LAKE SPAWNING BY AN ITEROPAROUS SALMONINE: THE MATING SYSTEM OF BROOK TROUT (Salvelinus fonthalis)

PAUL JAMES BLANCHFlELD

A thesis submitted to the Faculty of Graduate Studies in fulfilment of the requirements for the degree of

Doctor of Philasophy

Graduate Program in Biology York University North York. Ontario, Canada

May 1998 National Libmy Biithèque nationale du Canada Acquisitions and Acquisitions et Bibliographie SeMces se- bibliographiques

The author has granted a non- L'auteur a accordé une licence non exclusive licence allowhg the exclusive permettant à la National Lhrary of Cdto BiblotMque nationale du Canada de reproduce, loan, distribute or sell reproduire, prêter, distriiuer ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or electronic formats. la fome de microfiche/nlm, de reproduction sur papier ou sur format électronique.

The author retains owsmhip of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts from it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation. LAKE SPAWNING BY AN ITEROPAROUS SALMONINE: THE MATING SYSTEM OF BROOK TROUT (Salvelinus fontinalis)

by Paul James Blanchfield

a dissertation submrtted to ?he Faculty of Graduate Studies of York University in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Permission has been granted to the LlBRARY OF YORK UNIVEEISITY !O lend or sefl copies of this dissertalion. 10 the NATIONAL LIBRAAY i3F CANADA Io microfiim fhis dissertation and to iend or seil copies of the film. and io UNIVERStFf VICROF~LMSto ~ublishan abstract of this dissertation

The author reserves orhe: oublication righfs. and neirher rhe dtsserta!ion nor extensive extracts from lt niqy be Drinted or orherwise reDroduced wirhout ihe author's written permission ABSTRACT

The study of mating systems focuses on the interactions of males and females

from the perspective that al1 individuals compete to maximise their reproductive

success. Although there has been extensive study of reproductive patterns in

salmonine fishes (salmon, trout and char). there is limited understanding of the

reproductive interactions that occur under natural conditions. Istudied the

reproductive stmtegies of lake-spawning brook trout, Salveleius fonthalis, and

relate these findings to current mating system theoiy.

Spawning was characteriseci by cornpetition arnong females for spawning

sites that contain upwelling groundwater. Extensive reuse of spawning sites and

fernale removal expriments indicated that groundwater sites wre a limiting

resoutce. Males competed for access to females and were preçent in greater

numbers than females throughout the spauming season. Mate searching by

males induded many repeat visits to females and allowed males to predict

female readiness to spawn. Large body size conferred a reproductive advantage

to males searching for mates, but searching behaviours became restricted Mth

an increasingly male-biased operational sex ratio.

Peripheral males exerted a mating cost to dominant males and females.

The potential for stolen fertilisations was greatest for males paired with large fernales due to the presence of numerous peripheral males. Egg cannibalism by

peripheral males was greatest for fernales paired with small males. Latency to spawn by fernales increased when paired with relatively small males, and resulted in females obtaining a large?spawning partner and size-assortative mating.

Large females exert a strong influence on the seasonal timing of spawning. Small females (45.0 cm FL) spawned significantly later during the year that large females were detained (1996) compared to the previous two years. The delay by srnall females may be a tactic to decrease brood loss through nest superimposition or gain access to high quality spawning sites. Very few spawning sites (AH)used by females during the four years of this study, had rates of groundwater flow that wuld result in offspring suivival (Le. MO m~m"min*'). Thus, reproductive suucces of female and male brmk trout may be constrained by the limited number of spawning sites that have high rates of groundwater flow in this population. This projed wuld not have been possible if not for the guidance and support of my advisor, Mark Ridgway. Mark, many thanks for dl your advice and friendship over the last several years. I appreciate the support and opportunities you have given me throughout my graduate career. I thank my CO-supefvisor, Don

McQueen, for his insights into my project and for the use of his lab. 1 am especially grateful to have had Bddget Stutchbury as a committee member.

Bridget, thank you for your advice and critical comments of my work, along with the fnendship and optimism you have given me throughout the course of my

PIi.D. I thank my committee and examining members, Drs. B. Fenton, 6.

Loughton, S. MacDonald and G. Power, for constructive comments on my work.

A project of this nature wuld not have been possible without the help of dedicated field assistants. I thank Alison Little, Kim Hughes, Kim Mandzy. Kurt

Schryer, Loci Flavelle and Scott Milne for excellent support, often under miserable conditions. Fish lab regulars, namely Greg Betteridge, Doug Brown and Gary Ridout. provided invaluable advice and field assistance over the last few years. A special aianks to Doug for fixing (almost) everything Ibusted. I am indebted to Jackie King for providing lively help in the field dong with mernorable sledding and minnowtrap endeavours. I am grateful to staff at the Algonquin Fisheries Assessrnent Unit for advice and Nom Quinn for permits to work in the park. Special thank to Adrienne Dome and Susan Cameron for their patient administrative dealings. I am indebted to the Long-Tem Ecologicaf Research

Project of the Ontario Ministry of Natural Resources for providing stipend and research money. I thank Harkness Laboratory of Fisheries Research for providing equipment and accommodation. My research was funded by York

Feldwork and Research Cost Funds and a Natural Sciences and Engineering

Research Scholarship to Mark Ridgway. I also received support from a York

University Entrance Scholarship and a Dean's Academic Excellence Scholarship.

I am grateful for support from York to attend conferences.

Friends at Harkness. York and elsewhere provided much encouragement and ample distraction during the course of this degree. The setting and the people made my tirne at Harkness most enjoyable. Discussions over wffee with

Pete Biro and Jackie King were always an interesting mix of emlogy and life. I am fortunate to have becorne friends with several exceIlent behavioural ecologists, namely Lalita Acharya. Diane Neudorf, Jane Waterman and Richard

Wagner; al1 of whom have contributeci to my education in some way. Valerie

Blazeski, Sylvie Bouchard, Jenna Dunlop. Lawrence Giraudi and Jim Roth made the stay at York enjoyable.

Fellow lab mates, Chris Davis, Eric Derners, Brian Petri and Jim Rusak. provided helpful discussion and comments on many aspects of ecology. As well,

I thank Brian and Chris from respite from my thesis when needed - most wmmonly in the form of fishing trips. Thanks al- to Brian and Eric for Dorset hockey and art lessms.

I am especially thankful to my family for support of this project Menthere has been no history of such endeavours. I thank my parents. Christopher and

Jeanne for teaching me patience and tenacity. I am grateful to my extendeci family; Mark, Kate, Shannon, Connor, Quinn and Carl. Jennifer, Amy, Adam and

Elliot BlancMield, for support Thanks Mark and Kate for help in the field and Cal for graphical expertise. I am indebted to Alison for her patience and underçtanding, as wII as assistance over the last several years.

I dedate thiç thesis to the memory of Davki Tumbull - a good fniend who jnirOduC8dme to research h Algonquh.

viii TABLE OF CONTENT'S

List of Tables ...... xii List of Figures ...... xiii

CHAPTER ONE: General introduction

General introduction ...... 2 Salmonine mating systerns ...... 10 Brook trout mating system ...... 13 Thesis objectives ...... 15

CHAPTER TWO: Reproductive timing and use of redd sites by lake- spawning brook trout (Salvelinus fontinalis)

Abstract ...... 18 Introduction ...... 19 Methods ...... 22 Study site ...... 22 Adult observations ...... *...... 22 Redd selection ...... 24 Statisücal analyses ...... 27 ...... Resuits ...... 28 Adult observations ...... 28 Recfd selection ...... 36 Discussion ...... o....I...... 42 Acknowiedgments ...... 53 CHAPTER THREE: Wesearching behaviour: the role of male body sb and breeding competition

&&rad ...... 55 Introduction ...... 56 Metf\ods ...... 61 Male mate searching behaviour ...... 61 Seasonal patterns of male rnovements and behaviour ...... 62 The operational sex ratio ...... 66 Reults ...... 67 Male mate searching behaviour ...... 67 Seasonal patterns of male movement and behaviour ...... 72 The operational sex ratio ...... 74 Oiscussion ...... 80

CHAPTER FOUR: The cost of peripheral males in a brook trout mating system

Abstrad ...... 89 Introduction ...... 90 Mettlocfs ...... 93 Field observations ...... 93 Vide0 analyses ...... 95 Rgsults ...... 96 Discussion ...... *...... 104

CHAPTER FIVE: Size-based May in breeding-schedule in brook trout: e remval expriment

Absfract ...... 114 Introduction ...... 115 Methods ...... 120 Detention of large fernales ...... 120 Factors that may affect the timing of spawning ...... 121 Female size, nest depth and brood loss ...... 124 Results ...... 125 Discussion ...... 132 Foiced postponement of breeding due to cornpetition ...... 133 Size-specific tactics in breeding schedule to avoid brood destruction ...... 134 CHAPTER SEA resourcsbased mating system ra...... 140 Introduction ...... 141 Methods ...... 144 Location of spawning sites ...... 144 Measurement of groundwater ...... 144 Egg suNiVal experiment ...... 146 Resuits ...... 149 Discussion ...... 152

CHAPTER SEVEN: Summary

Surnmary ...... 158

Literature Ci...... 164

Appendix One ...... 180 LIST OF TABLES

Measurements (mean and ranges) of groundwater flow and conductivity of brook trout spawning sites and random sites ...... 37

The relationship between body size (FL) and various mate searching behaviours during 30 min observation periods of male brook trout ...... 71

Seasonal differences in the number and activity of male and female brook trout in relation to the breeding pefiod ...... 76 Seasonal differences in male movement patterns and behaviour in relation to breeding period ...... 78

Comparison of per capita aggression by the dominant male in relation to the number of peripheral males present ...... 97 Comparison of the body sizes of the mating pair and number of peripheral males preçent for spawnings that invoived peripheral males ...... 101

Factors which may affect the annual timing of spawning by female brook frout ...... 123

The annual number of females present, active and proportion active during three breeding seasons ...... 128 Spearman rank correlation coefficients of the relationship betvuwn spawning site use by females and the rate of groundwater flow ...... 151 A cornparison of studies in which seepage meters were used to detemine groundwater flow at brook trout redds ...... 193 A cornparison of methods used to determine groundwater fiow at br& trouf redds ...... 194 UST OF FlGURES

Scheme descnbing the influence of resource and female distribution in mating systems ...... -. 4

A synthesis of mating system theory describing the interactions between sexes with respect to sexual selection and parental car8 ...... 7

Map of study site showing the location of Scott Lake, Algonquin Park, Ontario ...... 26 The mean number of male and female brook trout obsedon the spawning grounds in 1994 and 1995 ...... 30 Number of days that individual male and female brook trout were observed on the spawning grounds ...... 31 Daily timing of actual spawning by brook trout observed during the 1994 and 1995 spawn ing seasons ...... 33 Daily number of spawnings and water temperature at Scott Lake in 1994 and 1995 ...... - ...... 34 Cumulative precipitation at Scott Lake as of 1 September 1994 and 1995

...... O o...... o...... m...... 35 Number of spawnings that occurred at spawning sites in 1994 and 1995

...... o...... m...... * ...... 39

Number of redds wnstructed by spawning brodc trout females of different body sizes in 1994 and 1995 ...... 40

The duration betwwn first and last spawning by females during the 1994 and 1995 breeding seasons ...... 41

Map of study site showing the cure area of use by a male brooù trout throughout one breeding season ...... 65 The pattern of mate searching behaviour by a male brmk trout during a 30 min observation penod ...... o...... 68 The frequency by which searching males made repeated visits to unoccupied sites and females ...... 70 Seasonal patterns of space use and behavioural obsewations of male brodc trout in relation to male body sire ...... -...... 73 . . Seasonai measures of breeding synchrony and operational sex ratio baçed on census swims during the 1994 and 1995 breeding seasons

...... O...... O...... ,...... o..... 75 Seasonal patterns of space use and behavioural observations of male brook trout in relation to the operationai sex ratio ...... 79 The potential mating success of peripheral males in relation to the number of peripheral males present ...... 98

The difference in pair size was cumpared for spawnings that involved only the spawning pair versus those with peripheral male participation

....o...... o...... o....,...... ) ...... 100

Comparison of female body size, dominant male size, and the number of males present at the time of spawning ...... 103 The difference between the mean body size of dominant males present prior to spawning and the dominant male present at spawning ...... 105

Comparison of potential loss of patemity in relation to the facundity of female brook trout ...... 108 The timing of spawning by small female brdtrout during control years and experimental year ...... 126

Arnong year cornparison of the cumulative number of spawnings by srnaIl femaie brook trout ...... o...... 127 The relationship between nest depth and female body size for brook trout and other salmonines ...... ,...... ,.....*...... 130 Brod loss through redd superimposition for small and large female brook trout ...... t...... 131

Map of Scott Lake indicating redd site locations recorded frorn 1993- 1996 ...... 145 The frequency of spawing site use and rnean groundwater seepage rates of spawning sites used in varying numbers of years at Scott Lake ..... 150 6.3 The proportion of eggs that emerged and survived in relation to rates of groundwater ffow ...... 153 ... 6.4 Logistic regression of the egg suivival ('yes' or 'no') in relation to rates of groundwater flow ...... 154 Al. 1 Tank trials of seepage meters to determine anomalous inflow of water over time ...... 189

A1.2 Field trials of seepage meters to determine anomalous inflow of water over tirne ...... , . 191 CHAPTER ONE:

GENERAL lNTRODUCTlON GENERAL INTRODUCTION

The study of mating systerns encompasses two broad areas - sexual seldon and parental are. Sexual selection focuses on individual variation in aspects of mating behaviour, such as courtship. mate choice, th8 presenœ and characteristics of pair bonds and the number and way in which mates are acquired. Parental car8 focuses on the form and duration of care by each sex

Current theory on mating systems is founded on sex differences in the relationship between fecundity and mating success (Bateman 1948), the role of parental Gare (Trivers 1972) and the spatial and temporal distribution of resources (Ernlen 8 Oring 1977). Essential to an understanding of mating systems is thaï we recognise mating not as a co-operatk venture benNeen the sexes, but where ind~dudscompete to maximise reproductive success

(Clutton-Brod 1989). Below I briefiy describe previous syntheses and present recent approaches to the operation of mating systems.

In their influential paper, Emlen & Oring (1977) indicated that the evohtion of mating systems is shaped pnmarily by the distribution in time and space of resources necessary for each sex to ensure successful reproduction. Fernales often need just one or a few matings to fertilise al1 their eggs. In cases where males provide only spem (Le. many mammals), female reproductive success is not limited by access to males. but by access to resources; the more resources a female can gather the greater the likelihood is of hec offspring surviking. Males, on the odher hand. have a much higher reproducüve potential than females. For males, reproductive success is limited by access to females; the more females a male can mate with, the greater his reproductive success (Baternan 1948). This difference in the factors lirniting the reproduction of the two sexes implies that female dispersion should primarily be influenced by resources while male dispersion should be influencsd by female dispersion (Davies 1991) (Fig. 1.1).

Previous studies of the dispersion model as a primary factor shaping mating syçtems have relied upon expriments in which resource distribution or female distribution has been manipulated (e.g. Ims 1988). Although these studies provide dear results, a convincing explanation for animal dispersion does not exist because it is often d*Mcult to identify resources and rneasure their distribution in the field (but se0 Cronin & Sherman 19ï7).

From its conception, the theory of malselection has accounted for the fad that the degree to which sexually selected traits are expressed is related to the type of mating system and parental roles (Darwin 1871). A primary focus in studies of reproductive patterns has been malemale cornpetition over access to females. since males commonly campete in drarnatic and conspicuous ways for females as a lirniting resource (see Anderson 1994). This perspective has m Direct i 1 Predation 1 1 Female dispersion I competim I Male dispersion

Benefits and costs of social living

Figure 1.1 Where males do not provide parental care, fernale reproductive success is lirnlted by resources. Thus, the dispersion of resources determlnes the dlsperslon of females (whlch may also be lnfluenced by predation and benefits and costs of social living). Male reproductive success is llmited by access to females, therefore, males compete for females directly or the resources females require (from Davies 1991). resulted in the tendency to dassify mating systems based on the degree to which mates could be monopolised as a resul of the spatial distribution of resources and the temporal availability of mates (i.e. environmental potential for polygamy;

Ernlen & Oring 1977). Many taxa (e.g. fishes) did not fit into these early classifications, which were primarily a product of avian- and marnmalian-bas4 wrk. Recent perspectives have focused on the role of females in shaping mating patterns (see Ahnesj6 et al. 1993). Recent advances in molecular genetics, especially DNA techniques whereby patemity and matemity can be routinely quantified in natural populations, have added greatly to our understanding of mating systems. In particular. the preponderance of extra-pair fertilisations

(EPFs) in 'monogamous' birds have revealed that fernales may have different social and genetic partnen (Birkhead & Msller 1992). Furthemore, EPFs appear to be under fernale control for at least some species (Wagner 1991; Neudorf et al. 1997). These findings have produceci renewed interest in the relationship betuwn confidence of paternity and male parental care (8.g. Whittingharn et al.

1992).

Trivers (1972) introduced the concept of parental investment and proposed that the relative expendiure of the sexes in aieir young is oie key variable controlling the operation of sexual selection. Parental investrnent includes investment into gametes. as well as any behaviour that benefits the young (Le. guarding, feeding). Recently, this framework has been extended and

induded in the concept of potential reproductive rate (PRR) (Clutton-Brock 1991;

Clutton-Brock & Vincent 1991; Clutton-Brock & Parker 1992). Essentially, the

PRR represents a %mebudget' measured in ternis of the number of offspring

parents cm produce within a given breeding season. The type and duration of

parental care by each sex will determine differences in the PRR. The sex that

inveçts less in parental Gare (usually males) will then be able to breed more

quickly and in tum have a higher PRR. In this way, the PRR predicts which sex is

the predominant cornpetitor for mates based on time allocation and, in effect, is a

measure of the operational sex ratio (OSR). defined as the ratio of females to

males that are ready to mate (Ernlen & Oring 1977). The link between PRR and

OSR is easy to visualise. Greater differences in PRR between the sexes will

result in a more biased OSR, and consequently biases in the OSR determine the

intensity of cornpetition for mates (Clutton-Brock 8 Parker 1992).

Currently, there is much debate as to the relative contributions of mating opportunity (ûateman 1948; Williams 1966). parental care (Trivers 1972) and the spatial and temporal distribution of resourceç (Emlen & Oring 19TI) in shaping mating systems (Arnold & Dwall 1994). However, what is dear is aiat an integration of the above features is necessary for any updated theoretical framework. It is important to appreciate that a conceptual framework that Gametes + Cam Mate cholce Cornpetition Quantity of mates

Competitbn

Mate cholce Genes + Resources

Figure 1.2 Reynolds' (1996) synthesis of breeding systems includes the costs and beneflts to both sexes from mate choice, cornpetition and advertisement. Sex differences in gamete production and parental care per brood determine potential rates of reproduction (PRR), which is influenced by environmental constraints (Le. clustering of resources). Differences in the PRR between sexes determine the costs and benefits of competing, advertising and chooslng mates. The OSR (operational sex ratio, see text) 1s determined by adult sex ratlo biases. integrates the various perspectives described above rnay still not appfy to aie diversity of observed meting systems. One example of an apdaied ih&retical f ramework for mating systems is Reynolds' (1996) synthesis that incorporates the costs and benefits to both sexes from mate choice, competition and adverthement (Fig. 1.2). This sdieme bonows from the concept of PRR and integrates how sex differences in the PRR influence the OSR. In addition. the

PRR cm be affected by environmental constraints. such as the dustering of resources or mates. This scheme is appealing because it provides a framework in which to address the roles of intra- and inter-sexual competition in shaping mating patterns and accommodates variation among indMduals in courbhip and competition.

Whatever theoretical framework we car8 to operate under, one of the main factors Iimiting our understanding of mating systems is the inability to observe interactions and choices by ind~dualsand to observe entire breeding populations. In general, such levels of detail are rare in studies of mating systems; however, there are exceptions such as the study of leks (H6gIund &

Alatolo 1995) and lek-like mating arenas (Wagner 1992). Briefly, leks are clusters of srnall territories where males congregate and display to attract mates.

Females visit these arenas for the sole purpose of obtaining matings and males do not provide access to any resources or parental care (H6gIund & Alatolo 1995). There are severai features unique to lek and lek-like mating systms that have consequentiy resulted in some of the greatest insights into-matechoice. semal selectÎon and the evolution of mating systems. First and forernost, breeding at leks tends to take place in relativeiy defined areas (or arenas). Thus, it is possible to mark and observe the adions of al1 (most) breeding individuals.

Leks also provide a finite boundary to a population, such that al1 indMduals congregate ai these arenas for breeding; however, rnovement among leks is known to occur (Hbglund & Alatolo 1995). Finally, the display sites of males contain no resources required by females and therefore the absence of male contributions cannot confound the obsewed choices.

In summary, the study af mating systems centres on the relative contributions of sewal selection, parental care and the spatial and temporal distribution of resources and mates. Recent theory attempts to accommodate these aspeds in an updated framewrk; howver, our understanding of mating systems is lirnited by the ability to observe individual interactions within entire breeding populations. SALMONINE MAT1NG SYSTEIUS

The salmonines (subfamily Salrnoninae) include the genera Oncorhynchus. SaIrno and Salvelinus; a classification that roughly corresponds to the salmons, trouts and chars, respeciively. The diversity of life history patterns within Mis group of fishes presents a great deal of intraspecific variability in mating systems. Briefly, some examples of the diversity exhibited by salmonines include: semelparous and iteroparous breeding; migratory versus resident (stream or lake) populations; spring and fall breeding seasons; and the evolution of alternative life history tactics.

Because salmonine reproductive behaviour is so diverse, I provide a description that best characterises the mating system for this group of fishes. I have kept this description brief since I provide more details and cornparisons in the chapters that follow. Males arrive on the spawning grounds first, but do not show signs of tenitoriality in the absence of active females (Foote 1990).

Fernales cornpete for spaw ing sites and dig pits (cdlecüvely called a 'r- into which depositeâ eggs are sirnultaneousiy fertiliseci by one or more males (0.g.

Keenleyside & Dupuis 1988). Fernales cover egg pits in succession and when finished defend the entire redd from re-use by other females (van den Berghe &

Gros 1984). Young emerge in spring. Male cornpetition for access to spawning fernales is intense and has resuited in dear semal dimorphism and in some cases,

the evolution of alternative mating tactics (Gros 1985). Generalty. the ability of a

male to gain access to females is based on body size (e.g. JBrvi 1990) or other

sexually selected charaders (Quinn & Foote 1994), and there is a first-male

advantage in fertilisation success (see Mjralnered et al. 1998; Chapter 4). One

exception to this general pattern of breeding is that d lake trout, Sahelhus

namaycush, in which there is neither female site preparation nor cornpetitin for

mates by males (Martin 1957).

Size and age at maturity of male salmonines has drawn considetable

attention and much effort has been put forth to explain the coexistence of two

distinct phenotypes; large males which compte for access to fernales and much

smaller males (sometimes a hundredth the weight; Fleming 1996) which attempt to

'sneak' fertilisations (G ross 19û5). These phenotypes are found in At lantic (Mmo

salar) and some Pacific (Oncorfrynchus W.) salmon and are te& 'pan' and

'jacks', respectively (see Gross 1984; Chapter 4). There is little evidence to suggest that these phenotypes are geneticaliy distinct, and thus they are fepresentaüve of a condiaonal strategy with alternative Me history tactics (Gross 1996).

Most previous research on the mating systems of salmonines has concentrated on Pacific salmon (Oncorhynchus spp.). Pacific Salmon are semelparous, have immense spawning runs (Le. millions) and primarily swwn in lotic (stteadriver) environments (see Groot & Margolis 1991).The large nurnber of individuals and necessity to conduct observations from shore'mak6s detailed investigation af these spawning populations difficult. In an effort to alleviate this problern, much of the field research in this area has entailed the examination of smali Stream tributary populations (e.g. van den Berghe & Gross 1984, 1986). or has relied on the use of artifidal strearn settings to simulate natural conditions

(8.g. Foote 1990; Fleming & Gross 1994). Although this latter option is a viable alternative, data derived in this manner should be viewed with caution as these animals do not have available to them the range of options. with respect to rnovement or mate dioice, that would be possible in the wild. In any =Se1 little quantitative information exists on individual reproductive behaviour under natural conditions for Pacific salmon. Recent contributions on lake spawning by eeye salmon (0. netka), however, provide some individual-based observations for this species (Quinn & Foote 1994; Hendry et al. 1995; Quinn et al. 1996).

Reproduction in Pacific salmon is characterised by a short breeding lifespan foilowed by death, little movement on the spawning grounds and alternative male phenotyp8s. The inability to observe interactions among marked individuals under natural conditions has led to controversy over the operation of mate choice in Pacific salmon mating systems (Sargent et al. 1986, 1988; Foote

1988a). Likewise, quantifying the range in quality of spawning habitat is difficult for these çystems. A consequence of the emphasis on Pacific &mon and

alternative male rnating tactics is that Our present perception of salrnonine

systems may not tnily reflect the range of mating patterns within this group of

fishes.

BROOK TROüT MATING SYSTEM

ihere are several characteristics unique to brook trout (Saivelinus fontindis)

which spawn in lakes that rnake them ideal assemblages in which to study

individual and population-level aspects of reproduction. The mst important

characteristic for this study is the distinct spawning requirement of upwelling

groundwater at spawning sites (Reiser & Wesche 1977; Wiel& MacCrimmon

1983: Ridgway & Blanchfield in press). fhe role of groundwater is crucial to the

study of brook trout mating systems for several reasons. First, these areas of

upwelling tend to be located in shallow areas and in dose proximity to one

another due to sub-surface geomorphology. As a result. the entire spawning

population congregates among these sites and can be easily observed. In addition, by conducting studies in headwater lakes, populations remain relatively closed to immigration. Studies confined to relatively small lakes (10-50 ha) allow for the identification of most indiduals in the breeding population (-100-120 fish). Furthemore, % is possible to quantrfy the rate of groundwafer flow at brook trout spawning sites (see Appendix 1). An increase in the rate of groundwater flow is predicted to increase egg suMval (e-g. rainbow trout, 0.mykss) (Coble

1962; Sowden & Powr 1985) and decrease incubation time (Ernbody 1934;

Power 1980). Therefore. groundwater flow presents a measurable resource in which to link habitat quality and female choice of spawning sites.

The consequeme of such a mating system, then, is that the actions of most breeding indMduals cm be monitored for an entire brdingseason. In general, such levels af detail are rare in studies d mating systems apart from leks, and have not been previously available for salmonines. lndividuakbased obse~ationsof breeding brook trout, under natural conditions. and the ability to quantify the resources essential for suwival of young will allow the opponunity for new insights into the mating patterns of salmonines. The resource-based mating system of brook trout provides a needed contrast to lek mating systems that have dominated field studies to date. THESIS OBJECTIVES

Br& trout are an ideal spscies in wtiich to examine the mating patterns of

çalmonines because of the ability to mark and observe an entire breeding

population. This allows booi for insights into reproduction by salmonines and the operation of mating systems. I first describe mathg in a lake population of brodc trout ni whidi the majodty of aduits are individually tagged and where the entire spawning population can be observed. In the chapters that follow, I develop and test hypotheses retated to: (1) male mate searching, space use and the role of the male body size in influencing male movements: (2) the cost of peripheral males and the influence of male and female mate choice; (3) social control of female breeding schedule and brood loss; and (4) the influence of resources in this mating çyçtem.

In this chapter I have presented a review of current literature reiating to mating system theory, followed by an OV~W~~Wof salmonine mating systerns. I have presented a case for why brook trout may provide new insights into the patterns of mating in salrnonines and suggested their relevanœ to an understanding of resource-based mating systems. In Chapter 2', i describe

This thesis contains two published manuscripts by Dr. M.S. Ridgway and myself (Chapter 2 and Appendix 1). The collection, analysis and interpretation af data contained in these manuscripts are my own work; CO-authorshipresuited from an exchange of ideas and editorial input. naturai reproduction by brook trout in lakes and indude study site infornation and detailed Methods that are common to al1 other chapters. More importantly, this chapter provides relevant background information to which the following four chapters will refer. Idescribe how male brook trout obtain mates through cornpetitive mate searching, and the influence of male body size on search behaviour (Chapter 3). In this chapter 1 aiso address how male movement patterns and mate searching behaviour are influenced by changes in the OSR as a result of female spawning synchrony. In Chapter 4,l examine the cost of peripheral males in this mating system and Iink this cost to male and female mate choice. Chapter 5 explores further the influence of females on the timing of breeding throug h a removal experiment. This field manipulation addresses whether cornpetition for resources can account for the patterns of breeding schedule generally observed for salmonines. In Chapter 6,1 examine the influence of resources in this brook trout mating system. I summarise the major findings of this thesis and provide general conclusions in Chapter 7. CHAPTER TWO:

REPRODUCTIVE TIMING AND SITE USE BY

LAKE-SPAWNING BROOK TROUT (SALVEUNUS FONnNALlS)'

Paul J. ~lanchfield'' and Mark S. ~idgwaf

' Department of Biology, York University, 4700 Keele St., North York. ON,

CANADA, M3J 1P3

Harkness Laboratory of Fisheries Research, Aquatic Ecosysterns Science

Section, Ontario Ministry of Natural Resources, Third Fioor North, 300 Water

St., Peterborough ON, CANADA, K9J 8M5

*Reprinted f rom the Canadian Journal of Fishenes and Aquatic Sciences volume 54, issue 4, pp. 747-756, with the permission of NRC Research Press. We provide a detailed description of a salmonine mating system based on daily observations of tagged individuals in a lakespavming population of brook trout

(Sahelinus fonthalis) throughout two breeding seasons. Actual spawning occurred over a period of -50 d. ûver 90% of spawning males were present soon after spawning commenceci and outnumbered females for the duration of the spawning peciod. The amount of time males and females remained on the spawning grounds increased with body size; however, males were present over a longer period than females of quivalent size. A distinct seasonal peak in spawning activity (-1 5 d) accounted for 58% and 84% (1994 and 1995) of al1 reproduction and was coincident wiai a dedine in water temperature below 11 OC and increased rainfall. Seledion of redd sites by fernale brook trout was determineci by groundwater flow which was significantly greater than at nonspawning sites. A preference for certain redd sites was observed, with 50% of spawnings occurring at 11 sites. The construction of multiple redds and duration in spawning actMty by females increased with body size. Extensive reuse of redd sites and rapid replacement of females during removal experirnents indicate that redd sites are a limiting resource. Among saIrnonines (salmons, trouts and chars), females prepare sites for spawning

(redds) and males cornpete vigorously for acoess to fernales (e.g. Keenleyside &

Dupuis 1988). Atthough not al1 species of salrnonines follow this genemi description

(0.g. Sahelmus namaycush; Martin 1957), the site-based cornpetitive mating systern in this group of fishes is the most frequently obsewed pattern of mating behaviour. In this mating system, mdes develop secondary sewal characteristics that show char maldimorphism and dichromatism dunng the reproductive season (Quhn & Foote 1994), and in some cases, the evolution of alternative mating strategies (Gross 1985).

Pacific salmon (pusOneorfrynchus) have becsme the mode1 systern for invesügaüng and describing these mating systems. In PacifÏc salrnon, femafe preparation of redds îs highly synchronous (van den Berghe & Gross 1986; Hendry et al. t 995), with females spawning and then defending one redd for -9 d (Hartman et al. 1964; van den Berghe & Gross 1986). These mating systems are further characterised by a relathrely short reproductive lifespan (7 - 12 d), limited movement of males among fernales (Hendiy et al. 1995), and terminal investment in reproduction as males and fernales die soon after spawning (Quînn & Foote

1994). Direct observations of ind~dualsin these mating SystemS are difficult, in part because of the size and distribution of the aduit population as well as the ditficutty of obseMng behaviour from shore. At a logistical levd, not identifying indidwls has led to some contrwersy about the operation of mate choice in coho saimon (Oncorfiynchus kisutch) (Sargent et al. 1986,1988; Foote 1988a). At a fundamental level, being unable to aççess the dispersion of the entire breeding population limits the application of patch choice rnodels as a theoretical frameinrork for site or mate choice (Sargent et al. 1986). ff breeding adults can not be fully sampled because of aieir dispersion then e>cperiments in enclosures can alleviate this problem to some ex!ent by providing an experimentai mode1 of a complete rnating system (e.g. Fleming & Gross 1994). Howvr, these manipulations cannot provide al1 the movement options facing free-ranging fish in a natural mating system. This is particularty true f some individuals, such as males using an alternative mating strategy, are moving throughout the spawning population based on the competitÏve abilities of other males and size of fernales at specific redd sites.

In any case, it is &en difiicult to delimit the effective population size in the large stream and lake systerns where PaMc salmon spawn.

The purpose of this stutudy is to describe mating in a lake population of brook trou (&h&P)us fmthalis) in which the rnajority of aduits are ind~duaiiytagged and where the entire spawning population can be observed. We present data on oie seasonality of redd site use and reproductive behaviour of lakespawning brook troout This data provides a sirong contrast with the literature on Pa&c salmon mating systems as well as one of the most extensive field descriptions of a salmonine mating systern basecl on an iteroparous population.

Information on the reproductive ecology of brodc trout is baseci largely on data collected in Stream (0-g. White 1930; Greeley 1932; Hanard 1932; Smith

1941; Neeôham 1961) and labofatory settings. Abiotic factors related to spawning substrat8 (Fraser 1985; Young et al. 1989), redd seledion (Wiitzel& MacCrimmon l983), and the requirement of groundwater seepage (Benson 1953; Reiser &

Wesche 1977; Cuny & Noakes 1995) charaaerise rnuch of what is known about brcnk trout reproductive ecology. Laboratory obsewafions have supported the generai observation of seepage flow as a preferred element of redd seledion (e.4

Webster 8 Eiriksdottir 1976) and have provided some descriptions of spawning behaviour (Hale 1968; Hale 8 Hilden 1969; Hokanson et al. 1973). Recent contributions on life history variation and the survival mst of reproduction provide the most cunent perspective on the reproductive ecology of brook trou (Hutchings

1994). METHODS

Study site

Scott Lake is a small, deep headwater lake (27.6 ha; maximum depth, 25 m) located within Algonquin Park, Ontario (4S02YN,78O431N) (Fig. 2.1). Scott Lake tends to be slightfy acidic (pH, 6.3-6.9), high in dissolvecl oxygen (8.4-9.6 mg-LI), with a conducavity of 26.6 p~-cm''at 25OC. The fi& community of Scott Lake is cornprisecl primarily of Culaea hoonsians. Pimephales promefas and Phoxinus spp. Catostomids are absent. Scott Lake supports a self-sustaining brook trout population last stocked with hatchery fish (Hill's Lake strain) in 1959. Despite hatchery plantings, Çcott Lake fish dosely resemble ancestral lineages, although hatchery influences cannot be completely nrled out (Danmann & Ihssen 1995).

Aduît observations

Brook trout wre caught using a 1.Som trapnet just prior to spawning (early

October 1994 and 1995) and thereafter were captured with dipnets by swirnmers

(mask and snorkel). Fish were anaesthetireci with tricaine methanesuifmate (MS

222) prior to tagging and measuring. A t-tag (Hallprint Co.,Australia) was inserted just beiow the dorsal fin of the fish and a uniquely coded disc was applied to this tag to allow for individual identification. At this time, fish were deteminsd to be ripe if eggs or milt wre extruded when gentle pressure was applied along the ventral surface. Fork lengths (FL) wre measured to the nearest millimetre and mass recorded to the nearest 10 g (Pesolam, 2.5 kg scale). Ail other field data wre recorded from underwater observations made by swimmers using drysuits, mask and snorkel.

Spawning was concentrated in one area of aie lake (Fig. 2.1); however, the entire shoreline of the lake was searched on tw separate occasions during the spawning period to detemine if spawning occurred elsewhere in the lake.

Only one redd site was found outside the main spawning area and this was monitored for the rest of the seasan. We conducted a census of the entire spawning area four %mesdaily (weather pemitting) once spawning activity cornmenced (1O October 1994 and 1995) until fish were no longer present on the spawning gmunds (6 December 1994) or the lake surface was frozen (24

November 1995). Three days wre missed during the 1995 season due to equipment failure. During each census the position and acüvity of eacb fish wre recorda on underwater dates by swimmers.

The term 'reddWhas been used to describe the nest or group of nests female salmonines prepare in Mich to spawn (Chapman 1988). This definition more commonîy describes the successive nests (or egg pockets) that a stream- spawning female digs and covers as she works her way upsiream. In Scott Lake, female choice of redd sites and spawning substrate did not always allow for the construction of contiguous neçts; therefore, we considered al1 non-contiguous nests in which a female spawned to be a separate redd.

The majority of spawnings occurred during the day and were often witnessed either by swimmers or via video camera, but more &en the presence of eggs in the nest or covering behaviour was used to distinguish a spawning act.

Spawnings were assumed to have occurred if redds that were open the previous evening were covered before Our arrival in the moming. If, during the first transect swim, we observed covering behaviour by the same female that was active and present at the redd the previous evening, then we considered the spawning act to have occurred just prior to ouf arrival. Often, the fernale that was active and present at a particular redd the previous evening was not found near the redd at which çhe was formerly active, although the redd was covered. In such instances we considered the spawning act to have ocwrred at some time after our departure the previous evening. In the text these spawning acts are referred to as "early" (before 9:00) and 'late" (after 18:00), respectively.

Redd selection

To determine the hydraulic preferences of spawning brook trout. we measured groundwater flow and conductivity at sites where we had observed spawning the previous season (1994, W3;1995, k22)and at randorn sites

(1994, MO;1995. k11). Random sites were located in close proximity (0.5-20 m) to redds where spawning had occurred and were of similar substrat0 composition. Of the 20 randomly-selected sites measured in 1994, 13 sites occurred YWithinn the spwning area; the remaining seven sites-were located elsewhere around the lake (Fig. 2.1). Measurernents of groundwater flow were made using plastic seepage meters just prior to the start of spawning in 1994 and

1995. Measurements of groundwater flow were made in Mplicate, lasted 2h, and used collection bags prefilled to 1 L (see Appendix 1 for full details).

A separate collection of groundwater was made without a prefilled bag to measure conductivity (microsiemens per centimetre at 25OC). Two conductivity meters were used during the 1994 season. Groundwater samples were measured with both meters to correct for differences in sensitivity in accordance with the more sensitive meter (Orionm model 140).

Measurements of depth, area and distance to shore were taken for al1 redds after spawning was complete. Estimates were made for four redds that were inaccessible due to f reezing in 1994. Area was calculaied by measu ring the longest ais and oie maximum width perpendicular to oie long axis for the entire cleared area. Water temperatures were recorded hourly throughout the spawning seasons at six locations within the spawning area using submersible temperature recorders (Ryan TempMentorsTY).Precipitation was measured continuaily from a weather station located on Scott Lake (Fig. 2.1). Precipitation prior to September would have little influence on spawning aaivities; therefore only fall precipitation Figure 2.1 Map of study site showing the location of Scott Lake, Algonquin Park, Ontario. The areas where spawning activity ocourred are shaded. Random groundwater sampllng sites from around the lake (dlamonds) and the weather station (W) are shown. data are included (1 September to 31 October 1994; 1 September to 15

November 1995).

To detemine if redd sites were limiting, a female rernoval expriment was

conducted at Stringer Lake (45026'N, 78O3UVV) in Algonquin Park. just following

peak spawning, on 7 November 1994. Stringer Lake was used for this

experiment since we did not wish to disturb my of the çpawning activity occurring

at Scott Lake. An active female was removed from her redd by a snorkeller with a

dip net. The removed fish was held in a container onshore Mile the redd was

obsewed. A female was considered to be replaced if a new female cornmencecl

redd construction actMties or actively defended the redd against other females.

We allowed a period of 1 h for females to be replaced. aven different fernales

were removed from their reddç and the tirne for each replacement record&. Ail

confined females were released to the spawning area immediately after their

redds were occupied by another female.

Statistical Analyses

We compared the onset of ripsness by males and females using a chi-

squared test. We used linear regression to analyse the number of days fish were

obsewed in relation to body size (In transformed). To be consistent between years, we recorded the arnount of time male and female brook trout were present

on the spavming area starting from the first day of spawning activity (10 October 1994 and1995). We compared the duration of time males and females spent on the spawning grounds in each year and within sexes between years using analysis of covariance (ANCOVA). We compared the rate of groundwater flow and conductivii at spawning sites versos unused spawning sites and random sites using the Mann-Whitney Utest (two-tailed). Because data were nonnormal. the number of redds used by individual fernales and the duration af actual spawning were analysed with a Spearman correlation. We followed a closed sequential test design to determine if redck were reocwpied (Cole 1962). Mean values (IlSD) are reported below.

RESULTS

Aduît observations

Brook trout were obsewed on or near the spawning grounds over a 64-d pehod in 1994 (30 September to 2 December), and a 614 period in 1995 (25

September to 24 November). Male and female brook trout were first observed congregating around a small inlet creek adjacent to the spawning area; however. no reproductive activity was obsenred at this time. Only one of the females caught prior to the first observed spawning was ripe. In contrast, 86% (f16%) of the males caught at aiis tirne were ripe. From the first day of spamhg, but pnor to peak spawning, 20% (f7%) of females and 98% (B%)of males caught were ripe. Throughout both spawning seam. more males were found to be ripe than fernales (1994, dkl, *=21.9, Pc0.0001; 1995, df =Il 2=63.5, Pc0.0001).

More males occurred on the spawning grounds than females. except for the very end of spawning (Fig. 2.2). In both years, peak numbers of males and females occurred at approxirnately the same tirne, &emndays 296 and 312

(22 October to 8 November), although a second peak in the nurnber of males occuned in 1995 (Fig. 2.2). Approximately 90% of al1 spawning males were present within 14 d following the first day of spawning, while only 60% of the spawning female population was present by this time.

There was a trend for larger fish to remain on the spawning grounds longer than srnaller fish (Fig. 2.3). lncreased duration on the spawning grounds with body size was significant for both males (M,f-=0.19, -.6, M.006) as

-Il as females (MlFd.27, k15.9, Pdûûl)during the 1994 season. This trend was not obçerved for males (W5.&0.002, kO.09, M.8)or females

(Ml,fd.06, 61-8, M.2)during the 1995 season. The duration of time spent on the spawning grounds was longer for males in 1994 than 1995 (31 f 12 d and 22 f 9 d respectively; k9.3, FL0.003),although no difference in duration was observed for females between years (1 1 f 6 d and 15 f 7 d respectively;

F2.7, M.l). Males, however, remained on the spawning grounds for a longer Day of the year

Figure 2.2 The mean (fl SD) number of male (solid circles) and female (open circles) brook trout observed on the spawning grounds in 1994 and 1995, beginning from the first day of spawning activity (day 283 is 10 October). Fork length (cm) Figure 2.3 Number of days that individual male (sdid cirdes) and female (open cirdes) brook trout were obswed an the spawning grounds during the 1994 and 1995 breeding seasons. period than females during both spawning seasons (1994, k100.8, Peû.001;

1995. k12.1, Pc0.001).

We observed spawning at al1 times of the day throughout both seasons, with a distinct peak in activity between 13:00 and 14:OO (Fig. 24). Eighty-nine percent of al1 spawning took place during the hours of daily obseivation (9:OO-

18:00), the remaining spawnings occurred either just prior to Our amval rearly";

3%) or someüme after our departure ('late"; 8%; Fig. 2.4).

The seasonal peak in spawning adivity (1994. days 296-310; 1995, days

296-312) (Fig. 2.5) coincided with peak counts of males and females (Fig. 2.2). In

1994 and 1995, peak breeding activity accounted for 58 and 84% of the total spawnings observed during aie season, respectively (Fig. 2.5). During peak spawning periods, water temperature in the spaming area decreased from 11.3 to 8.8OC in 1994. and from 10.3 to 5.g°C in 1995 (Fig. 2.5). In both years, peak spawning periods were preceded by substantial rainfall events such that 90 mm of rainfall had accumulateci (from 1 September) by this time (Fig. 2.6).

In general. spawning fernales were longer (1994, Ak47,372 16.5 cm;

1995. IW2,37.6 f 6.4 cm) and heavier (1994, M.656 I 395 g; 1995, M2,

705 f 381 g) than spawning males (FL: 1994, MO,33.4 f 7.7 cm; 1995, H5,

33.7 f 6.5 cm; mas: 1994, M.480 f 407 g; 1995, -,507 1: 351 g), akhough several small females were not captured. In each year a very small Midpoint of time interval (h)

Figure 2.4 Daily timing of actual spawning by brook trout observed during the 1994 and 1995 spawning seasons. "Early" refers to spawning estirnated to have taken place just prior to 09:OO; 'late" refers to spawning that occurred sometime after 18:00 (see text for details). peak f 1994

Day of the year

Figure 2.5 Daily number of spawnings at Scott Lake in 1994 and 1995. Peak spawning acüvity ocwrred over approximately the same period between years. Mean (solid line) water temperature (IlSD, broken lines) within the spawning area decreased throughout the spaming season. .-Sep 1 l-ûcî 21-0~3 10-Nov Date

Figure 2.6 Cumulative precipitation at Scott Lake as of 1 Septernber 1994 (solid circles) and 1995 (open cirdes). Peak spawning aMiby brook trout was preceded by rainfall events.

35 (SI4.0 cm FL), sexually mature male brook trout was camred on the spawning grounds and is induded in the above length data. Tmother similar-sized fish

(SI 4.0 cm FL) were also caught in 1994, but it was not possible to detemine their sex The stomachs of these two fish were full of eggs (P.J. Blanchfield, personal observation). These smaller fish were al- seen hiding dose to active redds, and wre never observeci as part of the hierarchy formed by males around fernales. men, these fish were aggressively attacked by active fernales.

Redd selection

Of the known spawning sites (obsewed the previous year) that wre measured for groundwater flow, 13 and 11 of these sites remained unused during the 1994 and 1995 spawning penods, respectively (Table 2.1). The flow rates of the unused sites wre not significantly lower than at redd sites where spawning occurred (1994, &13.10~51FL0.07; 1995, 411.l 1~55, M.7).

However, groundwater flow at spawning sites was significantly greater than at random sites located within the spawning area (1994, Qls,loF1 26, P~0.001;

1995, 411.1 1~11,k0.001) and around the lake (1994, L(7~,~70,Pe0.001). The conductivity of groundwater at redds was higher compared with random sites

(1994, QI~.IOJ=~,P

Redds were located in the shallow littoral zone (1994. 1.1 f 0.4 rn; 1995,

1.6 f 0.5 rn) and near shore (1994.6.6 f 4.6 m; 1995. 7.8-f 6.1 rn). Redds were

less than 1 m2 in area (1994.0.9 f 0.8 rn? 1995,0.7 I0.7 rn? resuiting in a total substrate area of 30 m2 in 1994 and 35 # in 1995. There waç evidence of female spawning activity (Le. dearings) in other areas; however. we obsetved spawning at only 32 and 47 redd sites in 1994 and 1995, respectively. In total, 60 reâd sites were used, of which 19 wwe used in both years (Fig. 2.7). Throughout the breeding seasons, there was an unequal distribution of spawning activity among redd sites. wDththe 11 most used sites accounting for half of the total spawnings (Fig. 2.7). Fernale brook trout often constructed more than one redd

(1994, 1.4 f 0.8; IWS. 1.7 f 0.8) (Fg. 2.8). The construction of multiple redds by spawning females increased with body size during the 1994 season (M,

M.44, &0.009), but not in 1995 (W3,W.039, FL0.9) (Fig. 2.8). In 1994 and 1995. female brook trout spent, on average, 3.6 f 2.7 and 4.1 12.6 d between their first and last obsewed spawning, reswvely. The number of days between first and la& spawning increased with female body sire (1994, M.

Ft-0.43. M.01; 1995, M3,A0.48, W.02) (Fig. 2.9). As well, most of the larger females spawned during the pend of peak activity. while only smaller females tended to spawn after this time (Fig. 2.9). Redd sites

Figure 2.7 Number of spawnings that occurred at redd sites in 1994 and 1995. Half of al1 spawnings occurred at 1 1 redd sites during the two breeding seasons. -

. > 30 35 40 45 50 55 Fork length (cm)

Figure 2.8 Number of redds constnicted by spawning brook trout fernales of âiierent body sizes in 1994 and 1995. 28aoa25 Day of the year

Flgure 2.9 Mean spawning date (circles) and the number of days between first and last spawning (lines) by individual females during the 1994 and 1995 breeding seasons. In ail seven removal trials females were replaoed within the allotted 1-h

time period (P<0.001) (Cole 1962). Active females removed from their redd wre

quiddy replaced by other females (12 f 7 min). In three trials, more than one female visited the redd prior to establishing dominance on that redd. In three additional trials, a single female was involved in reâû replacement. In another trial, the redd was visited twice, but we were unWe to determine 1 this was by the same female. Because they were not individually marked, it was diiicult to determine Y the replacing females were from neighbouring redds or "floaters". In one trial, a neighbouring female (A) started defending the unoccupied redd of the removed female (8) against other females. Later. we removed this female (A), and similar behaviour was shown by the initial female (B) after release. In another trial, a neighbouring female inspected a redd from which a female had been removed, but retumed to her original redd.

We have provided a detailed description of an iteroparous salmonine mating system, whbh was baseci on daily observations of the spawning actMties of marked indiiiduals throughout two breeding seaçons. Male recruitment to the spawning area was abrupt and occurred early in both seasons. In contrast, female recniitment occurred throughout the spawning periud with a continuai loss of spent fish, aithough there was a marked peak in spawning activity. These data, in combination with duration estirnates and evidence of npeness, show males to be sexually active over a much longer period than fernales. The cumulative distribution of males shows similar trends to Fraser's (1985) mark- recapture study in that little new recruitment of males occurs after peak spawning and few males leave aie spawning grounds.

The number of days male and female brook trout were present on the spawning grounds was dependent on body size during the 1994 season but not in 1995. One reason for a weakening of the body size duration pattern in 1995 was because spawning activity was obsenred for 11 fewer days than in 1994 and was more synchronous during the 1995 season. For males, this was reflected in the significantly greater mean number of days spent on the spawning grounds in

1994 (31 d) than in 1995 (22 d). There was no difference in amount of time spent on aie spawning grounds by females between years, which concurs with our above observations of continual recruitment into spawning area throoghout the breeding season. Another explanation for this difference between years may be due to increased breeding density in 1995 (Fig. 2.2). lncreaseâ breeding density in coho salmon was found to make a signifiant negative contribution to life-span due to increased cornpetition for resources (spawning sites for females; mates for males) (van den Berghe & Gros 1986). Since smaller male brook trout spent lestirne spawning than larger males (Fig. 2.3). we attribute the decline in

numbers of spawning males w observed to the disappearanc6 of small males.

In such a protracteci (some males obsenred for 64 d) and highly cornpetitive mating system, smaller males may incur higher reproductive costs from physical

injury due to mate cornpetition with larger males (Hutchings 8 Myers 1987;

Hutchings 1994). Smaller males may also incur greater energetic costs of reproduction relative to large males (Schmidt-Nielsen 1984). Post-reproductive energy deficits can be substantial in salronines (Jonsson et al. 1991) and are known to occur among stream-spawning brook trout (Cunjak et al. 1987;

Hutchings 1994). This combination of possible physical injury and energetic demand may be particularly acute for small males in a manner that is similar to

Arctic char, (Saî'velhus alpo?us) mating systems in lakes (Sigurj6nsd6ttir &

Gunnarsson 1989). In these Arctic char mating systems, as well as the Scott

Lake bro& trout mating system. satellite males can surround the dominant pair, and approach from any depth or direction. This non-linear arena contributes to continual aggressive satellite-satellite and dominant-satellite interactions, as well as extended assessrnent and fighting among males, which may be more costly to smaller males.

Both male and female brook trout were seen near and around the spawning area prior to the start of spawning. Many of the eariy-tagged females were not yet ripe. These same females wre routinely observed in stationary inactive groups within the spawning area for periods of up to 2 wk or more prior to spawning. Similar behaviour was observed with stream -brooktrout in which females did not frequent the spawning areas until ready to spawn, although they were often found in nearby pools (White 1930).

Actual spawning by brook trout occurred ove?50 d in 1994 and 43 d in

1995, with daily and seasonal peaks in spawning activity obseived during these periods. In contrast with a previous field study (Cuny & Noakes 1995; Curry et al.

1995). in which no actual spawning was obsenred, and a laboratory study (Hale

& Hilden 1969), in which spawning was obsenred at night, we observed brook trout spawning primarily during daylight hours, with peak actMty occumng around midday (13:ûû - 14:OO). Similar peaks in daily activity wre found in a spawning population of golden trout, (Oncuhynchus mykss aagabonita) (Knapp

& Vredenburg 1996).

A distinct seasonal peak in spawning activity occurred over -15 consecutive days, in which al1 of the larger females spawned. Peak activity was followed by an extendeci period of spawning by some of the srnallest females.

Large females were present on the spawning grounds eariy and were presumably able to secure a redd at any given time due to body size. This synchrony in spawning acüvity suggests that timing of spawning is an important aspect of fernale reproductive behaviour. Synchronous spawning by large females may also force some of the smaller females to delay fxeeding, similar to coho salmon (van den Berghe & Gross 1989; Fleming & Gross 1994).

Two hypotheses may account for synchronous spawning by females as a strategy (Knowlton 1979) to increase egg survivorship. First. it is possible that females are choosing to spawn synchronously, thereby lowering the operational sex ratio (OSR) as a strategy to reduce harassrnent (e.g. Boness et al. 1995) and egg predation by satellite males, since the number of males per active female should be fewer when many fernales are active (Quinn et al. 1996). Unlike Pacific salmon, which oease feeding pnor to spawning. egg cannibalism by satellite males has been observed in this and other lake-spawning populations of brook trout (see Chapter 4). Second. females prevent superimposition of their redds. and therefore increase brood survivorship. by spawning when other large females spawn. If nest depth increases with increasing female size, as in other salmonines (van den Berghe 8 Gross 1984; Knapp & Vredenburg 1996; Chapter

5),then a preferred üme to spawn would be with other large females. In this way, larger fernales compte for preferred redd sites, and smaller females could choose redd sites where large females are not spawning. Superimposition of redds by later-spawning small females may have little consequence for brood survival of larger fernales. Conversely. maller fernales may be choosing to delay spawning as a strategy to decrease redd superimposition by larger females

(Chapter 5) or obtain better mates (Knowlton 1979) if male mate choice is dependent on male size as in other salmonines (Foote 1988b). The proximate cues for synchroniring a peak in spawning activity may have been a combination of temperature and precipitation. Pflor.to spawning, daily variations in water temperature below 1I0C are thought to enhance egg viability in brook trout (Hokançon et al. 1973). At Scott Lake, spaming started when temperatures fell below 13OC, with most activity occumng as water temperature dedined below 11.3 and 10.3OC in 1994 and 1995, respectively.

Compelling evidence edsts for a significant. positive relationship between mean groundwater discharge rates and mean daily rainfall (Downing 8 Peterka 1978).

In 1994 and 1995, significant rainfall events (-30 mm) occurred juçt prior to peak spawning activity (Fig. 2.6). Presumably, groundwater flow increased during this perîod, and in doing so greater amounts of water (higher in conductivity than ambient lake water) would be present at redd sites. We suspect that an increase in groundwater flow in conjunction with a decrease in water temperature below

11°C stimulate brook trout spawning, and that these factors are responsible for the dramatic increase in spawning activity observed.

Natural reproduction by female brook trout occurred in areas with groundwater upwelling at a rate of 3.4-296.1 m~.rn'~.min-'.This gradient in flow is consistent with previous quantitative measures of flow rates at brook trout redds in Ientic and 10th systems (see Appendix 1). A highly significant difference in flow rates between redds and nonspawning sites located within the spawning area indicates that females were actively choosing sites associated with groundwater flow. Howver, some redd sites were reused, Mile other redd sites were not

used in consecutive years (1994-1 995). Redd sites that were reused had higher average groundwater seepage rates than those that were not reused in 1995

Men compared with 1994. These results indicate that reuse of redd sites may

Vary according to annual variaüon in seepage rates at redd sites. It is possible that female brook trout detect upwelling groundwater by its' chernical composition

(Le. high conductivity), but later spawning females may also use visual cues such as cleared areas from previous spawning aMi.We observed braok trout spawning over a wide range of grave1 substrate sizes as well as on an aggregation of waterfogged sticks and wood chips beside a beaver lodge, as previously observed in another population (Fraser 1982). This variation in spawning substrate suggests the need for groundwater Row at redds is more important than substrate composition in redd site seledon (Witzel&

MacCrimrnon 1983).

Among the salmonines, brook trout are not unique in Meir requirernent of groundwater upwelling at redds. The use of spring seepage areas by Arctic char

(Cunjak et al. 1986), brown trout (Salmo tNfta)((Hansen 1975), rainbow trout

(Oncothynchus mykss) ((Sowden & Power 1985), and sockeye salmon

(Oncorf7ynchus nerka) (Lorenz 8 Eiler 1989) has aiso been documented.

However. these salrnonines also spawn in areas wiaiout groundwater inflow.

Although al1 redds could not be measured, our data. and others (e.g. Curry & Noakes 1995). suggest that laksdwelling brook trout spawn sdely at sites of groundwater upwelling. The relative importance of upwelling groundwater ought to be greater in lake than in strearn systerns, since there is considerably less movement of water in Iakes during the oveMnter incubation period as a result of f reezing (Curry et al. 1995).

Cornpetition for limited sites availabie to spawning salmonines can lead to redd reuse and superimposition of redds (e-g. McNeil 1967). The maximum total area used for incubation by the spawning population of brook trout in Scott Lake was 35 rn2. Our data agrees with observations by Carline (1980) and others that lad< of suitable sites for redd construction is one likely factor limiting brook trout populations. The fact that active fernales are readily replaced once removed from their redd indicates that redd sites are a limiting resource and that certain sites are prefened. Such intense reuse of redd sites is most likely a consequence of limited sites (32 and 47 redd sites in total, 1994 and 1995. respectively) and contributes to the construction of multiple redds by fernales.

It has been suggested that seleetion of redd sites by fernale brook trout is regulated by competition for an opportunity to spawn in a limited area defined by discharging groundwater (Cuny 8 Noakes 1995). We agree that sites are limiting, as shown from redd superimposition data and rernoval experiments.

Hovuever, we argue that females do not compte solely for an opportunity to spawn. but instead, females compete for certain sites and competition is aff ected by the seasonai timing of spawning. If the selection of sites is determined solely by an opportunity to spawn, then we would expect approximatëly equal numbers of spawnings among al1 sites. In contrast. we found preferential use of certain sites during both breeding seasons (Fig. 2.7). Secondly, we bdieve that timing of spawning is a significant component regulating fernale cornpetition for redd sites.

The fact that the majority of spawning (up to 84%) occurs over a brief period suggests that the timing is very important.

Similar to oaier iteroparous salmonine species (0.g. Fukushima 1994) female brodc &out tend to deposit eggs in more than one redd. On average, female brook trout at Scott Lake constnicted 1.6 redds. Fernale Atlantic salrnon

(Salmo salar) and brown trout usually spawn in one redd (Crisp 8 Carling 1989); however, estimates of 8.4 and 5.7 redds per female. respectively. have also been suggested (Barlaup et al. 1994). The large number of redds per female estimated by Barlaup et al. (1994) may be biased since #ey sampled only stranded redds and did not observe spawning females in the field. As with Stream observations

(White 1930). brook &out females generally spamed only once in each egg pit. but multiple spawnings by an individual female can occur in a single egg pit

(Greeley 1932; Hanard 1932; P.J. BlancMield, personal obsewation).

Altemate life history strategies are hown to occur among some salmonines (Gross 1984). The presence of very smail (SI4.0 cm). semally mature males on the spawning grounds may be evidence of an altemate life history strategy for male brook trout. These srnaller fish were always seen hiding

among sticks in the proximity of active redds and were never observed at redds

which were located in open areas. As well, these fish were never observed as

part of the hierarchy fomed by males around females, and their behaviour was

similar to the sneaking behaviour documented for other salmonines (Gross

1985).

The ability to individually identify and obsewe a large proporh'on of a

naturally breeding population throughout a reproductive season is rare in

research on mating systems in general. However. the implications of such data

are that information about aspects of fish reproductive ecology that were

previously assumed may now be detennined in detail. This is especially tnie with

respect to the reproductive ecology of brook trout in lakes about which few

studies ex& Prior research on this subject does not incorporate undenvater

observations of tagged individuals. The ability to fully observe a spawning

population at an individual level allows for detailed measures of spawning

behaviour wiai respect to seasonal timing of breeding, duration on the spawning

area. and site choice. The ratio of male to female brook trout present on the

. spawning grounds at Scott Lake was skewed towards males as a result of

differences between males and females in the timing and duration of spawning.

Temporal and spatial (Le. multi-redd) tactics appear to be an important part of

female spawning behaviour in brook trout. Larger females tend to spawn synchronously and in more redds and over a longer time pend than srnaller fernales. The resulting differences in fitness associated with such behaviour warrant fuither investigation. The authors are especially grateful to Alison Little, Kim Hughes and Kim Mandzy who provided excellent field assistance and often endured unpleasant weather conditions duting the 1993, 1994 and 1995 fall seasons respecbvely. Staff at

HLFR, namely. Greg Betteridge, Doug Brown, Gary Ridout, and Kurt Schryer provided logistical and field support. Sharon Templeton contributed precipitation data. Advice for the removal expriment came from Dr. Bridge4 Stutdibuiy. We are thanMul to Dr. Geoff Power for insightful comments on a previous version of this manuscript, and to an anonymous reviewer. This study was supported by the

Scott Lake Long-Tem Ecological Research Program of the Ontario Ministry of

Natural Resources, and a Natural Sciences and Engineering Research Council of Canada operating grant (M.S.R.). CHAPTER THREE:

MATE SEARCHING BEHAVIOUR: THE ROLE OF MALE BODY SKE AND BREEDING COMPETITION In mating systems where males search for mates but provide no parental care, male reproductive success is detemined not oniy by the availability of mates. but by a males' ability to find a mate and prevent others from gaining access to this mate. Focal sampling of individually identified male brook trout. Salvelinus fontjinalis, sçhowed extensive site visitation that included many repeat visits to sites with females. These observations agree with the importance of a male's being able to predict female readiness to spawn as a significant cornponent of reproductive success. Seasonal mate searching behaviours wre positively related to male body size. Larger males likely have greater movement because they are dominant more often than smaller males upon encountering a female, and therefore have the opportunity to mate with more females. The proportion of observations in which males were moving and dominant, both measures of cornpetitive mate searching. decreased with an increasingly male-biased operational sex ratio, but were not influenced by the absolute number of receptive females. ûverall, these patterns show male size is a good predictor of mate searching behaviour but in addition, the operational sex ratio has the potential to influence space use in species where mate choice and mate searching are important components of reproductive success. INTRODUCTION

Mate choice is usually the final step of a proces that first invdves mate searching and assessrnent. lnterest in semai selection and mating systems has overwhelmingly focused on the aspect of mate choice by males and fernales (see

Andersson 1994). Aecently, this has been followed by a body of theoretical wrk predicting the process of mate searching and assessrnent that leads to mate choice (e.g. Gibson & Langen 1996; Luttbeg 1996; Reid 8 Starnps 1997).

Typicaily, theory is pattemed around females as the searching sex, in part due to the preponderance of leks as a mode1 study system (8.g. Trail & Adams 1989;

Rintamaki et al. 1995; Gibson 1996). While leks do provide the ability to individually identify and observe adive mate search and choice in the field, they are not representative of most other mating system types. Furthemiore, aieory based on fernales as the searching sex is most likely not applicable to situations where males search for mates. The reason for this difference is that mate quality generally detemines fernale reproductive success, whereas mating frequency

&en detemines male reproductive succes (Bateman 1948). This important difference in the factors that confer reproductive success suggests that the tactics of mate searching may also differ significantiy between sexes.

The main factors that affect male mating frequency in breeding systems where males search for mates, will be a males' ability to find a mate and prevent other males from gaining access to this mate. Therefore, mate searching behaviour which allowç for a greater encounter rate with recephe females

(Parker IWO, 1978; Davies & Halliday 1979; Schwagmeyer 1995) and factors that increase a males' acceçs to mates once found, such as large body sire (see

Andersson 1994), will confer a reproductive advantage to males. Current theory does not account for how factors, such as differences in cornpetitive ability, influence mate searching, cornparison and choice.

Temporal variability in the availability of mates can also influence male- male cornpetition in breeding systems (Ernlen & Oring 1977) and may be most pronounced in instances where males provide no parental are. In these casas, the potential rate at which males can reproduce usually far exceeds that of females (Clutton-Brodc & Vincent 1991; Clutton-Br& & Parker 1992). so the operational sex ratio (OSR), defined as the ratio of fertilisable females to çexually active males at any given time (Emlen & Oring 19T7). tends to be biased towards males as a consequence of limited mate availability. The OSR can becorne increasingly male biased when females arrive asynchronously, leading to greater cornpetition among males for females and greater variance in male reproductive success (Grant et al. 1995).

Seasonal reproduction tends to be characterised by a peak in breeding activity (breeding synchrony). Studies across a wide range of taxa have capitalised on this natural change in OSR within a breeding season, as a consequenœ of fernale breeding syndirony, to examine how competition for mates changes with OSR (8.g. anurans: Tejedo 1988: insects: Enders 1993; fishes: Vincent et al. 1994; mammals: Michener & Mclean 1996). Most commonly, studies relating cornpetition for mates with OSR focus on changes in intersemal selection, usualiy in the fom of fernale dioice, or nitrasemial selection such as male-male aggression (e-g. Kvarnemo et al. 1995). While the

OSR has been a reliaMe predictor of aggression and mating success (see review by Kvarnerno & Ahnesjô 1996). it is possible that the OSR may influence other aspects of reproductive behaviour. In mating systems where males search for fernales with which to mate. male movements may reflect changes in the temporal avdlability of females. In other words, the OSR may also be a reliable predictor of how ind~dualssearch for mates. not just the intensity of this competition as expressed in male aggression or mate choice.

I monitored the movements and behaviour of individually marked male brook trout. Seivelinus fonthalis, to examine the process of mate searching and how variation in male body size and OSR affects male mating behaviour. Lake breeding by brook trout provides a unique opportunity to rnonitor the patterns of movement by searching males. Females require groundwater flow at spawning sites (Appendix 1). which are intemiittently dumped along several hundred meters of shoreline (ma*mum density -1 site per 2.5 m2) (see Chapter 2, Fig.

2.1). lnd~dualfemales can prepare a site and spawn in less than 2 h, but often take considerably longer (Chapter 4). Thus, the spatial and temporal distribution of spawning females requires males to rnove among females inTorderto detemine female readiness to spawn and fernale body size (Schroder 1981).

There is exîensive variation in male fork length (-100%), and males compete in dominance hierarchies for spawning access. Males move through the spawning area in search of active females and larger males have the ability to displace smaller males from spawning females (Chapter 4).

The ability to predict female readiness to spawn and gain acceçç to fernales will be strong deteminants of mating succes for male brook trout.

Based on ihese factors, I generate predictions for size-related movement patterns by males. First, larger males rnove on account of their cornpetitive ability. Larger males will have less difficulty establishing dominance at any female encountered relative to other males (Quinn et al. 1996); therefore, their movernents will not be constrained by competitive interactions. Small males, susceptible to take-overs as a consequence of the movements of larger males, subsequently are forced to move in search of new females or to a peripheral position at the same female. The net result of the dependence of large and small male movements suggests that there could be no size-related patterns of searching behaviour by male brook trout.

The competitive abilities of larger males are, in part, based on their lower energetic costs of rnovement. Previous work has show that reproductive costs are higher among small males in other populations of brook trout (Hutchings

1994) and these greater energetic costs for small males are likeiy also related to a greater probability of physical injury (Hutchings 8 Myers 1987). Together these data suggest Viat male body size might constrain energetically costly cornpetitive mate searching behaviour (Kitano 1996). If male mate searching behaviour is a result of sizebased energetic constraints. then seasonal patterns of space use and movements should reflect this trend.

Seasonal variation in the anival of breeding fernales ont0 the spawning grounds allows for cornparison of the influence of OSR on male mate searching behaviour. The effect of OSR on male mate searching is sornewhat less easy to predict. However, because these seasonal changes in OSR are mediated by female breeding synchrony, I used the 'syndirony hypothesis' (Bryant & Grant

1995: Lindstr6m & Seppi3 1996) to make general predictions of space use by spavming males. This hypothesis predicts that during periods of low breeding synchrony (quivalent to a high OSR) males wili try to monopolise mates, whereas periods of high breeding synchrony (equivalent to a low OSR) should result in scramble competition among males. Thus, the synchrony hypothesis predicts that male movement will increase with decreasing OSR (i.9. less competition for mates). individually marked and followed the reproductive behaviour of most indMduals

(-85%) in a population of brook trout at Scott Lake, Algonquin Provincial Park,

Ontario, Canada (45O2WN, 78O437N) on a daiiy basis over two breeding seasons. Fish were captured prior to spawning using trapnets, and thereafter with dip nets by swimmers. Fish were anaesthetised prior to tagging and measuring. Fork lengths (FL; to the nearest mm) were recorded for al1 fish (for complete details of the study site and tagging procedure see Methods, Chapter

2)-

Male mate searching behaviour

I followed the movements of marked male brook trout for periods of up to

30 min. To examine male movement patterns, I chose males that were initially observeci moving between spawning sites. One assumption implicit to this wrk is that al1 male rnovement is a result of searching for mates. The predicüon that small males move as a result of take-overs still assumes a choice to move rather than to remain at a given spawning site as a peripherel male. During each observation of male movements the position and activities of the focal male, the time spent at specific sites and the presence and activity of spawning fernales at those sites were recorded on undenvater dates. Females were considerd active if behaviours assoclSSOClat8dwith nest site construction, egg deposition, or the covering of eggs were observed (Tautz & Groot 1975). 1 recorded whether movement between spawning sites was preceded by a male losing an agonistic encounter (chases, attacks and lateral displays) at the previous spawning site. In total 59 observations of male movements were conducted; howver, males that I was not able to obsenre for periods of about 30 min (Le. usually lost) or showed no movement between sites were excludeci from the analysis. Also, I induded only one observation period for a male that I had repeat observations on during the season. The mate searching observations of 20 males met the above criteria.

Observations of male movements were conducted over a period of peak breeding activity (26 Oct. - 9 Nov. 1995; see below).

Seasonal patterns of male movements and behaviour

I conducted a census of the entire spawning area four times daiiy (weaaier permitüng) from the first day of observed fernale spawning actMty until fish were no longer preçent on the spawning grounds or the lake surface was from (10

W.- 6 Dec. 1994; 10 Oct. - 24 Nov. 1995). 1 calculated the area individual males covered during the breeding season (referred to as the spawning range) based on their location during each census of the main spawning area. Males were occasionally observed at a distant spawning site (-350 m from the main spawning area; see Chapter 2, Fig. 2.1); however, because spawning was never considerably inflate areal estirnates (Wray et al. 1992). As well, I calculated the linear distance that males moved behNeen successive observations as another estimate of male mate searching actMty.

From census swims, I also calculated the percent of observations in which individual males were: 1) moving - observeci several times dunng the same census, or obseived between spawning sites but without a female; and 2) with an active female - they were either dominant or satellite in their position with respect to their proximity and defence of the female. Males wre considered dominant when they were closest to a given female and denying other males access to her

(Chapter 4). The percent of observations that males were dominant was calculated using only the observations when present with an active female (to represent mating cornpetition). I used linear regression to describe the relationships be-n seasonal spawning range, movement and dominance with respect to male body size (logetransformed). Spawning range was log. transformed and observational data presented as proportions (movement and dominance) wre arcsine transfomied. ANCOVA was used to detemine between year differences with respect to male body size. When betiiuieen year differences were not significant, analyses were conducted on combined data from males in both years. Bonferroni corrections were applied to these analyses. Figure 3.1 Map of study site showing the wre area of use by a male brook trout throughout one breeding season. Space use (Le. area) was hlculated as the sum of al1 polygons determined by Dirichlet tesselation (see inset for detailed view). The operational sex ratio

I calwlated the OSR during each census swirn as a perdnt of the number of active individuals; OSR=(the number of males around active females /

(number of males around active females + the number of active females)) x 100

(Kvarnemo & Ahnesj6 19%). thus excluding those males and fernales present on the spawning grounds that were not active. For males and fernales that were observed more than once duting a census swim (Le. moving, as defined above),

I induded only their initial observation with an active female (if any) in my calculation of OSR. Due to the nurnber of observations required to construct spawning ranges for male movernents, I was unable to treat the daily estimates of OSR as a continuous variable. lnçtead, to examine the relationship between

OSR and mate searching, I di~dedthe breeding season into three periods corresponding to levels of breeding actMty. In each year there was an obvious period of peak breeding activity, which I used to compare activity pdor to (early) and after (late) this period (Chapter 2, Fig. 2.5). 1 calculated the mean OSR for early (1994: 11 -22 Odober; 1995: 13-22 October), peak (1994: 23 OcZober - 6

November; 1995: 23 October - 8 November) and late periods (1 994: 7 November

- 1 December; 1995: 9-24 November).

Among males, the duration of time spent on the spawning ground increases with body ske (Chapter 2). Therefore, to control for differences in male condition which may occur as a result of diifferences in duration on the spawning grounds, I included only males that I had observed a minimum of 10 tirnes throughout each of the early, peak and late periods of the breediiigseason

(1994: Ak18; 1995: k14). Observations of these males were used to detemine aie size- related male patterns of seasonal movements described above. The re were no differences in male body size (FL) between years (1994: 39.1 i 1.7 cm;

1995: 35.8 + 1.8 cm; b1.3. ENS). Means (+ SE) are presented.

RESULTS

Male mate searching behaviour

Male brook trout showed considerable movement among spawning sites as well as frequent revisitation of sites. An example of one male's visits to spavming sites and females during a 30 min observation period is shown in Figure 3.2.

Males encountered females (mean: 12; range: 1-28) on approximateiy haif of their total visits to spawning sites (mean: 25; range: 12-43). However, these numbers inciude revisitation to sites, sudi that the number of different spawning sites visited (mm: 11 ; range: 5-20) and distinct females encountered (mean: 5; range: 1-11) are lower. I compared the number of different females visited by males during the observation period to the number of available females.

Available fernales were those that were present at spawning sites during the

census swim nearest in time to the male observation period. Males visited less

than one-third of the available females (29 I4%) dunng a 30 min observation

period, wtiich was significantly fewer fernales (5.1 I0.6) than were available

(18.5 t 1.1 ; Paired &test: b1 1.2. PcO.000001).

Revisitation to spawning sites was frequent, with sites being visited, on

average, more than twice (2.5 t 0.2 times) during the observation period (see

Fig. 3.2). 1 tested whether there was any ditference in the frequency by which

males revisited sites with and without females present. Males did not visit

spawning sites without females present (2.0 t 0.2 times) with any less frequency

than sites with females (2.6 * 0.2 times; Paired t-test: r;-1.6, %NS). However,

because male visitation to unoccupied spawning sites may have been influenced

by the recent presence of a female. I further categorised visits to unoccupied sites into ones that females had never been observed at during the observation period. Male frequency of visitation to these sites was signlicantly lower (1.5 t

0.2 times; Paired 'test: 1=-3.0, P4.01) than to sites where females were present

(Fig. 3.3).

I tested whether body size influenced the patterns of mate searching observed during the focal male observation periods. The proportion of al1 visits to a different spawning site that were preceded by a male losing an agonistic encounter (chases. ettacks and lateral displays) at the previous spawning site decreased significantly with male body size (Table 3.1). The proportion of Visitation

Figure 3.3 The frequency by which searching males (MO)made repeated visits to unoccupied sites (open bars) and females (shaded bars) during 30 min observation periods. Unoccupied sites refer to spawning sites at which females had never been obsenred during the obsewation period. Table 3.1 The relationship (Iinear regression) betwen body size (FL) and various mate searching behavioun for male brook trout followed for 30 min (MO). For calculation of the proportion of females Msited see text.

Mean (SE) i? F pt

Proportion of: Females visited 0.29 (0.04) 0.09 1.7 NS

Tirne spent (min): All females 15.5 (1.8) 0.04 0.7 NS

Active fernales 10.6 (1.4) 0.09 1.8 NS

Dominant 5.3 (1 -2) 0.59 24.5 4.001

Othe propottion of movement to a different spawning site that was preceded by a male losing an agonistic encounter (chases, attacks and lateral displays). %onferroni corrections used. different females visited during male observations (females viçited versus those available from census swim) was not influenced by male body size. There was no relationship between body size and the amount of tirne males spent with al1 females at spawning sites or with active females (Table 3.1). The amount of time spent as the most dominant male in the presence of active fernales was positively and significantly related to male body size.

Seasonal patterns of male movements and behaviour

There were signifiant size-related trends in space use and movement patterns by male brook trout. The spawning range of males increased with body size (pd.27; fi.~I1.3; P<0.01) (Fig. 3.4a), as did the average linear distance between successive observations (A2=0.26;Fl,-l 0.3; Pcû.01). Males were observed, on average, a linear distance of 30.0 m (f 2.3 m) between successive observations. The proportion of observations in which males were seen moving differed between years (64.3; P<0.05), but only in 1995 was there a significant increase in movernent with male body size (1994: p4.24; Fi,ls=5.0;

A0.013; 1995: ed.48; Fl,rr=11.3; Pe0.01) (Fig. 3.4b). Of the census swims in which male brook trout were associated with active females, the proportion of obsewations in which males were dominant increased significantly with male body size (@=0.69, FI,+--66.5, Pe0.000 1) (Fig. 3.4~). 0 * oc ci- E 4500 *. 9 w 1

Male fork length (cm)

Figure 3.4 The (a) spawning range of male brook trout, and percent of observations (b) moving and (c) dominant. increased with male fork length for 1994 (open circles) and 1 995 (solid circles). The operanional sex ratio

Breeding seasons were characterised by a marked period of 15-17 days that accounted for 58% and 84% of all spawning in 1994 and 1995, respectively

(referred to as the peak period; see also Chapter 2, Fig. 2.5). Very little spawning occurred prior to the peak date (early; 1994: 6%; 1995: 8%) and after the peak

(late; t 995: 8%), except during the late pefiod of 1994 (which was much longer than in 1995 due to warmer weather), where intermediate levels of spawning occurred (36%) (Fig. 3.5a).

Female breeding synchrony (Fig. 3.5b) influenced the OSR. Early in the season the presence of many males but few active females elevated the OSR

(1994: 81% males I1.1; 1995: 84% males I 0.6). During the peak pend the

OSR decreased due to synchronous breeding by females (1994: 75% males t

0.9; 1995: 71% males I 0.8). The OSR continued to decline after the peak period as a result of males leaving the spawning grounds at a higher rate than new females were recruited onto the spawning area (1994: 69% males I 1.3; 1995:

68% males +_ 1.8) (Fig. 3.5b; see also Chapter 2, Fig. 2.2). The daily maximum number of active females, males associated with active females and total males present on the spawning grounds were similar during aie early and late periods, and significantly lower than during the peak period during both breeding seaçons

(Table 3.2). Howw~~,in 1995, there were significantly fewer males present on Early Peak Late

OU 12 Oct 22 Nov 1 Nov 11 Nov 21 Dsc 1 Date

Figure 3.5 (a) Breeding synchrony was measured as the proportion of total spawnings occurring in each of three periods, early, peak and late for 1994 (open bars; k173spawnings) and 1995 (solid bars; k183 spawnings). Numben above bars represent the census swims during that period. (b) Daily estimates of operational sex ratio (mean f SE) based on census swims varied throughout the 1994 (open cirdes) and 1995 (solid cirdes) breeding seasons. An average operational sex ratio was determined for each period. Table 3.2 Seasonal differences in the number and actMty of male and female brook trout in relation to the breeding peflod. Data are presented as rneans (SE). For exact dates corresponding to each period see text.

Year Early Peak Late F

Number of days

Daily maximum: Active females

Males with active females

Total Males

Within a row, values with different superscripts diiered significantly (Tukey HSD posthoc test following ANOVA). the spawning grounds late in the season wmpared to eariy and peak periods

(Table 3.2).

Patterns of space use (i.e. spawning range) by male brook trout during early, peak and late penods did not correspond to seasonal differences in the

OSR (Table 3.3; Fig. 3.6a). However, the proportion of observations in which males were seen moving and as dominant are negatively associated with seasonal dierences in the OSR (Table 3.3; Fig. 3.6bPc).Conversely, the seasonal trends of male movement and mating opponunity (Le. dominance) do not correspond to seasonal patterns of female availability (see Tables 3.2, 3.3). Table 3.3 Seasonai differences in male movement patterns and behaviour in relation ta breeding period. For relationships between the operational sex ratio see Figure 3.6. Data are presented as means (SE). For exact dates corresponding to each period see text.

Period

Year Early Peak Late F P

Spawning range (m2) 1994 5393a (574) 1995 5519 (1 573)

Percent of observations: Moving 1994 7.8a (1 -5) 1995 12.ga (2-7) Dominant

Within a row, values with different superscripts differed significantly (LSD posthoc test for non-independent sarnples following repeated rneasures ANOVA). Operational sex ratio (% males)

Figure 3.6 The (a) spawning range of male brook trout, and percent of observations (b) moving, and (c) dominant in relation to the operational sex ratio. Means (ISE) and period (e=early; p=k; I=late) are given for 1994 (open cirdes) and 1995 (solid circles).

79 Large body size confers a cornpetitive advantage to male brook trout (Sahelinus fontinaiis) searching for mates. Although there was movement among spawning sites by al1 males, rnovement following adverse aggressive interactions was more cornmon for smaller males. ûver the course of the breeding season larger males moved among fernales more frequently, moved greater distances and had larger spawning ranges. The use of space among male brook trout (i.e. spawning range and percentage of observations moving) and their frequency in a position of dominance, was strongly related to male size. These results, and similar size- related male movements in other taxa (Borgia 1980; Otronen 1993), support the idea that male body size is important in mating systems characteriseci by males of greatly vaiying size searching for spatially distributeci females (but see

ûtronen 1995).

Searching male brook trout commonly made repeat visits to spawning sites and females. Current theory accounts for repeated visits to mates by the searching sex as a comparative tadic (Real 1990; Reid & Starnps 1997) or as a result of rece~ngimperfect information about mate quality during previous visits

(Luttbeg 1996). Most likely both of these processes occur when males compete and search for females. In the context of this brook trout mating system an important point to consider is that the spatiai distribution of mates and their readiness to spawn is constantly dianging. Male kno~kdg8af female readiness

to spawn, thergfore, is an important component of reproductive success

(Schwagmeyer 1994. 1995). Male brook trout revisited sites more frequenfly with females present (or sites in which females were recently present) than sites where no females were present. These movements suggest that male brwk trout track female readiness to spawn by way of extensive revisitation, but at the same time determine the spatial distribution of reproducüvely active females through visits to spawing sites.

Male brook trout visited less than a third (-30%) of females available. This may be due, in part, to the Iength of the observation period (30 min). However, mis seems unlikely since males travelled the entire length of the spawning area and more (SIkm) during sirnilar observation periods. An altemate explmation may be due to the availability of females. If the travel üme for males to encounter a new fernale is less than the oestrous period of a female. males are predicted to move between females (Whitehead 1990). Most obsewations of male movements occurred dunng peak breeding activity; a period in which females occupied many of the spawning sites. Thus, males may not have had to move far to find an active female, which may explain why they visited oniy a fraction of available females. Extensive site and female visitation by male brook trout ernphasises the diifference between mate searching tactics when reproductive success is determineci by mating frequency and not necessarily mate quality. Studies (kl1) in which females searched for mates resulted in them visiting approximately four males prior to mating (see Gibson & Langen 1996).

In contrast to the seasonal patterns af male movement. the frequency of site and fernale Visitation did not increase with male body size during the focal sampling period. However, movements as an outcome of competitive interaction may have obscured any sire-related pattern of mate searching behaviour. The frequency by whidi males rnoved to a different spawning site after an adverse aggressive interaction varied widely (0-60°), but was related to body size. Large males rnoved freely with little aggressive niteraction. whereas small males changed sites more frequently after having received an aggressive attack from other males. The net effect of these contrasting movement types may have resulted in the absence of any detectable sizs-related movement patterns.

Mate searching behaviour of large males is not restrlcted by their competitive ability; they will have less difficulty establishing dominance at any female encountered relative to other males (Quinn et al. 1996). Part of this cornpetitive ability cm be based on the lower energetic costs of movernent for larger males. During reproduction, male salmonines primarily lose Iipids from somatic tissue (Jonsson et al. 1991). In dolly varden, Salvelinus malma. somatic weight loss was show to be proportional for large and srnall males even though large males were reproducüvely active longer than small males (Kitano 1996).

For males, the duration of time present on the spawning grounds is related to body size in this population (Chapter 2). As well. the number of census swims in which male brook trout were absent from the spawning grounds (Le. resting;

Quinn et al. 1S6), on days when males were present, decreased with increasing body condition (pdl.35, 616.0, P4.0005). These data suggest that the energetic cost of reproduction is greater for small than large male brook trout

(see also Hutchings 1994). Greater energetic costs for small males are likely also related to more frequent aggressive interactions and a greater possibility of physical injuty (Hutchings 8 Myers 1987). If small males are energy limited during the breeding season, then perhaps they will use alternative tactics when searching for spatially distributed females compared to larger males (Whitehead

1990). Small male brook trout appear to confine their search to smaller areas ttian larger males. Thus, male body size might constrain energetically costly cornpetitive rnate searching behaviour, and in tum. affect male mating success

(Kitano 1996).

A second explanation for male rnovements is related to rnate choice. Mate choice experiments with Pacific salmon show males prefer females equal to or larger than themseives (Foote 1988b). Thus, based on relative body size, smaller males should be less choosy as al1 females are preferred, with the opposite being true for larger males. In this instance, larger males sample a larger area or move greater distances to find a preferred mate (Le. large female) relative to smaller males (Owens & Thompson 1994). Pair formation in this mating system is size-assortative as a consequence of male and female mate choice. indicating

that large males do indeed seek out similady sized mates (Chapter 4). As well,

large males search greater areas than small males through a combination of further and more frequent movements. Consequentiy, cornpetitive mate searching by male brook trout allows larger males to obtain higher quality fernales (i.e. more fecund) but they may have to search further and longer to find such females. lncreasing time and energy costs involved in mate searching are known to reduce mate selectivity in other fish (e.g. stickiebacks Gasterosteus aculeatus) (Milinski & Bakker 1992). The cost of mate searching is higher and the opportunity for gaining successful matings is lower for srnall cornpared to larger male brook trout. Thus, size-related seasonal movements and use of space by males may be a result of both energetic constraints and mate choice criteria

There are few studies of male movement on the spawning grounds for other salmonines. No size-related trends in pair formation or the mean distance moved between observations were observed for lake-spawning sockeye salmon,

Oncorhynchus nerka (Quinn 8 Foote 1994; Quinn et al. 1996). In addition, male sockeye salmon were observed much closer between sightings (2-5 m) and rarely moved a distance of 15 rn to a separate spawning area where females were more abundant (Hendry et al. 1995; Quinn et al. 1996). Stream-spawning male coho salmon, 0.kisutch, show a variety of mate search tactics (Healey &

Prince 1995). The largest males defend a segment of the stream and have a restricted range of movement (50 - 150 m) arnong the females within their

terntory. Smaller males are highly mobile and visit many wideiy dispersed

spawning groups in a given day (O m - >1 km). The smallest males, which exhibit

alternative breeding tactics (precocial or 'jack' males). hid near nests to sneak

spawning opportunities and also had limited movement (O - 100 rn). Schroder

(1981) determined that the number of eggs fertilised by male chum salmon, 0.

keta, searching for large females would not be greater than males that mated

with any female due to lost mating opportunities for choosy males. These studies

suggest that time spent mate searching is costly for males that have a limited

brwding lifespan. The extent of movement by male brook trout presents a strong

divergence from the 'static' impression that characterises most mating systems of

Pacific salmon.

The strategies of searching males in mating systems that preclude the

longrange evaluation of females will depend on time constraints related to lost mating oppominities (Schroder 1981; Sullivan 1994). If perceptual abilities are equal among males. then seasonal patterns of male movement suggest that larger males are able to gather more information or at least update this information more frequently than smaller males by visiting a greater number of fernales (Le. having a greater spawning range and moving more often).

Consequently, large males would be able to more accurately predict the readiness of females to spawn (Schwagmeyer 1995), a factor more important than fernale body size for mate choice in other salmonines (Schroder 1981).

Large males are able to displace srnaller males when their mates are close to spawning (Chapter 4), and would therefore minimise the time invested in each spawning event (Everson & Addicott 1982).

The prediction that OSR. mediated by fernale breeding synchrony, is an important factor influencing the intensity of malmale competition was not entirely supported in this study (Grant et al. 1995). As predicted, high OSR (low breeding synchrony) corresponded to little rnovement as males tried to monopolise mates eady in the season. and low OSR (high breeding synchrony) did relate to scramble competition among males, conesponding to greater movement throughout the breeding area. However, male mate searching behaviours were not conelated with the absolute number of spawning females, suggesting that cornpetition among males, not the availability of fertile fernales, is the most important factor influencing mate searching. While the OSR did not accurately predict male spawning range, male movements and dominance were negatively correlated with an increasingly malebiased OSR. If these patterns were a result of breeding synchrony, we would expct mate searching behaviour during the early and late periods to be sirnilar, since similar numbers of females were available during these two periods (Table 3.2). Instead, male rnovements and dominance were lowest during the eariy periods when the OSR waç greatest

(1:5, fernales:males), but greatest during the late periods when the OSR was considerably less malsbiased (12.5 females:males). Spawning movements by male brook trout and other field studies support the general consensus that the

OSR is an important influence on the movements of males seeking mates (Davis

& Murie 1985; Tejedo 1988; Nelson 1995) and that individuals adjust their reproductive behaviour to the current OSR (Kvamemo et al. 1995).

In condusion, larger males appear to be at a distinct advantage over smaller males in that they are able to search larger areas and have a greater probability of being dominant upon encountering an active female. However, mate searching and mating opportunity (Le. dominance) appear to be influenced by the changing OSR, such that males move Iess during periods of greatest cornpetition. Considering these findings, male size is a good predictor of mate searching behaviour. In addition, the operational sex ratio has the potential to influence spaœ use in spedes where mate choice and mate searching are important components of reproductive success. CHAPTER FOUR:

THE COS1 OF PERIPHERAL MALES IN

A BROOK TROUT MATING SYSTEM ABSTRACT

A focus on the reproductive contributions of males exhibiüng alternative life history tactics has neglecled the role of statusdependent peripheral males in salmonine mating systems. Mating behaviour of brook trout, SalLelMus fontinalis, induding obseivations of spawning, were documented to detemine the mating costs of peripheral males to dominant males (kleptogamy) and fernales (egg cannibalisrn). The mating costs of peripheral males were substantial since more than haIf (56%) of al1 observed spawnings involved peripheral males. Males paired with large females experienced a greater incidence of kleptogamy due to increased numbers of peripheral males present. From patemity studies, I estimated that males which had peripheral males participate in spawning may fertilise, on average, equal numbers of eggs compared to males spawning solely with a smaller female. Fernales which had eggs eaten by peripheral males were paired with relatively smaller males. Latency to spawn by females hcreaçed when paired with a relatively small male, and resulted in females obtaining a larger spawning partner. The observed patterns of sire-assortative mating, kleptogamy and cannibalism are discussed in relation to mate choice for a population of brook trout. INTRODUCTION

Status-dependent selection within mating systems generally results in males that

occupy a 'satellite' or 'peripheral' position relative to that of dominant individuals

(Magnhagen 1992: Gross 1996). Males relegated to peripheral positions (from

here on termed 'peripheral males') can impose a number of fitness costs on

dominant males and fernales that indude parasitizing the effort of larger males by

stealing fertilisations (kleptogamy: e.g. mammals, Clutton-Brock et al. 1979;

arnphibians, Perill et al. 1978; insects. Crespi 1986; fishes, Taborsky 1994,

reducing mating opportunities (van den Berghe et al. 1989), and cannibalising

eggs deposited by fernales (Rohwer 1978; Maekawa & Hino 1990).

The mating systems of salmonine fish (salmon, trout and char) exhibit

various reproductive strategies that have received considerable attention with

respect to kleptogamy (Gross 1984,1986; Taborsky 1994). In this group of fishes,

males compete for access to fernales whom they guard intensely prior to

spawning in an attempt to be the sole spawning male (e.g. Keenleyside & Dupuis

1988). Patemity analyses have shown that males dosest to the fernale at the time of spawning fertilise most of the eggs (Schroder 1981; Chebanov et al. 1983;

Gros et al. in press). Males cornpete for close proximity to females on the &sis of body ske, as well as other traits such as hooked snout length (kype) and hump

height, independent of body ske (Fleming & Gross 1994; Quinn & Foote 1994).

The fitness costs mthin these mating systems have therefore been formulated around the loss of patemity through kleptogamy, largely based on the question of

reproductive contributions from precocial males following an alternative life

history tactic ('pan' in Atlantic salmon, Salmo salar. and 'jacks' in Pacific salrnon,

Oncorhynchus sp.: Jones & King 1952; Hutchings & Myers lm;Groot & Margolis

1991; Foote et al. 1997). The fitness costs associated with peripheral males

following a conditional strategy, where body size determines the position of

individuals relative to a spawning fernale, has received considerably less

attention. The cost of peripheral males for females, based on emer alternative or

conditional strategies. appears to be egg cannibalism (Maekawa & Hino 1987,

1990; Thomaz et al. 1997) but has generally remained undetected in this group

of fishes (Foote 1989).

The objective of ihis study is to determine the costs of peripheral males to

spawning males and females in the mating system of brook trout (Salvelinus

fonthalis). In this dudy I link the cost of peripheral males to male and female

choice of mates. Experimental studies (Foote 1988b; Foote & Larkin 1988) have

revealed that mate choice in salmonines is bas& on relative, not absolute. body

size, such that males prefer mates that are equal to or larger than themselves.

Male preference for females of equal or greater size corresponds to 'threshold'

models of mate choice (Janetos 1980; Gibson 8 Langen 1996) and for salmonines this threshold level varies depending on male size (Foote 19ûûb).

For male salmonines, the cornparison af mates can be a time-consurning procedure; the cost of which may be Iost mating opportunities due to searching (Chapter 3). Thus, a simple threshold-criterion of mate choice cm reduce rnate- searching costs. Sirnilar size-based experiments have notbeefi performed with female salmonines, so there is no indication whether females prefer absoluteiy larger males. However, field or enclosure studies demonstrate aiat females delay spawning when paired with relatively small males (Schroder 1981; Foote & Larkin

1988; Foote 1989). Where a size discrepancy occurs be-n a male and a larger female, delaying tactics by the female appear to result in a more equitable size based pairing be-n spawning fish (Schroâer 1981; Foote 8 Larkin 1988). This

Pr- of displacement of srnall males by larger males presumably amunts for the widespread observation of spawning pairs being roughly equal in size (e.g. sockeye salmon [O. ne&] Hanson & Smith 1967; dolly varden [Salvelinus malma]

Maekawa et al. 1993; Miyabe char [S. m. mjabeij Maekawa et al. 1994). However, this proœss of female choice under natural conditions has yet to be documented.

In salmonine mating systems larger females also tend to be accompanied by greater numbers of males when spawning (e.g. Sargent et al. 1986). The cost to males that prefer to spawn with large fernales is the attraction of many other males to these same fernales. Thus males pairing with large females have greater potentiai for loss of patemity to peripheral males. In addition to these mating costs of peripheral males, field observations suggest that egg cannibalism is also a CO& associated with the presence of peripheral males (Maekawa & Hino 1987). Female

Miyabe char deposl more eggs during their first spawning in a netrelative to later spawnings, in an apparent afiernpt to reduce egg cannibalism by peripheral males (Maekawa 8 Hino 1990).

I link rny predictions regarding the fitness costs associated with petipheral

males to the patterns of mate choice prevalent in rnating systems of salmonine fishes. For males paired with large females, I predid an increasd incidence of

kleptogamy by peripheral males. This is baseâ on the prediction that a posiüve

relationship between female size and the number of perïpheral males will exist

since large females are attractive to a greater number of males based on relative body size. I preûict that females paired with large maies will have a lower incidence of egg cannibalism by peripheral males, and that females actively choose larger mates by increasing Iatency to spawn when accompanied by relativety small males.

METHODS

Field Observations

I individually marked and followed the reproductive behaviour of most individuals in a population of brook trout at Scott Lake, Algonquin Provincial Park, Ontario

(45O2YN. 78O431N) on a daily basis over two breeding seasons (for complete details of the study site and tagging procedure see Methods, Chapter 2). Fork lengths (FL) of al1 fish were measured to the nearest millimetre. I conducted a census of the entire spawning area four times daily (weather pemitting) once spawning activity cornmencd (10 October 1994 and 1995) until fish were no longer present on the spawning grounds (6 December 1994) or the lake surface was frozen (24 Novernber 1995). During each census the position and activity of all fish were recorded on underwater dates by swimrners using drysuits. mask and snorkel. For each census. the operational sex ratio (OSR) was calculated as the mean number of males around active females (this measure excludes males on the spawning grounds that are not associated with active females aven though al1 males are sexually active or ripe) (Chapter 3).

In total, 45 spawning events (involving 39 females and 30 males) were observed by swimmers or recorded on videotape (~ony@Hi-8 videmarnera with

~mphibidunder-water housing). To determine a females' latency to spawn, I recorded the time from the first observation of female activity at a site until the actual time of spawning at that site. If females were not obsenred at the spawning site prior to spawning, then I assumed a time of 1.5 h, since census swirns were roughly every two hours. Likewise, females that took greater than one day to spawn were allotted a time of 10 h because little activity occurs at night (Chapter 2).

I considerd only instances in which peripheral males were observed to gape and quiver (Le. orgasm behaviour asçociated with the release of sperm;

Jones & Ball 1954) in the nest of the female during the spawning event, as spawnings in which there was a potentiai for loss of patemity. A doud of milt expelleci by peripheral males was also distinguishable to the observer (or noticeable on video). Often males wre seen diving into the nest at the moment of spawning, however unless gaping and quivering or çperm release were observed, these males were not considered as potential spawners. Likewise, I labelled spawnings as cannibalistic only when peripheral males were observeci eating eggs. My obseivations may underestirnate cannibalism because it occurred in the video recordings up to 30 min after spawning.

Vide0 Analyses

I analysed video records of spawnings (W25)for the frequency and duration of aggressive interactions by the dominant male five minutes imrnediately prior to spawning using The 0bserveP (v. 3.0; Noldus Information

Technology 1995). For males I recorded the following behaviour: 1) cross-over - male remains in close proximity to oie female (ROcm) and prevents acœss to female by other males; 2) non-guard - male away from female (>30cm) and does not prevent access to fernale by other males; 3) chase - male aggressively charges another male (including bites); and, 4) threat - male lateral displays at or attempts to charge other males. I divided the duration and frequency of aggressive behaviour by the number of peripheral males to determine per capita aggression.

I transfomed (log.) al1 body size and frequency of aggression data prior to parametrie analyses to conform to assumptions of homogeneity of variance.

Proportion data wre amine transformed. Means (I SE) are presented. RESULTS

The time dominant males spent in aggressive interactions was strongly infiuenced by the number of peripheral males, such that signifiicantly more per capita chases occurred when one or two males were present than with increaçing numbers of peripheral males (ANOVA: FZl5=7.1, Pe0.01) (Table 4.1 ).

There were no differences in per capita threat displays with the number of peripheral males.

The number of peripheral males releasing spem during spawnings increased with the number of penpheral males around active females (rs=û.61,

H2,P<0.00001) (Fig. 4.1 a) for spawnings in wh ich peripheral males were present. However. the proportion of peripheral males spawning did not Vary in relation to the number of penpheral males present (rs=0.0085,M2, NS) (Fig.

4.1 b). Aimost half (49%) of al1 spawnings with less than four peripheral males present had peripheral males participate in the spawning, whereas al1 (100%) spawnings with five or more peripheral males present were subject to peripheral males releasing sperm.

More than haif (56%) of al1 obseived spawnings involved kleptogamy or egg predation by peripheral males. The relative body size of a spawning pair appears to play a role in whether peripheral males participate in spawning events. Males were significantiy larger than their mates (5.0 + 1.3 cm; Wilcoxon matched pairs test: 2=3.1, Pcû.01) only for spawnings in which peripheral males Table 4.1 Cornparison of mean per cap& aggression (SE) by the dominant male five minutes prior to spawning in relation to the number of peripheral males present.

Peripheral males

Per capita aggression (5 min observation period):

Chase

Threat

Means with different superscript letters are significantiy different (ANOVA and Tu key post-hoc tests). Proportion of Number of peripheral males spawning peripheral males spawning O O O O O O hr P b ko I 1 I 1 t did not participate in kieptogarny or egg predation (mg. 4.2). Peripheral males

were obseived to release sperm in 47% (21145) of the witnessed spawning

events, with one-third of these spawnings (7/21) involving both spem release

and egg predation by peripheral males (Fig. 4.2). There was no difference in the

ske of the dominant male or difference in pair size for spaininings in which

peripheral males participated versus thos8 spawnings in which the dominant

male was the only one to release sperm. However, spawnings in which

peripheral males were obseived to release spem involved signficantly larger

females and greater numbers of peripheral males (Table 4.2). The size of male

aggregations in which peripheral males participated during spawning (4.1 i 0.40)

was greater than the average size of male aggregations in the spawning area on

that day (i.e. OSR; 3.1 r 0.24; Paired fitest: -3.1, Pcû.01). In contrast. the

aggregation size of males which had no peripheral males releasing spem (2.0 I

0.27) was less than the average size of male aggregations in the spawning area

on that day (3.0 I0.20; Paired &test: hd.6, Pe0.01). For spawnings in which

peripheral males were absent, fernales were significantiy srnaller (34.4 i 1.2 cm,

k13; 63.1. PeO.01) than females with peripheral males present (40.3 I1.1 cm.

#=32).

Egg predation by peripheral males occurred in almost onequarter (24%)

of witnessed spawning events. For spawnings in which there was no egg

predation, female body size and the number of peripheral males present did not differ compared to spawnings in Mich penpheral males participateci in egg Pair Cannibalism Kleptogamy Kleptogamy & Cannibalism

Figure 4.2 The difference in pair size (male FL - female FL; Xand SE) was compared for spawnings which involved only the spawning pair versus those with peripheral male participation (sperm release only, egg predation only and both çpem release and egg predaüon). Cabgories for which differences in pair size were significant are marked (~~0.05;Wilcoxon matched pairs test).

predation. However, spawnings in which penpheral males participateci in egg

cannibalism involved significantly smaller dominant males, resulting in a

significant difference in pair size (Table 4.2).

Assortative mating was apparent (@=0.28, Fi,a=l 6.1, Pc0.001) (Fig.

4.3a) although males were larger than their mates (Paired t-test: &=2.51,

Pç0.05). The number of males around active female brook trout increased with

female body size (pa.26, F1,m=1 5.0. Pc0.001) (Fig. 4.3b). The number of

penpheral males present around active females did not change during the hours

leading up to spawning (r,-0.14, =7, ENS).

Females which were at one time paired with a relatively small male, over

the course of spawning activity. took a longer time from the start to cornpletion of

spawning (7.2 + 0.9 h) than did females alwayç paired with males of equal or

greater size (2.7 f 0.7 h; &=3.9, Pc0.001). As well, females paired with males

smaller than themselves were larger (41.O f 1-5 cm) than females paired with

males of equal or greater size (36.5 f 1.1 cm; &=2.4, R0.05). I examined to what extent latency to spawn by females was a result of absolute fernale size or relative size of the spawning pair (Le. mate choice). Only the dierence bet\iifeen female boây size and the body size of the male she was originally pair& with resulted in a significant increase in latency to spawn (Forward step-wise multiple regression partial correlation: H.38, P<0.01). Female fork length (cm)

Figure 4.3 The relationship between female body size and (a) the size of dominant male present, and (b) the number of males around females at the time of spawning (N=45). The lines represent the best fit obtained with Model II regression (size of dominant male: Y= 1.1 3x-2.58; size of male aggregation: Y=O926x-7.07). I compared the size of the dominant male present during census swims prior to spawning with the size of the dorninant male present durhg spawning.

The dominant male present four to six hours prior to spawning was, on average,

5.8 f 3.5 cm smaller than the dominant male at spawning (Fig. 4.4). The dÏfference in size between early males observed as dominant during census swims and the dominant male at spawning decreased as the time to spawning came closer (r,=-0.29, M8, Pe0.05) (Fig. 4.4), thus the size of the dominant male increased as spawning time approached. Most females (80%) that were paired with a relatively small male during the first observation of spawning activity eventüally spawned with a male larger than this original male.

DISCUSSION

Male and female brook trout incurred mating costs through kleptogamy and egg cannibalisrn, respectively, in the presence of statusdependent peripheral males.

For males, the potential for kleptogamy &y peripheral males was great with peripheral males releasing spem in neariy haîf (47%) of al1 spawnings. Kieptogamy was a resuit of large females attracting a relatively large aggregation of peripheral males. Female brook trout delayed spawning when paired with relatively small males, and in doing so obtained larger mates. It appears that the choice of relaüvely large males by females is related to a reduced incidence of egg Hours prior to spawning

Figure 4.4 The difference between the mean body size of dominant males (f SE) present during transect swims prior to spawning and the dominant male present at spawning (represented as the dashed iine passing through zero). cannibalism. Alaiough 1 was not able to quant@ the exact proportion of eggs

eaten by peripheral males. brood cannibalism was cornmonly observed in this

population, and is likely the cost driving female choice for larger males.

Natural selection favours large body size for egg production in female

salmonines (van den Berghe & Gross 1989; Fleming & Gross 1994). Because female quality (fecundity and egg biomass) increases with body size, males mating with large females have the potential to sire many offspring. Thus, large male brook trout face a paradox when mating; they prefer to spawn with similar- sized (Le. large) females. However, large females attract rnany males which dominant males are unable to defend against (Quinn et al. 1996). thereby increasing the opportunity for loss of fertilisation. Chases by dominant males, which are the most effective way to keep peripheral males away from the spawning fernale, decreased with increasing numbers of peripheral males.

Interestingly, iess than hatf (37%) of ail peripheral males present release sperm at the moment of spawning, and the probability of a peripheral male releasing sperm did not diifer significantly among groups with varying numbers of peripheral males present.

Why then, do large males seek out large females for spawnings in which they will most likely lose partial patemity' I provide a simple cost-benefit analysis of male rnating success with respect to the number of eggs fertilised. From fecundity estirnates of wild brook trout (Vladykov 1956). 1 calculated the mean number of eggs available to males paired with large females based on the size of these females (41.1 cm from Table 4.2; Line A, FÏg. 4.5). 1 then calculated the mean nurnber of eggs available to males whidi did not have peripheral males release spen during spawning based on the size of fernales to which these males were paired (36.5 cm from Table 4.2; Une B. Fig. 4.5). While I do not know the exact proportion of eggs fertilised by peripheral males, a review of salmonine patemity studies revealed that dominant males, on average, fertilise roughly threequaiters (72%) of a given fernales' eggs during a spawning event in which peripheral males participate. Assuming a 17039%loss of patemity (k 95% confidence intervals from published studies; see arrows on Fig. 4.5). aien males in this study which had the potential to los8 patemity may fertilise, on average, an quivalent number of eggs if they had solely fertilised the eggs of a smaller female (Fig. 4.5). Therefore, males paired with large females have the opportunity to sire many offspring, although peripheral males may reduce this rneasure of success due to an increased incidence of kleptogamy. It is important to note, however, that male mating success ought not to be considered solely in ternis of th8 number of eggs fertilised. Other studies have show that over and above greater fecundity, male salrnonines mating with large fernales enjoy greater reproductive success by producing larger young (Ferguson et al. 1995;

Hayashizaki et al. 1995) and having a greater proportion of these young survive

(see van den Berghe & Gros 1984,1986,1989; Fleming 8 Gross 1994).

Therefore benefits accrued by males spawning with large females may be greater than realised, even if the absolute number of eggs fertiliseci are reduced Female fork length (cm)

Figure 4.5 Fecundity (Le. number of eggs) of female brook trout increases exponentially with body size (from Vladykov 1956). The potential fertilisation success of dominant males which did (Line A) or did not (Line B) have peripheral males participate in spawning is based on the mean size of females with which they were paired (see Table 4.2). Males paired with larger females (Line A) sMer a 17.39% loss of patemity (anows) to peripheral males. Loss of patemity (giey area) was calculated as the 95% confidence intervals from patemity studies (success of the dominant male - Schroder 198 1: 71 -7%; Chebanov et al. 1983: 71 -2%; Hutchings & Myers 1988: 77.0%; Jordan & Youngson 1992: 892%; Foote et al. 1997: 57.8%; G ross et al. in press: 65.0%)- through kleptogamy by peripheral males.

Large body size andlor other çecondary semal characters used in mating

competition among male salmonines leads to greater access to females (Fleming

& Gross 1994; Quinn & Foote 1994; Kitano 1996). Male body size in brook trout

corresponds to greater mating opportunity in this system. The ex&entof the

spawning area covered as well as the frequency of observations moving and as

dominant increase with male body size (Chapter 3). The patterns of cornpetitive

mate searching and pair formation I observed agree with observations of Pacific

salmon which show that male mate choiœ is dependent on female readiness to

spawn and female size (Schroder 1981; Quinn et al. 1996). 1 agree that an

association between female size and male aggregation size is not, in itself, direct

evidence of male mate choice (Foote 1988a). However, aie fact that male

aggregation size remained constant over the course of spawning suggests that

large females attract more males than small females. and that aggregation size is

not a product of passive accumulation of males around females.

Female mate choice plays an important role in detemining the observed patterns of egg predation in this population of brook trout. Experimental studies have shown that females prefer mates equal to or greater in body size than themsefves (Foote 1989). For females, which remain site-attacheci, this choice is expressed through aggression towards the dominant male (Kiiano 1996). a delay in site preparation (Schroder 1981; Foote & Larkin 1988), or the deposition of fewer eggs (Foote 1989) when paired Ath a relatively small male. By delaying site preparation, females paired with relatively small males increase their likelihood of attracting a larger male with which to spawn (Foote & Larkin 1988).

My results are the first to confirm that female delaying tactics in salmonines do indeed result in females gaining a larger mate. In addition, these findings corne from observing reproduction under natural conditions and therefore lend strong support ?O the role of female mate choice in prornoting size-assortative mating that results in spawning fernales and males of equivalent size. This pattern of spawning pairs of equivalent size is commonly observed among salmonines

(Hanson & Smith 1967; Schroder 1981; Foote & Larkin 1988; Maekawa et al.

1994; this study).

There is a growing body of evidence linking female preference for large males with greater survival of eggs in fish where males make a significant parental investment. In these cases male parental care leads to a greater hatching success of eggs due to the ability of large males to defend eggs from predators (Downhower & Brown 1980; Bisazza & Marconato 1988; Côte & Hunte

1989). Uniike Pacific salmon, which cease feeding during the breeding season, this brook trout mating system is characterised by a high degree of egg predation by peripheral males (24% of spawnings). Although males in this group of fishes are generally not known to provide extended parental care, female preference for relatively large males appears to play a role in detemining which spawnings incurred egg predation. Overall, males were significantly larger than their mates for spawnings in wtiich peripheral males did not consume recently spawned eggs. Thus. for this population of brook trout one proximate mechanism for female choice of relatively larger males may be the apparent reduction in the incidence of brood cannibaiism. In Pacific salmon. female choice of relatively large males is accounted for by a good genes argument and is based on evidence of faster growth rates and greater size at maturity by progeny of large male pink salmon, 0.gorbuscha (Beacham & Murray 1988). However. there is still no evidence for female preference of absolutely versus relatively larger mates in salmonines.

Within comparisons of spawnings with peripheral male participation

(sperm releaçe and egg cannibalism) versus those with no peripheral male participation, I included spawnings in which no peripheral males were present

(13/45). 1 argue that spawnings that occurred when no peripheral males were present represent an active choice by males to avoid certain spawning pairs, and thus should be included in my analysis of the behaviour of peripheral males.

Because male brook trout move and choose among fernales on the spawning grounds (Chapter 3). female body size and the number and size of male competitors will most likely be important deciding factors for male choice of females. Evidence that female body size is an important factor in male mate choice comes from the finding that females were significantly smaller for spawnings in which peripheral males were absent.

In summary. I have shown that peripheral males exert mating costs to males and females. and that these costs are linked to mate choiœ for this population of brook trout. A fow proportion of individual peripheral male participation in spawning and an estimated low gain of patemity by these males, suggest that male rnating success could be more greatly skewed towards large males than previously realised. Female choice may be a strong determinant of male rnating success in salmonine mating systerns and could very well detemine the distribution of condition-dependent mating behaviour among males (Henson

& Wamer 1997). If females spawn more readily in the presenœ of relatively large males. as 1 have shown, then the potential reproductive rate of males may Vary widely (Clutton-Brock & Parker 1992). Obsewations of mate searching would provide much needed data on the choice of females by males of varying size. CHAPTER F1VE:

SIZE-BASED DELAY IN BREEDINGSCHEDULE IN BROOK TROUT:

A REMOVAL EXPERIMENT ABSTRACT

Among salmonines (trout, salmon, char), there is a consistent trend for large females to spawn More small ones within a breeding season. This pattern may be related to cornpetition for limited spawning sites or. alternatively. srnall females may delay reproduction as an adaptive tactic to minimise brood loss through subsequent reuse of nest sites. I tested these two hypotheses by detaining the largest females in a marked population of lakespawning brook trout (Salvelinus fonthalis)).The cornpetiüon hypothesis predicts small fernales would breed earlier in the absence of large females, whereas the adaptive tactic hypothesis predicts delayed breeding. Consistent with the second hypothesis, small females spawned significantly later during the experirnental year (1996) wmpared to th8 previous two years after controlling for among year differences in abiotic factors. Furthemore, brood loss was significantly reduced in the absence of large fernales. Within seasons, brood loswas significantly higher for small than large females for one year. Site choice may also play an important role in reproductive succes. Here I show that small females increase reproductive succesç by delayed reproduction and suggest Mat this adaptive tactic may be socially mediated. INTRODUCTION

A comrnon principle defining Our understanding of breeding syncbrony is that

reproduction is timed for the greatest resource availability to young (e.g. insects:

van Dongen et al. 1997; birds: Martin 1987; fish: Cushing 1969). The strong

selective forces acting on timing of breeding predicts that most individuais in a

population should breed at approximately the same time (Schultz et al. 1991).

However, minthis defined season for reproduction, individuais in animal

populations show predictaMe patterns of variation in the timing of breeding. men

larger or older individuais tend to precede smaller or younger ones within a

breeding season (e.g. marnrnals: Reiter et ai. 1981; birds: Morton 8 Derrickson

1990; fish: Miranda & Muncy 1987). Previously, variation in the timing of breeding

as a result of age- or size-related phenology has been explained by an energetic constraint argument; individuals are constrained to teproduce at a given time

because they must build up suffident energy reserves prior to reproduction (e.g.

Ridgway et al. 1991). Recent contributions on size-related timing of breeding show that, in addition to energetic constraints, smaller ind~dualscan delay breeding as an adaptive tactic to maximise reproductive success (Schultz et al.

1991). Thus. once individuais have acquired suffident energy stores, they may süil delay reproduction because doing so will result in greater reproductive success than by breeding eady.

A different perspective, somewhat removed frorn an energetic constraints argument, is that size-based variation in breeding schedule is socially mediated.

For example, smaller individuals may b8 outcompeted for limiting resources &en associated with breeding (i.8. nesüng sites: Stutchbury & Robertson 1988; mates: Downhower & Brown 1981). Small individuals breed later in a season since their body size places thern at a cornpetitive disadvantage relative to larger indMduals. Altematively, smaller individuals may delay breeding as an adaptive tactic to maximise reproductive success. For instance, small females choosing to breed outside of the time when large indMduals are breeding may result in offspring being sired by better quality mates (Knowfton 1979). Consistent with ideas of mate choice, smaller, inexperienced indïviduals may postpone brding to copy the mate choice(s) of larger (older) ind~duals(Gibson & Hdglund 1992).

80th examples reveal potential situations that could readily result in a size- related trend in reproductive timing irrespective of energetic constraints.

Because fish have indeteminate growth. and fecundity is related to body size (0.g. Vladykov 1956). they are ideal ~0d8kin which to test the role of body size in detemining timing of breeding. This pattern of large females spawning prior to srnaIl females has been observed in a variety of fish species (see Schultz et al. 1991) but is especially apparent among salmonine fishes (salmons, trouts and chars): chinook salmon Oncorhynchus tshawcha (Neilson & Banford

1983); coho salmon 0.kisvtd, (Fleming & Gross 1994); sockeye salmon O. nerka (Quinn & Foote 1994); California golden trout O. mykiss aguebonita

(Knapp & Vredenburg 1996); cutthroat trout O. clarki (Ball& Cope 1961 in Knapp & Vreden burg 1996); brown trout SaIrno trutta (Eiliott 1984); brodc trout

Sahelinus fontindis (Chapter 2); and dolly varden SalLeinus malma (Kitano

1996). Within thiç group of fishes, females dig a series of pits (collectively called a redd) into which eggs are deposited and fertilised by one or more males.

Females then cover aiese pits over and may defend the nest area. Most cornrnoniy, spawning takes place during autumn and young (fiy) ernerge from nests the following spring. It is generally maintained mat timing of spawning occurs at a time that will result in hatching of eggs at an optimal time for survival of fry (Heggberget 1988; Webb & McLay 1996). The consistent trend for large females to precede smaller ones within this group of fishes, as wll as relatively synchronous spawning by large females (0.g. Chapter 2). indicates that there is selection for early spawning. This site-based pattern of seasonal reproduction may help secure better feeding tenitories for offspting in spring and in tum a greater body size (Mason 8 Chapman 1965).

There are several observations characteristic of breeding female salmonines that would suggest sizebased tactics are important. The first observation relevant to this study is that female competition for breeding sites can be intense. One perspective used to define the intensity d breeding competition by salmonines has been to quantify the selection on various morphological traits (see van den ûerghe & Gross 1989; Fleming 8 Gross 1994).

For fernale who saimon, breeding competition accounts for almost half of natural selection, suggesting it is equaliy as important as fecundity in the selection for body size (van den Berghe & Gross 1989). Urger female Pacific salrnon acquire prefened territories under direct cornpetition wiai small females (Foote 1990) and smaller females are more frequently displaced from nest sites than large females

(van den Berghe & Gros 1989). The body size (fork length) of breeding females in a population can vaiy greatly (>100%) and egg burial depth increases with female body size (ûttaway et al. 1981; van den Berghe & Gross 1984; Crisp &

Carling 1989; Kitano & Shimazaki 1995). Furthemore. spawning sites are typically used by more than one female (called superimposition: McNeil 1964; van den Berghe & Gross 1989; Fleming 8 Gross 1994; Quinn 8 Foote 1994;

Knapp & Vredenburg 1996). Thus, small females are at a distinct disadvantage wi?h respect to cornpetition for spawning sites and, in addition, have a greater probability of their nests being destroyed by larger females (van den Berghe &

Gross 1989). The above observations suggest small females spawn later as a direct result of competition for spawning sites (Fleming & Gross 1993, 1994).

Altematively, small females may choose to delay spawning until larger females have completed in order to reduce the chance of having their nests dug up and their broods Io&.

I conducted an expedment involving the detention of roughly haff of the large females (7h 6) in a dosed, lake-spawning population of brook trout to address these hypotheses. Although this experiment focuses on the social mediation of breeding schedule, the nuIl hypothesis, that timing of breeding by small females does not change with the detainment of large fernales, wuld be a direct result of site-based energetic conçtraints. In many fish species, large

individuals are able to spawn earlier in the season because they have less of an

energy deficit wrnpared to small indiiiduals (Schultz et al. 1991). If timing of

spawning by female bruok trout is relateâ to energy reserves, then the

detainment of large females should have no affect on the spawning schedule of

çrnaller-shed indMduals (Ridgway et al. 1991). In contrast, the competition or

'territorial limitation' hypothesis predicts srnall females would breed earlier in the

absence of large fernales (Watson & Moss 1970). Extensive muse of spawning

sites as well as rapid replacement of females during removal experiments has

shown spawning sites can be lirniting for brook trout (Chaptec 2). If site limitation,

as a result of direct competition with larger fernales, causes small females to

postpone spawning, then the detention af large females should result in smaller females spawning earlier than in previous years. Finally, the adaptive tactic hypothesis predicts delayed breeding by small females. Because small females have a greater probability of their broods being destroyed by large females throogh superimposition (van den Berghe 8 Gr- 1989). small females may delay spawning as a tactic to increase reproductive success. tf delayed spawning is a tactic to minimise brood loss then the detention of large females should result in smaller females spawning later than in previous years as they awal large females to reproduce. Most individuals in a population of lake-spawning brook trout wre marked and their reproductive actMties were docurnented on a daily basis during three breeding seasons, IWM996 (for complete details of the study site and tagging procedure see Methods, Chapter 2). Fork lengths (FL) af ail fish were measured to the nearest millimetre. I conducteci a census of the entire spawning area four times daily (weather permitting) once spawning actity commenced (10 Octoôer

1994 and 1995; 15 October 1996) until fish were no longer present on the spawning grounds (6 December 1994) or the lake surface was frozen (24

November 1995; 19 November 1996). During each œnsus the position and activity of all fish were recordeci on undenvater slates by swimmers (mask and snorkel). For those fish which I was unable to mark, or Io& their tag during the breeding season, I estimated fork length (to within 50 mm) based on the size of other marked individuals present.

ûetention of large females

In 1996 1 installed a pen (1.8 m x 1.8m x 2.4 m) made of wire mesh (2.5 cm) prior to any spawning by female brook trout (7 October 1996). The pen was placed approxirnately 4 rn away from the greatest density of spawning sites in the study lake. During the pre-spawning trapnetüng period. I placed the largest females captured into the pen (10.15 October 1996). Large females that I did not catch ni the trapnets were later caught as they commenced spawning act~ties

by swimmers with a dip net or gill net (22-27October 1996). Fernale brook trout

placed in the pen were marked and measured in the same manner as al1 other

fish (see Chapter 2). In total seven fish were placed in the pen (mean: 47.8 cm

FL; range: 42.9-52.3 cm FL) and held until 12 November 1996, menthey were

released.

The experiment, although involving the manipulation of only a few

individuals, was designed to detect a mating system level response. Hence, I

sacrifiœd the power of a multi-lake study for a detailed investigation based on

daily observations needed to detect changes in the timing of spawning.

Factors that may affect the timing of spawning

Prior to an among year comparison of breeding phenology it was necessary to account for several factors mich are known to affect the timing of

reproduction in fishes (see Table 5.1). If larger fish precede smaller ones dunng a breeding season, then spawning populations consisting of larger fish will spawn earlier than populations consisting of smaller fish (Carçcadden et al. 1997). Thus,

I compared the size of breeding females among years (ANOVA). A second factor that may affect the timing of reproduction among years is pre-spawning water temperature (Le. growing season: Coulson 1981 ; Carscadden et al. 1997).

Because I did not record water temperatures outside of the spawning season at

Scott Lake, I compared mean daily air temperatures from a local weather station (-60 km east of Scott Lake) durhg the prespawning period (17 May4 7

September 19944996; ANCOVA). Within a breeding season, the amount of precipitation accumulated and water temperature are thought to affect the timing of spawning (see Chapter 2). Because the amount of precipitation is related to groundwater flow (Downing & Peterka 1978). 1 examined both of these variables.

Precipitation data from Scott Lake was not available for 1996; however, since mean daily precipitation at Scott Lake and the local weather station were highly correlated (1994: ~4.54,P4.000001, M4;1 995: rs=0.49. Pcû.000001, k161). 1 compared mean daily precipitation prior to spawning (September;

ANOVA) from weather station data. I rneasured groundwater flow at spawning sites just prior to breeding at Scott Lake using seepage meters (see Appendix 1;

Ridgway 8 Blanchfield in press for detailed methods) and compared rate of groundwater flow at spawning sites in which measurements wre made in all years (AH4; repeated measures ANOVA). Finally, I cornpared mean daily water temperatures (recorded hourly by six ~emp~entors~;see Chapter 2 for details) from the spawning area at Scott Lake from the start of spawning activity (10

October until end of spawning season). since water temperature may play a role in determinhg the timing of spawning of salmonines (Henderson 1963;

Hokanson et al. 1973; Heggberget 1988). For al1 analyses of covariance, the covariate was day of the year and the independent factors were years (1994-

1996). TaMe 5.1 Factors that may affect the annual timing of spawning by female brook trout were anaiysed. Means (SE) are pcesented

Year Variable 1994 1995 1996 F P

Female FL (cm) 37.2 37.0 38.0 0.44 0.64 (0.96) (0.94) (0.70)

Mean pre-spawning 19.9 21.3 20.9 2.2 0.1 1 daily temperature (OC) (0.47) (0.47) (0.37)

Mean daily spawning 1.8 3-1 2.6 0.19 0.82 precipitation (mm) (0.79) (1 -3) (0.88)

Groundwater flow 37.8 38.6 49.5 0.67 0.52 (m~~rn~~=rnin~') (20.2) (17.8) (29.7)

Spawning water 9.0" 7.gb 7.2' 222.4 <0.0001 temperature (OC) (0.34) (0.47) (0.35)

Values with different superscripts represent means that are significantly different using Tukey HSD test with unequal sample sites following Meir respective analyses (see Methods for details). Female size, nest depth and brood loss

I determined the relationship between female body size and nest depth since one hypothesis is that smaller fernales delay spawning as a size-based tactic to decrease brood loss through nest destruction. Nest depth was measured as the difference between the edge (unexcavated) and deepest part of the excavated spawning site (immediately prior to or after spawning) in relation to the top of the water column for 26 spawnings in 1996 (see DeVries 1997).

By following the daily breeding acüvities of marked females. I determined the number of spawnings that occuned at a specific site, the order in which they occurred and the size of each female spawning. 1 estimated nest destruction similar to van den Berghe and Gros (1989), with the exception that there is lime need to consider overlap of ne& since spawning sites are so site specific in this lake population (Chapter 2). Instead, 100% egg loss was inferred if a subsequent female spawning at a site was of equal or greater length than any previous female. I did not include brood loss caused by the release of the large females for the year of the detention experiment (1996).

1 transformed (loge) al1 data pnor to parametric analyses to conform to assomptions of homogeneity of variance. Means (r SE) are presented. RESULTS

Of the factors that may cause annual differences in the timing of reproduction by spawning fernale brook trout, only water temperature during spawning was significant (Table 5.1). To account for this differenœ among years, I used a thermal sumrnation sirnilar to degree-days in other studies (e-g. Ridgway et al.

1991). 1 chose 15OC as a base temperature (a temperature at which females were present on the spawning grounds in al1 years) from which I subtracted the mean daily water temperature. Thermal summation was calculated as the sum of the absolute value of the difference between mean daily water temperature and

15°C. The thermal surnmation for each spawning was recordeci to compare the breeding phenology of srnall female brook trout (45.0 cm FL) among years

(ANOVA). Small female brook trout showed a delay in timing uf spawning in 1996 wmpared to the two previous breeding seasons (&11=36.7, Pcû.000001) (Fig.

5.1). This trend is perhaps more clearly expresseci as the cumulative proportion of spawings by small fernales among years (Fig. 5.2). During aie year in which large f males were detained from spawn ing (1W6), there were signif icantly greater numbers of fish present than in one of the other spawning years (1994)

(Table 5.2). Proportionally, however, there were significantly fewer active fish present on any given day during the expimental year than during either of the previous two breeding seasons (Table 5.2). Thermal summation

Figure 5.1 The mean spawning time (arrows) for small females was significantly later during the experimental year (shaded) based on thermal sumrnation (see text). Thermal sumrnation

Figure 5.2 Among year comparison of the cumulative proportion of spawnings by small female brook trout W45.0 cm FL). Table 5.2 Annual mean (SE) number of females present. active and proportion active were calculated from the average of daily transect swims.

Year Females 1 994 1995 1 996 F P

Present 12.0" 17.1ab 19.gb 4.4 ~0.05 (1 -1 (1 -8) (1-7) Active

Proportion Active 0.2ga 0.3Ia O.@ 8.5

Values with different superscripts represent means that are significantly different using Tukey HSD test with unequal sample sizes following ANOVA. These results appear consistent with the size-based tactic hypothesis; srnall females postpone breeding to increase some component of reproductive success. Here I examine whether there is a selective advantage for srnall females to delay spawning in response to the detention of large females. Within this population of brook trout there is extensive reuse of spawning sites by females; 78% and 72% of spawning sites were used by more than one female during the 1994 and 1995 breeding seasons, respectively. As well, the body size of breeding females varied by roughly 100% (range: 97-1 13%) within each year

(1994: 24.3-51.5 cm FL; 1995: 26.3-5 1.9 cm FL; 1996: 24.5-52.3 cm FL). Finalty, nest depth increased with female body size (fiS4=10.7,Eû.0032, Pd.31) (Fig.

5.3a) in a manner similar to other salmonines (Fig. 5.3b).

The above factors suggest that small females are at a distinct disadvantage with respect to size-related superimposition. However, small females incurred significantly greater brood loss only during the 1995 bfeeding season (1994: ENS; 1995: P4.0006 [one-tailed test of percentages; StatSoft

19961) (Fig. 5.4). For small brook trout, a significantiy higher incidence of brood loss ocairred during the years when large females were present (1994: 34%.

Ir0.022; 1995: 46%, P4.00065) compared to when they were absent (1996:

23%) (Fig. 5.4). .=O Salmon

Female size (cm) figure 5.3 Nest depth increased significantly with female body size for (a) brook trout (y=-2.991 + 0.0309~).and (b) in a fashion similar to other salmonines (dolly varden: Kitano & Shimazaki 1995; coho salmon: van den Berghe & Gross 1 984 1. Year

Figure 5.4 Within year brood loss through redd superimposition was greater for small (<45.0 cm FL; open bars) than large (245.0 cm FL; closed bars) female brook trout for one year (1995). Among years, srnall females lost significantly fewer broods during the year large females were absent (1 996). DISCUSSION

This expriment addressed two hypotheses for why small female brook trout,

Sahelinus fontinal&, spawn later in the breeding season than large females.

Since small females did not spawn eariier relative to small females in years without manipulation, my reçults do not support the hypothesis that small females spawn after large females due to cornpetition for spawning sites. Instead, delayed breeding durhg the manipulation year provides support for the hypothesis that size-based breeding tactics can increase reproductive success.

Furthemore, I provide evidence that factors other than energetic wnstraints can account for these tactics.

Within the salmonines, small females face several options with respect to site choice and timing of spawning: (1) spawning eariy in high quality territories will increase the likelihood of brood loss through nest destruction by later spawning larger females; (2) spawning early in poor quality temtories can decrease brood loss through nest destruction, but may result in reduced su~val and later emergence of offspring; or (3) spawning later in high quality territories decreases the likelihood of brood loss but will result in later emergence of offspring (se8 Myers 1986). In light of what is known about breeding cornpetition by female salmonines, the findings of this experiment provide new insight into the tradsoffs between timing of breeding and site choice related to brood su~val. Forced postponement of breeding due to competiüon

Sizsrelated cornpetitive asymmetries in competition for breedmg sites

appear to exist among female brook trout, such that smaller females are more frequently displaced from nest sites than large females. If competition for nest sites was responsible for the timing of breeding, the absence af large females on the spawning grounds should promote earlier spawning by smaller females

(Watson 8 Moss 1970). The detainment of large females failed to show that early spawning, larger females excludecl smaller females from nesting sites. In addition, the absence of large females did not advance the timing of breeding for srnaller females. Thus, the territorial limitation hypothesis of Watson 8 Moss

(1970) does not account for the spawning pattern in which large fernales preceded smaller ones for mis population of br& trout. Observations of breeding Pacific salmon provide good evidence that the trend for small females to spawn after larger ones is a result of size-based cornpetitive asymmetry.

Selection studies have found that success in breeding competition by female coho salmon is alrnost equally as important as egg production in the evolution of fernale body size (van den Berghe & Gross 1989). In addition to fernale body sire, caudal-pedunde (tail) depth was also a character select& for in breeding competition, influencing burst swimming used in female-female aggression

(Fleming & Gross 1994). Small (or hatchery) females attempting to breed at the same time as larger females incurred greater displacement from territories (van den ûerghe & Gross 1989; Foote 1990), or suffered greater delays in the onset of breeding, resulting in a decreased ability to spawn al1 eggs before death

(Fleming & Gross 1993, 1994). In contrast to the observations of Pacific salmon, the results of this experiment suggest that the imminent possibility of brood loss through nest destruction by the confinement of large, unspawned female brook trout rnay be responsibk for the delayed spawning by srnaller fish. High variability in territory quality may also account, in part, for the observed breeding delay. If tenitory quality is crucial for brood survival (se8 Chapter 6). then small females may choose to spawn later in the high quality sites used by large females as opposed to breeding early in poor quality habitat.

Sire-specHic tactics in breeding schedule to avoid brood destruction

This lake-spawning population of brook trout provides a real divergence in the patterns of site use that one might expect from the literature on Pacific salmon. Limited spawning sites (-40), the use of multiple spawning sites by females and high variation in site quality, related to groundwater flow, together account for much of the reuse of sites in this population (Chapter 2; Ridgway &

Blanchfield in press). A positive relationship between site quality (rate of groundwater fbw) and site use by females suggest that the reuse of sites is not at al1 a random process (Chapter 6). Extensive site reuse (72.78%) in this population did not result in an equivalent frequency of nest destruction (32938%) due to the spawning of large females preceding smaller females. ln any case, these estimates of nest destruction are arnong the highest reported for salmonines and suggest that cornpetition for spawning sites is intense.

Larger female body site in Pacific salmon allowed access to higher quality spawning territories and the added seleclion on caudal-peduncle depth influenced nest suwival due to increased egg buriai depth (Foote 1990; Fleming

& Gross 1994). Oespite this fact, srnaller female coho salmon did not attempt to avoid cornpetition by delaying breeding (van den Berghe & Gross 1989).

Intereçtingly, destruction of nests in this coho çalmon study (18029%) is much greater than those reported for other studies of salrnonines in which large females were observed to breed before smaller mes (coho: 8%. Fleming 8

Gross 1994; sockeye [O. neka]: 11 %, Quinn & Foote 1994; California golden trout [O. mykss aguabonita]:4%, Knapp & Vredenburg 1996). Fleming and

Gross (1994) propose that a shorter duration of spawning activity (and thus nest guarding by females) and lovuer variation in territory quality in their enclosure experiments may account for this apparent dichotomy in site reuse among studies of coho salrnon.

Greater numbers of spawning fish did not result in a delayed onset of breeding or significantly higher levels of nest destruction for a study of Pacific salmon (Fieming & Gross 1993). Significantly higher rates of nest destruction for small versus large fernale brook trout occurred during a year of greater spawning synchrony (1995). This result is in contrast to Pacific salmon in which pst- spawning nest defence by females may lead to less opportunity for site reuse during shorter breeding seasons (Fleming & Gross 1994). The use of multiple spawning sites in conjunction with lirnited site defence may allow for greater nest destruction in shorter versus prolonged breeding çeasons in brook trout. The lower incidence of nest destruction during the year when most large females were confined lends strong evidenœ that these individuals pose a severe threat of brood loss to smaller females spawning in the same sites. Furthemore. these results suggest that adaptive delaying tactics can minimise brood loss.

Breeding competition for spawning sites, tenitory quality and nest destruction appear to be factors shaping the timing of reproduction in salmon ines. Reproduction by lake t rout, Sahelinus namaycush, provides an example of a salmonine mating systern in which there is no competition for spawning sites and no egg burial by females. For breeding lake trout, smaller females precede larger ones on the spawning grounds (Martin 1957). This pattern has also been observed in Atlantic cod, Gadus rnohua, and may be the consequence of a delay in maturation for larger individuals related to a longer penod required for size-independent production of gonadal tissue (Hutchings &

Myers 1993).

The later spawning tirne observed for brook trout during the year large females were detained does not appear to be a result of among year differences in environmental conditions. The influence of water temperature on timing of spawning is thought to be relatively unimportant for salmonines, as long as temperatures are within preferred ranges (Bye 1984). The differences in water temperature observed among breeding seasons should not result in different overall rates of gonadal development in brook trout (Henderson 1963). However, the use of thermal sumrnation to account for significant differences in water temperature among years is justified by the fact that within and arnong populations, spawning times are significantly later and more protracted in warmer versus wlder conditions (0.g. Heggberget 1988; Webb & McLay 1996). The timing of spawning at Scott Lake reflects these generalities; spawning was more protracted dunng a wamer (1994) than colder (1995) breeding season (see

Chapter 2; Fig. 2.2). Thus, based on the differences in pattern of breeding and water temperature between these two years, one wouid expect the timing of breeding in 1996 to most closely follow that of 1995, in the absence of any manipulation of females.

Variation in reproductive success based on size- related spawning phenoiogy demonstrates that delays in breeding schedule can be explained by a combination of energetic constraints and adaptive tactics that increase reproductive success for smaller individuais. In the surfperch, Micromettus mhhus, delays in conception by small females resulted in the production of larger broods (Schultz et al. 1991). For srnallmouth bass, Micropterus dolumieu, delays in nesting by small males may lead to greater lifetime reproductive success (Ridgway, unpublished data). For this population of brook trout, the detainment of several large females had a dramatic effect on the timing of breeding for smaller females. Small fernale brook trout postponed spawning in response to the presence of large, unspawned fernales, presumably as a tactic to maximise egg survival. The tradeoff between site choice bemnearly and late spawning small individuals and estimates of reproductive success in relation to territory quality are future directions for research. CHAPTER SIX:

A RESOURCE-BASED MATING SYSTEM I quantifid the relationship between resource quality (Le. rate of groundwater

flow at spawning sites) and offspring survival for a population of lakespawning

brook trout, Saivelinus fontinalis The few spawning sites used by females in

each of the four years of this study (AH1). had significantly higher rates of

groundwater flow when cornpared to sites not used in each year. Egg survival

(emerged + alive) was related to the rate of groundwater flow at the site the eggs

were buried. Egg survival did not increase substantially in sites with rates of

groundwater flow above -20 ml-m" -min-'. I tested the hypothesis that female

quality matches a gradient in resource quality. My results provide partial support for the prediction that female quality (Le. body size) and frequency of site use

correlate with the gradient in resource quality at spawning sites. A threshold in the benefb associated with increasing groundwater flow may obscure size-

related patterns of site use with resource quality. The reproductive success of females and males may be wnstrained by the limited number of spawning sites that have high rates of groundwater flow in this lake population of brook trout. INTRODUCTION

The spatial distributions of animals reflect direct responses of individuals to their environment and to conspedfics. This is especially true of mating systems, which are based on interactions among indiduals and the spatio-temporal distribution of resources necessary for each sex to ensure successful reproduction (Emlen &

Oring 1977). Because of sex differences in factors that confer reproductive success, the dispersion of resources will determine female dispersion, which in tum will determine male dispersion (se8 Chapter 1). Studies in which the distribution of resources and females were manipulated provide strong evidence for the dispersion model as a primaiy factor shaping mating systems in which males provide no parental care. For example. the addition of surplus food to areas resulted in a change in female home range (Ostfeld 1986). Likewise, manipulating the location of females resulted in a change in the home range of males, but not vice-versa (Ims 1988). Because male dispersion does not influence fernale dispersion, we assume that the spatio-temporal distribution of resources determines the location of females. Although both of these examples provide support for the dispersion model, they also illuminate the problem of identifying and measuring the resources sought by fernales to maximise reproductive success.

Beyond the general consensus that the spatio-temporal distribution of resources influences female dispersion (e.g. Deutsch 1994), 1 would argue that there is little evidence to predict how individual females distribute themselves

with respect to reçource quality. For females, the importance of resources will be

greatest when males do not provide parental car8 or other resources (Davies

1991). Thus, it followç that females should dioose the highest quality resources

to maximise offspring survival. In situations where there exists substantial

variation in the quality of females (i.8. size, fecundity) and resources. one would

predict female quality to equate with resource quality. This hypothesis remains

largely untested because of the diiiculty involved in identÏfying and quantrfying

resources in the field. Part of this diffiwlty may be a resuit of male contributions

in many mating systems. When males defend or provide resources, it is diïicult

to separate the contributions af the male (i.0. genes. protection) from that of the

resource. In any case, we lack a convincing explanation for how individual

fernales distribute themselves among resources of varying quality since it is

difficult to measure resource selection and quality in the field.

The mating system of brook trout, Salvelinus fontinalis, which spawn in

lakes, provides a unique opportunity to quantify the resources sought by females.

Fernale brook trout compte for access to spawning sites in which upwelling

groundwater is present (see review in Appendix 1; Ridgway & Blanchfield in

press). The flow of groundwater at spawning sites increases the su~valof young salmonines by providing a stable environment and removing metabolic

wastes from developing embryos (Sowden & Power 1985). Similar processes

apparently occur in lakes wtiere sockeye salrnon, Oncutt?ynchusne&, spawn, although water movement is a result of wave action (Leonetti 1997). For brook trout, the period of time required for development to a free-swimming stage depends rnainly on water temperature (Embody 1934; Godin 1982). Since groundwater is generally warmer (8°C) than ambient water temperatures during the incubation period (2°C).groundwater flow also shortens developmental time

(Appendix 1). Thus, groundwater is beneficial because it increases the survival of offspring and decreases developmental time (Embody 1934).

The objective of this study is to determine the relationship between female quality and resource quality. Female body size and the rate of groundwater flow at spawning sites are variable for this population (seChapter 2). 1 test the hypothesis that the distribution of females in the breeding population, with respect to size, is conelated to the distribution of spawning habitat quality. I predict that the (i) number of spawnings. (ii) the number of females spawning, and (iii) aie body size of spawning females will increase with increasing groundwater flow at spawning sites. These predicüons are based on the assumption that the rate of groundwater flow increases the survival of eggs in this population of brook trout. I test this assurnption. Location of spawning sites

I located spawning sites at Scott Lake based on the digging, guarding or covering activity of fernales. In the absence of fish. the presence of recently deared substrate with a small mound indicated a recently covered spawning site. I determined the distance of each spawning site from shore and the depth of each spawning site (to the nearest cm) with a tape measure. Spawning sites were mapped with one observer located in the water above a site and a second person holding the measuring tape on shore. The distribution of spawning sites was mapped each year for four years (1993-1 996). 1 determined the use of spawning sites among years by fernale brodc trout from frequent visits dunng

1993, and from daiiy visits during the following three breeding seasons (1994-

1996). The marking of individual spawning sites and the use of permanent shoreline markers from which to take measurements allowed me to determine whether redds were constructed in the same sites among years (Fig. 6.1).

Measurement of groundwater

I measured groundwater flux at spawning sites in three of four years

(1994-1 996) by the seepage meter method described in Appendix 1. In 1996,l made additional measures at redd sites using smaller seepage meters (25-cm diameter) where the regular meters (57-cm diameter) would not fit. 1 correcteci for any biases in groundwater flow due to the sire of the meter with triplicate measures using both the large and small meters. The ratio of flow into large versus small meters at the same site provided an unbiased estimate of groundwater ffow as opposed to using only the difference in area ~overdby the meters based on aie size of the meters (Ridgway & Blanchfield in press).

There was no signifiant difference in the rate of groundwater flow at spawning sites among years (see Table 5.1 for repeated measures ANOVA).

Rates of groundwater flow were pooled for al1 years at each site and related to use in each of the year classes. I documenteci sitespecific number of spawnings, number of females and maximum size of female spawning in relation to site quality for oiree breeding seaçons (1994 - 1996).

Egg su~valexperirnent

In the fail of 1997,l conducted an expriment using egg hatchboxes

(VibeRm) to determine the effect of groundwater on egg survival at Scott Lake. I measured groundwater flow at nine previously used spawning sites and at one random site located dose (4m) to other spawning sites.

Three ripe females of similar size (37.4 - 41.1 cm FL) were captured and anaesthetised lightly with MS-222. Using gentle pressure, I stripped the eggs of each female into a separate bowl. I feitilised the eggs of each female using the milt from two males (which were also lightly anaesthetised) and added a slight amount of water. All fish were retumed to the lake unharmed and the eggs were incubated overnight in the lake. The following day al1 dead eggs were removed. and 100 eggs were placed in each Vibert box I attached three Vibert boxes together and each box contained the eggs of a different female. I used a second type of Vibert box since there were not enough of al1 one type. This second style of box (Ak2)was similar to the other type, except that al1 three crosses could be contained separately within one box. There were not equal numbers of eggs available from ail females, thus, each set of three boxes did not always contain eggs from al1 three females (k2),and one box contained only 57 eggs.

The boxes with eggs were placed in a cooler and each set of three boxes was placed into one spawning site (6 Nov. 1997). 1 gently covered the boxes with substrate (usually srnaIl gravel with some sand) until they were no longer visible and there was a slight rnound over the spawning site, similar to when females cover. As well, I attached a small float with string to each set of boxes prior to burial (se8 below). I replaced the mesh covet over the buried egg boxes to prevent any disturbance from site reuse by spawning females.

The egg boxes were left for a period of almost three months (88 d) prior to retrieval (2 Feb. 1998). Holes were drilled through the ice above egg box sites.

The mesh covers were removed from over the site and each set of three boxes was gently pulled to the surface by the float, using a long pole. Boxes were immediately placed in coolers and once ail boxes were retrieved, eggs were counted and scored as to suwival and development stage. Eggs were considered dead or alive prirnarily based on pigment; dead eggs tend& to be cloudy whitdorange, whereas IN8 eggs were deep orange in colour. I categorised the stage of development of dl eggs (categories: pre-eyed; eyed; yolk sac) whether alive or dead. Eggs were preserved in alcdiol and were rechecked at a later date using a microscope. No ditference in the number of eggs alive was found between these two counts.

Some egg boxes did not contain their original full complement of eggs. It is possible that some eggs may have fallen out during the retrievaf process.

However, if egg loss were a result of methodology, I would expect roughly the same losses from al1 boxes. This was cleafly not the case. More likely is that missing eggs represent the development to a free-swimming stage (note: the egg boxes are designed with dots such that the eggs can not fall out, but newly hatched youngof-year can swim through the slots). I considered rnissing eggs as young that had emerged only if there were other eggs (alive or dead) at the yolk sac stage (Le. the stage just prior to emergence). Thus, I calculated egg su~val as follows:

egg sumal= number of eggs alive + number of emerged young (7)

Egg suMval did not dier among females for the sites (AM) in which the eggs of al1 three females were present ( M.35, P-0.71, ANOVA). RESULTS

Females constructed redds in a total of 92 separate sites over a four year period

at Scott Lake (1993-1996) (Fig. 6.1). However, not al1 sites were used in each

year. The few spawning sites used by females in each of the four years (fil 1)

had significantly higher rates of groundwater flow when cornpared to sites not

used in each year (Mann-Whitney: b23.0, M.ûû1) (Fig. 6.2).

The relationship between resource quality and fernale quality was

determined using the largest female obsetved spawning at a given site. The

relationship between groundwater flow and the maximum size of female

spawning at a given site was positively correlated during one breeding season

(Table 6.1). Interestingly, the year in which I found a significant relationship

between female body size and groundwater flow is the year that large females were delayed from breeding (1996; see Chapter 5). The relationships between groundwater flow and the number of females spawning and number of spawnings at a site were significantly and positively correlated for two of the three breeding seasons (Table 6.1).

The proportion of eggs that ernerged was related to the rate of groundwater flow at the site the eggs were buried (Fig. 6.3a). As well, egg sunrival (eq. 1) increased with the rate of groundwater flow (Fig. 6.3b). Eggs 3 4 Years used

Figure 6.2 The frequency of spawning site use (open bars) and mean groundwater seepage rates (+1 SE; solid bars) of spawning sites used in varying numbers of years at Scott Lake. The number of redd sites in which groundwater estimates were made are presented. Table 6.1 Speanan rank correlation coefficients of the relationship be-n variables associateci with site use by females and the rate of groundwater flow (poded among years, see Methods); Sample sizes (number of spawning sites from which data were obtained) are shown beneath the correlation coefficients.

Year Variables 1994 1995 1996

Number of fernales 0.56 0.34 0.64' (20) (1 8) (18)

Number of spawnings 0.52' 0.4 1 0.82' (22) (20) (21

Maximum female size 0.06 0.19 0.67' (21 ) (1 9) (21 1

'significant correlation between groundwater flow and variable (Pe0.05) placed in sites with high rates of groundwater flow showed greater development than eggs in sites of lesser flow. However, egg survival did not increase substantially in sites with rates of groundwater flow above -20 rn~-m"-min-' (Fig.

6.3b). This relationship is shown best as a logistic regression of survival ('yes' or

'no') compared to groundwater flow rate (2~4.0,df=l. PcO.00001) (Fig. 6.4).

Overall, egg suMval was low (mean t SE: 20.7 t 4.6%). If sites that had no survival are excluded from this measure, then average egg survival is 36.5% (I

5.7%).

DISCUSSION

I quantified the relationship between resource quality and offspring survival for a population of lake-spawning brook trout, Salvelinus fontinalis. l tested the hypothesis that female quality matches a gradient in resource quality. My results provide partial support for the prediction that female quality (i.0. body size) and frequency of site use correlate with the gradient in resource quality at spawning sites.

Within seasons, a positive relationship between resource quality and the frequency with which females spawned at these sites occurred during two breeding seasons. I observed a similar relationship behnreen groundwater ftow at spawning sites and the size of the large& female that spawned at those sites Rate of aroundwater flow (rn~m-~mrnin-')

Figure 6.3 The proportion of eggs that (a) emerged and (b) sunrived (emerged + alive) in relation to rates of groundwater flow (+1 SE). I 1 I @ 1 O 5 10 15 20 25 30 35 40 Rate of groundwater flow (rn~=rn-~.rnin-')

Figure 6.4 Logistic regression of egg su~val('yes' or 'no') in relation to rates of groundwater flow. Logistic regression modei: yiexp( 3.59+ (0.284)'~) /(1 +exp(-3.59+(0.284)*~)). only during the year in which large females were detained from spawning (1996; see Chapter 5). Why the relationship between resource quality and choice of sites by spawning females was so strong during the year I detained large females but not significant in other years is not immediately apparent. One explanation for why the relationship between site quality and female size or site use was not significant may be due to the gradient in resource quality. At Scott

Lake, resource quality (groundwater flow) was not continuous but reached a threshold (-20 rnLm'* omin-')nat which point increased flow did not substantially improve egg survival (Fig. 6.3b). A threshold in resource quality could weaken the relationship between site quality and female quality since ail females spawning at sites with groundwater flow rates above the threshold will have similar reproductive success (Le. offspring survival).

This is the first study to show that the rate of groundwater flow influences the suMval and development of brook trout eggs in the field. For salmonines, roughly 15% of eggs survive to the alevin (free-swimming) stage (reviewed by

Godin 1982). The proportion of eggs su~vingduring this study (-21%) was much less than reported for other studies of brook trout (90%. McFadden 1961 ;

79%. Shetter 1961; 70%, Gdswold 1967). This may be due to naturally lower survhl at Scott Lake, or perhaps handling during the experiment reduced egg viability. Developmental times required to reach the freeswimming stage did not show a similar threshold as a result of groundwater flow. Thus, oie benefit to female brook trout that choose sites with high flow rates is a combination of greater survival and shortef developmental times for offspring. The benefiis of

early emergence are related to territov acquisition for strearn salmonines (0.g.

Maçon & Chapman 1965). Although brook trout in this lake do not follow the territorial mode1 (Biro et al. 1997) early emergence likely results in the ability to feed for a longer period of time and gain benefits associated with large body size at the end of the first summer of life (Le. greater overwinter survival. decreased predation).

Spawning sites used in each of the four years of this study had significantly higher rates of groundwater flow than sites used intermittently dunng this time period. These sites of high flow account for only 12% of al1 spawning sites used during the four years of this study. In any given year, these high flow sites represent no more than onethird of available spawning sites. In this study, there was virtually no egg suwival at rates of groundwater flow less than -1 2 rn~~rn*~*min? Thus, females that spawn at sites with low rates of groundwater flow may not be cuntributing to the population. The observation that resource quality is a good predictor of offspring survival suggests that the delay in spawning by smaller females may not only reduce brood loss through nest destruction, but allow access to high quality spawning sites (see Chapter 5).

Many studies have been able to show a link between the dispersion of females and resources (see Davies 1991) and this study is no exception; the dispersion of groundwater detemines the dispersion of females. Beyond this generalisation, however, lies the basis for female choice of sites. Higher quality sites result in the survival of greater numbers of offspring. It is important not to underestimate the importance of this seemingly simple observation. The ability to quantify the resources that confer reproductive success to females provides a unique opportunity to Iink female choice of resources and resource quality.

Although the distribution of females in the breeding population, with respect to size, was not correlated to the distribution of spawning habitat quality in a11 years, this study provides strong evidence for the significance of resources to females. CHAPTER SEVEN:

SUMMARY SUMMARY

Although there has been extensive study of reproductive patterns in salmonine fishes (salmon, trout and char), we are still only beginning to understand the dynamics and implications of the reproductive interactions that occur on the spawning grounds (Fleming 1996). The reason for this limited understanding is due, in part. to a lack of detailed observations on marked individuals under natural conditions. The objective of this thesis is to provide a detailed description of a salmonine mating system and to relate these findings to current mating system theory. The reproductive strategies of brook trout (Salvelinus fontinalis) are examined from an evolutionary-ecological perspective. whereby the behaviour of individuals is viewed as an outcome of cornpetition to maximise reproductive success. Unlike sernelparous Pacific salmon. it was not possible to estimate lifetime reproductive success for this iteroparous population of brook trout. I also was unable to determine the sunrival and reproductive success of individuals throughout all four years of this study due to the problem of extensive tag loss in each year.

1. I base my description of this brook trout mating system on daily observations

of tagged individuals. Spawning occurred over an extended period (-50 d)

during which males outnumbered fernales (Figs. 2.2,2.4). Males were ripe prior to females and spent a longer duration on the spawning grounds

compared to females of similar length (Fig. 2.3).

2. This is the first study to show that the selection of spawning sites by females is

determined by groundwater flow in lakes (Table 2.1). Spawning sites were

used unequally, with half of ail spawning acüvity occurring at only 18% of

available sites (Fig. 2.7). Removal expiments indicate that sites are a limiting

resource. The frequency of multiple redds and duration in spawning activity by

fernales increased with body size (Figs. 2.8,2.9). Large females spawned

synchronously over a brief period (-1 5 d).

3. Focal sampling of 20 males showed extensive site Visitation, with a greater

frequency of repeated visits to sites with females (Fig. 3.3). Males visited

significantly fewer females than were available (30%). Repeat visits to females

are explained as a tactic used by males to predict female readiness to spawn.

4. Seasonal patterns of space use and cornpetitive mate searching behaviours

(Le. moving and dominant) by male brook trout were positively related to male

body size (Fig. 3.4). Competiüve mate searching behaviours were not

influenced by the absolute number of receptive females, but decreased, as the

operational sex ratio (OSR) became increasingly male-biased (Tables 3.2, 3.3;

Fig. 3.5). The OSR has the potential to influence space use in species where mate choice and mate searching are important components of reproductive

SUCCBSS*

5. Male brook trout paired with large fernales experienced a greater incidence of

kleptogamy due to increased numbers of peripheral males present (Table 4.2).

I estimated that males which had peripheral males participate in spawning rnay

fertilise, on average, equal numbers of eggs compared to males spawning

solely with a smaller fernale (Fig. 4.5).

6. Fernales which had eggs eaten by peripheral males were paired with smaller

males (Tabie 4.2). Latency to spawn by females increased when paired with a

relativeiy small male, and resulted in females obtaining a larger spawning

partner (Fig. 4.4). Delayed spawning by females paired with relaüvely small

males promotes patterns of sizeassortative mating (Fig. 4.3a).

7. Large females influence the seasonal timing of spawning. Small fernale brook

trout (45.0 cm FL) spawned significantly later during the year that large

females were detained (1 996) compared to the previous two years (Figs. 5.1,

5.2). Brood loss through nest destruction was significantly reduced in aie

absence of large femafes (Fig. 5.4). 8. The few spawning sites used by females in each of the four years of this study

(AHl), had significantly higher rates of groundwater flow when compared to

sites not used in each year (Figs. 6.1,6.2).

9. Egg sutvival (emerged + alive) was related to the rate of groundwater flow at

the site the eggs were buried (Fig. 6.3). Egg survival did not increase

substantially in sites with rates of groundwater flow above -20 mLm" -min"

(Figs. 6.3, 6.4).

10. The relationships between groundwater flow and the (i)number of females

spawning and (ii)number of spawnings at a site wre significantly and

positively correlated during two breeding seasons (Table 6.1). The size of

spawning female was positively related to groundwater flow only during the

year in which large females were delayed from breeding (1996). A threshold

in the benefits associated with increasing groundwater flow rnay obscure

size-related patterns of site use with resource quality.

11. Reproductive success of females and males may be constrained by the

limited number of spawning sites (AB)that have high rates of groundwater

flow (>20 rn~-rn-~-minœ') in this lake population of brook trout (Figs. 6.2, 6.3). 12. This study is the first to describe the mating systern of brook trout based on

individual observations. Resource quality determines spawning site selection

by fernales. The spatial distribution of spawning sites (i.e. groundwater flow)

detemines the distribution of fernales. The location of spawning fernales. in

tum, detemines where males search for mates. Ahnesjt) 1, Vincent ACJ, Alatolo R. Hailiday T & Sutherland WJ (1993) The role of females in influencing mating patterns. Behav. Ecol. 4: 187-189.

Andersson M (1994) Sexual selection. Pfinceton: Princeton University Press.

Arnold SJ & Duvall D (1994) Animal mating systems: a synthesis based on selecüon theory. Amer. Nat. l43:317-348.

Ball OP & Cope OB (1961) Mortality studies on cutthroat trout in Yellowstone Lake. U.S. Fish and Wildlife Service Research Report 55.

Barfaup BT, Lura H, Saqrov H & Sundt RC (1994) Inter- and intra-speafic variability in female salmonid spawning behaviour. Can. J. Zool. 72:636-642.

Bateman Al (1948) Intra-sexual selection in Dmsophiia. Heredity 2:349368.

Beacham TD & Murray CB (1988) A genetic analysis of body size in pink salmon (Oncorhynchus gorbuscha). Genome 30:s 1-35.

Benson NG (1953) The importance of ground water to trout populations in the Pigeon River, Michigan. Trans. N. Amer. Wildl. Conf. 18264281. van den Berghe EP & Gross MR (1 984) Fernale size and nest depth in cdio salmon (OncoIhynchus khutch)).Cm. J. Fish. Aquat. Sci. 41 204-206. van den Berghe EP & Gross MR (1 986) Length af breeding life of coho salrnon (Oncodqmchus kisutch). Cm. J. 2001. 64: 1482-1 486. van den Berghe EP & Gross MR (1989) Natural selection resulting from female breeding competition in a Pacific salmon (coho: Oncorf,ynchuskisutch). Evolution 43: 125-1 40. van den Berghe EP, Wemerus F & Wamer RR (1989) Fernale choice and the mating cost of peripheral males. Anim. Behav. 38:875-884.

BirMiead TR & Msller AP (1992) Spem competition in birds: evolutionary causes and consequemes. London: Academic Press.

Biro PA, Ridgway MS & Noakes DLG (1997) The centraCplace territorial mode1 does not apply to space-use by juvenile brook charr Salvelinus fontbal& in lakes. J. Anim. Ecol. 66:837-845. Bisana A & Marconato A (1 988) Female mate choice, male-male competition and parental care in the river bulihead, Cottus gobio L,(Pisces: Cottidae). Anim. Behav. 36: 1352-1360.

Borgia G (1980) Semal competition in Scatophaga stemmri& size- and density- related changes in male ability to capture females. Behaviour 75:185-206.

Boness DJ, Bowen WD & lverson SJ (1995) Does male harassrnent of females contribute to reproductive synchrony in the grey seal by affecting materna1 perfomance? Behav. Ecol. Sociobiol. 36: 1-1 0.

Biyant MJ & Grant JWA (1995) Resource defence. monopolization and variation of fmiess in groups of female Japanese rnedaka depend of the synchrony of food arrival. Anim, Behav. 491469-1 479.

Bye VJ (1984) The role of environmental factors in the timing of reproductive cycles. ln Fish reproduction: strategies and tactics (Ed. by Potts GW & Wootton RJ) pp. 187-205. London: Academic Press.

Carline RF (1980) Features of successful spawning site development for brook trout in Wisconsin ponds. Trans. Amer. Fish. Soc. 109:453-457.

Carscadden J, Nakashima BS & Frank KT (1997) Effects of fish length and temperature on the timing of peak spawning by capelin (Mallotus vilusus). Can. J. Fish. Aquat. Sci. 54:781-787.

Chapman DW (1 988) Critical review of variables used to define effects of fines in redds of large salmonids. Trans. Amer. Fish. Soc. 11 7: 1-21.

Chebanov NA, Vamavskaya NV & Vanavskiy VS (1983) Effectiveness of spawning of male sockeye salmon, Oncori9ynchus ne- (Salmonidae), of differing hierarchical rank by means of genetic-biochemical markers. J. Ichthyol. 23:s 1-55.

Clutton-Br& TH (1989) Mamrnalian rnating systerns. Proc. Royal Soc. Lond. B 236:339-372.

Clutton-Brock TH (1991) The evolution of parental are. Princeton: Princeton University Press.

Clutton-Brock TH & Vincent ACJ (1991) Çexual selection and the potential reproductive rates of males and females. Nature 351 :58-60. Clutton-Brock TH & Parker GA (1992) Potential reproductive rates and the operation of semai selection. Quar. Rev. Biol. 67A37-456.

Clutton-Brock TH, Albon SD. Gibson RM & Guinness FE (1979) The logical stag: adaptive aspects of fighting in red deer (Cervus elaphus L).Anim. Behav. 27:211-225.

Coble DW (1962) Influence of water exchange and dissolveci oxygen in redck on suwival of steelhead trout embryos. Trans. Amer. Fish. Soc. 90:469-474.

Cole L (1962) A closed sequential test design for toleration experirnents. Ecology 43:749-753.

Côte IM & Hunte W (1989) Male and female mate choice in the redlip blenny: why bigger is better. Anim. Behav. 38:78-88.

Coulson JC (1981) A study of the factors influencing the timing of breeding in the grey seal Halichoenrs grypus. J. Zod. 194:55Mil.

Crespi BJ (1986) Size assessment and alternative fighting tactics in Elaphrothnps tuberculatus (1 nsecta: Thysanoptera). Anim. Behav. 34: 1324- 1335.

Crisp DT 8 Carling PA (1989) Obsenmtions on siting. dimensions and structure of salmonid redds. J. Fish. Biol. 34: 119-1 34.

Cronin EW & Sherman PW (19TI) A resource-based mating system: the orange- rumped honeyguide. Living Bird 15:s-32.

Cunjak RA, Power G 8 Barton DR (1986) Reproductive habitat and behaviour of anadromous arctic char (Sahelrirus alphus) in the Koroc River, Quebec. Nat. Can. 113:383-387.

Cunjak RA, Curv RA & Power G (1987) Seasonal energy budget of brook trout in streams: implications of a possible deficit in early winter. Trms. Amer. Fish. SOC. 11 6~8 17-828.

Curry RA & Noakes DLG (1995) Groundwater and the selection of spawning sites by brook trout (Sahelinus fontinalis). Can. J. Fish. Aquat. Sci. S2:1733- 1740.

Curry RA, Noakes DLG & Morgan GE (1995) Groundwater and the incubation and emergence of brook trout (Sahelinus fontinalis). Can. J. Fish. Aquat. Sci. 52:1741-1749. Cushing DH (1969) The regulatity of the spawning season of some fishes. J. Cons. int. Explor. Mer. 33:81-97. .. . .

Danzmann AG 8 Ihçsen PE (1995) A phylogeographic survey of brwk charr (Sah/eIhusfontha1rS) in Algonquin Park, Ontario bas& upon mitochondrial DNA variation. Mol. Ecol. 4:681:697.

Darwin C (1871) The descent of man. and selection in relation to sex London: Murray.

Davies NB (1991) Mating systems. In An introduction to behavioural ecology, 3rd edn (Ed. by Krebs JR & Davies NB) pp. 263-295.Oxford: Blackwell Scientific Publications.

Davies NB & Halliday TR (1979) Cornpetitive mate searching in male cornmon toads, Bufo bh.Anim. Behav. 27: 1253-1267.

Davis LS & Murie JO (1985) Male territoriality and the mating system of Richardson's ground squinels (Spennophilus riizhardsonir). J. Mamm. 66:268- 279.

Deutsch JC (1994) Leldang by default: female habitat preferences and male strategies in Uganda kob. J. Anim. Ecol. 63: 101-1 51.

DeVries P (1997) Riverine salmonid egg butial depths: a review of published data and implications for scour studies. Can. J. Fîsh. Aquat. Sci. 54: 1685- 1698. van Dongen S, Backeljau T. Matthysen E & Dhondt AA (1997) Synchronization of hatching date with budburst of individual host trees (Quercus robur) in the winter moth (Operophtem brumata) and its fitness consequences. J. Anim. €COI.66:113-121.

Downhower JF 81 Brown L (1 980) Mate preferences of female mottled sculpin, Cottus bairdi. Anim. Behav. 28:728-734.

Downhower JF, Brown L (1981) The timing of reproduction and its behavioral consequences for mottled sculpins, Cottus bairdi. In Natural selection and social behavior (Ed. by Alexander RD, Tinkle DW) pp. 78-95. New York: Chiron.

Downing JA & Peterka JJ (1978) Relationship of rainfall and lake groundwater seepage. Limnol. Oceonogr. 25821-825. Eastman JR (1993) Idrisi, version 4.1. Worcester: Clark University.

Elliott JM (1984) Numerical changes and population regulation in young migratory trout Salmo Main a Lake District stream, 1966-83. J. Anim. Ecol. 53~327-350.

Embody GC (1934) Relation of temperature to the incubation perïod of eggs of four species of trout Trans. Amer. Fish. Soc. 64:281-289.

Emlen ST & Onng LW (19TI) Ecology, semal selection, and the evolution of mating systems. Science l97:215-223.

Enders MM (1993) The effect of male size and operational sex ratio on male mating success in the cornmon spider mite, Tetranychus urticae Koch (Acari: Tetranychidae). Anim. Behav. 46:835-846.

Everson PR & Addicott JF (1 982) Mate selection strategies by male mites in the absence of intersemal selection by fernales: a test of six hypotheses. Can. J. ZOO^. 60:2729-2736.

Ferguson MM, Liskauskas AP & Danzmann RG (1995) Genetic and environmental correlates of variation in body weight of brook trout (Salvelhus fonthalis). Can. J. Fish. Aquat. Sci. 52:307-314.

Fleming IA (1996) Reproductive strategies of Atlantic salmon: ecology and evolution. Rev. Fish Siol. Fish. 6:379-416.

Fleming IA & Gross MR (1993) Breeding succeçs of hatchery and wild coho salmon (Oncorhynchus kisutch) in campetition. Ecol. Appl. 3:230-245.

Fleming IA & Gross MR (1994) Breeding competition in a Pacific salmon (coho: Oncorfiynchus kisutch): measures of natural and sexual selection. Evolution 48:637-657.

Foote CJ (1988a) Male mate choice in fishes: a review of the evidence in coho salmon. Anim. Behav. 36: 1228-1230.

Foote CJ (1988b) Male mate choice dependent on male size in salmon. Behaviour 106:63-80.

Foote CJ (1989) Female mate preference in Pacific salmon. Anim. Behav. 37:721-723. Foote CJ (1990) An experimental comparison of male and fernale spawning territoriality in a pacific salmon. Behaviour 115:283-314.

Foote CJ & Larkin PA (1988) The foie of male choice in the assortative mating of anadromous and non-anadromous sockeye salmon (Onwrhynchus nerka). Behaviour 1 06:43-62.

Foote CJ, Brown GS & Wood CC (1997) Spawning success of males using alternative mating tactics in sockeye salmon, Oncmhynchus nerka. Can. J. Fish. Aquat. Sci. 54: 1785-1 795.

Fraser JM (1982) An atypical brook charr (Salvelinus fontinalis) spawning area. Env. Biol. Fish. 7:385-388.

Fraser JM (1985) Shoal spawning of brook trout, Salvelinus fontinalis, in a Precambrian shield lake- Nat- Can. 1 t 2:163-174.

Fukushima M (1994) Spawning migration and redd construction of Sakhalin taimen, Hucho penyi (Salmonidae) on northem Hokkaido Island, Japan. J. Fish Biol. 44:877-888.

Gibson RM (1996) Female choice in sage grouse: the roles of attraction and active comparison. Behav. €cd. Socîobiol. 3955-59.

Gibson RM & H6glund J (1992) Copying and sexual selection. Trends Ecol. Evol. 7229-231.

Gibson RM & Langen TA (1996) How do animals choose their mates? Trends Eco~.Evo~. 11:468-470.

Godin J-GJ (1982) Migrations of salmonid fishes during early life history phases: daily and annual timing. In Proceedings of the salmon and trout migratory behavior symposium (Ed. by Brannon EL & Salo €0)pp. 22-50. Seattle: University of Washington Press.

Grant JW, Bryant MJ & Soos CE (1995) Operational sex ratio, mediated by synchrony of fernale amval, alters the variance of male mating success in Japanese medaka. Anim. Behav. 49:367-375.

Greeley JR (1932) The spawning habits of brook, brown and rainbow trout, and the problem of egg predators. Trans. Amer. Fish. Soc. 62:239-247.

Griswold BL (1967) Some aspects of the shore spawning of brook trout (Salvelinus fontinalis, Mitchell). M.Sc. thesis, University of Maine. Groot C & Margolis L (1991) Pacific salmon life histones. Vancouver: UBC Press.

Gross MR (1984) Sunfish, salmon, and the evolution of alternative reproductive strategies and tactics in fishes. In Fish reproduction: strategies and tactics (Ed. by Potts GW & Wootton RJ) pp. 55-75. London: Academic Press.

Gross MR (1985) Disruptive selection for alternative life histories in salmon. Nature 313:47-48.

Gross MR (1996) Alternative reproductive strategies and tactics: diversity wïthin sexes. Trends Ecol. Evol. 11 :92-98.

Gross MR, Neff BD 8 Fleming IA (in press). Reproductive inferiofity of hatchery salmon. Science.

Hale JG (1968) Observations on brook trout, Salvelnus fontinalls, spawning in 10-gallon aquaria. Trans. Amer. Fish. Soc. 97:299-300.

Hale JG & Hilden DA (1969) Spawning and some aspects of eariy life history of brook trout, SaMnus fontinalis (Mitchell), in the laboratory. Trans. Amer. Fish. Soc- 98:473-477.

Hansen DWM (1972) Reproductive interactions betw&n the brook trout and splake of Redrock Lake, Ontario. M.&. thesis, Universw of Toronto.

Hansen EA (1975) Some effects of groundwater on brown trout redds. Trans. Amer. Fish. Soc. 104: 100- 11 0.

Hanson Al & Smith HD (1967) Mate selection in a population of sockeye salmon (Oncuthynchus nerka) of mixed age groups. J. Fish. Res. Bd. Can. 24:1955- 1977.

Hartman WL Merrell TR & Painter R (1964) Mass spawning behavior of sockeye salmon in Brooks River, Alaska. Copeia 1964:362-368.

Hayashizaki K, Hirohashi M & Ida H (1995) Effect of egg size on the characteristics of embryos and alevins of diurn salmon. Fish. Science 61 :177-1 90. Hazzard AS (1932) Some phases of the life history of the eastem brook trout, Saivelinus fontinalis. Trans. Amer. Fish. Soc. 62344-350.

Healey MC & Prince A (1995) Scales of variation in life history tactics of Pacific salmon and the consenration of phenoiype and genotype. Amer. Fish. Soc. Symp. 17: 176-184. Heggberget TG (1 988) Timing of spawning in Norwegian Atlantic salmon (Sdrno salar). Cm. J. Fish. Aquat. Sci. 45:845-849.

Henderson NE (1963) Influence of Iight and temperature on the reproductive cycle of the eastem brook trout, Salvelinus fontinalis (Mitchell). J. Fish. Res. Bd. Can. 20:859-897.

Hendry AP, Leonetti FE & Quinn TP (1995) Spatial and temporal isolating mechanisms: the formation of discrete breeding aggregations of sockeye salmon (Oncorhynchus nerka). Can. J. Zool. 73:339-352.

Henson SA & Wamer RR (1997) Male and female alternative reproductive behaviors in fishes: a new approach using intersemal dynamics. Ann. Rev. EcoI. Syst. 28:571-592.

Hoglund J & Alatalo RV (1995) Leks. Princeton: Princeton University Press.

Hokanson KEF, McComick JH, Jones BR & Tucker JH (1973) Thermal requirements for maturation, spawning, and embryo survival of the brmk trout, Salvelinus fontinalis. J. Fish. Res. Bd. Can. 30:975-984.

Hutchings JA (1 994) Age- and sizespecific costs of reproduction within populations of brook trout. Sahelinus fontinalis. Oikos 70:12-20.

Hutchings JA & Myen RA (1987) Escalation of an asyrnmetric contest: mortality resulting from mate cornpetition in Atlantic salmon, Salmo salac Can. J. Zool. 65~766-768.

Hutchings JA & Myers RA (1988) Mating success of alternative maturation phenotypes in male Atlantic salmon, Salmo salar. Oecologia 75: 169-1 74.

Hutchings JA & Myers RA (1993) Effect of age on the seasonality of maturation and spawning of Atlantic mi, Gadus momua, in the northwest Atlantic. Can. J. Fish. Aquat. Sci. 50:2468-2474.

Ims RA (1988) Spatial dumping of sexually receptive fernales induces space sharing among male voles. Nature 335541-543.

Janetos AC (1980) Strategies of female mate choice: a theoretical analysis. 8ehav. Ecol. Sociobiol. 7: 107-112.

Jawi 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 84:123132. Jones JW & King GM (1952) The spawning of the male salmon parr (Salmo salar Linn. juv.). Proc. Zool. Soc. Lond. B I22:6lS-619.

Jones JW & Bal1 JN (1954) The spawning behaviour of brown trout and salrnon. Anim. Behav. 2: 103-114.

Jonsson N, Jonsson B & Hansen LP (1991) Energetic cost of spawning in male and female Atlantic salrnon, Salmo salm J. Fish. Biol. 39:739-744.

Jordan WC & Youngson AF (1992) The use of genetic marking to assess the reproductive succeçç of mature male Atlantic salmon parr (Salmo selar, L) under natural spawning conditions. J. Fish Biol. 41 :613-618.

Keenleyside MHA & Dupuis HMC (1988) Courtship and spawning competition in pink salmon (OncuIhynchus gorbuscha). Can. J. Zool. 66:262-265.

Kitano S (1996) Sb-related factors causing individual variation in seasonal reproductive success of fluvial male Dolly Varden (Salvelinus malma). Ecol. Freshw. Fish 59-67.

Kitano S & Shimazaki K (1995) Spawning habitat and nest depth of female Dolly Varden SaMnus malma of different body size. Fish. Science 61 3'76-779.

Knapp RA & Vrendenburg VT (1996) Spawning by California golden trout: characteristics of spawning fish, seasonal and daily timing, redd characteristics, and microhabitat preferences. Trans. Amer. Fish. Soc. 125:519-531.

Knowlton N (1979) Reproductive synchrony, parental investment, and the evolutionary dynamics of semal selection. Anim. Behav. 27:IO22-1033.

Kvamemo C & Ahnesj6 1 (1996) The dynamics of operational sex ratios and competition for mates. Trends Ecol. Evol. 11 :404-408.

Kvamemo C, Forsgren E & Magnhagen C (1995) Effectç of sex ratio on intra- and inter-sexual behavior in sand gobies. Anim. Behav. 50:1 455-1 461.

Leonetti FE (1997) Estimation of surface and intragravel water flow at sockeye salmon spawning beaches in llimna Lake, Aisaka. N. Amer. J. Fish. Mgrnt. 17: 194-201.

Lindstrom K & Seppa T (1996) The environmental potential for polygyny and semal seldon in th8 sand gay, PomatoSd)istus minutus. Proc. Royal Soc. Lond., B 263: 1319-1 323. Lorenz JM & Eiler JH (1989) Spawning habitat and redd characteristics of sockeye salmon in the glacial Taku River, British Columbia and Alaska. Trans. Amer, Fish. Soc. 118:495-502.

Luttbeg B (1996) A comparative Bayes tactic for mate assessment and choice. Behav. Ecol. 7:45 1-460,

Maekawa K & Hino T (1987) Effect of cannibalism on alternative life histories in charr. Evolution 41 :1 120-1 123.

Maekawa K & Hino T (1990) Spawning tactics of female Miyabe char (Salvelinus maha miyabei) against egg cannibalism. Can. J. Zool. 68:889-894.

Maekawa K, Hino T. Nakano S & Smoker WW (1993) Mate preference in anadromous and landlocked Dolly Varden (Saheiinus malma) females in two Alaskan Stream. Can. J. Fish. Aquat. Sci. 50:2375-2379.

Maekawa K & Nakano S & Yamamoto S (1994) Spawning behaviour and size- assortative mating of Japanese charr in an artificial lake-inlet stream system. Env. Biol. Fishes 39: 109- 1 1 7.

Magnhagen C (1992) Alternative reproductive behaviour in the cornmon goby, Pomatoschistus microps an ontogenetic gradient? Anim. Behav. U:f82-184.

Martin NV (1957) Reproduction of lake trout in Algonquin Park. Ontario. Trans. Amer. Fish. Soc. 863231 -244.

Martin TE (1987) Food as a limit on breeding birds: a lifshistory perspective. Ann. Rev. Ecol. Syst 18:453-487.

Mason JC & Chapman DW (1 965) Significance of early emergence, environmental rearing capacity, and behavioral ecology of juvenile coho salmon in stream channels. J. Fish. Res. Bd, Cm. 22173-190,

McFadden JT (1961) A population study of the brook trout, Sahrelinus fonthalis. Wildl. Monog. 7: 77pp.

McNeil WJ (1964) Quality of the spawning bed as it relates to survival and growth of pink salmon ernbryos and alevins and time of fry emergence. In 16' International Congress of Zoology, 1963. Proc. Internatl. Cong. Zool. l6:2#.

McNeil WJ (1967) Randomness in distribution of pink salmon redds. J. Fish. Res. Bd. Can. 24: 1629- 1634. Michener GR & McLean IG (1996) Reproductive behavior and operational sex ratio in Richardson's ground squirrels. Anim. Behav. 52:743-758.

Milinski M & Bakker TCM (1992) Costs influence sequential mate choice in sticklebacks, Gasterosteus aculeatus. P roc. Royal Soc. Lond. B 250:229-233.

Miranda LE & Muncy RJ (1987) Recruitment of young-of-year largemouth bass in relation to size structure of parental stock. N. Amer. J. Fish. Mgmt. 7: 131 -137.

Mjralnerrad IB, Fleming IA. Refseth UH 8 Hindar K (1998) Mate and spen cornpetition during multiple-male spawnings of Atlantic saimon. Can. J. Zool. 76:7WS.

Morton ES & Derrickson KC (1990) The biological significance of age-speafic return schedules in breeding purple martins. Condor 92:l O4&lO5O.

Myers RA (1986) Game theory and the evolution of Atlantic salmon (Salmo sala4 age at maturation. Can. Spec. Publ. Fish. Aquat. Sci. 89:53-61.

Needham PR (1961) Observations on the natural spawning of eastern brook trout. Calif. FÏsh Game 47:27-40.

Neilson JD 8 Banford CE (1983) Chinook salmon (Oncorhynchus tshawyfscha) spawner characteristics in relation to redd physical features. Gan. J. Zool. 61~1524-1531.

Nelson J (1995) lntrasexual cornpetition and spacing behaviour in male field voles, Microtus agrestis, mder constant female density and spatial distribution. Oikos 739-14.

Neudorf DL, Stutchbury BJM & Piper WH (1997) Covert extraterritorial behavior of fernale hooded warblers. Behav. Ecol. 8595-600.

Ostfeld RS (1986) Temtonality and mating system of California voles. J. Anirn. Eco~.55:691-706.

Otronen M (1993) Size-related male movement and its effect on spatial and temporal male distribution at fernale oviposition sites in the fly Dryomyza anilis. Anim. Behav. 46:731-740.

Otronen M (1995) Male distribution and mate searching in the yellow dung fly Scathophaga stercoraria: corn parison between paired and un paired males. Ethology 100:265-276. Ottaway EM, Carling PA, Clarke A & Reader NA (1981) Obsewations on the structure of brown trout, Salmo tnrtta Linnaeus, redds. J. Fish. Biol. 19593- 607.

Owens IPF & Thompson DBA (1994) Sex differences, sex ratios and sex roles. Proc. Royal Soc. Lond., B 258:93-99.

Parker GA (1970) The reproductive behaviour and the nature of sexual selection in Scatophaga çtercoraria L (Diptera: Scatophagidae). II. The fertilization rate and the spatial and temporal relationships of each sex around the site of mating and oviposition. J. Anim. Ecol. 39:205-228.

Parker GA (1 978) Searching for mates. In An introduction to behavioural ecology. 3rd edn (Ed. by Krebs JR 8 Davies NB) pp. 214-245. Oxford: Blackwell Scientific Publications.

Perill SA, Gebhardt HC & Daniel R (1978) Sexual parasitism in the green tree f rog Hyla cineria. Science 200: 1 17411 80.

Power G (1 980) The brook charr, Saivelinus fontinalis. ln Charrs, salmonid fishes of the genus Saivel!nus (Ed. by Balon EK) pp. 141-203. The Hague: W. Junk bv Publishers.

Quinn TP & Foote CJ (1994) The effects of body size and sexual dimorphism on the reproductive behaviour of sockeye salmon, Oncofiynchuç nerka. Anim. Behav. 48:EI -761.

Quinn TP, Adkinson MD 8 Ward MD (1996) Behavioral tactics of male sockeye salmon (Onwrhynchusne-) under varying operational sex ratios. Ethology 102:304-322.

Real L (1990) Search theory and mate choice. 1. Models of single-sex discrimination. Amer. Nat. 136:376-405.

Reid ML & Stamps JA (1997) Female mate choice tacücs in a resource-based mating system: field tests of alternative models. Amer. Nat. 150:98-121.

Reiser DW & Wesche TA (19ï7) Determination of physical and hydraulic preferences of brown and brook trout in the selection of spawning locations. Comp. Rep., Water Res. Series No. 64, 100 pp.

Reiter J, Panken KJ & LeBoeuf BJ (1981) Female cornpetition and reproductive success in northem elephant seals. Anirn. Behav. 29:670-687. Reynolds JD (1996) Animal breeding systems. Trends Ecol. Evol. 11 :68-72.

Ridgway MS & Blanchfield PJ (in press) Brook trout spawning areas in lakes. Ecol. Freshw. Fish.

Ridgway MS, Shuter BJ 8 Post EE (1991) The relative influence of body size and territorial behaviour on nesting asynchrony in male smallrnouth bas, Micropterus dolomkui (Pisces: Centrarchidae). J. Anirn. Ecol. 60:665-681.

Rintamaki PT, Alatalo RV, Hôglund J & Lundberg A (1995) Mate sampling behaviour of bladc grouse fernales (Tetmo tetrk). Behav. Ecol. Sociobiol. 37:209-215.

Rohwer S (1978) Parent cannibalism of offçpring and egg raiding as a courtship strategy. Amer. Nat. 1l2:42S44lO.

Sargent RC, Gross MR & van den Berghe EP (1986) Male mate choice in fishes. Anim. Behav. 34:545-550.

Sargent RC, Gross MR & van den Berghe EP (1988) Male mate choice in fishes: a reply to Foote. Anim. Behav. 36: 1230-1233.

Schmidt-Nielsen K (1984) Scaling: why is animal size so important? Cambridge: Cambridge University Press.

Schroder SL (1981) The role of semal selection in detemining the overall mating pattern and mate choice in churn salmon. Ph.D. thesis, University of Washington.

Schultz ET, Clifton LM & Wamer RR (1991) Energetic constraints and sizebased tactics: the adaptiue significance of breeding-schedule variation in a marine fish (Embiotocidae: Mkrometrus minimus). Amer. Nat. 138:1408-1430.

Schwagmeyer PL (1994) Cornpetitive mate searching in thirtwn-lined ground squirrels (Mammalia, Sciuridae): potential roles of spatial memory. Ethology 98:265-276.

Schwagmeyer PL (1995) Searching today for tomorrow's mates. Anim. Behav. 50:759-767.

Shetter DS (1961) Su~valof brook trout from egg to fingerling stage in two Michigan trout streams. Trans. Amer. Fish. Soc. 90:252-258. Sigurj6nsdottir H & Gunnarsson K (1989) Alternative mating tactics of ardic charr, Sa~elhusa@hus, in Thingvallavatn, lceiand. Env. Biol. Fish. 26: 154 176.

Smith OR (1941) The spawning habits of cutthroat and eastem brook trouts. J. Wildl. Mgmt. 5:461-471.

Sowden TK & Power G (1985) Prediction of rainbow trout embryo suMvaf in relation to groundwater seepage and particle size of spawning substrates. Trans. Amer. Fish. Soc. 1l4:804-812.

StatSoft, Inc. (1 996) STATISTICA for Windows, version 5.0 [Cornputer program manual]. Tulsa, OK: Statsoft, Inc.

Stutchbury BJ & Robertson RJ (1988) Within-season and agsrelated patterns of reproductive performance in female tree swallows (Tachycineta bicolor). Can. J. ZOOI.66:827-834-

Sullivan MS (1994) Mate choice as an information gatherhg process under time coristraint: implications for behaviour and signal design. Anim. Behav. 47:141- 151.

Tabo rsky M (1994) Sneakers, satellites, and helpe rs: parasitic and cooperative behavior in fish reproduction. Adv. Study Anim. Behav. 23: 1-100.

Tautz AF & Groot C (1975) Spawning behavior of churn salmon (Oncorhynchus keta) and rainbow trout (Salmo gairdnen). J. Fish. Res. Bd. Cam 32:633-642.

Tejedo M (1988) Fighting for females in the toad Bufo calamita is affected by the operational sex ratio. Anim. Behav. 36: 1765-1769.

Thomaz D, &al1 E & Burke T (1997) Alternative reproductive tactics in Atlantic salmon: factors affecting mature parr success. Proc. R. Soc. Lond. 8 264:2 19-226.

Trail PW & Adams ES (1989) Active mate choice at cock-of-therock leks: tactics of sampling and cornparison. Behav. Ecol. Sociobiol. 25283-292.

Trivers RL (1 972) Parental investment and sexual selection. In Sexual selection and the descent of man, 1871-1 971 (Ed. by Campbell B) pp.136-179. Chicago: Aldine. Vincent ACJ, Ahnesjo I & Berglund A (1994) Operational sex ratios and behavioural sex differences in a pipef ish population. Behav. Ecol. Sociobiol. 34:435-442.

Vladykov VD (1956) Fecundity of wild speckled trout (Salvelinus fontinalis) in Quebec lakes. 3. Fish. Res- Bd. Can. 13:799-841.

Wagner RH (1991) Evidence that female razorbills control extra-pair copulations. Behaviour 1l8:157-169.

Wagner RH (1992) Extra-pair copulations in a lek: the secondary mating system of rnonogarnous razorbills Behav. Ecol. Sociobiol. 31 :63-71.

Watson A & Moss R (1970) Dominance, spacing behaviour and aggression in relation to population limitation in vertebrates. ln Animal populations in relation to their food resources (Ed. by Watson A) pp. 167-218, Symposia 10 of the British Ecological Society. Mord: Blackwell Scientific Publications.

Webb JH & Mclay HA (1996) Variation in the üme of spawning of Atlantic salmon (Saimo salar) and its relationship to temperature in the Aberdeenshire Dee, Scotland. Can. J. Fish. Aquat. Sci. 5327342744

Webster DA & Eiriksdottir G (1976) Upwelling water as a factor influencing choice of spawning sites by brook trout, Salvelinus fontinalis. Trans. Amer- Fish. SOC. 10541 6-421 .

White HC (1930) Some observations on the eastem brook trout (S. fontinalis) of Prince Edward Island. Trans. Amer. Fish. Soc. 6O:f 01 -1 08.

Whitehead H (1990) Rules for roving males. J. Theor. Biol. 145: 355-368.

Whittingham LA, Taylor PD 8 Robertson RJ (1992) Confidence of patemity and male parental care. Amer. Nat. 139: 11 1 5-1 125.

Williams GC (1966) Adaptation and natural selection. Princeton: Princeton University Press.

Whel LD & MacCrimmon HR (1983) Redd-site selection by brook trout and brown trout in southwestern Ontario streams. Trans. Amer. Fish. Soc. 1l2:76O-ï7 1. Wray S. Cresswell WJ & Rogers D (1992). Dirchlet tesselations: a new, non- parametric approach to home range analysis. ln Wildlife telernetry. remote monitoring and tracking of animals (Ed. by Priede IG & Swift SM) pp. 247- 255. Chichester: EIlis Horwood Ltd.

Young MK. Hubert WA & Wesche TA (1989) Substrate alteration by spawning brook trout in a southeastem Wyoming Stream. Trans. Amer. Fish. Soc. 1 18:379-385. APPENDlX ONE:

USE OF SEEPAGE METERS TO MEASURE

GROUNDWATER FLOW AT BROOK TROUT REDDS*

Paul J. ~lanchfield'~and Mark S. ~idgw#

' Department of Biology, York University, 4700 Keele St., North York, ON,

CANADA, M3J 1P3

* Harkness Laboratory of Fisheries Research. Aquatic Ecosysterns Research

Section, Ontario Ministry of Natural Resources, Box 5000. Maple, ON,

CANADA, L6A 1S9

'Reprinted frorn Transactions of the American Fisheries Society, volume 125, pp. 81 3-818, with the permission of the American Fisheries Society. AnornaIous influxes of water into unfillecl colledion bags can greatly overestimate volume and flow rate data from çeepage meters. From static tank trials, initially empty collection bags (4,500 mL capacity) attached to seepage meters gained

çignificantly more water relative to bags prefilled to 1,000 mL Data from a study of groundwater flow at redds of brook trout Salveilnus fontmalis in Scott Lake, Ontano, indicate that the use of unfilled bags biases seepage meter data. At these redds, the anornalous influx of water into unfilled bags was significant (intercept of regression equation, y=275 mL); however, this influx was sufficiently reduceâ when prefilled bags were used ml). Our data suggests that even at high flow rates

(22-169 rn~-rn-~*min-'),seepage measures can be inflated by an order of magnitude when initially empty bags are used. Because of this anomaly, previous measures of groundwater flow at brodc trout redds with unfilled bags are probabiy not representative of natural flow rates. Our estimates of groundwater flow at brook trout redds in Scott Lake (6-296 rn~-rn-~-min-')are ver-similar to the range in groundwater flow found in lake and stream redds (444û rn~=m'~~rnin~')by other methods. We suggest the use of prefilled collection bags (to 1,000 ml) and conformity in measurement units (rnl-rn-**min-')when measuring groundwater flow wÎth seepage meters. INTRODUCTION

Spawning areas of brook trout Saivelmus fonthalis are generally characterised by groundwater seepage which provides elevated temperatures for em bryo development (Embody 1 934). Adult females appear to select these areas to construct redds and deposl eggs (Benson 1953; Witzel& MacCrimrnon 1W3).

Although the association bemnspavming sites and groundwater seepage has long been recognised (8.g. White 193û),few studies have quantitatively measured naturally occumng rates of seepage at brook bout redds.

A variety of methods have ben used in the past to deted the sespage sites associated wÏth brook trout spawning. Various field studies have used differences

ôetween ambient Stream temperature and redd site temperature during fall and winter (White 1 930; Benson 1953; Fraser 19&), field obsewations of adjacent riparian seepage areas (Hazard 1932; Witzel & MacCrirnmon l983), dispersion of dye placed on the substrate (Fraser 1982), standpipes (Reiser & Wesche 1Qn), piezometers (Curv 1993), and seepage meters (Carline 1980; Snucins et al. 1992).

Only the last three methods provide quantitative measures of groundwater flow.

A seepage meter is the end section of a barre1 (15 cm long x 57 cm in diameter), the open end of which is pushed into the substrate. The upper dos& end is bored; a stopper is placed in the hole, and a plastic tube is inserted through the stopper. An empty plastic bag attached to the exposed end af the tube is used to collect gioundwater amples (Lee 1977). The seepgge meter allows a direct measure of the upward flux of water at the substrate-water interface but provides no information on the hydraulic conditions beneath the substrate (Lee 1977).

Seepage meters are inexpensive, easy to install. and provide data on water quantity as well as large numbers of sarnples in a relatively short amount of time

(Lee 1977; &langer & Mikutel1985). However, one anomalous feaiure oJ the rnethod has been detectd; an ernpty plastic bag attacha to a seepage meter creates a hydraulic potential resulting in a short-ten influx of water into the bag

(Shaw & Prepas 1989). This influx of water appears to be due to the mechanical properties of the bag and can seriously bias estirnates uf groundwater flow by the

çeepage meter rnethod. Using plastic bags partiaily filled with water can alleviate this problern (Shaw & Prepas 1989).

The 0bjeCtiiive of this study was to charaderise the flow rates of natural brook trout redds in a lake and to determine the potential bias of the seepage meter method when groundwater flow is recorded at brook trout redds This bias is particularly acute at low flow rates and rnay diminish with increasing flow rates

(Shaw & Prepas 1989). Because groundwater seepage in brook trout redds can be relatively high (Carline 1980; Snucins et al. 1992). the potential bias may be relativeiy small. Aiso. collection time may be relatbeiy short where high flow rates exiçt; a situation that Shaw and Prepas (19û9) did not fully address. In a more general mntext, the use of seepage meters may becorne more wideçpread given their ease of use and the growing interest in the role of groundwater discharge as a cornponent of fish habitat Therefore, we believe the solution to the problem of anornafous data kinggenerated Menthe çespage meter rnethod is used needs

to be more wideiy appreciated (Shaw & Prepas 1989).

METHODS

We deteminecf the amount of short-term anomalous influx of water into seepage

meters in the lab and field. Seepage meters were constructed similar to Lee

(1977), the main difference king the drums were made of plastic and the dosed

end hPsl a 3.6 cm hole for the rubber stopper and attachent of the bag

assembly. Other differences to note are as follows: (1) No-namem polyethyfene

bags, 4.500 mL capacity, were attached with an elastic band to TygonN tubing

(length=ô.O cm, 0.80 cm inner diameter); and (2) a plastic oonnector (length=6.3 cm, 0.65 cm inner diameter) tapered at each end (0.80-1.3 cm outer diameter) was ernbeddeû flush to the bottom (depthd.5 cm) of a number 8 rubber stopper.

In both the lab and field trials, seepage meters were allowed to equilibrate before insertion of the rubber stopper assembly, which was attached about 15 min before attachment of the bag and tubing assernbly. For trials with both unfilled and prefilled bags. the end of the flexible Tygon tubing was blocked by a finger and th8 tubing was gently pushed over the tapered plastic connector to achieve a snug fit. To remove the plastic bags, the tubing was pinched and pulled off the plastic connector and a finger placed over the open end of the tubing so no water could escape. The resulting volumes were measured to the nearest milliliter using 500- and 1,-mL graduated cylinders. We used the flow rate calculation of Shaw and Prepas (1989) where q, in ml-rn-2min*'.is calculated as q=

3.92~Vlt;AV is the change in volume of water (ml) in the polyethylene bag when the bag is removed after a given penod of time. t (min). The constant, 3.92, converts the area covered by the seepage meter (2,550 cm2) to 1 m2.

Tank Expeflment

Tank experiments were conducted at Hahess Laboratory of Fisheries

Research, Ontario, 1-2 September 1994. Three seepage meters were placed in a

6304,rectangular tank (length, 2.01 rn; width, 0.57 m; depth, 0.55 m) filled with iake water and were left for about 16 h before measuring water flow. One measure of seepage at each seepage meter was recorded at 5, 10, 15,30.45 and 60 min with unfilled polyethylene bags and bags prefilled to 1,000 mL The tank was refilled and allowed tu settle between subsequent time trials. The seepage meters rested on the bottom of the tank but did not create a seal, such that water could move into the meters.

Field study

Field measures of groundwater seepage were conducted at Scott Lake,

Algonquin Provincial Park, Ontario (4502WN,78O431N). 5-21 September 1994.

Scott Lake is 28 ha in surface area, has a maximum depth of 24 m and wntains a naturally reproducing population of brook trout. In total, four seepage meters

were placed in raswhere brook trout had previously spawned in 1993 (PJB,

personal observation); sites ranged in depth from 1.4 to 1.6 m. The meters wre

inserted by divers to a depth of 12 cm beneath the substrate surface and alloweâ

to settle for a pend of about 16 h before seepage was measured. Seepage

meters were sampled at 5. 10, 15,30,45, 60 and 120 min. For cornparison to

Shaw and Prepas (1Q89), sitespecific data were pooled by time period and were

used in linear regressions of V on t for unfilled bags (AH03) and bags prefilled to

1,000 ml(k77).

For tank and field trials, we tested the nuIl hypotheses that (1) the slope of

the regression of volume on time was equal to zero (linear regression). and (2) the volume of water entering into prefilled and unfilled bags was equal (analysis

of covariance). All analyses were performed on untransformed data.

Groundwater flow was measured at 36 sites just before brook trout

spawned in Scott Lake, 4-17 ûctober 1994. Twenty-three sites represent actual

redds where spawning took place during the 1993 season. The remaining 13 sites represent random sites chosen close to spawning redds and contained sirnilar substrate. Spawning occuned at 10 of the 23 redds during the 1994 season. For comparative purposes, we consider the redds where spawning occuned as 'used" (k10) and grouped the remaining sites as "unusedm(M6).

Seepage meters were installed in the same manner as described for the field expriment. All rneasurements of groundwater flow were made in triplicate and lasted 120 min; collection bags were prefilled to 1,000 mL At one spawning site. the collection time was redoced to 15 min due to extensive flow. One seepage meter remained at one site during the entire collection period to detemine the exient of day-to-day variability in groundwater flow. For comparative purposes we converted other published data of groundwater flow at brodc trout redds. oMained from seepage meters attached to unfilled bags, from units of centirneters per hour to milîiliters per square meter per minute (Carlin0 1980; data wre provided by E. Snucins). We calculated Our flow rate estimates in the units just noted to produce a conversion factor (166.6). To present unbiased masures of groundwater flow, we divided the flow rates from other studies by a factor of

9.57 because this is the ratio of mean volume of water present in unfilied bags

(215 ml) versus prefilkd bags (23 ml) from Our 5-min time period

(unfilled:prefilled=9.57:1). This tirne period was most representative because previous measures of groundwater flow were collecteci over periods of 1-7 min

(Carline 1980). We multiplied other published data of groundwater fiow at brook trout redds, colleaed by different methodology, by Our conversion factor of 166.6. to change units of centirneters per hour to units of milliliters per square meter per minute (Reiser & Wesche 1977; Curry 1993). We prefer the latter measure of flow. because this incorporates an aççociated volume which is far more understandable. RESULTS AND DlSCUSSlON

There was no linear increase in water volume with time for unfilled bags in a static environment (60.18; -.5; df=1.16; W.ûû; Figure 1A). The volume of water present in unfilled bags increased as a function of tirne up to t45min and thereafter showed a marked decrease in volume at 60 min. Overall there was a significantly greater influx of water into unfilled bags than bags prefilled to 1,000 mL (k109.3, df=1,33;P4.001). The additional volume of water present in prefilled bags was relatively constant with tirne. Over the comparable time period

(up to k6û min), our data are consistent with the tank experiments of Shaw and

Prepas (1989) when unfilled bags are used. Tank trials performed by Shaw and

Prepas (1989) with bags prefilled to 1,000 ml, showed an increase in V with t. ln contrast, in our study, V did not increase significantly with t (K0.04;M.61; df=1,16; kû.44) suggesting that attachment of the bag to the seepage meter regulates the flow of water entering the prefilled bag (Figure 1A). The most likely explanation is that water must travel from a much greater distance to reach the bag and is sornewhat restricted by the bottom of the seepage meter resting on the tank floor. As flow rate is a function of volume, there was essentially a logarithmic decrease in Row rate with time when unfilled bags were used (Figure

1B). The pattern was similar with prefilled bags, but to a much lesser degree, whereby the flow rate at 60 min was 1.3 rn~-rn-*~rnirf'(Figure 18). These data suggest that an equilibnum is reached at this point and that prefilling 4,500niL 1- I 1 O 10 20 30 40 50 60 Time (min)

Figure 1 (A) Volume of water collected and (B) flow rate of initially empty (O) and prefilled (a)bags in a water-filled tank versus time intenmi. Each circle is the mean (f SD) of three trials, one per seepage meter. bags to 1,000-rnL is a sufficient volume to reduœ short-term influxes during collection periods of 60 min.

Volume of water collected increased linearly with t in field triais for both unfilled and prefilled bags; however, the volume collected differed greatiy (Figure

24: unfilled V = 275 + 3.95t (Fd.76; -1 6.9; df=l ,1 O1 ; Pc0.001) compareci with prefilled V = 34 + 2.93t (?=0.80; L298.6; df=1,75; Pc0.001). If we compare the yintercepts of the regression equations for unfilled w75) and prefilled

(y=&) trials, aiere is an exceçs of 241 mL entering the unfilled bags. This is consistent with the regression analyses of data from Narrow Lake, Alberta (Shaw

& Prepas 1989), where the difference in anomalous short-terni (30 min) influxes of water from unfilled (237 ml) and prefilled (9 BI)trials was 228 mL The wmpatibility of these two studies provides further evidence that the use of initially ernpty bags on seepage meters can seriously overestimate rates of groundwater flow-

Our data show that anomalous influxes of water occur at relatively high seepage rates as well as at low seepage rates. Data from Buffalo Lake (Shaw &

Prepas 198g) indicated that volume of water in initially empty bags increased linearly with t V = 106 + 2.23t (dk2; pd.97; P~0.05)~such that the yintercept of

106 mL was not significantly greater than zero (df=2; k1.8; A0.1). From these data they detemined that a "high" flow rate (8.7 rnl-m-*-min-')may not affect short-term influxes of water into unfilled bags. However, from our volume data there was a significantly greater influx of water into unfilled bags than bags Time (min)

Figure 2 (A) Volume of water cullected and (B) flow rate of initially empty (O) and prefilled (e)bags at four brook trout spawning redds versus time interval. Each cirde represents the mean (f SD) of 2-21 seepage measures pooled per time period (initialiy empty: N=103; prefilled: k77). prefilled ?O 1,000 mL (k735.3; dk1.177; Pe0.001) with flow rates ranging from

22 to 169 ml-m%nin-'. Therefore, it is important to note that the bisis

substantial even with the relatively high flows at redds. In fact, the mean volume

recordeci from the unfilled bags (424 I158 ml) is sçsentialiy triple that of the

prefilled bags (152 I 106 ml). The critical finding of this study is that when

collection bags (4,500 ml) are prefilled (to 1.O ml), flow rate remains constant

over t (Figure 28).

Wiih prefilled bags, our rates of groundwater flow at brook trout redds in

Scott Lake (6-296 rn~-rn-~min-')appear to be considerably lower Men compared to previous seepage meter studies in lentic systerns (Table 1). This may, in part, be due to the generality of our conversion factor, or perhaps seepage rates in

Scott Lake are naturaliy lower. Flow rates of about 500 mL in 1-7 min observed by Cadine (1980) are very high and rnay not be directly comparable due to the varying sizes of seepage meters used. Flow rates reported by Snucins et al.

(1992) were detemined frorn April to July and were inflated due to spring snowrnelt and increased levels of precipitation. Mean groundwater flow is positively and significantly correlated with mean daily precipitation (Downing &

Peterka 1978). Measures of groundwater flow at brook trout redds detemined with standpipes and pietometers were in the range of 4-340 rn~~rn~~-rnin"(Reiser

& Wesche 1977; Curry 1993), which is very similar to the gradient in groundwater flow we found at Scott Lake, 6-296 mlmq-min*' (Table 2). This lends further evidence that the use of prefilled bags with seepage meters provides a more Table 1 A cornparison of three studies in which seepage meters were used to determine groundwater flow at brook trout redds. Groundwater flow was measured at redds ("used") and at nearby random sites Cunusedn). Row rates from other studies have been reduced by a factor of 9.57 to account for the collection of groundwater with unfilled bags.

Number of Groundwater Flow (rn~-rn-*~min-') Study Spawning Sites Mean + SD Range

This study usa N= 10 49 I 83 6 - 296 unused N=26 716 0-22

Carline used N=5 430 I294 200 - 780 (1 980) unused N=11 40 f 23 17 - 139

Snucins et al. used N= 1 (1 992) Table 2 A cornparison of three methods used to detemine groundwater flow at brook trout redds.

G roundwater Flow Aquatic (ml-m"-min") Study System Method Range

This study lake seepage meters 6 - 296 (prefilled bags)

Curry lake and mini- 4 - 279 (1993) st rearn piezometers (bundles) Reiser and st ream standpipes 4-340 Wesche (1 977) realistic rneasure of natural groundwater flow at brook trout redds.

Piezorneters provide the most extensive analysis of hydraulic conditions in groundwater discharge areas (Lee 1988). However, their use depends on substrate conditions for proper installation and &en requires an assumption of hornogenous subçtrate to facilitate discharge calculaüons (Lee 1977; Cuny 1993). The use of piezometers to examine groundwater discharge in brook trout spawning habitat cm be difficult in some cases because redds can be found in a variety of substrates

(Carline 1980; Fraser 1982) and may not be sensitive enough to detect differences in flow arnong spawning redds (0.g. Curry 1993). Seepage meters mera mudi larger area of substrate than piezometers and incorporate a much greater area of any redd when deterrnining groundwater seepage in brook trout spawning habitat.

Although seepage meters (modified barml ends) have limited use. they provide an excellent method for seepage estimates in terms of water quantity (Downing &

Peterka 1978; Belanger & Mikutel 1985). if used with prefilled collection bags

(this study; Shaw & Prepas 1989). We suggest the use of 4,500-mL collection bags prefilled to 1,000 mL and collection periods of 60-120min for measuring groundwater flow at brook trout redds with seepage meters. Shorter collection periods (10-15 min) are recommended for redds with high flow rates. The authors are espedafly grateful to Kim Hughes and Jackie King who provided field assistance. We thank Doug Brown for logistical support as well as advice.

We are grateful to P. Bisson, R. Mackereth. R. Shaw and E. Snucins for helpful reviews of the manuscript. This study was supported by the Scott Lake Long-

Tem Ecological Research Program of the Ontario Ministry of Natural Resources. REFERENCES

Belanger, T.V., and D.F. Mikutel. 1985. On the use of seepage meters to estimate groundwater nutrient loading to lakes. Water Resources Bulletin 2 1:265-272.

Benson, N.G. 1953. The importance of ground water to trout populations in the Pigeon River, Michigan. Transactions of the Norai Arnerican Wildlife Conference 18:269-281.

Carline, R.F. 1980. Features of successful spawning site development for brook trout in Wisconsin ponds. Transactions of the Arnerican Fisheries Society 109:453-457.

Curry, R.A. 1993. Hydrogeology of brook charr (Sahelhus fontinaiis) spawning and incubation habitats: linking aquatic and terrestrial ecosystems. Doctoral dissertation. University of Guelph, Guelph. Ont.

Downing, J.A., and J.J. Peterka 1978. Relationship of rainfall and lake groundwater seepage. Limnology and Oceanography 23:821-825.

Embody, G.C. 1934. Relation of temperature to the incubation pend of eggs of four species of trout. Transaction of the American Fisheries Society 64281- 289.

Fraser, J.M. 1982. An atypical brook charr (Sahelinus fontinalis) spawning area. Environmental Biology of Fishes 7:3û5-388.

Fraser, J.M. 1985. Shoal spawning of brook trout, Salvelinus fontinalis, in a Precambrian shield lake. Le Naturaliste Canadien 112: 163-174.

Hanard, AS. 1932. Some phases of the life history of the eastern brook trout, Sahelinus fontinalis. Transactions of the Arnerican Fisheries Society 62:344- 350.

Lee, D.R. 1977. A device for measuring seepage flux in lakes and estuaries. Limnology and Oceanography 22: 140-1 47.

Lee. D.R. 1988. Six in situ methods for study of groundwater discharges, p. 556- 566 in The proceedings of the international symposium on interaction between groundwater and surface water. Sigma tryck, Lund. Reiser, D.W. and T.A. Wesche. 1977. Determination of physical and hydraulic preferences of brown and brook trout in the selection of spawning locations. Water Resources Research institute, Completion Report, Water Resources Series No. 64. Laramie, Wyoming.

Shaw. RD.. and E.E. Prepas. 1989. Anomalous. short-term influx of water into seepage meters. Limnology and Oceanography 34: 1343-1 35 1.

Snucins, E.J., RACurry, and J.M. Gunn. 1992. Brook trout (Salvelinus fonthalis) embryo habitat and timing of alevin emergence in a lake and a stream. Canadian Journal of Zoology 70:423427.

White, H.C. 1930. Some observations on the eastem brook trout (S. fontinalis) of Prince Edward Island. Transactions of the Arnerican Fisheries Society 60:101-108.

Wiiel, LD., and H.R. MacCrimrnon. 1983. Redd-site selection by brook trout and brown trout in southwestern Ontario strearns. Transactions of the American Fisheries Society 112:76O-ïil. lMAGE EVALUATIO N TEST TARGET (QA-3)

ec ii lI[E L; - L, 11111A LLL IlIll-"

APPLJED - INIAGE . lnc = t 653 East Main Street --*- - Rochester. NY 14609 USA ------Phone: 716/482-0300 ------Fax: 71612805989

Q 1993. Wied Image. IK. Ail RitsResewed