UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A genetic analysis of the mating system of sockeye salmon Mahranvar, Ladan 2002

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

Item Metadata

Download

Media
831-ubc_2003-0017.pdf [ 5.45MB ]
Metadata
JSON: 831-1.0090762.json
JSON-LD: 831-1.0090762-ld.json
RDF/XML (Pretty): 831-1.0090762-rdf.xml
RDF/JSON: 831-1.0090762-rdf.json
Turtle: 831-1.0090762-turtle.txt
N-Triples: 831-1.0090762-rdf-ntriples.txt
Original Record: 831-1.0090762-source.json
Full Text
831-1.0090762-fulltext.txt
Citation
831-1.0090762.ris

Full Text

A G E N E T I C A N A L Y S I S O F T H E M A T I N G S Y S T E M O F S O C K E Y E S A L M O N by L A D A N M E H R A N V A R B.Sc, The University of Toronto, 1999 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF ZOOLOGY accept this thesis as conforming to the required standard. / D r . Mic^aeiC. Healey^Supervi^or Dr. Dolph Schluter, Committee Member Dr. Martin L. Adamson, Random Examiner Dr. Michael Doebeli, Chair T H E UNIVERSITY OF BRITISH COLUMBIA NOVEMBER 2002 © Ladan Mehranvar, 2002 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Z^oc-tog^  The University of British Columbia Vancouver, Canada Date Nwavtber ^ i 2^>&2-DE-6 (2/88) ABSTRACT It has now been widely shown that the genetic mating system of an organism determined using genetic markers can be very different to the social mating system based on behavioural observations. However, it is less well known what particular characteristics make an individual more or less successful than expected from the observed system. I present the first attempt to measure individual genetic reproductive success, as evidenced by microsatellite parentage assignment, and to compare this measure to the behavioural mating success in sockeye salmon, Oncorhynchus nerka. Approximately 40% of the variation in the genetic reproductive success of males was accounted for by the behavioural estimates of dominance and consort behaviour, whereas no reasonable behavioural estimate of female genetic success, measured as her genetic mate quality, was found. There was a high variance in individual reproductive success in males, and a high number of mates acquired by both males and females, suggesting that multiple mating partners is a rule rather than an exception. There was also a significant correlation between the number of genetic mates acquired and the number of offspring sired in males. This study also investigated whether any traits (size, shape, energy content, longevity) predicted the social mating success of males and females. I found little evidence supporting the prediction from previous studies of a significant correlation between individual size and social mating success for either males or females. Instead, body shape (exaggeration of secondary sexual characteristics) and longevity predicted dominance and consort behaviour in male sockeye salmon, and pre-spawning energy content predicted female longevity and social mating success. Finally, I present path models for both males and females describing the causal relationships among life history traits, social mating success and genetic reproductive success (or genetic mate quality in females) based on my previous results. Overall, these results indicate that this species' mating system is more flexible than previously thought and suggest that the behavioural estimates of mating success are not always successful predictors of the genetic reproductive success. Whether this low congruence is due to a high sneaking rate or to poor estimates of behavioural or genetic spawning success is unknown. i i T A B L E O F C O N T E N T S A B S T R A C T i i T A B L E OF CONTENTS i i i LIST OF T A B L E S v LIST OF FIGURES vi A C K N O W L E D G E M E N T S vii CHAPTER 1: G E N E R A L INTRODUCTION 1 Literature Cited 5 CHAPTER 2: B E H A V I O U R VERSUS GENETIC M E A S U R E S OF M A T I N G SUCCESS 7 Introduction 7 Methods 10 Study site 10 Morphology and energy measurements 10 Behavioural data collection 11 Embryo collection 12 Genetic analyses 12 D N A extractions 12 Microsatellite screening and PCR conditions 13 Parentage assignment 13 Social mating success indices 16 Genetic reproductive success indices 21 Statistical analyses 22 Results 24 Sample collection 24 Microsatellite markers 24 Paternity analysis 27 Parental combinations 27 Behavioural and genetic measures 32 Behavioural versus genetic mating success 36 Discussion 36 Literature Cited 49 i i i CHAPTER 3: FACTORS AFFECTING SPAWNING B E H A V I O U R 53 Introduction 53 Methods 57 Statistical analyses 58 Results 59 Characteristics of adult sockeye salmon 59 Male allocation trade-offs and social mating success 61 Female allocation trade-offs and social mating success 64 Discussion 67 Male allocation trade-offs and social mating success 69 Female allocation trade-offs and social mating success 72 Literature Cited 75 CHAPTER 4: THE IMPORTANCE OF DIRECT A N D INDIRECT COMPONENTS OF S E X U A L SELECTION 79 Introduction 79 Methods 81 Path analytic models for male fitness 84 Path analytic models for female fitness 88 Results 90 Path analytic models for male fitness 90 Path analytic models for female fitness 93 Discussion 97 Path analytic model for male fitness 101 Path analytic model for female fitness 104 Literature Cited 106 CHAPTER 5: G E N E R A L DISCUSSION I l l Literature Cited 114 iv LIST OF TABLES Table 2.1 Microsatellite loci descriptions and PCR amplification conditions 14 Table 2.2 Criterion used to compute the consort index for males 18 Table 2.3 Criterion used to compute the consort score for females 20 Table 2.4 Number of alleles, allele size range, observed and expected heterozygosities, and the probability of exclusion for each locus 26 Table 2.5 The genetic mating patterns in each study enclosure 28 Table 2.6 Analysis of Variance for testing assortative mating 31 Table 2.7 The social mating patterns of males 33 Table 2.8 Correlation measures among behavioural and among genetic mating success indices in males 34 Table 2.9 Correlation measures among behavioural and among genetic mating success indices in females 35 Table 2.10 Regression models for genetic reproductive success measures in males 40 Table 3.1 Mean values, range and coefficients of variation of phenotypic traits and energetic content of adult fish 60 Table 3.2 Correlation matrix for all measured variables in all fish 63 Table 3.3 Summary of the multiple regression analyses on male social mating success 65 Table 3.4 Summary of the multiple regression analyses on female social mating success 68 Table 4.1 Description of the models for the male breeding system 87 Table 4.2 Description of the models for the female breeding system 91 Table 4.3 Summary of the goodness of fit tests for male breeding system models 92 Table 4.4 Direct and total effects for the best male breeding system model 94 Table 4.5 Summary of the goodness of fit tests for female breeding system models 96 Table 4.6 Direct and total effects for the best female breeding system model 98 LIST OF FIGURES Figure 2.1 Allele frequencies at each locus for all individuals sampled 25 Figure 2.2 Measures of the genetic reproductive success of males 29 Figure 2.3 Measure of the frequency of the proportion of mates acquired by females 30 Figure 2.4 Scatter plots of genetic reproductive success versus dominance in males 37 Figure 2.5 Scatter plots of genetic reproductive success versus consort score in males 38 Figure 2.6 Scatter plots of genetic reproductive success versus # of social mates in males 39 Figure 3.1 Relationship between morphological traits and energy content and between energy content and longevity in males 62 Figure 3.2 Relationship between caudal peduncle depth and between energy content and longevity in females 66 Figure 4.1 Hypothesized causal scheme for the relationships between phenotypic traits, longevity, social mating success, and genetic reproductive success for males 85 Figure 4.2 Hypothesized causal scheme for the relationships between phenotypic traits, social mating success, and mate choice for females 89 Figure 4.3 Representation of the path diagram corresponding to the best model for the mating system of male sockeye salmon 95 Figure 4.4 Representation of the path diagram corresponding to the best alternative model for the mating system of female sockeye salmon 99 VI ACKNOWLEDGEMENTS There are so many people who have taught me something, influenced me in some way, touched me some how, or encouraged me along the way. So many people. I cannot begin to justly thank them all. I began my Master's project knowing very little about the Pacific Salmon mating system, and I am leaving here with a tiny bit more knowledge about the fish... a little more insight to add to the observations of what we perceive to be "the truth". I came here knowing much less about the professor with whom I was going to work for the next few years, and I am leaving here satiated and a little wiser. M y supervisor, Michael Healey, has become my friend, colleague, teacher, confidante, and my mentor. During the past three years I have learned so much from Mike, from his lectures in fisheries science, life history evolution and statistics, to photography, literature and culture... a kindred spirit whom I had the opportunity to get to know throughout my short visit on this side of the world. People come into our lives and paint our otherwise black and white world with some colour. Mike was that person. He is full of hope, full of encouragement, full of comfort and full of love. I survived many thinking attacks with Mike's unconditional support. I survived fieldwork with comforting words from Mike. More than the science he shared with me, however, I was reaffirmed that by being a scientist you can also be a person with ideas about politics, a person with a passion for photography, a person who thinks beyond what he studies. Mike brought me up when I was down, encouraged me to go forward with my heart, and helped me untangle uncertainties. It was the little things that Mike revealed that made him such an incredible supervisor, and more importantly, a complete human being. I have learned to think. I have learned to listen. I have learned to concentrate. And I have learned to love. I thank my lucky stars for having coincidently introduced me to Mike on that very first visit to U B C almost four years ago. Mike, thank you for being you. It has been a most interesting, rewarding and enriching experience. I greatly benefited from the helpful comments and thoughts and suggestions provided by my committee members: Rick Taylor, Scott Hinch, Dolph Schluter, and especially Sally Otto. Allison Barnes, the Zoology graduate secretary also deserves a big thank you for all her hard work, her charisma and her contagious laughter. A special thank you to Kyle Young, who went beyond all sense of duty, and to Julian Olden who helped me with my thesis from thousands of kilometres away. Thank you both for your help and especially for your friendship. vii M y fieldwork would not have been possible without the assistance of Andrew Lotto and Yuho Okada. Andrew and Yuho did as much work, i f not more than, myself. Sitting in the cold, wet weather, Yuho observed and recorded, while I sat inside shivering, gazing out, marvelling at the willpower possessed by this man. Many others helped me get through the field season- Jody Frolek, Dave Patterson, Eric Lotto. Andrew Hendry prepared me, both with words of encouragement and with facts and figures, before plunging out into the salmon world. Glenn Crossin also provided helpful ideas and advice throughout my thesis. Working within the spawning channels was made possible with the help of the local N'quatqua Band, in particular Harry O'Donaghey, the manager of Gates Creek Spawning Channel, Willard Abraham, and Harry Leo Junior. I am grateful to Victor Ewert, the manager of Weaver Creek Spawning Channel and his sidekick Wayne Charlie, who helped me with everything I ever needed in the field. And of course, the little fish- all of those who cooperated, who didn't contract the parvicapsula parasite, those that survived and spawned, and fought and did all the "natural" fishy stuff I was hoping they would do. Thank you so much. M y lab work was a long and enduring stretch. From the Taylor lab to the Ritland lab, there are many souls to whom I am indebted. Rick Taylor generously offered me space and good karma in his lab. He became a good friend and a good teacher. And in there, the few spirits I met will hopefully travel alongside me throughout my life- Patrick Tamkee, Emily Rubidge, Anna Elz, Allan Qostello- thanks for your friendship, your good loving, and all our academic and non-academic talks. M y lab assistants- Melanie Yee and Mike McDermid- thank you for all your help and patience. Thank you to Cole Burton, Steve Latham, Mike Stamford, Jason Ladell, Don McPhail for your presence, your words, your thoughts. In my journey from the Taylor lab to the Ritland lab, I stopped over at a lab to complete my bomb calorimetry. Gilles Galzi, and Siva: without you the bombing of my fish samples would have been impossible. Merci beaucoup for your teachings. Finally, with Carol Ritland leading the way, I took up a corner at the Ritland lab. Working in a lab became a completely new experience, and my time in the Ritland lab was both efficient and fun. Carol, I am thankful and greatly appreciative for all the help that you offered, all your encouraging words, your enthusiasm for my project, and your general positive attitude. I would also like to thank Allyson Miscampbell for all her help in and around the lab. Cherdsak Liewlaksaneeyanawin and Washington Gapare- coming in on the weekends was tough, but together, we made it through. vii i M y research was funded through a Natural Sciences and Engineering Research Council strategic grant to M . C. Healey, Tony Farrell and Scott Hinch. I was financially supported by an NSERC PGS A award and a teaching assistantship from the Department of Zoology. This may seem to be the end- the end because I have thanked my supervisor, my field crew, and my lab crew, but that was just the preface. M y friends- from Tarun Nayar, who taught me to live life through music, to Thomas Bell, whose friendship will always remain a special part of my life. Jordana Tzenova- we learned so much about Vancouver-living together- from our very first days in the sushi bars, to our last days, drinking coffee downtown. Emily Rubidge, you are my best friend, my roommate, my lab mate, meeting you has made life so much more beautiful. Anna Elz, together we struggled with life, struggled with academics, struggled with people in general, and in the end I learned so much with you. Thank you. Arturo Orellano- a bosom friend, a friend who understands, who listens, and who loves. I'm glad we shared so much with each other, including all those coffee breaks, the films, and the Cuban music. Olivier Cheneval (the male version of me) is there a recipe for happiness? And of course Hilary Lindh, my office mate who heard just about every word of my thesis and my presentations twenty times while I read out loud before the final days, and still remains my friend. May we all re-unite in the mountains somewhere- somewhere warmer, somewhere with less rain, and with louder music. Of course, my list of friends would not be nearly complete without Nathan Gabriel (Aureliano Segundo) Taylor. You have so much love, so much substance, and so much passion for both the known and the unknown. M y life in Vancouver would not be complete without having met you. Your heart is full- and I thank you for having squeezed me in there. You mean so much to me. Thank you for supporting me through the toughest times in graduate school, for helping me to relax, and for working through my thesis with me. It would have been nearly impossible without you by my side. And then I have friends and family on the opposite side of the world, whom I often visit in my dreams. M y madar, ameh's, khaleh's, dayee's, Nassreen, and all of my cousins. Merci for everything, especially to my Madar, for her unconditional love and support for the way we have chosen to lead our lives. Hooman Vatankhah- my jigar- your spirit lives within me and travels with me always. And finally my best friend, Jannice Friedman, with whom I have grown for the ix past decade. We have helped colour each other's lives with different shades and I hope it will remain this way forever. It is near the end, yet the most important few people in my life have not been mentioned. M y family. It continually amazes me that I survived living on this side of the country without them next to me. Yasi- thank you for all your encouraging words, for all your love and support, for coming out here at the worst of times and reassuring me that a mother is all one needs. Sanaz-you are life. M y life. Thank you for all your thoughts. Baba- your support allowed me to pursue my dreams- thank you. And of course, Hooman-1 cannot wait to see how you will carve your way through life. Thank you. So much. General Introduction Adaptation increases the frequency of certain traits within a population that enhance the survival or reproductive success of individuals expressing those traits (Darwin 1859). It is a central theme in evolutionary biology as it involves the roles of natural selection and sexual selection in promoting evolutionary change (Mayr 1963). Explaining why such traits have evolved the way they have, how these traits have come to coadapt, and whether selection has pushed the population towards some optimal tactic for a particular condition remains the basis of the general phenomena to be investigated. Each year, millions of juvenile Pacific salmon (genus Oncorhynchus) leave their natal streams and rivers of northwestern North America in their outward migration to feeding grounds in the ocean (Clarke 1995). After periods ranging from a few months to several years mature adult salmon leave the ocean, and enter rivers along the coast of Asia and North America, migrating upstream to their natal spawning locations to complete their life cycle. Adult salmon undergo severe energy constraints from the point they enter freshwater systems, when they stop feeding, to the end of reproduction, when they die (Groot and Margolis 1991). Following entry into freshwater, adults undertake a demanding upstream migration to their natal streams or lakes, utilizing up to 85% of their somatic energy (Brett 1995; Hendry and Berg 1999; Crossin 2002) depending to a large degree on their migration distance. During this time, a portion of their energy is used for the development of secondary sexual traits, despite the demand for the energy required for the arduous effort to reach their spawning sites (Brett 1995; Hendry 1998). The most exaggerated features are the elongated snout and hump in males, the former functioning as a weapon during fighting (Gross 1984), and the latter shielding the bearer against sneaking males while spawning, and protecting it from attacks while fighting. Females tend to develop less exaggerated features, although some development of their snout, hump and tail does occur 1 (Quinn and Foote 1994). Because of the one-time opportunity to reproduce (semelparity), Pacific salmon do not need to invest any energy into future reproduction. The limited time period and the availability of breeding grounds in which salmonids can spawn successfully create a highly competitive arena for the acquisition of limited resources (e.g. Foote 1990; Fleming and Gross 1994; Quinn et al. 1996). Females compete for access to optimal spawning and nesting sites (Hanson and Smith 1967; Schroder 1982; Gross 1985; Foote 1990; Fleming and Gross 1994; Quinn and Foote 1994), where they purge the substrate to construct a series of nests, collectively called a redd. Immediately after each spawning event, females bury their eggs and start excavating another depression for another pocket of eggs, which may be fathered by the same or some different male. Thus, each nest contains only a portion of the female's eggs. Larger females have increase egg production (Fleming and Gross 1994; Jonsson et al. 1996; Hendry et al. 2001; Kinnison et al. 2001; Crossin 2002) and tend to be the winners of intrasexual competition for preferred nesting sites (Foote 1990; Fleming and Gross 1994), which may be indicative of reproductive status or overall fitness. They remain in their territory and guard their redd until their death from destruction by later-arriving females (Van den Berghe and Gross 1989; Foote 1990). Males do not partake in nest construction, nor the guarding of the redds, but are constantly competing with other males for access to nesting females (Schroder 1981; Foote 1990; Quinn and Foote 1994; Quinn et al. 1996). The outcome of most male-male competitions is a dominance hierarchy, with the largest male often occupying the most dominant position (closest to the female) (Quinn and Foote 1994). The other subdominant males hold satellite positions, downstream or lateral in relation to the mating pair, sneaking in to fertilize eggs at the moment of spawning (Chebanov 1980; Schroder 1981; Gross 1985; Foote 1990; Foote et al. 1997). 2 Pacific salmon are thought to mate with more than one partner (Hanson and Smith 1967; Gross 1985), but the difficulty of observing extra-pair fertilization and nocturnal behaviour in the wild has left the mating system incompletely described. The link between the social system of Pacific salmon and their fitness is similarly unclear. Pacific salmon form dominance hierarchies, and the importance of various phenotypic traits in the establishment of these hierarchies have been studied (Foote 1990; Fleming and Gross 1994; Quinn and Foote 1994; Healey and Prince 1998), yet the role of these dominance hierarchies in fitness remains ambiguous. Many studies have focused on the social mating system of this species, but none have expanded this to include the genetic mating system of the fish. Genetic markers, such as microsatellite D N A , provide one way of exploring mating structures (Queller et al. 1993; Avise 1994; Hughes 1998). These markers can be used to identify individuals, parentage and kinship, and mating systems at large. Local gene flow, such as the genetic distance among animals or plants in different regions, or the history of large-scale population migrations can be studied using microsatellites. Temporal patterns of historical population processes and the evolutionary mechanisms underlying genetic structure can also be studied with the use of these markers (Queller et al. 1993). Microsatellites, consisting of tandem repeats of very short nucleotide sequences, are an important tool for Mendelian problems due to their highly polymorphic and codominant nature, their selective neutrality, and their relative ease of amplification and scoring (Queller et al. 1993; Jarne and Lagoda 1996). The objective of my thesis was to investigate the mating structure of sockeye salmon (Oncorhynchus nerka) with the use of microsatellite markers. I studied sockeye salmon in a small stream tributary to the lower Fraser River. In my second chapter, I examine the breeding 3 system of sockeye salmon by estimating the variance in male genetic reproductive success and female mate choice from the paternity of fertilized eggs using microsatellite D N A . From my paternity results, I tested whether assortative mating exists and how prevalent multiple mating is in the population. I present a comparison of field estimates of mating success with individual success as measured by D N A fingerprinting, in an attempt to evaluate the social versus the genetic partner of these fish. The variance in male reproductive success has implications for female mate choice and male-male competition (Reynolds 1996), which takes me to my third chapter. The focus of my third chapter is largely based on the factors that influence intra- and intersexual competition among males and females. M y goal was to first test for reproductive trade-offs, and then to evaluate the importance of specific life history traits to the social mating success of the fish. Here, only behavioural observations were used, in an attempt to compare previous observations on this species to my results, when genetic data is unavailable. Finally, to link all the pieces of the mating system together, I employed path analysis, which tests whether a specific model is an acceptable description of the hypothesized causal relationships among a set of variables. M y fourth chapter is devoted to various models I developed describing the mating system of sockeye salmon. I have used all the information from previous chapters to connect life history characteristics, behavioural mating success and genetic reproductive success in an attempt to demonstrate the components that influence fitness in these fish. By using microsatellite markers to investigate these components of mating structure, I have attempted to further our understanding of the mating system and life history traits influencing sockeye salmon fitness. I have also made an assessment of the genetic variance in reproductive 4 success in sockeye salmon and modeled the mating system to describe it, providing a starting point for further research into the factors influencing the fitness of Pacific salmon. Literature Cited: Avise, J.C. 1994. Molecular markers, natural history and evolution. Chapman and Hall, New York. Brett, J. R. 1995. Energetics. In Physiological ecology of Pacific salmon. (Groot, C , Margolis, L . and Clarke, W. C , eds), pp. 1-68. Vancouver: U B C Press. Clarke, W.C. 1995. Preface. In: Physiological ecology of Pacific Salmon (Groot C, Margolis L , Clarke WC, eds). Vancouver, BC: University of British Columbia Press. Crossin, G. 2002. Selection for a bioenergetic-morphological feedback mechanism in migrating adult salmon is related to migratory distance and elevation. MSc thesis, University of British Columbia, Vancouver. Darwin, C. 1859. The origin of species by means of natural selection, or The Preservation of favored races in the struggle for life. London J. Murray, London. Fleming, L A . , Gross, M.R. 1994. Breeding competition in a Pacific salmon (coho: Oncorhynchus kisutch): measures of natural and sexual selection. Evolution 48:637-657. Foote, C.J. 1990. A n experimental comparison of male and female spawning territoriality in a Pacific salmon. Behaviour 115:283-314. Foote, C. J., G. S. Brown and C. C. Wood. 1997. Spawning success of males using alternative mating tactics in sockeye salmon, Oncorhynchus nerka. Canadian Journal of Fisheries and Aquatic Sciences 54:1785-1795. Groot, C , and L. Margolis. 1991. Pacific salmon life histories. U B C Press, Vancouver. Gross, M . R. 1984. Sunfish, salmon, and the evolution of alternative reproductive strategies and tactics in fishes. In Fish reproductions: strategies and tactics. (Wootton, R. and Potts, G., eds), pp.55-75. London: Academic Press. Gross, M.R. 1985. Disruptive selection for alternative life history strategies in salmon. Nature 313:47-48. Hanson, A.J . , Smith, H.D. 1967. Mate selection in a population of sockeye salmon {Oncorhynchus nerka) of mixed age groups. Journal of Fisheries Research Board of Canada 24:1955-1977. Healey, M . C . , Prince, A . 1998. Alternative tactics in the breeding behaviour of male coho salmon. Behaviour 135:1099-1124. 5 Hendry, A . P. 1998. Reproductive energetics of Pacific salmon: strategies, tactics, and trade-offs. PhD thesis, University of Washington, Seattle. Hendry, A . P. and O. K. Berg. 1999. Secondary sexual characters, energy use, senescence, and the cost of reproduction in sockeye salmon. Canadian Journal of Zoology 77:1663-1675. Hendry, A . P., T. Day and A. B. Cooper. 2001. Optimal size and number of propagules: Allowance for discrete stages and effects of maternal size on reproductive output and offspring fitness. American Naturalist 157:387-407. Hughes, C. 1998. Integrating molecular techniques with field methods in studies of social behaviour: a revolution results. Ecology 19:383-399. Jarne, P., Lagoda, P J .L . 1996. Microsatellites, from molecules to populations and back. Trends in Ecology and Evolution 11:424-429. Jonsson, N . , B. Jonsson and I. A . Fleming. 1996. Does early growth cause a phenotypically plastic response in egg production of Atlantic salmon? Functional Ecology 10:89-96. Kinnison, M . T., M . J. Unwin, A . P. Hendry and T. P. Quinn. 2001. Migratory costs and the evolution of egg size and number in introduced and indigenous salmon populations. Evolution 55:1656-1667. Mayr, E. 1963. Animal Species and Evolution. Harvard University Press, Cambridge, M A . Queller, D.C., Strassmann, J.E., Hughes, C.R. 1993. Microsatellites and kinship. Trends in Ecology and Evolution 8:285-288. Quinn, T.P., Adkison, M.D. , Ward, M . B . 1996. Behavioural tactics of male sockeye salmon (Oncorhynchus nerka) under varying operational sex ratios. Ethology 102:304-322. Quinn, T.P., Foote, C.J. 1994. The effects of body size and sexual dimorphism on the reproductive behaviour of sockeye salmon, Oncorhynchus nerka. Animal Behaviour 48:751-761. Reynolds, J.D. 1996. Animal breeding systems. Trends in Ecology and Evolution 11:68-72. Schroder, S. 1981. The role of sexual selection in determining the overall mating patterns and mate choice in chum salmon. PhD thesis, University of Washington, Seattle. Schroder, S. 1982. The influence of intrasexual competition on the distribution of chum salmon in an experimental stream. In: Proceedings of the salmon and trout migratory symposium (E.L. Brannon & E.O. Salo, eds), pp. 275-285. University of Washington Press, Seattle. Van den Berghe, E. P. and M . R. Gross. 1989. Natural selection resulting from female breeding competition in a Pacific salmon (coho: Oncorhynchus kisutch). Evolution 43:125-140. 6 Chapter 2- Behavioural versus genetic measures of mating success Introduction Differential success in survival and in reproduction exists among individuals of a species or a population (Darwin 1859). In sexual organisms, the ability to find a member of the opposite sex and to persuade him or her to copulate is of utmost importance to the genetic fitness of the individual. When Darwin introduced the concept of sexual selection- a measure of the differential mating success that exists among members of the same sex in some trait that enhances their ability to acquire a mate- he was well aware of the struggle that must exist among individuals of one sex for access to the opposite sex (Darwin 1859). Intrasexual interactions among males for access to limited, ovipositing females and among females for access to resources necessary for the care of their offspring are expected to have a strong competitive component (Darwin 1859; Emlen and Oring 1977). Intersexual interactions exist in mate choice, where one sex struggles to gain access to the opposite sex by displaying his/her dominance, large body size, or colourful plumage. Because in many animal systems males maximize their fitness by siring as many offspring as possible and females are limited by the number of eggs they can make, there is a struggle among the males for access to a limiting resource. Throughout this reproductive struggle among members of the same sex, intense fighting and displaying may take place (Darwin 1871). This struggle increases when the availability of the limiting sex is low, as in polygamous systems, or when the animals are constrained to a single or short breeding season (Emlen and Oring 1977). Because fighting for access to females is costly, the victor of male-male competition is often superior in quality and the one which females may prefer (Cox 1977; Berglund et al. 1996). This variance in the reproductive success of both sexes is critical to understanding the evolution of reproductive traits and the processes of selection in natural populations. 7 Pacific salmon (genus Oncorhynchus) are an ideal species for the study of breeding competition both logistically and biologically. Their ability to home to precise and predictable spawning locations at specific times (Quinn et al. 1987), the exaggerated sexual dimorphism that exists in most species, developed with limited energy acquired at sea, and the constraint of being semelparous and, thus, unable to rely on future years for reproduction, make salmon a good model species for life history studies. The limited carrying capacity of the spawning grounds in some populations (Groot and Margolis 1991), the fact that males contribute nothing but genes to their mates and their offspring, and the cessation of food intake upon arrival into freshwater, provide a good opportunity to study a system with a number of controls that are inherent in their natural life history. Many studies of mating systems infer the reproductive success of species from behavioural observations made in the natural environment of the fish (Hughes 1998; and references therein). The dominance of a salmon is often assessed by his aggressive acts toward other males or his consort position (proximity and interaction to ovipositing female). It has been suggested that aggressive males and consorts fertilize the great majority of eggs in salmonids (Schroder 1982; Chebanov et al. 1983; Maekawa and Onozato 1986; Quinn and Foote 1994). Recently, there has been considerable progress in the study of mating systems with the use of molecular markers at the individual level, and it is now possible to distinguish between the social and the genetic mating partner on individuals (Hughes 1998). Studies employing molecular markers to infer reproductive success have shown that behavioural observations alone are not always an accurate measure of true parentage (Hughes 1998; Coltman et al. 1998; Coltman et al. 1999; Lebas 2001). Behavioural observations have many limitations, for instance, not witnessing extra-pair parentage or nocturnal behaviour. Genetic data on birds have shown that many species that were 8 considered to be monogamous engage in extra-pair copulations (Moller and Birkhead 1993). Likewise in mammals, a high proportion of offspring were often not sired by the observed social partner (Amos et al. 1993; Coltman et al. 1999). The use of genetic parentage assignments in polygynous species has found both high (Pemberton et al. 1992; Altmann et al. 1996; Abell 1997; Gullberg et al. 1997) and low (Amos et al. 1993; Coltman et al. 1999; Lebas 2001) associations between observed behavioural mating success and the more reliable genetic reproductive success. For many species, including sockeye salmon, however, we do not know the quantitative or qualitative relationship between the social mating success of individuals and their genetic reproductive success. If these estimates are strongly correlated, substantial insight into the evolutionary biology of these fish could be gained using inexpensive field observations when molecular techniques cannot be used. In the present work, I assess the breeding system of sockeye salmon (Oncorhynchus nerka) with the use of microsatellite markers and provide the first direct estimates of the variance in male reproductive success by determining paternity of fertilized eggs. Reproductive success of females is a function of their egg production (both fecundity and egg size), egg viability, guarding of nests, and offspring survival to reproductive age. The variance in female reproductive success is not directly assessed in this study, as only a portion of the eggs per female was analysed. Therefore, female mate choice was used as a female indicator of reproductive success. This assumes that female fitness is, in part, influenced by the quality of the males who fertilize her eggs. I also tested whether assortative mate preference (based on body length, secondary sexual character size or relative energy content) exists in this population. I then present a comparison of field estimates of male and female mating success with individual success as measured by D N A fingerprinting. I investigated whether mating success estimated from parameters observed on the spawning grounds (i.e. reproductive lifespan, dominance and 9 the percent of time a fish was sighted to be in a consort position) was equivalent to realized reproductive success (the number of mates and the proportion of offspring sired in males, and the number of mates and mate quality in females). To avoid confusion, I use the term 'social' mating success when referring to the mating success based on the behavioural observations and 'genetic' reproductive success when referring to the reproductive success based on D N A analyses. Methods Study site Weaver Creek Spawning Channel is located approximately 100 km upstream from the mouth of the Fraser River in southwestern British Columbia and empties into the Harrison River, a major tributary of the lower Fraser River. The channel, built in 1965, is 2930 m long and 6.1 m wide; and provides optimal conditions for the breeding of pink, chum and sockeye salmon adults and the survival of developing embryos and hatchling alevins. Sockeye salmon enter the Creek from September until late October. Spawning occurred from early October until early November. Two months prior to the start of the run, a 40m x 3m enclosure divided into four equal sections was installed near the middle reach of the channel. The enclosure was constructed of a wooden frame covered with 5 cm vexar mesh that isolated the enclosure from the rest of the channel and the four enclosure units from each other. A l l the field work in the present study was completed in one spawning season, 2000. Morphology and energy measurements I used a dip-net to capture mature sockeye salmon entering the spawning channel (24 females and 28 males in total). A l l fish selected for the study were in good physical condition (no scarring or fungal growth present). The females had not spawned but were ripe. Each fish was 10 anaesthetized using a mixture of acetic acid and baking soda (Prince et al. 1995), tagged with colour-coded disks attached by a pin through the musculature below the dorsal fin, and measured. I measured body length (tip of snout to fork of tail), snout length (tip of upper jaw to middle of eye), and body depth (anterior insertion of dorsal fin to bottom of abdomen, perpendicular to lateral line). To estimate the energy density of each fish, I removed a piece of muscle tissue weighing approximately l g from the dorsal hump. Crawford et al. (1977) and Hendry et al. (2001) have shown this to be a successful technique for non-lethal tissue sampling in large salmonids, and Hendry et al. (2001) have shown that these samples provide a reasonable measure of mass-specific energy content as estimated by bomb calorimetry. These samples were dried for 36 hours at 90°C to determine percent dry mass (dry mass:wet mass). Because a negative linear relationship exists between the moisture content and the fat content of sockeye salmon (Brett 1995; Crossin 2002), I used this relationship to estimate the relative fat content of the fish in my study. I have called this variable the pre-spawning energy content. Behavioural data collection After tagging and measuring, the fish were held in a recovery pen within the spawning channel and were released on the same day into one of the four enclosures. Each enclosure had a total of thirteen fish (seven males and six females). Five- or ten-minute tape-recorded behavioural observations were collected randomly during daylight hours, once a day for each fish in each enclosure. There were two observers, and each was restricted to observing individuals in two out of the four enclosures for the duration of the study. During observing periods, the observer stood on the stream bank adjacent to the channel and spoke into a tape recorder. The recorded observations were later transcribed. This continued until the females within the enclosure had 11 spawned. The total length of observations varied among fish from five to ten recordings, each five or ten minutes in duration (i.e. each fish had at least 25 minutes and at most 60 minutes of observing time). The behaviours recorded included chases, charges and bites for both sexes, lateral and posture displays among males, and courting displays of males toward ovipositing females (see Healey and Prince 1998 for a detailed description of these behaviours). The frequency of different behaviours was standardized to ten minutes in order to account for the difference in total observation times. Reproductive life span, or the total number of days each fish survived within the spawning channel was also noted. Embryo Collection One month following the spawning of the study fish the nests were excavated. The gravel and stones were removed manually until the egg pockets were exposed. The eggs were then collected using a turkey baster, placed in metallic cylinders, and submerged under gravel in the channel until all nests were excavated. Four to six nests were excavated per redd, where each redd represents the collection of nests from one female. Approximately 12 eggs per nest were collected randomly. When all the eggs were collected, the cylinders were packed into coolers with stream water, transported to a laboratory, placed into Heath trays and incubated at 9-11°C for one month. Dead (or cloudy) eggs were removed from each sample on a weekly basis to avoid contamination (mortality was recorded and was low, ranging from 0-10% of eggs per sample). After one month, eggs were removed from the incubators and placed in 95% ethanol. Eggs with a visible white embryo were used in further analyses. Genetic analyses DNA extractions- Tissue from all adult sockeye (fin clips) and 2-4 embryos per nest pocket (60-64 embryos per enclosure) were analysed. D N A was extracted using standard proteinase K 12 digestion/salt-based purification using the Gentra Systems (Minneapolis, MN) Puregene D N A isolation kit. The D N A was resuspended in 50ul of TE solution (lOmM Tris, I m M E D T A in H2O; pH 8.0) and stored at -20°C until used. D N A concentrations were determined using spectroscopy, and working stocks of 50ng/ul were made for each sample. Microsatellite screening and PCR conditions- Nineteen microsatellite loci were scored for all the parents. From these, six loci were chosen for their high degree of polymorphism. Initially, the tetranucleotide loci Otsl03, One 108, One 109, OnellO and RT212 and the dinucleotide locus Ssa85 were examined on half of the offspring samples. It was noted, however, that Otsl03, One 108, OnellO and RT212 were sufficiently polymorphic for reliable paternity assignment. Thus, the remaining individuals were amplified with only these microsatellites, and the data presented in this study refer to these four microsatellites. Initial PCR amplification for each primer was carried out in 10pl volumes [50ng of D N A template, 0.2 m M each dNTP, 1.0 pmol each primer, 1.5 m M MgCL;, 1.0 U Taq D N A polymerase (GibcoBRL), lOx reaction buffer (20mM Tris-HCl pH 8.4, 50 m M KCI), and 1.0 pmol of M13-29 primer] using a Robocycler Gradient 96 (Stratagene). The forward primer for each of Otsl03 and RT212 and the reverse primer for each of One 108 and OnellO were synthesized with an additional modified 19 or 20-bp (forward labelled and reverse labelled, respectively) M l 3 tail added to the 5' end of the oligonucleotide (Steffens et al. 1993). A n M l 3 primer, with an identical sequence, is directly labelled to the infrared fluorophore, IRD41, which is used as the labelled primer for detecting the microsatellites (Oetting et al. 1995). Microsatellites were amplified using the PCR conditions as outlined in Table 2.1. Parentage assignment- Products were size fractionated using gel electrophoresis and pattern visualization, performed by a LI-COR automated fluorescent D N A sequencer (Middendorf et al. 13 Table 2.1 Microsatellite loci descriptions and PCR amplification conditions. The notation 94°C (20 s) means 20 seconds at 94°C. A l l reactions were held at 4°C at the end of the amplification. Locus Reference Repetition Primer sequence (5'-3') Amplification conditions Otsl03 Nelson and Beacham, 1999 ( G A C A ) F: A G G C T C T G G G T C C G T G R: T G A T A T G G T G T G A T A G C T G G 94°C (2 min), 58°C (1 min), 72°C (20 s) 25x [94°C (20 s), 58°C (20 s), 72°C (20 s)] 72°C (5 min) Onel08 Olsen JB et al. 2000 (ATCT) 2 i F: T G C A G A G C C A T A C T A A A C C A R: A A G A A T T G A G A G A T G C A G G G 95°C (2 min) 6 cycle "touchdown" profile: 94°C (1 min), 61°C-56°C (30 s), 72°C (15 s) 23x [94°C (30 s), 56°C (30 s), 72°C (15 s)] 72°C (30 min) O n e l l O Olsen JB etal. 2000 ( T A G A ) 21 F: C C T C C A T T T C A A T C T C A T C C R: A C A G A G A A C A G T G A G G G A G C same as One 108 RT212 Bentzen P, ( A A A A ) unpublished F: A C T C A C C T A A C C C T G T C A G C A A T G 95°C (2 min) R: T G A A A G G G A T G G G T T A T T A T A C A G C C C 3x [95°C (1 min), 66°C (30 s), 70°C (1 min)] 25x [95°C (10 s), 66°C (30 s), 70°C (1 min)] 70°C (45 min) 14 1992) (LI-COR, Inc., Lincoln, NE). Raw electrophoretic data were stored in TIFF format using the image manipulation subprogram of the Base ImaglR software package (LI-COR, Inc.). Band size analysis was performed using the RFLPScan software (Scanalytics, Billerican M A ) . Genotype identification by hand was carried out after the initial RFLPScan size determination to ensure consistent and accurate band sizing. Parental assignment was determined by CERVUS version 2.0 (Marshall et al. 1998). CERVUS is a simulation program that assigns parentage to a particular male or female if the likelihood ratio is large relative to the likelihood ratios of alternative males or females given a level of statistical confidence. The program allows an error rate parameter that may result from non-paternity, erroneously typed paternal, maternal, or offspring genotypes, marker mutations or the presence of null alleles (Marshall et al. 1998). A null allele is any allele that cannot be detected by the assay used to genotype individuals at a particular locus, and most often occurs because of mutations in one or both primer binding sites. The frequency of any possible null allele is estimated for each locus in C E R V U S . The presence of null alleles is sufficient to prevent effective amplification of the microsatellite allele. Therefore, it is often worth excluding loci with high null allele frequencies from parentage analysis. The allele frequency analysis shows the expected heterozygosity for each locus. This is the unbiased expected heterozygosity (He) calculated from allele frequencies assuming Hardy-Weinberg equilibrium. Deviations from Hardy-Weinberg equilibrium are assessed using a chi-square goodness-of-fit test, which compares observed genotype frequencies with expected genotype frequencies that are calculated from allele frequencies assuming Hardy-Weinberg equilibrium (Marshall et al. 1998). The paternity tests in this study were carried out with the following parameters: 10,000 cycles, 13 candidate parents (the female parent was known for each offspring), 100% of candidate parents sampled, 96-98% of loci typed, and 1% of loci typing error. Parentage was assigned at a 15 relaxed confidence level of 90% and a strict confidence level of 95%. Parental assignment was confirmed by the process of exclusion, where the genotypes of candidate parents were compared against the offspring's genotype (taking the mother's genotype into account). Candidate parents are excluded if a mismatch occurs at one or more loci. From the computed likelihood of each offspring genotype for each pair of parental genotypes, I was able to obtain the number of offspring assigned to each parental pair and hence to each individual male. By establishing all successful pairs of spawners to which an offspring had been assigned, I was also able to obtain an estimate of the number of mates for each adult fish, and the proportion of the fish sired by each male within an enclosure. Social mating success indices I analysed four indices of male social reproductive success. The first was a dominance index for each male within an enclosure. I considered both the aggressiveness index and displaying index from taped observations. The aggressive act frequencies based on taped observations were used to calculate the aggressiveness index (difference between acts toward and away from focal male). The following behaviours were used to compute this index: chases (one fish chases another that is retreating), charges (one fish rushes toward another that is holding still), and bites (one fish bites another). Each fish was given a rank from one (least aggressive) to seven (most aggressive) for the entire spawning period based on the aggressiveness index. The displaying frequencies were used to calculate the displaying index (difference between displays toward and away from focal male). The following behaviours were used to compute this index: lateral displays (male erects fins and holds body stiffly in response to another male) and posture display (two males swim side by side with bodies tilted up). As with the aggressiveness index, each male was ranked from one (least displaying acts) to seven (most displaying acts) for the entire 16 spawning period. The average of the two indices was taken as the dominance rank for each male in each enclosure. I called this index the dominance index. The second index was a measure of consort behaviour (proximity and interaction with ovipositing female). Male consorts actively court ripe females and are either the only male with a female or are dominating access to a female, as evidenced by their proximity and interaction with the female, and their aggressive interactions with nearby males (Quinn and Foote 1994). A male courts an ovipositing female by quivering to her (vibrating while next to a female on her redd). A male seen courting a female all day was given a score of 1. A male courting a female for half of the observations in one day was given a score of 0.5. A male holding in one position, away from females, or sneaking into the nests of other pairs all day was given a score of 0. Departures from these three alternatives were scored accordingly. Table 2.2 gives narrative examples of actual males in the study and how they were scored based on their consort behaviour. Each fish was given a consort score for each day based on taped observations. The scores were averaged over the entire spawning period. I called this index the consort score. The third index was the length of time that an individual male lived on the spawning ground. Presumably, the longer a male is alive the greater his chance to fertilize eggs. I called this index longevity. The fourth index was the number of social partners each male had during the spawning period. I calculated the proportion of social mates acquired, and called this index the proportion of social mates. For females, I calculated fives indices of social reproductive success. First, I calculated a consort score for each female. Each time a female was observed she was given a score based on her partner's consort status. Thus, each observation allowed me to give a female a score based on 17 Table 2.2 The criteria used to compute the consort index for male sockeye salmon. Male consort score 0 0.25 0.50 0.75 1.0 Sneakers Holding outside of female territory range Next to female but no sign of courting Courting female for lA day Courting female for Vi day, next to another female but not courting for V2 day Courting female all day 18 the consort male she was with and these were averaged to get the final score. For instance, a female was given a score of 1 i f she was with a consort male (score of 1) for the entire day and a score of 0.5 i f she was with a consort male (score of 1) for only half of that day. Table 2.3 shows the various scenarios used in computing the consort index for females. I called this index the consort score. The second index was an aggressiveness index and was tabulated using tape-recorded observations. The aggressive act frequencies were used to calculate the aggressiveness index (difference between acts toward and away from focal female), as was done with the male aggressiveness index. This was a measure of territoriality as females were aggressive to neighbouring nesting females and occasionally to males. The aggressive interactions of females with males were also considered when calculating their territoriality index. Based on this territorial index, each female in an enclosure unit was given a rank from one (least territorial) to six (most territorial). This index was called the territoriality index. The third social reproductive success index (termed weighted consort score) was a score based on the consort score and social status of the males with which she consorted. The consort score of the female (described above) was multiplied by the social rank of the attending male to give a weighted consort score for each time the female was observed. These weighted scores were summed to obtain the overall score for the female. I called this index the weighted consort score. The length of time that the female was alive on the spawning grounds was the last social success index used. This index was called longevity. The fifth index was the proportion of social partners each female had during the spawning period. This index was called the proportion of social mates. 19 Table 2.3 The criteria used to compute the consort index for female sockeye salmon. Note that all scores are based on the consort score of the male with whom she associated. Female consort score 0 0.25 0.50 0.75 1.0 No partner 0.25 male courts her 0.5 male courts her lA time 1.0 male courts her lA time 0.5 male courts her 1.0 male courts her l/ 2 time, 0.5 male courts her 'A time 1.0 male courts her for entire day 1.0 male courts her occasionally 20 Genetic reproductive success indices I analysed two indices of male genetic reproductive success. First, I calculated the proportion of offspring sired by each male. Second, I computed the proportion of genetic mates with which each male mated. The genetic success of females was slightly more complicated as all female fish in the study spawned 100% of their eggs. As a result, I have used measures of the quantity and quality of the males with whom the female spawned. First, the proportion of genetic mates with which a female mated was calculated. This measure was called the proportion of genetic mates acquired. The remaining two indices of female success take the quality of the male partner into account, thereby, measuring the female's choice in mate quality. I have assumed that better quality males are those that fertilize relatively more eggs and those that have a higher dominance ranking. Thus, the second female success index (called Fitness 2) was based on the genetic reproductive success of her male partners. It was calculated as the sum of the genetic reproductive success of the male with whom she mated, based on the proportion of eggs sired by that male in that enclosure, times the number of eggs sired by that male with that female. That is, 7 2Z (total number of female fs eggs sired by male i x proportion of eggs sired by male i) i=\ The third female success index (called Fitness 3) was based on the social status of her male partners. It was calculated as the sum of the rank of the male multiplied by the number of eggs sired by that male. That is, 7 X (dominance rank of male i x total number of female fs eggs sired by male i) i=l Both Fitness 2 and 3 were standardized to account for the different numbers of eggs that were analysed per female. 21 Statistical Analyses One male and four females died within two days after the start of the study. These fish were not included in any analyses. In the case of two females, neither had spawned. The third female and the one male spawned once before dying on the first night of the study. Since the cause of their death was unknown, I did not include them in the study. Furthermore, because of their premature death, behavioural observations were not taken. To justify the pooling of all males and all females, an A N O V A was performed to test the differences among the four enclosures for each of the measured variables. A l l statistical tests were conducted in SAS (version 8.2). I used the results from the parentage assignment procedures to examine the extent of multiple spawning and multiple paternity. I also considered the potential for assortative mating. The parental combinations I examined statistically as indices of assortative mating were: female body length versus male body length, female snout length versus male snout length, and female pre-spawning energy content versus male pre-spawning energy content. Since each female mated with more than one male (and each male mated with more than one female), the observed data points represent multiple observations of the same female (or male), and are therefore not independent. I analysed the data consisting of multiple Y values at each X value using an analysis of variance with replication to determine, for example, whether there are differences in the body length of males among female body lengths (i.e. i f longer females preferred longer males). If the null hypothesis that the population regression is linear is not rejected, a one-way analysis of variance is performed to test: H0: (3 = 0, using F = regression MS/residual MS, with a critical value of Fa(i),i,(N-2)- If the A N O V A is not significant, this indicates that the means are not 22 significantly different from each other, and it would seem unlikely that a regression line fitted to these data would have a slope significantly different from zero (Zar 1999). I used Spearman rank correlations to relate the social mating success indices to each other and the genetic reproductive success measures to each other among the males and the females. To test the ability of the social mating success measures to predict the genetic reproductive success measures, I employed linear regression models. Each genetic measure was tested for dependency on each of the four (for males) or five (for females) behavioural indices. Within the males, I regressed longevity on the dominance index (or consort score) and then used the residuals to see if longevity, independent of dominance (or consort score), was related to the genetic reproductive success. In addition, each variable was tested for normality using the Kolmogorov-Smirnov test statistic (at the 0.05 a level). Intercorrelations of social and genetic mating success measures, and regression analyses using the social mating success measures as determinants of the genetic reproductive success measures, were adjusted for multiple comparisons using sequential Bonferroni correction tests. The sequential Bonferroni correction ranks the tests (correlation or regression) from the highest to the lowest P values obtained. It then divides the upper limit of the significance level of the individual test by the number of sequential tests being conducted in the study. For example, the P value for each of the correlation tests among the five behavioural measures for females (therefore 10 comparisons) was compared to a set at 0.05 (for the first test), 0.05/2 for the second test, and so on. The last test in the series was compared to a P value of 0.05/10, or 0.0005. A l l significant tests reported in this study are, therefore, corrected for multiple comparisons. 23 Results Sample collection Parental assignment was examined on a redd-by-redd basis, where each redd represents the collection of nests from one female. Originally, an attempt was made to keep each nest within a redd (representing one spawning event) separate from the others to determine extra-pair fertilization events, as this was observed during behavioural recordings. However, due to the dispersion of eggs during egg- burying, the potential for egg movement after egg laying, and sampling errors while collecting eggs, all offspring from one mother were combined for the analysis. Offspring from each redd was sired by one mother and one or more fathers. Microsatellite markers High levels of polymorphism were observed at the four microsatellite loci. The total number of alleles observed per locus ranged from 17 at locus Otsl03 to 33 at locus RT212 (Fig. 2.1 and Table 2.4). As shown in Table 2.4, observed per-locus heterozygosities (H 0) in the adult and offspring samples ranged from 0.966 (RT212) to 0.821 (OnellO); expected heterozygosities (He) ranged from 0.908 (RT212) to 0.775 (OnellO). There was no evidence of null alleles segregating or departures from Hardy-Weinberg equilibrium. Although the individual exclusion probabilities (the average probability of excluding a single unrelated candidate parent from parentage of a given offspring) were not high for each locus, the combined exclusion probability (total exclusionary power) across all loci was very high (0.993). Paternity was assigned for 243 of the 247 offspring with 95% confidence. Additional offspring were assigned to a male with 90% confidence. As all other potential male parents had a mismatching allele(s) with the offspring in question, the male with 90% confidence was considered to be the true father. 24 0.3 0.25 . >• O 0.2 . § 0.15. g" 0.1. ^ 0.05 . 0 ._ Ots103 - • • l l l l . - I I I . • -151 155 159 163 167 171 175 179 163 187 191 195 199 203 207 211 215 allele size (bp) 0.2 51 0.15 § 0.1 £ 0.05 0 One108 185 189 193 197 201 205 209 213 217 221 223 225 229 233 235 237 241 245 249 253 allele size (bp) 0.5 1 >» 0.4 -O § 0 3 • = 0.2-£ o.i-o One110 . • • I I I . . . . . 222 226 230 234 238 242 246 250 254 258 262 266 270 274 278 282 286 290 294 allele size ( b p ) 0.12 , >, 0 1 • 0.08 . 2> 0.06 . er o.o4. ,fc 0.02 -0 -c c RT212 I I I illlllll.l lll-l.l. J I • allele size (bp) Figure 2.1 Allele frequencies at each locus for all individuals sampled. 25 Table 2.4 Number of alleles, allele size range, observed (H 0) and expected (He) heterozygosities, and the probability of exclusion (E) for each locus in the population. Locus No. alleles Allele size range H 0 H e E (bp) Otsl03 17 151 -215 0.902 0.841 0.684 One108 20 185 -253 0.885 0.877 0.745 OnellO 19 222 -294 0.821 0.775 0.594 RT212 33 321 -641 0.966 0.908 0.804 Average 22 - 0.894 0.850 0.993* * Total exclusionary power (combined exclusion probability) across all loci 26 Paternity analysis Individual male reproductive success was measured as the total number of offspring assigned to each male adult in the analysis. Table 2.5 displays the data on mating patterns in each enclosure. The proportion of offspring sired by each male ranged from 0 to 0.52, with a mean of 0.15 (variance = 0.11) (Fig 2.2). Thus, there was a relatively high variance in this measure of reproductive success among males. I also determined the number of mates for each adult based on the genetic parental pairs. The number of mates acquired by each male ranged from 0 to 4 mates, with a mean of 1.85 and a variance of 1.06. Similarly, the proportion of mates acquired by each male ranged from 0 to 0.80, with a mean of 0.38, and a variance of 0.22 (Fig 2.2). For females, the number of mates ranged from 1 to 4, with a mean of 2.5 and a variance of 1.10. The proportion of mates acquired ranged from 0.14 to 0.67, with a mean of 0.37, and a variance of 0.17 (Fig 2.3). There was a high variance in the number of mates with whom each sex spawned based on these genetic analyses. Parental Combinations The males and females from all four enclosures were pooled (males with males, females with females), as there was no significant difference in the mean and variance of the measured variables (all P > 0.05). There was no evidence of any assortative mating based on the three morphological and physiological characters considered. Both female and male mate preference was tested, where the former is based on the mate preference of females and the latter, on the mate preference of males. The results are summarized on Table 2.6. 27 Table 2.5 The genetic mating patterns in each study enclosure. The number in each box refers to the number of eggs sired by each male for the female in question. The first column (#mates) refers to the number of social mates acquired by females. Note that females with a total of 0 eggs were eliminated from the study, as were male 1 and female 6 from enclosures 1 and 2, respectively (see text for details). Enclosure 1 # mates male 1 male 2 male 3 male 4 male 5 male 6 male 7 total female 1 2 0 0 4 4 4 0 4 16 female 2 1 0 0 16 0 0 0 0 16 female 3 2 0 0 0 6 0 10 0 16 female 4 3 1 0 13 0 2 0 0 16 female 5 n/a 0 0 0 0 0 0 0 0 female 6 n/a 0 0 0 0 0 0 0 0 total 1 0 33 10 6 10 4 64 Enclosure 2 # mates male 1 male 2 male 3 male 4 male 5 male 6 male 7 total female 1 2 0 6 0 6 0 0 0 12 female 2 2 0 6 0 2 4 0 0 12 female 3 2 0 2 • 1 4 5 0 0 12 female 4 1 0 0 2 4 3 0 3 12 female 5 1 3 0 0 0 0 0 9 12 female 6 n/a 0 0 0 0 0 3 0 3 total 3 14 3 16 12 3 12 63 Enclosure 3 # mates male 1 male 2 male 3 male 4 male 5 male 6 male 7 total female 1 1 0 0 0 0 0 10 0 10 female 2 2 0 0 0 0 0 10 0 10 female 3 2 0 3 0 0 0 0 7 10 female 4 1 0 10 0 0 0 0 0 10 female 5 2 0 0 0 8 0 0 2 10 female 6 2 0 0 6 4 0 0 0 10 total 0 13 6 12 0 20 9 60 Enclosure 4 # mates male 1 male 2 male 3 male 4 male 5 male 6 male 7 total female 1 1 4 0 4 0 4 0 0 12 female 2 2 1 0 0 3 5 0 3 12 female 3 2 0 0 0 4 0 0 8 12 female 4 1 1 6 3 2 0 0 0 12 female 5 2 0 4 0 2 0 6 0 12 female 6 n/a 0 0 0 0 0 0 0 0 total 6 10 7 11 9 6 11 60 28 Z 6 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 proportion of eggs fertilized by males Figure 2.2 Measures of the genetic reproductive success of male sockeye salmon: frequency distribution of the proportion of eggs fertilized (above) and the proportion of mates acquired (below) by males. 29 proportion of mates acquired by females Figure 2.3 The frequency distribution of the proportion of mates acquired by females. 30 Table 2.6 Summary of Analysis of Variance for testing Ho: p = 0 in mate preference. The first three tests measure female preference based on body length, snout length, and pre-spawning energy content of her genetic male partners. The last three tests measure male mate preference for the same variables in his genetic female partners. Source of SS df MS F variation Female preference Body length Total 332.795 49 Linear regression 1.1426 1 1.1426 Residual 331.652 48 6.9094 0.1654 Snout length Total 15.6258 49 Linear regression 0.7362 1 0.7362 Residual 14.8896 48 0.3102 2.3732 Pre-spawning energy content Total 114.713 49 Linear regression 2.6229 1 2.6229 Residual 112.090 48 2.3352 1.1232 Male preference Body length Total 186.535 49 Linear regression 4.4899 1 4.4899 Residual 182.045 48 3.7926 1.1839 Snout length Total 6.6418 49 Linear regression 0.3472 1 0.3472 Residual 6.2946 48 0.1311 2.6478 Pre-spawning energy content Total 52.2597 49 Linear regression 0.1002 1 0.1002 Residual 52.1595 48 1.0867 0.0922 Note: None of the above tests are statistically significant at the 0.05 alpha level. 31 Behavioural and genetic measures The variables that did not follow a normal distribution (Fitness 2 and proportion of social mates in females, and proportion of social mates in males) were rank transformed to conform to the assumptions of parametric statistics (i.e. followed a normal distribution after transformation). The transformed variables were used for subsequent analyses. The four individual male mating success indices used in the analysis are displayed in Table 2.7. In this table, I have shown both the consort score per fish for each day the fish was observed and the average consort score that was used in the analysis. I have also shown the dominance ranking of each fish, the total number of days alive on the spawning grounds, and the number of social mates acquired. The males on this table are congruent with the males on Table 2.5 (the number of offspring sired by each male). In males, the dominance index was positively related to the percent of time he was in a consort position (n -27,r = 0.885, P < 0.0001). Both dominance and consort score were highly correlated with the proportion of social mates acquired (n = 27, r = 0.610, P = 0.0007; n = 27, r = 0.649, P = 0.0003, respectively). However, the correlations between longevity and all other social mating success measures were non-significant, although positive (Table 2.8). The two indices of genetic reproductive success in males were highly correlated (n = 27, r = 0.667, P = 0.0001). Among the females, none of the behavioural indices of mating success was correlated (see Table 2.9) except for consort score and the weighted consort score (n = 20, r = 0.739, P = 0.0002). Within the genetic measures, Fitness 2 and Fitness 3 in females were highly correlated (n = 20, r = 0.736, P = 0.0002). The number of social mates acquired by females is shown on Table 2.5. 32 Table 2.7 The social mating patterns of males in each study enclosure. The following are summarized in the table: the consort score per day, the average consort score for the entire spawning period, the dominance index, the longevity and the number of social mates acquired by each male. Enclosure 1 male 1 male 2 male 3 male 4 male 5 male 6 male 7 Dayl n/a 0 1 1 0 1 1 Day 2 n/a 0 1 0.5 0 0.5 1 Day 3 n/a 0.5 1 0.5 0 n/a 1 Consort score n/a 0.17 1 0.67 0 0.75 1 Dominance index n/a 1 6 2 3 4 5 Longevity n/a 8 8 6 7 5 10 # of social mates n/a 0 3 3 0 2 2 Enclosure 2 male 1 male 2 male 3 male 4 male 5 male 6 male 7 Dayl 0 1 0 1 0.25 0 1 Day 2 0 0.5 1 0.25 0 1 1 Day 3 0 0.25 0.25 1 1 0.25 0 Day 4 0.25 0.25 0.25 1 1 0.25 n/a Consort score 0.06 0.5 0.38 0.81 0.56 0.38 0.67 Dominance index 1 5 2 7 6 3 4 Longevity 6 10 8 9 9 9 5 # of social mates 0 2 1 3 1 1 1 Enclosure 3 male 1 male 2 male 3 male 4 male 5 male 6 male 7 Dayl 0 1 1 1 0 1 1 Day 2 1 1 1 1 0 1 1 Day 3 0 1 0 0 n/a 1 1 Consort score 0.33 1 0.67 0.67 0 1 1 Dominance index 2 5 3 4 1 6 7 Longevity 10 14 9 7 4 12 12 # of social mates 1 2 2 1 0 2 1 Enclosure 4 male 1 male 2 male 3 male 4 male 5 male 6 male 7 Dayl 0 1 0 0 1 0 0.5 Day 2 0.5 1 0.25 0.25 1 0 1 Day 3 0 1 0 0 1 0.25 0.5 Day 4 0.25 1 0 0 n/a 0 1 Consort score 0.19 1 0.06 0.06 1 0.06 0.75 Dominance index 4 7 2 1 5 3 6 Longevity 8 6 5 9 5 9 11 # of social mates 1 2 0 0 1 1 2 33 Table 2.8 Correlation measures among behavioural mating success indices and among genetic reproductive success indices in males. Behavioural/Genetic measures Correlation coefficient P-value Dominance vs. consort score 0.8847 < 0.0001* Dominance vs. longevity 0.3948 0.0415 Dominance vs. proportion of social mates 0.6104 0.0007* Consort score vs. longevity 0.2629 0.1852 Consort score vs. proportion of social mates 0.6487 0.0003* Longevity vs. proportion of social mates 0.2303 0.2477 Proportion of offspring sired vs. proportion of genetic mates acquired 0.6382 0.0003* * Statistically significant at the P = 0.05 level (Bonferroni corrected). 34 Table 2.9 Correlation measures among behavioural mating success indices and amon genetic reproductive success indices in females. Behavioural/Genetic measures Correlation coefficient P-value Territoriality vs. consort score 0.0307 0.8979 Territoriality vs. longevity -0.0460 0.8472 Territoriality vs. wt consort score -0.1369 0.5650 Territoriality vs. proportion of social mates -0.0257 0.9143 Consort score vs. longevity 0.3182 0.1715 Consort score vs. wt consort score 0.7394 0.0002* Consort score vs. proportion of social mates -0.2005 0.3966 Longevity vs. wt consort score 0.3272 0.1590 Longevity vs. proportion of social mates -0.1482 0.5330 Wt consort score vs. proportion of social mates -0.3527 0.1272 Proportion of genetic mates acquired vs. Fitness 2 -0.3292 0.1564 Proportion of genetic mates acquired vs. Fitness 3 -0.3291 0.1565 Fitness 2 vs. Fitness 3 0.7364 0.0002 * Statistically significant at the P = 0.05 level (Bonferroni corrected) 35 Behaviour versus genetic mating success Dominance was a reasonable predictor of both genetic measures of reproductive success (dominance vs. proportion of offspring: r 2 = 0.396, df = 25, P = 0.0004; dominance vs. proportion of mates: r 2 = 0.176, df = 25, P = 0.0296) (Fig. 2.4). Consort score was also a reasonable predictor of both genetic measures (consort score vs. proportion of offspring: r = 0.368, df = 25, P = 0.0008; consort score vs. proportion of mates: r 2 = 0.147, df = 25, P = 0.0480) (Fig. 2.5). However, with the sequential Bonferroni corrections, only the proportion of offspring sired can be predicted by the two measures of social mating success. The proportion of social mates acquired was highly correlated with the proportion of offspring sired (r2 = 0.328, df = 25, P = 0.0018) (Fig. 2.6). Longevity was positively but not significantly correlated with genetic reproductive success. To test whether longevity was significant when controlling for dominance, the residuals from the appropriate regressions were related to the two measures of genetic reproductive success. In neither case was longevity a significant determinant of either of the two reproductive success measures. Table 2.10 summarizes the results for this section. For females, none of the social mating success measures was a good predictor of the genetic reproductive success measures. Longevity was a reasonable predictor of Fitness 3 (F\,\s = 4.52, Fitness 3 = 2.4285 + 0.2576(longevity), r 2 = 0.2006, df = 18, P = 0.0477), however, with the Bonferroni correction, it is not statistically significant. Discussion The primary objective of this study was to report the first direct estimates of individual reproductive success in sockeye salmon based on the number of eggs fertilized in open 36 2 3 4 5 6 dominance index (rank) o 1.00 -enet red 0.80 - • • • O) '5 E T 0.60 - • • • o c g es act 0.40 . • • • • • es act • • • o ra 0.20 . • • • 1^  = 0.18 Q. O E • P = 0.0296 L_ Q. 0 00 I I | V W 1 t I 0 1 2 3 4 5 6 7 8 dominance index (rank) Figure 2.4 Scatter plots of the genetic reproductive success measures as predicted by dominance in males. 37 0.60 •o 0.50 -o c sire 0.40 -o •E O) c 0.30 -o 'c a o Q. 10 0.20 -i_ a Off 0.10 .. 0.00 -. 3 4 5 consort score (rank) 3 4 5 consort score (rank) Figure 2.5 Scatter plots of the genetic reproductive success measures as predicted by the consort score in males. 38 1^  = 0.33 P = 0.0018 0.1 0.2 0.3 0.4 0.5 0.6 proportion of social mates acquired 0.8 <B -o a> !: cn "5 o S" c « o tn t * o ra 1.00 -r 0.80 , 0.60 - • 0.40 ! 0.20 | » • 0 00 | • 1 0.1 w 1 0.2 0.3 0.4 0.5 0.6 proportion of social mates acquired 1^  = 0.13 P = 0.064 0.7 0.8 Figure 2.6 Scatter plots of the genetic reproductive success measures as predicted by the proportion of social mates acquired in males. 39 Table 2.10 Least squared regression models relating genetic reproductive success measures to social mating success measures in males. Dependent (genetic) variable versus independent (behavioural) variable F i , 2 5 model r 2 P-value Proportion of offspring sired (Gl) vs. Dominance (D) 16.41 G l = 0.0121 + 0.0348(D) 0.3963 0.0004 Percent Consort (C) 14.57 G l = 0.0130 +0.0346(C) 0.3683 0.0008 Longevity (L) 1.37 G l = 0.0650 + 0.0101(L) 0.0518 0.2535 Longevity (L2) (independent of dominance) 0.02 G l = 0.1476-0.0014(L2) 0.0008 0.8901 Proportion of social mates (S) 12.22 G l = 0.1476+ 0.0668(S) 0.3283 0.0018 Proportion of mates acquired (G2) vs. Dominance (D) 5.32 G2= 0.1937 +0.0469(D) 0.1756 0.0296 Percent Consort (C) 4.32 G2= 0.2038 + 0.0443(C) 0.1474 0.0480 Longevity (L) 0.14 G2= 0.3212+ 0.0067(L) 0.0056 0.7114 Longevity (L2) (independent of dominance) 0.27 G l = 0.3759-0.0100(L2) 0.0105 0.6108 Proportion of social mates (S) 3.76 G l = 0.3759 + 0.0852(S) 0.1306 0.0640 40 competition within study enclosures under otherwise natural conditions. I established two measures of paternal reproductive success, and three indices of maternal success based on the assignment of offspring to candidate parents. In addition, I tested the prediction of a positive association between behavioural measures of mating success and realized genetic reproductive success (mate choice in females). The microsatellite approach was efficient in establishing parent-offspring relationships. Using only four polymorphic microsatellite loci in combination with a maximum-likelihood method (Marshall et al. 1998), and knowing the mother of each offspring, I was able to assign 100% of the offspring to a single parental pair, with a confidence level of 95%. M y results demonstrate a high incidence of multiple mating in both sexes. Given the small number of offspring that I sampled, the proportion of mates acquired is almost certainly underestimated, and my estimates of variance in reproductive success should be considered conservative. A few recent population studies on fish have evaluated the variance of individual genetic reproductive success based on sexually mature adults. Gross and Kapuscinski (1997) found a high variance in individual reproductive success for male smallmouth bass, where 5.4% of all spawning males produced 54.7% of the offspring collected. Garant et al. (2001) found the ratio between the variance and the mean number of assigned offspring was 8.9 for females and 8.6 for males in Atlantic salmon. And Neff (2001) found the mean paternity of the 38 parental males was 78.9%, with a range of 26-100% in bluegill sunfish. M y results are consistent with these studies and with speculations from previous behavioural studies on Pacific salmon suggesting that adult male behaviour is focused towards access to matings. This can create a great variation in mating success among males as a few males have primary access to ovipositing females (Schroder 1982; Fleming and Gross 1994; Foote el al 1997; Hamon et al. 1999). The occurrence of multiple partners in female Pacific salmon spawning is less well documented. I 41 found that females spawned with multiple partners and that the mean and variance in partners was as high for females as for males. Given that sampling of fertilized eggs within an individual redd was partial, the estimate of multiple mating is a minimum estimate. It was not possible to differentiate between simultaneous contributions to single nests and consecutive male contributions to sequential nests, as there was movement of eggs within the nests both from natural causes and during egg collection. Multiple mating within a spawning event may be unavoidable, as the dominant male cannot both chase away other males and fertilize eggs during actual spawning. Similarly, the female cannot completely control who is releasing sperm while she is releasing ova. The goal of both the female and consort may be to exclude other partners. On the other hand, the benefit of multiple mating (from one spawning event to the next) to the sex whose reproduction is limited by the opposite sex (often the males) is easy to understand. The switch in dominant partner between spawning events may be more interesting in regards to fitness consequences and may reflect more on female choice. The fitness advantages of multiple mating by females are less obvious and may exist for a variety of reasons. Multiple partners may convey a form of fertilization insurance, in the case where a single male may be sterile or has low fertility. It may be a result of bet-hedging, increasing the likelihood that at least some progeny will emerge and live to adult stages, or it may be a tactic to generate greater genetic diversity among progeny, which may have a selective advantage (Avise 1994; Barlaup et al. 1994; Fleming 1996). Multiple mating may also reduce inbreeding or reduce the competition among genetically dissimilar siblings (Reynolds 1996). Whatever the cause, multiple mating was just as frequent in females as it was in males. 42 Perhaps the answer to multiple mating by females is contained in the concept that mate choice may yield genetic benefits to the offspring. For example, previous studies on Pacific salmon have suggested that females seem to prefer large, well developed, and aggressive males (Schroder 1981; Foote 1989). Such morphological and behavioural traits are costly and often opposed by selection for survival (Darwin 1871). If it is only the individuals of relatively high quality that are able to bear the cost of dominance, the position in the hierarchy per se or traits indicating dominance will reliably reflect certain aspects of mate quality (Qvarstrom and Forsgren 1998). Thus, males with a relatively large expression of one or more of these indicator traits will be at an advantage. Indeed, this was the general trend, where females tended to mate with the more dominant males and those who displayed strong consort behaviour. When it is obvious that only one male is superior to the others, the female in question could choose to mate only with that male. In this case, she would have a fitness advantage over the female who mates repeatedly with different "inferior" males. This was the case with four females who mated repeatedly with one male only (see Table 2.5). These males were often ranked high based on their dominance index and were always courting a female. The remaining 16 females had more than one genetic partner, suggesting that most males fathered at least a small proportion of the offspring. Indication of assortative mating has been found in Pacific salmon (Hanson and Smith 1967; McCart 1970; Foote 1988). This was not the case in my study. Foote (1988) demonstrated that male mate choice is dependent on male size, where males of all sizes discriminate against females smaller than themselves, but not between females of their own size and larger. There was no clear pattern of mate pairing in my study based on overall body size, body shape and pre-spawning energy content, suggesting that a range of individual tactics was employed. A failure to detect evidence of assortative mating may have been due to the enclosure design, as males and 43 females could only mate with the fish within their enclosure, or due to the range in female body size in my study. The range may not have been as great as those found in previous studies where size-assortative mating was detected. Indeed, in my study there was no female that was larger than the smallest male in the same enclosure. Thus, males could not afford to discriminate against females smaller than themselves. In a recent study using D N A minisatellite profiling on Atlantic salmon, Taggart et al. (2001) also found no clear pattern of mate pairing among spawners. On the one hand, it is quite possible that the use of genetic analysis provides a more reliable estimate of mating systems and actual mate preference, and therefore, studies using only observational data should be viewed with caution. On the other hand, D N A profiling also has its limitations in that not all spawning events are usually analysed due to time and the cost of genetic work, the same constraint that often applies to behavioural observations as well. Studies on mating success in Pacific salmon have focused on behavioural measures of success, assuming that large, aggressive or consort males fertilize most of the eggs laid by the female (Schroder 1982; Chebanov et al. 1983; Gross 1985; Maekawa and Onozato 1986; Hutchings and Myers 1988; Foote 1990; Quinn and Foote 1994). Field observations provide only a limited view of the extent and pattern of spawning. As indicated in this study, multiple mating seems to be the rule rather than the exception. Typically, studies examining the mating success of organisms through observational data may experience difficulties in observing covert mating behaviour (extra-pair fertilization or sneaky male strategies) and nocturnal behaviour (Coltman et al. 1999). Behavioural indices of mating success were found to be a reliable indicator of the realized genetic reproductive success in male sockeye salmon. Dominance, consort score and the number of social mates were significant determinants of the genetic reproductive success of males 44 (measured as the proportion of offspring sired). However, these indices explained only 33-40% of the variation in male genetic reproductive success. Studies that have examined the predictive value of behavioural observations in mammals and lizards have shown that as much as 90% of the variance in male reproductive success could be predicted from behavioural data (Pemberton et al. 1992; Gullberg et al. 1997). Unlike the present study, however, where only a small fraction of the entire clutch of eggs was analysed, these studies analysed most of the fertilized offspring. As a result, sampling error must be taken into account. Biologically, the measures of behavioural mating success could be poor indicators of social success. Perhaps the behavioural mating success measures I used were correlated with an unmeasured factor that would have explained the variation in genetic reproductive success more accurately. The fact that males seen in consort positions, engaged in more aggressive acts, or with a high number of social partners were more likely to be the sire than other males means that observing associations between individuals provides useful information for predicting paternity. Although on the surface, this conclusion may suggest that more intensive observations will lead to useful predictions about paternity, measures of behaviour alone can sometimes be misleading in predicting realized reproductive success. In this study, dominance, consort score, or the number of social mates may underestimate the success of sneaker males (or extra-pair fertilizations), as indicated in Figures 2.4 and 2.5 (Also see Tables 2.5 and 2.7). One male, who was ranked 1/7 on the dominance scale, 2/7 on the consort scale, who did not have a single social mate and was also the smallest male in the whole study (body length = 62.30 cm), was quite successful in mate acquisition (80%) and egg fertilization (18%). In fact, he acquired the most mates relative to all other males in all enclosures (with the exception of one other male who also acquired 80% of the females as mates). The use of any of the behavioural observations alone would have indicated a low mating success for this particular male, quite different from his realized reproductive 45 success. The finding that the number of social mates and the actual number of genetic mates acquired by males are not highly correlated further suggests that there is a gap between the two kinds of measures (behavioural versus genetics). Past studies focusing on behavioural observations alone may, thus, be poor indicators of reproductive success in males. The three behavioural measures (dominance, consort score and number of social partners) used to assess male mating success were highly correlated, suggesting that dominant males are often found in consort positions, and acquire relatively more social mates. The use of any of these three measures may suffice in the assessment of social mating success. For any of the measures, great care must be taken in the collection of the data. A male holding close to a female but outside her redd is not necessarily in a consort position, he may be there only as an observer or because of a lack of space. In addition, although a male may not be aggressive, he may still be in a consort position for the spawning period, not because he is a weak competitor, but because his position in the hierarchy was established without the need to fight. These scenarios could hold true only in the study enclosures where the male population is fixed. In an open spawning area, however, there are usually large unattached males roaming around testing the consorts so a dominant male would not remain a consort or acquire a high number of social mates for long without having to fight to maintain his position. So, although the three measures may not always be associated, they are approximately equivalent when assessing mating success in the long term. Similarly, the two measures of genetic reproductive success (proportion of offspring sired and proportion of mates acquired) were highly correlated, where the proportion of mates acquired was associated with how successful a male was at siring offspring. If one is interested in the relative contribution of individual males to subsequent generations, the proportion of offspring sired may be a more accurate estimate of reproductive success. However, in situations where 46 most offspring die before emergence (70+% in salmon) and where destruction of redds may be a somewhat random occurrence, spreading your offspring among a lot of nests may be as important to fitness as fertilizing a lot of eggs in one nest. The number of mates acquired may be inflated for a sneaker male, who manages to fertilize the eggs of many females, but in so doing, only a few eggs in each case are actually fertilized. His reproductive success in terms of the number of mates acquired may be high, but the number of offspring he actually sired could be low. Thus, the trade-offs that distinguish sneaker males from dominants may be that the dominant males try to fertilize a lot of eggs whereas the sneaker males try to get a lot of mates. Ideally, one would assess both these measures. For the female mating system, two social mating success measures were highly correlated (consort score vs. weighted consort score: r = 0.74). Whereas the consort score was a measure of the percent of time a female was the partner of a consort male, the weighted consort score took into account the social status of the males with which she consorted. As consort score and dominance rank were highly correlated in males, it is no surprise that these female measures are also related. Thus, it is reasonable to conclude that the consort score may be sufficient to assess the social mating success of females, as it seems to be both reliable and easier to record than the weighted consort score. The comparison of the behavioural mating success to the genetic reproductive success was more challenging with the females. Defining measures of genetic success for females that go beyond fecundity is problematic when actual contribution to future generations cannot be measured. As an alternative, I derived indices of the genetic success of females from the reproductive success of the males with whom they mated. This approach assumes that fitness consequences for the female flow, in part, from the quality of the males who fertilize her eggs. Longevity, a 47 behavioural index, was the only reasonable predictor of female genetic success indexed by the dominance ranking of her genetic partners (Fitness 3), although it was not significant after a Bonferroni correction test. Biologically, it seems reasonable to suggest that the longer a female is alive, the longer she can defend her nests from being dug-up by other females (Quinn and Foote 1994). This may therefore have positive fitness consequences as the probability of her nest being destroyed is minimized the longer she lives. The length of spawning life may be indicative of the fitness of a female, which may influence the quality of the mate with whom she mates. M y results indicate that in the general sense paternity can be predicted from the behaviour of competing males. The specific relationships between male dominance, consort scores or number of social partners and genetic mating success presented here are probably specific to the particular situation of my study. Nevertheless, considerable variation in genetic reproductive success was unexplained by behaviour even in the constrained environment of my spawning enclosures. Despite the highly developed behaviours and social dominance morphology of these animals, reproductive success may still be highly stochastic. The concept of female reproductive success in terms of the kinds and numbers of males that fertilize her eggs has not been much explored in the literature on mating success in salmon. Based on work on other species in which female mate choice is more apparent, I have proposed that female success can be indexed by reference to the quality of the males who fertilize her eggs. For females, longevity seems to be an acceptable indicator of her genetic mate choice. However, since none of the indices were statistically significant, alternatives to genetic fingerprinting in assessing mate choice of females seem to be unreliable. 48 Literature cited: Abell, A . J. 1997. Estimating paternity with spatial behaviour and D N A fingerprinting in the striped plateau lizard, Sceloporus virgatus (Phrynosomatidae). Behavioural Ecology and Sociobiology 41:217-226. Altmann, J., S. Alberts, and S. A . Haines. 1996. Behaviour predicts genetic structure in a wild primate group. Proceedings of the National Academy of Sciences of the USA 93:5797-5801. Amos, W., S. Twiss, P. Pomeroy, and S. Anderson. 1993. Male mating success and paternity in the grey seal, Halichoerus grypus, a study using D N A fingerprinting. Proceedings of the Royal Society of London, Series B 252:199-207. Avise, J. C. 1994. Molecular Markers, Natural History and Evolution. Chapman & Hall, London. Barlaup, B. T., H . Lura, H . Saegrov, and R. C. Sundt. 1994. Inter- and intra-specific variability in female salmonid spawning behaviour. Canadian Journal of Zoology 72:636-642. Berglund, A. , A . Bisazza, and A. Pilastro. 1996. Armaments and ornaments: an evolutionary explanation of traits of dual utility. Biological Journal of the Linnean Society 58:385-399. Brett, J. R. 1995. Energetics. In Physiological ecology of Pacific salmon. (Groot, C , Margolis, L. and Clarke, W. C , eds), pp. 1-68. Vancouver: U B C Press. Chebanov, N . A . , N . V . Varnavskaya, and V. S. Varnevskii. 1983. Assessment of successful participation in spawning of male sockeye salmon Oncorhynchus nerka (Walbaum) (Salmonidae) of different hierarchical status with the help of genetic-biochemical markers. Voprosy Ikhtiologii 23:114-11%. Coltman, D. W., D. R. Bancroft, A . Robertson, J. A . Smith, T. H . Clutton-Brock, and J. M . Pemberton. 1999. Male reproductive success in a promiscuous mammal: behavioural estimates with genetic paternity. Molecular Ecology 8:1199-1209. Coltman, D. W., W. D. Bowen, and J. M . Wright. 1998. Male mating success in an aquatically mating pinniped, the harbour seal (Phoca vitulina), assessed by microsatellite D N A markers. Molecular Ecology 7:627-638. Cox, C. R., and B . J. Leboeuf. 1977. Female incitation of male competition: a mechanism in sexual selection. American Naturalist 111:317-335. Crawford, B. A . , S. A . Leider and J. M . Tipping. 1977. Technique for rapidly taking samples of skeletal muscle from live adult steelhead trout. Progressive Fish Culturist 39:125. Crossin, G. 2002. Selection for a bioenergetic-morphological feedback mechanism in migrating adult salmon is related to migratory distance and elevation. MSc thesis, University of British Columbia, Vancouver. Darwin, C. 1859. The origin of species by means of natural selection, or The preservation of 49 favored races in the struggle for life. London J. Murray, London. Darwin, C. 1871. The descent of man, and selection in relation to sex. J. Murray, London. Emlen, S. T., and L. W. Oring. 1977. Ecology, sexual selection and the evolution of mating systems. Science 197:215-223. Fleming, I. A . 1996. Reproductive strategies of Atlantic salmon: ecology and evolution. Reviews in Fish Biology and Fisheries 6:379-416. Fleming, I. A . , and M . R. Gross. 1994. Breeding competition in a Pacific salmon (coho: Oncorhynchus kisutch): measures of natural and sexual selection. Evolution 48:637-657. Foote, C. J. 1988. Male mate choice dependent on male size in salmon. Behaviour 106:63-80. Foote, C. J. 1989. Female mate preference in Pacific salmon. Animal Behaviour 38:721-723. Foote, C. J. 1990. A n experimental comparison of male and female spawning territoriality in a Pacific salmon. Behaviour 115:283-314. Foote, C. J., G. S. Brown, and C. C. Wood. 1997. Spawning success of males using alternative mating tactics in sockeye salmon, Oncorhynchus nerka. Canadian Journal of Fisheries and Aquatic Sciences 54:1785-1795. Garant, D. , J. J. Dodson, and L. Bernatchez. 2001. A genetic evaluation of mating system and determinant of individual reproductive success in Atlantic salmon (Salmo salar L.). The Journal of Heredity 92:137-145. Groot, C., and L. Margolis. 1991. Pacific salmon life histories. U B C Press, Vancouver. Gross, M . L . , and A . R. Kapuscinski. 1997. Reproductive success of smallmout bass estimated and evaluated from family-specific D N A fingerprints. Ecology 78:1424-1430. Gross, M . R. 1985. Disruptive selection for alternative life history strategies in salmon. Nature 313:47-48. Gullberg, A . , M . Olsson, and H . Tegelstrom. 1997. Male mating success, reproductive success and multiple paternity in a natural population of sand lizards: behavioural and molecular genetics data. Molecular Ecology 6:105-112. Hamon, T. R., C. J. Foote, and G. S. Brown. 1999. Use of female nest characteristics in the sexual behaviour of male sockeye salmon. Journal of Fish Biology 55:459-471. Hanson, A . J., and H . D. Smith. 1967. Mate selection in a population of sockeye salmon (Oncorhynchus nerka) of mixed age group. Journal of Fisheries Research Board of Canada 24:1955-1977. Healey, M . , and A . Prince. 1998. Alternative tactics in the breeding behaviour of male coho salmon. Behaviour 135:1099-1124. 50 Hendry, A . P., O. K. Berg and T. P. Quinn. 2001. Breeding location choice in salmon: causes (habitat, competition, body size, energy stores) and consequences (life span, energy stores). Oikos 93:407-418. Hughes, C. 1998. Integrating molecular techniques with field methods in studies of social behaviour: a revolution results. Ecology 19:383-399. Hutchings, J. A . , and R. A . Myers. 1988. Mating success of alternative maturation phenotypes in male Atlantic salmon, Salmo salar. Oecologia 75:169-174. Lebas, N . 2001. Microsatellite determination of male reproductive success in a natural population of the territorial ornate dragon lizard, Ctenophorus ornatus. Molecular Ecology WA93-203. Maekawa, K. , and H . Onozato. 1986. Reproductive tactics and fertilization success of mature male Miyabe charr, Salvelinus malma miyabei. Environmental Biology of Fishes 15:119-129. Marshall, T. C , J. Slate, L . E. B. Kruuk, J. M . Pemberton. 1998 Statistical confidence for likelihood-based paternity inference in natural populations. Molecular Ecology 7:639-655. McCart, P. J. 1970. A polymorphic population of Oncorhynchus nerka in Babine Lake, British Columbia. University of British Columbia, Vancouver. Middendorf, L. R., J. C. Bruce, R. C. Bruce, R. D. Eckles, D. L . Grone, S. C. Roemer, S. C. Sloniker, G. D. Steffens, D. L. Sutter, J. A . Brumbaugh, and G. Patonay. 1992. Continuous, on-line D N A sequencing using a versatile infrared laser scanner/electrophoresis apparatus. Electrophoresis 13:487-494. Moller, A . P., and T. R. Birkhead. 1993. Cuckoldry and sociality: a comparative study of birds. American Naturalist 142:118-140. Neff, B. 2001. Genetic paternity analysis and breeding success in Bluegill sunfish (Lepomis macrochirus). The Journal of Heredity 92:111-119. Nelson, R. J., and T. D. Beacham. 1999. Isolation and cross species amplification of microsatellite loci useful for study of Pacific salmon. Animal Genetics 30:228-229. Oetting, W., H . Lee, D. Flanders, G. Wiesner, T. Sellers, and R. King. 1995. Linkage analysis with multiplexed short tandem repeat polymorphisms using infrared fluorescence and M13 talked primers. Genomics 30:450-458. Olsen, J. B., S. L . Wilson, E. J. Kretschmer, K . C. Jones, J. E. Seeb. 2000. Characterization of 14 tetranucleotide microsatellite loci derived from sockeye salmon. Molecular Ecology 9:2185-2187. Pemberton, J. M . , S. D. Albon, F. E. Guinness, T. H . Clutton-Brock, and G. A . Dover. 1992. 51 Behavioural estimates of male mating success tested by D N A fingerprinting in a polygynous mammal. Behavioural Ecology 3:66-75. Prince, A . , S. Low, T. Lissimore, R. Diewert, and S. Hinch. 1995. Sodium bicarbonate and acetic acid: an effective anesthetic for field use. North American Journal of Fisheries Management 15: 170-172. Quinn, T., C. Wood, L . Margolis, B. Riddell, and K. Hyatt. 1987. Homing in wild sockeye salmon (Oncorhynchus nerka) populations as inferred from differences in parasite prevalence and allozyme allele frequencies. Canadian Journal of Fisheries and Aquatic Sciences 44:1963-1971. Quinn, T. P., and C. J. Foote. 1994. The effects of body size and sexual dimorphism on the reproductive behaviour of sockeye salmon, Oncorhynchus nerka. Animal Behaviour 48:751-761. Qvarnstrom, A . , and E. Forsgen. 1998. Should females prefer dominant males? Trends in Ecology and Evolution 13:498-501. Reynolds, J. D. 1996. Animal breeding systems. Trends in Ecology and Evolution 11:68-72. SAS 2001. SAS/STAT Windows version 8.2. SAS Institute, North Carolina, USA. Schroder, S. 1981. The role of sexual selection in determining the overall mating patterns and mate choice in chum salmon. PhD thesis, University of Washington, Seattle. Schroder, S. 1982. The influence of intrasexual competition on the distribution of chum salmon in an experimental stream. In: Proceedings of the salmon and trout migratory symposium (E.L. Brannon & E.O. Salo, eds), pp. 275-285. University of Washington Press, Seattle. Steffens, D. , S. Sutter, and S. Roemer. 1993. A n alternate universal forward primer for improved automated D N A sequencing of M13. BioTechniques 15:580-581. Taggart, J., I. McLaren, D. Hay, J. Webb, and A. Youngson. 2001. Spawning success in Atlantic salmon (Salmo salar L.): a long-term D N A pro filing-based study conducted in a natural stream. Molecular Ecology 10:1047-1060. Zar, J. H . 1999. Biostatistical analysis (4th edition). Prentice Hall, Inc. Upper Saddle River, New Jersey. 52 Chapter 3- Factors affecting spawning behaviour Introduction The most prominent period of an organism's life is reproduction, when resources are directed away from acquiring food and avoiding predators and toward acquiring mates and breeding successfully (Stearns 1989, 1992; Roff 1992). Reproduction is central to fitness, for the contribution of an organism to future generations depends on its reproductive success (Stearns 1977). Understanding reproduction is a focal biological problem. Within constraints intrinsic to the organism, reproductive traits are subject to trade-offs because of energy limitations (e.g. secondary sexual characters, intrasexual aggression, and parental care; for a review in fishes, see Wootton 1990). As a consequence, the underlying theoretical problem in the evolution of the model for reproduction is to predict how various traits (behavioural, morphological and physiological) will be combined by evolution to maximize the organism's lifetime reproductive output. Because of the many energetic constraints associated with reproduction, the trade-offs associated with this stage are most evident in organisms with a low energy intake and a high energy use during this critical stage (Hendry and Berg 1999). Pacific salmon (genus Oncorhynchus), which are anadromous and semelparous, provide a good model by which to explore these trade-offs because they experience severe energy constraints during reproduction. Feeding ceases upon arrival in freshwater, sometimes as much as several months before spawning begins. Reproductive investment and breeding take place at the expense of stored energy, and all fish die after a single spawning season (Gilhousen 1980; Brett 1995). As they must complete their upstream migration and spawning with the energy reserves they bring from the oceanic feeding grounds, salmon face conflicting demands for that energy once they have reach freshwater. Because they do not have to invest energy into future reproduction, they can afford to allocate a large portion of their fuel for the upstream migration 53 and the development of their gonads and extreme secondary sexual characteristics. However, some energy must be saved for use on the spawning grounds, where males expend energy fighting to gain access to females and females expend energy to construct and guard their spawning nest site. The strict semelparity of these species and the limitation on energy available for reproduction intensify natural and sexual selection on energy allocation schedules to maximize reproductive success (Fleming and Gross 1994; Quinn and Foote 1994; Hendry and Berg 1999). During the upstream migration female Pacific salmon develop large ovaries and to a lesser extent secondary sexual traits, whereas males develop very exaggerated secondary sexual characteristics (an enlarged hooked snout and a humped back). From river entry to spawning ground arrival, sockeye expend up to 20% of their body proteins and up to 86% of their body lipids, representing a 53% loss of energy stores (Crossin 2002). Thus, once on the breeding grounds, salmon have already used up a large portion of their energy reserves, limiting the energy available for the completion of their life cycle. Crossin (2002) has shown that from spawning ground arrival to death, sockeye utilize up to 25% of the remaining somatic proteins and 50% of their lipids. This represents a loss of an additional 30% of their gross energy, consistent with the work of Hendry and Berg (1999) who have shown that females lose 74% and males lose 66% of their total somatic energy between freshwater entry and death. Trade-offs in the allocation of limited energy into various reproductive components is, thus, expected in Pacific salmon. Space, time and energy constraints combine to make reproduction an extremely competitive activity in salmon. Male salmon exhibit high rates of aggression and courtship display to gain temporary access to nesting females (Quinn et al. 1996; Quinn and McPhee 1998), both of which are energetically costly (Neat et al. 1998; Hendry and Berg 1999; Hendry et al. 1999). The 54 exaggerated secondary sexual traits in males are consistent with sexual selection favouring individuals with long hooked snouts (Fleming and Gross 1994) and deeper bodies (Quinn and Foote 1994). The former functions as a weapon during fighting (Gross 1984), and the latter is thought to shield the bearer against sneaking males while spawning and to protect it from attacks while fighting. The chance of being the victor of male-male competition increases with size (Foote 1990; Fleming and Gross 1994; Quinn and Foote 1994). The winner is typically the most dominant male and he then tends to occupy the position closest to the nesting female, thus increasing his reproductive success (Schroder 1982; Chebanov et al. 1983; Fleming and Gross 1994). The competitive scene among females is also intense on spawning grounds. Females compete for access to optimal nesting sites (Schroder 1982; Gross 1985; Fleming and Gross 1989; Foote 1990; Quinn and Foote 1994). Their traits may also be indicative of direct sexual selection, because the same secondary sexual traits (hooked snouts especially) influence competitive ability in males and in females (Fleming and Gross 1994). Larger females are often the winners of intrasexual competition for preferred nesting sites (Foote 1990; Fleming and Gross 1994). Female size is also important in reducing egg loss through nest destruction because larger females dig deeper nests, in part due to their deeper caudal peduncles (Van den Bergh and Gross 1989; Fleming and Gross 1994). Some studies have shown that larger salmonid females also tend to make a proportionately larger investment in gonad development (Fleming and Gross 1994; Jonsson et al. 1996; Hendry et al. 2001; Kinnison et al. 2001; Crossin 2002), while data on chinook salmon indicates that larger females actually make a proportionately smaller investment in gonads than smaller females (Healey 2001). Thus, population or species-type may, in part, influence gonad investment. Both gonad investment and intrasexual competition are energetically costly. 55 M y primary objective in this study was to measure trade-offs among different aspects of reproductive investment in sockeye salmon (Oncorhynchus nerka). The traits under consideration were secondary sexual characteristics, energy reserves (fat), and spawning life span. Caudal peduncle depth was also considered in females. Increased expression of any of these traits should be associated with increased reproductive success and, therefore, increased fitness. As all reflect a resource allocation decision either before or during spawning, all are involved in the trade-offs associated with reproduction. Analysis of the relationships among these traits and reproductive success should give insights into the way energy is allocated to maximize fitness in sockeye salmon. For instance, increased investment in the development of secondary sexual traits or stored energy without compromising the investment into the other trait should increase the fitness of an individual. Whereas excess energy reserves may aid in the competition for mates, exaggerated secondary sexual traits may increase the chance of winning those contests. I expected to find a trade-off between pre-spawning energy reserves and secondary sexual characteristic development. On the other hand, because bigger fish are more aggressive, and because aggression increases energy expenditure and the possibility of injury in fish, those fish with exaggerated traits should have a decreased reproductive life span. Furthermore, energy stores at the start of breeding may also be positively correlated with life span. The second goal of this study was to evaluate quantitatively the importance of specific life history traits to the spawning behaviour (social mating success) of the fish and to determine the traits that are favoured by sexual selection during this highly competitive phase. Field observations of spawning behaviour that were found to be reasonable predictors of genetic reproductive success (see Chapter 2) were used to study the phenotypic components determining 56 the outcome of success among both males and females. For males, dominance, consort score and the number of social mates acquired were the social mating success measures used. For females, territoriality, longevity (an estimate of the duration of nest guarding), and a measure of the consorting of females with different status males were used for social mating success measures. Methods A complete description of the overall study design, including the study site, study species, morphological and physiological measurements, behavioural data collection, and social mating success indices can be found in Chapter 2 (pages 10-20). A n additional morphological variable was measured for use in the present study. The caudal peduncle depth, at its shallowest point, was measured for all study fish. Three indices of male social mating success were used. The first was a dominance index for each male within an enclosure. Based on this index, each male was ranked from one (least dominant) to seven (most dominant). This index is called the dominance index. The second index was a measure of consort behaviour. Each male was given a consort score based on his courting behaviour and proximity to ovipositing females. This index was called consort score. The third measure was the number of social mates acquired by each male, called the proportion of social mates acquired. Three indices for the social mating success of females were measured, based on the results from Chapter 2 (P values from regression models were close to the alpha level). The length of time that the female was alive on the spawning grounds was the first social success index used (termed longevity). The second mating success index for females was a score based on the consort score and social status of the males with which she consorted. This measure was called 57 the weighted consort score. The third index was a territoriality index, based on aggressive counts toward and away from the focal female. This measure was used based on previous work on sockeye salmon suggesting that body size influences intrasexual competition in females (Van den Berghe and Gross 1989; Foote 1990). This index was called territoriality. Statistical Analyses One male and four females were not used in the study (see Chapter 2, page 22 for details). A l l statistical tests were conducted in SAS (version 8.2). M y criterion for the existence of a trade-off was the presence of a significant negative correlation between pairs of morphological variables (body length, snout length, body depth, and caudal peduncle depth), between morphological variables and energy content, and between longevity and all other traits. To examine if any differences exist between males and females in these correlations, I compared the sex specific correlation coefficients using the Fisher z transformation (Zar 1999). To determine if variation in shape as a secondary sexual characteristic, independent of length, was associated with social mating success, I calculated the residuals of the body length-snout length, body length-body depth, and body length-caudal peduncle depth relationships. These three new variables (i.e. residuals) were used in the remainder of the analyses. I refer to these variables as their original terms, snout length, body depth, and caudal peduncle depth. To examine the extent to which social mating success in males was associated with phenotypic traits, dominance was regressed onto the variables: longevity, body length, snout length, body depth and pre-spawning energy content. Similarly, to examine the extent to which consort score 58 and the proportion of social mates acquired could be predicted from phenotypic traits and longevity, a multiple regression analysis was performed for each, using the same variables as above. To determine what characteristics of females most influenced their social mating success, I regressed the three measures of female social mating success (longevity, weighted consort score, and territoriality) onto the variables body length, snout length, body depth, caudal peduncle depth, and pre-spawning energy content. Intercorrelations of morphological traits, energy content and longevity, and regression analyses predicting social mating success measures, were adjusted for multiple comparisons using sequential Bonferroni correction tests (as described in Chapter 2). Results Characteristics of adult sockeye salmon Males ranged in body length from 623 to 718 mm and in pre-spawning energy content from 15% to 22% (Table 3.1). Snout length, body depth, and caudal peduncle depth were significantly correlated with body length in males (n = 27, r = 0.60, P = 0.0009; n = 27, r = 0.72, P < 0.0001; n = 21, x- 0.66, P = 0.0002, respectively). Females ranged in body length from 584 to 646 mm and in pre-spawning energy content from 18% to 22% (Table 3.1). Females were generally smaller than males, less variable in size, and dramatically different in snout length and body depth. Although females did not have highly developed secondary sexual characteristics, the snout length, body depth, and caudal peduncle depth were significantly correlated with body length (n = 20, r = 0.47, P - 0.0385; n = 20, r = 0.50, P = 0.0250; n = 20, r = 0.82, P < 0.0001, respectively). In general, longer fish of both sexes had larger snouts, deeper humps and caudal 59 Table 3.1 Mean values, range and coefficients of variation (C.V.) of the phenotypic traits and energetic content of adult male and female sockeye. Traits Mean ± 1 SD Range C.V. Males Body length (mm) 673.7 ± 25.0 623-718 3.71 Snout length (mm) 96.7 ± 5.3 87-110 5.48 Body depth (mm) 183.0 ± 13.4 148-209 7.32 Energy content (%) 18.2 ±1 .7 15.1-21.6 9.34 Caudal peduncle depth (mm) 57.7 ± 2.7 53-62 4.68 Females Body length (mm) 612.4 ±20.3 584-646 3.31 Snout length (mm) 60.2 ±3 .8 55-66 6.31 Body depth (mm) 152.7 ±8 .4 137-165 5.50 Energy content (%) 19.9 ±1 .0 18.1-21.7 5.03 Caudal peduncle depth (mm) 54.7 ± 2.3 51-60 4.20 60 peduncles, as was expected. Bigger fish have longer mouths and are deeper in the body irrespective of secondary sex characteristic development. Male allocation trade-offs and social mating success In males, all morphological variables were negatively correlated with pre-spawning energy content, but only snout length was statistically significant after the Bonferroni correction (n = 27, r = -0.50, P = 0.0086, Fig 3.1). Longevity was not significantly correlated with any of the morphological traits (body depth versus longevity: r = -0.36, P = 0.0666 had the strongest association). Pre-spawning energy content and longevity, however, were positively correlated (n = 27, r = 0.46, P = 0.0167) (Fig 3.1). Table 3.2 summarizes the correlation results among all measured variables within males and females. Correlations among independent variables make regression statistics difficult to interpret, so I reduced the variables used in the regression model to three independent ones, snout length (a shape measure), body length (a size measure) and longevity. Both body depth and pre-spawning energy content were significantly correlated with snout length (n = 27, r = 0.41, P = 0.0334; n = 27, r = -0.46, P = 0.0169, respectively) and with longevity (r = -0.38, P = 0.05; r = 0.46, P = 0.0167, respectively). Snout length, and not body depth, was retained in the analyses since snout length increases in size almost 2.5 times more than body depth does between freshwater entry and the beginning of spawning, suggesting that snout length may be experiencing stronger directional selection (Hendry and Berg 1999). A l l three variables contributed significantly to dominance (R2 = 0.38, df= 23, P = 0.0099), with longevity being the most important factor (F = 8.97, P = 0.0065). The multiple regression model for consort score was also significant (R2 = 0.31, df = 23, P = 0.0352). Snout length (F = 61 80 85 90 95 100 105 110 115 Snout length (mm) O 13 -I 1 1 1 1 1 1 1 1 140 150 160 170 180 190 200 210 220 Body depth (mm) 13 14 15 16 17 18 19 20 21 22 23 Pre-spawning energy content (%) Figure 3.1 Relationship between morphological traits and pre-spawning energy-content (%) and between pre-spawning energy content and longevity (days) in male sockeye salmon. 62 CO CD co CD > C o co o <u fc! o o CD 43 » co CD N •g I co 13 43 60 •a a , o —^» CD H CO CD > CD CD a CD o 03 CO CD 13 > CD 43 H CO e > oc a o o u u o> B H a s •o o. •o B U a •a >, o 03 «—i ^ d © oo m CN CN d d Si ©. M ° d o V •si- vo o o d d oo O O OS © •sT © d o d <=> 2 8 8 § d ° 0 0 ro O d d Os OS o oo oo © O 3 © d <=> oo r--r~ CN o t--d o CN £ 1 d o II d o d d ° d <= £ oo oo cN 2 m . d ° d © d © s 3 d ° CO CD > CD i-=3 co 03 CD 43 a S o a d ° d ° d ° I D 0 0 m 0 0 o in o o 0 0 T—i so '—1 o CN d d d d d ca cS x •a 13 S S3 O 'M 13 o U CN — s a H 43 a >-> o PQ d ° CTs ( N C ^ O ^ d 0 0 ^ CN § CN g 0 0 H d v 43 a — •a o 03 c o a 0/3 a. •a >•> •d o pa w a s a. s es U d ° a o a o u oc a 11 CN . d ° v OC 8 O 5.37, P = 0.0298), and longevity (F = 5.36, P = 0.0299) contributed almost equally to the regression. The model for the proportion of social mates acquired was significant (R = 0.33, df= 23, P = 0.0250). Body length was the single statistically significant contributor (F = 7.90, P = 0.0099). Table 3.3 summarizes the results from this section. Female allocation trade-offs and social mating success Similar to what I observed for males, longevity and energy content were significantly correlated in female salmon (n = 20, r = 0.54, P = 0.0136) (Fig 3.2); females with a higher energy content tended to live longer on the spawning grounds than fish with a lower energy content. There was a significant correlation between caudal peduncle depth and energy content (n = 20, r = 0.50, P = 0.0235) (Fig 3.2). Caudal peduncle depth was the only morphological variable correlated with energy content in females. The next strongest relationship was between body length and energy content (n = 20, r = 0.38, P = 0.0965), suggesting that bigger females tended to have a higher percentage of pre-spawning energy content. A significant difference existed in the correlation coefficient of body length and energy content between males and females (Z= 1.97, P < 0.05), where the former has a negative association and the latter, a positive one. Other correlations did not differ between males and females. Body depth was removed from the multiple regression analysis of female social mating success. Body depth is not hypothesized to affect female breeding success. Deeper females may look more 'threatening' or resist charges by other females more effectively, but unlike snout length, body depth does not increase in size during trait development in females and thus, it may not be sexually selected (Hendry and Berg 1999). Pre spawning energy content, body length, snout length, and caudal peduncle depth were retained for multiple regression analyses. 64 Table 3.3 Summary of the multiple regression analyses on male social mating success. R 2 is the coefficient of determination, and P is the regression coefficient. Traits R 2 3 P-value Dependent variable: Dominance Longevity (days) 0.4010 0.0065 Body length (mm) 0.0246 0.0741 Snout length 0.1747 0.0358 Model 0.3841 0.0099 Dependent variable: Consort score Longevity (days) 0.3191 0.0299 Body length (mm) 0.0169 0.2236 Snout length 0.1868 0.0298 Model 0.3065 0.0352 Dependent variable: # social mates Longevity (days) 0.1141 0.0991 Body length (mm) 0.0183 0.0099 Snout length 0.0576 0.1514 Model 0.3285 0.0250 65 Figure 3.2 Relationship between caudal peduncle depth (mm) and pre-spawning energy content (%) and between pre-spawning energy content and longevity (days) in female sockeye salmon. 66 The multiple regression analysis indicated that longevity was associated with the phenotypic traits of females on the spawning grounds (R2 = 0.47, df = 15, P = 0.0392). However, only pre-spawning energy content was a significant indicator (F = 6.12, P = 0.0258). The model for predicting weighted consort score in females was similar to the longevity model above (R -0.50, df = 15, P = 0.0263). However, none of the variables was statistically significant in the model. The simple regression of weighted consort score on pre-spawning energy content was highly significant (r2 = 0.35, df = 18, P = 0.0063). The multiple regression of territoriality in females was non- significant (Rz = 0.04, df = 15, P = 0.9619). Table 3.4 summarizes the results for the female social mating success regressions. Discussion Organisms are faced with trade-offs throughout their existence, a cost paid in fitness when a beneficial change in one character is linked to a greater detrimental change in another. The trade-offs involved during reproduction are the most prominent, where costs are often paid in terms of survival and breeding success (Stearns 1989). In Pacific salmon, this trade-off is so severe that the species die after a single bout of reproduction. Investment in reproductive traits are further constrained by the fact that the fish do not feed after returning to freshwater and must complete their migration and spawn with the energy reserves brought from the ocean. Increased energy use during migration can only be accomplished at the expense of energy available for egg production, nest construction and nest defence in females and secondary sexual development and competition for mates in males (Hendry and Berg 1999). Thus, the fish face several trade-offs in energy allocation for reproduction. 67 Table 3.4 Summary of the multiple regression analyses on female social mating success. Traits R 2 3 P-value Dependent variable: Longevity Energy content Body length Snout length Caudal Peduncle Model 0.9677 0.0258 0.0082 0.6616 -0.1910 0.0834 -0.3067 0.2993 0.4690 .... 0.0392 Dependent variable: Weighted Consort score Energy content .... 0.5357 0.1176 Body length .... 0.0275 0.0899 Snout length .... -0.0362 0.6764 Caudal Peduncle .... 0.3187 0.1958 Model 0.4999 .... 0.0263 Dependent variable: Territoriality Energy content .... -0.2578 0.5732 Body length .... -0.0041 0.8486 Snout length .... -0.0266 0.8247 Caudal Peduncle .... 0.0903 0.7858 Model 0.0375 .... 0.9619 68 Male allocation trade-offs and social mating success Among males, all morphological measurements were strongly correlated; bigger males generally had longer snouts and deeper bodies. From freshwater entry to the spawning grounds, male body depth has been shown to increase by 19% and snout length by 47.7% in short distance migrating sockeye salmon, suggesting that these traits are sexually selected (Hendry and Berg 1999). The first trade-off I considered was between the exaggeration of secondary sexual traits and other energetic demands such as maintenance metabolism in freshwater during spawning. Since these extravagant characteristics cost energy to construct, it seems reasonable to assert that sexual selection is favouring the development of secondary sexual traits, whether it is through female choice or intrasexual competition among males (Quinn and Foote 1994; Fleming and Gross 1994; Quinn et al. 2001). The expected negative correlation between these variables and energy content was found. Hendry et al. (2000), however, did not find this correlation. However, their fish were held in freshwater for their entire existence, fed unnatural food throughout, and not allowed to migrate or spawn. Trade-offs may not be evident in circumstances such as these since the fish do not experience the upstream migration, where most of their energy may be utilized. The trade-off suggests that males entering the spawning grounds with a low percentage of energy content were bigger in overall size. This seems to contradict the work of Crossin (2002) who found that energy density was highest in the smallest bodied salmon stocks. Crossin's work, however, was based on river entry energy density, and not spawning channel entry, which is the focus in my study. Fish arriving at the mouth of the Fraser River may not show a trade-off between energy content and body size. However, since bigger male fish have more developed secondary sexual traits, they utilize more energy (or utilize energy less efficiently) for body development and upstream swimming during migration. Their laterally 69 compressed body shape makes them less maneuverable in small streams and in fast water (Quinn et al. 2001) and may, thus, result in less relative energy reserves once on the spawning grounds. While some trade-offs can have immediate consequences, in many cases the relative costs and benefits of alternative strategies or tactics may only become evident in the long-term, both from the acquisition and the allocation of resources (Morgan et al. 2002). Species with variable life history patterns may possess more than one favourable pattern of resource acquisition and allocation. Fish arriving at the mouth of the Fraser River may be in similar condition. The trade-off between size and energy that I found could be based on different allocation schedules, where some males expend more energy in trait development while others do not. I propose the existence of two life history tactics within male sockeye salmon. Fish that are bigger in size due to their genetics, or an interaction of genes and the environment, may have a different resource allocation priority. In sockeye salmon, fish that are initially bigger can develop exaggerated secondary sexual traits more readily because of their already larger size. Smaller fish do not have this advantage. They can develop their secondary sexual characteristics to a certain degree, within physiological constraints, but they will never outgrow their larger conspecifics. But since energy is required for other activities (such as migration and longevity on the spawning ground) I propose that these smaller fish focus more on surviving the migration and reaching the spawning sites in order to maximize their genetic fitness. They cannot develop huge traits because of their relatively smaller size, but they can save that energy and live a longer life on the spawning grounds, outliving the bigger males. Energy allocation schedules to maximize reproductive success, thus, may differ between large and small males, with large males spending more on secondary sexual characters and small males more on the length of life. 70 The social mating success of male sockeye salmon was found to be principally a function of three components: snout length, body length, and longevity on the spawning grounds, in agreement with speculations made in previous studies (Quinn and Foote 1994; Fleming and Gross 1994; Quinn et al. 2001). Large body size facilitates male dominance, increasing both access to spawning females and winning male-male competitions. Male intrasexual competition is the most likely evolutionary explanation for the extravagant traits in salmon (Darwin 1871; Fleming and Gross 1994; Quinn and Foote 1994), reinforced by the fact that females select their mates based on the outcome of these competitions. Aggressive males chase, charge and bite each other with the aid of their jaws and recurved teeth. The exaggeration of the snout increases the effectiveness of this weapon during competition (suggested by Quinn and Foote 1994). The smallest individuals become excluded from mating opportunities in this way. Longevity was also an important determinant of male social mating success. The longer the male lives on the spawning grounds, the greater his chances are at obtaining fertilizations. The importance of longevity may be exaggerated in my study design because the mating assemblage is not open to new males. In a natural spawning environment, newly arrived large males may simply evict any long living small males. However, smaller males also tend to arrive at the spawning grounds later on average than large males, which may give them an advantage with any late spawning females. It seems reasonable to conclude that smaller males have indeed adopted a different (and successful) reproductive tactic that includes a suite of behavioural and morphological variables. The observation that higher energy content correlates with a longer reproductive lifespan is itself an important piece to the puzzle of why all these fish do not increase in size to the limits imposed by their physiological architecture, and instead invest less energy in trait development and more in energy stores on the spawning grounds. 8 71 Female allocation trade-offs and social mating success As with the males, female sockeye salmon showed strong relationships among morphological variables. Females were generally smaller than males, less variable in size, and had much smaller snout lengths and body depths (Table 3.1). Increased body size is important for both a higher total egg production and for a higher quality nest site (Fleming et al. 1997; Fleming 1998; Quinn et al. 1995; Steen and Quinn 1999; Hendry et al. 2001). Larger females tend to have deeper caudal peduncles, which is used for digging during nest construction and appears to influence the depth of egg burial (Fleming and Gross 1994), potentially reducing the vulnerability of nests to destruction (Fleming 1998). They have also been found to dominate smaller females in competition for preferred sites (Foote 1990). Larger females tend to have a larger proportional increase in egg number than in egg size (Wood and Foote 1996; Crossin 2002; Hendry et al. 2001; Kinnison et al. 2001), in agreement with reproductive investment theory, which predicts a trade-off between fecundity and egg size (Smith and Fretwell 1974; Jonsson et al. 1996). In the present study, I found no significant correlations between pre-spawning energy content and body size or shape, similar to findings by Adams et al. (1995) and Hendry et al. (2000). I did, however, find a significant positive correlation between caudal peduncle depth and pre-spawning energy content. In contrast to males, all the associations were positive in direction, indicating the absence of reproductive trade-offs at maturity in females. A comparison of the correlation coefficients indicated that body length versus energy content differed significantly between the sexes (P < 0.05), but body shape versus energy content did not. In fact, a picture very different from that seen in males emerges with respect to the predicted trade-off between body size and pre-spawning energy reserves in females. Large amounts of energy are allocated to egg production in females (Ballantyne et al. 1996; Hendry et al. 2000). However, contrary to 72 expectation, larger females, which are expected to be more fecund had higher, not lower, energy content. Given that studies on female salmonids suggest that large females put relatively more energy into gonads, evidence of a trade-off between these variables would be expected. Several hypotheses may be proposed to explain this negative result. One possible reason for the failure to detect a trade-off between size and energy content is that individuals may vary in the amount of energy they acquire at sea, and this variation can create correlated variation in energy allocation (Hendry et al. 2000). High-energy individuals may be more able to invest into multiple traits than can low-energy individuals, and thus obscure the expected trade-off (de Jong and van Noordwijk 1992; Hendry et al. 2000). This may be a result of genetics or an interaction between genetics and the environment. Furthermore, it has been suggested that energy reserves are depleted less rapidly in larger fish than in smaller fish due to their specific metabolic rate being inversely proportional to body size. For the same distance and time, the relative swimming speed of larger fish is substantially less (Weatherley and Gil l 1995). Therefore, a positive association between body size and energy content would be expected. A third, more interesting, explanation is the idea that animals adjust their acquisition and allocation of food in relation to their body size, food availability, or predation risk (Holtby and Healey 1990). Females may be more constrained to a minimum body mass for successful reproduction than males, where being too small or too big may both have negative effects on their fitness. Indeed, the smaller variation in female size provides evidence of stronger stabilizing selection compared to the larger variation seen in males (see Table 3.1). If food is limiting, a female may maximize her fitness by investing any food she acquires into body mass rather than energy stores for future use, so that she can grow to the minimum body size required for successful breeding. However, if a female is able to find an abundance of food, and has, 73 therefore, achieved the necessary (optimal) body size, she would benefit from storing the surplus energy into fat reserves rather than her body size. This is because by achieving a body size greater than the optimal, she may be reducing her overall fitness (e.g., a large body size may restrict access to home streams). Thus, the positive relationship between body size and energy reserves found in my study could be the consequence of fish adjusting their allocation of energy in relation to the biotic and abiotic constraints imposed on them (Holtby and Healey 1990). The same arguments can be made for males. However, the difference may lie in the way the energy is used from river entry to spawning ground entry, where larger males use more energy for the development of their secondary sexual traits and end up with relatively less upon arrival on the spawning ground, compared to females. The social mating success of females indicated that pre-spawning energy was the single, most important factor influencing the success of females (measured as longevity on the spawning grounds). Females with higher energy reserves also tend to have a longer reproductive life span (r = 0.54, P < 0.05), which can again be attributed to energy-rich and energy-poor individuals. The former invest more energy into multiple traits (e.g. caudal peduncle depth) as a result of early life and ocean life development, and therefore, have a higher fitness. The latter invest much less into various traits because of a lack of food sources (or genetics) prior to spawning. Energy-poor individuals may face more pronounced trade-offs and at the end have a lower genetic fitness. It is important to note that these energy-poor individuals have reached the minimum body size to successfully migrate, compete and spawn on the breeding grounds, but are energy-poor only relative to the biggest and toughest females with whom they are in competition. 74 In contrast to other studies (Van den Berghe and Gross 1989, Fleming and Gross 1989; Foote 1990), except for Schroder's study on chum salmon, O. keta (1982), I found no association between female morphology and territoriality. However, unlike the previous two measures of female mating success (longevity and weighted consort score), in this model both snout length and caudal peduncle depth had a positive (but weak) effect on territoriality. The failure to find a statistically significant effect could be masked by the enclosure design in my study. The females had ample space for both the construction and the guarding of their nests. Thus, each female had the opportunity to build a nest without encroachment by neighbouring or newly arriving females. Unlike natural streams where new females continually enter and attempt to take over a nesting site due to limited resources, these females did not need to defend their nest site from new females entering the system as the number of fish did not increase (and often decreased due to pre-spawning mortality). This may have contributed to the fact that no association between body size and territoriality was found. Literature Cited: Adams, C. E., F. A . Huntingford and M . Jobling. 1995. A non-destructive morphometric technique for estimation of body and mesenteric lipid in Arctic charr: a case study of its application. Journal of Fish Biology 47:82-90. Ballantyne, J. S., F. Mercure, M . F. Gerrits, G. Van Der Kraak, S. McKinley, D. W. Martens, S. G. Hinch and R. E. Diewert. 1996. Plasma nonesterified fatty acid profiles in male and female sockeye salmon, Oncorhynchus nerka, during the spawning migration. Canadian Journal of Fisheries and Aquatic Sciences 53: 1418-1426 Brett, J. R. 1995. Energetics. In Physiological ecology of Pacific salmon. (Groot, C , Margolis, L . and Clarke, W. C , eds), pp. 1-68. Vancouver: U B C Press. Chebanov, N . A . , N . V . Varnavskaya and V . S. Varnevskii. 1983. Assessment of successful participation in spawning of male sockeye salmon Oncorhynchus nerka (Walbaum) (Salmonidae) of different hierarchical status with the help of genetic-biochemical markers. Voprosy Ikhtiologii 23:774-778. Crossin, G. 2002. Selection for a bioenergetic-morphological feedback mechanism in migrating adult salmon is related to migratory distance and elevation. MSc thesis, University of 75 British Columbia, Vancouver. Darwin, C. 1871. The descent of man, and selection in relation to sex. J. Murray, London. de Jong, G. and A. J. van Noordwijk. 1992. Acquisition and allocation of resources: genetic (co)variances, selection, and life histories. American Naturalist 139:749-770. Fleming, I. A . 1998. Pattern and variability in the breeding system of Atlantic salmon (Salmo salar), with comparisons to other salmonids. Canadian Journal of Fisheries and Aquatic Sciences 55:59-76. Fleming, I. A . and M . R. Gross. 1989. Evolution of adult female life history and morphology in a Pacific salmon (coho: Oncorhynchus kisutch). Evolution 43:141-157. Fleming, I. A . and M . R. Gross. 1994. Breeding competition in a Pacific salmon (coho: Oncorhynchus kisutch): measures of natural and sexual selection. Evolution 48:637-657. Fleming, I. A . , A . Lamberg and B. Jonsson. 1997. Effects of early experience on the reproductive performance of Atlantic salmon. Behavioral Ecology 8:470-480. Foote, C. J. 1990. A n experimental comparison of male and female spawning territoriality in a Pacific salmon. Behaviour 115:283-314. Gilhousen, P. 1980. Energy sources and expenditures in Fraser River sockeye salmon during their spawning migration. International Pacific Salmon Fisheries Commission Bulletin 22:1-51. Gross, M . R. 1984. Sunfish, salmon, and the evolution of alternative reproductive strategies and tactics in fishes. In Fish reproductions: strategies and tactics. (Wootton, R. and Potts, G., eds), pp.55-75. London: Academic Press. Gross, M . R. 1985. Disruptive selection for alternative life history strategies in salmon. Nature 313:47-48. Healey, M . C. 2001. Patterns of gametic investment by female stream- and ocean-type chinook salmon. Journal of Fish Biology 58:1545-1556. Hendry, A . P. and O. K. Berg. 1999. Secondary sexual characters, energy use, senescence, and the cost of reproduction in sockeye salmon. Canadian Journal of Zoology 77:1663-1675. Hendry, A . P., O. K . Berg and T. P. Quinn. 1999. Condition dependence and adaptation-by-time: breeding date, life history, and energy allocation within a population of salmon. Oikos 85:499-514. Hendry, A . P., A . H . Dittman and R. W. Hardy. 2000. Proximate composition, reproductive development, and a test for trade-offs in captive sockeye salmon. Transactions of the American Fisheries Society 129:1082-1095. Hendry, A . P., T. Day and A. B. Cooper. 2001. Optimal size and number of propagules: 76 Allowance for discrete stages and effects of maternal size on reproductive output and offspring fitness. American Naturalist 157:387-407. Holtby, L . B. and M . C. Healey. 1990. Sex-specific life history tactics and risk-taking in coho salmon. Ecology 71:678-690. Jonsson, N . , B. Jonsson and I. A . Fleming. 1996. Does early growth cause a phenotypically plastic response in egg production of Atlantic salmon? Functional Ecology 10:89-96. Kinnison, M . T., M . J. Unwin, A . P. Hendry and T. P. Quinn. 2001. Migratory costs and the evolution of egg size and number in introduced and indigenous salmon populations. Evolution 55:1656-1667. Morgan, I. J., I. D. McCarthy and N . B. Metcalfe. 2002. The influence of life-history strategy on lipid metabolism in overwintering juvenile Atlantic salmon. Journal of Fish Biology 60:674-686. Neat, F. C , A . C. Taylor and F. A. Huntingford. 1998. Proximate costs of fighting in male cichlid fish: the role of injuries and energy metabolism. Animal Behaviour 55:875-882. Quinn, T. P. and C. J. Foote. 1994. The effects of body size and sexual dimorphism on the reproductive behaviour of sockeye salmon, Oncorhynchus nerka. Animal Behaviour 48:751-761. Quinn, T. P., A . P. Hendry and L. A . Wetzel. 1995. The influences of life history trade-offs and the size of incubation gravels on egg size variation in sockeye salmon (Oncorhynchus nerka). Oikos 74:425-438. Quinn, T. P., M . D. Adkison and M . B. Ward. 1996. Behavioural tactics of male sockeye salmon (Oncorhynchus nerka) under varying operational sex ratios. Ethology 102:304-322. Quinn, T. P. and M . V. McPhee. 1998. Effects of senescence and density on the aggression of adult female sockeye salmon. Journal of Fish Biology 52:1295-1300. Quinn, T. P., L . A . Wetzel, S. Bishop, K. Overberg and D. E. Rogers. 2001. Influence of breeding habitat on bear predation and age at maturity and sexual dimorphism of sockeye salmon populations. Canadian Journal of Zoology 79:1782-1793. Roff, D. A . 1992. The evolution of life histories: theory and analysis. Chapman and Hall. New York. SAS 2001. SAS/STAT Windows version 8.2. SAS Institute, North Carolina, USA. Schroder, S. L . 1982. The influence of intrasexual competition on the distribution of chum salmon in an experimental stream. In Proceedings of the salmon and trout migratory symposium (Salo, E. O., eds), pp.275-285. Seattle: University of Washington Press. Smith, C. C. and S. D. Fretwell. 1974. The optimal balance between size and number of offspring. American Naturalist 108:499-505. 77 Stearns, S. C. 1977. The evolution of life history traits: a critique of the theory and a review of the data. Annual Review of Ecology and Systematics 8:145-171. Stearns, S. C. 1989. Trade-offs in life-history evolution. Functional Ecology 3:259-268. Stearns, S. C. 1992. The Evolution of Life Histories. Oxford University Press. Oxford. Steen, R. P. and T. P. Quinn. 1999. Egg burial depth by sockeye salmon (Oncorhynchus nerka): implications for survival of embryos and natural selection on female body size. Canadian Journal of Zoology 77:836-841. Van den Berghe, E. P. and M . R. Gross. 1989. Natural selection resulting from female breeding competition in a Pacific salmon (coho: Oncorhynchus kisutch). Evolution 43:125-140. Weatherley, A . H . and H . S. Gi l l . 1995. Growth. In Physiological ecology of Pacific salmon. (Groot, C., Margolis, L . and Clarke, W. C , eds), pp.101-158. Vancouver: U B C Press. Wood, C. C , and C. J. Foote. 1996. Evidence for sympatric genetic divergence of anadromous and nonanadromous morphs of sockeye salmon (Oncorhynchus nerka). Evolution 50:1265-1279. Wootton, R. 1990. Ecology of teleost fishes. Chapman and Hall. New York. Zar, J. H . 1999. Biostatistical analysis (4th edition). Prentice Hall, Inc. Upper Saddle River, New Jersey. 78 Chapter 4- The importance of direct and indirect components of sexual selection Introduction Pacific salmon display complex patterns of reproduction involving a suite of morphological and behavioural characteristics. The species are sexually dimorphic and display sex specific breeding behaviour. Within sexes, certain traits tend to be correlated, whether it is among morphological traits, or between morphological and behavioural traits. This has been best described for males in which large males develop large dorsal humps, hooked jaws and characteristic colouration and fight for access to females whereas small males do not develop this extreme morphology, adopt cryptic or female colouration and access females by sneaking (Schroder 1982; Gross 1985; Healey and Prince 1998). It is widely accepted that these characteristics are correlated with fitness but there have been few studies to demonstrate this association (but see Holtby and Healey 1986, Van den Bergh and Gross 1989). Furthermore, the causal relationships, i f any, among these traits have never been worked out. Ecologists have long recognized that morphological traits tend to co-occur in correlated sets. The same is true of behavioural traits. Darwin (1872) referred to this as the "correlated variation" of attributes. He argued that i f humans were continually to select, and thus increase, any particular characteristic of some life form, they would most certainly and inevitably modify other parts of the structure unintentionally, due to the existence of the laws of correlation. Observations of such correlations in organismal biology have been studied and dissected for a long time. So, when Georges Cuvier, the great 19th century comparative biologist, was given a fossil encased in limestone, he was able to look at only the teeth and predict the structure of the pelvis (Shipley 1999). Such patterns of association underlie the causal explanations proposed by many in the fields of psychology, sociology, biology and other areas of research. 79 Correlation analysis is perhaps the most widely used means to explore life history variation and trade-offs in biological systems. Unfortunately, much less attention has been paid to testing hypothesized causal relationships drawn from such correlations (Hilborn and Mangel 1997; Shipley 1999). The correlations alone provide very weak evidence for the presumed causal relationships because the underlying structure and dynamics of the system are so complex (Smith et al. 1997). Finding rigorous ways to test these causal relationships is a major challenge in organismal biology. If one were to infer the interaction of dependent and independent variables from prior research or knowledge (e.g., the correlations described above), then one can potentially draw one or more mathematical or predictive models of the system. The reliability of each model can then be tested against the alternatives, given the available data (Hilborn and Mangel 1997; Shipley 1999). One way of testing such a priori causal relationships is through structural equation modeling. Almost a century ago, Sewall Wright (1920) examined the patterns of covariation between various traits of guinea pigs. From his selection experiments, he proposed a set of direct and indirect causal links among the traits. Since its inception, this form of path analysis has relied on dependency associations, in which the investigator specifies an ordered sequence of variables (A to B to C), and then partitions the correlation based on those pathways. Although Wright's method could quantify the linkages in a hypothesized causal pathway, it did not provide an overall test of model fit to the data. This limitation has been overcome in structural equation modeling (Grace and Pugesek 1998, Shipley 1999). 80 At present, path analysis (or structural equation modeling, the more general area of linear modeling techniques) is used in many research areas, from the social sciences to chemistry and biology. Although ecologists have been slow to adopt path analysis, and structural equation modeling in general, the use of it has increased substantially in recent years (Kingslover and Schemske 1991; Mitchell 1992, 1993, 1994; Wootton 1994a, 1994b; Shipley 1997; Grace and Pugesek 1998; Johnson 2002), although some authors are still critical of the method and its applications (Smith et a. 1997; Petraitis et al. 1996). Path analysis provides precisely the technique needed to explore salmon breeding systems as it allows one to partition the simple correlations among a set of variables according to a particular working model and their causal relationships (A causes B, B in turn causes Q . In this chapter, I construct a set of models using path analysis to explore life history characteristics and trade-offs in sockeye salmon (Oncorhynchus nerka). Previous work on salmonids has shown that greater size and exaggerated secondary sexual traits are positively correlated with variability in reproductive success (Schroder 1981, 1982; Gross 1985; Holtby and Healey 1986; Foote 1989, 1990; Fleming and Gross 1989, 1994; Quinn and Foote 1994; Quinn et al. 1995; Healey and Prince 1998). No study, however, has yet quantified the direct and indirect contribution of various reproductive traits to the genetic reproductive success in male and female salmon. I use structural equation modeling to identify specific factors that have shaped such dimorphism and variability in reproduction. Methods A total of 27 males and 20 females were used in this study. A detailed summary of the study design can be found in Chapter two (pages 10-21). Male and female fitness measures were used in a path analysis to estimate the relative importance of phenotypic and behavioural traits in 81 determining male genetic reproductive success and female mate choice. Path analysis starts with a defined structure of hypothesized causal relationships (paths), often represented as causal diagrams, and then uses the covariance among all the variables to estimate the contribution of each path to a trait (Petraitis et al. 1996). The total effect of an independent variable on a dependent variable is quantified by the sum of its direct path coefficients and all the potential indirect path coefficients (see Kingsolver and Schemske 1991; Mitchell 1992; Shipley 1997, 1999, for a fuller description of the methodology). Prior to a goodness of fit test, there may be no strong reason to be dissatisfied with the model of interest, particularly when some R2 values are large, and several path coefficients within the model are statistically significant. Measures of multiple determination (R2) indicate the proportion of observed variance explained by each equation (Mitchell 1992). Thus, each dependent variable has an associated R2 value, which tests the adequacy of the equation (or independent variables) in predicting values of the dependent variable, but does not address the model as a whole (Biddle and Marlin 1987). In other words, like a multiple regression model, a significant R indicates that a significant proportion of the variance in the dependent variable is accounted for by the independent variables within the equation. Path coefficients correspond to the standardized partial regression coefficients of multiple regression, and so they represent the effect of a variable when all other variables are held constant (Li 1975). Each path coefficient is tested for significance against the null hypothesis that it is equal to zero (Mitchell 1992). This is the end point for multiple regression analysis, but is only part of the analysis when using path analysis. Structural Equation Modeling (SEM), an extension of path analysis, tests whether a specific model, represented as a path diagram, is an acceptable description of the underlying causal 82 mechanisms or, in other words, the covariation among the traits within the model (Gomez and Zamora 2000). This is achieved by a goodness-of-fit test, in which the observed covariance matrix is compared with that expected if the model was true (Mitchell 1992, 1993). When structural equation models are tested, the null model that one is trying to reject is often the model that the researcher believes to be the correct one (Shipley 1997). Thus, a significant goodness-of-fit test indicates that the model is a poor description of the observed covariance among the variables, while failure to reject a model (a non-significant goodness-of-fit test) is often interpreted as evidence in favour of the model (i.e. the pattern of covariance predicted by the model is not distinguishable from that observed) (Mitchell 1994; Shipley 1997). As with any model testing, an insignificant goodness-of-fit test does not prove that the right model has been found. It merely indicates that the model and the set of coefficient estimates are consistent with the observed covariances. One way to represent causal mechanisms of selection is to incorporate in the path diagram intermediate variables that relate phenotypic traits to fitness (X-+R-+ W). Here, selection on some trait X results from the effect o f X o n some intermediate variable R that in turn affects fitness ^(Kingsolver and Schemske 1991). By design, not all combinations of factors, or all possible interactions between factors, are evaluated. I focused on those paths and combinations of paths that represent plausible hypotheses from life history theory (Roff 1992 provides a review). I use the S E M program LISREL (Joreskog and Sorbom 2002), which solves for path coefficients using maximum likelihood, least-squares, or other methods. The models are solved by a set of constraints representing the path diagram, and are based on the observed covariance matrix of the measured variables. 83 Path analytic models for male fitness I hypothesized about the causal links among seven reproductive traits; body length, secondary sexual trait development (snout length is used as a surrogate for secondary sexual trait development), body energy content, longevity on the spawning grounds, dominance, consort score, and their effects on genetic fitness (proportion of eggs fertilized and proportion of mates acquired) of the males, building an a priori basic path-analysis model. This model expresses the causal paths that relate male traits indirectly to male fitness, via their effects on the intermediate variables, dominance and consort score. Figure 4.1 is a representation of the first basic path model, illustrating the causal linkages of the male breeding system. The methodology used in measuring the above mentioned variables are found in Chapter two (pages 10-21). Consort score is a measure of the percent of time a male spends as a consort to a receptive female. During behavioural observations, I noted that females had at least partial mate choice, chasing away less dominant, smaller or inferior males. Indeed, other studies have shown female mate choice (Foote 1988). Chum salmon females (O. keta) spawn more rapidly when accompanied by larger males, delaying spawning when attended by smaller males (Schroder 1981), as do female kokanee (Foote 1988, 1989), and Atlantic salmon (Jarvi 1990). Thus, in this study I have used consort score as an indicator of female choice. The proportion of mates acquired and the proportion of offspring sired were treated as response variables (fitness components). Dominance and consort score were treated as intermediate variables directly related to fitness (collectively called social mating success measures). I also hypothesized that dominance will have an effect on consort score, which relates the aggression and displaying of males to the consorting behaviour shown toward females. In this way, intrasexual competition is related to intersexual competition (or female mate choice). As independent variables affecting dominance and consort score, I used three phenotypic traits: body length, snout length (body length-84 in 0 0 corrected) and pre-spawning energy content, and reproductive life span (termed longevity). These four variables will be collectively called life history variables. I build three basic models and considered two nested models for each basic one (Table 4.1). The main difference between each of the three basic models (and their nested alternatives) was the life history variables considered. The first model considered body length, snout length and longevity, the second model considered snout length, pre-spawning energy content and longevity, and the third model considered only snout length and longevity as independent variables related to the social mating success measures. LISREL computes an estimate of the unknown regression coefficient of interest, and also the associated standard error and a t-value (the estimate divided by its standard error), both of which are associated with the path coefficient of interest. If a reported t-value is greater than a certain critical value, the null hypothesis that the associated parameter is equal to 0 is rejected. As a rule of thumb, ^-values greater than 1.96 are taken to indicate statistical significance (Mueller 1996). The nested models were built on the basis of reducing the parameters to be estimated, but still within biological reasoning. A l l paths from the life history variables to consort score were fixed at zero (i.e. the path was eliminated from the basic model) for the second nested models. The paths from social mating success measures (dominance and consort score) to genetic fitness were reduced from four to two paths in the third nested models. The same approach was taken for each of the three basic models in building the two nested models. Table 4.1 describes the paths that were eliminated in each of the alternative models. For each model a maximum likelihood function was used in a goodness of fit test. This likelihood is used to generate a test statistic that is distributed approximately as x • The degrees of freedom (d.f.) for the %2 test are calculated as the difference between the total number of 86 Table 4.1 Description of the models considered for the male breeding system. Each model is described relative to its most basic model, i.e. Nested 2a is nested within Model 2. Model Model description (paths constrained to zero from basic model) Model 1 No constraints Nested l a a. Body length, snout length and longevity to consort score b. Consort score to proportion of mates acquired Nested lb a. Body length, snout length and longevity to consort score b. Consort score to proportion of mates acquired c. Dominance to proportion of offspring sired Model 2 No constraints Nested 2a a. Snout length, pre-spawning energy content and longevity to consort score b. Consort score to proportion of mates acquired Nested 2b a. Snout length, pre-spawning energy content and longevity to consort score b. Consort score to proportion of mates acquired c. Dominance to proportion of offspring sired Model 3 No constraints Nested 3 a a. Snout length and longevity to consort score b. Consort score to proportion of mates acquired Nested 3b a. Snout length and longevity to consort score b. Consort score to proportion of mates acquired c. Dominance to proportion of offspring sired 87 unique entries in the covariance matrix and the total number of coefficients estimated in the model (d.f. = n (n + l)/2 - m, where n is the number of variables in the model and m is the total number of estimated coefficients). The nested model, with its additional restrictions, should have a larger x 2 than the basic model, and the d.f. should be larger than for the basic model, because fewer coefficients are estimated for the more restricted model. The difference between the x,2 values calculated for any two nested models is also distributed as %2 and can be used to test whether a model nested within another model gives as good a fit to the data. The degrees of freedom of this latter x 2 are the change in the degrees of freedom between the two models (Shipley 2000). A significant difference in the goodness-of-fit indicates that the additional constraints have significantly reduced the model's ability to fit the data, or in other words, the more general model (i.e. basic model) is better at describing the causal relationships among the variables. If, on the other hand, the difference in the goodness-of-fit is non-significant, both models are equally appropriate, but the simpler one may be favoured on the principle of parsimony, although it is at the discretion of the researcher (Mitchell 1993; Shipley 2000). For 2 2 models that are not nested, the model with the smaller x is preferred, because a smaller x implies that the predicted covariance more closely matches the observed covariance (Hayduk 1987). Path analytic models for female fitness The causal relationships in the female breeding system included six reproductive traits: body length, body energy content, longevity on the spawning grounds, weighted consort score, and their effects on genetic mate choice (mate quality and mate rank). This model expresses the causal paths that relate female traits indirectly to female mate quality, via their effects on the intermediate variables, longevity and a weighted consort score. Figure 4.2 is a representation of the first basic path model, illustrating the causal linkages of the female breeding system. The 88 methodology used in measuring the variables is found in Chapter two (pages 10-21). The variable "mate quality" is equivalent to Fitness 2, and "mate rank" to Fitness 3 as described in Chapter 2. The two mate choice variables were treated as response variables. Longevity and weighted consort score were treated as intermediate variables directly related to mate choice (collectively called social mating success measures). And finally, variables affecting longevity and weighted consort score were body length and pre-spawning energy content (life history variables). I build two basic models and considered one nested model for each basic one (Table 4.2). The one difference between the two basic models (and their nested alternative) was the life history variables considered. The first model considered both life history variables, whereas the second model considered only pre-spawning energy content. The nested models were built on the basis of reducing the parameters to be estimated, but still within biological reasoning. The paths from social mating success measures (longevity and weighted consort score) to mate choice were reduced from four to two paths in the nested models. The same approach was taken for each of the two basic models in building the nested models. Table 4.2 summarizes the paths that were eliminated in each of the alternative models. Results Path analytic models for male fitness A l l three basic models were non-significant, suggesting that they were sufficient models, as they explained the observed covariance among the traits measured (See Table 4.3 for details). The difference between the x 2 's for each of the nested models and the basic model, and between the two nested models indicated that the additional constraints (in the nested models) improved the 90 Table 4.2 Description of the models considered for the female breeding system. Model Model description (paths constrained to zero from basic model) Model 1 No constraints Nested l a a. weighted consort score to mate quality b. longevity to mate rank Model 2 No constraints Nested 2a a. weighted consort score to mate quality b. longevity to mate rank 91 Table 4.3 Summary of the goodness of fit for the models in the male breeding system. The name, the degrees of freedom (d.f.), the x 2 test statistic, and the P value for each model are shown. Each nested model also includes a x 2 test between the basic model and the nested model in question (Basic-Nested), and between the two nested models (Nested-Nested). Model d.f. P value Basic-Nested Nested-Nested (d.f., X 2 ) (df, X 2 ) Model 1 7 11.45 0.120 Nested l a 10 12.81 0.234 3, 1.359* Nested lb 12 13.91 0.307 5, 2.456** 2, 1.097* Model 2 7 13.13 0.069 Nested 2a 10 14.77 0.141 3, 1.644* Nested 2b 12 15.32 0.224 5,2.193** 2, 0.549** Model 3 5 9.75 0.083 Nested 3 a 7 10.72 0.151 2, 0.977* Nested 3b 9 11.55 0.240 4, 1.800** 2, 0.823* Note: See Table 4.1 for description of the alternative models. *P>0.50 ** P > 0.75 92 model's ability to fit the data. Thus, nested models lb, 2b, and 3b best described the relationships among the traits within the model. Since these three models are not nested, the model with the smaller x 2 is preferred. Model 3b was the simplest, most parsimonious nested model with a non-significant goodness of fit test (x2 = 11.546, d.f. = 9, P = 0.24, LISREL). The path coefficients relating the parameters were significant, and the model did not violate model assumptions (Table 4.4 and Fig 4.3). For this model, path coefficients for snout length to dominance and longevity to dominance were respectively, 0.363 (P < 0.05) and 0.459 (P < 0.01). The path coefficients for dominance to consort score, dominance to the proportion of mates, and consort score to the proportion of offspring sired were respectively, 0.895 (P < 0.0001), 0.419 (P < 0.05), and 0.607 (P < 0.001). Snout length and longevity explained 29% of the total variability of dominance in males (R2 = 0.290). Dominance explained 80% of the variability of consort score (R2 = 0.801) and 18% of the variability of the proportion of mates acquired (R2 = 0.175). Consort score explained 37% of the variability in the proportion of offspring sired (R2 = 0.368). The proportion of the variation in the dependent variable that is not explained by the given model is calculated as (1- R2). Path analytic models for female fitness The first basic model was non-significant (x2 = 12.00, df = 6, P = 0.062) and, therefore, provided an acceptable fit to the data. The second basic model, however, did not explain the observed covariances among the traits measured (x 2 = 10.12, df = 4, P = 0.039). Table 4.5 gives a detailed summary of the results for each of the tested models. Both nested models significantly improved the fit of the model to the data (P > 0.95 in both cases) since the difference in the x test between 93 13 CD CD O u S3 O "0 CD o &0 g '-3 CD <u 1-1 , p CD s CD 43 00 g IS •a o cn CD T3 o W H H-> o 1 W Q -t-> o CD Q T Hi « 13 o cn 15 o c o '-6 o O M . II CD -t-> +-* cd £ I cn ^ T3 a> in 'In t)0 s PH o <+H o (3 o '£ o PH o 00 a PH c£2 cn CD * o CD CD CD In O O </) •e o M (3 O CD CD t-i O CD cn •c o cn S3 O U a <H-H o op r+H CHH O O cn •c o cn S3 O U cd CD o o cn •c o cn 13 o\ ol o p W H H H Q Q Q Q 13 CD T3 cn S3 CD .2 S3 « >—i »> # O N o C N C N m O N o © * * * & 2 3 o ^ ° i m N O o o © © * * * * O N C O * o * * C N — i o © * . # : t— . o NO * * * m O N oo * * * . ro ON • NO TN • m © © CD CD CD 43 00 S3 CD > CD 3 00 * © S3 ^ O o cn a I T ) 0 0 N O C O o oo o O N C N 0^  O N O i - i o o o C O O P d o V PH O d V PH O V PH * * # # * * * to Table 4.5 Summary of the goodness of fit for the models in the female breeding system. The name, the degrees of freedom {d.f.), the x 2 test statistic, and the P value for each model are shown. Each nested model also includes a x 2 test between the basic model and the nested model in question (Basic-Nested). Model d.f. X 2 P value Basic-Nested {df, X 2 ) Model 1 6 12.00 0.062 Nested l a 8 12.07 0.148 2,0.07* Model 2 4 10.12 0.039 Nested 2a 6 10.19 0.117 2,0.07* Note: See Table 4.2 for description of the alternative models. *P>0.95 96 the basic model and the nested model (in both cases) was non-significant. Again, as the two models, lb and 2b, are not nested, the model with the smaller %2 is preferred (model 2b). According to model 2b (%2 = 10.19, df = 6, P = 0.117) (Table 4.6 and Fig. 4.4), pre-spawning energy content had a statistically significant influence on longevity and weighted consort score (pie = 0.542, P < 0.01; p c e = 0.589, P < 0.005). Weighted consort score had a significant influence on female success as indicated by her genetic mate's ranking (p mrc = 0.498, P < 0.02). Within this model, pre-spawning energy content explained 35% of the total variation of the weighted consort score in females (R2 = 0.346) and 29% of the variability in longevity (R2 = 0.293). Longevity explained only 8% of the variability in mate quality (R2 = 0.079) and the weighted consort score explained 25% of mate rank (R2 = 0.248). Discussion The primary objective of this study was to develop a set of a priori hypotheses describing the mating system of sockeye salmon based on life history theory and to parameterize each model by means of data presented in previous chapters. These models allow me to identify likely causes of the genetic fitness of Weaver Creek sockeye salmon. I have shown how integrating various components of the breeding system of sockeye salmon, natural history observations and path analysis can be used to infer the importance of direct and indirect pathways through the mating system, and thereby efficiently deriving a functional web of interacting factors. A series of path models was developed for the male and female sockeye salmon breeding systems, each consisting of three (for males) or two (for females) comprehensive models (the maximum number of plausible pathways among measured variables), and two (for females) or six (for males) nested, alternative models. A l l models included traits under direct selection with direct 97 00 PS '3 I ex CO PS o PX CO > CD 00 a o CD O CD c n > > c n 00 PS 13 <u CD t-c x> c d 3 CD JS -4-* a CD W H c d O -*-» T3 w Q +-» o CD )-i • r—l Q NO — CS H CD CD 'S o CD CD PS -4—* id CD 13 CD I-c d P S C T CD CD CD 1 -O O CO •e o c o PS o CD T3 CD -*-» -C op 'S CD t- i O O c n •c o c o PS o CD co T3 O 6b PS CD PS o O H o CD H c d 3 I-H c d 1 J -CD I d W Q I d f > p? o W Q CO P S CD TJ-PS CD O H CD Ti-PS * OO CO : O N O N • •<*• CN d d C N r n oo : ui C N • — i d ' d * oo O N d C N oo C N d * * * O N oo # # C N in oo C N O N o N O O N C N C O JF o • r " c o > PS CD g oo 8 CD l-c O 00 P S Y P S « J 3 c c d ^ c o 00 _ +-> CD CD O ^ C fH PH CD 0 0 O N o o v V ^ » * * * * and indirect pathways to fitness traits. I used the chi-squared statistic to identify those models (i.e. hypotheses) that were most consistent with the data. While the use of path analysis has increased in ecology over the past few years (Wootton 1994a, 1994b; Grace and Pugesek 1997; Bart and Earnst 1999), Petraitis et al. (1996) and Smith et al. (1997) have argued that the method has been misused in several studies. Petraitis et al. (1996) argue that i f the path coefficients are to be unbiased, the covariance structure in the sample must be the same as the covariance structure in the population. Hence, non-random and complex sampling and experimental designs (e.g. stratification) may lead to differences in sample and population covariance. Pseudo-replication was also a problem that occurred in some of the studies reviewed by Petraitis et al. (1996). These problems did not occur in my study, as I did not use non-random nor complex experimental designs, and each fish was used only once in the analysis. Perhaps the most frequent criticism was that in a large number of studies the sample size was too small for stable parameter estimates. Sample size should be at least five times larger than the number of estimated paths to ensure reliable results (Spirtes 1993; Hoyle 1995; Petraitis et al. 1996; Shipley 2000). Based on this rule, my sample of 27 males allows me to have five free parameters, and my sample of 20 females allows me to have four free parameters in my path models. Any path model with more estimated parameters would have to be interpreted cautiously. Characterizing a selective regime using point estimates of fitness measures and phenotypic measures of life histories presents problems in the way we attempt to explain evolutionary processes. The problem is that phenotypic data tells us little about the genetic basis for life history differences among individuals or populations. Studies that target the genetic component of phenotypic variation are essential i f we are to infer that natural selection has been operating 100 (Endler 1986). In this study, I have inferred that phenotypic traits, whose genetic basis is unknown, have evolved to maximize the genetic fitness of the individual. A n alternative hypothesis, for instance phenotypic plasticity, could be the operating force in salmonids, where the phenotype changes as a result of the environment in which the fish find themselves. Evolution can still play an important role in the shaping of this alternative hypothesis, however. In the case of phenotypic plasticity, it would be operating on the genetic basis of phenotypic plasticity as a trait itself, and not directly on the phenotypic traits of the organism. Furthermore, assuming that I have measured selection on observed phenotypic associations between characters and fitness measures, the findings may still be biased by unmeasured external factors (Rausher 1992). Limitations aside, path analysis is a useful tool for performing inferential tests in non-experimental complex ecological systems (Grace and Pugesek 1998; Shipley 1999). Path analytic model for male fitness A l l three comprehensive models for males gave a statistically acceptable fit to the data. However, the nested model 3b had fewer paths, improved the fit to the data significantly, and was both more parsimonious and did not violate the rule of unstable parameter estimates as a result of small sample size. I discuss the results from this model only. The model of the male breeding system excluded the paths from dominance to the proportion of offspring sired and from consort score to the proportion of mates acquired. Also, it did not include direct paths from the life history traits (longevity and snout length) to consort score. One path coefficient that was highly significant (P < 0.0001) in all models was the path linking dominance to consort score. This indicates that males that are highly competitive are also predominantly in a consort position, as typically observed in studies of sockeye reproductive behaviour (Schroder 1982; Gross 1984, 1985; Quinn and Foote 1994; Foote et al. 1997; Quinn 1999). 101 Intrasexual interactions among males for access to limited, ovipositing females, are expected to have a strong competitive component (Darwin 1859; Emlen and Oring 1977). In my study, longevity and snout length (an arbitrary variable chosen to indicate secondary sexual trait development) were important contributors to successful competition among males, suggesting that there is strong directional selection for increasing snout length, or highly developed secondary sexual traits and life span on the spawning grounds. Males that exhibit high levels of aggression and who are winners of male-male competition are of superior quality and are often the same males chosen by females (holding consort positions throughout the breeding season). In this model, dominance accounted for 80% of the total variation in the time a male was in a consort position. Intersexual interactions exist in mate choice, where one sex (often the male) attempts to influence members of the opposite sex (females) by displaying his dominance, large body size, or colourful plumage, which may have been established by the fate of former interactions with rivals, i.e. intrasexual selection (Jarvi 1990). It may seem natural that intrasexual competition will favour the evolution of social dominance hierarchies or any traits that may evolve as indicators of dominance, such as large body size or size of weaponry (Jarvi 1990), and that intersexual interactions will reinforce this evolution. However, in a review of male competition and mate choice studies, Qvarnstrom and Forsgren (1998) challenge this claim. They argue that the outcome of one may not necessarily reinforce the other in the evolution of a certain trait. For instance, winners of male-male competition (i.e. "dominant" males) in the sand goby (Pomatoschistus minutus) do not provide better parental care than the losers (Forsgren 1997), a trait that females tend to prefer over dominance when selecting a mate (Qvarnstrom and Forsgren 1998, provide a review of examples where females do not prefer dominant males). In sockeye salmon, the two seem to work in concert, where intrasexual competition favours the evolution of 102 exaggerated traits and longevity, and female choice (measured as consort score) reinforces this evolution through its strong association with dominance. Body shape and reproductive life span impose similar selection pressures on dominance. Since males with larger snout lengths are generally bigger in overall size (see results from previous chapter), selection seems to be favouring bigger males. This may be a unique attribute to short-distance migrating populations as they invest more of their energy into body development and less into body fuel to complete their migration (Hendry and Berg 1999; Crossin 2002). However, longer distance migrating fish may also experience selective pressure for larger body size within the constraints imposed by their breeding environment (i.e. the optimal body size in long distance migrating fish may be substantially smaller than in short distance migrating fish). Because of the negative association between snout length and longevity, the model indicates that bigger fish have a shorter reproductive life span. Longer living males (often the fish with higher pre-spawning energy content) may be shifting up in the dominance hierarchy. This could be a result of bigger males using up most of their energy in fights and, therefore, dying sooner and, thus, providing the opportunity for smaller males to move up the hierarchy. Of course, this may be a result of the study design as there were no newly arriving males that could fight the smaller, longer living males. The model suggests that dominance has a significant direct effect on the proportion of mates acquired (P < 0.05). This association may, in part, reflect the success of smaller sneaker males, who live longer lives on the spawning ground, thereby increasing their reproductive success by spreading out their sperm among many mates. The variation in the proportion of mates acquired explained by dominance, however, is relatively low (18%). As suggested in the previous chapter (Chapter 2) the measure of the proportion of offspring sired may be a better estimate of the true 103 reproductive success of males. Males that are most often in a consort position sire the most offspring (P < 0.001) and these same males are often among the highest ranking males in terms of dominance. Since both measures of fitness are only estimates of the true success, the results generally support the hypothesis that intra- and inter- sexual competition both contribute to the fitness of male sockeye salmon. The proportion of offspring sired seems to be the more reliable of the two measures as it more accurately estimates the actual contribution of males to future generations. In summary, male traits, specifically snout length and longevity, affect intrasexual competition directly through their influence on dominance and indirectly through their influence on female choice (consort behaviour). The direct and indirect effects appeared to be of approximately equal importance in determining genetic reproductive success. Also, the ranking of the males based on social interactions in the spawning arenas was a decent measure of genetic reproductive success, as measured by the proportion of offspring sired (33 - 40% explained). Sampling error in both the estimate of reproductive success and behaviour measures could potentially explain the variation in the genetic success measures that is not accounted for by the socially successful males on the spawning grounds. Perhaps by using a completely natural setting where male number is not fixed, and observing behavioural interactions more frequently and increasing the number of analysed offspring, the social mating success measures would explain more of this variation. Overall, the genetic results confirm and extend previous behavioural observations suggesting that dominant and consort males fertilize the majority of the eggs in sockeye salmon. Path analytic model for female fitness In the path models I developed for the female breeding system, I found that females seem to benefit greatly from higher energy reserves once they arrive on the spawning grounds. The 104 models each related female phenotype to their genetic mate choice, via its effects on the observed mating success (measured as longevity and weighted consort score). M y results suggest that although female size is important in overall female fitness, it may act as an indirect determinant through its positive association with pre-spawning energy content (see chapter three for more details). The path coefficients from pre-spawning energy content to both mating success measures were statistically significant (P < 0.01), indicating strong selection for increasing energy content in female sockeye salmon. Although many studies have examined the relative importance of female body size in female reproductive success under natural and experimental conditions in salmonid species (Van den Berghe and Gross 1984, 1989; Crisp and Carling 1989; Kitano and Shimazaki 1995; Fleming et al. 1997; Steen and Quinn 1999; Hendry et al. 2001; Healey 2001), no previous study has quantified the relative importance of energy content in female reproductive success. Through its effects on nest quality, construction, and guarding, and egg production, female body size has consistently been related to superior fitness. My results indicate that it is high-energy females that live longer lives, are visited by more dominant males, and also have a higher mate choice success. These individuals presumably have a higher fitness. The positive link between body size and relative energy content (see Chapter 3 and also Crossin 2002) may suggest that larger females have higher overall condition. Both measures can be used as indicators of female fitness. The measure of female fitness in my study is based on the quality of her mates, and not directly on the female's reproductive success. The use of two different fitness measures could, alternatively, explain the difference in my results from previous studies. High-energy individuals have a double advantage in that they are visited more often by dominant males and, once they have completed spawning, are able to defend their nests for a longer period of time from disturbance by other females. It seems likely that there is strong selection for 105 increasing body energy in females. Once they have achieved a minimum body size and invested enough energy into the development of their gonads, the benefits of increasing body energy seem to outweigh the increase in any other phenotypic trait (body length or secondary sexual traits). M y results suggest that hypothesis testing through path analysis can identify potential traits under directional selection under field conditions. More generally, path analysis provides both an exploratory and confirmatory approach to the study of individual variation of selection in ecological systems. Literature Cited: Bart, J. and S. L . Earnst. 1999. Relative importance of male and territory quality in pairing success of male rock ptarmigan (Lagopus mutus). Behavioral Ecology and Sociolobiology 45:355-359. Biddle, B. J., and M . M . Marlin. 1987. Causality, confirmation, credulity, and structural equation modeling. Child Development 58:4-17. Crisp, D. T. and P. A . Carling. 1989. Observations on siting, dimensions and structure of salmonid redds. Journal of Fish Biology 34:119-134. Crossin, G. 2002. Selection for a bioenergetic-morphological feedback mechanism in migrating adult salmon is related to migratory distance and elevation. MSc thesis, University of British Columbia, Vancouver. Darwin, C. 1859. The origin of species by means of natural selection, or The preservation of favored races in the struggle for life. London J. Murray. London. Darwin, C. 1872. The expression of the emotions in man and animals. London J. Murray, London. Emlen, S. T. and L. W. Oring. 1977. Ecology, sexual selection and the evolution of mating systems. Science 197:215-223. Endler, J. A . 1986. Natural selection in the wild. Princeton University Press, Princeton. Fleming, I. A . and M . R. Gross. 1989. Evolution of adult female life history and morphology in a Pacific salmon (coho: Oncorhynchus kisutch). Evolution 43:141-157. 106 Fleming, I. A . and M . R. Gross. 1994. Breeding competition in a Pacific salmon (coho: Oncorhynchus kisutch): measures of natural and sexual selection. Evolution 48:637-657. Fleming, I. A . , A . Lamberg and B. Jonsson. 1997. Effects of early experience on the reproductive performance of Atlantic salmon. Behavioral Ecology 8:470-480. Foote, C. J. 1988. Male mate choice dependent on male size in salmon. Behavior 106:63-80. Foote, C. J. 1989. Female mate preference in Pacific salmon. Animal Behavior 38:721-723. Foote, C. J. 1990. A n experimental comparison of male and female spawning territoriality in a Pacific salmon. Behavior 115:283-314. Foote, C. J., G. S. Brown and C. C. Wood. 1997. Spawning success of males using alternative mating tactics in sockeye salmon, Oncorhynchus nerka. Canadian Journal of Fisheries and Aquatic Sciences 54:1785-1795. Forsgren, E. 1997. Female sand gobies prefer good fathers over dominant males. Proceedings of the Royal Society of London B: Biological Sciences 264:1283-1286. Gomez, J. M . and R. Zamora. 2000. Spatial Variation in the Selective Scenarios of Hormathophylla spinosa (Cruciferae). The American Naturalist 155:657-668. Grace, J. B. and B. H . Pugesek. 1998. On the use of path analysis and related procedures for the investigation of ecological problems. The American Naturalist 152:151-159. Gross, M . R. 1984. Sunfish, salmon, and the evolution of alternative reproductive strategies and tactics in fishes. In Fish reproductions: strategies and tactics. (Wootton, R. and Potts, G., eds), pp.55-75. London: Academic Press. Gross, M . R. 1985. Disruptive selection for alternative life history strategies in salmon. Nature 313:47-48. Hayduk, L . A . 1987. Structural equation modeling with LISREL. The Johns Hopkins University Press, Baltimore. Healey, M . C. 2001. Patterns of gametic investment by female stream- and ocean-type chinook salmon. Journal of Fish Biology 58:1545-1556. Healey, M . C. and A. Prince. 1998. Alternative tactics in the breeding behaviour of male coho salmon. Behavior 135:1099-1124. Hendry, A . P. and O. K. Berg. 1999. Secondary sexual characters, energy use, senescence, and the cost of reproduction in sockeye salmon. Canadian Journal of Zoology 77:1663-1675. Hendry, A . P., T. Day and A. B. Cooper. 2001b. Optimal size and number of propagules: Allowance for discrete stages and effects of maternal size on reproductive output and offspring fitness. American Naturalist 157:387-407. 107 Hilborn, R. and M . Mangel. 1997. The ecological detective. Princeton University Press. Princeton, New Jersey. Holtby, L. B. and M . C. Healey. 1986. Selection for adult size in female coho salmon (Oncorhynchus kisutch). Canadian Journal of Fisheries and Aquatic Sciences 43:1946-1959. Hoyle, R. H . 1995. Structural equation modeling. Sage. London. Jarvi, T. 1990. The effects of male dominance, secondary sexual characteristics and female mate choice on the mating success of male Atlantic salmon Salmo salar. Ethology 84:123-132. Johnson, J. B. 2002. Divergent life histories among populations of the fish Brachyrhaphis rhabdophora: detecting putative agents of selection by candidate model analysis. Oikos 96:82-91. Joreskog, K. G., and D. Sorbom. 2002. LISREL 8.52 (Student edition). A Guide to the Program and Applications. Scientific Software, Inc., Mooresville. Kingsolver, J. G. and D. W. Schemske. 1991. Path analyses of selection. Trends in Ecology and Evolution 6:276-280. Kitano, S. and K. Shimazaki. 1995. Spawning habitat and nest depth of female dolly-varden Salvelinus malma of different body size. Fisheries Science 61:116-119. L i , C. C. 1975. Path analysis: a primer. Pacific Grove, C A . Mitchell, R. J. 1992. Testing evolutionary and ecological hypotheses using path analysis and structural equation modelling. Functional Ecology 6:123-129. Mitchell, R. J. 1993. Path analysis: pollination. In Design and Analysis of Ecological experiments. (Scheiner, S. M . and Gurevitch, J., eds), pp.211-232. U S A : Chapman and Hall. Mitchell, R. J. 1994. Effects of floral traits, pollinator visitation, and plant size on Ipomopsis aggregata fruit production. American Naturalist 143:870-889. Mueller, R. O. 1996. Basic principles of structural equation modeling. Springer-Verlag New York, Inc. Petraitis, P. S., A . E. Dunham and P. H . Niewiarowski. 1996. Inferring multiple causality: the limitations of path analysis. Functional Ecology 10:421-431. Quinn, T. P. 1999. Variation in Pacific salmon reproductive behaviour associated with species, sex and levels of competition. Behavior 136:179-204. Quinn, T. P. and C. J. Foote. 1994. The effects of body size and sexual dimorphism on the reproductive behaviour of sockeye salmon, Oncorhynchus nerka. Animal Behavior 48:751-761. 108 Quinn, T. P., A . P. Hendry and L. A . Wetzel. 1995. The influences of life history trade offs and the size of incubation gravels on egg size variation in sockeye salmon (Oncorhynchus nerka). Oikos 74:425-438. Qvarnstrom, A . and E. Forsgen. 1998. Should females prefer dominant males? Trends in Ecology and Evolution 13:498-501. Rausher, M . D. 1992. The measurement of selection on quantitative traits: biases due to environmental covariance between traits and fitness. Evolution 46:616-626. Roff, D. A . 1992. The Evolution of Life Histories. Chapman and Hall. New York. Schroder, S. L. 1981. The role of sexual selection in determining the overall mating patterns and mate choice in chum salmon. PhD thesis, University of Washington, Seattle. Schroder, S. L. 1982. The influence of intrasexual competition on the distribution of chum salmon in an experimental stream. In Proceedings of the salmon and trout migratory symposium (Brannon, E. L . and Salo, E. O., eds), pp.275-285. Seattle: University of Washington Press. Shipley, B. 1997. Exploratory path analysis with applications in ecology and evolution. American Naturalist 149:1113-1138. Shipley, B. 1999. Testing causal explanations in organismal biology: causation, correlation and structural equation modelling. Oikos 86:374-382. Shipley, B. 2000. Cause and Correlation in Biology. Cambridge. Smith, F. A . , J. H . Brown and T. J. Valone. 1997. Path analysis: a critical evaluation using long-term experimental data. American Naturalist 149:29-42. Spirtes, P., C. Glymour and R. Schemes. 1993. Causation, prediction, and search. New York. Steen, R. P. and T. P. Quinn. 1999. Egg burial depth by sockeye salmon {Oncorhynchus nerka): implications for survival of embryos and natural selection on female body size. Canadian Journal of Zoology 77:836-841. Van den Berghe, E. P. and M . R. Gross. 1984. Female size and nest depth in coho salmon {Oncorhynchus kisutch). Canadian Journal of Fisheries and Aquatic Sciences 41:204-206. Van den Berghe, E. P. and M . R. Gross. 1989. Natural selection resulting from female breeding competition in a Pacific salmon (coho: Oncorhynchus kisutch). Evolution 43:125-140. Wootton, J. T. 1994a. The nature and consequences of indirect effects in ecological communities. Annual Review of Ecology and Systematics 25:443-466. Wootton, J. T. 1994b. Predicting direct and indirect effects: an integrated approach using 109 experiments and path analysis. Ecology 75:151-165. Wright, S. 1920. The relative importance of heredity and environment in determining the piebald pattern of guinea pigs. Proceedings of the National Academy of Science 6:320-332. 110 General Discussion I have attempted to demonstrate a blending of evolutionary ecology, behavioural ecology, molecular genetics, even a little physiology, and some statistical analyses to provide another piece to the natural life history of salmonids, or more specifically, the mating system of sockeye salmon from the Weaver Creek spawning channel. In itself, it is by no means a solution to any conservation dilemma, nor a unique piece of theoretical work. The present work, while specific to sockeye salmon, is an example of the integration of various areas of biology, and may parallel the opportunities that are available in other areas of fundamental ecology in the hope to learn more about the natural world that we live in. A range of characteristic behaviours was identified, D N A profiling was possible, and life history characteristic tested as a result of the study design. D N A fingerprinting proved to be reliable and capable of resolving the adequacy of observational studies on salmon breeding patterns. It is through the integration of various areas of ecology and evolution that we are able to dissect and unravel the patterns and mechanisms responsible for the shaping of natural systems. Sockeye salmon was used as a model organism to evaluate the many facets of salmon reproductive biology that have remained ambiguous, but that are possible to resolve by the genetic profiling methodologies now available. After arriving on the spawning grounds, female sockeye salmon select and prepare nest sites, competing for the best locations. Within study enclosures, female breeding morphology was not an apparent product of sexual selection arising from female competition for oviposition sites, as has been suggested by earlier work on this species (e.g. Fleming and Gross 1994). Pre-spawning energy content, however, had a positive association with both life span on the spawning grounds, and the social status of the male with whom the female chose as her social partner, both 111 indicators of female fitness. Relative energy content may, thus, be the trait favoured by natural and sexual selection in female sockeye salmon. Males, on the other hand, arrive on spawning grounds with elaborate breeding morphologies and secondary sexual characters, developed during their return migration to their natal stream. The range in their morphologies and energy content were greater than those in females, suggesting a more intense selection on male characters. Males provide no parental care, but rather spend most of their time engaged in competition and displaying acts to win access to ovipositing females. This results in size-structured dominance hierarchies, where snout length and longevity seem to be the primary determinants. These same dominant males tend to spend most of their time in a consort position, actively courting ripe females. Interestingly, there was a trade-off between body morphology, including secondary sexual characters, and energy content, where bigger males had relatively lower energy at the start of the breeding period than smaller males. The smaller males, however, lived longer on the spawning grounds, suggesting an alternative tactic employed by smaller-sized male salmon. At the moment of egg deposition, subdominant males may attempt to fertilize some of the female's eggs (see Gross 1984; Keenleyside and Dupuis 1988), as observed in a few cases in my study. And in almost every case, a female had multiple partners throughout her reproductive period (i.e. more than one male contributed per redd), suggesting that female promiscuity is the rule, rather than the exception. This was true of males as well. The female then buries the eggs after each spawning event, excavates another nest, and spawns until all of her eggs are deposited. There was no evidence of assortative mating between pairing individuals, suggesting either a more dynamic setting, or a failure to detect such pairings as a result of the study design. 112 Large-bodied males and males that lived longer had consistently higher dominant rankings. Aggression (and displaying acts), or intrasexual competition, on the other hand, directly influenced mate choice, where more aggressive males were found in consort positions. Intra-and intersexual competitions were positively associated with the genetic fitness of male sockeye. Within the female breeding system, energy content seemed to be the most important contributor to longevity on the spawning grounds and the status of the social partner with whom she chose to associate. Both of these were associated with the female's genetic mate choice. Pacific salmon species are likely to provide excellent model organisms in studies of evolutionary ecology, where information of maternity and paternity are fundamental necessities. One important caveat is the number of eggs deposited by females, which demands vast amounts of resources, such as time and genetic supplies. In spite of this, my study suggests that when male reproductive success is indexed by the number of mates acquired or the number of offspring sired, behavioural estimates of dominance are likely to provide reliable approximations of male fitness. When female mate choice is indexed by the social status of her genetic partner, behavioural estimates of consort score weighted by her social partner's status provide adequate approximations. Not surprisingly, a marriage of behaviour and genetic techniques will offer the most reliable and adequate approach to the mating structure of organisms. The current study is among the very few now available on genetic mating systems in salmonid populations. Although there have been a few studies on Pacific salmon (Foote et al. 1997; Bentzen et al. 2001; Berejikian et al. 2001), most studies have focused on Atlantic salmon (e.g. Moran et al. 1996; Thomaz et al. 1997; Moran and Garia-Vazquez 1998; Garant et al. 2001; Taggart et al. 2001). Parentage analysis can quantify variance in individual reproductive success, examine the possible determinants of individual fitness, or provide means of studying 113 mate choice. It more accurately examines the evolution of alternative reproductive life histories. It provides data needed to distinguish between opposing hypothesis regarding evolutionary stable strategies (i.e. alternative, mixed, and conditional strategies) (Gross 1996). Evolutionary and behavioural ecologists can now calculate genetic relatedness and gene flow within wild populations, in an attempt to answer questions related to homing, kinship and inbreeding (e.g. Bentzen et al. 2001). Combining nuclear markers with other genetic markers, such as mitochondrial D N A , is useful for better understanding historical gene flow patterns in salmonid populations (e.g. Tessier et al. 1995; Redenbach and Taylor 2002). A major goal of future studies of genetic mating systems could be a focus on population- and species- specific systems. Comparing mating patterns in species-specific or geographically distinct populations that differ in ecological factors, phenotypic features, and migration difficulty will provide additional insight into important evolutionary processes involved in mating system evolution and sexual selection. Literature Cited: Bentzen, P., J. B. Olsen, J. E. McLean, T. R. Seamons, and T. P. Quinn. 2001. Kinship analysis of Pacific Salmon: insights into mating, homing, and timing of reproduction. The Journal of Heredity 92:127-136. Berejikian, B. A . , E. P. Tezak, L . Park, E. LaHood, S. L . Schroder, and E. Beall. 2001. Male competition and breeding success in captively reared and wild coho salmon {Oncorhynchus kisutch). Canadian Journal of Fisheries and Aquatic Sciences 58:804-810. Fleming, I. A . , and M . R. Gross. 1994. Breeding competition in a Pacific salmon (coho: Oncorhynchus kisutch): measures of natural and sexual selection. Evolution 48:637-657. Foote, C. J., G. S. Brown, and C. C. Wood. 1997. Spawning success of males using alternative mating tactics in sockeye salmon, Oncorhynchus nerka. Canadian Journal of Fisheries and Aquatic Sciences 54:1785-1795. Garant, D., J. J. Dodson, and L. Bernatchez. 2001. A genetic evaluation of mating system and determinant of individual reproductive success in Atlantic salmon (Salmo salar L.). The Journal of Heredity 92:137-145. Gross, M . R. 1984. Sunfish, salmon, and the evolution of alternative reproductive strategies and 114 tactics in fishes. In Fish reproductions: strategies and tactics. (Wootton, R. and Potts, G., eds), pp.55-75. London: Academic Press. Gross, M . R. 1996. Alternative reproductive strategies and tactics: Diversity within sexes. Trends in Ecology and Evolution 11: A92-A98. Keenleyside, M . H . , and H . M . C. Dupuis. 1988. Courtship and spawning competition in pink salmon (Oncorhynchus gorbuscha). Canadian Journal of Zoology 66:262-265. Moran, P., and E. Garcia-Vazquez. 1998. Mulitple paternity in Atlantic salmon: a way to maintain genetic variability in relicted populations. The Journal of Heredity 89:551-553. Moran, P., A . M . Pendas, E. Beall, and E. Garcia-Vazquez. 1996. Genetic assessment of the reproductive success of Atlantic salmon precocious parr by means of V N T R loci. Heredity 77:655-660. Redenbach, Z. , and E. B. Taylor. 2002. Evidence for historical introgression along a contact zone between two species of char (Pisces: Salmonidae) in northwestern North America. Evolution 56:1021-1035. Taggart, J., I. McLaren, D. Hay, J. Webb, and A. Youngson. 2001. Spawning success in Atlantic salmon (Salmo salar L.): a long-term D N A profiling-based study conducted in a natural stream. Molecular Ecology 10:1047-1060. Tessier, N . , L . Bernatchez, P. Presa, and B. Angers. 1995. Gene diversity analysis of mitochondrial D N A , microsatellites and allozymes in landlocked Atlantic salmon. Journal of Fish Biology 47:156-163. Thomaz, D. , E. Beall, and T. Burke. 1997. Alternative reproductive tactics in Atlantic salmon: factors affecting mature parr success. Proceedings of the Royal Society, London, Series B 264:219-226. 115 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0090762/manifest

Comment

Related Items