumably because encounters were more frequent and search behaviour was reinforced. Mergansers nearly sated with smolt also dove less often (.07/15 s ± .03, 95% C.I.) than those with 2 h cc < 2 (.14/15 s ± .03, p < .01, x2). These trends are evident in a l l t r i a l s with coho smolt but cannot be discerned in any of the t r i a l s with coho fry. In fact, search intensity increased, at f i r s t , with fry consumption (Fig. 19B). Even at low fry density (25 - 50), search intensity did not decline as birds became sated; clearly this is not just a consequence of more frequent reinforcement due to a higher average density of fry. The fact that capture rate is predicted by level of satiation but not by search intensity suggests that hunger acts on later events in the feeding, chain (links 3, 4 or 5 in Fig. 17). 68 Fig. 17. Schematic representation of feeding feedback channels. 1. reinforcement of search behaviour 2. reduced search intensity 3. reduced reactive distance 4. reduced pursuit success 5. more prey escape grasp 6. more prey discarded chain showing potential with satiation SEARCH « ENCOUNTER PURSUIT CAPTURE 4 SUBDUE r SWALLOW < I © © © © HUNGER © 7 0 F i g . 18. Search i n t e n s i t y vs time since l a s t pursuit. 95% C.I. and regression (using midpoints of i n t e r v a l s ) computed af t e r arcsine transformation. y = 38.8 - l . l x , n = 10, r = - .87 Fi,8 = 24.1, p < 0.001 "D CD CD CD e U) TD O CD SL 50-cu E o \ o CO Z LU o or < LU CO 40-30-20-10 0 •(vVrt.^/.'-'.•rv-'/i'. •/•'•v'.*V-~ ••'•.••;-'rf S. ij, •:, v-./j;;-'-. . • ' ^ 3 , . - i . . . . • :":^ .f.v4'T!; >':--.?^j;:p;.T. ;%''-x'.V': -^' • r - V . - v s . '••i.'.:'.''v!i' ',«'•.•"•• iYv' .':-"'> .*'•<" ...*'"<•* •. SIS' 0 8 10 TIME SINCE LAST PURSUIT (min) 7 2 Fig. 19. Relationships between search intensity and degree of satiation with respect to prey size and density. A - coho smolt; B - coho smolt and fry. Dots - high density; circles - low density. Mean search intensity with 95% C.I. is given for discrete consumption levels in A but for midpoints of intervals in B. 7 3 CUMULATIVE CONSUMPTION (g) DURING PREVIOUS 2h 7 4 B COHO FRY =5 60-CD O l c_ CD E W "D O CD CD E >-CO LU r-Z 20I •25-50 fry/enclosure T / 0 0 2 0 0 - 500 fry/enclosure . - 0 x o or < LU CO A . 0 80 0 AO CUMULATIVE CONSUMPTION (g) DURING PREVIOUS 2h 120 160 Pusuit and Capture: Mergansers close to satiation pursued fish less frequently than did hungry mergansers (p < .05, x2 } pig. 20). Neither hunger nor smolt density had a significant effect on pursuit success (the proportion of pursuits resulting in capture) or on the proportion of smolt that managed to wriggle free after being captured. Pursuit success was slightly reduced in the lower enclosure where more cover was available but the difference was not s t a t i s t i c a l l y significant (p > .20, x2, Fig. 21). There is no evidence that smolt previously exposed to mergansers eluded pursuit better than those with no experience (p > .75, x2). 7 6 Fig. 20. Frequency of pursuit vs level of satiation and coho smolt density in lower enclosure. A l l fish previously exposed; flock size <^ 2. Stipple -mergansers near satiation; white - hungry birds. Numbers refer to number of 15 s intervals of observation; 95% C.I. given for Poisson distribution. 77 CH 0 - 1 | smolt eaten during W\ 2 - 3 ) P r e v ' o u s 2 hr LO \ to ZD LO o r ZD 0_ LL O o r LU CQ .15-.10-.05-0 saturation effect: p <.001 density effect: p= .01 74 n=350 LOW DENSITY (5-25 smolt) 211 146 HIGH DENSITY (50-200 smolt) 78 Fig. 21. Pursuit success with respect to cover available to coho smolt and their previous exposure to merganser attack. White - smolt previously exposed; stipple - smolt not previously exposed. Numbers refer to number of attacks observed; 95% C.I. given for binomial distribution. 79 CO CO LU o o 3 CO CO 3 CO rx 3 D_ LL O rx o D_ o rx Q_ 1.0" .8-0-19 n=37 UPPER ENCLOSURE | | previous exposure no previous exposure p >.20 19 47 LOWER ENCLOSURE 201 i w 2£ ALL DATA 80 DISCUSSION Factors Influencing Hunting Performance: Merganser feeding success was influenced by fi s h density, cover available to the fish, their previous exposure to merganser attacks, and the merganser's hunger level. A type II (asymptotic) functional response was observed under each experimental condition; there was no evidence of threshold responses to cover or previous exposure in either enclosure. This result i s surprising. It suggests that the available cover was not depleted as fish density increased. Although pursuit frequency was reduced in the enclosure with more cover, pursuit success was not significantly different between the enclosures. This implies that either mergansers encountered fis h less frequently, or alternatively, that they attacked more carefully, to ensure success, when better cover was available. But i f encounter frequency is reduced by cover, a threshold response is expected as fi s h density is increased; at some point, the cover becomes inadequate to conceal a l l f i s h and the attack rate must increase. Differences in vulnerability probably arise, therefore, through differences in the quality rather than quantity of cover. That i s , mergansers must search longer or more carefully to achieve an advantage that ensures a favourable probability of capture where good cover is available. The undercut banks, found only in the lower enclosure, appear to have provided more protection than the sunken logs, branches and debris common to both enclosures. Alternatively, i t is possible that more subtle differences between the enclosures were responsible for the discrepancy in capture rates. 81 Fish with previous exposure to mergansers were less vulnerable than those with no previous exposure. Again, attacks were not less successful, merely less frequent, on exposed fi s h . The difference in vulnerability before and after exposure was probably not a result of the removal of especially vulnerable fi s h . If this were true, the difference should not have been apparent after only one exposure at high fish density since a constant proportion, and thus, a greater number of fish, would have been especially vulnerable. Rather, i t is likely that, once alerted to danger, fish utilized cover or behaved so as to frustrate the merganser's ab i l i t y to init i a t e attack. Similarly, Ginetz and Larkin (1976) found that sockeye salmon fry (Oncorhynchus nerka) were less susceptible to predation by rainbow trout (Salmo gairdneri) after previous exposure to trout or after passing through a counting apparatus. In the latter case, reduced vulnerability must have developed through experience rather than a r t i f i c i a l selection, because a l l fry survived the treatment. As mergansers became satiated, their attack rate declined even when search intensity did not. Assuming their visual acuity was undiminished, mergansers must have reacted less often to encounters when partly satiated. Neither pursuit success nor time required for pursuit decreased with decreasing hunger so i t seems implausible that mergansers pursued only fish that appeared easy to catch. More likely, mergansers responded to fish within a smaller and smaller reactive f i e l d as hunger was sated. Hunger-dependent reactive fields have been demonstrated for several other species (Holling 1966, Mace 1983, Ware 1972). 82 Search i n t e n s i t y declined as mergansers became satiated with coho smolt, although i t increased b r i e f l y following encounters with prey. No such trend was observed for mergansers feeding upon coho f r y . This absence of c o r r e l a t i o n between search i n t e n s i t y and coho f r y consumption may be due to the cumulative e f f e c t s of ingestion and digestion on gut capacity, coupled with a non-linear (or threshold) r e l a t i o n s h i p between gut capacity and motivation to search. The larger smolt would accumulate more quickly in the gut than would f r y because digestion acts continuously whereas ingestion occurs only at i n t e r v a l s . Consequently, a gut-capacity threshold, such as might trig g e r reduced search a c t i v i t y , would be exceeded more quickly while feeding on smolt than on f r y . It i s also apparent that newly-arrived mergansers foraged less i n t e n s i v e l y during the f r y experiments than during the e a r l i e r smolt experiments. Unfortunately, the hunger l e v e l of newly-arrived birds could not be determined. It i s possible that with experience at the enclosure s i t e , mergansers anticipated the urge to forage and arrived at the s i t e less hungry than during previous, exploratory v i s i t s . A l t e r n a t i v e l y , feeding conditions outside the enclosure s i t e may have become more favourable; c e r t a i n l y wild salmonid f r y were more numerous within Rosewall Creek during the l a s t week of t r i a l s . But given the uncertainty regarding hunger l e v e l upon a r r i v a l at the enclosure s i t e , a more rigorous analysis of differences in search motivation for smolt and f r y i s not possible. No evidence was found to support claims that mergansers forage co-operatively (see White 1957, Huntingdon and Roberts 1959, M i l l e r 1973 regarding the commmon merganser, and Des Lauriers and Brattstrom 1965, the red-breasted merganser). Although mergansers were commonly observed to 83 forage in close proximity and, occasionally, along parallel search paths, their feeding rate was no higher under these conditions than when foraging alone. Mergansers usually move continuously while foraging. Hence, members of a flock wishing to remain together for reasons unrelated to foraging efficiency (e.g. to reduce the risk of predation) can be expected to move along parallel paths especially when following a depth contour. Co-ordinated foraging behaviour was most often observed among groups of birds (particularly juveniles) searching along the shoreline of estuaries; i t was rarely seen on rivers. Moreover, search paths seldom remained co-ordinated once fi s h schools were encountered. Alternatively, i t is possible that mergansers do benefit by co-ordinated search behaviour but that this merely compensates for lost efficiency due to indirect interference. In other words, given that an individual is foraging near conspecifics, i t may be necessary to co-ordinate its search behaviour to maintain a satisfactory feeding rate. It is perhaps more surprising that mergansers feeding in flocks were, on average, no less successful than solitary birds. Direct interference through agonistic interactions was rarely observed except among males in the presence of mature females (Chap. 3). Indirect interference due to prey disturbance is probably minimized by moving continuously while foraging. In any case, there seem to be no net costs or benefits in terms of feeding efficiency, associated with flock size - at least over the range of flock sizes commonly encountered. Flock size is probably limited, not by decreases in foraging efficiency, but through local depletion of prey and the need to continuously explore new areas. 84 Prey Selection and Appetite: Coho smolt were eaten more frequently than coho fry when stocked together in the enclosures. Other investigators (Salyer and Lagler 1940, Elson 1962, Alexander 1979) have concluded that mergansers select large salmonids over small, based on the predominance of large fish in stomach samples compared with their availability in the habitat where the mergansers had been shot. This size-selection is consistent with expected energy gain given the time required to pursue and swallow prey of each size class, and the probability of successful attack. However, the higher predation rate on smolt can be explained equally well in terms of their greater conspicuousness without postulating preference. In fact, after accounting for apparent size, there appears to be residual selection in favour of fry, but given the confidence intervals on estimates of af and a s, the discrepancy is probably not significant. S t i l l , i t may be significant that af/a s approximates l f / l s very closely; this relationship suggests that the probability of detecting prey depends only on length, rather than long-sectional area. Clearly, the perceptual a b i l i t i e s of a predator must be well understood to distinguish, with confidence, between selection due to differential v i s i b i l i t y and selection due to preference. In general, a merganser's daily energy gain during the experiments was constrained by the time required for digestion, not by hunting ab i l i t y . Approximately 1 h is required to digest a 40 g coho smolt so that daily consumption would not exceed 500 g in an average (12 h) day. This maximum estimate agrees well with previously published estimates of daily consumption: these range from 250 g for captive, immature birds which lost 85 weight (Latta and Sharkey 1966, Miller 1973) to 306 - 440 g for captive and tamed mergansers that maintained weight (White 1957). It is clear from the feeding rates observed in this study that mergansers could satisfy daily appetites of 400 g at relatively low fish densities - < 33 smolt /100 m2 depending on the amount of cover and previous exposure to mergansers (Fig. 11). Densities of wild coho smolt in productive streams are typically 17 - 67/100 m2 (Mundie and Traber 1983). During releases of hatchery-reared fish, smolt densities are often increased by 2 - 3 orders of magnitude. In other words, mergansers feeding exclusively on juvenile salmonids w i l l find adequate food resources in a l l but the least productive streams. 86 CHAPTER 3: PROFITABILITY AND PATCH CHOICE INTRODUCTION A forager must make decisions about when and where to concentrate i t s search e f f o r t . At the l e a s t , i t should adopt some r u l e which allows i t to avoid r e l a t i v e l y unprofitable feeding conditions. Decision rules f o r terminating foraging bouts have been of c e n t r a l i n t e r e s t i n most discussions of optimal foraging theory (e.g. Charnov 1976, Pyke et a l . 1977). Mobile predators are confronted by patchy dispersion of prey i n space as w e l l as time. Where patches of high prey density are d i f f i c u l t to locate or assess from a distance, the predator must sample. A r e a - r e s t r i c t e d searching behaviour w i l l improve foraging e f f i c i e n c y i f prey i s clustered (Tinbergen et a l . 1967); f o r example, the p r o b a b i l i t y of searching t e r r i t o r y adjacent to a previous capture point, can be enhanced through changes i n speed of search and turning angle (Smith 1974). A l t e r n a t i v e l y , i f cues to in d i c a t e patch q u a l i t y are a v a i l a b l e , an e f f i c i e n t forager should choose among patches by considering, f i r s t , the cost i n terms of time and energy that w i l l be incurred by t r a v e l l i n g between patches (e.g. Cowie 1977) and second, the r e l i a b i l i t y of the cues. The 'best' decision rule for leaving a patch depends on the d i s t r i b u t i o n of prey among patches and on what the forager knows about that d i s t r i b u t i o n (Iwasa et a l . 1981, McNair 1982). Ollason (1980) shows by way of an hydraulic memory model, how animals unfamiliar with the d i s t r i b u t i o n of prey among patches might forage e f f i c i e n t l y by leaving a patch whenever It feeds less 87 successfully than i t remembers doing. This result is a clever reformulation of the area-restricted search strategy where the decision rule for termination depends on previous feeding rates and the persistence of memory. Where patches are distinct but known to vary widely in quality, a 'giving-up-time' (GUT) rule i s appropriate - regardless of whether patches are depleted of prey by the forager (Krebs 1974, Iwasa et a l . 1981). Additional information that suggests a particular patch is superior to others should alter the decision rules in favour of longer v i s i t s to that patch (McNair 1982). In particular, the presence of a group of conspecifics or competitors would often convey information about patch quality (i.e. 'local enhancement', Thorpe 1963). For a predator like the common merganser whose feeding success is not markedly reduced through interference by conspecifics (Chap. 2), the presence of a flock might bias i n i t i a l patch choice and also create expectations that alter decision rules influencing residence time at the patch. In this chapter, I investigate how mergansers respond to differences in pr o f i t a b i l i t y among patches at two spatial scales, using a conceptual model of patch choice outlined in Fig. 22. F i r s t , I examine responses to p r o f i t a b i l i t y in terms of movements within and between adjacent enclosed sections of a natural stream stocked with coho smolt at different densities. Second, I consider the combined enclosures to be a larger patch within the river system where alternative patches include freshwater reaches farther upstream, and more importantly, the intertidal zone of the estuary. Responses to pr o f i t a b i l i t y at the enclosure site were measured in terms of the duration and frequency of v i s i t s over successive t r i a l s . 88 Fig. 22. A conceptual model for patch choice. 1. memory of previous foraging success 2. functional response 3. decision rule based on current p r o f i t a b i l i t y 4. local enhancement 5. mutual interference/facilitation 6. aggressive/gregarious interactions 89 PREVIOUS EXPERIENCE 90 METHODS Study S i t e and Observation Method: The movements and v i s i t i n g habits of mergansers were monitored during experiments to investigate t h e i r feeding rate on coho smolt and f r y . Details pertaining to s i t e , design of strea- enclosures and f i s h stocking schedules are given i n Chap. 2 (Figs. 3A and 4, Table I ) . Stakes along the boundaries of each enclosure served to delineate quadrants within the enclosure when sighted in l i n e from the b l i n d . The quadrants were roughly equal i n size for the upper and lower enclosures (80 m2) but smaller (50 m2) in the middle enclosure. One observer recorded movements of fo c a l i n d i v i d u a l s within enclosures (from quadrant to quadrant), and between enclosures, in conjunction with relevant foraging behaviour, using a Digitorg event recorder (Chap. 2). A second observer monitored times of a r r i v a l and departure and f i s h consumption for each merganser that v i s i t e d the enclosure s i t e . The number of birds in the v i c i n i t y was recorded at the beginning and end of each t r i a l day. The census area included the lower kilometer of Rosewall Creek (20% of the length accessible to salmon), and the i n t e r t i d a l and offshore zones within a 1 km radius of the r i v e r mouth. Interpretation of Movements on Enclosures: Mergansers us u a l l y moved continuously while foraging. The speed and 91 d i r e c t i o n of a search path determined, on average, how quickly i t crossed from one quadrant to another. Hence, the d i s t r i b u t i o n of search time per quadrant provided a convenient measure of a r e a - r e s t r i c t e d searching behaviour. Giving-up-times (GUT) were defined as the time elapsed since capturing a f i s h , (but not n e c e s s a r i l y eating i t ) , and leaving the enclosure to continue foraging elsewhere. If no f i s h were captured the GUT was considered to be the ent i r e period spent foraging on the enclosure. To be included i n the GUT a n a l y s i s , a merganser must have foraged continuously (no pauses > 30 s) from the time of entering the enclosure, or since resuming search a c t i v i t y following a pause > 5 min, u n t i l i t s departure from the enclosure. A l l data concerning movements pertain to s o l i t a r y or paired birds, i . e . f l o c k size <^ 2. Because the male of a breeding p a i r tends to follow i t s mate, only data for the female were analyzed, except i n comparisons of the duration of foraging bouts; the sequence and timing of movements by members of a pair were usually very s i m i l a r , although the duration of foraging a c t i v i t y varied widely within the p a i r . Decoy Experiment: A d d i t i o n a l data pertaining to the r o l e of f l o c k s i z e i n determining a r r i v a l rate was obtained during an experiment with decoys at the mouth of the L i t t l e Qualicum R. ( F i g . 3B). Polyurethane decoys were anchored i n a swimming (non-foraging) posture i n a r t i f i c i a l f l o c k s of 4, 8 and 16 decoys of which 25% had male breeding plumage. Time of a r r i v a l and departure and the proportion of birds f l y i n g overhead that landed i n a 100 X 40 m zone surrounding the decoys was monitored from a raised b l i n d 50 m d i s t a n t . The experiment was conducted on 8 d over a 3 wk period (June 10 - 12, 17, 19 and 92 24 - 26, 1981) with two observation sessions per day beginning 6 h apart. This schedule ensured that t i d e l e v e l and current were s i m i l a r during the f i r s t and l a s t week but reversed during the second week due to the 2 wk c y c l e of semi-duirnal t i d e s . Moreover, t i d e l e v e l was s i m i l a r but with current reversed, during the two sessions of each day. A c o n t r o l (no decoys) and experimental p e r i o d , each of 80 min d u r a t i o n , was scheduled during both morning (09:00 - 11:30 h PST) and afternoon (13:00 - 15:30 h) sessions. The f l o c k s i z e treatments were arranged i n random sequence except that no decoys were set out during the f i r s t period (09:00 - 10:20 h) of each day. 93 RESULTS Patch Assessment: Search Pattern Within an Enclosure: Search was distributed evenly over a l l quadrants within each enclosure, regardless of depth, after taking into account differences in surface area among quadrants (p > .25, x2). Search time within quadrants appeared to increase with fi s h density (Fig. 23) as would be expected i f either search speed was reduced or search paths became more circuitous in response to an increased frequency of encounter with prey. The trend is s t a t i s t i c a l l y significant only for the lower enclosure (p < .05, x2) but i t is comparable, with fewer data, in the upper enclosure (p < .10). A l l fish had been previously exposed to mergansers in this comparison. Further, durations < 10 s were excluded to avoid including 'skittering-pursuit' across quadrant boundaries - an event more lik e l y to occur at higher f i s h density. Search Pattern Among the Enclosures: GUT was not significantly different among enclosures (p > .25, x 2) nor between levels of satiation (p > .25, x2); neither were there any obvious differences with respect to flock size nor fi s h exposure, although too few data were available for an adequate test. Moreover, the distributions of GUT showed no change with increasing smolt density (p > .25, x 2) and overall, were consistent with that expected from a random departure process with no 'ageing' 94 Fig. 23. Time spent foraging within quadrants of enclosure vs coho smolt density. A l l fish previously exposed. Arrows indicate means. Upper - p < .10, x2, 6 df; lower - p < .05, x2, 6 df. >-u ~z. LU ZD O LU cr L i . 0.6 1.2 1.8 24 3.0 3.6 X = 0.69 X = 0.87 LOWER smolt density = 5 . n = 94 p <.05 10 n = 28 25 n = .48 50 - 2 0 0 .n = 185 1.2 2.4 3.6 4.8 6.0 7.2 DURATION (minutes) 96 effect: Pr (departure between t D and t^) = " * x dx [3.1] where

0.25, x2; B - p > 0.25, x2. 86 99 surface area was probably of l i t t l e consequence; but i f they did not, as suggested above, then duration of search within an enclosure may have depended on surface area so that a correction factor has been applied. It i s evident from F i g . 5 (Chap. 2) that the upper enclosure always provided the highest capture rates (0.06 - 0.26/30 s for smolt densities of 5 - 25) whereas the middle enclosure provided the lowest (0.03 - .04/30 s for smolt de n s i t i e s of 10 - 25). As expected, foraging bouts tended to be longest on the upper enclosure, of intermediate duration on the lower enclosure and shortest on the middle enclosure ( F i g . 25, p < .025 a f t e r correction for surface area, x2). Additional evidence indicates that mergansers were able to remember differences in p r o f i t a b i l i t y from day to day. The frequency with which mergansers i n i t i a t e d foraging bouts on the upper enclosure increased s t e a d i l y over the course of the experiments (p < .005, x2, F i g . 26), even though the majority of birds arrived at the enclosure s i t e by swimming or f l y i n g upstream and, therefore, encountered the lower enclosure f i r s t . It would seem that 1 mergansers developed an expectation of higher p r o f i t a b i l i t y i n the upper enclosure over the course of several weeks. However, there i s no evidence that such an expectation resulted in longer GUT ( F i g . 27) as predicted by an optimal GUT model (McNair 1982). 100 Fig. 25. Comparison of duration of foraging bouts on lower, middle and upper enclosures. Fish density <^ 25 coho smolt, a l l previously exposed. Arrows indicate means. 1 0 1 30 20 10 0 ^"v --'w*-iv V V.*- ••• -o LU 20-O LU o r 10 0 10-0 r x = 1.5 l.rV-.7 V - T C MIDDLE n = 47 p < .001 ~! 1 1 r LOWER n = 39 x = 3.6 • T i i i i r UPPER n = 20 x = 4.6 - F T T ~i r l — — i — — i r 0 6 8 DURATION (min) i r 10 » . 1 12 102 Fig. 26. Trend in choice of enclosure on which foraging bouts are initiated. Males of pairs are excluded; 95% C.I. given for binomial distribution. NO. OF FORAGING BOUTS INITIATED % FORAGING BOUTS INITIATED ON UPPER _s. NJ CO N ) * ^ C O C O o o o o o o o o o o I i i i I i i i i i i i L_ CO N ) 104 F i g . 27. D i s t r i b u t i o n s of GUT on upper enclosure with respect to experience. White - early experience (May 12-24, mean = 2.2, n = 13); black - l a t e r v i s i t s (May 14-24, mean = 2.1, n = 22); s t i p p l e - expected frequency i f departures are at random (exponential d i s t r i b u t i o n corresponding to o v e r a l l mean). D i s t r i b u t i o n s do not d i f f e r s i g n i f i c a n t l y (p > .5, x2, 4 d f ) . 1 0 5 >-o z LU 0.8-o LU 0.6 or • May 1 - 12 , n= 13 • May 14-24, n= 22 expected if random p > .50 LU 0.4 > LU or 0.2 0 0 - 2.5 2.5-5.5 GUT (min) 5.5 106 A r r i v a l s at the Enclosure Site E f f e c t of Previous Experience: The frequency of v i s i t s to the enclosure s i t e increased from May 1 to May 26 when the enclosures were stocked with coho smolt, yet decreased to the pre-stocking l e v e l during May 28 - June 8 when predominately coho f r y were stocked ( F i g . 28). A r r i v a l rates were calculated from observations while no mergansers were present at the enclosure s i t e to avoid possible bias from s o c i a l i n t e r a c t i o n s . A square-root transformation was used to s t a b i l i z e variances since a r r i v a l s occurred at random ( i . e . Poisson d i s t r i b u t i o n ) . The f i r s t few days of increasing a r r i v a l rates co r r e l a t e with increased census counts following stocking of the enclosures; thereafter, census counts tended to decline as expected from seasonal trends (Chap. 4) whereas a r r i v a l rates continued to increase. The decrease i n a r r i v a l frequency from May 28 to June 8 i s correlated with the census counts which declined more r a p i d l y than expected from seasonal trends. Although some birds v i s i t e d the enclosure s i t e several times a day (as in f e r r e d from a few recognizable i n d i v i d u a l s ) , a r r i v a l rate was not correlated to either f i s h density (p > .25, F test) or to time of day (p > .10, Wilcoxon signed-rank t e s t ) . Therefore, I conclude that mergansers responded to the enclosure s i t e over longer periods of time than one day. It appears that mergansers were able to remember the l o c a t i o n of the enclosure s i t e while concentrations of coho smolt were a v a i l a b l e , but to 'unlearn' the response once less p r o f i t a b l e d e n s i t i e s of coho f r y were stocked. More than a week of reinforcement was required to acheive the maximum response. 107 Fig. 28. Trends in arrival rate (A) and abundance (C) during enclosure experiments at Rosewall Creek, May 1 - June 8, 1981. Arrival rates are included only i f no birds were present at the site to avoid bias from social interactions. Dotted line shows the expected seasonal decline (Chap. 4). 108 A • < 2 . 5 h observation • 2.5 - 4.9 h • 2=5.0 h 1.04 LU < — or JC >.% • E - 0 . 5 or < , ENCLOSURE E X P ' T S 0 Q LU o —-J - c n CO B 10-CO LL ~ 5 0-co £ co -° 10 LU o smolt (40 g) fry (2 g) and smolt i I I II i J - 1 expected seasonal decline 30 10 JUNE 1 0 9 E f f e c t of Flock Size: More rapid i d e n t i f i c a t i o n of p r o f i t a b l e s i t e s should be possible through information conveyed by the behaviour of conspecifics ( i . e . l o c a l enhancement). A r r i v a l rates while no birds were present on the s i t e were lower than when at least one merganser was present. However, because mergansers were not abundant i n the v i c i n i t y , i t i s necessary to consider the number of birds on the s i t e which have been removed from the pool of mergansers l i k e l y to a r r i v e ; the difference i s s t a t i s t i c a l l y s i g n i f i c a n t (p < .05, Wilcoxon-Mann-Whitney test) provided the l o c a l 'population' l i k e l y to v i s i t the enclosure s i t e does not exceed the maximum census count (15 bi r d s , average count = 7 ) . Of course, t h i s observation does not demonstrate l o c a l enhancement; mergansers may be responding to other s t i m u l i independently of each other. Additional evidence for l o c a l enchancement was obtained from an experiment with a r t i f i c i a l flocks near the mouth of the L i t t l e Qualicum R. The proportion of mergansers f l y i n g overhead that landed i n the immediate v i c i n i t y of the decoys increased from 43% when no decoys were present, to 100% for a flock of 16 decoys. S i m i l a r 1 y , mixed flocks of decoys and mergansers attracted 62% of overhead f l i g h t s at fl o c k sizes £ 4, and 100% at flocks of 17 - 19. The p r o b a b i l i t y of landing was approximately l i n e a r i n flock s i z e using an arcsine-square root transformation for the proportions (r = .81, p < .01, F i g . 29). It i s possible that flocks composed ex c l u s i v e l y of r e a l b irds would have produced more dramatic r e s u l t s , or that foraging birds may be more a t t r a c t i v e than immobile ones. Even so, many birds appeared to be deceived by the decoys, at least i n i t i a l l y . 110 F i g . 29. Proportion of f l i g h t s overhead that landed at the study plot with respect to the size of an a r t i f i c i a l f lock already present. Dots - decoys only; c i r c l e s - mixed decoys and r e a l mergansers (depicted at midpoint of i n t e r v a l s ) . The regression i s l i n e a r for (arcsine) transformed proportions: y = 34.9 + 2.58x, n = 9, r = 0.81 Fl,17 = 13.2, p < 0.01 n = 57 flights overhead p <.01 112 Departures from the Enclosure Site: The average v i s i t at the enclosure site lasted 38.2 min (mode <10 min, maximum = 270 min) of which an average of 14.2 min (maximum = 109 min) were spent foraging (n = 138 with males of pairs excluded). Visits in the afternoon (11:00 - 16:00 h PST) were similar in duration to those in the evening (16:00 - 20:00 h, p > .10, Wilcoxon signed-rank test). Non-foraging activities predominated for v i s i t s > 10 min. Effect of Flock Size: Aggressive interactions seldom precipitated departures among mergansers at the enclosure site. Three classes of agonistic behaviour were distinguished: attempted piracy while foraging, chase and display behaviour and forced copulation. The frequency of occurrence of each behaviour and i t s effect on probability of departure is summarized with respect to flock size in Table IV. Attempts at piracy never caused the departure of either contestant and were initiated equally often by both sexes. Chases and displays were initiated by mature males on a l l but 1 of 69 occurrences; of these, most (85%) were initiated by paired males (p < .05, x 2 ) . Departure resulted from only 9% of interactions involving chase/display behaviour. Although the frequency of such encounters increased with the number of males present, the proportion leading to departure decreased; overall, the departure rate due to chase/display behaviour was independent of the number of males present and of total flock size. This may be due to greater site tenacity associated with higher pro f i t a b i l i t y that, in turn, accounts for the presence of the larger flock. The departure rate due to chase/display behaviour was low (0.24/h) 1 1 3 compared to the rate of departure i n the absence of aggression (1.4/h). In contrast, 54% of forced copulations caused departure of the female; a l l attempts were directed at s i n g l e , mature females. Forced copulations were no more frequent within larger flocks because single mature females were never observed in flocks with more than one mature male. However, there i s no evidence that single mature females avoided s i t u a t i o n s where forced copulations might occur - 42% of these incidents occurred when the female arrived at the enclosure s i t e while the aggressor was already present. Single males accounted for most attempts (76%, p < .05, x2) s and were more persistent than paired males. The p r o b a b i l i t y of departure was also independent of flock size i n the absence of agonistic i n t e r a c t i o n s . This conclusion i s based on two analyses that control for the e f f e c t of f i s h density ( i . e . p r o f i t a b i l i t y ) : f i r s t , a comparison of the average d a i l y departure rate for s o l i t a r y vs flocked birds (p > .10, Wilcoxon signed-rank t e s t ) , and second, a comparison of the duration of i n d i v i d u a l v i s i t s spent alone or mostly in company (p > .50, x2, Table V). In the l a t t e r t e s t , i t was necessary to account for an inherent bias due to the greater l i k e l i h o o d of another merganser a r r i v i n g during a long s o l i t a r y v i s i t which r e s u l t s in that v i s i t being r e c l a s s i f i e d as 'group' v i s i t . D e t a i l s accompany the table. E f f e c t of P r o f i t a b i l i t y : Because the a v a i l a b i l i t y of f i s h regression of v i s i t duration on smolt varied among the enclosures, the density ( F i g . 30) includes data for Table IV: Summary of aggressive interactions with respect to flock s i z e . PIRACY CHASE/DISPLAY FORCED COPULATION RATES Number Number Causing Time Number Number Number Number by Number Number Depart. Flock Observ. by Causing by Causing Unpaired Causing per h per h Size (min) Number Males Departure Number Males Departure Number Males Departure per Bird per Bird 2 1114 1 1 6 6 4 0.2 0.1 3 327 16 16 1 2 2 1 1.1 0.1 4 566 4 2 29 29 4 5 2 1.0 0.2 5 239 3 2 11 11 0.7 0 6 54 2 9 9 1 2.0 0.2 7 50 3 3 0.5 0 ALL 2350 10 4 0 69 68 6 13 8 7 0.8 0.1 115 Table V: Comparison of v i s i t duration when s o l i t a r y and in fl o c k with respect to p r o f i t a b i l i t y . If the presence of other birds does not a f f e c t v i s i t length ( H 0 ) , the d i s t r i b u t i o n of v i s i t durations for s o l i t a r y and flocked birds w i l l be s i m i l a r a f t e r c o r r e c t i n g for the l i k e l i h o o d of another b i r d a r r i v i n g during s o l i t a r y v i s i t s . H Q i s not rejected. FREQUENCY DISTRIBUTION OF VISIT LENGTHS P r o f i t a b i l i t y (time to f i r s t capture) Duration of V i s i t (min) A l l V i s i t s (flocked + s o l i t a r y ) Expected Number So l i t a r y a Observed Number S o l i t a r y Test Results < 2 min < 20.0 20 - 39.9 40 - 59.9 > 60.0 32 9 7 13 20.38 4.25 2.01 3.37 19 4 5 2 x2 = 5.2 3 df p > .10 Total 61 30.0 30 > 2 min < 20.0 51 2 0 - 3 9 . 9 5 4 0 - 5 9 . 9 7 >_ 60.0 11 T o t a l : 74 27.51 27 2.00 1 x2 = 1.6 2.07 2 3 df 2.41 4 p > .50 34.0 34 a The frequency for a l l v i s i t s m u l t i p l i e d by a cor r e c t i o n factor equal to the p r o b a b i l i t y of an a r r i v a l i n interva 1 ( t i , t 2 ) : t2 f ~ x co r r e c t i o n factor = IVe dx where

-o LU 1.0-0.8-0.6 a LU ^ 0.4 LL •no smolt c ap tu red n=25 >1 smolt cap tu red within f i r s t 30 s n = 41 p <.02 0.2-0 LJI H 0 -0.5 0.5-1.0 1.0-2.0 2.0-3.0 DURATION OF VISIT (h) DURATION OF VISIT ( h ) I Z I 122 DISCUSSION Patch Assessment: Mergansers demonstrated an a b i l i t y to respond to differences in p r o f i t a b i l i t y among the enclosures, despite the fact that prey were mobile and encountered at random. A l l o c a t i o n of search time to the most p r o f i t a b l e enclosure also increased over the 5 wk period. Very few b i r d s , however, foraged e x c l u s i v e l y on the most p r o f i t a b l e enclosure. This search pattern indicates a trade-off between the need to sample and the advantage of e x p l o i t i n g p r e v i o u s l y - p r o f i t a b l e areas. Extensive sampling i s adaptive where p r o f i t a b i l i t y changes in an unpredictable way (Smith 1974, Krebs and Cowie 1978). In natural s i t u a t i o n s , mergansers would seldom encounter t r a n s i t i o n s i n prey density as abrupt as i n these experiments; moreover, such concentrations of prey would be depleted r a p i d l y in nature whereas constant de n s i t i e s were maintained throughout t h i s study. It i s not s u r p r i s i n g , therefore, that mergansers continued to forage on the less p r o f i t a b l e enclosures. The evidence suggests that mergansers did not follow a GUT decision r u l e in t h e i r responses to prey density. GUT i n t e r v a l s from l a s t capture to departure from the enclosure, appear to have been randomly d i s t r i b u t e d and independent of f i s h density. GUT did not increase a f t e r mergansers had developed an expectation of higher p r o f i t a b i l i t y on the upper enclosure, as predicted by McNair (1982). S t i l l , mergansers were less l i k e l y to leave an enclosure, or a quadrant within an enclosure, at high f i s h density than at 1 2 3 low f i s h density. These observations indicate that mergansers exhibit a r e a - r e s t r i c t e d searching behaviour following encounters with prey. Indeed, random search, r e s t r i c t e d near the locus of the l a s t encounter with prey but not influenced by enclosure boundaries, would generate the observed (random) d i s t r i b u t i o n of GUT i n t e r v a l s . Mergansers do not seem to recognize the enclosure boundaries as defining d i s t i n c t patches. This i s understandable because the enclosure boundaries are highly a r t i f i c i a l - f i s h movement i s not r e s t r i c t e d within natural streams. Mergansers are perhaps more l i k e l y to i d e n t i f y gradients i n current, gravel texture or incident l i g h t i n t e n s i t y as patch boundaries.. In general, ascertaining what i s perceived as a 'patch' or a 'food item' i s a major d i f f i c u l t y i n applying optimization theory to f i e l d studies (Schluter 1981, Zach and Smith 1981). The foregoing example i l l u s t r a t e s that cues related to p r o f i t a b i l i t y are associated with v i r t u a l l y a l l foraging patches. To say that patches vary widely i n p r o f i t a b i l i t y i s to say that the perceived cues are inappropriate or u n r e l i a b l e . S i m i l a r l y , i f the p r o f i t a b i l i t y of patches i s high l y predictable, the cues defining a patch must be r e l i a b l e . Iwasa et a l . (1981) have demonstrated that the optimal decision rules for a l l o c a t i n g search time among various patches depends on the d i s t r i b u t i o n of prey among patches and on how much the forager knows about this d i s t r i b u t i o n . This i s equivalent to saying that the appropriate decision rule depends on the choice and r e l i a b i l i t y of perceived cues and on the 'prior model', i n a Bayesian sense, with which the forager interprets these cues. From t h i s viewpoint, i t i s clear that natural s e l e c t i o n w i l l operate to improve the a b i l i t y to u t i l i z e appropriate cues when making decisions. In 124 natural s i t u a t i o n s , foraging animals are inundated with p o t e n t i a l cues, some r e l i a b l e and some of only momentary usefulness. Hence, we should expect search strategies to be mixed and f l e x i b l e , to involve d i s t i n c t (optimal) choices with respect to some cues but not others. Foraging models generally predict behaviour more accurately and c o n s i s t e n t l y i n simple test s i t u a t i o n s than i n f i e l d studies where a wider v a r i e t y of cues are a v a i l a b l e (Zach and Smith 1981). Of course, hypotheses about the choice of cues and p r i o r model constitute part of the 'constraint set' i n foraging models, and as such, are subject to i n v e s t i g a t i o n by the optimization approach (Maynard Smith 1978). F a i l u r e to observe predicted decision-making behaviour i s usually taken as evidence that the costs and benefits of foraging behaviours have been misunderstood (Ollason 1980), or that the time frame f o r optimization i s inappropriate (Katz 1974). However, suboptimal behaviour may only i n d i c a t e that the test animals are u t i l i z i n g cues that are l e s s r e l i a b l e than those perceived by the experimenters (e.g. Dawkins and Brockmann 1980). A l t e r n a t i v e l y , the forager may be using a d i f f e r e n t p r i o r model by which i t int e r p r e t s these cues. No doubt, mergansers foraging on the experimental enclosures perceived the netting that delimited the patches (for a human observer); but i t would seem t h e i r p r i o r model of prey d i s t r i b u t i o n could not allow f o r r e s t r i c t e d movement of f i s h within a flowing stream. Why might animals use poor cues i f better ones are available? The value of using cues e f f i c i e n t l y to s e l e c t p r o f i t a b l e patches must be evaluated i n terms of the consequences of foraging without cues. E f f e c t i v e use of cues w i l l reduce the time and energy wasted while foraging i n areas of low prey density. However, at small s p a t i a l scales t r a v e l and sampling costs w i l l be small; a r e a - r e s t r i c t e d searching behaviour without cues w i l l be almost as 125 efficient as direct recognition of favourable spots. It is doubtful whether the benefits from the latter would offset the developmental costs associated with increased perceptual or cognitive a b i l i t i e s . At larger spatial scales where energy expenditure during search is substantial or where search time is at a premium, efficient use of cues w i l l become more advantageous. Similarly, Dawkins and Brockmann (1980) attempt to explain the 'misuse' of cues by digger wasps (Sphex ichneumoneus) in terms of 'game theory'. Arrival and Departure: Both previous experience and the presence of other mergansers influenced the frequency of v i s i t s to the enclosure s i t e . Arrival rates increased steadily during the smolt experiments, but decreased again during the less profitable t r i a l s with fry. This reinforcement of foraging site occurred slowly; thus, arrival rates w i l l be relatively unaffected i f prey density varies substantially in p r o f i t a b i l i t y over periods less than 1 wk. The local enhancement effect due to the presence of other mergansers is more important. The response appears to be linear in flock size at least for decoy and mixed flocks (i.e. decoys and birds) < 20. The duration of v i s i t s at the enclosure site varied greatly but was significantly influenced by p r o f i t a b i l i t y . V i s i t durations increased exponentially with prey density, and decreased exponentially with time-to-first-capture after arriv a l . The variation in length of v i s i t s was largely due to variation in the duration of non-foraging activity at the enclosure site. Presumably, this variation can be attributed to individual differences in the allocation of non-foraging time to other a c t i v i t i e s . 1 2 6 S o c i a l interactions at the s i t e had l i t t l e e f f e c t on the duration of v i s i t s , except those involving forced copulation. Agonistic encounters while foraging were infrequent and b r i e f ; very few resulted i n departure of one of the p a r t i c i p a n t s . Moreover, there i s l i t t l e evidence that mergansers experienced decreased p r o f i t a b i l i t y while foraging i n groups (Chap. 2). Discounting aggression, the duration of v i s i t s was s i m i l a r for both s o l i t a r y birds and those in f l o c k s . O v e r a l l , the p r o b a b i l i t y of departure was independent of flock size over the range of flock sizes observed. Evidently, mergansers follow decision rules for departure independently of one another. (A breeding pair has been considered as one ' i n d i v i d u a l ' i n t h i s context). This i s a fortunate r e s u l t , for i t greatly s i m p l i f i e s the task of extending the model of patch choice from the l e v e l of i n d i v i d u a l to the l e v e l of population (Chap. 4). The majority of birds fed to s a t i a t i o n and departed many minutes a f t e r they had ceased foraging. Yet c l e a r l y , the duration of v i s i t s (including both foraging and non-foraging a c t i v i t y ) was related to f i s h density. What, then, prompted mergansers to leave the enclosure s i t e ? The decision rule for departure i n th i s context cannot have been a simple one such as those often proposed (e.g. GUT, residence time or y i e l d s t r a t e g i e s , see McNair 1983) because foraging a c t i v i t y had already ceased. Motivation to leave the s i t e was probably a complex balance of c o n f l i c t i n g responses to p r o f i t a b i l i t y experienced at the s i t e r e l a t i v e to other s i t e s , degree of s a t i a t i o n and motivation to engage in non-foraging a c t i v i t i e s elsewhere. Because each factor w i l l have d i f f e r e n t p r i o r i t y among ind i v i d u a l s with d i f f e r e n t experience, i t i s not su r p r i s i n g that time spent at the enclosure s i t e varied as much as i t did. Even so, i t i s clear from F i g . 28, 30 and 31 1 2 7 that the memory of past foraging success (or perhaps expectation of future success) influences the decision to leave and to return to a particular site. I conclude that at this larger spatial scale, mergansers assess the p r o f i t a b i l i t y of a foraging patch relative to alternative sites. Simple, mechanistic decision rules do not describe the birds' response adequately at this scale. Nevertheless, the typical response to p r o f i t a b i l i t y at larger spatial scales can be described and related to the dispersion of mergansers within a population. This is the subject of the following chapters. 128 PART I I DISPERSION I N RELATION TO THE A V A I L A B I L I T Y OF J U V E N I L E SALMONIDS 129 CHAPTER 4: THE AGGREGATIVE RESPONSE INTRODUCTION The search behaviour of individual predators is well adapted to exploit local concentrations of their prey (e.g. Smith and Sweatman 1974, Chap. 3). Consequently, predators that forage independently of each other should accumulate at favourable feeding sites; ample evidence supports this expectation (e.g. Rusch et a l . 1972; Readshaw 1973, O'Connor and Brown 1977, Phelan et a l . 1978, Bell 1980, Village 1982). This relationship between the dispersion of prey and predator populations, is referred to as an 'aggregative response' (Readshaw 1973). For species that feed on valuable crops or game, there is much practical incentive to understand factors that govern the aggregative response and thus, the intensity of predation in specified areas. The common merganser poses a serious threat to salmonid fisheries, but only when i t is abundant on the freshwater reaches of salmonid-producing streams or lakes; elsewhere, mergansers do l i t t l e damage (Salyer and Lagler 1940; White, 1957; also Chap. 6). The objective of this chapter is to identify conditions that lead to undesirable concentrations of mergansers in salmonid-rearing areas. It is hoped that, by understanding when and how aggregations arise, efforts can be made to curtail their size and duration, and hence, to reduce their impact on the prey population. Two steps are involved in the investigation: f i r s t , i t i s essential to determine the size of the effective population - the number of mergansers within ranging distance of the site in question; second, the aggregative response must be described in 130 terms of changes in p r o f i t a b i l i t y at the s i t e . The e f f e c t i v e population size depends both on the density of birds in the v i c i n i t y and on the distance they venture each day i n search of food. As with most waterfowl, merganser populations undergo dramatic seasonal changes in abundance and sex composition associated with non-feeding behaviour (e.g. moults, migration and reproduction). Consequently, the e f f e c t i v e population size w i l l depend on the season. D a i l y and t i d a l cycles might also govern the timing of feeding a c t i v i t i e s and thereby a f f e c t ranging distance. Daily and seasonal trends in abundance and composition are examined in the f i r s t section of t h i s chapter; t y p i c a l ranging patterns are inferred from f l i g h t and foraging a c t i v i t y and from the movements of marked mergansers. The simplest approach to modeling the aggregative response i s to assume that each merganser forages independently on the basis of i t s own information and decision rules (Chap. 3). This i s the essence of the equilibrium f l o c k - s i z e (EFS) model proposed by Krebs (1974) whereby birds accumulate at a favourable s i t e u n t i l the expected number of a r r i v a l s i s balanced by those expected to leave. Under the model, the decision to j o i n or desert a flock i s made independently by i n d i v i d u a l birds or perhaps small groups of i n d i v i d u a l s such as breeding pairs - whatever units are appropriate for measuring group s i z e . This does not mean that decisions are independent of flock size or other information conveyed by the group. In f a c t , the e f f e c t s of s o c i a l i n t e r a c t i o n s or l o c a l enhancement are accommodated e a s i l y i n the following general form of the model: = a(F)(N-F) - d(F,P)F [4.1] dt 131 where F represents f l o c k s i z e , N the e f f e c t i v e population, P the p r o f i t a b i l i t y at the s i t e and where a and d are functions defining the p r o b a b i l i t i e s of a r r i v a l and departure r e s p e c t i v e l y . Just as the functional response to prey density depends on the p r o f i t a b i l i t y of al t e r n a t i v e prey, the aggregative response to a s p e c i f i e d s i t e w i l l depend on the q u a l i t y of a l t e r n a t i v e s i t e s . In other words, [4.1] is not a repeatable function unless the a l t e r n a t i v e s i t e s are also s p e c i f i e d . It i s possible, i n p r i n c i p l e , to extend the model to many areas, although to my knowledge t h i s has not been attempted. The simple form has been applied with considerable success for feeding flocks of great blue herons (Ardea herodias, Krebs 1974) and house sparrows (Passer domesticus, Barnard 1980) and stochastic versions were evaluated for yellow-eyed junco flocks (Junco phaeonotus, Caraco 1980). Because mergansers spend r e l a t i v e l y l i t t l e time foraging, the EFS model cannot be applied s t r i c t l y to foraging f l o c k s ; rather, I assume that p r o f i t a b i l i t y i s assessed while foraging, but that departures depend on recent foraging success, and are equally l i k e l y , whether the birds are foraging or not (Chap. 3). Qu a l i t a t i v e predictions of the EFS model are evaluated i n r e l a t i o n to merganser f l o c k - s i z e d i s t r i b u t i o n s . The common merganser also e x h i b i t s an aggregative response to enhanced downstream migrations of juv e n i l e salmonids. The response i s evident through increased abundance on enhanced streams, as compared to neighbouring streams, and through increased d e n s i t i e s on freshwater as opposed to t i d a l waters within a single r i v e r system. Accordingly, the EFS model has been adapted to predict the size of aggregations ( i . e . c o l l e c t i o n s of merganser flocks) on a small r i v e r system. 1 3 2 The predicted trends in aggregation which track juvenile salmonid densities, are compared with observed trends for two streams with salmonid enhancement f a c i l i t i e s . 1 3 3 METHODS Study Area: Eight salmon-producing streams on the east coast of Vancouver Is. were censused r e g u l a r l y during 1980 - 1982. The location and relevant s t a t i s t i c s for each are presented in F i g . 2 and Table VI. Four neighbouring streams -Englishman R. (ENG), L i t t l e Qualicum R. (LQ) , Big Qualicum R. (BQ) and Rosewall Creek (RC) were studied more i n t e n s i v e l y than the others during A p r i l to August. Salmon hatcheries are situated on the l a t t e r three and scheduled f i s h releases from these provide an opportunity to monitor the responses of mergansers to repeated, large-scale perturbations of juvenile salmonid density. R e l i a b l e data on downstream migration of wild f i s h i s a v a i l a b l e from counting fences on the BQ and Quinsam r i v e r s , and for a previous year (1979) on LQ ( L i s t e r , D.B. & Assoc. 1979); rough estimates of wild production were also calculated from annual spawner counts (Marshall et a l . 1977) averaged over the past 30 yr using conversion factors (eggs per spawner, surviv a l of eggs to f r y or smolt) from L i s t e r , D.B. & Assoc. (1979) and Minaker et a l . (1979). I consider that estimates from spawner counts provide a s a t i s f a c t o r y index, given that d e n s i t i e s of j u v e n i l e salmon i n streams with hatcheries are 2 - 3 orders of magnitude greater than those in unenhanced streams. Census Method: Censuses were undertaken on foot by two observers (with binoculars) working simultaneously in d i f f e r e n t zones of a single r i v e r system to cover Table VI: Summary of s t a t i s t i c s pertaining to the size and juvenile salmonid production of r i v e r s censused, 1980 - 1982. JUVENILE MIGRATION Area (km2) of (JSI xlO 6) River System Drainage Area (km2) Length (km) Accessible to Salmon 3 I n t e r t i d a l Within 1 km of Mouth Yr Wild Spawnersb (xl O 3 ) A l l Fry Smo 11 and March - May Fry Total Quinsam-Lower Campbell 0 2000 d 25.6 1.4 82 12.4 8.0 24.4 28.0 Englishman 283 16.0 0.8 81 2.5 3.0 3.8 4.0 82 3.8 4.0 L i t t l e Qualicum 0 246 11.5 1.1 80 6.6 20.0 20.4 27.0 81 9.0 9.6 17.0 Big Qualicum 0 150 e 10.4 0.6 80 6.5 67.0 74.9 93.0 81 60.0 87.3 95.0 82 44.0 55.6 76.0 Tsable 107 5.1 1.4 80 0.8 3.0 3.0 3.0 81 3.0 3.0 Rosewall Creek 0 45 3.8 1.5 80 0.4 0.8 1.8 2.5 81 0.9 1.6 Wilfred Creek 33 3.8 0.9 82 0.5 0.5 0.5 0.6 Nile Creek 18 3.9 0.4 80 0.3 0.2 0.2 0.2 a Marshall et_ a l . 1977 k index of wild smolt migration (30 yr avg. for coho, chinook and steelhead spawners, Marshall et_ al_. 1977) 0 streams with hatchery d drainage area below dam i s 260 km2 e drainage area below dam i s 44 km2 135 the designated area as quickly as possible. During 1980, attention was focussed on the estuary and very lowest reaches of freshwater where mergansers were most common. BQ was censused r e g u l a r l y to 5 km, and occa s i o n a l l y , to 10 km upstream of the r i v e r mouth; LQ was censused to 2.5 km and RC, to 1 km upstream. More extensive t r a i l s were cut in 1981 to allow regular coverage to 5 km upstream on BQ, LQ, and ENG, to 10 km on the Quinsam-Lower Campbell (Q-LC) system i n 1982, and to 1 km on RC, Wilfred Creek (WC; 1982 only) and Tsable R. (TS). Censuses were .conducted at i n t e r v a l s of < 3 d during l a t e A p r i l - August of 1980 at various combinations of tide and time of day. During 1981, counts were made every 4 days, A p r i l through July, and less frequently i n other months; weekly censuses were conducted from June to September in 1982. Each census required 2 - 3 h. During 1981, census times were scheduled to provide maximum contrast between time of day and tide e f f e c t s since the semi-diurnal tides follow a 2 wk c y c l e . Mergansers are e a s i l y counted; they are conspicuous by colour and habit, v o c a l i z e when alarmed, thereby a l e r t i n g the observer, and yet, are nor e s p e c i a l l y wary. The greatest source of error i n counts arises from birds on the wing which may or may not have been counted already. Fortunately, only a small proportion of birds were recorded as f l y i n g during a t y p i c a l census (< 5%). The following convention was adopted to resolve uncertain cases: birds f l y i n g into previously censused t e r r i t o r y , from without, were recorded whereas those f l y i n g out of previously censused areas were not counted, and subtracted i f others of matching sex and maturity were seen l a t e r in the d i r e c t i o n of t r a v e l . This procedure ensured that abundance was not - overestimated. Records were kept of sighting l o c a t i o n to within 0.2 km, number and size of fl o c k s , sex r a t i o , p a i r i n g and maturity status (where 1 3 6 p o s s i b l e ) , a c t i v i t y and the presence/absence of bands or any recognizable marks. A c t i v i t y categories included foraging (submerging head i n search of prey or consuming prey), preening, i n t e r a c t i n g with conspecifics (display, chase or copulation), sleeping, r e s t i n g ( a l e r t but inacti v e ) and f l y i n g . In the discussion that follows, statements about foraging or f l y i n g a c t i v i t y r e f e r to the proportion of birds that were engaged i n the s p e c i f i e d a c t i v i t y during the census. A flock was defined as a group of birds with a t y p i c a l nearest-neighbour spacing < 10 m. Using t h i s scale as a guideline, flocks were usually obvious; a s i n g l e , large flock was recorded wherever birds were very active so that spacing changed r a p i d l y (e.g. during feeding f r e n z i e s ) . Marking Method: Twenty-one mergansers were banded and marked i n d i v i d u a l l y with p i c r i c acid s t a i n (1980) or colour-bands (1981). The birds were captured in size 15 mist nets (10 cm mesh) suspended across the stream. A l l but one were captured and released on BQ. Crop and g u l l e t contents were obtained from 17 of 20 birds to which a 1% aqueous solution of antimony potassium t a r t r a t e had been administered at dosages of 10 cc/kg as described by Prys-Jones et a l . (1974). Birds marked with s t a i n i n 1980 did not r e t a i n t h e i r marks past the wing moult i n l a t e summer and could not be recognized i n d i v i d u a l l y i n the f i e l d i n subsequent years. Index of Juvenile Salmonid Density (JSI): For analysis of the aggregative response, i t was necessary to rank 1 3 ? releases of juvenile salmonids in order of p r o f i t a b i l i t y . This is straight-forward where a l l fish released are of similar size. However, where several different salmonid species and size groups are released (e.g. BQ), some assumption about relative p r o f i t a b i l i t y is required. A simple pooled estimate of density would underestimate the contribution of larger fish (up to 90 g) in comparison to emergent fry (0.3 - .4 g); on the other hand, a pooled estimate of biomass would probably overestimate the availability of the larger fish. Capture rates of mergansers feeding on 2 g coho fry and 40 g coho smolt have been evaluated in earlier experiments (Chap. 2). Coho smolt were taken more frequently than fry but the size-selection could be explained on the basis of an apparent size model (Chap. 2): [4.2] where a^ a n < j a^ a r e the rates of successful search for coho fry and smolt and 1^ and 1 are the lengths of fry and smolt respectively. A juvenile salmonid index (JSI) was computed assuming size-selection is described adequately by [4.2] as follows: [4.3] where n£ is the number and 1^ the length of salmonid type i and l£, the average length of an emergent chum fry (3.8 cm). Unfortunately, the model, [4.2], could not be tested rigorously; the data available indicate af/a s > ( l f / l s ) 2 [vis. af/a s = .37, ( l f / l s ) 2 = .13] which implies that the JSI may 138 exaggerate the importance of large f i s h . The r e l a t i o n s h i p between se l e c t i o n by density, biomass and apparent size i s i l l u s t r a t e d i n F i g . 59. Species differences are ignored. Flock-Size D i s t r i b u t i o n s : Additional data pertaining to f l o c k formation and f l o c k - s i z e d i s t r i b u t i o n s were obtained during two consecutive experiments, one on RC during May 4 - 2 2 , 1981 and another at the mouth of LQ on June 10 - 26, 1981. During the f i r s t experiment, stream enclosures were stocked with known dens i t i e s of coho smolt (Chap. 2). The number of mergansers at the enclosure s i t e was observed from a b l i n d and recorded at 20 min i n t e r v a l s following a 3 h e q u i l i b r a t i o n period. As foraging bouts were t y p i c a l l y < 10 min, each census was treated as an independent r e p l i c a t e of the flock-formation process. In the second series of observations, flock sizes at the mouth of the LQ r i v e r were recorded at 20 min i n t e r v a l s from 10:30 - 13:00 h and 14:00 -16:30 h on June 10 - 12, 17, 19, and 24 - 26. Overall abundance on the estuary could be recorded simultaneously from a raised b l i n d . A breeding pair i s treated as a single ' i n d i v i d u a l ' in the following discussions of aggregation and flock size because members of the pair usually t r a v e l together; i n c l u s i o n of both members would exaggerate the response, and v i o l a t e the assumption of independence in the EFS model. The t y p i c a l flock size (TFS) i s calculated as £ x i 2 / £ x i where x i i s the size of the i t h f l o c k observed. It represents the f l o c k size i n which the 'average' i n d i v i d u a l occurs (jarman 1974). The parameter, k, of negative binomial 139 distributions was estimated iteratively such that X n i l o s e I 1 + — ) = 2 & [ 4 - 4 ] k / 1 k+n. I where n^ is the observed frequency of flock size i , and x is the mean flock size (Southwood 1978). For truncated negative binomial distributions, k = [ p x - f*(l)] / (1 - p) [4.5] where p = x [1 - f*(l)] / S2 [4.6] and f * ( l ) is the relative frequency of solitaries and s2 is the sample variance of flock size (Brass 1958). 140 RESULTS Trends in Dispersion Unrelated to Salmonid Density Population Density and Composition: Mergansers were most abundant i n the study area throughout March and A p r i l just a f t e r the peak period of northerly migration (Bellrose 1978). Trends in abundance were generally s i m i l a r from year to year within streams although they d i f f e r e d among streams ( F i g . 32). Most day to day v a r i a t i o n in estimates of abundance i s probably due to movement of mergansers to and from the census area; I believe that only a minor proportion of t h i s v a r i a t i o n can be at t r i b u t e d to sampling error. The r e c i p r o c a l trends in abundance i n F i g . 32 are associated with differences in salmonid production among the three major r i v e r s (see below). Accordingly, I pooled data for these r i v e r s to reveal o v e r a l l trends corresponding to seasonal breeding, moulting and migratory a c t i v i t i e s ( F i g . 33). The combined census represents the great majority of mergansers within a 20 km radius of the LQ estuary. Common mergansers nest near streams or lakes and often use c a v i t i e s in trees or undercut banks (Munro and Clemens 1937). The breeding pairs form early i n the spring; egg-laying begins between early A p r i l and early June (Bellrose 1978, Chap. 5) and continues for about two weeks for an average-sized clutch (Erskine 1972). The pairs break up as soon as incubation begins and the males leave the v i c i n i t y to moult on protected estuaries or lakes (Bellrose 1978). This accounts for the change in sex r a t i o during June 141 Fig. 32. Trends in abundance of mergansers on 3 adjacent river systems (tidal and freshwater zones). A - Englishman; B - L i t t l e Qualicum; C - Big Qualicum. Circles - 1980; dots - 1981; half-dots - 1982. Freshwater census of L i t t l e Qualicum in 1980 to 1 km upstream only; a l l other counts to 5 km upstream. 00 Q or CD LL O or LU CD 1 0 0 -5 0 -0 -o 1 9 8 0 • 1 9 8 1 » 1 9 8 2 5 0 -0 -1 5 0 -1 0 0 -5 0 -\ m » 0 0_° / \ t/3" * — « ^ J»^*0 < V V o °oo 0 o °ooO° oo 0 o/ o o\ / V o o oo% 0- T 1 1 r MARCH APRIL o1-o o o o o o o MAY B c T 1 1 1 1 1 T1-1 1 1 r JUNE I JULY I AUGUST 143 ( F i g . 33). Moulting males were r a r e l y observed at any of the study s i t e s , but most (22) were sighted on the extensive and sheltered t i d a l f l a t s at the mouth of TS. Downy young emerged between late May and August and tended to remain on or near t h e i r natal stream u n t i l able to f l y at about 10 weeks of age. The pooled estimate of abundance declined s t e a d i l y from A p r i l through June. This seasonal decline begins well before the pairs break up (June) and presumably r e f l e c t s the pattern of northerly migration. The subsequent increase in abundance i s due, in part, to the i n c l u s i o n of juvenile b i r d s , > 50 d of age, that are not e a s i l y distinguished from adult female and immature b i r d s . However, females and immatures (distinguished by t h e i r a b i l i t y to f l y ) also became more numerous during July and August. Flocks of mixed adult and j u v e n i l e mergansers were c h a r a c t e r i s t i c a l l y large and cohesive during t h i s period. Census counts were not correlated with either time of day, t i d e height or cloud cover (p > .25, ANOVA). Ranging Behaviour: F l i g h t a c t i v i t y was much reduced in June compared with previous months (p < .05, x2 } F i g . 34). F l i g h t s were even less frequent during J u l y and August, but t h i s may be because juveniles and moulting birds could not always be distinguished in the censuses. No seasonal trends were evident with respect to foraging a c t i v i t y . Both f l i g h t a c t i v i t y and foraging a c t i v i t y varied widely among censuses. S u r p r i s i n g l y , no s i g n i f i c a n t influence could be a t t r i b u t e d to time of day, tide height or cloud cover whether weighted by census or number of b i r d s . 144 F i g . 33. Seasonal trends i n abundance, p a i r i n g , sex r a t i o and brood emergence of mergansers during 1981. Counts from Big Qualicum, L i t t l e Qualicum and Englishman r i v e r s are combined. Juveniles (> 50 d) are included in counts aft e r June 7 (dashed) v e r t i c a l l i n e ) ; no birds < 50 d are included. NO. BROODS HATCHING 146 Fig. 34. Seasonal trend in flight activity of mergansers in 1981. Counts from Big Qualicum, L i t t l e Qualicum and Englishman rivers are combined. Numbers refer to numbers of census days; arrows indicate means. Trend is significant (p < 0.05, x2, 3 df). PROPORTION FLYING 148 Foraging a c t i v i t y i s summarized in r e l a t i o n to tide height and time of day in F i g . 35; the corresponding figure for f l i g h t a c t i v i t y ( F i g . 36) pertains only to the period from March to May in l i g h t of decreased a c t i v i t y during the summer months. Movements of Marked Mergansers: Of 19 adults banded i n 1980 and 1981, at least 12 were resighted on a t o t a l of 53 occasions. Almost h a l f (45%) of resightings occurred more than one month aft e r release. Two i n d i v i d u a l s were resighted a f t e r 697 and 741 d at large. In both 1981 and 1982, one brood on BQ was accompanied by a female (or possibly a d i f f e r e n t b i r d each year) banded on the same stream i n 1980. Movements of i n d i v i d u a l birds are summarized in Table VII. At least 80% of marked birds resighted i n 1980 and 43% in 1981 were observed to v i s i t neighbouring r i v e r systems between LQ and RC. Several birds banded i n May -June of 1980 were observed to make t r i p s back and forth between adjacent streams. Despite more extensive census coverage and more frequent resightings, no birds banded in A p r i l of 1981 were observed to make return t r i p s to and from BQ. Moreover, these birds appeared to l i n g e r on the BQ system for a longer period a f t e r release before venturing to neighbouring streams. In the following section, i t i s argued that due to salmonid enhancement p r a c t i c e s , the BQ system i s e s p e c i a l l y a t t r a c t i v e to mergansers e a r l y in the season during the chum f r y (Oncorhynchus keta) migration when i n t e r t i d a l species (e.g. Cottidae spp) are r e l a t i v e l y less p l e n t i f u l . I f t h i s i s true, the r e s t r i c t e d movement of marked birds during A p r i l , 1981 i s e a s i l y explained. 149 F i g . 35. Foraging a c t i v i t y i n r e l a t i o n to time of day and t i d e height at Rosewall Creek and B i g Qualicum R, March - November, 1980 and 1981. White -low t i d e (< 2.5 m); s t i p p l e - high t i d e (> 2.5 m). Arrows i n d i c a t e estimates based on fewer than 100 b i r d s . 150 CD < or o LL 0.6 0.4 0.2 0 or o o_ o or o_ 0.6 0.4-0.2-R 0 S E W A L L n = 1797 b i rds i 1 BIG QUAL ICUM n = 3 5 6 8 0 0400 0700 1 1000 1 1300 1600 1900 2200 T IME OF DAY 151 Fig. 36. Flight activity in relation to time of day and tide height at Rosewall Creek and Big Qualicum R, March - May 1980 and 1981. White - low tide (< 2.5 m); stipple - high tide (> 2.5 m). Arrows indicate estimates based on fewer than 50 birds. 1 5 2 CD LL 0.3 0.2 0.1 0 or o o_ o or Q_ 0.3-0.2-0.1-0 ROSEWALL n = 1139 birds BIG QUALICUM n = 2427 v f , ' •i*S I •>.o, 1 0400 0700 1000 1300 1400 1900 2200 TIME OF DAY 1 5 3 T a b l e V I I A : Summary o f r e s i g h t i n g s o f me rgan se r s banded i n 1980. A s t e r i s k s i n d i c a t e s i t e o f r e l e a s e on day 0. BAND DATE OF NUMBER SEX RELEASE RC LOCATION RESIGHTED AND DAYS ELAPSED SINCE RELEASE NC BQ LQ 101 M May 5 1 3 a < r * , 6 - -+7 102 F/I May 5 o r 108 June 13 397 <-* , 3 5 8 103 104 F/I May 30 F/I May 30 -•13 36 ,35 ,22 * 1 * *19 106 F/I June 1 13,11 « -+ 1,5 103, 104 F/I see o r 106b above 276,313 • 3 6 2 , 3 7 0 , 3 7 6 r - 4 1 4 c < 1 U-74ld Number o f R e s i g h t i n g s ( t o t a l = 21) Number Banded i n 1980: 8 a d u l t s , 2 j u v e n i l e s Number o f I n d i v i d u a l s R e s i g h t e d : <6 a d u l t s , 0 j u v e n i l e s % R e s i g h t e d O u t s i d e BQ A r e a : 80 - 100% a p o s s i b l e c o n f u s i o n w i t h number 107 banded June 2 on BQ b i n d i s t i n g u i s h a b l e a f t e r f i r s t m o u l t c w i t h b r o o d , 1981 d w i t h b r o o d , 1982 1 5 4 Table VIIB: Summary of resightings of mergansers banded i n 1981. Asterisks i n d i c a t e s i t e of release on day 0. BAND DATE OF NUMBER SEX RELEASE RC 109 114 115 116 117 120 122 F M F F A p r i l 8 A p r i l 8 A p r i l 8 A p r i l 11 A p r i l 11 A p r i l 23 A p r i l 23 LOCATION RESIGHTED AND DAYS ELAPSED SINCE RELEASE BQ *,1,5,9,13, 17,21,37,41, 57,319,375 *,1,13,19,23 *,1,5,9,29 *,6,10,26, 59,70 *,2a -•59 7 (shot) -*48,50,55b> 63,69 Number of Resightings ( t o t a l = 32) 25 Number Banded i n 1981: 11 adults, 0 juve n i l e s Number of Individuals Resighted: 7 adults % Resighted Outside of BQ Area: 43% a possible confusion with number 117 h possible confusion with number 116 1 5 5 Dispersion In Relation To Juvenile Salmonid Density Reciprocal Movements Between River Systems: D i f f e r i n g trends in abundance among neighbouring streams evident in F i g . 32 can be explained to a large extent, by differences in j u v e n i l e salmonid d e n s i t i e s . Reciprocal r e l a t i o n s h i p s become obvious when the proportion of the pooled census (BQ, LQ, and ENG) observed on each of the 3 contributing streams i s plotted against the juvenile salmonid index (JSI) for BQ ( F i g . 37). Although BQ i s the smallest of the 3 streams in terms of drainage area, flow and length accessible to salmon, i t s annual production of salmonids i s 2 - 5 times greater than LQ - the next most productive system. In t h i s a n a l y s i s , f i s h d ensities have been summarized by month because precise information i s not a v a i l a b l e regarding stream residence times for smolt released from the hatcheries. Females with broods were excluded since t h e i r a c t i v i t i e s are confined to areas accessible to the f l i g h t l e s s broods, and the i r i n c l u s i o n would tend to obscure any movements between r i v e r systems. Reciprocal Movements between Fresh and T i d a l Waters: The proportion of birds observed on freshwater as opposed to t i d a l waters i s also correlated with the salmonid index within each of the 3 larger r i v e r s studied ( F i g . 38). Juvenile salmonids are r a r e l y consumed by mergansers on salt-water but comprise a large proportion of th e i r d i e t on freshwater reaches of salmonid-producing streams and lakes (White 1957, pers. observ.). It is highly probable, therefore, that the r e l a t i o n s h i p s i n F i g . 38 indicate an aggregative response to juvenile salmonids within r i v e r systems. An 1 5 6 F i g . 37. Reciprocal r e l a t i o n s h i p s in merganser abundance on t i d a l and freshwater zones of 3 neighbouring systems with respect to juvenile salmonid migration from Big Qualicum R. Numbers re f e r to number of census days; 95% C.I. are given for normal d i s t r i b u t i o n of arcsine-transformed data; regressions are li n e a r for transformed data: BQ y = 2.7 + 0.627x, n = 31, r = 0.81 F l , 2 9 = 5 7- 3> p < ° - 0 0 1 LQ y = 45.9 = 0.463x, n = 31, r - -.74 F l , 2 9 = 3 5' 2> P < 0 , 0 0 1 ENG y = 31.0 - 0.175x, n = 31, r = -.38 F i 9 q = 4.8, p < 0.05 \ 157 .5 .5 .5 0 4 r -BIG QUALICUM p < .001 LITTLE QUALICUM p < .001 i r. ENGLISHMAN p < .05 L " $ — 0 20 40 JUVENILE SALMONID MIGRATION IN BIG QUALICUM. 1981 (monthly tota l , JS I * 10 6)~ 1 5 8 F i g . 38. Relationships between the proportion of mergansers on freshwater and ju v e n i l e salmonid migration by month, March - June. Freshwater census coverage li m i t e d to 2.5 km above r i v e r mouth i n order to include 1980 data from L i t t l e Qualicum R. Numbers re f e r to number of census days; note log scale on abscissa. 95% C.I,. are given for normal d i s t r i b u t i o n of arcsine-transformed data; regressions are l i n e a r for transormed data: BQ y = 44.73 + 9.36 ( L o g 1 0 j s i X 10 5), n = 61, r = 0.30 F l 59 = 5 ' 9 2 ' P < 0 , 0 2 5 LQ y = 4.78 + 18.30 ( L o g ^ JSI X 10 5), n = 33, r = 0.49 F l 31 = 9 - 9 4 , p < ° - 0 0 5 ENG y = 22.3 + 13.16 ( L o g ^ j s i X 105), n = 29, r = 0.55 F, , 7 = 11.48, p < 0.005 BIG QUALICUM n = 61 p<.025 (1980-83) LITTLE QUALICUM (1980-81) n = 33 p <.005 1 1 • I 1 1 r-r-TI 1 1 1 1 1 1- iT] 1 1 1 1 •> I I • I 0.1 0.5 1.0 5 10 50 100 JUVENILE SALMONID MIGRATION (month ly t o t a l , JSI * 10 6) 160 a l t e r n a t i v e i n t e r p r e t a t i o n might be that the c o r r e l a t i o n arises merely because nesting occurs on freshwater and happens to coincide with the downstream migration of salmonids. This i s untenable, however, because during the periods of peak downstream migration ( A p r i l - May), the majority of the merganser population present does not breed ( F i g . 33) - p a r t i c u l a r l y as each breeding pair counts as only one i n d i v i d u a l i n t h i s a n a l ysis. P r e d i c t i n g Flock Size: The equilibrium f l o c k - s i z e (EFS) model predicts that, on average, flocks w i l l be larger on a more p r o f i t a b l e patch than on a less p r o f i t a b l e one provided s u f f i c i e n t time has elapsed for equilibrium to be achieved. If a r r i v a l s and departures occur at random, the EFS w i l l also be a random v a r i a b l e . Caraco (1980) has shown that where the p r o b a b i l i t y of a r r i v a l i s a l i n e a r function of flock size and the p r o b a b i l i t y of departure i s constant, the EFS follows a negative binomial d i s t r i b u t i o n . In terms of equation [4.1], i f the expected value of the a r r i v a l function, E[a(F)] = a + 0F, and the expected value for the departure function E[d(F,P)] = 7 at constant p r o f i t a b i l i t y (P), where a , /? , and 7 are p r o b a b i l i t i e s for a Poisson process, then the EFS has a negative binomial d i s t r i b u t i o n with parameters = a / ( 7 - 0 ) and k = a/p . F l o c k - s i z e d i s t r i b u t i o n s observed at the enclosure s i t e on RC are presented in F i g . 39 with respect to the maximum density of smolts available each day. As expected, flock size was, on average, s i g n i f i c a n t l y larger when 100 - 200 smolt were stocked than when a maximum of 10 - 50 were av a i l a b l e (p < .001, x2). Also, the f l o c k - s i z e d i s t r i b u t i o n s were not s i g n i f i c a n t l y 161 F i g . 39. Comparison of observed and expected d i s t r i b u t i o n s of equilibrium flock sizes with respect to coho smolt density stocked i n enclosures at Rosewall Creek, 1981. Black bars - observed frequency; white bars - expected frequency under corresponding negative binomial d i s t r i b u t i o n . A - low smolt density (max. 10 - 50, x flock = 0.83, k = 16.1, p > 0.05); B - high smolt density (max. 100 - 200, x flock = 1.92, k = 2.50, p > 0.10). Flock censused at 20 minute i n t e r v a l s a f t e r 3 h e q u i l i b r a t i o n period. Open arrows indicate means, s o l i d arrows, the t y p i c a l flock s i z e . Difference between dens i t i e s i s s i g n i f i c a n t (p < 0.001, x2). 162 2 4 F L O C K S I Z E 163 different from those under the corresponding negative binomial distributions (p > .05, and p > .10, x 2, for low and high smolt density repectively). It is doubtful, however, that the probability of arrival was a s t r i c t l y linear function of flock size at RC. F i r s t , the merganser population at RC is small in comparison to that at LQ; a maximum of 15 mergansers was observed within 1 km of the enclosure site during the experiments. Unless mergansers from farther afield were l i k e l y to v i s i t the si t e , the probability of arrival would have been reduced ( i . e . curvilinear) as flock size increased (maximum observed = 6). Second, arrival rates in the absence of a flock, increased gradually over the course of the experiments as a result of experience (Chap. 3, Fig. 28). To some extent this would merely compensate for the seasonal decline in abundance expected over the period May 4 - 24. The signficance of these deviations from linearity i s not clear, but they could be expected to spread out the EFS distribution. Smolt density treatments were alternated randomly throughout the experimental period so that any trends in arrival rates would not invalidate that comparison. Flock-size distributions at the mouth of LQ during mid- to late June are summarized by week in Fig. 40. The availability of prey (mostly Cottidae) at the site i s unknown but assumed to be comparable over the 16-day period. Tidal action may have influenced prey ava i l a b i l i t y but observation periods were scheduled 6 h apart so that both phases of the tide would be monitored. Moreover, tide action was virtually identical during the f i r s t (June 10 - 12) and third period (June 24 - 26), but reversed during the second (June 17, 19). Wind and cloud cover did not differ appreciably in any of the 3 periods. In contrast to the RC experiment, merganser density was high at LQ and arrival probabilities were close to linear, as illustrated by an 164 Fig. 40. Comparison of observed and expected distributions of equilibrium flock sizes at the mouth of the L i t t l e Qualicum R, 1981. Black bars -observed frequency; white bars - expected frequency under corresponding negative binomial distribution. A - June 10 - 12 (n = 89, x flock = 2.57, k = 0.87, p > 0.25); B - June 17, 19 (n = 50, x flock = 2.16, k = 1.40, p > 0.25); C - June 24 - 26 (n = 120, x flock = 5.09, k = 0.26, p < 0.001). White arrows indicate means, black arrows, the typical flock size. A 40-20" 0 1 J U N E 10 -12 • observed • expec ted n=89 T - N B t e W i ) x = 2.57 1 k = 0.87 p>.25 J U N E 17, 19 40" 20 0 n = 50 x = 2.16 k = 1.40 p >.25 J U N E 2 4 - 2 6 60-n = 120 x = 5.09 k = 0.26 p >.001 F L O C K S I Z E 166 experiment with decoys conducted simultaneously in an adjacent area upstream of the r i v e r mouth (Chap. 3, F i g . 29). The f l o c k - s i z e d i s t r i b u t i o n s for the f i r s t two periods do not d i f f e r s i g n i f i c a n t l y from negative binomial d i s t r i b u t i o n s expected for the EFS (p > .25, x2). Flock sizes were s u b s t a n t i a l l y larger during the l a s t week in June; moreover, the d i s t r i b u t i o n could not be f i t t e d to a corresponding negative binomial d i s t r i b u t i o n (p < .001, x2). Although the d i s t r i b u t i o n of small flocks (<_ 6 birds) in the l a s t week was s i m i l a r to that i n early weeks, larger flocks of up to 20 birds appeared which created a bimodal d i s t r i b u t i o n . I t i s inferred from t h i s , that birds i n the large flocks during l a t e June are behaving gregariously - not independently as assumed by the EFS model. Most l i k e l y , t h i s r e f l e c t s a seasonal change associated with reduced f l i g h t a c t i v i t y noted during June -August. Flock-size data were also c o l l e c t e d during regular census coverage of a l l streams during 1981. The d i s t r i b u t i o n of mergansers by flock size on fresh and t i d a l waters of the 3 major systems i s summarized for the peak period of juvenile salmonid migration ( A p r i l 1 - May 31) i n F i g . 41A and for months subsequent to downstream migration (June 14 - Sept. 14) i n F i g . 41B. An equal number of census days are included in both figures so that abundances are d i r e c t l y comparable. Only gross comparisons can be attempted at t h i s l e v e l , because so l i t t l e i s known about v a r i a t i o n s in p r o f i t a b i l i t y among d i f f e r e n t s i t e s within the r i v e r systems. Even so, i t i s clear that both abundance and f l o c k s i z e were greater on BQ during the salmonid migration than on e i t h e r of the other larger streams. Only on BQ during the months of salmonid migration did f l o c k sizes on freshwater exceed those on t i d a l waters. After the salmonid migration, the largest flocks and the greatest density of mergansers 16? F i g . 41. Comparison of merganser flocks on fresh and t i d a l waters i n 1981. Black bars - freshwater; s t i p p l e d bars - t i d a l waters; white - j u v e n i l e mergansers. A - during juvenile salmonid migration ( A p r i l 1 - May 31); B -af t e r j u v e n i l e salmonid migration (June 14 - Sept. 14). Arrows indicate t y p i c a l f l o c k s i z e s . 168 100 50 ENGL I SHMAN n = 83 freshwater tidal waters n = 144 L ITTLE QUALICUM n = 145 n =164 BIG QUAL ICUM n =762 ! n = 213 10 20 30 40 F L O C K S IZE 50 60 169 5 0 0 50 B i ENGL ISHMAN n = 31 f r e shwate r t t idal wa te r s • juveni les n = 208 Q or G O O 5 0 0 LU CQ 1 t L ITTLE QUALICUM n = 5 6 n = 529 5 0 0 5 0 1 t BIG QUAL ICUM n = 128 n = U 8 0 10 20 30 40 F L O C K S I ZE 50 60 1?0 occurred on the LQ estuary. Also, note that flock sizes on t i d a l waters from mid-June to mid-September were larger than during A p r i l - May even though de n s i t i e s there were a c t u a l l y lower on ENG and BQ during the l a t e r months. Flock sizes became e s p e c i a l l y large (up to 56) on the LQ during June 14 -Sept 14, which confirms the trend observed in the previous experiment ( F i g . 42). Pr e d i c t i n g Aggregations - A Generalized EFS Model: The Approach: Although predictions of the EFS model have proven q u a l i t a t i v e l y correct for merganser fl o c k s , not only the size d i s t r i b u t i o n of f l o c k s , but also the number of flocks i n a s p e c i f i e d area i s of concern. Of course, these two quantities are interdependent so that stochastic models become awkward to develop or apply (Cohen 1972, Morgan 1976). A simpler, a l t e r n a t i v e , approach i s to regard the c o l l e c t i o n of f l o c k s , i t s e l f , as a large 'flock' or aggregation which grows and decays according to immigration and emigration by the remaining l o c a l population not included within the aggregation. In taking this approach, i t i s necessary to evaluate the following q u a n t i t i e s : a) the e f f e c t i v e population size i n the v i c i n i t y (N) - a l l birds equally l i k e l y to j o i n an aggregation, b) the p r o b a b i l i t y that an i n d i v i d u a l (member of N) w i l l approach the s i t e of aggregation during an i n t e r v a l of 1 day ( Xj), c) the p r o b a b i l i t y that having approached the s i t e , an i n d i v i d u a l w i l l j o i n the aggregation ( A 2 ) , d) the p r o f i t a b i l i t y at the s i t e of aggregation (P), 171 F i g . 42. Seasonal trend in abundance of mergansers in units of equilibrium L aggregation model. Males of breeding pairs and females with broods are excluded. Census covered freshwater (to 5 km) and t i d a l zones of Englishman, L i t t l e Qualicum and Big Qualicum r i v e r s during 1981. Regression i s l i n e a r for log-transformed data with 95% C.I. for predicted values: Log y = 5.04 - 0.012x, n = 25, r" = -.83 F. = 50.0, p < 0.001 1 7 3 e) the p r o b a b i l i t y of departure for an i n d i v i d u a l member of the aggregation (d) given p r o f i t a b i l i t y P. The Model: A d e t e r m i n i s t i c , discrete-time model was constructed to make predictions about trends in the aggregative response of mergansers to hatchery releases of j u v e n i l e salmonids at RC and BQ. Expressions for assumptions (a) to (e) above are as follows: E f f e c t i v e Population Size = Nfc = N^e W t [4.6] where t indicates time in days and a) = 0.012. This expression i s required because of the seasonal decline in abundance during the period of juvenile salmonid migration. The parameter, co , was estimated by regressing the pooled census data for ENG, LQ, BQ against time during March - June, 1981; again each pair counted as one i n d i v i d u a l and females with broods were excluded (p < .001, F t e s t , F i g . 42). The p r o b a b i l i t y of j o i n i n g the aggregation, having already approached the s i t e , was assumed to be of the same form as for j o i n i n g a f l o c k , since an approaching b i r d should be able to assess both the number and size of flocks from overhead. Pr (jo i n s | approach) = A 2 = a + 0 A [4.7] where a and 0 are constants and A i s the size of the aggregation. If the number of approaches per day follows a Poisson d i s t r i b u t i o n with parameter, Al, then the number of birds j o i n i n g the aggregation w i l l follow the compound Poisson d i s t r i b u t i o n with parameter Al A2- T h e p r o b a b i l i t y of j o i n i n g the 1?4 aggregation at least once in a day is therefore one minus the zero frequency as follows: Pr (joining) = a(A) =1 - e " X l ( " + P A > [A.8] where , a , P > 0, A ^ 0. Parameters a and P were given values of 0.4 and 0.03 respectively since the observed response to decoys ( a = .35, P = .03, Chap. 3) defines minimum values. The number of departures was determined at the end of each day after a l l arrivals had taken place to f a c i l i t a t e computation. Because interference effects could not be demonstrated for mergansers foraging in flocks of up to 25 individuals (Chap. 2) and aggressive interactions appear to play a small role (Chap. 3), the probability of departure was assumed to be independent of the size of the aggregation and to depend only on p r o f i t a b i l i t y . The duration of v i s i t s was found to increase exponentially with fish density (Chap. 3, Fig. 32), so the probability of departure from an aggregation was assumed to follow the reciprocal relationship: — j, p Pr (departure) = d(p) = pe [4.9] where P, V > 0, P ^ 0. Pr o f i t a b i l i t y was assumed to be proportional to the juvenile salmonid index (JSI); the availability of non-salmonid fishes was ignored. Residence times of smolt released from the hatcheries are known only approximately. At RC, steelhead smolt were assumed to emigrate at a rate of 25% per day based on personal observations of how long they remained conspicuous in the stream. Estimates of smolt residence times for BQ were obtained from Mace (1983 - Appendix). 1 7 5 F i t t i n g The Model: During the i n i t i a l t r i a l s , parameters X^ t p and v were assigned values which seemed pla u s i b l e and provided an adequate scale of response i n simulating responses to f i s h releases at R C in 1980. A s a t i s f a c t o r y response was obtained with \\ ~ 0.5, p = 0.8 and v - 0.05. Since the B Q and R C systems d i f f e r g r e a t l y i n s i z e , i t was also necessary to scale the e f f e c t i v e population sizes and p r o f i t a b i l i t y estimates accordingly. The e f f e c t i v e population size i n the v i c i n i t y of B Q was taken to be the pooled census estimate for B Q , L Q , and E N G from F i g . 42 and No = 150 where to ^ s March 20. For R C , N Q was assumed to be the maximum number recorded in the v i c i n i t y following a large smolt release (31 birds on A p r i l 29, 1980) corrected for seasonal decline; hence N Q = 31/e~^0 w = 50. P r o f i t a b i l i t y was scaled such that the predicted maximum aggregation si z e coincided c l o s e l y with maxima for the observed trends i n abundance for years with most census coverage (1980 at R C and 1981 at B Q ) . This required that the JSI be 25 times higher at B Q than at R C . Predictions for 1981 at R C and 1980 at B Q were generated without further adjustment of parameter values. Pre d i c t i o n s : Predicted and observed trends are presented i n F i g . 43 and F i g . 44 for R C and B Q res p e c t i v e l y . It i s scarcely s u r p r i s i n g that predicted trends matched observed trends for one f i s h release at R C during 1980; t h i s was v i r t u a l l y assured by se l e c t i n g appropriate Xj, p and v values. It i s encouraging, however, that predicted trends track observed trends so c l o s e l y over a l l f i s h releases ( F i g . 43A). Observed responses were less than predicted during June (days 72 - 102) in both 1980 and 1981. Merganser abundance was higher than expected during the enclosure experiments (days 42 - 80) in 1981 but the comparison i s inappropriate because small numbers of 176 smolt were made available at high density during this period (Fig. 43B). Predicted trends at BQ in 1981 (Fig. 44B) matched the general features of the observed trends but again tended to overestimate abundance in late May through June. In this case, the predicted curve was fitted only by scaling p r o f i t a b i l i t y ; other parameters are identical to those used in the RC predictions. The match in 1980 for BQ (Fig. 44A) is surprisingly good considering that the predicted trend was generated without any reference to that observed, and that fish releases in 1980 differed considerably with respect to both size and time of release from those in 1981. 1 7 7 Fig. 43. Comparison of observed and predicted trends in abundance of mergansers at Rosewall Creek. A - 1980; B - 1981. Abundance is in units of the aggregation model (i.e. breeding males and females with broods excluded). Circles - paired birds d i f f i c u l t to distinguish (3 assumed); cross - atypical, cohesive flock included (corresponding dot represents typical, small, scattered flocks only). Stippled fish release - smolt; white - fry. 179 Fig. 44. Comparison of observed and predicted trends in abundance o mergansers at Big Qualicum. A - 1980; B - 1981. Abundance is in units of th aggregation model (i.e. breeding males and females with broods excluded) Stippled fish release - smolts; white - fry. 181 DISCUSSION Evaluating Assumptions of the Aggregation Model: P r o f i t a b i l i t y : Throughout the preceding analysis, p r o f i t a b i l i t y has been represented by a juvenile salmonid index (JSI), to provide a consistent assumption about size s e l e c t i o n by mergansers. The precise formulation of the index i s debatable, but i t i s c l e a r that some index i s required; moreover, the appropriate s i z e - s e l e c t i o n assumption must l i e between the bounds of s e l e c t i o n in proportion to biomass and s e l e c t i o n i n proportion to number ( F i g . 59). Available evidence from coho smolt and f r y s e l e c t i o n experiments (Chap. 2) suggests that the JSI may overestimate the importance of large f i s h . I f t h i s i s true, p r o f i t a b i l i t y has been underestimated during the f r y migrations ( l a t e March - early May) r e l a t i v e to the l a t e r smolt migrations. Such an error would tend to accentuate, not diminsh, the r e l a t i o n s h i p s between p r o f i t a b i l i t y and merganser dispersion ( F i g . 37 and 38). The JSI assumption has v i r t u a l l y no e f f e c t on the RC predictions because a l l f i s h released form the RC hatchery were of comparable siz e excepting a small release of large (1 g) chum f r y l a t e i n 1980. However, predicted trends on BQ might be i n f l a t e d l a t e i n the season r e l a t i v e to trends during the e a r l i e r migrations. The A r r i v a l Function: A r r i v a l rate i s assumed to depend on e f f e c t i v e population size and a l o c a l enhancement term determined by f l o c k (or aggregation) s i z e . Population size was estimated em p i r i c a l l y for both the RC and BQ areas. By adopting i n d i v i d u a l estimates for the two areas, rather 182 than a single estimate for the en t i r e c o a s t l i n e , i t has been assumed, i m p l i c i t l y , that other factors besides f i s h density influence dispersion. Indeed, mergansers were always more abundant on BQ than RC even during October and November, long a f t e r j u venile salmon migration had ceased and subsequent to departure of most migrating mergansers (x = 15.0 + 2.6 [1 SE] on BQ and 2.5 +_ 2.5 on RC). No attempt has been made to i d e n t i f y the s i g n i f i c a n t factors but i t seems l i k e l y that physical c h a r a c t e r i s t i c s of the streams such as width, length and current v e l o c i t y would be relevant cues e a s i l y perceived by mergansers. The decline i n abundance from A p r i l through June, used in pre d i c t i n g aggregation size ( i . e . equation 4.6), was estimated from combined (1981) data for BQ, LQ and ENG. Thus, one might argue that the predicted trends for BQ in 1981 are not independent of the observations. But the exponential decline i n abundance over the f u l l season i s apparent only in the composite data (compare F i g . 32 and 33). Furthermore, the r e c i p r o c a l r e l a t i o n s h i p s betwen BQ, LQ and ENG indicate that other processes besides continuous decline operate within the BQ system. The same decline parameter, based on the composite 1981 data, was used to generate predictions for RC and for BQ in 1980; hence, these predictions are independent of the observed trends. The l o c a l enhancement term in [4.7] controls the i n i t i a l rate of accumulation ( i . e . speed of response) following an increase i n p r o f i t a b i l i t y . But the p r o b a b i l i t y of j o i n i n g the aggregation, given an approach, reaches c e r t a i n t y for aggregations exceeding 20 b i r d s . Thus, the l o c a l enhancement term can only have a s i g n i f i c a n t e f f e c t for small 183 aggregations such as those on RC and those following i n t e r v a l s of low p r o f i t a b i l i t y on BQ. The p r o b a b i l i t y of approach, \ l , i s the more c r i t i c a l parameter determining number of a r r i v a l s and consequently, the responsiveness of the aggregation to increased p r o f i t a b i l i t y . It i s l i k e l y that reduced f l i g h t a c t i v i t y , such as observed from June to September would decrease the rate of approach. Because Xj was a constant in the simulations, predicted a r r i v a l rates and hence aggregation s i z e , may have been excessive l a t e i n the season ( a f t e r day 80 in simulations). Indeed, the discrepancy between predicted and observed trends could be reduced in simulations for RC and for BQ in 1981 by decreasing X^ in accordance with observed f l i g h t a c t i v i t y . The Departure Function: Two parameters determine the p r o b a b i l i t y of departure: the f i r s t , p , merely scales the departure rate whereas the second, v , governs the reponse to p r o f i t a b i l i t y . Scaling p r o f i t a b i l i t y , P, independently for BQ and RC i s equivalent to f i t t i n g v i n d i v i d u a l l y . The residence time of f i s h following release determines how P (or equivalently V ) changes over time. Accordingly, one must ask how much of the f i t in F i g . 43 and 44 i s due to assumptions about changes i n p r o f i t a b i l i t y and how much r e s u l t s from the model having the correct dynamics. It i s c l e a r that p r o f i t a b i l i t y i s related to both prey density and the v u l n e r a b i l i t y of prey (e.g. Chap. 2). Because BQ and RC d i f f e r greatly in s i z e , e s p e c i a l l y in stream volume below the hatchery, i t i s not s u r p r i s i n g that more f i s h are required to produce a given response in BQ than in RC - some sca l i n g i s necessary. The BQ hatchery i s located approximately twice as far upstream as the RC hatchery and BQ appears to be 2 - 3 times wider and perhaps 184 3 times as deep, on average than RC. As a very rough estimate, a given quantity of f i s h would be up to 18 times more concentrated in RC than i n BQ. Thus, the s c a l i n g factor of 25 adopted to f i t the model i s not too d i f f e r e n t from that expected on the basis of d i l u t i o n alone, although other factors such as cover and t u r b i d i t y may influence v u l n e r a b i l i t y . Although excellent cover i s a v a i l a b l e in 5 or 6 pools below the hatchery on RC, there are lengthy intervening sections of shallow r i f f l e s where cover for large f i s h i s poor. Moreover, the water i n RC i s p a r t i c u l a r l y c l e a r . Water depth in BQ appears to be more consistent throughout i t s length below the hatchery so that f i s h may pass through fewer bottlenecks where they become e s p e c i a l l y vulnerable to predation. Freshwater residence time of salmonids following release varies among species (Tom B i l t o n , pers. comm.) and with size and date of release ( B i l t o n et a l . 1982). My observations of the decline in abundance of steelhead smolt in reference pools and of the incidence of coded-wire tags i n otter spraints at RC ( i n preparation) indicate that while most f i s h l e f t during the f i r s t few days, some smolt remained in the stream for up to a month following release. In the RC simulations, I have assumed that f i s h emigrate at the rate or 25% per day; in other words, 50% would have l e f t the stream within the f i r s t 3 d but 0.1% would s t i l l remain af t e r 26 days. Although t h i s exponential model describes my observations adequately, i t i s equally possible that the majority of f i s h emigrate during the f i r s t few days but that those remaining are depleted by predators at a constant rate ( i . e . l i n e a r decrease in abundance). Smolt den s i t i e s following release in BQ have been calculated by Mace (1983 - Appendix) who found that 185 emigration of coho and chinook salmon was best described in two phases - an exponential decline in abundance during the f i r s t two-thirds of the migration followed by a slower, l i n e a r decline during the remainder of the migration. It t h i s were also true for steelhead trout at RC, p r o f i t a b i l i t y would decline more r a p i d l y , immediately following a release but remain at a low to intermediate value for a longer period than has been assumed; consequently, aggregations would peak sooner but disperse less quickly (under the EFS model) and i f anything, the f i t to observed trends i n F i g . 43 would be improved. S e n s i t i v i t y of Predictions to Parameter Values: Predictions of the aggregation model are most influenced by v a r i a t i o n s in parameters X} and v . Isopleths for residual variance (mean sum-of-squared errors) are plotted for a range of values of Xj and v in F i g . 45A. The plot indicates that a wide range of combinations of these two parameters f i t the data equally w e l l ; in other words, the absolute value of either parameter i s not c r i t i c a l to the model's performance. A second technique for evaluating robustness of a model i s to draw parameter values randomly from p r i o r d i s t r i b u t i o n s determined by independent data or subjective estimates (Don Ludwug, pers. comm.). With respect to the present model, three parameters, Xi, v and p , are of primary concern -the others are r e l a t i v e l y well defined by independent observation. Uniform p r i o r d i s t r i b u t i o n s were assigned to each within the following l i m i t s : one-half to twice the value of Xi and v giving best f i t , i . e . 0.25 - 1.0 and 0.035 - 0.14 re s p e c t i v e l y and 0.6 - 1.0 for p (the p r o b a b i l i t y of leaving the v i c i n i t y within one day given zero p r o f i t a b i l i t y ) . The r e s u l t s 186 Fig. 45. Sensitivity of the aggregation model to variations in parameter values with respect to predictions at Rosewall Creek, 1980. A - contour plot of mean sum of squared residuals with respect to values of parameters X, (arrival) and v (departure); the f i t to data is highly significant (p < .001) for MSS values < 36; residual sum of squares = total sum of squares along dotted line. B - predicted trends from combinations of parameters X, , V and p drawm randomly from uniform prior distributions (see text). 188 of 5 t r i a l s with combinations of A j , \) and p drawn randomly from within these intervals are illustrated in Fig. 45B. A l l trajectories l i e within the shaded region; 4 of the 5 runs were s t a t i s t i c a l l y significant (p < .001, F test). Hence, i t is concluded that the aggregation model is robust to variations in parameter values. Yet, there appears to be a systematic decline in the aggregative response throughout the season that is not captured by the model regardless of parameter values. This anomaly is attributed to a seasonal change in ranging distance or social behaviour as evidenced by decreased flight activity and larger flock sizes during the summer months. . • v Conclusions: An aggregation model, driven by a simple response to pro f i t a b i l i t y , captures many qualitative features of the temporal disperson of mergansers on salmon-producing streams. The predictive power of the model does not reflect on the truth or falsehood of the individual relationships comprising the model - these relationships have been validated independently by direct experiment or observation (Chap. 2 and 3). Rather, errors in prediction should help to establish the relative importance of these relationships and to reveal whether other, significant variables have been overlooked. More reliable predictions might have been possible had estimates of prof i t a b i l i t y been based on better information regarding prey size selection and the availa b i l i t y of non-salmonid prey in estuaries. Nevertheless, the results are encouraging, for they suggest that the most important processes determining the aggregative response have been identified. 189 Two problems are l i k e l y to be encountered i n applying the model outside the context of th i s study. F i r s t , mergansers demonstrated an a b i l i t y to learn the loc a t i o n of p r o f i t a b l e feeding areas (Chap. 3); the e f f e c t of previous experience was ignored i n the model because the learned response was slow in comparison with the time-scale for flu c t u a t i o n s i n prey density. This condition i s s a t i s f i e d only during periods when f i s h are being released form hatcheries frequently, but not continuously. In general, the e f f e c t of experience cannot be ignored. Second, the performance of the model i s dependent on mergansers dispersing independently i n t h e i r search for favourable feeding areas. This condition appears to be s a t i s f i e d during the breeding season but i t i s probably v i o l a t e d at other times of the year. For example, the f l o c k - s i z e d i s t r i b u t i o n at LQ became bimodal i n late June and f l i g h t a c t i v i t y declined s t e a d i l y during the summer months. P r o f i t a b i l i t y may be e s p e c i a l l y important to mergansers during the breeding season both because of the addi t i o n a l energy required by egg-laying females and because the d i s p e r s a l a b i l i t y of merganser broods i s l i m i t e d . The broods remain within or near t h e i r natal stream u n t i l able to f l y at about 10 wk of age (Chap. 5 and 6). Thus, the need to locate concentrations of prey that w i l l not be depleted before the broods are well-developed, may lead to greater exploration and enhanced responsiveness to p r o f i t a b i l i t y among breeding p a i r s . At other times of the year, p r o f i t a b i l i t y may be less c r u c i a l to f i t n e s s , since the rate of energy a c q u i s i t i o n by adult mergansers i s constrained, t y p i c a l l y , by diges t i v e rate and not hunting a b i l i t y (Chap. 2). For these birds, the anti-predator advantages of s o c i a l behaviour may more than compensate for add i t i o n a l expenses i n search time while remaining i n flocks on less 190 profitable feeding grounds. Similarly, winter flocks of shelducks (Tadorna tadorna) do not always feed in the most profitable areas although they avoid unfavourable areas (Buxton 1981). Further research is required to extend flock size models to multiple patches; only then w i l l they have general u t i l i t y . Recent studies have shown that mallards (Anas platyrhynchos, Harper 1982) and feral pigeons (Columba l i v i a , Lefebvre 1983) distribute themselves in groups among patches such that each experiences similar feeding success (the 'ideal free' distribution, Fretwell 1972). However, in these experiments, patches were close together and relative p r o f i t a b i l i t y could be assessed rather easily. On a larger spatial scale, such as between river systems, sampling is necessarily more expensive; since stakes are higher, the consequences of patch choice decisions should have a greater effect on fitness. The extension of theory, orignally derived on small spatial scales, to larger scales i s a necessary process in testing our understanding of behaviour and population ecology. Moreover, such investigation is necessary to achieve results of practical benefit for the management of natural resources. 191 CHAPTER 5: NESTING DESPERSION AND THE AGGREGATIVE RESPONSE INTRODUCTION The aggregative response of predators to patches of high prey density i s , in general, a compensatory or s t a b i l i z i n g force i n predator-prey interactions ( R o l l i n g 1959a). M o r t a l i t y due to aggregating predators i s increased where prey i s concentrated and therefore, reduced where prey i s less common. In e f f e c t , patches of low prey density become refugia for the prey population (Hassell 1978). This compensatory property of non-random search i s diminished i f aggregations of predators do not track prey density c l o s e l y over time. Lag time (hysteresis) i n the a r r i v a l phase of the aggregative response reduces m o r t a l i t y at high prey density, at least i n i t i a l l y , whereas lags during departure phase may lead to unusually high mortality at low prey density. Consequently, behavioural a t t r i b u t e s of predators that cause lags i n response time may influence the s t a b i l i t y of the predator-prey system as a whole and w i l l determine the pattern of mortality for prey at s p e c i f i e d s i t e s . Both paired (breeding) and unpaired common mergansers congregate at favourable feeding s i t e s during the nesting season (Chap. 4). It i s reasonable to ask whether greater nesting a c t i v i t y might also occur near these locations as a r e s u l t of the aggregative response of po t e n t i a l breeding p a i r s . Such a change in dispersion of reproductive a c t i v i t y need not af f e c t the predator's rate of increase - i t i s not a reproductive response i n the usual sense (e.g. Readshaw 1973). Nevertheless, reproduction requires a committment to the nesting s i t e that i s not exhibited by non-breeding 1 9 2 individuals; the female merganser and its brood w i l l continue to forage on the natal stream for up 10 wk after hatching, at which time the young are able to fl y . Breeding pairs might be expected to forage close to their nests throughout the egg-laying and incubation period (the central-place foraging hypothesis, Covich 1976, Orians and Pearson 1979). Consequently, increased breeding activity due to aggregation of breeding pairs at profitable feeding sites would greatly prolong the aggregative response. A committed aggregative response by mergansers would be of special importance for salmonid populations because merganser broods feed heavily on fish that inhabit freshwater during the summer months (Chap. 6). Young mergansers raised on hatchery-enhanced streams cannot feed on protected hatchery stock; instead, they must rely on wild salmonids and alternative species. In principle, a committed aggregative response to hatchery fish released over short duration, perhaps early in the nesting season, could lead to unusually intense predation on wild stocks over a period of many weeks. The objective of this study was to identify factors that limit merganser brood densities on the small coastal river systems of eastern Vancouver Island. Several alternative hypotheses were examined: (1) the size of the potential breeding 'population' limits brood production below the carrying capacity of nesting areas; (2) the carrying capacity of a river system is limited by the number of safe nesting locations or territories and thus, proportional to the area of suitable nesting habitat accessible from the river; (3) carrying capacity depends on the availability of prey, and in large part, upon the density of juvenile salmonids. In the latter case, p r o f i t a b i l i t y might be assessed prior to choosing a nest site, or on the basis of previous experience in raising broods. A committed aggregative response 1 9 3 could arise only i f brood densities were typically lower than the carrying capacity of the river system (hypothesis 1) or i f breeding pairs over-estimated the carrying capacity based on experience prior to nesting. 1 9 4 METHODS Study Area: Eight salmon-producing streams on the east coast of Vancouver Island were censused r e g u l a r l y for merganser breeding pairs from l a t e May to September, 1980 - 1982. The streams were selected to provide contrast i n size (drainage area), wild salmonid den s i t i e s and hatchery-enhanced salmonid production. Deta i l s pertaining to the study area, salmonid production and extent of census coverage are given i n the Methods section of Chap. 4 and Table VI. Breeding Pair Census: The cohesive behaviour of breeding pairs permits easy recognition i n most s i t u a t i o n s . Maturity of females was confirmed, where possible, by examining the si z e and demarcation of the cheek patch (Erskine 1971). Whenever p a i r i n g status remained uncertain, as for example during feeding frenzies or when ac c i d e n t a l l y flushed, no pairs were recorded; hence breeding pair counts w i l l tend to underestimate the true abundance. Maximum counts have been used for comparison of breeding density for th i s reason. Brood Census: In contrast to adult mergansers, the broods are d i f f i c u l t to census. Broods alarmed at close range fl e e across the water and are conspicuous but those sensing the presence of the observer from a distance, sneak s i l e n t l y 195 under cover and are extremely d i f f i c u l t to detect. Consequently, success in sighting broods depends on stealth. During this study an observer walked upstream on t r a i l s cut well back from the stream and emerged quietly at strategic viewpoints to scan for broods. Census by canoe, although convenient, was concluded to be unreliable because broods would often have opportunity to hide. Location, activity, age and number of young was recorded for each brood. Age was estimated from size and plumage as described by Erskine (1971). Hatching dates were back-calculated from estimated age at f i r s t sighting. Estimates of Brood Production: Individual broods were identified throughout the rearing period by plotting estimated age against elapsed time; correctly aged broods followed diagonals in the plot. The number of downy young per brood varied from 1-18 and was helpful in distinguishing different broods of similar age. Occasionally, distinct broods grouped together to form a single, composite brood accompanied by only one female. Composite broods were usually easy to detect because the ducklings differed in age. In two cases on BQ, composite broods were too large to have been formed from previously identified broods; accordingly, a single 'hypothetical' brood of the required size was added to the estimate of broods produced. Average clutch size was estimated from the number of downy young in broods identified at < 14 d of age. Daily survival rate for downy young < 30 d old was calculated from reliable counts of downy young in broods seen > 3 times. Survivorship over the f i r s t 30 d was estimated from the daily estimates assuming a constant mortality rate. 1 9 6 Only a fraction of broods known to exist were observed on each census. The final estimates of brood production depend on both the extent and frequency of census coverage. At least 20 censuses were conducted on each stream during the brood-rearing period (May 15 - August 15) in 1980 and 1981. These censuses covered the tidal waters and between 20 to 50% of the freshwater length accessible to salmon upstream from the estuary (1 - 5 km depending on stream). Weekly counts on tidal waters and > 30% of accessible freshwater length were undertaken during 1982 based on the results from more frequent counts in the previous 2 yr (see below). 1 9 7 RESULTS R e l i a b i l i t y of the Brood Production Estimates Much of the analysis to follow r e l i e s on r e l a t i v e l y small differences in brood production on neighbouring streams. It i s important, therefore, to examine the r e l i a b i l i t y of the brood production estimates. These estimates depend on the frequency and extent of census coverage because only a f r a c t i o n of broods known to exist were sighted during each census. A brood may have escaped detection for two reasons; f i r s t , i t may have moved upstream into t e r r i t o r y not covered by the census and second, i t may have been wary enough to hide from the observer. Census Coverage: The r e l a t i o n s h i p between the extent of census coverage upstream from the estuary and the f i n a l estimate of brood production i s i l l u s t r a t e d i n F i g . 46 for streams censused frequently (> 20 times) during May 15 - August 15, 1980 and 1981. Estimates r i s e to an asymptote at 20 - 30% coverage of length accessible to salmon. This r e s u l t s from the tendency of broods to move downstream as they grow (Chap. 6, White 1957). In p r i n c i p l e , a l l broods could be counted near the r i v e r mouth provided censuses were conducted often enough. In p r a c t i c e , however, d i f f i c u l t i e s would arise i n d i s t i n g u i s h i n g broods of s i m i l a r age and size when not i d e n t i f i e d separately within a single census. 198 Census Frequency: The effect of census frequency at 20 and 30% coverage is shown in Fig. 47. Again, an asymptote is reached at approximately 5 census days per month on a l l rivers except BQ in 1981 - but even this estimate appears to be within 1 - 2 broods of its asymptotic value. Practical considerations determine the optimal tradeoff in sampling effort between the amount of territory covered and the frequency of counts. Weekly counts of > 30% of accessible length were deemed apropriate for the larger rivers censused in 1982 (Q-LC, ENG, BQ) on the basis of the relationships in Fig. 46 and 47. 199 Fig. 46. Relationship between census coverage and estimated brood production. Open symbols - 1980; solid symbols - 1981. At least 20 census days were scheduled from May 15 - Aug. 15. BQ - Big Qualicum R.; LQ - L i t t l e Qualicum R.; ENG - Englishman R.; RC - Rosewall Creek. 200 % OF FRESHWATER ACCESSIBLE TO SALMON CENSUSED (IN ADDITION TO TIDAL WATERS) 201 Fig. 47. Relationship between census frequency and estimated brood production. Open symbols - 1980; closed symbols - 1981. Triangles - 20% and circles - 30% of accessible freshwater length censused upstream from river mouth. 2 0 2 203 Factors Limiting Brood Production Hypothesis 1 - Breeding Population Limits Brood Density Below Carrying Capacity of Streams: The peak period of egg-laying and incubation on eastern Vancouver Is. occurs in May (Fig. 48), assuming that egg-laying begins 45 days prior to brood emergence (Bellrose 1978). Approximately half of breeding pairs had begun laying eggs by May 1 and 95% had begun by June 1. Thus the maximum density of breeding pairs in May should provide the best index of brood production because virtually a l l successful pairs w i l l have established nests by the end of May. If the density of breeding pairs on a given stream is greater in April than in May, i t follows that brood production on that stream cannot be limited solely by the abundance of breeding pairs. The comparison of pre-nesting densities (April) and nesting densities (May) in Fig. 49 indicates that 'surplus' breeding pairs are evident on BQ in both years censused. Clearly, other factors limit brood production on this system. The fact that no surplus breeding pairs are observed on the neighbouring streams suggests that either mergansers were able to assess stream carrying capacity prior to nesting or that breeding densities are, typically (in the absence of fis h hatcheries), less than carrying capacities. Hypothesis 2 - Brood Carrying Capacity is Determined by the Size of Streams: The maximum number of breeding pairs observed on freshwater during May increased linearly with the distance censused upstream from the river mouth 204 Fig. 48. Distribution of egg-laying/incubation activity back-calculated from hatching dates. It is assumed that 45 <1 are required to lay and incubate a clutch. 206 F i g . 49. Comparison of breeding-pair density before and during peak egg-laying period. Open symbols 1980; black symbols - 1981. Census coverage: RC and LQ (1980) to 1 km upstream; LQ (1981), BQ and ENG to 5 km upstream. 2 0 ? j 208 ( F i g . 50). This observation suggests that breeding pairs space themselves out along the length of the stream. Moreover, the slopes of the r e l a t i o n s h i p were si m i l a r for each r i v e r (BQ, LQ, and ENG) in 1981, although the slope for BQ was steeper i n 1980. This implies that the density of breeding pairs i s s i m i l a r on the freshwater reaches > 1 km above the r i v e r mouth; thus, the number of breeding pairs above 1 km was proportional to the size of the system as expected under Hypothesis 2. However, the number of breeding pairs on the lower kilometer of freshwater ( i . e . the intercept i n F i g . 50) i s s u b s t a n t i a l l y higher on BQ and RC than on LQ and ENG. I a t t r i b u t e t h i s r e s u l t to differences i n salmonid density (Chap. 4). Fish production from hatchery f a c i l i t i e s situated 1.3 and 0.7 km upstream on the former two streams comprises 75 and 80%, r e s p e c t i v e l y , of the o v e r a l l j u venile salmonid migration in terms of biomass. In contrast, the recently-constructed hatchery and spawning channel on the LQ system, situated 3.4 km upstream, produces only 25% as much as the BQ f a c i l i t i e s ; there i s no hatchery on ENG although a r e l a t i v e l y small number of smolt (2% of the biomass released at BQ) were transported there by truck. It should be noted that breeding pairs observed on t i d a l waters are not included in the preceding comparison. The d i s t r i b u t i o n of maximum counts of breeding pairs by zone within the BQ, LQ and ENG systems i s presented in F i g . 51. Breeding pairs were most abundant near the r i v e r mouth of a l l systems; on BQ, they were concentrated on the lower kilometer of freshwater as opposed to t i d a l waters due to the greater a v a i l a b i l i t y of juvenile salmonids. Even so, breeding pair d e n s i t i e s near the r i v e r mouth were as high or higher on BQ than on either of the other streams. Because BQ i s a smaller system, the o v e r a l l breeding pair density ( t i d a l and freshwater) must have been greater than on the other systems. 2 0 9 Fig. 50. Comparison of breeding-pair density on freshwater reaches of neighbouring streams during peak egg-laying period. Maximum counts are plotted against distance censused upstream from stream outlet. Open symbols - 1980; solid symbols - 1981. CENSUS COVERAGE ( km upstream from river mouth ) 211 Fig. 51. Dispersion of breeding-pairs during peak egg-laying period. These are maximum counts per zone and therefore, are not additive. The 'ALL' category represents the maximum count for a l l areas censused. 2 1 2 C O rx < Q_ CD D LU LU or CD LL O fx LU CD 12-8-4-0-12-8-4-0-12-8-4-0-ENGLISHMAN 1981 •v.-a ft-•J.>4---.' LITTLE QUALICUM 1981 BIG QUALICUM 1981 BIG QUALICUM 1980 x < 16-12-8-4-0-TIDAL 1 2 3 4 5 v ss " km UPSTREAM ALL 213 I conclude that breeding pairs are dispersed at a low but comparable density on the upper reaches of a l l streams. To this extent, brood production is expected to increase linearly with the length of the stream and i t s tributaries. However, breeding pairs also congregate near the river mouth apparently in response to favourable feeding conditions there. It is not known how many of these pairs have established nests upstream but presumably many have because paired birds frequently move upstream from the estuary and vice versa. Thus, an additional component of brood production, not related to stream size, remains to be explained. Hypothesis 3 - Brood Carrying Capacity is Related to Salmonid Production: Multiple regression analysis was performed to examine correlations between estimated brood production and drainage area (hypothesis 2) and juvenile salmonid production. Drainage area was thought to best reflect the overall length of a stream and i t s numerous tributaries - that i s , to give the best index of available nesting territory. Drainage area above dams that are impassable to broods (such as occur on Q-LC and BQ) has not been included. Both independent variables were significant (Fig. 52); residuals after the drainage area effect has been removed are plotted against juvenile salmonid migration in Fig. 52B and 52C. Only salmonids available to the breeding pairs during A p r i l and May - before a l l pairs are committed to nesting sites - were considered in Fig. 52B. Alternatively, , in Fig. 52C, the average spawning escapement of wild salmonids whose progeny inhabit freshwater (mostly coho, (). kistuch) was used as an index of typical wild fry densities resident in freshwater during summer months and thus, available to the growing broods. Residuals (after drainage area effect removed) were also regressed against 214 total juvenile salmonid production computed in standardized units (JSI - see Chap. 4, Methods). A l l regressions are significant, but total and early salmonid migration indices give a much better f i t to the data (p < .0001, r 2 = .90 for both) than the index of wild production available to the broods (p < .001, r 2 = .77). These data are insufficient to draw firm conclusions as to when or how salmonid density influences carrying capacity but there remains l i t t l e doubt that salmonid production is an important correlate of brood density. It is conceivable that mergansers evaluate the food resources available in potential nesting areas on the basis of early juvenile salmon migrations. Wild smolt migrations (which peak in May) would provide an index for the density of resident fry available to broods later in the current year provided wild smolt production i s similar from year to year. Of course, the index i s only appropriate for unenhanced streams but i t is unlikely that mergansers can distinguish between wild and hatchery-reared smolt. Data are available to test two predictions of this 'committed aggregation' hypothesis. F i r s t , breeding pairs which choose nest sites early in April would have committed themselves to a site before the spring fry migrations had begun; only pairs not yet committed by late May would be able to select nesting sites on the basis of salmonid availability during the current year. If breeding pairs do u t i l i z e such an index to assess carrying capacity, a greater proportion of late-nesting birds (i.e. after mid-May) should be observed on enhanced streams. The distribution of hatching dates for rivers with > 4 broods are summarized in Fig. 53. The mean hatching date was later on BQ than on the other streams for 3 yr in succession. More birds began laying eggs 2 1 5 F i g . 52. Relationship between merganser brood density, drainage area of r i v e r system and ju v e n i l e salmonid production. A - multiple regression on drainage area and e a r l y (March - May) downstream salmonid migration a v a i l a b l e to mergansers p r i o r to egg-laying; dots indicate observations, c i r c l e s indicate predicted value. B - p a r t i a l regression of residuals ( a f t e r drainage area e f f e c t has been removed i n A) on early salmonid migration. C —- p a r t i a l regression on an index of resident salmonid density available to merganser broods ( i . e . average spawning escapement of salmon whose progeny inhabit freshwater over summer - mostly coho). Regressions: A and B: No. of broods = 0.769 + 0.029 (area, km2) + 0.110 (JSI X 106) n = 14, r = 0.95, * 2 , l l = 51'2> P < 0 - 0 0 0 1 C: No. of broods = 2.42 + 0.009 (area, km2) + 0.827 (coho spawn. X 103) n = 14, r = 0.88, F 2 n = 19.4, p < 0.001 Q-LC 1 — T- 1 1 1 1 ' 0 50 100 150 200 250 300 350 DRAINAGE AREA (km2 ) ACCESSIBLE TO BROODS 2 1 7 10 8 6 U + 21 B v LQ • BQ • ENG • Q-LC • WC, NC. IS, RC t = 6.83 p .0001 0 25 50 75 100 TOTAL JUVENILE SALMONID MIGRATION (JSI*106) 0 5 10 15 WILD SPAWNER INDEX (COHO * 103) 218 a f t e r May 15 (assuming a 45 day egg-laying/incubation period) on BQ than on the other streams (combined) but the difference i s only marginally s i g n i f i c a n t (p = .05, x2) Second, mergansers choosing nest s i t e s on the basis of early salmon migrations w i l l be deceived about the pr o d u c t i v i t y of (natural) salmon populations i n hatchery-enhanced streams. Food resources for the broods would then be less than expected and may l i m i t fledging success. Estimates of the number of downy young hatched and reared and of survivorship on major r i v e r s are summarized in Table VIII. The number of young reared was r e l a t i v e l y constant for each stream i n comparison to the number of broods emerging and t o t a l number of downy young hatched. In other words, s u r v i v a l was higher when fewer downy young were hatched. This trend i s preserved for independent estimates of survivorship based on d a i l y rates of su r v i v a l for well-documented broods on BQ and LQ ( F i g . 54). Of course, there i s no d i r e c t evidence that survivorship i s related to food supply, so other explanations are possi b l e . However, the committed aggregation hypothesis cannot be rejected on the basis of either p r e d i c t i o n ; more d i r e c t tests are required. 2 1 9 Fig. 53. Comparison of hatching dates in neighbouring streams. Arrows indicate means; broods represented to lef t of dashed vertical line were initiated prior to May 15. 220 Table VIII; Estimates of merganser brood production and survivorship on major streams censused, 1980 - 1982. Estimated Ducklings/Brood 3 Total Ducklings" S u r v i v a l 0 Number of River Year Broods Mean Std. Dev. N Hatched Reared P] P ? N ENG 81 11 8.5 4.3 10 94 59 .63 .62 5 82 7 9.5 2.1 4 67 59 .88 — — Q-LC 82 13 6.1 2.5 9 79 60 .76 .85 2 LQ 80 d 8 11.4 5.8 5 91 69 .76 — —. 81 10 7.9 .4.2 8 79 63 .80 .86 6 BQ 80 11 7.9 5.1 10 87 61 .70 .57 4 81 11 6.5 3.2 8 72 51 .71 .93 2 82 8 8.0 3.6 7 64 55 .86 1.00 2 a Size of broods at f i r s t sighting provided < 2 wk old. b Number hatched = avg. ducklings per brood multiplied by the number of broods; number reared = sum of number i n each brood at l a s t s i g h t i n g . c ?i = estimated number hatched/estimated number reared; duration not constant. P2 = mean proportion surviving from < 2 wk to > 5 wk of age (N broods observed) d Census coverage may not have been adequate; 2.5 km (20% of access, length) X 20 d (May 31-Aug. 1). 222 Fig. 54. Survivorship of merganser downy young during 1980 and 1981. Only broods observed at least 3 times are considered. A - broods reared on Big Qualicum system; B - broods reared on L i t t l e Qualicum system. The 1980 BQ brood with low survivorship (open triangles) was motherless after 15 d and is not included in the survivorship estimate. 2 2 3 20] A 15- \ 1980 11 BROODS 87 HATCHED S = 66 % oo oo—o-o-o-o—o 10- V \ 5 •-T- V - - v - w — v — v — V 0 15H o- -o-o 1981 11 BROODS 72 HATCHED S = 76 % 1CH V — v v • — e . 5 • — o V — V -0 -1 r -i r 0 10 20 30 40 50 60 DAYS AFTER HATCHING B o o 1980 8 BROODS 91 HATCHED S = 79 % DAYS AFTER HATCHING 2 2 5 DISCUSSION I have concluded that the dispersion of common merganser nests i s rel a t e d to the a v a i l a b i l i t y of juvenile salmonids in Vancouver Island streams. A s i m i l a r r e l a t i o n s h i p has been reported for the red-breasted merganser in Norway: Rad (1980) observed that breeding pairs inhabited only lakes which supported populations of three-spined stickleback (Gasterosteus aculeatus) or trout (Salmo spp); breeding pairs were not seen on morphologically-similar lakes lacking these species. Nesting dispersion i s also related to food resources among other species of duck (e.g. l o n g - t a i l e d duck, Clangula hyemalis, Pehrsson 1974 and goldeneye, Bucephala clangula, Eriksson 1978). But how might this r e l a t i o n s h i p occur? Rad (1980) and White (1957) argue that a p l e n t i f u l supply of small f i s h i n shallow water i s required for breeding success among mergansers; both authors conclude that f i s h density i s a proximate factor influencing nest s i t e s e l e c t i o n . The committed aggregation hypothesis advanced here, i s one mechanism by which f i s h density could operate as a proximate f a c t o r . Although breeding success i s influenced by food supply in many wild b i r d populations (e.g. Lack 1968, Newton 1980), there i s less experimental evidence that breeding density depends on food supply. For example, nesting de n s i t i e s of red-winged blackbirds (Agelaius phoeniceus) increased following supplemental feeding experiments (Ewald and Rohwer 1982) but crows (Corvus corone) and magpies (Pica pica) did not show the predicted response (Yom-Tov 1974, Hogstedt 1981). There i s much i n d i r e c t evidence that breeding density i s limited by 226 food supply among raptors (Galushin 1974, Parker 1974, Mclnvaille and Keith 1974, Newton 1976, Village 1982). The discrepancies between these studies are probably associated with the intensity of predation - whether territories ensure food supply or space out the nests as a defense against predators (Hinde 1956). Because mergansers nest in cavities and the young leave the nest within a few days of hatching, predation is unlikely to be a factor limiting nesting density. Alternatively, the number of mergansers nesting on a stream may be maintained at or below carrying capacity through a recruitment mechanism which does not invoke f i s h density as a proximate factor. Common mergansers tend to return to the vi c i n i t y of their natal stream to breed (Erskine 1972) and there is some evidence to suggest that individuals nest repeatedly on the same stream (Chap. 4). If fledging success is dependent on the availability of small f i s h , then more first-time breeders would return to streams where small f i s h were plentiful. In addition, mature females which experience poor breeding success in one year might seek out different nesting sites in subsequent years. Such behaviour has been reported for the sparrow hawk (Accipter missus, Newton and Marquiss 1982) and the goldeneye (Bucephala clangula, Dow and Fredga 1983). In either case, an equilibrium nesting density is guaranteed provided food supply limits brood survival. The process is best visualized in terms of a recruitment curve (e.g. Ricker 1954) where carrying capacity is determined by the intersection of the curve with the diagonal, replacement line (Fig. 55A). Streams with differing carrying capacities are represented by different curves of the same form. Under these assumptions, nesting densities for a population near equilibrium w i l l tend to l i e along the replacement line; that i s , under the 227 Fig. 55. The recruitment hypothesis: A - Equilibria exist at points where recruitment curves characteristic of each river system cross diagonal replacement line; for equilibrium breeding populations, the number of juvenile birds produced should be proportional to breeding density. Dotted line connects points of equal slope (i.e. equal reproductive success per breeding pair) for this family of curves; the less productive river, i i , is more favourable than river i i f the latter is more f u l l y occupied (e.g. in zone II whereas i i l i e s in zone I ) . B - Best estimates of juveniles produced vs breeding density (i.e. number of broods) for major rivers censused (data from Table VIII). No linear relationship is evident. 228 229 recruitment hypothesis, a linear correspondence is expected between nesting pairs and subsequent recruits, on average, within data collected from streams differing in carrying capacity. Of course, this relationship may exist anyway i f food resources do not limit the survival of broods. S t i l l , the prediction provides a way to f a l s i f y the recruitment hypothesis since a non-linear relationship would indicate that nesting density influences survival but that the number of breeding pairs is not at equilibrium with respect to stream carrying capacity. Juvenile mergansers disperse and mingle with those from neighbouring streams once they are able to f l y ; there is no reason to presume differential mortality among 'breeding stocks' once they have l e f t their natal streams. ,Consequently, recruitment to a particular stream can be regarded as the number of birds reared there, multiplied by a mortality rate common to a l l stocks (i.e. an arbitrary constant which merely determines slope of the recruitment curve). These data are plotted in Fig. 55B for streams censused in this study. There is no positive correlation between recruits and nesting density, in contradiction to the recruitment hypothesis. It is unlikely that a l l streams have comparable carrying capacity ( i . e . that the data indicate a single equilibruim) because drainage area accessible to the broods ranges from 44 km2 for BQ to 283 km2 for ENG; thus, the recruitment hypothesis seems improbable. The observed pattern in Fig. 55B is consistent, however, with the committed aggregation hypothesis even i f the density of breeding pairs in the larger v i c i n i t y i s constrained by recruitment due to philopatry. Hatchery releases of juvenile salmon would attract additional breeding pairs 230 on some rivers (i.e. BQ and Q-LC) yet would not result in a proportional increase in recruitment from those streams because hatchery fi s h are unavailable to the broods; the horizontal skew in Fig. 55B would result i f competition for food reduced brood survival such that overall recruitment increased only marginally. The preceding analysis suggests that salmonid density i s , indeed, a proximate factor in determining nesting dispersion of mergansers. It would be useful to know more precisely how, and when, mergansers select their nesting sites. For example, there is evidence that the size of territories defended by nuthatches (Sitta europaea) is based on an assessment of seed densities in the autumn, in anticipation of food shortages during the winter (Enoksson and Nilsson 1983). Such information would help to ascertain the extent to which salmonid enhancement f a c i l i t i e s attract additional breeding pairs thereby causing, indirectly, increased mortality for wild salmonid populations. Two observations suggest that the majority of mergansers rely on previous experience in choosing nest sites: f i r s t , yearling mergansers have been observed to examine potential nesting cavities during the late summer (White 1957 and pers. observ.); second, back-calculations from hatching dates indicate that many pairs are committed to nesting sites before the peak of juvenile salmonid migration. Nevertheless, brood density was correlated much more closely with overall (i.e. enhanced) salmonid production than with just natural production (Fig. 52). It seems probable that mergansers with prior experience in a nesting area would be influenced by their previous reproductive success whereas those exploring unfamiliar areas must make decisions according to the prevailing level of food resources. 231 The availability of food for the broods depends both on i n i t i a l f i s h density and the number of broods exploiting the resource. Consequently, the stakes for nest site decisions are much lower for early nesters than for late nesters. Most streams w i l l provide sufficient food for at least one merganser brood. There is no evidence that mergansers defend nesting territories; although broods and breeding pairs are typically spaced apart on streams (Fig 50, also Foreman 1976, White 1957), individual broods roam freely on the stream and do not maintain their relative position. Without t e r r i t o r i a l behaviour, early nesting birds would have no control over prey exploitation rate by other broods. Presumably a l l breeding pairs would experience the effects of exploitation but, clearly, late-nesting birds would suffer the most from depleted food resources; late-nesting birds should choose nest sites more carefully. At some breeding density, fledging success w i l l be higher on a stream with less f i s h . For example, the reproductive success of herring gulls (Larus argentatus) nesting in inferior habitat was found to be equivalent to that of gulls nesting in superior areas for analogous reasons (Pierotti 1982). The dotted, line in Fig. 55A joins an isocline of equally good nesting locations (i.e. choice of recruitment curve). One would expect late nesters to be in the best position to evaluate fish production and breeding pair density. But they are also more lik e l y to be deceived by salmon enhancement practices. Further study of the behaviour of individual (marked) birds is required to ascertain how these decisions are made. Any additional broods reared as a consequence of the aggregative response to enhanced salmon migrations, w i l l increase predation of wild juvenile salmonids. However, the intensity of predation on wild salmonids 232 w i l l depend on the extent of a l t e r n a t i v e feeding o p p o r t u n i t i e s w i t h i n the r i v e r system; where a l t e r n a t i v e prey are p l e n t i f u l , the pressure on w i l d salmonids w i l l be reduced. Nevertheless, predation by merganser broods i s p o t e n t i a l l y more se r i o u s f o r salmonid f i s h e r i e s than i s predation by the a d u l t s . The magnitude and s i g n i f i c a n c e of t h i s m o r t a l i t y i s examined w i t h respect to the B i g Qualicum System i n Chapter 7. PART III MERGANSER PREDATION AND SALMONID ENHANCEMENT V 234 CHAPTER 6 - MERGANSER PREDATION ON JUVENILE SALMONIDS DURING THEIR SEAWARD MIGRATION INTRODUCTION The common merganser i s a large and e f f i c i e n t predator of juvenile salmonids (Chap. 2); moreover, i t i s highly mobile and w i l l congregate and nest near p r o f i t a b l e feeding areas (Chap. 3 - 5 ) . Not s u r p r i s i n g l y , the merganser has been widely regarded as a menace to sport f i s h populations since the early part of t h i s century (e.g. Leonard and Shetter 1936, Beach 1937, Huntsman 1941, Elson 1962). Other investigators (e.g. Munro and Clemens 1937, Salyer and Lagler 1940) have emphasized that despite i t s great p o t e n t i a l as a predator of salmon, the merganser feeds more commonly on 'coarse' fishes and only threatens salmon and trout populations under exceptional conditions. The d e c i s i o n to enhance salmonid populations through f a c i l i t i e s such as spawning channels and hatcheries that create l o c a l concentrations of j u v e n i l e salmon many times greater than found in nature, introduces a new facet to t h i s controversy: Does mo r t a l i t y due to mergansers decrease as salmon populations are enhanced because the predators become swamped by prey (depensatory mortality) or increase because mergansers congregate and feed in larger and larger flocks (compensatory mortality)? The purpose of t h i s chapter i s to present quantitative estimates of salmonid mortality due to mergansers on a hatchery-stocked stream. The question of compensation vs depensation i s examined i n l i g h t of these estimates of mortality over a wide range of salmonid density. 2 3 5 The impact of predation by mergansers on fish a b l e salmon stocks w i l l depend not only on the proportion eaten, but on the opportunities for compensatory growth and mortality during the subsequent l i f e of surviving f i s h . I f , as widely suggested (Larkin 1974, MacLeod 1977, Peterman 1978), ocean s u r v i v a l and growth i s independent of salmonid density (over t y p i c a l values), a l l seaward migrants are equally valuable i n terms of t h e i r respective, fishable stocks; s p e c i f i c a l l y , mortality of hatchery-reared smolt or of f r y emigrating from a r t i f i c i a l spawning channels, w i l l cause a proportionate reduction i n the abundance of returning adults. Hence, i t i s desirable to minimize predation on these cohorts. On the other hand, there i s evidence that s u r v i v a l of salmonids that rear i n freshwater i s lim i t e d by the amount of suitable habitat (Chapman 1962,1965), p a r t i c u l a r l y during periods of low flow (Marshall and Br i t t o n 1982) and over winter (Holtby and Hartman 1982, Tschaplinski and Hartman 1982). Predation on these f i s h early i n the summer may have l i t t l e consequence for the eventual size of the cohort migrating to sea the following spring. For t h i s reason, attention i s r e s t r i c t e d to the former case - predation by adult mergansers during the period when juvenile salmonids migrate to sea ( A p r i l - June). The pote n t i a l for predation on stream-resident salmonids (salmonid f r y and parr) by merganser broods during June through September i s considered i n the following chapter. 2 3 6 METHODS Study Area: Merganser density was censused on two neighbouring streams (the Big Qualicum and L i t t l e Qualicum) at 1 - 4 d i n t e r v a l s during the downstream salmon migration ( A p r i l - July) in 1980 and 1981. Hatcheries and spawning channels e x i s t on both r i v e r s . Physical data and salmon production s t a t i s t i c s for each system are summarized i n Table VI; census techniques are described in Chap. 4. Fry emigration was enumerated d a i l y at counting fences maintained by Dept. of F i s h e r i e s and Oceans personnel. Smolt reared at the BQ hatchery were counted p r i o r to release but did not pass through the counting fence; however, Mace (1983 - Appendix) evaluated residence times within the r i v e r and has calculated smolt density (by species) following releases in 1979 - 1982. Computations for Predation: Daily appetite, functional response and s i z e - s e l e c t i o n c h a r a c t e r i s t i c s of adult mergansers are reviewed in Chap. 2. D a i l y consumption of f i s h was found to be 0.4 kg per adult. Mergansers observed on freshwater were assumed to feed e x c l u s i v e l y on salmonids; those on t i d a l waters, to feed only on other species. Although mergansers move f r e e l y between these two environments, census estimates are assumed to r e f l e c t the proportion of time spent in each. This i s computationally equivalent to assuming that only birds sighted on freshwater consumed salmonids during that day. Evidence i s presented below to support t h i s assumption. 2 3 7 Mortality of salmonids on any specified day was computed as follows: mortality of species i (%) = p^C / w ^ X 100% [6.1] where N is the number of mergansers on freshwater, C is consumption per bird per day (0.4 kg), n£ i s the number and w£, the average weight of prey species i migrating downstream that day and pi is the proportion of species i in the diet. Where a single prey species is overwhelmingly abundant compared to alternative prey, pi was assumed to be unity. Otherwise, three assumptions were made in order to bound mortality estimates for each size class: f i r s t , a lower bound for predation on large fish (smolt) was computed by assuming prey as taken in proportion to its density, regardless of size; second, an upper bound for predation of large fish was found by assuming prey was selected in proportion to its respresentation by biomass and f i n a l l y , a 'best' estimate was based on selection by apparent size (details in Chap. 2). In other words, p. = v.n. / J v.n. [6.2] r i i i I l where v^ = l implies selection in proportion to density, v i = wi implies selection in proportion to biomass; v i = ( l i / l f ) 2 under the apparent size selection model in which l i = length of prey species i and If is the length of the standard for comparison (chum fry, If = 3.8 cm). These relationships are illustrated in Fig. 59. 238 Single day estimates of mortality are calculated where pf 1 and plotted against prey density (n£). Where pi < 1, bounds on mortality were computed for the entire migration. Estimates of merganser abundance for the intervals between census days were obtained by interpolation. Estimates of fry density were obtained from counting fences on BQ and LQ. However, daily estimates of smolt densities were available only for BQ from diving surveys by Mace (1983, and pers. comm.). 239 RESULTS Dispersion of Foraging A c t i v i t y : Merganser density was highest on the lower kilometer of the BQ system throughout the downstream migration in both 1980 and 1981 ( F i g . 56). This i s the zone immediately below the BQ hatchery. Density declined farther upstream such that, at most, an addi t i o n a l 10 - 15% of the t o t a l count to 5 km would be expected on the remaining headwaters below the dam at Home Lake. Foraging a c t i v i t y was not correlated with time of day or tide height (Chap. 4). Moreover, the proportion of the t o t a l ( r i v e r system) count foraging on t i d a l waters did not change s i g n i f i c a n t l y with time of day (p > .10, Wilcoxon-Mann-Whitney t e s t , F i g . 57A) or tide height (p > .40, Wilcoxon-Mann-Whitney te s t , F i g . 57B). Hence, I conclude that the dispersion of mergansers on fresh as opposed to t i d a l waters ( F i g . 56) also r e f l e c t s the dispersion of foraging a c t i v i t y within the r i v e r system throughout the day. Mergansers observed on t i d a l waters r a r e l y ate salmonids; i n t e r t i d a l fishes - ' predominantly sculpins (Cottidae) and blennies (Stichaeidae or Pholidae) - comprised the majority of t h e i r d i e t . Those foraging on freshwater on BQ and Rosewall Creek (Chap. 2) were observed to feed almost e x c l u s i v e l y on salmonids while these were abundant. Fi s h remains were obtained from the crops and g u l l e t s of 14 mergansers captured on freshwater reaches of the BQ system and treated with an emetic drug (Methods, Chap. 4). Of these samples, only 2 (14%) revealed evidence of non-salraonid fi s h e s ; both contained a mixture of marine species and salmonid "fry. Undoubtedly, 240 Fig. 56. Dispersion of mergansers on the Big Qualicum R, April through June, 1980 and 1981. Numbers refer to number of census days; 95% C.I. calculated from arcsine-transformed data. The hatchery is located 1.3 km upstream from river mouth. 1980 r -Z ZD O o < o LL O z o h-or o o_ o or .2 0 .6-0 T T n = 22 (census days) 1981 TIDAL 0 1 \ 2 3 U 5 / km UPSTREAM 242 Fig. 57. The proportion of mergansers (of river-system count) foraging on the Big Qualicum estuary with respect to time of day (A) and tide height (B). A l l counts in comparison B began between 11:00 - 14:00 PST. 243 CO or LU < CD -z. CD < or o LL o o o LL O or o Q_ O or .20-.15 .10-^ .054 A o 1980 1981 p >.10 MORNING 10:00 h MID - DAY 10: -16:00 h EVENING 16:00 h PST B .15 .101 .051 all counts 11-00 -tt:00 h p >.10 0 < 2 m > 2 m TIDE HEIGHT (range 0.3-3.5 m) 244 freshwater sculpins are also eaten (Munro and Clemens 1937), but probably only when salmonids are r e l a t i v e l y scarce. Accordingly, estimates of predation i n the following analyses are based s o l e l y on freshwater counts. I also assume that mergansers on freshwater feed to s a t i a t i o n e x c l u s i v e l y on salmonids during the downstream migrations. This assumption establishes an upper bound but also provides a c r e d i b l e estimate during the periods when f i s h are released form hatcheries at very high concentration. Estimates of M o r t a l i t y : The most r e l i a b l e estimates of mortality have been computed for s i t u a t i o n s where the prey can be regarded as a single species by v i r t u e of i t s overwhelming abundance r e l a t i v e to other prey; that i s , p^ 1 with reference to equation [6.1]. These situ a t i o n s occurred during the chum f r y (0. keta) migration on the BQ and LQ systems - both of which have a r t i f i c i a l spawning channels and f r y enumeration f a c i l i t i e s - and during the coho (0. kisutch) and chinook (0. tshawytscha) smolt releases from the BQ hatchery in 1980 ( F i g . 58). Unfortunately, the coho and chinook releases overlapped i n BQ during 1981 so that the problem of s i z e - s e l e c t i o n cannot be circumvented. Despite the aggregative response exhibited by mergansers, mortality always decreased with increasing f i s h density. The e f f e c t of aggregation i s apparent, however, because estimates of d a i l y m o r tality l i e along a t r a j e c t o r y that i s more nearly l i n e a r and whose slope i s less negative than would be expected, were merganser density constant ( s o l i d l i n e s ) . The aggregative response also declines over the course of the season; estimates f a l l c l o s e r to the l i n e of constant density as the season progresses from April-May 245 Fig. 58. Relationships between % mortality and salmonid density over periods when a single species is very abundant relative to alternative prey. A - peak chum fry migration in Big Qualicum R. (Apr. 4 - May 3, 1981); B - peak chum fry migration in L i t t l e Qualicum R. (Apr. 4 - May 3, 1981); C - coho smolt and D - chinook smolt migration in Big Qualicum R . Solid lines indicate expected relationship i f merganser density remained constant. Overall estimates of mortality are computed by interpolation from daily estimates. OO 3 0 -o r LU 00 < O 2 0 -o r LU L U ZD Q >-I— 10-• m • A : BQ April U - May 3 o v e r a l l m o r t a l i t y * 6 . 1 % 30-2 0 10 B : LQ April U - May 3 o v e r a l l m o r t a l i t y - 6 . 6 % 0.2 0.4 0.6 DAILY CHUM FRY MIGRATION (*10b) o r o < X < 8-6-4-2-0 20 C:BQ COHO May 26-June 6 40 6 0 8-j 6 4 2 Too" D: BQ CHINOOK June. 8 - July 5 200 300 SMOLT DENSITY FOLLOWING RELEASE FROM HATCHERY (*103) ON Table IX: Estimates of salmonid mortality (%) by species due to adult mergansers during downstream migration in the Big Qualicum R., April 15 - July 15, 1980 and 1981. SIZE-SELECTION ASSUMPTION SPECIES Density 1980 1981 Apparent Size 3 1980 1981 Biomass 1980 1981 RANGE Emergent fry (chum, coho) (.33 g) 3.8 6.0 3.5 5.7 3.3 5.2 3.3 - 6.0 Reared chum fry (.9 g) 6.9 7.2 8.4 6.9 - 8.4 Reared chinook smolt ( 6 g) 1.3 0.9 1.2 0.8 1.2 0.7 . 0.7 - 1.3 Reared coho smolt (17 g) 1.2 0.8 1.5 1.0 1.7 1.1 0.8 - 1.7 Reared steelhead smolt (45 - 90 g) <.l <.l 0.3 0.8 2.1 5.6 <.l - 5.6 a JSI units (see text). 248 ( F i g . 58A and B) to May-June ( F i g . 58C) and f i n a l l y to June-July ( F i g . 58D). Bounded estimates for mortality on each species over the en t i r e migration period are summarized for BQ in Table IX. Each estimate represents maximum mort a l i t y under a d i f f e r e n t assumption about prey s i z e - s e l e c t i o n (see F i g . 59 and Methods). The s i z e - s e l e c t i o n assumption i s c r i t i c a l where emergent chum f r y (0.3 - 0.4 g) and large steelhead (Salmo gairdneri) smolt (45 - 90 g) are vulnerable simultaneously. In th i s case, mortality estimates for steelhead smolt with s e l e c t i o n i n proportion to biomass i s 2 orders of magnitude greater than for se l e c t i o n i n proportion to density. For other size comparisons, the estimates are more robust. In no case did mortality exceed 10%. Fish released or migrating e a r l y i n the year ( A p r i l - early May) sustained maximum losses, under the apparent size model, i n the range of 0.3 - 7.2%, whereas, those released i n lat e May - July suffered maximum losses of only 0.8 - 1.5%. It should be noted that these estimates are based on census counts to 5 km upstream of the r i v e r mouth; the t o t a l freshwater count, and thus maximum mortality i s l i k e l y to be 10 - 15% higher than computed for Table IX. E f f e c t of Size and Spacing of Fish Releases: The hatchery manager must release a large number of reared f i s h over a short season. Due to the number of f i s h involved and the e f f i c i e n c y with which mergansers capture salmonids at even low density (Chap. 2, F i g . 11), i t is i n e v i t a b l e that mergansers w i l l be able to feed to s a t i a t i o n under any fe a s i b l e release schedule. If merganser density were constant ( i . e . independent of f i s h density) and other things equal, the best release strategy i s c l e a r : a l l f i s h should be released simultaneously so as to allow 249 Fig. 59. Graphic representation of size-selection assumptions for computations in Table X (see pg. 237 for details). 2 5 0 2 5 1 mergansers the shortest possible time in which to gorge. Unfortunately, salmonids released from rearing ponds do not leave the stream immediately or p r e d i c t a b l y (Mace 1983 - Appendix) and mergansers do congregate wherever feeding opportunities are favourable. Given these constraints, i t i s not so obvious which release schedule w i l l minimize mortality by mergansers. It seems possible that a series of small releases may a t t r a c t fewer predators than a single large release. i The e f f e c t of size and spacing of f i s h releases on mortality was investigated in simulations using an equilibrium aggregation model described i n Chap. 4. Parameter values used in these simulations are i d e n t i c a l to those used i n other simulations for BQ where predicted trends compared favourably to observed trends in merganser abundance (Chap. 4, F i g . 44). The number of f i s h remaining i n freshwater following a release was assumed to decrease exponentially with time and two p l a u s i b l e parameter values were tested: z = .4 and .6 (for time measured in days) corresponding to mean residence times of 1.7 and 2.5 d r e s p e c t i v e l y . In each scenario, the manager must release 50,000 smolt within 20 d. He can do t h i s by allowing a single large release or up to 10 smaller (equal-sized) releases spaced as far apart as possible within the 20 d. The r e s u l t s of these t r i a l s are presented in F i g . 60A. The single release proved superior to a l l other schedules. This i s because most f i s h reach the sea before the merganser aggregation reaches i t s equilibrium s i z e . When releases are small and numerous, equilibrium i s achieved quickly following each release and the trade-off between the magnitude of the equilibrium aggregation and the number of times t h i s occurs i s roughly equivalent; consequently, the mortality curve reaches an asymptote. In another t r i a l , the e f f e c t of spacing was evaluated for equal-sized releases 2 5 2 (Fig. 60B). Again, the single large release (i.e. 0-spacing) is the best strategy. Because some time is required for the aggregation to build to equilibrium whereas fish density declines monotonically following release, mortality w i l l always be low i n i t i a l l y then increase dramatically as fewer and fewer fish remain vulnerable. In other words, fish exhibiting longer residence times following release are at far greater risk than those that emigrate quickly. 2 5 3 Fig. 60. Relationships between mortality expected due to merganser aggregation and the size (A) and spacing (B) of releases of hatchery fish. Dots - fish residence time parameter, z, = 0.4; circles - z = 0.6. 15 10 0 A — i 1 1 1 1 r 0 2 U _T 1 1 i — T — 6 8 10 NO. OF RELEASES IN 20 d AT MAXIMUM SPACING CD o — O 10 5H 0 0 2 U 6 8 INTERVAL (d) BETWEEN U EQUAL-SIZED R E L E A S E S 2 5 5 DISCUSSION The estimates of mortality due to mergansers hinge on a number of assumptions regarding the dispersion of foraging a c t i v i t y , diet and size s e l e c t i o n of prey. Before drawing conclusions about the impact of merganser depredations, i t i s prudent to consider whether these assumptions are j u s t i f i e d and how well they can be generalized to other s i t u a t i o n s . F i r s t , i s i t reasonable to ignore predation of juvenile salmonids by mergansers foraging on estuaries? There i s a great deal of information a v a i l a b l e regarding gut contents of mergansers feeding on a wide v a r i e t y of h a b i t a t s . V i r t u a l l y a l l these studies (White 1936, 1937, 1957; Munro and Clemens 1936, 1937, Salyer and Lagler 1940; Lindroth 1955; Latta and Sharkey 1966; Timken and Anderson 1969; M i l l e r 1973, Alexander 1979) indicate that mergansers eat salmonids wherever they are conspicuous r e l a t i v e to other species. In p a r t i c u l a r , White (1939) observed that the proportion of salmonids comprising the diet of red-breasted mergansers declined from 100% to 73% to 12% for birds k i l l e d on the headwaters, lower reaches and upper estuary, r e s p e c t i v e l y , of the Margaree R., Nova Sc o t i a . These data concur with my observations of d i e t among common mergansers on Vancouver Island streams. In t h e i r review a r t i c l e , Salyer and Lagler (1940) concluded that mergansers threaten salmon populations only when concentrated on productive salmon-rearing waters. These authors also reported that mergansers wintering on lakes and streams in Michigan tend to congregate on the lower reaches of r i v e r s , as observed on Vancouver Island during the spring and summer. 256 Second, assumptions about d a i l y consumption and s i z e - s e l e c t i o n of prey require substantiation. White (1957) reported the average d a i l y consumption of a tame, but wild, male merganser, to be 440 g or 38% of body weight; 4 other captive birds i n his study averaged 380 g/d or 30% of body weight. Other studies with captive birds suggest lower consumption rates (ranging from 19 - 26% of body weight, Latta and Sharkey 1966, M i l l e r 1973), but a l l these birds l o s t weight i n c a p t i v i t y ; indeed M i l l e r concluded that 450 g/d was reasonable (on metabolic grounds) for wild, active mergansers. Daily consumption was estimated to be 290 - 450 g or 20 - 40% of body weight depending on sex, by extrapolation from ingestion rates of wild birds feeding on coho smolt of known size (Chap. 2). Thus, the estimate of 400 g/d, used i n the present c a l c u l a t i o n s , seems j u s t i f i e d . Less information i s available regarding the s i z e - s e l e c t i o n of prey by mergansers. In general, mergansers appear to choose a disproportionate number of large f i s h compared with sizes t y p i c a l l y a v a i l a b l e (Salyer and Lagler 1940, Elson 1962, Alexander 1979). Mergansers also selected 40 g coho smolt over coho f r y when both were, stocked i n large, r e l a t i v e l y - n a t u r a l stream enclosures (Chap. 2). Selection i n the l a t t e r study could be explained i n terms of d i f f e r e n t i a l v i s i b i l i t y of prey; this r e s u l t provides the basis for the apparent-size s e l e c t i o n assumption i n the present an a l y s i s . Mergansers given a choice of trout ranging from 10 - 20 cm in small tanks (where apparent-size s e l e c t i o n e f f e c t s would have been n e g l i g i b l e ) exhibited a preference for f i s h of smallest g i r t h (Latta and Sharkey 1966). Evidently, s e l e c t i o n of large f i s h i n natural s i t u a t i o n s cannot be interpreted as preference. And c l e a r l y , the apparent-size assumption i s superior to the al t e r n a t i v e s at either extreme - s e l e c t i o n 2 5 ? in proportion to number or in proportion to biomass. The problem deserves -further study, however; size differences among species and age classes of emigrating salmonids can be very large and the s i z e - s e l e c t i o n assumption i s c r i t i c a l for estimating mortality when many size classes are vulnerable simultaneously. I f more were known, i t might be possible to buffer valuable species by releasing them in the presence of cheaper and more vulnerable spec i e s . Juvenile salmon are vulnerable to predation for a period of days, or at most, weeks during t h e i r downstream migration. A predator must be very abundant to i n f l i c t appreciable mortality on large populations which are at r i s k for so short a time. Mergansers rank among the largest ( i n terms of appetite) and most e f f i c i e n t predators of juvenile salmon; what i s more, they are r e l a t i v e l y common and congregate wherever salmon density i s high. Yet the o v e r a l l m o r tality due to mergansers i s depensatory - mergansers are simply swamped by the output from spawning channels and hatcheries. M o r t a l i t y due to mergansers i s very u n l i k e l y to have exceeded 8% for any p a r t i c u l a r stock during downstream migration (where s i z e - s e l e c t i o n e f f e c t s can be ignored). This represents a substantial number of f i s h . But since m o r t a l i t y i s depensatory and in view of the d i f f i c u l t y and expense of r e s t r i c t i n g or c o n t r o l l i n g merganser abundance on salmon streams (e.g. Elson 1962, Miegs and Rieck 1967) - not to mention j u s t i f i a b l e concerns on esthetic (Borgeson 1979) and e c o l o g i c a l grounds (Campbell 1979) - i t i s probably more reasonable to produce 5% more f i s h and accept losses to mergansers. Merganser predation could be minimized by reducing the period over which f i s h are vulnerable ( i . e . s i n g l e , large, forced releases) and by 258 delaying f i s h releases as much as possible. Other considerations w i l l constrain a manager's a b i l i t y to adopt these t a c t i c s but the trade-offs can be evaluated (see F i g . 60). The o v e r a l l success of such t a c t i c s w i l l depend upon the a b i l i t y of young salmon to disperse once at sea and upon the actions of estuarine and marine predators. I t i s s t i l l e s s e n t i a l to learn where, bottlenecks exist ( i n terms of migrant s u r v i v a l ) and whether these bottlenecks can be relieved . rather than merely postponed to a subsequent stage of development. It seems improbable that other freshwater predators impose more serious mortality on j u v e n i l e salmon during t h e i r downstream migration in Vancouver Island streams. While estimating merganser abundance, I had opportunity to census the density of other p o t e n t i a l freshwater predators. Otters, mink, heron, eagles and k i n g f i s h e r s are a l l e f f i c i e n t predators with large appetites (Alexander 1979) but were probably less common than mergansers on Vancouver Island. These species may cause s i g n i f i c a n t mortality among stream-resident salmonids which are vulnerable for long duration but t h e i r depredations are u n l i k e l y to be serious during the b r i e f period of downstream migration. Gu l l s , p a r t i c u l a r l y Bonaparte's g u l l s (Larus Philadelphia) are numerous and feed on juvenile salmon under favourable conditions following hatchery releases. However, g u l l s are opportunistic predators and with one exception, noted below, the d i s t r i b u t i o n and size of g u l l flocks appears to be unrelated to the downstream migration of salmon in the streams I have censused ( v i z . Rosewall Creek, N i l e Creek, L i t t l e Qualicum and Englishman r i v e r s ) . Mace (1983) has documented an aggregation response by Bonaparte's g u l l s to the chinook smolt migrations through the Big Qualicum estuary, over several consecutive years. But Big Qualicum 259 appears to be an exceptional case, perhaps because of the unusually high d e n s i t i e s of chinook smolt released there; Mace found no evidence of aggregation i n response to chinook smolt releases from the Capilano River hatchery during the same period. Even on the Big Qualicum, mortality due to predation by g u l l s was depensatory. F i n a l l y , there are few resident piscivorous fishes within Vancouver Island streams. Sculpins (e.g. Leptocottus armatus and Cottus asper) prey upon migrant salmon f r y but the i r impact i s thought to be small (Mace 1983, M. Jones pers. comm.). Although predation by lar g e r , stream-resident salmonids may cause appreciable losses of emigrating j u v e n i l e salmon i n some streams (e.g. Meacham and Clark 1979), there i s l i t t l e evidence to suggest that mortality i s compensatory. Probably no single species of freshwater predator i s capable of i n f l i c t i n g compensatory mortality on juvenile salmonids migrating downstream. I f th i s i s true, i t follows that predation by a l l freshwater predators, acting i n concert, must also be depensatory and that salmon populations cannot be lim i t e d by predation during t h e i r seaward migration; the vulnerable period during migration i s simply too short - at least for r e l a t i v e l y small, coastal r i v e r s . However, depensatory predation during seaward migration may help to maintain (temporarily) patterns of dominance among d i f f e r e n t year classes or populations of P a c i f i c salmon (Ricker 1962, Larkin 1971). Salmon runs severely reduced by catastrophe or overfishing w i l l experience higher mortality due to predators than populations near equilibrium and hence, may take much longer to return to equilibrium abundance than populations only s l i g h t l y perturbed from equilibrium. S i m i l a r l y , depensatory mortality w i l l favour s u r v i v a l of f i s h in unusually large runs and may provide bonuses for salmonid enhancement programs. 260 If salmon abundance is limited mortality probably occurs over a longer alevins and juveniles which rear in potential role of merganser broods in stream-resident salmonids is considered by predation in freshwater, the duration through predation of eggs, freshwater before smolting. The limiting freshwater production of in the following chapter. 261 CHAPTER 7 : PREDATION BY MERGANSER BROODS ON STREAM-RESIDENT JUVENILE SALMON INTRODUCTION Mergansers have been recognized as a potential threat to salmon and trout populations wherever they occur in concentration on wintering grounds or r a i s e broods during the summer months (e.g. White 1936, 1957; Beach 1937; Salyer and Lagler 1940; Elson 1950, 1962). The most convincing evidence that common mergansers l i m i t salmon production comes from two studies on A t l a n t i c streams. A t l a n t i c salmon smolt production was doubled following j u s t one year of merganser and k i n g f i s h e r control on Forest Glen Brook, Nova Scotia (White 1939, Huntsman 1941). Because A t l a n t i c salmon are vulnerable to merganser predation in freshwater for t y p i c a l l y 2 yr, but often 3 yr, greater increases could be anticipated from longer term co n t r o l experiments. The Forest Glen experiment was discontinued because i t was not possible to control for differences in spawning escapement over consecutive years. However, a second more intensive study was undertaken on a 16 km section of the P o l l e t t R. in New Brunswick (Elson 1950, 1962). In t h i s experiment, wiers were constructed to prevent natural spawning runs and known de n s i t i e s of salmon fry were planted from hatcheries over a 9 yr period. P r i o r to the control program, mergansers foraged on the r i v e r throughout the i c e - f r e e seasons but the most intense predation was due to merganser broods. Elson estimated that predation by broods during June through September was equivalent to that by adult mergansers throughout the remainder of the year. Merganser control began in the s i x t h year of the 262 study after monitoring salmon survival over the previous 5 yr; a l l predation by broods and 80% of predation by adults was eliminated. Salmon survival to smolting was 300 - 500% greater during the next 4 yr while protected from merganser predation; other fishes also increased in abundance. The fry of three species of Pacific salmon (chinook, Oncorhynchus tshawytscha; coho, 0. kisutch and sockeye, 0. nerka) also inhabit freshwater rearing areas for up to 3 yr (typically 1 yr) before migrating to sea as smolt. Because these rearing periods are generally shorter than for Atlantic salmon the impact of predation is expected to be less severe. Nevertheless, broods of the common merganser are sufficiently abundant on Vancouver Island streams to cause high mortality on stream-resident juvenile salmonids. There is also evidence that merganser broods are more abundant on streams where salmon populations have been enhanced by hatchery and spawning channel f a c i l i t i e s . I have suggested (Chap. 5) that nesting density is higher on these streams because breeding pairs congregate on streams with enhanced fry and smolt migrations early in the spring before egg-laying has begun. Enhanced stocks of salmon are reared in protected ponds and are not vulnerable to predation by merganser broods; consequently, any additional broods raised on these streams may lead to unusually severe mortality for wild salmonids. The purpose of this chapter is to demonstrate the potential for predation of wild salmonid stocks by common merganser broods on the coastal streams of Vancouver Island. Mortality estimates are derived from observations of the dispersion, biomass and foraging activity of merganser 263 broods between late May and early September. However, the significance of this mortality, in terms of its impact on subsequent smolt production, requires further investigation. f 264 METHODS Study Area: The density of mergansers and t h e i r broods was censused on 8 streams on the east coast of Vancouver Island during 1980 - 1982. Det a i l s pertaining to study area and merganser census techniques are given in Chapter 4; brood census procedures are described i n Chapter 5. Data for 4 r i v e r s are presented i n t h i s chapter: 3 of these, Big Qualicum (BQ), L i t t l e Qualicum (LQ) and Englishman (ENG), are comparable i n size but d i f f e r greatly with respect to salmonid production (Table VI); the fourth system (Quinsam - Lower Campbell, Q-LC) i s much larger i n terms of discharge, supports a large wild salmonid population and an intermediate-sized hatchery population. Computations for Predation by Juvenile Mergansers: Predation of salmonids in t i d a l waters by juvenile mergansers appears to be n e g l i g i b l e (see below). It was therefore necessary to estimate the d i s t r i b u t i o n of brood biomass ( i . e . appetite) on freshwater over the rearing season. The number and age of downy young observed on freshwater during censuses was used in conjunction with growth curves for merganser ducklings (Erskine 1971) to compute the observed d i s t r i b u t i o n s of biomass. However, a l l broods known to exist were seldom sighted on a single census so that the observed d i s t r i b u t i o n greatly underestimates the p o t e n t i a l for predation. The estimated d i s t r i b u t i o n of biomass on freshwater was reconstructed from the h i s t o r y of i n d i v i d u a l broods. Broods missed during a census were assumed to 265 have been upstream of the census area or hiding on freshwater i f last seen on freshwater, or on tidal waters i f last seen there. The number of downy young per brood was assumed to be equal to the number recorded at next sighting. These conventions provide conservative estimates in that missed broods were more lik e l y to have been on freshwater where i t is comparatively easy to escape detection and because the size of broods declines due to mortality (Chap. 5). Cumulative biomass estimates over the rearing period were obtained by interpolating over intervals between censuses in 1980 and 1981. Estimates are less reliable for 1982 because counts were less frequent and the cumulative estimate is the average (estimated) biomass of weekly censuses multiplied by the duration of the rearing period. Daily appetite for juvenile mergansers was estimated from the frequency, duration and success of foraging bouts of broods observed at the mouth of the LQ river (site illustrated in Fig. 3B). Small sculpins (primarily 0+ age class staghorn sculpins, Leptocottus armatus) were plentiful at this site and comprised most of the ration of the broods observed. Sculpin density was assumed to be constant over the periods of observation - June 10-12, 17, 19 and 24 - 26, 1981. Feeding success was evaluated without d i f f i c u l t y from a raised blind approximately 50 m distant using 12x binoculars. In addition, the position and activity of each brood on the estuary was recorded at 20 min intervals. A l l broods could be identified readily by the age and number of downy young; age ranged from 10 - 35 d and number per brood, from 1 - 15. Data regarding capture rate and the proportion of time spent foraging were used to compute daily ration of small downy young. Daily consumption was estimated for other age groups by interpolation between this figure and estimates for fully-fledged immature birds. The estimate of daily 266 consumption, as a proportion of body weight, corresponding to the' mean age of the biomass distributions on freshwater was multiplied by the estimated biomass to find potential consumption. Next, mortality of resident coho was computed for the duration of the brood-rearing period subsequent to coho fry emigration (June 10 - August 25) by making assumptions about the proportion of coho fry in the diet, and fall-winter survivorship of uneaten parr. These assumptions are outlined in a later section. t ( 26? RESULTS D i s p e r s i o n o f F o r a g i n g A c t i v i t y : Me r gan se r b r o o d s f o r a g e d t h r o u g h o u t the f r e s h w a t e r and t i d a l zones o f a l l t h e s t r eams s t u d i e d . However, a g r e a t e r c o n c e n t r a t i o n o f b r ood b i o m a s s , and hence a p p e t i t e , was o b s e r v e d on t he l o w e r f r e s h w a t e r r e a c h e s o r t i d a l w a t e r s ( F i g . 6 1 ) . Me r gan se r b r ood s were more f r e q u e n t l y seen on f r e s h w a t e r on t h e BQ s y s tem t h a n on n e i g h b o u r i n g s y s t e m s . Young d u c k l i n g s f r e q u e n t e d t he f r e s h w a t e r r e a c h e s more commonly t h a n o l d e r d u c k l i n g s ; as t he b r ood s grew t h e y spen t an i n c r e a s i n g p r o p o r t i o n o f t ime on t i d a l w a t e r s . I n t e rms o f b i o m a s s , howeve r , t h o s e o f i n t e r m e d i a t e age (25 - 30 d ) were p redom inan t on f r e s h w a t e r ( F i g . 6 2 ) . The re were no o b v i o u s d i f f e r e n c e s i n t he age/b iomass d i s t r i b u t i o n o f b r ood s on f r e s h w a t e r among the 4 r i v e r s s t u d i e d . Not a l l b r o o d s were seen d u r i n g each c e n s u s . E s t i m a t e s o f ( t o t a l ) b r ood b i omas s on f r e s h w a t e r were r e c o n s t r u c t e d f rom the h i s t o r y o f i n d i v i d u a l b r o o d s ( s ee Me thod s ) and a r e summar ized i n F i g . 6 3 . F l u c t u a t i o n s i n b iomas s t oward t h e end o f the r e a r i n g s ea son r e s u l t f r om l a r g e b r ood s o f n e a r l y - f l e d g e d j u v e n i l e s mov ing between f r e s h and t i d a l w a t e r s . I n t e rms o f b i o m a s s , 51 - 71% o f b r ood deve lopment t ook p l a c e on f r e s h w a t e r on BQ compared t o 7 - 45% on LQ and 13 - 34% on ENG. S i n c e t he age/b iomass d i s t r i b u t i o n on f r e s h w a t e r i s s i m i l a r among t h e s e s t r e a m s , i t f o l l o w s t h a t a g r e a t e r p r o p o r t i o n o f b r ood s o f a l l ages u t i l i z e d the f r e s h w a t e r h a b i t a t on the BQ s y s t em t h a n d i d b r o o d s on LQ o r ENG. 268 Fig. 61. Biomass and dispersion of merganser broods (< 50 d) observed on 3 neighbouring streams: A - Big Qualicum R; B - L i t t l e Qualicum R; C - Englishman R. S t i p p l e - 1980; shaded - 1981; white - 1982. 269 0 - 1.0 km MAY JUNE JULY AUGUST 2 7 0 "a o VI CD o >->-o a to CO < o I—I CQ 10 H B o 5 0 5 0 -•—v 201 10 0 MAY 2.5 - 5.0 km (1981 only) -i ~ n r 1.0- 2.5 km ~i r T r 0-1.0 km TIDAL ZONE E H 3 1981 1980 JUNE JULY —1 3 r AUGUST 2 ? 1 5 0 •2.5-5.0 km 1.0-2.5 km 101 _ 0 1 r 0-1.0 km • r 20 H 10 H TIDAL ZONE \MM 1981 MAY JUNE JULY AUGUST 272 F i g . 62. Age d i s t r i b u t i o n s of merganser brood biomass ( i . e . a p p e t i t e ) observed on freshwater. A - Big Qualicum, 1980 - 1982; B - L i t t l e Qualicum (1980-1981), C - Englishman (1981-1982) and D - Quinsam-Lower Campbell system (1982). Arrows i n d i c a t e means; numbers r e f e r to number of census days, May 25 - Aug. 25. BIOMASS ON FRESHWATER (kg) 2?4 Fig. 63. Estimated biomass of merganser broods (< 50 d) on freshwater reaches of 3 neighboring streams. Census coverage was to 5 km upstream of river mouth. Stipple - 1980; black - 1981; white - 1982. Fluctuations are due to movements of large, nearly-fledged broods moving between fresh and tidal waters. Percentage figures refer to proportion of total biomass reared on freshwater. 276 Diet: It i s d i f f i c u l t to observe predation by broods on freshwater because they move continuously while foraging and are soon l o s t from sight. My observations on BQ suggest that resident salmon f r y and parr, p a r t i c u l a r l y coho, comprise most of the diet of broods of a l l ages foraging on freshwater. This i s not su r p r i s i n g because coho f r y are also the most conspicuous prey to a human observer. C e r t a i n l y , merganser downy young have no d i f f i c u l t y capturing salmonid f r y . A newly-hatched merganser (< 4 d) was able to pursue, k i l l , and aft e r some d i f f i c u l t y , swallow two 2 g coho parr from among 10 parr stocked in a tank approximately 2 m in diameter, and 0.2 m deep. On the BQ system, I have observed a brood of 18 (approx. 40 d old) consume 13 salmonid fry i n < 20 s. Crop and g u l l e t contents from 2 juveniles approximately 45 d ol d , captured on freshwater on the BQ system, Aug.10, 1980, reveal that both birds had fed on salmonid f r y ; no evidence of other f i s h species was found. No doubt, sculpins and aquatic insects are also eaten by broods on freshwater (Munro and Clemens 1937), but the importance of salmonid f r y in the d i e t should not be underestimated wherever f r y are abundant (Lindroth 1955, Elson 1962, Alexander 1979). On t i d a l waters, the foraging a c t i v i t y of broods i s comparatively easy to monitor. Small sculpins and f l a t f i s h (Bothidae or Pleuronectidae) were the most frequently consumed prey on the LQ estuary during my observations there June 10 - 26, 1981. Juvenile mergansers (< 50 d) were never observed to feed on salmonids during t h i s time. 3 2 7 7 Daily Consumption: Broods foraged throughout the observation periods, from 09:30 - 12:30 h and 15:30 - 18:30 h PST (Fig. 64A). The proportion of time spent foraging declined with age for broods 10 - 35 d (p < .005, F test, Fig. 64B). However, capture rate for small sculpins was not correlated with age (p > .25, t test) nor brood size (Chap. 2, Fig. 16). On average, juvenile mergansers foraging at the mouth of LQ consumed 0.29 i .01 (95% C.I.) young—of—the—year sculpins per minute. Assuming an average weight of 1.5 g/fish (derived from length distributions, Mace 1983 and a length-weight relationship, Mason and Machidori 1976), the average rate of consumption while foraging was 0.4 g/min. Although small (< 20 d) and large ducklings were equally successful at catching small sculpins, the older ducklings were able to supplement their diet with prey items too large for the young birds. Consequently, estimates of daily consumption based on the average capture rate of small sculpins, while foraging, multiplied by the proportion of time foraging during 12 h of daylight, w i l l underestimate the appetite of larger juveniles. I assume that the estimate is representative of ducklings at 10 d of age and have corrected for older juveniles by interpolating between this estimate, and the average daily appetite reported for fully-fledged, immature birds held in captivity (White 1957, Latta and Sharkey 1966). These interpolated estimates of consumption are expressed as a percent of body weight (Fig. 65). Estimates of Mortality: In order to estimate mortality of salmonids due to merganser broods, i t was assumed, as for adults, that salmonid fry are eaten only by juveniles 278 F i g . 64. Foraging a c t i v i t y of merganser broods on the L i t t l e Qualicum estuary with respect to time of day (A) and age of brood (B). Numbers in A refer to the number of broods observed. Regression i n B i s l i n e a r for arcsine-transformed data: y = 54.1 - 0.628x, n = 21, r = -.62, F. . q = 11.8, p < 0.005 % TIME SPENT FORAGING BY INDIVIDUAL BROODS % BROODS FORAGING cn o o o _i i i i—i—u. -V o . o o ro o o o o CO o o oo o o •'.•15* 3 I O GO on ro ^3 O oo > ro 280 Fig. 65. Daily consumption by merganser ducklings with respect to age. Solid line - interpolated estimates using observed capture rates of 0+ age sculpins by 11 d old ducklings (Fig. 64) and reported estimates for fully-fledged, immature birds. Dotted lines - upper and lower bounds from bioenergetic calculations (see text). Other symbols indicate consumption by captive birds. / • White 1957 (open circle - wild bird) • Latta and Sharkey 1966, Miller 1973 • Atkinson and Hewitt 1978 extrapolated from consumption in wild modal age on freshwater X 20 30 40 AGE (d) metabolic requirement 50 v 60 FULL SIZE 282 foraging on freshwater. The average age (weighted by biomass) of juvenile mergansers on freshwater is 27 d (Fig. 62A and B) which corresponds to an average daily consumption, C _> 40% of body weight. The potential appetite (A) of broods on freshwater over the rearing period is given by : A = C £ B.t./ w. [6.3] . 1 1 i l where B^ is the average biomass of broods, w^ , the average weight of coho fry and t.£, the duration (in days) of the i t n time period (approx. 1 mo. each). The proportion of coho fry in the diet, D, must also be specified; a value of 50% (by weight) has been selected arbitrarily (as an example) but seems reasonable. It is also necessary to consider the expectation of overwinter survival so that predation of fry can be evaluated in terms of smolt production. Estimates of survival rate from July to April in 3 Oregon streams average 19% (Chapman 1965); survival rates over the same period in Carnation Creek, Vancouver Is. averaged 24% (17 - 35%, 95% C.I.) from 1970 - 1981 (Holtby and Hartman 1982). If survival rate, S, is independent of density, consumption by broods can be computed as follows: consumption ('smolt-equiv') = DSA [6.4] These values are tabulated for D = 0.5 and S = 0.2 (with 95% C.I., .17 - .35) in Table X. Under these assumptions, mortality for the BQ system ranges from 18 - 31% of realized coho smolt production. Wild smolt migrations are not enumerated on either LQ or ENG so that (percentage) mortality cannot be estimated. Table X: Estimates of predation of coho parr by merganser broods on freshwater, June 10 - August 25. Number of Smolt--Equiv. Eaten 3 (xlO 3) Observed Number of Wild Smolt Parr Eaten S = .2 S = .17 - .35 Production'7 Estimated River Year (D = .5) (mean) (95% C.I.) (xlO 3) Mortality BQ 80 69 14 12 - 24 51.2 .26 81 82 16 14 - 29 85.0 .18 82 111 22 19 - 39 71.3 .31 LQ 80 14 3 2 - 5 81 43 9 7 - 15 ENG 81 39 8 6 - 14 82 50 10 9 - 18 — — a Consumption of smolt-equivalents = ^ 1 1 CDS w i where t^, B^ and w^ are the number of days, the average biomass of broods on freshwater, and the average weight of coho parr in period (month) i , respectively. C is the appetite of broods (40% of body weight/d), D, the proportion of coho parr in the diet, and S, the overwinter survival of uneaten parr from July to April. b from fence counts at BQ in the following year. Wild production from areas below the main fence was not estimated in 1981 - 1983 and is not included. Previous estimates indicate that > 90% of total wild production passes through main fence, although recent stocking of a lower tributary (Hunt's Creek) has probably reduced this figure. 284 DISCUSSION In c a l c u l a t i n g mortality due to merganser broods, I have attempted to underestimate, rather than overestimate, rates of predation. I t i s very u n l i k e l y that estimates of the biomass of broods on freshwater exceed the true values since conservative procedures were used in the computations (see Methods). The greatest uncertainty arises in estimating the d i e t and d a i l y consumption of merganser ducklings and the prospects for s u r v i v a l of uneaten f i s h during the remainder of the freshwater rearing period. Each of these assumptions warrants further consideration. Stomach contents have been analyzed from over 4000 juvenile and adult mergansers foraging in a v a r i e t y of habitats (White 1936, 1937, 1957; Munro and Clemens 1936, 1937; Leonard and Shetter 1936, Lindroth 1955, M i l l s 1962, Timken and Anderson 1969, M i l l e r 1973, Alexander 1979). V i r t u a l l y a l l these studies indicate that mergansers eat whatever f i s h are abundant; but on productive salmonid-rearing waters, juvenile salmonids tend to be selected over other f i s h species (Salyer and Lagler 1940, Lindroth 1955, Elson 1962, Alexander 1979). Brown trout (Salmo trutta) and creek chub (Semotilus atromaculatus) were eaten more frequently by captive juvenile mergansers than mottled sculpins (Cottus b a i r d i ) of s i m i l a r size (Latta and Sharkey 1966). Information on the diet of merganser ducklings i s less extensive but i s s t i l l convincing. A l l 20 merganser ducklings, up to one-half grown, captured on the Margaree and Cheticamp r i v e r systems, Cape Breton Island, 285 contained salmonids but no other fishes (White 1936, 1937). Salmon and trout are the dominant fish species in these rivers. Fish remains in stomachs from 118 ducklings shot on 6 New Brunswick streams which support a diverse freshwater fish fauna, were classified as 62% minnows (Cyprinidae), 30% salmonids and 7% suckers (Catostomidae). Insects comprised less that 50% by volume, on average, of the diet of newly-hatched ducklings < 110 g and were not common in the stomachs of the older ducklings (White 1957). Cyprinids are generally absent from Vancouver Island streams (Scott and Crossman 1973) and, except for sculpins, salmonids predominate in the streams I have studied. Accordingly, there is no evidence to suggest that juvenile salmonids would comprise < 30% of the diet of young mergansers on these streams; in fact, the data from Cape Breton Island where salmonids predominated, and my own observations on BQ, suggest that salmonids comprise the majority of the diet by weight. If this is true, estimates of mortality based on 50% representation in the diet, by weight, w i l l err on the low side. Daily consumption by young mergansers can be estimated in three ways: f i r s t by considering the proportion of time spent foraging and the average rate of ingestion while foraging; second, by calculating energy requirements from information about growth and metabolic rates, and f i n a l l y , by monitoring the consumption of juveniles raised in captivity. Foreman (1976) reported that 45% of broods censused on rivers in northern California were foraging (excluding those fleeing when f i r s t sighted). This is in agreement with the mean (44%) of data presented in Fig. 64 which provide the basis for my estimate of daily consumption. I am 286 not aware of any other data regarding t y p i c a l rates of ingestion by merganser ducklings while foraging. Basal metabolic rate for non-passerine birds can be expressed in terms of body weight: M = 78.3W.724 where M = kcal/24 h and W i s body weight in kilograms (Lasiewski and Dawson 1967). Young mergansers at the mode of the biomass/age d i s t r i b u t i o n on freshwater (27 d) weigh about 450 g. Accordingly, t h e i r BMR i s approximately 44 kcal/24 h. I f 45% of the daylight period (12 - 16 h) were spent a c t i v e l y foraging, and another 25% swimming and preening (Foreman 1976, pers. observ.) at a metabolic rate averaging 3 - 5 times BMR (Yom-Tov 1974), and the rest of the time was passed i n a c t i v e l y at BMR, the o v e r a l l metabolic rate would be on the order of 75 - 130 kcal/24 h. At t h i s age, mergansers increase in weight by roughly 20 g/d (Erskine 1971) which corresponds to an addi t i o n a l c a l o r i c requirement (at 1.25 kcal/g, Yom-Tov 1974) of 25 kcal/24 h. Hence, the t o t a l d a i l y c a l o r i c requirement, assuming 80% a s s i m i l a t i o n e f f i c i e n c y (King and Farner 1961), would be 125 - 190 k c a l . Since f r e s h l y - k i l l e d salmon f r y y i e l d 0.91 kcal/g (Brett 1973, a f t e r subtracting energy l o s t to s p e c i f i c dynamic a c t i v i t y - 30% for proteins and 10% for l i p i d s , Harper 1971), d a i l y consumption by 4 wk old mergansers should be at least 140 - 210 g or 31 - 47% of body weight. Although t h i s procedure provides only a gross approximation, i t does show that a d a i l y consumption of 40% of body weight i s not u n r e a l i s t i c from bioenergetic considerations. Additional support comes from f i s h consumption by young mergansers in c a p t i v i t y . Four merganser ducklings captured at 160 g (about 1 wk old) consumed, on average, 47% of th e i r body weight in f i s h over the next 3 d 28? (White 1957). Daily fish consumption by another tame, but wild, juvenile bird (1150 g) averaged 38.5% of i t s body weight (White 1957). Red-breasted mergansers, hatched and reared in captivity, consumed 74 - 83% of body weight/d during their f i r s t 20 d, 43 - 49% of body weight/d from 21 - 40 d of age and 25 - 35% of body weight/d from 41 - 60 d (from data in Atkinson and Hewitt 1978). These data are summarized in Fig. 65 to permit comparison with the curve I have derived from observations of wild birds. The derived curve may overestimate consumption by very young birds although activity, and hence appetite, is li k e l y to be higher among wild birds than those retained in cages. Bounds from bioenergetic calculations (as described above) are also plotted against age in Fig. 65. The estimate of 40% of body weight at the modal age on freshwater is supported by a l l the alternative data. If young ducklings, in the wild, consume 60 - 80% of their body weight/d, again, mortality due to broods w i l l have been underestimated. It is evident from the preceding discussion that predation by merganser broods must be regarded as a significant source of mortality among wild salmon parr populations in the BQ system. Yet following a survey of streams throughout British Columbia, Munro and Clemens (1937) concluded that predation by mergansers was not a serious problem for Pacific salmon populations. I do not dispute their conclusion with regard to predation of seaward migrants (Chap. 6); however, their estimates of merganser brood density are extremely low in comparison with estimates from the present study. While i t is conceivable that nesting densities have increased over the past 45 years, i t seems more likely that their census was inadequate. The sampling curves in Fig. 47 indicate that a river must be censused frequently (at least once a week) to obtain reliable estimates. Also, in 288 analyzing stomach samples, Munro and Clemens did not d i s t i n g u i s h between predation on freshwater where salmonids predominate i n the d i e t and predation in t i d a l waters where other prey are taken. I f most of t h e i r specimens were c o l l e c t e d on t i d a l waters (where mergansers are most conspicuous and least wary) the incidence of salmon in stomachs would have been underestimated. Elson (1962) has also expressed doubts about Munro and Clemen's conclusion with regard to the impact of mergansers on stream-resident salmonids in the Cowichan R. system. The p o t e n t i a l for predation by merganser broods i s considerably higher on BQ than on either LQ or ENG. In part, t h i s i s due to higher nesting density on t h i s stream, possibly because breeding pairs aggregate there in response to enhanced fr y and smolt migrations during March - May (Chap. 5). But also, within the BQ system, a greater proportion of broods are found on freshwater as opposed to t i d a l waters. The BQ estuary i s small in comparison with those of LQ and ENG, so that high freshwater d e n s i t i e s on BQ may r e f l e c t a s c a r c i t y of a l t e r n a t i v e prey for the broods. If this i s true, then m o r t a l i t y on wild salmonids in BQ might be unusually intense as a consequence of the combined e f f e c t of high nesting density and s c a r c i t y of non-salmonid prey species. A l t e r n a t i v e l y , broods may forage more in t e n s i v e l y on freshwater on the BQ system because wild salmonids are more abundant there than at either LQ or ENG. (Unfortunately, wild smolt migrations are not enumerated re g u l a r l y i n the l a t t e r streams). In t h i s case, (percentage) mortality on BQ would be representative of other streams. It i s not clear whether mortality due to merganser broods has any e f f e c t on the eventual size of smolt migrations in Vancouver Island 289 streams. Because the broods k i l l salmonid parr during the summer months, there is ample opportunity (30 - 40 wk) for compensation to occur i f overwintering survival is further limited by food or habitat (e.g. Chapman 1965). Subsequent mortality appears to depend largely on the minimum rearing area during periods of low flow (Marshall and Britton 1982) and physical displacement during freshets (e.g. Scrivener and Andersen 1982). Because the risk of physical displacement during freshet depends on the availability of cover (Tschaplinski and Hartman 1982), and the risk of behavioural displacement during low flow depends on availability of suitable rearing territory (Chapman 1962, Hartman 1965), mortality in both cases w i l l be determined by the number of 'surplus' individuals and w i l l , therefore, be compensatory. Provided merganser broods remove only surplus individuals, their predation may have l i t t l e consequence for subsequent smolt production. Merganser-control experiments on Atlantic streams proved very successful, but there are important differences between the Atlantic and Pacific situations. Atlantic salmon remain in freshwater for 2 - 3 yr (as do steelhead trout in Pacific streams); even i f mortality during the f i r s t winter reduced resident populations below the level determined by brood predation, additional predation during the next summer would diminish the cohort s t i l l further. Also, freshet mortality is probably higher in Pacific than Atlantic streams. For these reasons, i t seems unlikely that merganser broods limit the freshwater production of Pacific salmon (excluding steelhead trout) to the same extent that they appear to limit survival of Atlantic salmon in some rivers. Nevertheless, the mortality due to broods is substantial and further investigation is warranted. At present, there are not sufficient biological grounds to justify routine elimination of 290 merganser broods. Further study is required to discover the manner in which flow regimes, cover, food supply and predation interact to influence survival in freshwater. To this end, i t would be worthwhile controlling predation by merganser broods as part of a more general, experimental plan whereby other sources of mortality are also controlled. 2 9 1 CHAPTER 8 : CONCLUDING REMARKS Throughout t h i s study, I have advocated the importance of understanding predation in terms of the a v a i l a b i l i t y of prey. Many factors influence the i n t e n s i t y of predation in p a r t i c u l a r s i t u a t i o n s , but the basic responses to prey density are fundamental to a l l s i t u a t i o n s . The concepts of functional response and aggregative response have been cen t r a l to attempts at b i o l o g i c a l control of insect pests for many years (Watt 1968, Hassell 1980). But s i m i l a r investigations have r a r e l y (and only recently) been conducted with regard to predators of juvenile salmonids, despite the importance of the resource, the number of p o t e n t i a l predators and widespread enhancement p o l i c i e s that perturb salmonid de n s i t i e s by orders of magnitude (but see Peterman and Gatto 1978, Woodsworth 1982, Mace 1983). I hope that the present study w i l l provide a more general perspective from which to assess predation by mergansers. Some of the most general conclusions relevant to salmon management are reviewed below. Mergansers are e f f i c i e n t predators of j u v e n i l e salmonids and tend to congregate wherever f i s h are abundant. M o r t a l i t y during the seaward migration of enhanced salmon stocks w i l l , i n general, be depensatory because only a f r a c t i o n of the o v e r a l l migration (cohort) i s at r i s k on a p a r t i c u l a r day. The s i t u a t i o n i s quite d i f f e r e n t wherever a f i s h stock i s vulnerable for many days; hatchery f i s h reared in ponds or planted in lakes at high density, without protection, may experience very high mortality (e.g. Beach 1937, Elson 1962). Moreover, this mortality w i l l probably be compensatory due to the aggregative response exhibited by mergansers. Merganser broods 292 may also i n f l i c t compensatory mortality on stream-resident salmonids as a result of two mechanisms for aggregative response: f i r s t , the broods are like l y to spend more time foraging on freshwater for salmonids (as opposed to alternative species in t i d a l waters) when salmonids are abundant; second, the number of broods raised on a river system i s correlated with the availability of small f i s h . Merganser ducklings are capable of removing a large proportion of stream-resident salmonids; accordingly, further attention should be given to the implications of mortality caused by merganser broods, especially on enhanced or naturally-productive salmon streams. Merganser-removal experiments are not useful unless there is rigorous control over the many other, potentially important variables that influence salmon survival -i otherwise, results w i l l be d i f f i c u l t to interpret and generalize to other systems. A functional components approach i s recommended to c l a r i f y mechanisms by which predation, food supply and abiotic factors interact to determine limits for salmonid production in freshwater; such knowledge would be of general s c i e n t i f i c and practical value. Tactics to minimize predation can be adopted once properties of the aggregative response are understood. Mergansers forage as individuals (at least during the nesting season) and therefore, accumulate at profitable sites slowly, but predictably, compared with other species that forage in social groups (e.g. Bonaparte's gulls, Mace 1983). Short, forced fish releases w i l l usually escape serious predation by mergansers. In contrast, f i s h that tend to remain in freshwater following release w i l l not only be at risk longer, but w i l l become increasingly vulnerable as more mergansers 2 9 3 discover the site and fewer fish remain to share the risk. Aggregations build more slowly and grow less large in June than in April because merganser abundance declines throughout the spring; consequently, fish released late in the spring w i l l sustain less mortality due to mergansers. Aggregations w i l l build to equilibrium for fish populations continuously exposed to predation. Yet, there may s t i l l be tactics to reduce the equilibrium size of these aggregations. The present analysis is inadequate to predict responses to several patches, each with different (and changing) rank in pr o f i t a b i l i t y . It seems lik e l y that the aggregative response w i l l be constrained at some larger spatial scale. As an extreme example, suppose the salmon productivity of every stream on Vancouver Island were doubled; would the density of mergansers increase through immigration from other geographic areas? The implication, here, is that enhancement programs may be more successful i f efforts are concentrated rather than dispersed. At present, not enough is known about the behaviour of mergansers, nor other predators of salmon, to answer questions on these larger spatial scales. Routine control of merganser populations cannot be justified on biological grounds until limits to their dispersal are established. Further research would be required to determine how far afield mergansers venture while searching for food. Limited data from birds banded during this study suggest that mergansers frequently travel > 10 km/d, yet most marked birds were observed near the site of their release up to 2 yr later. With greater knowledge regarding the behavioural constraints on dispersal, i t should be possible to develop multi-patch models of aggregation. 2 9 4 The greatest obstacle to understanding ecological phenomena at large spatial scales is probably not increased complexity, but rather the d i f f i c u l t y of controlling confounding variables; the expense and uncertainty involved in manipulating largescale features of the environment are usually prohibitive. Nevertheless, ecological theory can only provide useful tools for resource management i f hypotheses developed in laboratory and small fi e l d studies are evaluated in larger contexts. The Salmonid Enhancement Program undertaken by the Canadian Department of Fisheries and Oceans offers many opportunities and 'natural experiments' to test predator-prey, and other ecological theory, on a scale that has seldom been possible. The present study is a small, but hopefully significant, contribution towards these larger objectives. 295 LITERATURE CITED Alexander, G.R. 1979. Predators of f i s h i n coldwater streams. pp. 153-170 in R.H. Stroud and H. Clepper (eds.) Predator-Prey Systems i n F i s h e r i e s Management. Sport Fishing I n s t i t u t e , Washington, D.C. Altmann, J . 1974. Observational study of behaviour: sampling methods. Behav. 49: 227-267. Atkinson, K.M. and D.P. Hewitt. 1978. A note on the food consumption of the red-breasted merganser. Wildfowl 29: 87-91. Barnard, C.J. 1980. Equilibrium flock s i z e and factors a f f e c t i n g a r r i v a l and departure in feeding house sparrows. Anim. Behav. 28: 503-511. Bartholomew, G.A. 1942. The f i s h i n g a c t i v i t i e s of double-crested cormorants on San Francisco Bay. Condor 44:13-21. Beach, U.S. 1937. The destruction of trout by f i s h ducks. Trans. Amer. F i s h . Soc. 66:338-342. B e l l , G.P. 1980. Habitat use and response to patches of prey by desert insectivorous bats. Can. J. Zool. 58:1876-1883. Be l l r o s e , F.C. 1978. Ducks, geese and swans of North America. Second e d i t i o n . Stackpole Books, Harrisburg, Penn. B i l t o n , H.T., D.F. Alderdice and J.T. Schnute. 1982. Influence of time and size at release of juvenile coho salmon (Qncorhynchus kisutch) on returns at maturity. Can. J . F i s h . Aquat. S c i . 39:426-447. Borgeson, D.P. C o n t r o l l i n g predator-prey r e l a t i o n s h i p s i n streams. pp. 425-430 in R.H. Stroud and H. Clepper (eds.) Predator-prey systems in f i s h e r i e s management. Sport Fishing Inst i t . , Washington, D.C. Brass, W. 1958. Simplified methods of f i t t i n g the truncated negative binomial d i s t r i b u t i o n . Biometrika 45:59-68. 296 Brett, J.R. 1973. Energy expenditure of sockeye salmon, Oncorhynchus nerka, during sustained performance. J. F i s h . Res. Bd. Can. 30:1799-1809. Buxton, N.E. 1981. The importance of food in the determination of the winter flock s i t e s of the shelduck. Wildfowl. 32:79-87. Campbell, K.P. 1979. Predation p r i n c i p l e s i n large r i v e r s : A review. pp. 181-191 i n R.H. Stroud and H. Clepper (eds.) Predator-Prey Systems i n F i s h e r i e s Management. Sport Fishing I n s t i t . , Washington, D.C. Caraco, T. 1980. Stochastic dynamics of avian foraging f l o c k s . Amer. Nat 115(2):262-275. Chapman, D.W. 1962. Aggressive behaviour in juvenile coho salmon as a cause of emigration. J . F i s h . Res. Bd. Can. 19(6):1047-1080. Chapman, D.W. 1965. Net production of juvenile coho salmon in three Oregon streams. Trans. Am. F i s h . Soc. 94(l):40-52. Charnov, E.L. 1976. Optimal foraging, the marginal value theorem. Theor. Popul. B i o l . 9:129-136. Charnov, E.L., G.H. Orians and K. Hyatt. 1976. E c o l o g i c a l implications of resource depression. Amer. Nat. 110(972):247-259. Cock, M.J.W. 1978. The assessment of preference. J. Anim. Eco l . 47:805-816. Cohen, J.E. 1972. Markov population processes as models of primate s o c i a l and population dynamics. Theor. Popul. B i o l . 3:119-134. C o l l i e r , G. and C. Rovee-Collier. 1981. A comparative analysis of optimal -foraging behavior: laboratory simulations, pp 39-76 in A.C. Kamil and T.D. Sargent (eds.). Foraging Behavior: E c o l o g i c a l , E t h o l o g i c a l and Psychological Approaches. Garland STPM Press, New York. Confer, J.L. and P.I. Blades. 1975. Omnivorous zooplankton and planktivorous f i s h . Limnol. Oceanogr. 20(4):571-579. Covich, A.P. 1976. Analyzing shapes of foraging areas: some eco l o g i c a l and economic theories. Ann. Rev. E c o l . Syst. 7:235-257. 297 Cowie, R.J. 1977. Optimal foraging in great t i t s (Parus major). Nature 268:137-139. Dawkins, M. 1971. Perceptual changes in chicks: another look at the 'searching image' concept. Anim. Behav. 19:566-574. Dawkins, R. and H.J. Brockmann. 1980. Do digger wasps commit the Concorde fallacy? Anim. Behav. 28:892-896. Des Lauriers, J.R. and B.H. Brattstrom. 1965. Cooperative feeding behaviour in red-breasted mergansers. Auk 82:639. D i l l , L.M. 1974. The escape response of the zebra danio (Brachydanio rerio) II. The effect of experience. Anim. Behav. 22:723-730. Dow, H. and S. Fredga. 1983. Breeding and natal dispersal of the goldeneye Bucephala clangula. J. Anim. Ecol. 52:681-695. Elson, P.F. 1950. Increasing salmon stocks by control of mergansers and kingfishers. Fish. Res. Bd., Can. A t l . Progr. Rep. 51:12-15. Elson, P.F. 1962. Predator-prey relationships between fish-eating birds and Atlantic salmon. Fish. Res. Bd. Can. Bull. No. 133. Enoksson, B. and S.G. Nilsson. 1983. Territory size and population density in relation to food supply in the nuthatch Sitta europaea (Aves). J. Anim. Ecol. 52:927-935. Eriksson, M.O.G. 1978. Lake selection by goldeneye ducklings in relation to abundance of food. Wildfowl 29:81-85. Errington, P. 1946. Predation and vertebrate populations. Q. Rev. Biol. 21:144-177. Erskine, A.J. 1971. Growth, and annual cycles in weights, plumages and reproductive organs of goosanders in Eastern Canada. Ibis 113:42-58. Erskine, A.J. 1972. Populations, movements and seasonal distribution of mergansers. Canadian Wildlife Service Report Series No. 17:35pp. 298 Ewald, P.W. and S. Rohwer. 1982. Effects of supplemental feeding on timing of breeding, clutch size and polygyny in red-winged blackbirds. J. Anim. Ecol. 51:429-450. Foreman, L.D. 1976. Observations of common merganser broods in NW California. C a l i f . Fish and Game 62(3):207-212. Fretwell, S.D. 1972. Populations in a seasonal environment. Monographs in Population Biology 5:218 pp. Galushin, V.M. 1974. Synchronous fluctuations in populations of some raptors and their prey. Ibis 116:127-132. Gass, C.L. 1977. A di g i t a l encoder for f i e l d recording of behavioural, temporal and spatial information in directly computer-accessible form. Behav. Res. Meth. and Instrum. 9(1):5-11. Ginetz, R.M. and P.A. Larkin. 1976. Factors affecting rainbow trout (Salmo gairdneri) predation on migrant fry of sockeye salmon (Oncorhynchus nerka). J. Fish. Res. Bd. Can. 33(l):19-24. Hafner, H., V. Boy and G. Gory. 1982. Feeding methods, flock size and feeding success in the l i t t l e egret, Egretta garzetta and the squacco heron, Ardeola ralloides in Camargue, Southern France. Ardea 70:45-54. Harper, D.G.C. 1982. Competitive foraging in mallards: 'ideal free' ducks. Anim. Behav. 30:575-584. Harper, H.A. 1971. Review of Physiological Chemistry. Thirteenth edition. Lange Med. Publ., Los Altos, California. Hartman, G.F. 1965. The role of behavior in the ecology and interaction of underyearling coho salmon (Oncorhynchus kisutch) and steelhead trout (Salmo gairdneri). J. Fish. Res. Bd. Can. 22(4):1035-1081. Hassell, M.P. 1978. The dynamics of arthropod predator-prey systems. Monographs in Population Biology 13:237 pp. Hassell, M.P. 1980. Foraging strategies, population models and biological control: a case study. J. Anim. Ecol. 49(2) :603-628. 299 Hinde, R.A. 1956. The biological significance of the territories of birds. Ibis 98:340-369. Hogstedt, G. 1981. Effect of additional food on reproductive success in the magpie (Pica pica). J. Anim. Ecol. 50:219-229. Rolling, C.S. 1959a. The components of predation as revealed by a study of small mammal predation of the European pine sawfly. Can. Entomol. 91:293-320. Holling, C.S. 1959b. Some characteristics of simple types of predation and parasitism. Can. Entomol. 91:385-398. Holling, C.S. 1961. Principles of insect predation. Ann. Rev. Entomol. 6:163-182. Holling, C.S. 1966. The functional response of invertebrate predators to prey density. Mem. Entomol. Soc. Can. 48:1-86. Holtby, L.B. and G.F. Hartman. 1982. The population dynamics of coho salmon (Qncorhynchus kisutch) in a west coast rain forest subjected to logging, pp. 308-347 in G.F. Hartman (ed.). Proc. Carnation Creek Workship: A 10 Year Review. Feb. 24-26, 1982. Malaspina College, Nanaimo, B.C. Huntingdon, E.H. and A.A. Roberts. 1959. Food habits of the merganser in New Mexico. New Mexico Dept. Game and Fish Bull. 9:36 pp. Huntsman, A.G. 1941. Cyclical abundance and birds versus salmon. J. Fish. Res. Bd. Can. 5(3):227-235. Iwasa, Y., M. Higashi and N. Yamamura. 1981. Prey distribution as a factor determining the choice of optimal foraging strategy. Am. Nat. 117(5):710-723. Jarman, P.J. 1974. The social organization of antelope in relation to their ecology. Behav. 58:215-267. Katz, P.L. 1974. A long-term approach to foraging optimization. Am. Nat. 108:758-782. 300 King, J.R. and D.S. Farner. 1961. Energy metabolism, thermoregulation and body temperature, pp. 215-288. in Marshall, A.J.. (ed.) Biology and Comparative Physiology of Birds, Vol. 2. Academic Press, New York. Krebs, J.R. 1974. Colonial nesting and social feeding as strategies for exploiting food resources in the Great Blue Heron (Ardea herodias). Behaviour 51:99-131. Krebs, J.R. and R.J. Cowie. 1976. Foraging strategies in birds. Ardea 64:98-116. Kruuk, H. 1972. The Spotted Hyaena. University of Chicago Press, Chicago. Lack, D. 1968. Ecological Adaptations for Breeding in Birds. Methuen, London. Larkin, P.A. 1971. Simulation studies of the Adams River sockeye salmon. J. Fish. Res. Bd. Can. 28:1493-1502. Lasiewski, R.C. and W.R. Dawson. 1967. A re-examination of the relation between standard metabolic rate and body weight in birds. Condor 69:13-23. Latta, W.C. and R.F. Sharkey. 1966. Feeding behaviour of the American merganser in captivity. J. Wildl. Manage. 30(1):17-23. Lefebvre, L. 1983. Equilibrium distribution of feral pigeons at multiple food sources. Behav. Ecol. Sociobiol. 12:11-17. Leonard, J.W. and D.S. Shetter. 1936. Studies on merganser depredations in Michigan trout waters. Trans. Amer. Fish. Soc. 66:335-337. Lindroth, A. 1955. Mergansers as salmon and trout predators in the River Indasalven. Fish. Bd., Sweden: Rep. Inst. Freshwat. Res., Drottningholm 36:126-132. Lister, D.B. and Assoc. Ltd. 1979. Juvenile salmon downstream migration study at L i t t l e Qualicum R., B.C. Unpubl. Rep. for Dept. Fish, and Oceans. Vancouver, B.C. 301 Mclnvaille, W.B. and L.B. Keith. 1974. Predator-prey relations and breeding biology of the great horned owl and red-tailed hawk in central Alberta. Can. Field-Nat. 88:1-20. MacLeod, J.R. 1977. Enhancement technology: a positive statement. pp. 137-147 in D.V. E l l i s (ed.). Pacific Salmon Management for People. West Geogr. Series Vol. 13, University of Victoria, B.C. McNair, J.N. 1982. Optimal giving-up times and the marginal value theorem. Am. Nat. 119(4):511-529. McNair, J.N. 1983. A class of patch-use strategies. Am. Zool. 23:303-313. Mace, P.M. 1983. Predator-prey functional responses and predation by staghorn sculpins (Leptocottus armatus) on chum salmon fry (Oncorhynchus keta). Unpub. Ph.D. thesis, University of British Columbia, Vancouver, B.C. Marshall, D.E. and E.W. Britton. 1980. Carrying capacity of coho streams. Unpub. manuscript. Fisheries and Oceans, Enhancement Service Branch, Vancouver, B.C. 33 pp. Marshall, D.E., R.F. Brown, V.D. Chahley and D.G. Demontier. 1977. Preliminary catalogue of salmon streams and spawning escapements of St a t i s t i c a l Area 14 (Comox - Parksville). Dept. Fish. Oceans,Canada Pac/d-77-1. Mason, J.C. and S. Machidori. 1976. Populations of sympatric sculpins, Cottus aleuticus and Cottus asper, in four adjacent salmon-producing coastal streams on Vancouver Island. Fish. Bull. 74:131-141. Maynard Smith, J. 1978. Optimization theory in evolution. Ann. Rev. Ecol. Syst. 9:31-56. Meacham, CP. and J.H. Clark. 1979. Management to increase anadromous salmon production, pp. 377-386 in R.H. Stroud and H. Clepper (eds.) Predator-Prey Systems in Fisheries Management. Sport Fishing Instit. Washington, D.C. Miegs, R.C. and C.A. Rieck. 1967. Mergansers and trout in Washington. Proc. 47th Ann. Conf. West Assoc. State Game Fish Comm.:306-318. 302 Mill e r , S.W. 1973. Predation in warm water reservoirs by wintering common mergansers. Proc. 27th Ann. Conf. SE Assoc. State Game Fish Comm.:243-252. Mi l l s , D.H. 1962. The goosander and red-breasted merganser as predators of salmon in Scottish waters. Scotland Dept. Agric. Fish., Freshwater and Salmon Fish. Res. 29:10 pp. Minaker, B.A., F.K. Sandercock and L.I. Balmer. 1979. Big Qualicum River report, 1974-1975. Fish. Mar. Serv. MS Rep. No. 1528, 131 pp. Morgan, B.J.T. 1976. Stochastic models of grouping changes. Adv. Appl. Prob. 8:30-57. Moynihan, M. 1962. The organization and probable evolution of some mixed species flocks of neotropical birds. Smithsonian Misc. Collections 143(7):1-140. Mueller, H. 1974. Factors influencing prey selection in the American kestrel. Auk 91:705-721. Mundie, J.H. and R.E..Traber. 1983. Carrying capacity of an enhanced side-channel for rearing salmonids. Can. J. Fish. Aquat. Sci. 40:1320-1322. Munro, J.A. and W.A. Clemens. 1936. Food of the American merganser (Mergus merganser americanus) in British Columbia. Can. Field-Nat. 50(3):34-36. Munro, J.A. and W.A. Clemens. 1937. The American merganser in British Columbia and i t s relation to the fish population. Fish. Res. Bd. Can. B u l l . No. 55:50 pp. Murdoch, W.W. 1973. The functional response of predators. J. Applied Ecol. 10:335-342. Neave, F. 1953. Principles affecting the size of pink and chum salmon populations in British Columbia. J. Fish. Res. Bd. Can. 9:450-491. Newton, I. 1976. Population limitation in diurnal raptors. Can. Field-Nat. 90(3):274-300. 303 Newton, I. 1980. The role of food in limiting bird numbers. Ardea 68(1-4):11-30. Newton, I. and M. Morquiss. 1982. Fidelity to breeding area and mate in sparrow hawks, Accipter misus. J. Anim. Ecol. 51:327-341. O'Connor, R.J. and R.A. Brown. 1977. Prey depletion and foraging strategy in the oystercatcher (Haematopus ostralegus). Oecologia 27:75-92. Ollason, J.G. 1980. Learing to forage - optimally? Theor. Pop. Bi o l . 18:44-56. Orians, G.H. and N.E. Pearson. 1979. On the theory of central place foraging, pp. 155-177 in D.J. Horn, B.R. Stairs and R.D. Mitchell. Analysis of Ecological Systems. Ohio State Univ. Press. Ostfeld, R.S. 1982. Foraging strategies and prey switching in the California sea otter. Oecologia 53:170-178. Parker, G.R. 1974. A population peak and crash of lemmings and snowy owls on Southamplon Is., N.W.T. Can. Field-Nat. 88:151-156. Patten, B.G. 1971. Increased predationby the torrent sculpin, Cottus rhotheus, on coho salmon fry, Oncorhynchus kisutch during moonlight nights. J. Fish. Res. Bd. Can. 28:1352-1354. Pehrsson, 0. 1974. Nutrition of small ducklings regulating breeding area and^ reproductive output in the long-tailed duck, Clangula hyemalis. Proc. Int. Cong. Game Biol. 11. Peterman, R.M. 1978. Testing for density-dependent marine survival in Pacific salmonids. J. Fish. Res. Bd. Can. 35(11):1434-1450. Peterman, R.M. and Gatto, M. 1978. Estimation of functional responses of predators on juvenile salmon. J. Fish. Res. Bd. Can. 35(6):797-808. Phelan, F.J.S. and R.J. Robertson. 1978. Predatory responses of a raptor guild to changes in prey density. Can. J. Zool. 56:2565-2572. 304 Pietrewicz, A.T. and A.C. Kamil. 1981. Search images and the detection of cryptic prey: an operant approach, pp. 311-331 in A.C. Kamil and T.D. Sargent (eds.). Foraging Behavior: Ecological, Ethological and Psychological Approaches. Garland STPM Press, New York. Pi e r o t t i , R. 1982. Habitat selection and i t s effect on reprodutive output in Newfoundland. Ecology 63(3):854-868. Prys-Jones, R.P., L. S c h i f f e r l i and D.W. MacDonald. 1974. The use of an emetic in obtaining food samples from passerines. Ibis 116:90-94. Pyke, G.H., H.R. Pulliam and E.L. Charnov. 1977. Optimal foraging: A selective review of theory and tests. Q. Rev. B i o l . 52(2):137-154. Rad, Olav. 1980. Breeding distribution and habitat selection of red-breasted mergansers in freshwater in western Norway. Wildfowl 31:53-56. Readshaw, J.L. 1973. The numerical response of predators to prey density. J Applied Ecol. 39:342-351. Ricker, W.E. 1954. Stock and recruitment. J. Fish. Res. Bd. Can. 11:59-623. Ricker, W.E. 1962. Regulation of the abundance of pink salmon populations, pp. 155-201. in N.J. Wilimovsky (ed.) Symposium on Pink Salmon. H.R. MacMillan Lectures in Fisheries. Instit. of Fish., University of British Columbia. Rogers D.J. and M.P. Hassell. 1974. General models for insect parasite and predator searching behaviour: interference. J. Anim. Ecol. 43:239-53. Rusch, D.H., E.C. Meslow, P.D. Doerr and L.B. Keith. 1972. Response of great-horned owl populations to changing prey densities. J. Wildl. Manage. 36:282-292. Salyer, J.C.,II and K.F. Lagler. 1940. The food and habits of the American merganser during winter in Michigan, considered in relation to fish management. J. Wildl. Manage. 4(2):186-219. 305 Schluter, D. 1982. Seed and patch selection by Galapagos ground finches: relation to foraging efficiency and food supply. Ecology 63(4):1106-1120. Scrivener, J.C. and B.C. Andersen. 1982. Logging impacts and some mechanisms which determine the size of spring and summer populations of coho salmon fry in Carnation Creek, pp. 257-272 in G.F. Hartman (ed.). Proc. Carnation Creek Workship: A 10 Year Review. Feb. 24-26, 1982. Malaspina College, Nanaimo, B.C. Scott, W.B. and E.J. Crossman. 1973. Freshwater Fishes of Canada. Fish. Res. Bd. Can. Bull. 184:966 pp. Shetter, D.S. and G.R. Alexander..1970. Results of predator reduction on brook trout and brown trout in 4.2 miles of the North Branch of the Au Sable River. Trans. Am. Fish. Soc. 99(2):312-319. Smith, J.N.M. 1974. The food searching behaviour of two European thrushes. I. Description and analysis of search paths. Behav. 48(3-4):276-302. Smith, J.N.M. and H.P.A. Sweatman. 1974. Food searching behaviour of titmice in patchy environments. Ecology 55:1216-1232. Snedecor, G.W. and W.G. Cochran. 1980. Statistical Methods. Seventh Edition. Iowa State Univ. Press, Ames, Iowa. i Southwood, T.R.E. 1978. Ecological methods. Second edition. Chapman and Hall, London. 524 pp. Sutherland, W.J. and P. Koene. 1982. Field estimates of the strength of interference between oystercatchers, Haematopus ostralegus. Oecologia 55:108-109. Thorpe, W.H. 1963. Learning and Instinct in Animals. Second edition. Methuen, London. Timken, R.L. and B.W. Anderson. 1969. Food habits of common mergansers in the north central United States. J. Wildl. Manage. 33:87-91. Tinbergen, N., M. Impekoven and D. Franck. 1967. An experiment on spacing-out as a defence against predation. Behav. 28:307-321. 306 Tschaplinski, P.J. and G.F. Hartman. 1982. Winter distribution of juvenile coho salmon (Qncorhynchus kisutch) in Carnation Creek and some implications to overwinter survival, pp. 273-288 in G.F. Hartman (ed.). Proc. Carnation Creek Workship: A 10 Year Review. Feb. 24-26, 1982. Malaspina College, Nanaimo, B.C. Visser, M. 1982. Prey selection by the three-spined stickleback. (Gasterosteus aculeatus L.) Oecologia 55:395-402. Ward, F.J. and P.A. Larkin. 1964. Cyclic dominance in Adams River sockeye salmon. Prog. Report II, Internat. Pacific Salmon Fish. Comm., New Westminister, B.C. Ware, D.M. 1971. Predation by rainbow trout: the effect of experience. J. Fish. Res. Bd. Can. 28:1847-1852. Ware, D.M. 1972. Predation by rainbow trout (Salmo gairdneri): the influence of hunger, prey density and prey size. J. Fish. Res. Bd. Can. 29:1193-1201. Werner, E.E. and D.J. Hall. 1974. Optimal foraging and the size selection of prey by the bluegill sunfish (Lepomis macrochirus). Ecology .55:1042-52. White, H.C. 1936. The food of kingfishers and mergansers on the Margaree R., Nova Scotia. J. Bi o l . Bd. Can. 2(3):299-309. White, H.C. 1937. Local feeding of kingfishers and mergansers. J. Biol. Bd. Can. 3(4):323-338. White, H.C. 1939. The food of Mergus serrator on the Margaree R., N.S. J. Fish. Res. Bd. Can. 4(5):309-311. White, H.C. 1957. Food and natural history of mergansers on salmon waters in the maritime provinces of Canada. Fish. Res. Bd. Can. Bull. No. 116:63 pp. Woodsworth, E.J. 1982. The predatory functional response of the prickly sculpin (Cottus asper) to density of sockeye salmon (Qncorhynchus nerka) fry. Unpub. M.Sc. thesis, University of British Columbia, Vancouver, B.C. 307 Yom-Tov, Y. 1974. The effect of food and predation on breeding density and success, clutch size and laying date of the crow (Corvus cprone). J. Anim. Ecol. 43:479-498. Zach, R. and J.N.M. Smith. 1981. Optimal foraging in wild birds? pp. 95-109 i n A.C. Kamil and T.D. Sargent (eds.). Foraging Behavior: Ecological, Ethological and Psychological Approaches. Garland STPM Press, New York.