UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Determinants of fitness in an island population of song sparrows Hochachka, Wesley Michael 1990

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

Item Metadata

Download

Media
831-UBC_1990_A1 H63.pdf [ 7.42MB ]
Metadata
JSON: 831-1.0098774.json
JSON-LD: 831-1.0098774-ld.json
RDF/XML (Pretty): 831-1.0098774-rdf.xml
RDF/JSON: 831-1.0098774-rdf.json
Turtle: 831-1.0098774-turtle.txt
N-Triples: 831-1.0098774-rdf-ntriples.txt
Original Record: 831-1.0098774-source.json
Full Text
831-1.0098774-fulltext.txt
Citation
831-1.0098774.ris

Full Text

D E T E R M I N A N T S OF FITNESS IN A N ISLAND P O P U L A T I O N OF SONG SPARROWS By W E S L E Y M I C H A E L H O C H A C H K A B .Sc , The University of Alberta, 1984 M . S c , The University of Alberta, 1985 A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T OF T H E R E Q U I R E M E N T S F O R T H E D E G R E E OF D O C T O R OF P H I L O S O P H Y in T H E F A C U L T Y OF G R A D U A T E STUDIES (Department of Zoology) We accept this thesis as conforming to the required standard T H E U N I V E R S I T Y OF BRITISH C O L U M B I A October 1990 © Wesley Michael Hochachka, 1990 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Z o o l o g y The University of British Columbia Vancouver, Canada D a t e O c t o b e r 11, 1990 DE-6 (2/88) T A B S T R A C T Patterns and causes of variation in the reproductive success of Song Sparrows (Melospiza melodia) are investigated in this thesis. The two general patterns looked for were: inter-annual variation in reproductive success, and repeatability of reproduc-tive success of individual parents. The specific problems addressed were: (1) whether intra-seasonal variation in reproductive success was the result of differences in the quality of parents or their territories; (2) whether nutritional condition of nestlings affected their subsequent survival; (3) whether variation in morphology of adult spar-rows was influenced by the conditions under which birds grew up; and (4) given the patterns discovered in the first three sections, how trade-offs between present and future reproduction constrain the effort expended by adult sparrows in reproduction. Data used in this thesis came from a long-term (1975-present) descriptive study of the population of Song Sparrows living on Mandarte Island, B .C . , Canada. Data on survival, reproductive success, and nestling and adult morphology were all available. The approach taken in the thesis was to search for systematic variation in the data, and from these patterns to make inferences about cause and effect. The following conclusions are made: (1) the intra-seasonal decline in clutch size, observed in populations of many species of birds, was the result of poor birds or birds on poor quality territories both nesting later and laying smaller clutches; (2) nestlings in better nutritional condition had a higher probability of survival while under the care of their parents, than nestlings in poor nutritional condition; (3) the probability of survival of offspring after they left their parents' care was lower for young born later in the year, but this pattern is not caused by variation among parents or their territories (contrary to the cause of seasonal decline in clutch size); (4) morphology of birds as adults was affected by the environment that birds grew up in, with nutritional condition and population density while a nestling both affecting adult morphology; (5) the effort that parents expend on reproduction is constrained by ability to vary reproductive effort with date of nesting and parental age. T A B L E OF C O N T E N T S Section Page A B S T R A C T ii LIST OF T A B L E S vii LIST OF F I G U R E S ix A C K N O W L E D G M E N T S x C H A P T E R 1. I N T R O D U C T I O N 1 Summary 1 BACKGROUND 2 GENERAL THEMES 3 THESIS ORGANIZATION 5 C H A P T E R 2. S E A S O N A L D E C L I N E IN R E P R O D U C T I V E P E R F O R M A N C E 8 Summary 8 INTRODUCTION 9 METHODS 10 Study Site and Data 10 Hypotheses 12 Rationale for Analyses 13 Statistical Procedures : 17 RESULTS . . 20 Clutch Size 20 Survival of Dependent Young . 24 Recruitment of Independent Young 32 DISCUSSION 32 iii C H A P T E R 3. D E T E R M I N A N T S A N D C O N S E Q U E N C E S OF N E S T L I N G C O N D I T I O N 40 Summary 40 INTRODUCTION 41 METHODS 44 General Methods 44 Statistical Procedures 45 Measuring Nutritional Condition of Nestlings 45 RESULTS 49 Nestling Condition and Survival 49 Parental Ability and Nestling Condition 54 Nestling Sex, Condition and Survival 59 DISCUSSION 59 Annual Variation in Nestling Survival 62 Adjustment of Reproductive Rate 62 Nestlings' Sex and Survival 65 Conclusions 65 C H A P T E R 4. T H E E N V I R O N M E N T A L C O M P O N E N T O F M O R P H O L O G I C A L V A R I A T I O N 67 Summary 67 INTRODUCTION 68 METHODS 69 RESULTS 73 Density-Dependent Morphology 73 Selection on Environmental Components of Size 86 Selection for Parental Care-. 87 Correlations Among Traits 88 iv Parental Care and the Origin of Environmental Variation 88 DISCUSSION 91 Natural Selection and Genetic Response 92 Environmental Variaiton and Adaptive Response 96 Implications 97 C H A P T E R 5 . H O W M U C H S H O U L D R E P R O D U C T I O N C O S T ? 99 Summary 99 INTRODUCTION 100 THE STUDY ANIMAL 102 THE COST OF PRODUCING EXTRA EGGS 102 Reproductive Effort Varies With Female Age 105 Reproductive Effort Invariant with Female Age 109 Summary 114 MULTIPLE COSTS OF REPRODUCTION 114 ESTIMATED COSTS FOR ADDITIONAL SPECIES 119 CONCLUSIONS 121 Variation in Adult Quality , 121 Detecting Costs of Reproduction 121 Considerations for Future Research 122 C H A P T E R 6. C O N C L U D I N G D I S C U S S I O N 124 Summary 124 RECAPITULATION OF THEMES 125 Inter-Annual Variation 125 Repeatability Within Females 127 EVOLUTION OF LIFE-HISTORY 128 IMPLICATIONS FOR FURTHER RESEARCH 129 v Variation Within Broods 129 Multiple Costs of Reproduction 132 L I T E R A T U R E C I T E D 134 A P P E N D I X 1. E S T I M A T I N G R E P R O D U C T I V E S U C C E S S 147 Summary 147 vi L I S T O F T A B L E S Table Page 2-1. Seasonal trends in reproductive success in the sample of years used in the statistical analyses, and in all years for which data are available 19 2-II. Components of reproductive performance in Song Sparrows 21 2-III. Laying date and clutch sizes of yearling and 2+ year old Song Sparrows, for both first and second broods 25 2- IV. Frequency of second nests failing to hatch all eggs, as a function of whether young were fledged from the first nest of the year 29 3- 1. Condition of nestlings and survival to independence 51 3-II. Relationship between nestling condition and date of hatch 55 3-III. Frequency of survival of nestlings to independence '. 57 3- IV. Condition of surviving male and female nestlings 60 4- 1. Variation in heritability among years 74 4-II. Correlations between annual mean values of traits 77 4-III. Correlations between mean offspring size and population density in the year of birth 79 4-IVa. Coefficients of directional survival selection on size and the environmental deviation in size 82 4-IVb. Coefficients of directional survival selection on bill size and the environmental component of bill size 83 4-Va. Coefficients of directional reproduction selection on size and the environmental component of size 84 4-Vb. Coefficients of directional reproduction selection on beak size and the environmental component of beak size 85 4-VI. Correlations between environmental components of size 89 vii 4- VII . Directional selection gradients for overwinter survival of juvenile females from 1975-78 cohorts combined, and for reproductive success of females in 1979 95 5- 1. Correlations of reproductive parameters for birds at different ages I l l 5-II. Estimates of reproductive parameters and theoretical minimum costs of reproduction for four species of birds 120 viii L I S T OF F I G U R E S Figure Page 2-1. Graphical presentation of the analyses used to separate the hypotheses 15 2-2. Seasonal variation in clutch size for 1978 23 2-3. Probability of hatching all eggs in a nest in 1978 27 2-4. Seasonal variation in offspring mass for 1983 31 2- 5. Expected probability of recruitment of independent young as a function of date of initiation of laying 34 3- 1. Relationship between nestling mass and wing length 48 3- 2. Nestling condition and survival 53 4- 1. Relationship between population density in the year of birth, and mean size of female offspring produced 76 5- 1. The theoretical minimum cost of adding one extra egg to each nest (reproductive effort variable with age) as a function of the date that nesting is started in the current year 108 5-2. The theoretical minimum cost of adding one extra egg to each nest (reproductive effort invariant with age) as a function of date of initiation of the first nest of the year 113 5- 3. The theoretical minimum reduction in survival of current offspring for an increase in clutch size of 1 egg for each nest, plotted as a function of date of initiation of first nest 118 6- 1. Effect of variance in nestling condition on parents' reproductive success 131 ix A C K N O W L E D G E M E N T S The most important acknowledgement is of my parents, who set me on my path through life. They instilled in me an interest and wonder of things not made by hu-mans. They encouraged and supported my curiosity and my education, both informal and formal. I am not sure that I am the intended product of their efforts, but I thank them for everything. I have been privileged to associate with those present over the course of my studies at U B C . Foremost in this group is my supervisor, Jamie Smith. He has been generous with time and resources, but most importantly with trust in my abilities. The other members of my research committee — Lee Gass, Dolph Schluter, Tony Sinclair, and Nico Verbeek — all aided greatly in my work. My thought and word have been rendered more lucid, and my perspectives challenged and (sometimes) changed. Jamie Smith and Peter Arcese made my research possible by introducing me to the Song Sparrows, and allowing me access to the data collected by them and their assistants. It is impossible to thank every individual for their contribution to this thesis; many people, though not forgotten, must remain un-named. I apologize for your anonymity. Thanks to all those who abided me for long periods while working on Mandarte: Alice Cassidy, Marie-Josee Houde, Gwen Jongejon, Anne Labbe, Marlene Machmer, Arne Mooers, and David Westcott. I was not the most congenial of people after several days of sitting on ladders, and the tolerance and diligence of these people not only made my fieldwork possible but also more enjoyable. In the office John Eadie, Dick Repasky, Locke Rowe variously prodded, poked, provoked, and listened to me more times than I was worth. Conny Askenmo and Richard Pettifor commented on incarnations of Chapter 2. Rauno Alatalo provided an unpublished manuscript for, Gayle Brown helped in data munching towards, and Anne Hedrick and Arie van Noordwijk made suggestions on Chapter 4. Thanks to the BDC's denizens for making it such a pleasant home. Andre Breault, Nancy Butler, Bob Gregory, Charlene Higgens, Xavier Lambin, Beth Scott, and Peter Watts helped keep thinks in perspective. Thank-you all. I am grateful for the Tseycum and Tsawout Indian bands for allowing us the priv-ilege of living and working on Mandarte Island. The Natural Sciences and Engineering Research Council of Canada funded this research through operating grants to Jamie Smith, and a post-graduate scholarship to myself. Personal support was also received from the Department of Zoology, U B C , in the form of a teaching assistantship. x 1 C H A P T E R 1 I N T R O D U C T I O N Summary - The form and content of this thesis was influenced by the fact that my research forms part of a long term research project. The thesis describes some factors that affect the reproductive rate of a population of Song Spar-rows (Melospiza melodia). Two general themes pervade the thesis: (1) an examination of inter-annual variation in reproductive parameters, and (2) a search for consistent differences in ability among parents. The four central chapters discuss: (1) the causes of intra-seasonal variation in reproductive success, (2) the relationship between nestling nutritional condition and sub-sequent survival, (3) factors influencing environmental variance in the adult size of birds, and (4) an analytical method for estimating the relationship between an adult's reproductive rate and subsequent survival. 2 This thesis is concerned with the evolution of life-history. Below, I explain the perspective from which I examined this general topic. BACKGROUND The greatest single influence on this thesis is that the work was done as part of a long-term study. By "long-term study" I mean a study that has continued longer than the lifespans of several generations of the study animal. In the present case, the data come from a population of Song Sparrows (Melospiza melodia) that live on Mandarte Island, B .C . , roughly 25 km N N E of Victoria. This population has been studied since 1975, and the majority of sparrows from 14 annual cohorts have been born and died over the period of observation. One principal reason for conducting protracted studies is to measure lifetime reproductive success of individuals, the best available measure of Darwinian fitness (Newton 1989, p. 1). The rationale is that variation in fitness can be correlated with variation in other traits, allowing us to identify the factors that have shaped the forms and lives of the study organism. However, as Grafen (1988) notes, there are limits to what can be inferred from descriptive data. These data tell us about present selective pressures, not the past selection that led to current form. Results of many long-term studies have recently been published (e.g. Woolfenden and Fitzpatrick 1984, Koenig and Mumme 1987, Clutton-Brock 1988, Newton 1989), and the results of these studies have produced some general statements about repro-ductive success. The data show that individuals are far from equal: in most cases the majority of individuals in a population do not leave any descendants while a small fraction (15-30%) of individuals actually maintains the population (Newton 1989, pp. 441-445). Longevity (Clutton-Brock 1988, pp. 473-479) and the survival of offspring after the end of parental care (Newton 1989, pp. 447-449) are the most important components of lifetime reproductive success of those individuals surviving to repro-duce. 3 These studies also make two observations that provide the two linking themes in this thesis. One observation is that lifetime reproductive success is very dependent on when an animal is alive (Newton 1989, p. 454). Animals born to different cohorts deal with different environments, and may require different characteristics in order to succeed (e.g. Gibbs and Grant 1987). The second observation is that some animals may be blessed with a "silver spoon" (Grafen 1988): some individuals in a cohort may survive longer, produce more progeny each year, and consistently have higher survival of their progeny — all due to some accident of rearing or environment (e.g. Hogstedt 1980, 1981) with no genetic basis. These successful individuals could be successful, no matter what cohort they were born into. Thus, the traits that we study may not be independent of each other (e.g. Reznick 1985). G E N E R A L T H E M E S This thesis examines several factors that may or do affect the reproductive success of adult Song Sparrows. My aim is not to catalogue all influences on reproductive success, but to test whether some recent ideas apply to Song Sparrows. Two general themes underlie the material presented here. While these two themes are not central to the body of the thesis, nevertheless they have shaped my research. One theme is the presence of inter-annual variation in reproductive success and in factors that influence reproductive success. I have not explored the consequences of this annual variation. Instead the thesis documents the presence of inter-annual varia-tion and describes patterns in this variation. For instance, we already know that much of the variation in reproductive success in the Mandarte population of Song Sparrows is the result of changes in population density between years (e.g. Arcese and Smith 1988, Hochachka et al. 1989). This thesis looks for density-dependence in traits for which it has not been previously documented. As well, I look for patterns that are qualitatively, although not quantitatively, similar among years. Chapter 2 shows that the pattern of intra-seasonal variation in reproductive rate (e.g. clutch size, survival 4 of fledged offspring) differed between years. However, one pattern — lower probabil-ity of survival of later born offspring — predominated. Inter-annual variation in the nutritional condition of nestlings is described in Chapter 3. Variation in nutritional condition of nestlings was not density dependent. As in Chapter 2, a pattern emerges that is relatively independent of annual variation: nestlings in better nutritional con-dition had a higher probability of survival. Chapter 4 looks at inter-annual variation in morphology, particularly environmentally induced variation in morphology. Body size was affected by the population density in the year of a bird's birth. Although morphology is occasionally correlated with reproductive success in this population of Song Sparrows (Schluter and Smith 1986a), body size was not a consistent predictor of reproductive success. Finally, Chapter 5 uses the patterns noted in the previous three chapters to explore the "cost" of reproduction (increased mortality, or decreased future reproductive success) experienced by Song Sparrows on Mandarte Island. The second theme is a search for evidence of "silver spoons". I am focusing on whether some parents are consistently more successful than others, either because of intrinsic differences or due to occupying territories of consistently different quality. This pre-disposition for success may not have a genetic basis. Detecting consistently high or low reproductive success is central to Chapter 2. In this chapter, I show that some females consistently lay larger clutches than others. However, the survival of these eggs to produce breeding offspring depends on the date the eggs were laid, and not on any intrinsic differences among parents. Chapter 3 shows that parents do not repeatedly produce nestlings in either good or poor nutritional condition — again, no "silver spoon". Likewise, Chapter 4 shows that parent sparrows do not consistently produce offspring either larger or smaller than their expected, genetically determined size. These results indicate that, aside from clutch size, the results of sequential nesting attempts are relatively independent. This knowledge was used in constructing the model used to estimate reproductive costs in Chapter 5. 5 In summary, this thesis provides ample evidence of inter-annual variation in re-productive success and morphology of Song Sparrows, but little indication that birds are either consistently good or poor parents. T H E S I S O R G A N I Z A T I O N As noted above, this thesis covers four distinct topics that are loosely linked by two themes. While there is continuity between chapters, the Introduction and Discussion of each chapter focuses on a separate topic of current interest in ecology. Chapters 2 - 5 were written first as independent scientific papers, and have been occasionally modified by the inclusion of extra information on subsidiary issues in the thesis versions. As well, basic information about Mandarte Island and the study is only provided in the Methods in Chapter 2. Chapter 2 deals with an issue that has long puzzled ornithologists (e.g. Klomp 1970), the observation that the size of clutch produced in a population declines the later the clutch is laid in a season. More recently, it has also been noted that the probability of survival of offspring to the next year also generally declines, the later that young are born (e.g. Newton and Marquiss 1984, Nilsson and Smith 1988). Both of these observations suggest that there is an advantage for parents to nest as early as possible in a breeding season, and that natural selection should favour early-nesting birds. Given consistent selection for early nesting, any genetic variance in laying date should disappear: observed variation in laying date should not have a genetic basis. However, date of laying often has a relatively high heritability (van Noordwijk et al. 1981, Findlay and Cooke 1982). A recent paper by Price et al. (1988) provides a possible explanation for this paradox, but only circumstantial evidence is provided to support their case. They suggest that early laying -per se is not advantageous, but that some third factor (e.g. availability of food on a territory) causes early laying and higher reproductive success to co-occur. Chapter 2 of this thesis is the first explicit test of Price et al.'s (1988) hypothesis of which I am aware. 6 At issue in Chapter 3 is whether the amount of effort that parents expend in raising young can influence the probability of their offspring's survival. Many studies show that nestlings that are better fed also have a higher probability of survival (e.g. Garnett 1981, Smith et al. 1989). The demonstration is usually that nestlings that survived were heavier on average than those that died. However, there is still one important piece of information missing, if we actually want to find the optimal trade-off between number of offspring raised and the quality of each offspring. Smith and Fretwell (1974) showed that the optimal brood size is determined by the exact shape of the curve relating nestling mass (and presumably parental effort) to nestling survival. In Chapter 3, offspring mass — survival curves are calculated and implications for optimal brood size are discussed. Chapter 4 deals with another aspect of parental care: the influence that parents have on the morphology of their offspring. Studies have shown that a bird's size or shape influences its survival or reproductive success (e.g. Price et al. 1984, Gibbs and Grant 1987, Schluter and Smith 1986a). Hence an offspring's size can influence its parents' fitness. There are ample demonstrations that parents affect the morphology of their offspring through the genes the offspring inherit (e.g. Smith and Dhondt 1980, Boag 1983, Dhondt 1983, van Noordwijk et al. 1988). Another intriguing possibility has recently received attention: not only the genetic variance in offspring size, but also the environmental variance may be under the parents' control. Parents may be able to affect the size of their offspring through the amount of care (probably food) that parents give their offspring (Boag 1987). Offspring may also be "hard-wired" to reach different final sizes in different environments, through a genetic mechanism referred to as a reaction norm (e.g. Stearns 1989). Chapter 4 discusses whether adult Song Sparrows control the morphology of their offspring through the parents' influence in the environmental component of their offspring's size. Chapter 5 integrates information presented in the previous three chapters. A topic 7 that has recently entertained many ecologists is the trade-off that parents face between reproducing now, or lowering their present reproductive rate and potentially increas-ing their future reproductive success. Avian ecologists, in particular, have expended considerable effort in attempts to demonstrate the existence of a "cost of reproduc-tion" (e.g. see reviews in Linden and M0ller 1989, Partridge 1989). The question still produces heated debate (contrast Nur 1984a, b, with Pettifor et al. 1988). I believe that most people have missed the point. Whether there is a cost to reproduction is not the important issue. There has to be some cost. The important point is how large the cost is. We want to know whether the cost of reproduction is large enough to influence reproductive rate. I believe that empirical measures are not the best way to determine the cost of reproduction because prohibitively large sample sizes may be needed to accurately estimate the cost. Instead, I suggest that a better method is to estimate the cost of reproduction analytically from a few simple assumptions and basic demographic information about the population in question. At the very least, this method helps to develop an appropriate design for a field experiment to test for a cost of reproduction. To my knowledge, this approach has not been taken before. Chapter 5 uses the patterns identified in the previous three chapters in an analytical analysis of the cost of reproduction in Song Sparrows. I close (Chapter 6) with a brief return to the two themes noted above, summarizing the findings relevant to these two topics. Finally, I note some areas of further research suggested by the results of this thesis. 8 C H A P T E R 2 S E A S O N A L D E C L I N E I N R E P R O D U C T I V E P E R F O R M A N C E Summary - Intra-seasonal variation in reproductive success and nestling size were studied in a population of Song Sparrows (Melospiza melodia). I tested whether sea-sonal variation in reproductive success was due to differences among parents or terri-tories, or whether the variation was felt by all individuals in the population. Clutch size declined in the population through the breeding season, and the decline was due to some classes of females both nesting later and laying smaller clutches. Later laying and smaller clutches of yearling females was a major contributor to seasonal decline in clutch size, although seasonal decline in clutch size was also observed within age classes of females. Hatching success was roughly 10% lower for second broods than first broods, but did not vary with date that the nest was initiated. Survival of off-spring did not vary seasonally during the period that offspring were in the care of their parents. Although nestling size was greater for birds born later in the breeding season, the probability of independent offspring entering the breeding population was lower for young born later in the breeding season. The seasonal decline in recruitment of offspring was not due to differences in the quality of territories or parents. 9 I N T R O D U C T I O N Intra-seasonal variation in rate of reproduction, especially intra-seasonal decline in clutch size (e.g. Klomp 1970), is almost ubiquitous in birds. Probability of sur-vival of fledged young also commonly decreases for young hatched later in the season (e.g. Perrins 1966, van Haartman 1966, Cooke et al. 1984, Dow and Fredga 1984, Newton and Marquiss 1984, Nilsson and Smith 1988), although van Noordwijk et al. (1981) present a case where the relative success of early nests varied between years. Many hypotheses have been advanced to explain intra-seasonal variation in clutch size (e.g. Klomp 1970, Murphy 1986), but there have still been few explicit tests of these hypotheses (Murphy 1986). We are even unsure if seasonal variation in reproductive success is due to some factor that affects all individuals in a population, or to differ-ences among individuals or the territories that they occupy. Population effects could arise if seasonal trends existed in a factor like food abundance (e.g. Lack 1966), while individual effects could result from birds with low reproductive capability nesting later than more capable birds. Distinguishing between individual- and population-level causation of seasonal vari-ation is important if we are to understand the evolution of avian life-histories. For example, Price et al. (1988).present a model to explain why considerable genetic vari-ance in laying date is often observed, in spite of constant selection favouring earlier laying. They propose that laying date per se is not directly linked to reproductive success; rather, a third factor (e.g. availability of food) acts independently on both laying date and reproductive success. The only unequivocal way to find whether laying date is directly linked with reproductive success is to vary the laying date of a bird without altering any other aspect of the bird or its environment. If reproductive success conforms to that expected from the bird's new, altered laying date, then date and reproductive success are directly linked. If birds have the same reproductive success regardless of date of nesting, 10 then variation among parents or territories is implicated as the cause of intra-seasonal variation in reproductive success. While it is impossible to alter only laying date, an approximation is provided by multiple-brooded species; the same bird nests at two or more different dates in one breeding season, generally on the same territory. Individual birds will either have consistently high or consistently low reproductive success relative to other birds nesting at the same time, if intra-seasonal decline in reproductive success is the result of differences among parents or their territories. Likewise, if date is the major predictor of reproductive success, then adults will not consistently produce above average or below average nests. Instead, reproductive success will be directly related to date of nesting (see Figure 2-1). In this chapter, I present a detailed examination of intra-seasonal variation in several components of reproductive success of a multiple-brooded population of Song Sparrows (Melospiza melodia); I determine whether variation in reproductive success is linked with date, or whether laying date and reproductive success are independently controlled by a third factor. I examined variation in clutch size, and in survival of offspring from hatching until offspring entered the breeding population. A n intra-seasonal decline in clutch size of Song Sparrows was observed, and this decline was the result of differences between breeding sparrows in either their intrinsic quality or the quality of their territories. Young born later in the breeding season had a lower probability of recruiting into the breeding population, and this decline was related to date of nesting and not to differences among birds or territories. The decline in recruitment of offspring was due to events occurring after young left the care of their parents. There was no indication of seasonal variation in the survival of young sparrows while they were under the care of their parents. M E T H O D S Study Site and Data Song Sparrows were studied on Mandarte Island, off the south-east side of Vancou-11 ver Island, B .C . , Canada, roughly 90 km S of Vancouver and 25km N N E of Victoria. Mandarte is small, (roughly 100m x 700m) rocky island, covered by grassy meadows and shrub (in which the Song Sparrows live). Mandarte is isolated, roughly 1.3 km from the nearest neighbouring island. For a description of Mandarte, see Drent et al. (1964) and Tompa (1964); general study methods are outlined by Smith (1981a). This entire population has been studied intensively from 1960-63 (Tompa 1964), and from 1975 to the present (e.g. Smith et al. 1982, Schluter and Smith 1986a, Arcese 1989), although data from 1980 are incomplete. Data used in this thesis come from the pe-riod 1975-89. A l l adult sparrows were uniquely colour-banded, and their survival and breeding activity were closely monitored. Nearly all successful nests were located, with young being banded as nestlings. This allowed all nests and young to be assigned to specific parents and allowed survival of fledged young to be followed. This population is almost entirely self-contained, with only about 3% of all breeders being born off the island over the course of the study. Almost all sparrows in this population are multiple-brooded; 73% of females that successfully produced 1 brood of young started another nest during a breeding season. Many of the birds that did not attempt to raise another brood, ran out of time for another attempt. For example, one female produced 6 failed nests before successfully fledging young at the end of the breeding season. Thus, variation in the number of broods that pairs raise in a season does not appear to result from pairs attempting to raise different numbers of broods, but from all pairs attempting to raise as many broods as possible with different pairs experiencing different rates of nest failure. The maximum number of brood successfully raised in one year by a single pair is four (Smith 1982). Five sequential measures of reproductive rate were made for each nest: clutch size, probability of survival of eggs to hatching, probability of survival of hatchlings to banding age (generally when nestlings were 5-7 days of age, in a 10-11 day nestling 12 period), probability of survival of banded young to independence from parental care (ca. 30 days of age), and probability of recruitment of independent young to the breed-ing population (generally at 1 year of age). In addition, three measures of nestling size were taken: mass, wing length, and tarsus length. A l l size measures used in this study were mean values for all offspring in a nest, and all size data were corrected for variation in nestling age at the time of measurement, using linear regression. The nestling mass and wing length data were used to compute an index of nestling condi-tion, defined as the residual from a cubic regression of nestling mass on nestling wing length (see Chapter 3 for details). Hypotheses I looked for evidence of seasonal variation in each of my measures of reproductive performance, and attempted to determine whether variation was due to differences between adults/territories (i.e. individual effect) or to an effect that varied with time and influenced all individuals (i.e. population effect). I distinguished between these hypotheses using the following regression model. If seasonal variation in reproductive success is due to differences in quality of birds or their territories, a bird that is relatively successful in raising its first brood should also be relatively successful in raising a second brood; likewise, other birds should be unsuccessful in both first and second broods. Timing of first and second broods within years was strongly, positively correlated (r=0.68, n=245). Thus, if early first broods are more successful than late first broods, early second broods should be more successful than late second broods. The result would be significant seasonal decline in reproductive success, with first and second broods displaying parallel seasonal trends (Figure 2-la). Alternatively, if seasonal variation in reproductive success is solely a function of time, the success of a first brood should have no bearing on whether the second brood is successful. Thus, there should be significant seasonal variation in reproductive success without the trends for first and second broods being offset from each other (Figure 2-lb). K 13 variation occurs both among individual birds and to the population as a whole, the two sources of variation will be manifested by parallel trends in seasonal variation for first and second broods (as in Figure 2-la), but with first and second broods of any female not having the same reproductive success (i.e. the two series of letters in Figure 2-la would not be horizontally aligned). Note that this model assumes that if territory quality changes through the season (i.e. if food supply declines through the breeding season), this seasonal variation in territory quality is experienced by all breeding birds. This assumption appears valid for the Song Sparrows on Mandarte Island because of the island's small size (100 x 700 m), and relatively uniform habitat (Tompa 1964). Although the success of a first brood could affect the success of the second brood, there is no indication of this phenomenon in this population of Song Sparrows (Nol and Smith 1987). Rationale for Analyses The analysis of data presented two major difficulties; as a result, I will present the rationale for a number of the analyses presented in this chapter. The first problem was in the definition of the term 'brood'. Because over 40% of all nests failed to produce offspring old enough to be banded, a bird raising a second brood of young may have produced from 2 to 5 or more clutches of eggs. However, in order to provide a complete picture of the reproductive strategy of an individual bird, one must be able to link the clutch size of a nest, with the subsequent fates of the offspring from that clutch. Thus, the 'second clutch' of a bird should refer to the clutch that produced the second group of offspring that left their parents' care (the 'second brood'). Most nesting attempts that failed did so before young were banded. I have eliminated any nesting attempts that failed to produce at least 1 banded (six day old) offspring from the analyses that follow. A result of excluding nests that failed to produce banded young is that the same set of nests was used in all analyses in this chapter. Nest abandonment or starvation of nestlings almost never caused the failure of nests to 14 Figure 2-1. Graphical presentation of the analyses used to separate the hypothe-ses. Letters (a - e) represent different females, and subscripts represent first and second broods. Because time of first and second broods is correlated, variation in success of first broods will be mirrored by variation in success of second broods when this variation is caused by differences in the quality of parents or territories (a). When seasonal variation is due to some factor correlated with nesting date such as food supply, first and second broods will follow the same line (b). Note that this figure illustrates the expected patterns when a linear decline is observed; the causes of other patterns of seasonal variation can also be differentiated by the same method. Reproductive Success 0) / cr O 0) CD O • / o / Q. 0 / • / / cr ro 03 ro 0 • / / a ro O ro ro CD CO CQ 0) / CT / O of Q. cr CD o ro / CL ro / 0 ro 00 16 produce banded young (personal observation); this leaves predation as the probable cause of most nest failures. Probabilities of survival and recruitment reported in this chapter are higher than per nest survival rates if nests attacked by predators were included. Eliminating nests that failed before young reached banding age did not eliminate a source of seasonal variation in reproductive success; there was no seasonal variation in frequency of early failure of nests (P>0.90, logistic regression). The second problem was quantifying loss of offspring to look for seasonal variation in the rate of loss. This problem has a dual origin: statistical non-independence of offspring in a nest, and dependence of most measures of rate of loss of offspring on both number of young lost and number of young initially in the nest (e.g. the loss of 1 of 1 offspring is very different from the loss of 1 of 5). The probabilities of eggs and young surviving to independence from parental care were not statistically independent, showing a significant tendency towards either all offspring or only one offspring surviving from a nest (nests that failed completely are included in an analysis of this problem in Chapter 3, Table 3-III). Thus each nest had to be treated as a single data point in my analyses. Attempts to assign numerical values to the degree of loss of offspring run into the difficulty of deciding whether it is the number of offspring that are lost, or the percentage that are lost that is most relevant. Further, once this decision is made, the amount of loss depends on the number of young present to begin with; hence, separate analyses must be conducted for each starting brood size. In this study, I could not do this, due to small sample sizes for individual brood sizes. Therefore, nests were classified as either: fully successful (success=l) if all offspring survived the period in question (e.g. if all eggs produced hatchlings), or as partially successful (success=0) if any of the offspring failed to survive the period in question. Thus, the analyses of offspring survival (loss) address the question: 'what is the likelihood that parents will lose some (any) young during the period in question'. Although it is qualitative, this classification scheme does reflect quantitative differences between fully 17 and partially successful nests; fully successful nests had significantly higher absolute numbers of offspring than partially successful nests at all stages in the nesting cycle (ANOVAs, P<0.05). The probability of survival of an offspring from independence to recruitment was statistically independent of the survival of its siblings. The frequency distribution of the number of offspring recruiting from broods of a given size did not differ from that expected by chance (null distribution calculated by binomial expansion; G-tests, P>0.50; separate analysis for each brood size). Thus, individual offspring were treated as the data points in my analysis of survival of independent young to entry into the breeding population. Statistical Procedures Seasonal variation in clutch size and nestling size were examined using analysis of covariance ( B M D P 2 V ; Dixon et al. 1983), with date as covariate and year and brood number (first or second brood) as factors in the analysis. Survival data were treated as binomial (i.e. either 0 or 1), and were analyzed using stepwise logistic regression ( B M D P L R ; Dixon et al. 1983), with maximum likelihood ratios being used as the criteria for goodness of fit of the factors. Logistic regression is able to take data on survival of individuals (binomial data), and estimate the most likely probability of survival for the population. The analysis is akin to computing by hand probabilities of survival over periods of time each with a large number of observations; logistic regression calculates probabilities for infinitely small periods of time making use of information from time periods immediately around the one in question. Data from several years were combined in the analyses in order to enlarge sample sizes. Inter-annual variation was factored out in the statistical analyses. Data from 1977, 1978, 1982, 1983, and 1984 were used for the analyses of clutch size and probabilities of survival. These particular years were chosen as they were the years with the largest numbers of nests (n=44-79 nests). Years with smaller sample sizes were excluded 18 from the analyses because of the existence of annual variation: including many years with small sample sizes would make it less likely for real inter-annual variation to be detected (Type II error). For this reason, there was greater likelihood of confounding inter-annual and intra-seasonal variation in reproductive success if data from all years were combined in the A N C O V A s and logistic regressions. Data on nestling size came from 1983-86, as these were the only years for which full data sets were available at the time of analyses. Dates were standardized to a mean of 0 and standard deviation of 1 within each year, to control for differences in start and duration of breeding seasons (average date varied by up to 24 days between years, and s.d. ranged from 19.4 to 30.6 days). Only first and second broods were considered in most analyses, even though Song Sparrows in this population occasionally raise up to 4 broods of young in a season (Smith 1982). Too few third and fourth broods existed to include them in analyses, except for brief mention of the sizes of clutches from third broods. The results of the analyses of covariance and logistic regression are meaningful only if the sub-set of years used is representative of the entire data set. Graphical inspection of the data showed that patterns detected by A N C O V A and logistic regres-sion were typical of the patterns found in all years of this study. Table 2-1 summarizes the graphical results, presenting the numbers of years in which each reproductive pa-rameter (e.g. clutch size, probability of survival to independence) tended to increase and decline for later nests. For each parameter, ratios of increasing:declining years for the 5 years used in A N C O V A s and logistic regressions are similar to the ratios found for all years combined. Note that the data in Table 2-1 represent seasonal trends, and not necessarily patterns of statistically significant variation. Second broods were not strictly independent of first broods, because pairs of first and second broods were produced by the same females; however, statistical dependence probably had little effect on the interpretation of the analyses. Non-independence of clutch sizes from a female would result if the residuals from a regression of clutch size on 19 Table 2-1. Seasonal trends in reproductive success in the sample of years used in the statistical analyses ("sample"), and in all years for which data are available ("total"). The table shows the number of years that each of the parameters (e.g. clutch size, hatching success) increased and decreased for nests later in the year; seasonal trends for first and second broods are shown separately. Sample sizes vary between measures because years were omitted from the table when no clear seasonal trend was evident. The trends of increase and decline were determined from cubic splines (Schluter 1988) through the data; tendencies for seasonal variation within years are not necessarily statistically significant. increase First Brood decline Second Brood increase decline clutch size sample total 1 3 4 9 0 2 5 10 hatch success sample total 3 7 2 5 4 9 1 3 survival to banding sample total 2 5 2 5 2 6 2 4 survival to s a m P l e independence t M 2 4 3 8 2 4 3 7 recruiting success sample total 2 6 3 5 1 2 4 8 20 date were correlated for first and second broods of the same female. These residuals (after the effect of inter-year variability was factored out) were weakly and almost statistically significantly correlated (r=0.14, n=179, P ~0.06). This result indicates that birds that laid large first clutches were likely to lay large second clutches, although there was considerable variation around this pattern. Nestling size and condition were regressed against date, and the residuals from these regressions were compared for first and second broods of the same female. The correlations between first and second broods were also small and no correlation was statistically significant (all T<0.05, n=88, P>0.50). In addition, there were no significant relationships between the probability of survival of young from first and second broods of a female: 2x2 contingency tables comparing the frequencies of success and failure of first and second broods of the same female showed no significant relationship between the success of first and second broods, at any stage of the lives of offspring (all P>0.25). The results of the logistic regression analyses were not altered when the success of first broods was forced into the logistic regression analyses; thus only the results of the simpler analyses are presented here. R E S U L T S Clutch size Average clutch size in the population changed through the season. However, the pattern of intra-seasonal variation in clutch size of individual females was not the same as the pattern observed when the population was considered as a whole. For the population as a whole, clutch size declined significantly with date; after the effect of date was factored out, clutch size also varied significantly between broods ( A N C O V A ; Table 2-II). This result indicates that clutch sizes of individual females varied systematically through the season; individual females laid larger clutches in the middle of the season than either at the beginning or end. Birds that produced the earliest, largest first clutches, also produced the earliest, largest second clutches Table 2-II. Components of reproductive performance in Song Sparrows. Presented are results of analysis of covariance (clutch size and nestling size), and stepwise logistic regression (egg and offspring survival). Note that in no data set was the interaction between Year and Brood Number statistically significant. Nestling mass is the only nestling size variable shown; the patterns of results for nestling mass were the same as those for tarsus and wing length. Sample sizes are n=336 for measures of pre-independence reproductive success, n=647 for independent survival of independent offspring and n=271 for nestling size and condition (see Chapter 3 Methods for a detailed explanation of condition the index). Survival, Survival, Egg Hatching Nestling Clutch Size to Hatching to Banding Mass Survival, Survival, Nestling Banding to Independence Condition Independence to Recruiting Effect of Date Variation with Date P=0.0008 decrease P=0.50 P=0.32 P<0.001 P=0.11 increase P=0.25 P<0.001 decrease Differences Between Broods P=0.0005 P=0.03 Brood Number clutch second broods P=0.16 Variation with larger second lower hatching — P=0.29 P=0.04 higher for second broods P=0.92 P=0.56 Differences Between Years P<0.0001 P=0.005 P=0.30 P<0.0001 P=0.06 P<0.001 P<0.001 to 22 Figure 2-2. Seasonal variation in clutch size for 1978 (year with the single largest data set); results for other years are similar. Circles and solid line are for clutches from first broods; open diamonds and dashed line are for clutches from second broods. Lines drawn by separate linear regressions for first and second clutches. 24 (Figure 2-2). Therefore, I conclude that the seasonal decline in clutch size was due to variation among birds or the territories that they occupied (Figure 2-la). Two causes of seasonal variation in clutch size will be considered in turn. One commonly advanced explanation for intra-seasonal decline in clutch size in populations is that young birds typically nest later and lay smaller clutches (e.g. Klomp 1970). This was at least a partial explanation for the seasonal decline in clutch size of Song Sparrows. Yearling Song Sparrows laid significantly smaller and later clutches than older females (Table 2-III). If differences between yearling and older females were the sole reason for the variation in clutch size, then the analysis of covariance reported in Table 2-II should show no significant effects of date on clutch size, when the analysis is repeated separately for yearling and adult females. However, declines in clutch size with date were still detected in these analyses, although these declines only approached statistical significance (P=0.06 for yearlings, and P=0.08 for older females). Thus, age alone accounts for much of the observed intra-seasonal variation in clutch size. However, factor(s) in addition to age also contributed to the intra-seasonal decline of clutch size for first or second broods, in this population. An intra-seasonal decline in clutch size (for first or second broods) was observed only when the population was considered as a whole. The second clutches of individual birds were almost significantly larger than clutches for first broods, by an average of 0.13±0.7 eggs (n=166, P=0.02, paired i-test; with multiple comparisons, the critical value for accepting significant differences was P=0.017 with Bonferroni correction for 3 comparisons). As well, clutches from third broods were significantly smaller than second clutches of the same female (0.39±0.8 eggs, 72=31, P=0.008, paired i-test), but third clutches were not significantly smaller than first clutches (smaller by 0.13±0.8 eggs, n=31, P—0.35 paired tf-test). Survival of dependent young The probability of all eggs in a nest hatching declined significantly through the 25 Table 2-III. Laying date and clutch sizes of yearling and 2+ year old Song Spar-rows, for both first and second broods. Data are presented as x±a.^.(») . Note that dates are standardized to x=0, and s.d.=l in each year. Two-way A N O V A s on clutch size and laying date (done separately for data from first and second broods, with female age and year as main effects) showed year-ling females to lay significantly smaller and later clutches, for both first and second broods (P<0.03 in all cases). First Brood Second Brood yearling 2+ year yearling 2+ year female female female female Clutch Size 1977 3.3±0.7(21) 3.4±0.6(19) 3.9±0.5(13) 3.8±0.4(11) 1978 3.5±0.7(22) 3.9±0.5(22) 3.5±0.8(11) 3.8±0.6(15) 1982 3.5±0.5(17) 3.7±0.5 (9) 3.7±0.5(12) 3.7±0.5 (6) 1983 3.2±0.6(28) 3.3±0.6(22) 3.3±0.7(15) 3.5±0.6(15) 1984 3.0±0.5(21) 3.3±0.6(27) 2.9±1.0(13) 3.5±0.7(17) Laying Date 1977 -0.4±0.5(21) -0.8±0.4(19) 1.2±0.7(13) 0.7±0.5(11) 1978 -0.4±0.8(22) -0.7±0.6(22) 0.8±0.5(11) 1.0±0.6(15) 1982 -0.6±0.9(17) -0.9±0.5 (9) 0.7±0.5(12) 0.6±0.3 (6) 1983 -0.4±0.8(28) -0.8±0.9(22) 0.8±0.6(15) 0.3±0.7(15) 1984 -0.3±0.7(21) -0.9±0.6(27) 1.1±0.5(13) 0.7±0.4(17) 26 Figure 2-3. Probability of hatching all eggs in a nest in 1978; results for other years are similar. Symbols are as in Figure 2-2; points on top of the graph represent nests where all eggs hatched, and points on bottom are for nests where at least 1 egg failed to hatch. Probabilities determined by logistic regression. 0.60 Probability of Hatching 0.65 0.70 0.75 0.80 o o & IV) 1 3 o O 2 . CD o o 3-S _ o o" 3 CD O 28 breeding season, and this was entirely an effect of variation that affected the whole population. Hatching success did not consistently differ among birds. The only factor significantly related to hatching success was brood number; probability of hatching was significantly lower (ca. 10% lower) for second broods than for first broods, when data were analyzed using logistic regression (Table 2-II, Figure 2-3). The lower hatching success of eggs from second broods did not appear to be the result of the stress of raising a first brood, because hatching success was very similar for second nests whether the previous nest produced young of banding age, or failed to produce young (Table 2-IV). There was no significant seasonal variation in the probability that hatchlings would survive to banding age (Table 2-II). Neither did raising a previous brood of young affect nestling survival, because survival of the nestlings was not significantly related to brood number (Table 2-II). The average size of nestlings (both mass and structural size) increased through the breeding season. Increase in nestling size was not due to differences among parents or their territories. Nestling mass (Figure 2-4), and tarsus and wing lengths all increased significantly through the breeding season (Table 2-II), but none of the three measures of nestling size varied significantly between broods after the effect of date was corrected for. These results indicate that intra-seasonal increase in nestling size was not due to variation among adults or their territories (Figure 2-lb). In fact, as noted at the end of the Methods, there was no evidence that size of nestlings from second broods was related to size of young from the first brood of the same pair (sizes of young from first and second broods were statistically independent). Nestling condition did not vary significantly with date, but was higher for offspring from second broods (Table 2-II). Logistic regression showed no significant association between survival of offspring from banding to independence, and either date or brood number (Table 2-II). Thus, survival over the entire period that the young were in the care of their parents was not affected by the date of birth of the young, although nestling size (at a given age) 29 Table 2-IV. Frequency of second nests failing to hatch all eggs, as a function of whether young were fledged from the first nest of the year. There was no significant difference in hatching success of second nests with success of first nests (G=0.15, P>0.50). First Nest all eggs hatch Second Nest no fledglings produced fledgling produced 28 68 not all eggs hatch 13 39 30 Figure 2-4. Seasonal variation in offspring mass for 1983 (year with the largest data set); results for other years are similar. Data points are mean values for each nest; circles are for first broods, and open diamonds for second broods. Nestlings were 6 days of age when measured. Line drawn by linear regression. 32 increased through the breeding season. As there are no data on offspring condition at the time of independence, I could not test whether the quality of independent young was affected by date or brood number. Recruitment of independent young The probability of recruitment of independent offspring decreased significantly through the breeding season, but brood number did not influence recruitment (Table 2-II). Figure 2-5 presents the predicted probability of survival as a function of date. These results support the hypothesis that the seasonal decline in recruitment was some function of time that affects all juvenile birds, irrespective of the territories that they came from, or the ability of their parents to raise young (Figure 2-lb). The logistic regression predicts that roughly 50% of independent young from the first nests in the breeding season survived to enter the breeding population, but that recruitment from a year's last nests is only about 20% (Figure 2-5). One fact not revealed by the logistic regression is that the decline in survival of independent young with later hatching is much more consistent among years for second than first broods (Table 2-1), although the patterns presented in Table 2-II and Figure 2-5 present the best statistical fit to the data. DISCUSSION The later a nest was initiated, the fewer independent offspring survived to enter the breeding population in this population of Song Sparrows. This was due to clutch size, hatching success, and survival of independent offspring all decreasing through the breeding season. However, only the latter two effects truly occurred at the pop-ulation level. The decline in clutch size was due to variation among nesting birds, with a major source of variation being female age. Yearling females nested later and laid smaller clutches, as has been noted for other species (e.g. Klomp 1070, Perrins and McCleery 1985, Askenmo and Unger 1986, Murphy 1986). After female age was 33 Figure 2-5. Expected probability of recruitment of independent young as a func-tion of date of initiation of laying. The line was predicted by logistic regres-sion. The data points and curve shown here are for survival of young from the 1978 breeding season. Results for other years are similar. Points at the top of the graph represent surviving offspring, and lower points are offspring that failed to recruit. Circles are for offspring of first broods and diamonds for second brood juveniles. 34 35 accounted for, there was still evidence of a seasonal decline in clutch size. Murphy (1986) also found that female age was not the only factor that produced intra-seasonal decline in clutch size, in a study of Eastern Kingbirds. Thus, some difference among females, in addition to age, may affect clutch size. A difference in food supply among territories is one possibility; birds with a poor supply of food on their territories may lay smaller clutches, and begin nesting later in the breeding season. Arcese and Smith (1988) showed that clutch size of Song Sparrows on Mandarte Island was increased significantly and nesting advanced by supplemental feeding. Whether seasonal varia-tion in clutch size of other species could be the result of differences in the supply of food among territories is unclear; most food supplementation experiments (see Arcese and Smith 1988 for a review) have found that clutch size is not significantly affected by supplemental feeding. When variation was examined at the level of the population, clutch size decreased through the breeding season; however individual females' clutches tended (almost sta-tistically significant) to increase in size from first to second brood. The same result was found by Askenmo and Unger (1986) and less direct evidence of the same phe-nomenon has been found by others (see Perrins 1970). This tendency to increase clutch size occurred despite survival of young from second broods being lower than survival of young from first broods: both probability of hatching and survival of young from independence to breeding were lower for second broods (Table 2-II). If producing a relatively large clutch was very costly to the female, then the lower success of second broods could result in selection to decrease the sizes of second clutches. However, second clutches were not smaller, but larger than first clutches. Food supply may increase through the breeding season, so even though each egg from a second clutch is worth less to the parent (there is a smaller chance of producing a surviving offspring) the energetic stress of producing offspring from second broods is also lower. This ar-gument implies that the expected, evolved decrease in clutch size has been swamped 36 by environmental variation that favors an increased clutch size for second broods. Hatching success was not affected by date, but was significantly lower for second broods than for first broods (Table 2-II, Figure 2-3). Raising one brood did not lower hatching success of the subsequent brood, when hatching success following successful and unsuccessful previous nests are compared (Table 2-IV). However, it can still be argued that even producing a first clutch of eggs, whether these eggs hatch or not, is sufficient stress to lower the hatching success of subsequent nests. Although I have no data that bear on this question, Nol and Smith (1987) showed that birds raising 2 broods of young had significantly fewer independent offspring from third clutches in a season, relative to those birds that had previously produced one brood and a failed nest. Thus, there is some evidence that present nesting can influence the success of future nests of Song Sparrows. Survival of young sparrows did not vary with date or brood number throughout the entire period that the young remained in the care of their parents (Table 2-II). However, two lines of evidence indicate that conditions for raising young did not remain constant through the breeding season. First, nestling size at a given age was larger for birds born later in the breeding season (Table 2-II, Figure 2-4), suggesting that conditions improved through the season, although not enough to compensate for the "factor causing seasonal decline in survival of offspring. Second, survival of young was measured on a per nest basis, and as a result the number of offspring raised in a nest was not considered; the number of young raised did vary through the season as a result of seasonal variation in clutch size. The lack of seasonal variation in survival of offspring under parental care suggests that once clutch size has been determined, under normal circumstances there is no further adjustment in numbers of offspring raised. It is clear that the decline in recruitment of independent young was not due to variation in quality of parents or their territories, as parents were not able to produce 37 young with high (or low) survival in both first and second broods (Table 2-II). However, parental quality may play a role in declining survival of offspring in other species (e.g. Snow Geese, Cooke et al. 1984) or populations. Increase in nestling size through the breeding season suggests that, if anything, conditions for feeding young improved later in the breeding season. Clutch size of individual females increased slightly and almost significantly from first to second brood, which also suggests that more food was available later in the breeding season. Lack of seasonal decline in survival of offspring under the care of their parents further indicates that environmental conditions at the time the young were raised should not have affected their subsequent survival. Social interactions seem to be important determinants of fledgling survival in some species. Interactions with older juveniles appeared to decrease survival of Great Tits from second broods in the Netherlands (Tinbergen et al. 1985). Nilsson and Smith (1988) found that timing was important for Marsh Tits, because late hatching offspring were less likely to establish as dominant members of a winter flock. In Marsh Tits, establishment in winter flocks occurs only days after young leave their parents. However, the period immediately after independence may not be as critical for Song Sparrows, as Arcese and Smith (1985) found that retaining newly independent young in captivity for over 1 month (thus preventing them from becoming established in an area before late hatching young) did not affect their social dominance (ability to compete at food sources). The only factor correlated with date (and hence probability of survival) for Song Sparrows was social dominance (Arcese and Smith 1985). It remains unclear whether the higher survival of earlier young was actually caused by their higher dominance status. Potentially, both dominance and foraging efficiency increase over time so that early born young survive better because they are better at finding food in the face of a declining supply of food after the end of the breeding season. A final point to note is the presence of significant inter-annual variation in al-38 most all measures of nest success used in this chapter (Table 2-II). There are two consequences of the variation occurring between years. First, this study shows that birds that start nesting earlier have higher reproductive success than birds that begin nesting later in the year; this is a relative difference. Absolute measures of repro-ductive success show that reproductive success was determined more by what year(s) a bird lived than by when it began nesting within a year. Inter-annual variation is largely associated with population density (Arcese and Smith 1988, Hochachka et al. 1989, Arcese et al. in prep.). The second result of inter-annual variation is that the patterns of seasonal variation in reproductive success are not identical over all years. The patterns of seasonal variation presented in this chapter are those that generally occur in this population. However, some years exhibited opposite trends (e.g. increas-ing survival of independent young through the year; Table 2-1). Van Noordwijk et al. (1981) found that the pattern of seasonal variation in reproductive success of Great Tits varied so much that there could not be consistent selection for either early or late laying. For Song Sparrows, although the patterns of seasonal decline of clutch size and recruitment of offspring are typical of patterns found when all years are considered (Table 2-1), qualitative variation in seasonal patterns of reproductive success should not be ignored as such variation can affect evolution of laying date, and maintenance of genetic variance for these traits. In conclusion, the seasonal decline in clutch size observed in this population of Song Sparrows was attributable solely to variation among birds or their territories. As laying date did not contribute to variation in clutch size, the model of Price et al. (1988) explains how genetic variance in laying date can be maintained in the face of selection for earlier, larger clutches. However, declining survival of independent offspring was not due to variation among birds or territories, and thus the factor determining survival of offspring did not also determine laying date. Thus, there should be a genotypic response to selection for earlier laying. As a result, one would expect low 39 heritability of laying date in this population of Song Sparrows. Published estimates (van Noordwijk et al. 1981, Findlay and Cooke 1982, O'Donald 1983) have shown laying date to be moderately to highly heritable in several populations of birds (h2 from 0.44 to 0.85), although no significant heritability of laying date was detected for two of the four populations of Great Tits studied by van Noordwijk et al. (1981). The population of Song Sparrows on Mandarte Island did not show detectable heritability of laying date (h2 =-0.012, n=71, P=0.96; mother - mid-daughter regression), even though laying date is a repeatable trait at some ages (Table 5-1). The results presented in this chapter show that seasonal variation in reproductive success can be a complex phenomenon, resulting from both differences among birds or their territories, and seasonal variation in the environment that affects all members of a population. The generality of these finding is currently unclear, and careful study of other multiple-brooded species is required. 40 C H A P T E R 3 D E T E R M I N A N T S A N D C O N S E Q U E N C E S OF N E S T L I N G C O N D I T I O N Summary - I examined the relationship between nutritional condition and early survival of offspring in a population of Song Sparrows (Melospiza melodia). As nestling condition increased, probability of survival of nestlings to inde-pendence also increased in 7 of 8 years. Rate of increase in survival was a concave-down function of nestling condition in 6 of 8 years. Average nestling condition declined with increasing brood size, and tended to be higher for nestlings born later in a year. Individual parents did not consistently pro-duce nestlings either above or below average condition in consecutive nests in a year. Of those offspring that survived to independence, males had been in better condition as nestlings than had females. 41 I N T R O D U C T I O N Nutritional condition of nestlings can influence their probability of survival (e.g. Garnett 1981, Coulson and Porter 1985, Smith et al. 1989). Insofar as adult birds can affect nutritional condition of their offspring, adults can affect their reproductive success. If increased parental effort produces higher nutritional condition of nestlings, it clearly benefits parents to expend considerable effort in raising their offspring. How-ever, there will be a point beyond which increased parental effort no longer pays off (e.g. Smith and Fretwell 1974). Increased parental effort per offspring becomes un-productive either if reproductive success could be increased by raising a larger number of offspring but giving each of them less care, or if lifetime fecundity of parents is de-creased by increased effort expended on the current brood. Understanding the causes and consequences of variation in nutritional condition of nestlings is of interest because this variation affects reproductive success and potentially reproductive strategies of birds. Although many studies have shown that nestlings in higher nutritional condition have higher probability of survival than those in poorer condition (e.g. Garnett 1981, Coulson and Porter 1985, Smith et al. 1989), these demonstrations generally show only that nestlings that died were in poorer condition, on average. Quantifying the relationship between nestling condition and survival is important. Raising more than one nestling is advantageous only if the condition — survival curve is concave down: i.e. increasing nestling condition by a given amount results in ever smaller increases in survival (Smith and Fretwell 1974). Not only the qualitative form, but also quanti-tative changes in the relationship between years are important, because quantitative differences in the condition — survival curve between years will result in different opti-mal levels of parental care in different years. Although theory suggests the form of the relationship between condition and survival, there are very few empirical estimates of the relationship. Nur (1984b) quantified the relationship between nestling condition 42 and survival in two years in a population of Blue Tits. He notes that nestling condition and survival were only correlated in 1 year, but with only 2 years of data he could not assess how variable the relationship was. The function of high nutritional condition may influence parental decisions on pro-visioning nestlings, and may also produce annual variation in the relationship between nestling condition and survival. The role of nutritional reserves may be as a hedge against short-term food shortage (e.g. Garnett 1981); alternatively, nestling condition merely indicates the overall competence of parents or quality of their territories. If nutrient reserves can tide nestlings over unpredictable periods of food shortage (e.g. Lack 1956, O'Connor 1977a), then the optimal level of parental care depends on the frequency and duration of such periods of food shortage. A higher probability or longer period of food shortage will favor parents that produce nestlings with more reserves. Annual variation in severity of food shortage would produce a strong relationship be-tween condition and survival in some years, but not in others. Alternatively, variation in nestling condition may indicate a chronic shortage of food, the result of variation in overall parental competence and territory quality. If this is the case, survival should increase with higher nestling condition in all years. One way to distinguish between these alternatives is to examine how nestlings respond to the presence of abundant food. Nestlings may either put on a reserve of nutrients (increase in mass) or may grow structurally larger (increase mass and skeletal size) when food is abundant. In the latter case, selection is presumed to favor growth rather than insurance against potential future starvation (O'Connor 1977b). Adults may differ in their ability to raise offspring (e.g. Coulson and Porter 1985, Amundsen and Stokland 1990), and this can result in different optimal brood sizes for different adults. One indication that birds adjust their broods to their provisioning abilities is that nestling condition is invariant with brood size (Nur 1986). Birds able to raise many offspring have many nestlings in good nutritional condition, while 43 less competent parents have fewer nestlings but the nestlings will still be in good condition. Differences in ability to rear offspring may be the result of differences either between adults themselves, or (for territorial species) between the territories they occupy (Hogstedt 1980). If differences in ability to raise offspring are the result of differences in territory quality, then a bird may not be able to produce nestlings in the same condition in two consecutive nests. This would result if territory quality changes over time. However, the condition of nestlings relative to the condition of offspring from other territories should still remain constant, as long as the quality of all territories covaries. Nutritional condition of nestlings should be correlated between successive nests of the same parents, if some adults or territories are consistently better than others. The effect of nestling condition may vary with sex of the offspring. The larger sex clearly requires more parental care in order to reach maturity in species that are strongly sexually dimorphic, like many American icterids (e.g. Howe 1976, Cronmiller and Thompson 1981). A recent study (Smith et al. 1989) presents data suggesting that sexes differ in their food requirements as nestlings even in Great Tits, a relatively monomorphic species (males about 5% larger than females; e.g. McCleery and Perrins 1989, Table 3.7). Smith et al. (1989) found that male Great Tits that survived to independence had been heavier as nestlings than had females that survived to independence. In this chapter, I examine some potential influences on nestling condition and the trade-off between brood size and nestling condition for a population of Song Sparrows (Melospiza melodia). I first consider the appropriateness of two measures of nestling condition. I then quantify the relationship between nestling condition and nestling survival. Because I have data from eight breeding seasons, I look for inter-annual variation in the condition — survival relationship. I examine survival of offspring from the time they are in the nest until they leave their parents' care. Thus, I examine 44 whether the benefits of good condition are felt early in offspring's lives. I examine whether nestling condition varies with brood size or is repeatable between nests of the same parents; both sets of results bear on whether parents have different abilities or strategies for the care of their offspring. Finally, I look for evidence of sex related differences in the requirements of nestling Song Sparrows. M E T H O D S General Methods Nestlings were aged using either known date of hatching, or empirically deter-mined relationships between age and development. Young were generally banded at 5-7 days of age (in a 10-11 day nestling period). Nestling mass and wing length were measured once for each nestling, at the time of banding. Young become independent of parental care at roughly 30 days of age. Which offspring reached independence was determined by systematic census and mist-netting on the island. Most newly inde-pendent offspring remain on the island for the summer. Birds could not be sexed as nestlings. However, for the subset of nestlings that survived to independence, sex was determined by using either sex specific behaviour (e.g. singing), or by a discriminant function (that correctly "predicted" sex of over 90% of known sex birds) based on juvenile mass, wing length and tarsus length. My data come from the 1982-89 breeding seasons. Ages of nestlings at banding were not available for birds born in 1982. Mass was measured with an electronic balance, to ±0.05 g. Wing length is the measure of a flattened wing from wrist to the end of the wing tip, measured to the nearest 0.5 mm. Any bird known to be alive when 30 days of age or older is considered independent of parental care (hereafter "independent"). Survival was not often ascertained exactly at 30 days of age, but was known always by the end of October, before the onset of winter. Hence, although I do not know whether birds died before or immediately after independence, I do know whether death occurred in the first 1-5 months of a bird's life. "Population density" 45 is taken as the number of breeding females on Mandarte in the year in question; other possible measures of population size are highly correlated with the number of breeding females (Arcese et al. in prep). Statistical Procedures Most statistical analyses were conducted using SYSTAT (Wilkinson 1987). The exceptions were analyses of the probability of survival of nestlings. For these data the logistic regression module of B M D P was used (PLR; Dixon et al. 1983). Func-tions describing the relationship between probability of survival and nestling condition (Figure 3-2) were generated with cubic splines (e.g. Schluter 1988). As with logistic regression, splines are capable of taking binomial (i.e. 0, or 1; die or live) data for individuals, and producing probabilities of survival for a population. Measuring Nutritional Condition of Nestlings Two types of measures are used as indications of nutritional condition of nestling birds. One is simply the offsprings' mass at a given age (e.g. Smith et al. 1989). The other is some ratio of mass to structural size (e.g. wing length, tarsus length). The "ratio" that I used was the residuals from a cubic regression of nestling mass on nestling wing length (Figure 3-1). Data from nestlings of all ages were entered into the regression. The regression describes the expected mass of a nestling, as a function of its structural size. Residuals measure deviations from the expected mass, with positive residuals for nestlings that weigh more than expected, and negative residuals for lighter nestlings. The actual regression equation is mass = 7.122 — (0.228 x wing) + (0.0366 x wing2) — (0.000568 x wing3), where mass is measured in grams, and wing length in mm. The regression fit the data closely (r 2=0.78), and all regression coefficients were highly significant (P<0.001), except for the first order term (P=0.10). No attempt was made to correct for age-specific differences in nestling size, but the average condition index did not vary significantly with nestling age (1-way A N O V A P=1.589, df=6, 1055, P=0.147). In contrast, nestling mass is highly dependent on nestling age, as nestlings 46 grow extremely rapidly (an average of 1.8 g/day) between the ages of 3 and 9 days. Data from all years were entered into the regression, without correcting for possible differences in average condition among years, because I wanted to look for patterns of variation among years. Although the average residuals varied among years (see Results), inspection of the data suggested that these annual differences were the result of differences in the intercept of the relationship between mass and wing length and not a difference in slope. That is, the residuals indicated that if nestling condition was higher in one year, all nestlings of all ages were in better condition in that year. The two potential measures of nestling nutritional condition, mass and residuals from a mass — wing length regression, were correlated. Mass and residuals were sig-nificantly correlated at every age (r>0.54, P<0.001, separate correlations for each age from 4 to 8 days). Given the correlation between the two measures, they should be rel-atively inter-changeable, but the residuals have the advantage of not being dependent on nestling age. Nestling mass and the residuals were similar in their ability to predict nestling survival. For comparative purposes, regressions were performed only on data from nestlings measured at 6 days of age, because mass varies with age. Six day old nestlings were the largest age group in the data set (47.5% of all nestlings of known age). In logistic regressions, nestling mass was a better predictor (%2=8.6, df=l, P=0.003) of offspring survival than the residuals (x 2 =6.13 , df=l, P=0.01), although both mass and residuals were statistically significant predictors and neither provided informa-tion lacking from the other. I compared predicted probabilities of survival estimated from the logistic regression that used residuals as the predictive variable, with the probabilities from the regression using nestling mass as predictor. Separate logistic regressions were made of nestling survival against nestling mass, and against residuals from the mass — wing length regression; year was entered into the regressions. I then calculated the expected probability of survival for every nestling from the equation 47 Figure 3-1. Relationship between nestling mass and wing length. Data from all years and nestling ages are combined in the regression. Residuals from this regression (vertical distances from regression line to data point) are used as an index of nestling condition. 49 provided by each logistic regression. The two indices of nutritional condition pro-duced highly correlated predictions of offspring survival (7^=0.64). However, the slope of the regression through the two sets of predicted survivals differed significantly from one; the two indices produced quantitatively different predictions. Assuming nestling mass to be a more accurate estimator of nestling survival, I compared the predicted survival rates of nestlings from the two regressions. The logistic regression based on the residuals tended to under-estimate survival of birds in best nutritional condition (by about 5%), and over-estimate survival of birds in poorest condition (by roughly 7%). The probabilities of nestling survival predicted by nestling mass and residuals are linearly related, so variation in residuals provides a qualitatively accurate description of variation in nestling condition. Although my analyses suggest that nestling mass is a slightly better predictor of offspring survival, I used the residuals from a mass — wing length regression as my index of nestling condition because the residuals do not depend on nestling age. This allowed larger sample sizes in statistical analyses, thus increasing the power of statistical tests. The one instance in which mass and residuals provide contrary evi-dence of nestling condition will be discussed below. For the remainder of the thesis all references to "nestling condition" and "condition index" will refer to the residuals from the cubic regression of mass on wing length. R E S U L T S Nestling Condition and Survival Nestlings in better condition were more likely to survive, as seen in simple com-parisons of nutritional condition of nestlings that survived and died. Nestlings that survived to independence had had a higher condition index than nestlings that dis-appeared (Table 3-1; P=0.003), and there was also a significant difference in mean condition index between years (P=0.009) in a two-way A N O V A . The survival x year interaction was not statistically significant. Survival of offspring from independence 50 to entering the breeding population was not related to nestling condition. Year and date of birth were significant predictors of probability of survival to recruitment (both P<0.01, see Chapter 2), but the effect of nestling condition did not approach signifi-cance (P=0.43), in a logistic regression of survival from independence to recruiting. Within each year, the probability of nestling survival to independence generally increased with increasing nestling condition; the shapes of the curves generally fol-lowed the expected pattern. Figure 3-2 shows the pattern of variation in survival with nestling condition. The lines were produced by cubic splines (see Schluter 1988). In 7 of 8 years there was an overall increase in survival with nestling condition; the only exception (1984) showed a very small decline in survival with increasing condition. In 6 of 8 years the curves were concave down (strongly so in 3 of 8 years): increasing nestling condition in these years produced a continuously smaller increase in nestling survival. Thus, increasing the effort expended in provisioning nestlings yields dimin-ishing returns for adult sparrows, beyond a certain point. The positive effect of nestling condition on survival was not confounded by the size of broods that offspring came from. The average nutritional condition of nestlings was lower for larger broods (an average decline in index of 0.199 per added nestling P=0.015, df=3, 482, slope from an analysis of covariance of nest mean condition index against year, with brood size as covariate). However, only nestling condition, and not brood size, was a significant predictor of offspring survival (step-wise logistic re-gression). A n added effect of brood size was highly improbable (P>>0.99). Thus, it appears that brood size affects nestling condition, which in turn affects survival of nestlings. Young born later in a year tended to be in better condition than early born young, but the pattern was statistically significant in only 1 of 8 years (Table 3-II). The predicted condition index increased by up to 0.013/day (over a 119 day breeding season), resulting in a mean change in condition index of up to 1.5 (or change in 51 Table 3-1. Condition of nestlings and survival to independence. Condition was measured as the residuals from a cubic regression of nestling mass on wing length. A value of zero is the average expected mass for a given wing length. Data are presented as x ± s.d. (n). Surviving birds had had significantly higher condition indices as nestlings than birds not surviving past indepen-dence. Year Survived Died 1982 0.225 ± 1.212 (116) -0.445 ± 1.604 (10) 1983 0.469 ± 1.562 (149) -0.115 ± 1.967 (80) 1984 0.114 ± 1.304 (126) 0.138 ± 1.221 (73) 1985 0.007 ± 1.012 (120) -0.368 ± 1.508 (58) 1986 -0.067 ± 1.180 (74) -0.418 ± 1.105 (68) 1987 -0.072 ± 1.375 (73) -0.556 ± 1.402 (76) 1988 0.052 ± 1.417 (102) -0.156 ± 1.371 (44) 1989 -0.225 ± 1.464 (14) -0.310 ± 0.807 (10) 52 Figure 3-2. Nestling condition and survival. Condition was measured as the residuals from a cubic regression of mass on wing length. Survival data were binomial (0 if a nestling died before independence, 1 if a nestling survived to independence). Curves were fitted by cubic splines; splines were forced to be smooth lines by setting a relatively large smoothing parameter (In(lambda) = 0). Diamonds on each line denote the mean condition index for each year. 85 54 survival of ~ 10%) over a season. However, the seasonal trend explained only a small fraction of the observed variation in nestling condition (all annual r 2<0.042). The tendency for increasing condition of later born nestlings did not result in a detectably higher rate of nestling survival through a breeding season. In Chapter 2, I showed that survival of nestlings from banding to independence does not vary with date of birth in this population. Because survival to independence does not vary with date, the correlation between nestling condition and probability of survival (Figure 3-2) can not be a spurious correlation resulting from the seasonal increase in nestling condition. The average condition of nestlings varied significantly among years, and tended to correlate positively with the average probability of survival of nestlings of a given condition. Although variation in many reproductive parameters is related to variation in population density of Song Sparrows (e.g. Arcese and Smith 1988, Hochachka et al. 1989), the average condition of nestlings was not related to population density (r=0.025, rx=8, P=0.95). However, annual variation in nestling condition may be driven by some factor that also influences the probability of nestling survival. For each year, I took the expected survival of nestlings with a condition index of 0 (the overall average) as my indicator of survival potential of nestlings in a given year. Probabilities were taken from the splines shown in Figure 3-2. Years with highest survival on average also produced nestlings with the highest nutritional condition (r=0.66, n=8, P=0.07). More years of data are needed to confirm these patterns. Parental Ability and Nestling Condition In this section, I look for evidence that parents control the condition of their nestlings. I show that nestlings in the same brood were in similar nutritional con-dition and had a similar probability of survival. Territories with supplemental food produced heavier nestlings, although the condition index was not influenced by ad-ditional food. Neither parents nor territories without supplemental food consistently produced nestlings of a given mass. 55 Table 3-II. Relationship between nestling condition and date of hatch. Regression slope is the change in condition index per day, and "range of dates" is the total range of dates that birds hatched in a given breeding season. range Year slope P n r2 of dati 1982 <0.001 0.989 126 <0.001 68 1983 0.013 0.002 228 0.042 119 1984 0.004 0.250 199 0.007 92 1985 0.007 0.060 179 0.020 86 1986 0.009 0.102 142 0.019 67 1987 0.005 0.396 149 0.005 84 1988 -0.002 0.712 145 0.001 91 1989 -0.006 0.528 24 0.018 80 56 The probability of one offspring surviving from banding to independence was re-lated to the probabilities of its brood-mates' survival; either all or none of the young from a nest tended to survive. K the survival of brood-mates was statistically indepen-dent, then the frequencies of 0, 1, or more surviving offspring would be predicted by a binomial distribution (e.g. Sokal and Rohlf 1981, pp. 70-72). However, more cases where all or none of the offspring in a nest survived were found than were expected by chance (Table 3-III). This was true for broods of 3 or 4 young, but not broods of 2. These results were probably not due to nest predation, which appears to cause an all-or-none loss of brood-mates (personal observations), and the analyses in Table 3-III were conducted on only those nests from which at least 1 offspring fledged. Thus, nests experiencing predation were excluded from the analyses. If similar survival of nest-mates was due in part to similar condition, then the average condition of offspring should differ among nests. Average condition of offspring differed significantly among nests. Separate one-way A N O V A s were conducted for data from each year, and all nests with more than 1 nestling were included in the analyses. Average nestling condition differed highly significantly between nests in 7 years (all P<0.001), with the difference approaching significance in the eighth year (1989, P=0.058). Between 34% and 61% of variance (see Sokal and Rohlf 1981, p. 216) in nestling condition (depending on year) was due to differences among nests. Although condition of nestlings varied among nests, nestlings varied substantially in condition within nests too. Nevertheless, the signifi-cant variation among nests is a necessary pre-condition for some parents or territories to consistently produce offspring of better condition than other parents. One factor potentially affecting nestling condition is availability of food on terri-tories. The effect of food availability on nestling condition was tested experimentally by providing supplemental food in two breeding seasons (1985 and 1988; methods for 1985 are outlined in Arcese and Smith (1988); similar methods were followed in 1988). 57 Table 3-III. Frequency of survival of nestlings to independence. Nests with in-termediate numbers of nestlings surviving were less frequently observed than was expected by chance. Expected frequencies were calculated by binomial expansion. Brood size = 2 Observed Expected Number Independent 0 1 2 11 53 27 15.5 44.1 31.5 G=3.62, P=0.16 Brood size = 3 Observed Expected Number Independent 0 1 2 3 59 129 120 65 45.5 138.7 141.0 47.8 C=13.21, P « 0 . 0 0 1 Brood size Observed Expected Number Independent 0 1 2 3 4 27 99 86 89 44 18.7 80.3 129.1 92.2 24.7 G=35.94, P « 0 . 0 0 1 58 The average condition of food-supplemented and control nestlings was compared; ter-ritories adjacent to the supplemented territories were excluded from the analyses, as some of the adults from neighbouring territories were known to take food from the feeders. Nestling condition was not affected by supplemental feeding (P=1.61, df=l, 99, P=0.21), nor did condition vary between the two years in the experiment (P=0.66); the year x feeding interaction was not significant (P=0.26) in the two-way A N O V A . However, the effect of supplemental feeding was not statistically significant because of the nature of the index of nestling condition, and not because of a lack of effect of the additional food. Nestling mass and structural size were significantly increased by feeding (Arcese and Smith 1988, Smith and Arcese 1988). The condition index, being a ratio, would not be affected if both mass and wing length increased by roughly the same proportion with the provision of supplemental food. Thus, I conclude that parents with super-abundant food produced bigger nestlings, but these nestlings, at a given age, are not heavier for a given structural size. If food supply differs consis-tently among territories, then some parents could consistently produce faster growing offspring than other parents. Repeatability of nestling condition between successive nests of the same parents could be confounded by differences in average nestling condition and mass with brood size. As brood size increased, the average condition index for nestlings declined by 0.199, and the average mass dropped by 0.28g for each additional nestling present. Both effects were statistically significant (P<0.02). Parents did not consistently produce offspring of a given quality; this conclusion is the same whether based on repeatability of condition index or nestling mass. I looked for repeatability by correlating the average condition index (and mass) of first and second broods produced by the same parents (territory) in a year. Because nestling mass varies with age, I limited my analysis of repeatability of mass to only those cases where nestlings of first and second broods were 6 days old when measured; 59 sample size for the analysis using mass was substantially lower than for that using the condition index. For both condition index and mass, I corrected for variation with brood size. This was done by adding 0.199 x (brood size — 1) to each condition index, and adding 0.28<7 x (brood size — 1) to each nestling mass. The result was to correct all measures up to the values expected if all nestlings were in broods of one. For neither condition index (r=0.07, n=152, P=0.38) nor nestling mass (r=0.04, n=55, P=0.77) was there any evidence that some parents consistently produced better offspring than other parents. Nestling Sex, Condition and Survival A factor that might interact with nestling condition to influence survival is a nestling's sex. Unfortunately, birds were not sexed as nestlings, but only after reach-ing independence. Thus, sex ratio in the nest, and the sex of offspring that died before independence are unknown. For those nestlings whose sexes were subsequently deter-mined, a two-way A N O V A comparing the nestling condition index by sex and year revealed that surviving males had been in significantly better condition as nestlings than had surviving females (Table 3-IV; P=50.129, df=l, 550, P<0.001). The differ-ences were strong and relatively constant across years (P=0.79 for sex x year interac-tion). This result has two potential interpretations. First, sexual dimorphism in size or growth during the nestling period may have produced the difference in average con-dition. Second, male nestlings may have to be in better condition in order to survive than female nestlings. I cannot distinguish between these two possibilities. D I S C U S S I O N I conclude, like several other studies (e.g. Garnett 1981, Coulson and Porter 1985, Smith et al. 1989), that nestling condition is an important predictor of survival. This result is not universal, however. Other studies of passerines (e.g. Ross and McLean 1981, Wolf et al. 1988, Sullivan 1989) found no correlation between nestling mass and subsequent survival. Sullivan (1989) notes that there is no evidence of food limitation 60 Table 3-IV. Condition of surviving male and female nestlings. Condition was measured as the residuals from a cubic regression of nestling mass on wing length. A value of zero is the average expected mass for a given wing length. Data are presented as x ± s.d. (n). Surviving males had been in significantly higher condition as nestlings than had surviving females. Year Males Female 1982 0.503 ± 1.101 (43) -0.008 ± 1.273 (47) 1983 0.851 ± 1.609 (59) 0.207 ± 1.689 (51) 1984 0.557 ± 1.138 (49) -0.206 ± 1.168 (51) 1985 0.372 ± 0.980 (31) -0.036 ± 1.006 (61) 1986 0.450 ± 1.052 (36) -0.624 ± 1.069 (29) 1987 0.363 ± 1.251 (18) -0.200 ± 1.482 (17) 1988 0.298 ± 1.216 (44) -0.244 ± 1.559 (28) 61 during the breeding season for her study population, whereas the Song Sparrows on Mandarte are clearly food-limited in at least some years (Arcese and Smith 1988). I also know that nestling Song Sparrows in good condition survived better only during the period up to, or shortly after, independence, but survival from independence to recruitment into the breeding population did not vary with nestling condition. Sullivan (1989) showed more directly that one of the peak periods of juvenile mortality in Yellow-eyed Juncos is immediately after young leave their parents' care; death was the result of starvation which resulted from inefficient foraging. This may also hold for Song Sparrows. A store of food at independence may provide the buffer needed by young of some species to survive while learning how to forage, although this is apparently not so in Yellow-eyed Juncos (Sullivan 1989). The advantage of higher nutritional condition, and presumably better parental care, is at best short-lived once offspring leave their parents. Nestling Song Sparrows did not respond to supplemental feeding by increasing their mass:size ratio. Nestlings with additional food (i.e. higher levels of parental care) grew structurally larger, and did not put on reserves of nutrients. This conclusion is based on two observations. First, the condition index, a measure of mass relative to structural size, was not affected by supplemental feeding. Second, nestling mass and structural size were increased by feeding (Arcese and Smith 1988, Smith and Arcese 1988). Nestling Song Sparrows grew faster but did not have a higher mass to size ratio, when supplemental food was provided. These results suggest that putting on a nutrient store is less important than rapid structural growth (O'Connor 1977a) for nestling Song Sparrows. That nestling survival increased with increasing condition in 7 of 8 years also suggests that periods of food shortage, if present, are not rare. Instead, high nestling condition indicates a high overall level of parental care. Faster growth may result in offspring fledging at an earlier age; the less time offspring spend in the nest, the greater may be their survival (Sullivan 1989). Unnecessary fat stores can be 62 counter-productive because energy is diverted away from growth. Such costs include not only the energy stored, but also the metabolic energy needed to create a store of lipids; deposition of lipids is roughly twice as expensive energetically as production of the same mass of protein (i.e. muscle growth; Whitlow 1986, p. 261). Annual Variation in Nestling Survival It is improbable that an adult will have the same optimal level of parental care in different years, because of annual variation in the relationship between condition and survival (Figure 3-2). As expected (Smith and Fretwell 1974), the curves were most often concave down, although the effect was often subtle. Although the condition — survival curve was qualitatively similar among years, the exact curve varied from year to year. Thus, the optimal level of reproductive effort probably varies between years. Nur (1984b) suggested that nestling condition influenced reproductive success of Blue Tits in only some years. In my study, offspring in better nutritional condition had higher survival to independence in 7 of 8 years (Figure 3-2) over a wide range of conditions, including variation in population densities from 72 breeding females (in 1985) to 4 breeding females (in 1989). Thus, adult Song Sparrows generally benefit from producing offspring in good nutritional condition, although the exact benefit of producing nestlings in a high nutritional condition appears to vary among years (Figure 3-2). Adjustment of Reproductive Rate Analysis of the trade-off between offspring number and offspring condition (sur-vival) suggests that the optimal brood size should not vary among years, and should remain higher than the observed average brood size. If maximizing the number of off-spring surviving to independence were the only consideration, then optimal brood size could be calculated from: the relationship between brood size and nestling condition, and the relationship between nestling condition and nestling survival. Both of these relationships can be derived for the Song Sparrows. Nestling survival in each year was 63 predicted from the relationships shown in Figure 3-2. Regressions of nestling condition on brood size were calculated for each year, in order to predict nutritional condition of nestlings at brood sizes from 1 to 4. Evidence, presented below, suggests that the relationship between brood size and nestling condition is not an artifact of differences in the quality of parents or their territories. Calculating per offspring survival rates from the above two relationships, I found that the probability of survival of individual offspring varied less than 6% between brood sizes of 1 and 4. The optimal brood size was predicted to be 4 young in every year. For birds that laid clutches of 4 eggs, mean size of broods varied from 2.4 to 3.2 nestlings, with statistically significantly variation among years (1-way A N O V A F=2.142, #=13,348, P=0.01). The lack of concordance between observed and predicted average brood size indicates that nestling condition (through its relationship with survival of offspring) is not the only influence on the evolution of brood size, and/or that Song Sparrows cannot regulate the size of broods that they raise. These conclusions run counter to results from other studies. Pettifor et al. (1988) argue that Great Tits do not lay larger than observed clutches because of the decline in per offspring survival with increased brood size resulted in fewer breeding offspring when extra young were added to nests. Even adding one additional nestling to a brood of five to thirteen would result in fewer offspring surviving. Gustafsson and Sutherland (1988) also suggest that Collared Flycatchers do not raise more than the observed number of nestlings partially because the declining per-offspring survival rate with increased brood size outweighs the numerical advantage of raising extra nestlings. Again, the evidence came from a study in which brood sizes were manipulated. Thus, there are some species in which the decreasing rate of offspring survival with increased brood size is sufficient to regulate brood sizes at or near their observed values. Data from the Song Sparrows suggest that there are other species for which the trade-off between brood size and nestling condition is not the major control of the evolution of 64 brood sizes. Song Sparrows do not vary their brood sizes to produce offspring of a set nutri-tional condition. Several workers have suggested that because parents vary in ability, the sizes of their broods should also vary as an adaptive response (e.g. Coulson and Porter 1985, Pettifor et al. 1988). Thus, brood size and nestling condition should not be negatively correlated (Nur 1986). Some studies have shown that nestling mass is not related to natural brood size (e.g. Wolf et al. 1988, Sullivan 1989). However, larger broods contained nestlings in poorer condition in this population of Song Sparrows. This result is typical of studies in which birds could not control the sizes of the broods that they raised. Nestling mass significantly decreased with increasing brood size in 22 of 34 studies in which brood size was manipulated (summarized in Djikstra et al. 1990). My results suggest that Song Sparrows raising larger broods are not inherently able to raise more offspring. Hence, variation in brood size of Song Sparrows is not under the control of parents, but is imposed by external forces. One of the simplest controls of brood size, variation in clutch size, is very limited in Song Sparrows. Forty-three percent of all clutches are of 3 eggs, and 38% of clutches have 4 eggs (unpublished data). Much of this variation is accounted for by inter-annual changes in mean clutch size with population density (Arcese and Smith 1988, Arcese et al., in prep.) Perhaps, species that lay clutches of more varied size than Song Sparrows are better able to optimize their clutch and brood sizes. The lack of repeatability of nestling condition between broods in the same year on the same territory also suggests that Song Sparrows cannot control brood size to produce nestlings in optimal condition. Some adult sparrows consistently produce larger clutches than others (Chapter 2), so some parents or territories are consistently better than others. However, birds did not consistently produce nestlings in better or worse condition than other parents. The lack of significant repeatability of nestling condition also suggests that production of nestlings of a given nutritional condition is 65 not a genetically fixed trait in adults. These results were not an artifact of variation in brood size. Hence, brood size and parental effort did not vary with the result of producing nestlings in good nutritional condition. Instead, nestling condition varied due to changing brood size and (presumably) levels of parental care. Nestlings' Sex and Survival My data also suggest that male nestling Song Sparrows may require more parental care, or may follow a different pattern of growth than female nestlings either due to inherent differences in growth rate or differences in the amount of food they are given (Table 3-IV). This is in spite of adult Song Sparrows being relatively monomorphic: for measured traits, males average only 1.7% (bill length) to 10.1% (mass) larger than females. Both this study and previous work by Smith et al. (1989) on Great Tits note that males that survived to independence had been in better condition as nestlings than had surviving females. One interpretation is that male nestlings in poor condition are more likely to die than female nestlings. The other explanation is that male nestlings have a different growth strategy than females; male nestlings would have a higher mass at any wing length than females, if this were true. Stamps (1990) notes reasons for sex differences in growth strategies in sexually dimorphic species. The possibility of such differences in an apparently monomorphic species is intriguing. Conclusions This study demonstrates that nestling condition and nestling survival are posi-tively correlated in Song Sparrows. That nestling growth and not the condition index was affected by supplemental feeding suggests that higher nestling condition was a consequence of having more competent parents rather than an indicator of reserves used during periods of food shortage. Further research is needed into whether there are sex differences in requirements for growth of nestlings in relatively monomorphic species. Some basic correlate of sex, other than adult size (e.g. Howe 1976), may dictate sex differences in nutritional requirements during growth. 66 Two avenues of future investigation are especially interesting. One is the apparent inability of parents to adjust the sizes of the broods that they raise, which was only indirectly implied in this paper due to the descriptive nature of the data. There is need for more studies, similar to that of Nur (1986), that compare the effects of artificially induced variation in brood size to the effects of natural variation in brood size. The second area deserving study is variation in nestling condition within broods. Smith et al. (1989) noted that increasing brood size also increased the variance in nestling mass, but due to the small brood sizes of Song Sparrows there was no simple way that I could duplicate their analyses. If increased variation in nestling condition in larger broods is a general phenomenon, it may affect the optimal size of broods. Recent studies (e.g. Boyce and Perrins 1987, Yoshimura and Shields 1987) have noted how annual variation in reproductive success can affect optimal brood size. In a similar manner, intra-brood variation in nestling condition could also affect the evolution of brood size. 67 C H A P T E R 4 T H E E N V I R O N M E N T A L C O M P O N E N T OF M O R P H O L O G I C A L V A R I A T I O N Summary - Studies looking for natural selection generally assume that changes in mean phenotype are the result of changes in both genetic and environmental variance. I found that this assumption does not always hold in a population of Song Sparrows (Melospiza melodia). Heavier juveniles consistently survived better than light juveniles, but mass did not have any detectable additive ge-netic variance. The average mass and tarsus length of juveniles decreased but wing length increased, as population density increased. Density-dependent variation in size was solely the result of changes in the environmental com-ponent of size, and did not represent morphological evolution. The effect of environment on size was not a simple phenomenon with a single cause. Environmental variation in some traits (mass, wing length, and some bill measurements) was correlated with birds' nutritional condition as nestlings. 68 INTRODUCTION The environment in which a population lives can alter the distribution of any trait in two ways. First, it may change phenotypic variance in traits without altering genetic variance, a phenomenon known as phenotypic plasticity. For example, rotifer morphology can vary with the presence of chemicals in the water (e.g. Gilbert and Waage 1967); various insects have winged and non-winged forms with the induction of wings being related to population density (e.g. Denno et al. 1980); and the maternal condition of ungulates can affect the size, dominance, and reproductive success of their male offspring (Clutton-Brock et al. 1982). Environmental differences can produce differences in the average size and shape of individuals among populations of birds (e.g. James 1983, James and NeSmith 1988). The second way that environment can affect phenotype is through natural selection (see Endler 1986 for a recent review of demonstrations of natural selection). Many of the demonstrations of natural selection have concerned selection on morphology of birds (e.g. Bumpus 1899, Schluter and Smith 1986a, Gibbs and Grant 1987). In birds, natural selection is generally believed to result in morphological evolu-tion, because morphological traits of birds are heritable (e.g. Dhondt 1982, Alatalo et al. 1984, Gustafsson 1986). Annual variation in morphology in populations (e.g. Dhondt et al. 1979) has been assumed to result from alteration of genetic variance. However, recent work has shown that natural selection sometimes affects only the en-vironmental component of variation in size (van Noordwijk et al. 1988, van Noordwijk 1988, Alatalo et al. 1990). Van Noordwijk et al. (1988) showed that in a Dutch population of Great Tits (Parus major), survival of young shortly after fledging was determined by nestling mass (a heritable trait), with smaller individuals being more likely to die. Directional selection affected only the environmentally induced varia-tion; additive genetic variance was not altered, and natural selection did not produce evolution of the population's morphology. Alatalo et al. (1990) present similar data 69 for Collared Flycatchers. Fledgling flycatchers with short tarsi have lower survival, but this only eliminated flycatchers that had been in poor nutritional condition as nestlings. Given the recent interest in natural selection on avian morphology (e.g. Alatalo and Lundberg 1986, Gibbs and Grant 1987, Grant and Grant 1989), it is impor-tant that we learn more about the causes and consequences of environmentally based variantion in morphology of birds. In particular, we should learn whether avian mor-phology exhibits phenotypic plasticity, and if so, identify the factor(s) leading to this variation. In this paper, I describe the variation in morphology that has occurred over a 14-year period in a wild population of Song Sparrows (Melospiza melodia). We know that the mean size of juveniles varies among years (Smith and Zach 1979), and that natural selection on morphology occurs (Schluter and Smith 1986a) in this pop-ulation of Song Sparrows. Unlike several other studies in which variation in average size was observed (e.g. Bumpus 1899, Price et al. 1984, Grant and Grant 1989), the present paper describes variation that was not caused by sudden, dramatic changes in environmental conditions. This paper has three main goals: (1) to describe patterns in the annual variation of several morphological traits, (2) to determine whether an-nual variation in morphology can be explained solely as the result of variation in the environmental component of variance in morphology, and (3) to determine whether variation in the environmental component of these traits is a single phenomenon or has a different origin depending on the trait in question. M E T H O D S Juveniles were caught in mist nets from July to October of each year, and mea-surements were taken of: mass, wing length, tarsus length, bill length, bill width, and bill depth. A more detailed description of the measurements is presented in Smith and Zach (1979). Measurements were taken when juveniles were over 55 days old, after their size had reached an asymptote (Smith and Zach 1979). As only roughly 3% of 70 all Song Sparrows breeding on Mandarte Island were born off the island, there is a large sample of birds for which we know the sizes of their parents, and their survival and reproductive success. "Population density" is defined as the number of breeding females at the start of the breeding season in a given year. The analyses assume that the behavioural father (bird that fed offspring) was also the genetic father of offspring. Measurements of an individual's size are generally the means of several (generally 2-3) measurements taken over a period of a month or more. Although measurements were taken by several people, 5 people took most of the measurements over the entire 15 year study, one of whom (J .N.M. Smith) was present for the entire study which aided consistency in measurement among people. No attempt was made to correct measurements of mass for variation with time of day (Dhondt and Smith 1980), but most measurements were made on birds captured and measured in the early morning. Because moulting birds are heavier than non-moulting individuals (Dhondt and Smith 1980), masses from moulting birds were excluded. Most measurements came from birds 2 to 7 months old. Measurements of birds greater than 1 year old (e.g. breeding adults not captured as juveniles), were corrected to take into account slight variation with age (Smith et al. 1986). Correction involved modifying the observed measurements by the mean change in size known to occur as birds age. Thus, all references to "adult size" are to size of birds in their first fall. Data on adult mass and wing length are included for 1027 birds born in 13 different years (1975-1979, and 1981-1988). Adult tarsus length, and all three bill measurements of 565 birds born in 10 years (1975-1979, and 1984-1988) are used in this paper. Note that although this population of Song Sparrows has experienced two dramatic reductions in size (Arcese and Smith 1988, Rogers et al., in prep.), this chapter does not consider selection across these periods of population decline (in the winters of 1979-80 and 1988-89). Mass and wing length data for nestlings born 1982-1989 are used in some analyses (see Chapter 3 for details on nestling measures). 71 Sexes of birds were determined in one of two ways. If birds survived long enough to exhibit sex-specific behaviour (e.g. singing), they were classified according to be-havioural criteria. Otherwise, a discriminant function based on mass, wing length, and tarsus length (when available) was used to assign sex. The discriminant function correctly classified slightly more than 90% of individuals of known sex. I examined environmentally-induced variation in size apart from genetic variance, measuring environmental variation as the residuals from a regression of offspring size on mid-parent size. A mid-parent — offspring regression line represents the expected (additive genetic) resemblance between parents and offspring; deviations above or below this regression should largely represent environmental influences on offspring's size. In this regression, data from all years were combined, and data were not corrected for variation in average size among cohorts. A l l years were combined because I was interested in examining environmental effects on variation in average size among years. Calculating residuals separately for each year would mean that the average residual for each year was zero and inter-annual variation could not be detected. Data were not corrected for variation in size among cohorts for the same reason. Such corrections would eh'minate any variation in size among years, and it is this variation that I was looking for. I considered the possibility that four factors biased my estimate of environmental variation; these potential biases are: common environment inflating the heritability, variation in heritability of traits among years, differences in phenotypic variance be-tween sexes, and sexual dimorphism. There is a possibility that any resemblance is due to the common environment shared by parents and offspring, and is not genetically based. The result would be an under-estimation of the degree to which variation in size is environmentally based. Common environment is not a serious problem. In an ex-perimental swapping of offspring between nests on Mandarte, offspring only resembled their genetic parents, not their adoptive (Smith and Dhondt 1980). The second poten-72 tial bias is variation in heritability among years. Different heritabilities mean different regression slopes for each year's data; offspring calculated to be above their expected size when data for all years are combined may be below their expected size if residuals were calculated separately for each year. Calculated heritabilities do vary among years (Table 4-1), although small sample sizes within years make it impossible to determine whether the variation is due to real differences or to sampling error. I know of no way to correct for this bias. Differences in calculated heritabilities were slight for mass, and wing and tarsus lengths. Residuals from mid-parent — offspring regressions calculated separately for each year were highly correlated with the residuals from the regression when all years' data were combined (r>0.94, P<0.001 for each trait). For the beak size measurements, there was a greater discrepancy between the two methods for cal-culating residuals; correlation coefficients ranged from 0.055 to 0.28 (all n>200). It remains unclear whether these discrepancies are due to variation in heritability among years, or to sampling errors in the residuals calculated separately for each year. I have calculated environmental variation for all traits using regressions in which data from all years were combined, in order to present a consistent measurement of environ-mental variation. The third potential bias is caused by unequal phenotypic variances between sexes; although differences were small, there were statistically significantly larger phenotypic variances in males for all traits. The result is an over-estimation of heritability of males' traits, and under-estimation for females (Falconer 1981, p. 153). This makes comparison of the absolute values of environmental variation (residuals from mid-parent — offspring regression) between sexes inappropriate, but should not affect the relative values calculated within each sex. No comparisons of environmental variation between sexes were made, so the sex-related biases do not affect my results. The final possible bias is that female sparrows are slightly but significantly smaller than males. In order to correct for sexual dimorphism, the mean difference (over all years) in size between males and females was added to the measurement of each fe-73 male's size. The degree of sexual dimorphism varied among years, such that residuals from mid-parent — offspring regression will be slightly high in years of low population density and slightly low at high population densities. The effects of this bias will be noted below, where appropriate. The residuals from the mid-parent - offspring regression do not purely measure environmentally induced variance in offspring size for two reasons. First, parental size itself has a component of environmental variance, because there was no correction for differences in size between cohorts. Thus, mid-parent size is not a true measure of the breeding value of offspring. Second, even if parental size was purely determined by genes, residual variance from a mid-parent — offspring regression is composed of environmental variance, non-additive variance, and gene x environment interactions (Falconer 1981, p. 148). Assuming that non-additive effects do not vary substantially among years, so annual differences in the residuals will mainly be the result of changes in environmental variance and gene X environment interactions. For the remainder of this paper, residuals from mid-parent - offspring regressions will be referred to as the "environmental component" of size variation. R E S U L T S Density-Dependent Morphology The average size of offspring varied significantly among years for all traits, but traits did not co-vary. The mean size of offspring varied significantly among years (all traits P<0.001), as well as between sexes (all traits P<0.001), in two-way ANOVAs . Year x sex interactions were only significant for wing length and tarsus length (P<0.03 and P=0.001, respectively). Thus, for all but these two traits, size varied in unison for both sexes. Within sexes, annual variation was not always highly correlated among traits (Table 4-II). These results suggest that annual variation in morphology had more than one underlying cause. Annual variation in mass, wing length, and tarsus length was density-dependent, 74 Table 4-1. Variation in heritability among years. Heritabilities are from mid-parent — mid-offspring regressions. Sample sizes are given in parentheses. Data for mass are not shown because the overall heritability was not statistically significant. Wing Length Tarsus Length Bi l l Length Bi l l Depth Bi l l Width Overall 0.271(346)*** 0.375(267)*** 0.163(185)* 0.369(185)*** 0.364(154)*** 1975 0.106 (20) 0.564 (20)** 0.307 (21) 0.414 (21) 0.428 (21) 1976 0.311 (29) 0.380 (28)* 0.304 (29) 0.211 (29) 0.331 (29) 1977 0.318 (36) 0.502 (33)*** 0.298 (36) 0.467 (36)*** 0.145 (36) 1978 0.265 (34) 0.540 (37)*** 0.342 (38)* 0.423 (38)*** 0.266 (38) 1981 -0.050 (9) — — — — 1982 0.408 (19) — — — — 1983 0.044 (34) -0.046 (17) — — — 1984 0.222 (34) 0.036 (21) — — — 1985 0.342 (34) 0.377 (31)* -0.111 (8) 0.547 (8) 0.901 (4) 1986 0.309 (36) 0.418 (32)* -0.368 (12) 0.517 (13) 0.316 (5) 1987 0.340 (26) 0.173 (16) -0.038 (16) 0.232 (16) 0.915 (5)* 1988 0.270 (32) 0.362 (31)* 0.167 (24) 0.363 (24) 0.724 (16)*** *P<0.05, **P<0.01, ***P<0.001 75 Figure 4-1. Relationship between population density (here shown as the number of breeding females in the population) in the year of birth, and mean size of female offspring produced. Results are similar for male offspring. See Table 4-III for the correlations and sample sizes for both sexes. Lengths are in mm and mass in g; error bars are ±1 S.E. around the mean. Mean Tarsus Length (mm Mean Wing Length (mm) Mean Mass (g) 19.0 19.4 19.8 64.0 65.0 66.0 21.8 22.4 23.0 ro o o CO o 00 o T 1 1 1 1 r T 1 1 1 1 r 1 1 1 1 1 1 1 r i • 1 i • 1 77 Table 4-II. Correlations between annual mean values of traits. Values below the diagonal are for males, and above the diagonal for females. Sample sizes, identical for males and females, are n=13 for correlations among mass, wing length, and tarsus; and n=9 for all other correlations. Mass Wing Length Tarsus Length Bi l l Length Bi l l Width Bi l l Depth Mass — -0.352 0.600a -0.341 0.171 -0.076 Wing Length -0.736* — -0.320 0.667* -0.225 0.197 Tarsus Length 0.584a -0.470 — -0.223 -0.515 0.562 Bil l Length -0.079 0.078 0.350 — -0.113 0.203 Bil l Width -0.055 0.421 -0.192 -0.459 — 0.569 Bil l Depth 0.059 0.176 -0.593a -0.345 0.602 — P<0.10, *P<0.05 78 but not all traits varied in the same direction (Table 4-III, Figure 4-1). As density increased, mean mass and tarsus length decreased while mean wing length increased for both male and female offspring, but only the wing length — density correlation for females was statistically significant. A l l correlations between bill size and popu-lation density were relatively weak. The largest correlation for any bill measurement was roughly half that of the smallest correlation for a non-bill trait. Thus, further consideration of density-dependent variation in morphology is restricted to mass, wing length, and tarsus length. Inter-annual variation in mass, wing length and tarsus length is comparable in magnitude to variation from three other sources. Relative to sexual dimorphism (the average difference in size between males and females, all years combined), the range in average size of males between cohorts was 70%, 57% and 180% as great, for mass, wing and tarsus respectively. The range in average size among cohorts for females was 34%, 51% and 157% (again, for mass, wing, and tarsus) of the average sexual dimorphism. Relative to the overall variation in size within a sex, the range of mean sizes among cohorts is also large. Within each sex and for each trait (mass, wing, tarsus), the range in average size among cohorts was compared to the standard deviation (over all years) of the trait. In every case, except female mass, the range in average size was greater than one standard deviation in magnitude. Wing and tarsus lengths are known to change between the ages of 1 and 2 years for Song Sparrows on Mandarte (Smith et al. 1986). The range of annual variation in wing length was over 130% greater than average change with age; inter-annual variation in tarsus length was over 500% greater than age-related changes. The inter-annual variation in mass, wing length, and tarsus length are not trivial. There are three potential causes of density-dependent variation in morphology of juveniles: (1) environmental influence on offspring size, (2) natural selection on parents' reproductive success as a function of their size, and (3) natural selection on Table 4-III. Correlations between mean offspring size and population density in the year of birth. Sample sizes are given in parentheses. Mass Wing Length Tarsus Length Bi l l Length Bi l l Width Bi l l Depth Size In Fall Male Female 0.71(13)' -0.50 (9) -0.09 (9) 0.12(9) 0.17(9) 0.66(13)** -0.55 (9) 0.02 (9) 0.27(9) -0.02(9) Environmental Component of Size Male Female -0.47(13)a -0.40(13) 0.49(12)° 0.80(12)** -0.29(10) -0.36(11) Size In Spring Male Female 0.06(12) -0.09(12) 0.61(12)* 0.66(12)* 0.05(11) -0.07(11) aP<0.10, *P<0.05, **P<0.01 co 80 juvenile survival as a function of size. The mean size of offspring, irrespective of their genotype, would vary with density in the year of birth if environmental conditions affect size. Natural selection on parental reproductive success could cause variation in mean offspring size, if the direction or magnitude of selection is a function of den-sity. A l l traits considered in this paper except mass are heritable (Smith and Zach 1979, Schluter and Smith 1986b, unpublished data). Thus, mean offspring size can vary with population density if natural selection on parental reproductive success is a function of adult size and population density. Natural selection on juvenile sur-vival has already been demonstrated in this population of Song Sparrows (Schluter and Smith 1986a); whether the intensity of selection varies with population density, causing density-dependent variation in the average size of living offspring, will be con-sidered below. Environmental influences accounted for a large part of the density dependent vari-ation in wing length. If density-dependent variation in size was of environmental origin, then environmental variation in mass and tarsus length is predicted to be negatively correlated with density, and environmental variation in wing length should be posi-tively correlated with density. A l l correlations were in the expected direction (Table 4-III), but only variation in wing length was significantly correlated with population density. Note that all three correlations have a slightly negative bias for females due to the method used to correct for sexual dimorphism (see Methods). Correlations for male traits do not share this bias. There was no suggestion that natural selection on reproductive success, as a func-tion of parental size, contributed to the variation in mean size of offspring produced. For each year's data, I calculated the standardized directional selection differentials on adult size, for selection resulting from differential reproductive success (numbers of offspring recruited into the breeding population). Selection differentials (Table 4-IVa) were calculated for each trait (mass, wing, tarsus) separately for each year and sex, 81 following the procedures outlined in Schluter and Smith (1986a). Very few of the selec-tion differentials were statistically significant at P=0.05 (4 of 78, roughly the number expected by chance). For neither sex, and for none of the three traits, was there a significant correlation between population density and intensity of selection (all r<0.3, P>0.30). Although I could not directly test the hypothesis that natural selection on survival of offspring was density-dependent, there is indirect evidence that density-dependent selection was not responsible for the observed variation in the size of offspring in fall. Because offspring were measured when at least 60 days old, it is possible that size dependent natural selection occurred before this age. I have no way of testing this possibility. However, if selection pressures were similar before and after offspring were measured, then one should find that intensity of selection on over-winter survival is correlated with density. Differentials of directional selection on size due to variation in over-winter survival of juveniles (Table 4-Va) were calculated as described in Schluter and Smith (1986a). Selection differentials were not significantly correlated with density for either sex, or for any of the three traits (range P=0.13 to 0.96). However, some of the correlation coefficients were relatively strong: r=0.44, n = l l for male mass, and 7"=0.51, 7t=10 for female tarsus. These positive correlations between selection and population density are in the wrong direction to have caused the observed annual variation in size of juveniles in fall. Further evidence suggests that the intensity of selection on mass and tarsus length varied with density, even though the correlations (previous paragraph) were not statis-tically significant. Although the negative correlations between population density and average mass and tarsus length of juveniles in fall were relatively large, correlations between density in the year of birth and the average size of successfully over-wintering birds approached zero (Table 4-III). These data suggest that over-winter selection for larger individuals counteracted the density-dependent variation observed in the size T a b l e 4 - I V a . C o e f f i c i e n t s o f d i r e c t i o n a l s u r v i v a l s e l e c t i o n o n size a n d the e n v i r o n m e n t a l d e v i a t i o n i n size; t h e f i tness c o m p o n e n t w a s s u r v i v a l o f j u v e n i l e b i r d s o v e r t h e i r f i r s t w i n t e r . C o r r e l a t i o n s b e t w e e n s ize a n d t h e e n v i r o n m e n t a l c o m p o n e n t o f s ize w e r e p a r t i a l l e d o u t b e f o r e t h e i n t e n s i t y o f s e l e c t i o n o n t h e e n v i r o n m e n t a l d e v i a t i o n i n s ize w a s d e t e r m i n e d . 4 - I V a ) p r e s e n t s s e l e c t i o n o n b o d y t r a i t s , w h i l e 4 - I V b ) s h o w s d a t a o n b i l l s i ze . N o s e p a r a t e d a t a o n s e l e c t i o n o n m a s s a r e p r e s e n t e d , as m a s s w a s n o t h e r i t a b l e a n d t h e r e f o r e v a r i a t i o n i n m a s s w a s c o m p l e t e l y o f e n v i r o n m e n t a l o r i g i n . C o e f f i c i e n t s a r e t h e s l o p e s o f l i n e a r r e g r e s s i o n s o f s u r v i v a l (0 o r 1) o n s ize , s i ze b e i n g s t a n d a r d i z e d t o a m e a n o f 0 a n d s . d . o f 1. C o e f f i c i e n t s w e r e c a l c u l a t e d s e p a r a t e l y for e a c h y e a r a n d sex . L e v e l s o f s i g n i f i c a n c e w e r e f r o m t - tes t s , c o m p a r i n g t h e m e a n s i zes o f i n d i v i d u a l s t h a t d i e d a n d s u r v i v e d . N u m b e r s i n p a r e n t h e s e s a r e s a m p l e s izes . S i z e E n v i r o n m e n t a l C o m p o n e n t  W i n g T a r s u s M a s s W i n g T a r s u s L e n g t h L e n g t h L e n g t h L e n g t h Y e a r D e n s i t y M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e 1975 36 -0 .69 (58 )** - 0 . 8 6 ( 4 2 ) - 0 . 0 1 ( 5 8 ) - 0 . 1 3 ( 4 2 ) 0 .21(58) 0 .60(42) -0 .11 (18) -0 .10(13) - 0 .05 (18 ) - 0 . 1 3 ( 1 3 ) 1976 30 - 0 . 1 6 ( 4 1 ) 0 .02 (49 ) - 0 . 3 3 ( 4 1 ) - 0 . 2 0 ( 4 8 ) -0 .17(41) 0 .15(49) -0 .14 (39) -0 .05 (40) - 0 . 3 9 ( 3 5 ) * - 0 . 0 2 ( 3 7 ) 1977 4 5 0 .04 (52 ) 0 .04 (53 ) 0 . 17 (53 ) - 0 . 1 7 ( 5 5 ) 0 . 3 1 ( 5 3 ) ° 0 .12(54) -0 .21 (42) -0 .07 (41) 0 .14(40) - 0 . 2 4 ( 4 0 ) 1978 4 8 - 0 . 1 3 ( 7 3 ) 0 .08 (68 ) - 0 . 2 2 ( 7 3 ) ° - 0 . 1 7 ( 6 7 ) 0 .33(73)* 0 .35(68)** 0 .08(54) -0 .01 (43) - 0 . 2 5 ( 5 9 ) ° - 0 . 2 1 ( 4 9 ) 1981 18 - 0 . 1 3 ( 2 0 ) 0 .18(17) — — 0.03(20) 0 .09(17) -0 .02 (13) -0 .05 (10) — — 1982 28 - 0 . 0 3 ( 4 7 ) - 0 . 1 3 ( 5 0 ) - 0 . 0 9 ( 4 7 ) - 0 . 1 5 ( 5 0 ) 0 .23(47)* 0 . 1 8 ( 5 0 ) ° - 0 .07 (26 ) 0 .01(37) — — 1983 56 0 .02 (60 ) - 0 . 1 6 ( 5 5 ) - 0 . 1 3 ( 6 0 ) - 0 . 0 1 ( 5 5 ) 0 .11(60) 0 .23(55) 0 .14(43) 0 .42(40)* -0 .23 (14 ) - 0 . 6 3 ( 1 8 ) * 1984 53 - 0 . 1 0 ( 5 7 ) 0 .46(56)** 0 .17 (57 ) 0 .14(57) 0 .29(57)* 0 .53(57)** 0 .03(43) 0 .08(43) 0 .08(22) 0 .23(32) 1985 72 - 0 . 4 3 ( 3 9 ) ° - 0 . 3 0 ( 6 7 ) " 0 .03 (39 ) 0 .02(67) 0 .11(39) -0 .08 (67 ) - 0 . 5 6 ( 2 2 ) ° 0 .24(43) 0 .04(18) 0 .04(42) 1986 63 0 .21 (39 ) 0 .06(31) - 0 . 0 7 ( 3 9 ) 0 .51 (31 )* 0.51(39)** 0 .55(31)** -0 .09 (33) - 0 . 4 2 ( 2 5 ) ° 0 .03(29) 0 .70 (21 )* 1987 56 - 0 . 2 8 ( 2 6 ) - 0 . 1 7 ( 2 4 ) - 0 . 0 0 ( 2 6 ) - 0 . 3 0 ( 2 5 ) 0 .81(26)** 0 .59(24)** 0 .07(14) -0 .21(16) - 0 .04 (13 ) - 0 . 3 2 ( 1 4 ) ° i t P < 0 . 1 0 , * P < 0 . 0 5 , * * P < 0 . 0 1 T a b l e 4 - I V b . C o e f f i c i e n t s o f d i r e c t i o n a l s u r v i v a l s e l e c t i o n o n b i l l s ize a n d t h e e n v i r o n m e n t a l c o m p o n e n t o f b i l l s ize; s e l e c t i o n w a s o n s u r v i v a l o f j u v e n i l e b i r d s o v e r t h e i r f i r s t w i n t e r . D a t a a r e p r e s e n t e d as i n T a b l e 4 - I V a . S i z e E n v i r o n m e n t a l C o m p o n e n t  "TJuT BUI B U I B i l l B i l l B i l l L e n g t h W i d t h D e p t h L e n g t h W i d t h D e p t h Y e a r M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e 1975 0 .1 0 (1 8 ) 0 .10 (13 ) 0 . 03 (18 ) -0 .23 (13 ) - 0 . 1 6 ( 1 8 ) 0 .93(13) 0 .42(18)* 0 .08(13) 0 .06(18) 0 .72 (13 ) 0 .40 (18 )* 0 .57(13) 1976 0 .0 2 (3 9 ) 0 .27 (40 ) - 0 . 0 4 ( 3 9 ) 0 .26(40) - 0 . 3 9 ( 3 9 ) * 0 .34(40)* -0 .02 (39) 0 .10(40) 0 .01(39) 0 .16 (40 ) - 0 . 1 0 ( 3 9 ) 0 .04(40) 1977 0 . 0 1 (4 3 ) 0 . 3 6 ( 4 2 ) * 0 .14 (43 ) 0 .23(42) 0 .01(43) 0 .42(42)* -0 .16 (43) 0 .01(42) 0 .09(43) 0 .26 (42 ) 0 .14(43) 0 .24(42) 1978 0 .1 9 (5 9 ) 0 .47 (50)** 0 . 3 3 ( 5 9 ) * 0 .18(50) 0 .39(59)** -0 .18(50) 0 .01(59) 0 . 2 5 ( 5 0 ) ° 0 .20(59) 0 .15 (50 ) 0 .30(59) 0 .08(50) 1984 0 .18 (47 ) 0 . 2 2 ( 4 7 ) " 0 .00 (47 ) 0 .19(47) - 0 . 0 4 ( 1 6 ) 0 .04(24) — — — — — — 1985 0 .34 (25 ) 0 .03 (50 ) 0 . 5 0 ( 2 5 ) * - 0 . 3 5 ( 5 0 ) ° - 0 . 1 3 ( 2 5 ) 0 .09(50) — — — — — 1986 0 .61(33)** 0 . 4 1 ( 2 7 ) * 0 .83(34)** 0 .42 (28 )* 0 .28(32) -0 .11(28) 0 .47(10) 0 . 4 0 ( 1 2 ) ° - 0 . 21 (11 ) - 0 . 1 8 ( 1 2 ) -0 .32 (5) — 1987 0 .27 (16 ) 0 .08(16) 0 . 3 8 ( 1 6 ) ° - 0 . 03 (16 ) - 0 .20 (15 ) - 0 . 5 1 ( 1 4 ) ° - 0 .04 (10 ) 0 . 6 0 ( 1 0 ) ° 0 .13(10) - 0 . 4 4 ( 1 0 ) - 0 . 3 6 (6) — ° P < 0 . 1 0 , * P < 0 . 0 5 , * * P < 0 . 0 1 oo CO T a b l e 4 - V a . C o e f f i c i e n t s o f d i r e c t i o n a l s e l e c t i o n o n s ize a n d t h e e n v i r o n m e n t a l c o m p o n e n t o f s ize; f i tness c o m p o n e n t w a s r e p r o d u c t i v e s u c c e s s ( m e a s u r e d i n n u m b e r o f i n d e p e n d e n t y o u n g p r o d u c e d ) . C o r r e l a t i o n s b e t w e e n s ize a n d t h e e n v i r o n m e n t a l c o m p o n e n t o f s ize w e r e p a r t i a l l e d o u t , b e f o r e v a l u e s o f t h e i n t e n s i t y o f s e l e c t i o n o n t h e e n v i r o n m e n t a l c o m p o n e n t o f s ize w e r e c a l c u l a t e d . 4 - V a ) p r e s e n t s d a t a o n s e l e c t i o n o n b o d y t r a i t s , a n d 4 - V b ) p r e s e n t s s e l e c t i o n o n b i l l t r a i t s . C o e f f i c i e n t s a r e the s l o p e s o f l i n e a r r e g r e s s i o n s o f r e p r o d u c t i v e s u c c e s s o n s i ze , w i t h r e p r o d u c t i v e s u c c e s s s t a n d a r d i z e d t o a m e a n o f 1 w i t h i n e a c h y e a r a n d s ize b e i n g s t a n d a r d i z e d to a m e a n o f 0 a n d s . d . o f 1. C o e f f i c i e n t s w e r e c a l c u l a t e d s e p a r a t e l y for e a c h y e a r a n d sex . L e v e l s o f s i g n i f i c a n c e w e r e f r o m t - tes ts , c o m p a r i n g t h e m e a n s izes o f i n d i v i d u a l s w i t h r e p r o d u c t i v e s u c c e s s e i t h e r b e l o w , o r e q u a l t o o r a b o v e t h e p o p u l a t i o n a v e r a g e for t h e y e a r i n q u e s t i o n . N u m b e r s i n p a r e n t h e s e s a r e s a m p l e s i zes . S i z e E n v i r o n m e n t a l C o m p o n e n t  W i n g ' T a r s u s M a s s W i n g T a r s u s L e n g t h L e n g t h L e n g t h L e n g t h Y e a r D e n s i t y M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e 1975 36 -0 .22 (39 ) - 0 . 2 5 ( 3 4 ) ° 0 .14(39) 0 .17(33) 0 .02(38) 0 .14(33) — — — — 1976 30 -0 .01 (35 ) 0 .05 (27 ) - 0 . 1 1 ( 3 5 ) - 0 .07 (26 ) 0 .01(34) -0 .04 (27) -0 .41 (7) — -0 .08 (7) — 1977 4 5 0 .10(40) - 0 . 0 4 ( 4 0 ) 0 .16(40) -0 .02 (38 ) 0 .07(40) -0 .07 (40) - 0 .12 (24 ) - 0 .06 (20 ) - 0 .02 (22 ) - 0 . 0 3 ( 1 7 ) 1978 48 . 0 .01(41) 0 .06 (42 ) 0 .04(42) 0 .08(43) -0 .11 (42) 0 . 1 8 ( 4 3 ) ° - 0 . 11 (30 ) 0 .01(34) - 0 . 0 9 ( 2 7 ) - 0 . 1 0 ( 3 3 ) 1979 62 -0 .00 (65 ) - 0 . 0 0 ( 5 9 ) - 0 . 1 4 ( 6 6 ) 0 .21 (60)* -0 .05 (66) 0 .19(60) 0 .18(53) -0 .04 (54 ) 0 .06(53) - 0 . 1 7 ( 5 6 ) 1981 18 -0 .02 (15 ) - 0 . 1 4 ( 1 0 ) - 0 . 0 7 ( 1 3 ) -0 .18 (5) -0 .05 (16) 0 .15(11) — — — — 1982 28 0 .02(24) - 0 . 0 6 ( 2 0 ) 0 .13(12) 0 .19 (3) 0 .01(25) 0 .20(20) 0 .03(13) — — — 1983 56 0 .01(51) - 0 . 0 6 ( 4 8 ) - 0 . 1 2 ( 3 7 ) 0 .12(32) -0 .04 (52) 0 .14(48) 0 .14(30) -0 .02 (34 ) — — 1984 53 0 .08(52) 0 .09 (45 ) 0 .10(42) 0 .09(33) -0 .06 (52) 0 .21(45)* -0 .07 (29 ) - 0 .05 (35 ) — -0 .32 (8) 1985 72 -0 .13 (69 ) - 0 . 0 6 ( 5 9 ) 0 .09(61) -0 .15 (51 ) 0 .02(69) -0 .16 (60) - 0 .08 (45 ) - 0 .12 (45 ) 0 .23(13) - 0 . 0 6 ( 2 1 ) 1986 63 0 . 2 4 ( 6 4 ) ° 0 .03 (52 ) 0 .09(59) 0 .10(50) -0 .13 (64 ) 0 .14(52) 0 .33(46)* - 0 . 1 8 ( 3 9 ) * - 0 . 1 9 ( 1 8 ) - 0 . 2 6 ( 2 4 ) 1 9 8 7 56 0 .20(48) - 0 . 1 4 ( 3 9 ) - 0 . 0 6 ( 4 5 ) 0 .08(38) -0 .02 (48 ) - 0 . 3 3 ( 3 9 ) * -0 .01 (37 ) - 0 .20 (32 ) - 0 . 0 2 ( 2 1 ) 0 .18 (25 ) 1988 54 0 .28(44)* 0 .04 (46 ) 0 .07(43) 0 .18(46) 0 .09(44) 0 .01(46) 0 .07(34) 0 .11(34) -0 .01 (22 ) 0 .19(28) ° P < 0 . 1 0 , * P < 0 . 0 5 T a b l e 4 - V b . C o e f f i c i e n t s o f d i r e c t i o n a l s e l e c t i o n o n b e a k s ize a n d t h e e n v i r o n m e n t a l c o m p o n e n t o f b e a k s ize; f i tness c o m p o n e n t w a s r e p r o d u c t i v e s u c c e s s . D a t a a r e as p r e s e n t e d i n T a b l e 4 - V a . S i z e E n v i r o n m e n t a l C o m p o n e n t  " B U I B i l l B i l l B i l l B i l l B i l l L e n g t h W i d t h D e p t h L e n g t h W i d t h D e p t h Y e a r M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e M a l e F e m a l e 1977 0 . 1 2 (2 4 ) 0 .15(20) 0 . 1 1 ( 2 4 ) ° - 0 . 2 0 ( 2 0 ) 0 .04(24) -0 .06 (20) -0 .05 (24) -0 .31 (20 )* 0 .13(24) 0 .04(20) 0 .13(24) 0 .04(20) 1978 - 0 . 0 3 ( 3 1 ) - 0 . 0 3 ( 3 3 ) - 0 . 1 2 ( 3 1 ) 0 . 05 (33 ) -0 .05 (31 ) 0 .11(33) -0 .05 (31) -0 .22 (33) 0 .01(31) 0 .05(33) 0 .01(31) 0 .05 (33 ) 1979 0 .05 (55 ) - 0 . 2 0 ( 5 0 ) ° 0 .03(55) 0 . 0 7 ( 5 0 ) ° - 0 . 07 (55 ) 0 . 2 7 ( 5 0 ) ° - 0 .06 (55 ) -0 .08 (50) 0 .29 (55)* -0 .06(50) 0 .29 (55 )* - 0 . 0 6 ( 5 0 ) 1985 - 0 . 2 7 ( 1 9 ) 0 .06(24) -0 .17 (19 ) - 0 . 2 6 ( 2 4 ) 0 .07 (9) 0 .04(15) — — — — — — 1986 0 .02 (29 ) 0 . 1 5 ( 3 0 ) ° - 0 . 11 (29 ) 0 . 04 (30 ) 0 .13(17) -0 .10 (24) — — — — - — 1987 - 0 . 1 0 ( 3 2 ) - 0 . 1 3 ( 2 7 ) - 0 .18 (32 ) - 0 . 0 1 ( 2 7 ) 0 .56 (22)* -0 .01 (22) — — — — — P < 0 . 1 0 , * P < 0 . 0 5 oo Or 86 of juveniles in fall. In contrast, there was very little evidence that over-winter selec-tion on wing length varied with density (correlations between selection intensity and density were r=0.04 for males, and r=-0.02 for females); wing length remained highly correlated with population density for both fall, and over-wintering birds (Table 4-III). Selection on Environmental Components of Size In all years, selection favoured large mass, the result of better over-winter survival of heavier juveniles. I tested for consistency in the direction of selection by performing a one-sample Wilcoxon Signed-Ranks tests on the annual selection differentials (see Table 4-IV for data). The distribution of annual selection differentials was signifi-cantly greater than zero for mass of both males (P<0.01) and females (P<0.005); for each sex, directional selection favoured lighter individuals in only 1 of 11 years. The distributions of selection differentials did not differ significantly from zero for wing or tarsus of either sex, although for wing length of males selection favouring smaller wings approached significance (P<0.10). Selection favoured longer-billed individuals (both male and female) in all 8 years for which data were available (P<0.05). Because mass is not heritable in this population (Smith and Zach 1979, Schluter and Smith 1986b, unpublished data), the variation in mass was environmental or non-additive in origin. There was thus no need to calculate a separate environmental component of mass. For the other traits, I tested whether selection acted on the environmental components of size variation by calculating directional selection differentials on environmental varia-tion in size for each year (Table 4-IV). Because environmental variation in size, and size per se are correlated, I partialled out the effect of size on environmental variation before calculating selection differentials. These selection differentials were then used in Wilcoxon Signed-Ranks tests. Neither wing or tarsus exhibited consistent directional selection on environmental variance; too few years' data exist to statistically examine consistency in direction on the environmental component of beak dimensions. There was no clear evidence that the environmental component of size consistently 87 affected reproductive success, although the results of some statistical tests approached significance. As before, directional selection differentials for reproductive success (num-ber of recruiting offspring; Table 4-V) were calculated for both parental size, and the environmental component of size (correlations with absolute size partialled out). Only-female tarsus length showed a tendency for consistent directional selection, with longer-legged birds being more successful in 9 of 12 years (P=0.10; Wilcoxon Signed-Ranks test). However, the only environmental component of size that showed consistent se-lection was female wing length; in 7 of 9 years females with wings shorter than their mid- parents' were more successful (P<0.10; Wilcoxon Signed-Ranks test). Too few data exist to examine direction of selection on bill size. If size (or the environmental component thereof) consistently affected reproductive success, the effect was too slight to detect. Selection for Parental Care Variation in gene frequencies did not result from the annual variation in mean sizes of offspring produced, or from mass-dependent variation in over-winter survival of juveniles. However, gene frequencies in the breeding population could be altered if parents' ability to care for their offspring was genetically based, and if offsprings' final mass was determined during the period that young were under their parents' care. The genes for good parenting would be carried along because heavier offspring had higher over-winter survival. It was not possible to determine if parental care was genetically based, or if there was selection on the environmental component of size, before offspring were measured. However, if parental care has a genetic basis, the sizes of offspring produced must be repeatable. Adult size of offspring produced was not repeatable between nests of a parent, and thus parents (directly or through variation in territory quality) did not consistently control environmental variation in the sizes of their offspring. I looked for repeatabil-ity in environmental variation of adult size of birds from different nests of the same 88 parents within a year, in order to see whether parents/territories are part of the en-vironmental variation. Because most sparrows in this population produce at least 2 broods of young in a given year (Smith and Roff 1980), the pairs of broods compared were the first and second brood of parents in one year. Available data from all years were combined to increase sample sizes. To correct for annual differences in average environmental variation in size, all values for the environmental component of size were standardized to a mean of 0 and standard deviation of 1 within each year. The means of these corrected values for each brood were used as the data points in the analyses. There was no significant repeatability of environmentally determined variation in size between first and second broods, with only the repeatability for beak depth approach-ing statistical significance (r=0.24, n=46, P=0.11; correlations for all other traits were smaller, and with larger sample sizes). There is no evidence of a genetic basis for vari-ation in parental care, insofar as parental care affects nestling condition, as there was no statistically significant repeatability of environmental variation in offspring size. Correlations Among Traits It is necessary to look for correlations in the variation among traits, because the variation in all morphological traits considered in this paper may have a common envi-ronmental or genetic basis. Phenotypic and genetic correlations have previously been calculated for the Song Sparrows of Mandarte Island by Schluter and Smith (1986b). They used a sub-set of the data presented here, but the correlations in the entire data set differed little from those previously found. These previous analyses showed occa-sional negative correlations (both phenotypic and genetic), but the only statistically significant correlations were positive. New values for genetic and phenotypic correla-tions are not presented in this chapter. Correlations between environmental variation in the six traits are given in Table 4-VI; correlations between all traits were positive. Parental Care and the Origin of Environmental Variation As van Noordwijk (1988) notes, environmental variation in size and nestling condi-Table 4-VI. Correlations between environmental components of size. Data below the diagonal are for males, and those above it for females. Data are measurements from individual sparrows. Mass is the absolute value of mass (a non-heritable trait), while for all other traits data are residuals from mid-parent - offspring regression. Data are standardized by year; numbers in parentheses are sample sizes. Mass Wing Length Tarsus Length Bi l l Length Bi l l Width Bi l l Depth Mass — 0.20(380)** 0.18(299)** 0.23(198)** 0.30(198)** 0.02(175) Wing Length 0.03(351) — 0.11(288)° -0.13(189)a -0.02(189) 0.07(166) Tarsus Length 0.07(265) 0.09(259) — -0.01(192) 0.01(192) -0.05(169) Bi l l Length 0.30(193)** 0.18(187)* -0.02(185) — 0.43(198)** 0.08(175) Bi l l Width 0.39(194)** 0.19(188)** -0.07(186) 0.47(193)** — 0.35(175) Bi l l Depth 0.27(177)** 0.18(171)** -0.05(171) 0.25(177)** 0.37(177)* — P<0.10, *P<0.05, **P<0.01 90 tion may be synonymous. It is possible to explain the density-dependent environmental variation as simply another aspect of density-dependent variation in reproductive suc-cess of Song Sparrows (e.g. Arcese and Smith 1988, Hochachka et al. 1989), with "poorer" quality offspring being produced at higher densities. Alternatively, the in-fluence of density on phenotype may occur after the young leave their parents' care, when juveniles are in their final stages of growth. I examined this issue by looking for correlations between an index of nestling nutritional condition (the index used in Chapter 3) and environmental variation in the traits under investigation. Ideally, one should look for such correlations between years, but nestling data were available only for 5 (for bill traits) to 7 breeding seasons, and I was forced to look for correlations within years. As a result, I assume that variation in nestling condition within a year produces the same results as variation among years. Variation in the environmentally-determined component of size was correlated with nestling condition for only some traits. For both sexes, adult mass and nestling condition were correlated (both r>0.15, P=0.017, rc>120). Birds in better condition as nestlings also had longer tarsi as adults (r>0.17, P<0.05, TI>120 for both sexes). Bi l l width was weakly, positively correlated with nestling condition (r=0.3, P=0.035, n=51) for females, but not for males (P=0.85). Conversely, bill depth was correlated with nestling condition for males (r=0.5, P=0.003, re=31), but not females (P=0.59). For neither sex was there a significant correlation between environmental variation in wing length or bill length, and the index of nestling condition (all r<0.22, P>0.12, ri>50; data from all years combined). The overall lack of statistical significance was not due to variation among years in the correlation between nestling condition and the environmental component of size; when correlations were performed on each years's data separately, there were still no significant correlations. Thus, birds' mass, tarsus length, and some bill dimensions are correlated with their condition as nestlings. Wing length was not related to a bird's condition as a nestling. 91 DISCUSSION This paper shows that for the traits measured, environmental variability can not only dictate morphological variation among populations (James 1983, James and Ne-Smith 1988), but can also cause,systematic annual variation in morphology within populations. Environmental variation can alter the average size of offspring in the population and natural selection can act selectively on the environmental variance in size, without changing gene frequencies. Some of these results have previously been found by van Noordwijk et al. (1988) and Alatalo et al. (1990), who show that se-lection occurred on environmental variance in nestling size for populations of Great Tits and Collared Flycatchers, respectively. The time when selection occurred was not known in these two studies, but may have been very shortly after the young fledged and were still in their parents' care. In contrast, Song Sparrows were not measured until at least sixty days old in this study, when they had been independent of their parents for at least one month. Selection on mass of Song Sparrows had to occur even later in offspring's lives. Thus, my study shows that selection on environmental variance can occur well after offspring have left their parents' care. Two conceptual frameworks already exist for the phenomena described in this pa-per (van Noordwijk 1988). The first is the notion of nutritional condition. Differential survival as a function of environmental variation in mass is assumed in all cases where indices of body condition are used (e.g. Slagsvold 1982, Boersma and Ryder 1984, Arcese and Smith 1985). A bird that is in poor nutritional condition as it is growing is unlikely to achieve its maximum potential size. Hence, good (poor) nutritional condi-tion during growth will result in a final size that falls above (below) that predicted by a mid-parent - offspring regression (my measure of environmentally induced variation in size). Environmentally induced variation in size may be synonymous with nutritional condition during growth. The second framework in which my observations fall is that of reaction norms and 92 genotype-environment interactions (e.g. Stearns 1989). Reaction norms assume that the observed pattern of phenotypic plasticity is constrained, and that the constraint has a genetic basis. Decreased size of offspring that grow under poor nutritional condition is a reaction norm. Given a genetic basis, reaction norms can be altered by selection. Thus, we might expect that inter-annual variation in the size of Song Sparrows was adaptive. A major influence on variation in body size, but not bill size, was the dependence of mean size of offspring on population density in the year of birth. Density-dependent variation in morphology was also suggested for Darwin's finches (Boag 1983). In con-trast Alatalo and Lundberg (1986) found no evidence for density-dependence in tarsus length of Pied Flycatchers. Density-dependent variation in Song Sparrow size occurred in the environmental component of variation (Table 4-III). The most perplexing aspect of this variation in size was that although mass and tarsus length tended to decrease as population density increased, wing length increased with higher density. Lower aver-age mass and tarsus length with higher population density is explicable if offspring are in poorer nutritional condition when population density is high. High population den-sities represent poor conditions for reproduction in this population of Song Sparrows (Arcese and Smith 1988). However, the increase in average wing length with increas-ing density is a reaction norm that seems unrelated to nutritional condition. Within individual birds the correlations between environmental variation in mass, wing and tarsus were positive, although very weak for males (Table 4-VI), and the traits should vary in the same direction. I have not found a reasonable explanation for this result, although the possibility that longer-winged birds in high density populations, having lower wing loadings, are better adapted to long-distance movement (e.g. Chapman 1940) is tantalizing. Natural Selection and Genetic Response Although mass of Song Sparrows is not likely to evolve in the population of Song 93 Sparrows under study, there is potential for the evolution of bill length and tarsus length. Heavier birds had consistently higher over-winter survival, but mass cannot evolve because mass does not show significant heritability in this population of Song Sparrows (Smith and Zach 1979, Schluter and Smith 1986b, unpublished data). In contrast, bill length and tarsus length are heritable traits in this population (e.g. Schluter and Smith 1986b, Table 4-1). Schluter and Smith (1986a) showed that longer-billed, shorter-legged juvenile female sparrows had higher survival. With more data, it was also found that longer-billed juvenile males also consistently survived better than shorter-billed males (Table 4-IVb). Schluter and Smith (1986a) also found that females with short bills and long tarsi had higher reproductive success in one of five years they examined. Average bill and tarsus lengths could evolve given the heritability of, and selection on bill and tarsus length. I investigated whether selection acted on genetic or environmental variance in those instances where Schluter and Smith (1986a) identified natural selection occurring in this population of Song Sparrows. The data presented by Schluter and Smith (1986a) are a sub-set of the data used in this paper. Thus, I was able to re-analyze their data for those traits for which they found statistically significant natural selection, examining whether selection was on size per se or the environmental component of size. In each case where Schluter and Smith (1986a) identified a trait as undergoing selection, I calculated a selection gradient (e.g. Price and Boag 1987) with the two components of the gradient being trait size and environmental variation in trait size. Schluter and Smith (1986a) found selection for short tarsus and long bill in over-winter juvenile females (with combined data from 1975-78 cohorts of young); therefore, the component of a selection gradient that best explains selection for short tarsus will have a negative coefficient, and positive for long bill. Schluter and Smith (1986a) also found that in 1979 selection favoured longer-legged shorter-billed females, through higher reproductive success. The component of the gradient that best explains this 94 observation should be positive for tarsus, and negative for bill length. My analyses suggest that the instances of natural selection found previously (Schluter and Smith 1986a) were often selection on environmental, and not genetic variance. The gradients (f3) for selection of higher survival (Table 4-VII) indicate that selection for long bill and short tarsus was the result of selection on the environmental component of variation in these two traits. The f$ coefficients for selection on environ-mental variation were larger and in the expected direction, whereas the coefficients for size per se were smaller and in the opposite direction. For selection favouring higher reproductive success, the larger, positive j3 coefficient for selection on tarsus length was for selection on size, whereas the larger, negative coefficient for selection on bill length was for selection on the environmental component of bill length (Table 4-VII). Thus, my analyses suggest that of the instances of selection reported by Schluter and Smith (1986a), tarsus length was the only trait for which genetic variance was altered by selection. Note that differences in magnitude of (3 values may be difficult to inter-pret, because size and environmental variation in size are highly correlated (r>0.90) for each of the traits considered above and because none of the gradient coefficients are statistically significant (Price and Boag 1987). If my conclusions are correct, they are difficult to interpret in terms of nutritional condition. The analyses suggest that yearling females with environmentally induced short tarsi (i.e. resulting from poor nutritional condition during growth) survived bet-ter than birds experiencing better conditions during growth. Likewise, the analyses suggest that those females experiencing poorer conditions during growth (shorter bills) had better reproductive success in 1979. Given the non-significance of the analyses, these results should not receive too much weight by themselves. However, similar analyses to those conducted here should be performed for other species in which selec-tion has been observed on morphological traits. An analysis similar to mine indicated that only environmental variation in tarsus length was selected in Collared Flycatchers 95 Table 4-VII. Directional selection gradients (/?) for overwinter survival ("Sur-vival") of juvenile females from 1975-78 cohorts combined, and for repro-ductive success ("Reproduction") of females in 1979. P-values are the signif-icance levels for the partial regression coefficients presented in (3. Separate selection gradients were calculated for each trait; the two components of each gradient are size, and environmental variation in size. Tarsus Length Tarsus Environmental Variation n Bil l Length Bi l l Length Environmental Variation n Survival Reproduction P 0.15 0.59 -0.32 0.24 139 0.72 -0.53 44 P 0.06 0.15 -0.57 0.28 0.93 0.08 145 0.02 -0.23 50 0.98 0.75 96 (Alatalo et al., 1990). Environmental Variation and Adaptive Response Parents have reason for controlling environmental variation in the morphology of their offspring. It would be advantageous for parents, particularly those of a sub-optimal genotype, to alter the environmental component of their offspring's size to more closely match the optimal type if there is selection on a trait. This applies whether there is selection on the environmental component of size (e.g. selection for good nutritional condition), or selection on morphology per se. Parent sparrows can affect the mass, tarsus length, and to some extent bill sizes of offspring by influencing their offspring's nutritional condition. This was indicated by the significant corre-lations between the environmental component of these traits and nestling condition. Parents can affect more than just the genetic component of beak size, because offspring in the nest are dependent on their parents for food (and hence nutritional condition). This has been shown for other species (e.g. Ricklefs and Peters 1981, Alatalo and Lundberg 1986). However, adult sparrows could not affect the environmental com-ponent of variation in wing length, as there was no significant correlation between nestling condition and this trait. Adult Song Sparrows were not able to systematically influence the environmental variance in size of their offspring. Adults sparrows did not consistently produce young of a given size, even for those traits in which parents could influence environmental variation in offspring size. There were no statistically significant repeatabilities of environmental variation in bill or body size between first and second broods in a year (although the largest repeatability was for a bill measure). Because of this lack of repeatability, I conclude that differences among parents or the territories that they occupy do not allow some parents to consistently affect the environmental component of size so as to produce "better" offspring than other parents. There is no evidence of consistent differences among parents in their ability to raise offspring in this population 97 of Song Sparrows (Chapters 2, 3). Further support for the view that parental care does not set offspring size is the finding (Smith and Arcese 1988) that variation in the supply of food to offspring does not affect adult size of surviving offspring in this population of Song Sparrows. The reaction norms of body size with density did not produce the optimal size of offspring for each population density. There was no indication that lighter, shorter-legged, longer-winged birds were produced at high density because such birds have either a survival or reproductive advantage at high density. The intensity of selection on wing length from differential survival showed almost no correlation with popula-tion density. Heavier, longer-legged birds tended to be favoured at high densities — opposite to the observed variation in mass and tarsus length. Because population den-sity in the year of birth and in the subsequent breeding season are highly correlated (Hochachka et al. 1989), controlling the average size of offspring to match the opti-mum in the year of birth could enhance reproductive success of offspring. However, the intensity or direction of selection on size for higher reproductive success was not correlated with population density. Implications My results have two main implications for the study of natural selection and morphological change. First one cannot assume that natural selection always alters additive genetic variance, or that phenotypic variation is the result of evolution. This conclusion was previously reached by van Noordwijk (1988, van Noordwijk et al. 1988) and Alatalo et al. (1990). Although there are many methods for detecting natural selection (see Chapter 3 in Endler 1986), we must rule out phenotypic plasticity be-fore determining that selection leads to evolution. For example, Dhondt et al. (1979) explained a long-term decline in mean mass of juveniles in a population of Great Tits as the result of evolutionary changes brought on by increased population density. However, the present study showed that a decrease in average mass of offspring with 98 increased density resulted from changes in only the environmental component of size. It is necessary to demonstrate that selection altered genetic variance before an evolu-tionary response can be presumed. This conclusion does not apply only to traits with little or no heritable variation such as mass of Song Sparrows (e.g. Smith and Zach 1979). I also showed that annual differences in average wing and tarsus length were the result of changes in the environmental component of size. The second implication of this study is that even when selection alters genetic variance, it may not be possible to predict the response of the population in the next generation. This is because a selection event, especially a large one, may alter not only genetic variance in a population, but also the environment that the population occupies. This study also indicates the difficulty of predicting how environmental variance will be altered by changing conditions. Although mass and tarsus length decreased as population density increased, as might be expected ("runtier" offspring are produced when conditions are poorer), mean wing length of juvenile Song Sparrows increased with increased density (Figure 4-1). Thus, genotype-environment interactions are not always equivalent to variation in nutritional stress. Not all traits responded to the same environmental perturbations; for example, body but not bill size varied with population density. Environmentally induced variance was even manifested at different times in the development of young birds, with variation in mass, tarsus and bill (some traits) being affected while young were still in the nest, although wing length was not correlated with nestling condition. Further comparisons with other species are needed before the generality of these findings can be assessed. 99 C H A P T E R 5 H o w M U C H S H O U L D R E P R O D U C T I O N C O S T ? Summary - This chapter makes predictions concerning variation in costs of repro-duction from detailed knowledge of a population of Song Sparrows. I distin-guish variation in fecundity (e.g. clutch size) from variation in reproductive effort. Although the two terms may be related, they are not necessarily so. The principal aim of this paper is to illustrate a method for predicting min-imum costs of increasing clutch size. Three constraints on the optimal cost of reproduction are considered: intra-seasonal variation in reproductive suc-cess, variation in parental age, and interactions between different effects of increasing current clutch sizes. Predictions of reproductive cost can be used to determine whether experimental manipulation of clutch size should show a statistically detectable cost of reproduction. Fruitful avenues for future empirical research include investigating whether ability of adults to raise off-spring changes with age, and whether reproductive effort varies with season and parental age. 100 I N T R O D U C T I O N Life-history theory is based on the assumption of trade-offs between the net ben-efits of various activities in organisms' lives. For instance, in iteroparous organisms increased reproductive effort may decrease survival (e.g. Williams 1966a). There have been many recent attempts to demonstrate trade-offs in avian reproduction (see Linden and M.0ller 1989, and Partridge 1989 for recent reviews). Several studies have manipulated clutch or brood size to create variation in reproductive rate. Artificial variation in reproductive rate is created to control for environmental effects, which can otherwise obscure the expected trade-offs (van Noordwijk and de Jong 1986): high survival and high reproductive success are often positively correlated in nature (e.g. Hogstedt 1981, Smith 1981b, Nol and Smith 1987). Some experimental manipu-lations have shown that increased brood size results in increased adult mortality (e.g. Nur 1988, Dijkstra et al. 1990), or in lower reproductive success in subsequent years (e.g. R0skaft 1985, Gustafsson and Sutherland 1988). However, other studies have failed to show any effect of the manipulations (de Steven 1980, Finke et al. 1987, Korpimaki 1988, Pettifor et al. 1988). In some studies sample sizes may have been too small (e.g. de Steven 1980), but even with well over 100 nests Finke et al. (1987) found no cost to parents or effect of increased brood size on survival of juveniles for House Wrens. A general problem with these manipulations is that they do not affect one major energetic expense incurred by female birds, egg production (Nur 1984b). There is therefore need for an estimate of the cost of reproduction that does not depend on manipulating broods. An independent estimate of reproductive cost would aid in estimating sample sizes needed in brood-size manipulations or estimat-ing the cost of producing eggs. Here I suggest such a method. First, I assume that the observed reproductive rate maximizes fitness, and that fitness can be estimated by lifetime reproductive success (this requires that the population not be continuously increasing or decreasing in size; e.g. Horn and Rubenstein 1984). Under these assump-101 tions, observed clutch sizes are optimal because, although the production of extra eggs would increase current reproductive success, it would also decrease future reproductive success and therefore fitness. Thus, one can estimate the theoretical minimum cost of reproduction for an individual as being just greater than the benefit of the increased rate of current reproduction. The decrease in future reproduction (cost of reproduc-tion) can be presented in terms of decreases in survival of adults or other appropriate units. The goal of any exploration of optimality is to identify constraints (e.g. Maynard Smith 1978). This chapter focuses on three potential constraints on economic viability of costs of reproduction. First, is whether birds can vary their reproductive effort with the time in the season that they are nesting. Offspring born later in the year are less likely to recruit into the breeding population than early born young, in many species of birds (e.g. Cooke et al. 1984, Newton and Marquiss 1984, Nilsson and Smith 1988, Chapter 2). Thus, the economically acceptable cost of reproduction may vary with the date that birds start nesting in a season. The second constraint is the inherent flexibility of reproductive effort with age. The economically viable cost is a determined by a trade-off between present and future fecundity, and older birds may have a lower future fecundity than younger individuals. Hence, incurring a greater cost and increasing current reproductive effort may be more economical for older birds (Williams 1966a). However, although studies have suggested that reproductive effort varies with age (e.g. Pugesek and Diem 1990), the generality of this pattern as not been established. Some birds may, in fact, be constrained to a single level of reproductive effort throughout their lives. In this chapter, I explore the consequences of reproductive effort being both variable and invariant through a bird's life. The third constraint is that the cost of reproduction is not a single item. Parental mortality is one potential cost, lower future fecundity (without mortality rate being altered) is another cost. The mechanism causing mortality could be nutritional stress or predation (Lima 1987). 102 Further, parents may not suffer the cost of increasing current clutch size, but transfer this cost to their offspring; increased clutch size may not affect parents, but could only lower the survival of offspring. I examine the inter-relationship of these different responses to increased clutch size. In this chapter, the methods for calculating cost of reproduction are applied to a population of Song Sparrows (Melospiza melodia). As the exact constraints on the economical cost of reproduction are unknown, the chapter should be viewed as an exploration of potential patterns rather than a quantification of actual costs. The different constraints produce different predictions about how the actual cost of re-production should vary with season and adult age. These predictions will be useful for future tests of which constraints actually operate. Assuming that the appropriate constraints can be identified, the methods outlined here are useful in conjunction with experiments attempting to identify and quantify a cost of reproduction. This chapter contains an example of such a use; I briefly discuss estimates of the costs of reproduc-tion for three species of birds (House Wren, Great Tit , and Collared Flycatcher) for which attempts have been made to demonstrate a cost of reproduction. T H E S T U D Y A N I M A L Data used to estimate costs of reproduction come from the population of Song Sparrows living on Mandarte Island. Data used in this chapter come from the period 1975-89. Data required are: rates of nest failure, clutch size, survival of offspring while under parental care, and probability of offspring recruitment into the breeding population. T H E C O S T OF P R O D U C I N G E X T R A E G G S Clutch size can increase for two reasons, only one of which is relevant to estimating the cost of reproduction. Clutch size could increase if effort required per egg declined. For instance, clutch size may increase with female age (e.g. Klomp 1970) because foraging efficiency increases with age (e.g. Gochfeld and Burger 1984). Thus older 103 females could expend the same foraging effort but collect more nutrients for their effort than younger females, allowing them to lay larger clutches with no change in cost to the female. In the analyses that follow, I will not discuss this type of variation in clutch size. Instead, I will consider increases in clutch size that result in declines in adult survival or future reproductive success. In other words, I assume that clutch size per se is not directly related to reproductive cost, but that changes in clutch size relative to the clutch size that birds would normally lay are important. Given this perspective, I define "reproductive effort" as any expenditure (i.e. time, energy) into reproduction that can result in a decline in adult survival or future reproductive success. Another term that requires definition is "extra eggs". Throughout this chapter I will be comparing reproductive success of birds laying clutches of normal size with the reproductive success that would have resulted if one "extra egg" was laid. Thus "extra" eggs are entirely hypothetical. If a bird is observed to lay 4 eggs then the "extra egg" (not laid) would have been the fifth; if 3 eggs are laid then the "extra" would have been the fourth. This treatment is the result of the basic assumption of my analyses: that observed clutch size is optimal. Hence, under this assumption, "extra" eggs are never laid because this would result in sub-maximal lifetime reproductive success for the parents. To estimate the increase in reproductive success that would result from increased clutch size, I constructed a model that predicts reproductive success (measured in number of offspring recruiting into the breeding population) for a given clutch size. I did not use direct measurements of numbers of recruits produced by birds that laid clutches of different sizes, because there is evidence (e.g. Pettifor et al. 1988) that individuals laying different sized clutches are inherently capable of raising different numbers of offspring, instead, I estimated the parameters of the model as averages over all clutch sizes and over all years for which data were available (i.e. individual 104 eggs and not clutches were the units over which averages were taken). Parameters were estimated separately for each year and geometric mean values (e.g. Boyce and Perrins 1987, Yoshimura and Shields 1987) were used in the model, because there is annual variation in all reproductive parameters (e.g. Arcese and Smith 1988). The resulting model predicts the reproductive success expected for an average Song Sparrow in an average year, and ignores inter-individual variation in reproductive potential (Nol and Smith 1987). Combining data among years also does not take into account small differences in reproductive success with female age (Nol and Smith 1987, unpubl. data). The model is described in Appendix 1. For production of extra eggs to be economical, lifetime reproductive success (LRS) with additional eggs should be at least equal to LRS without the additional eggs. This can be expressed algebraically as Rp + ARP + (s - cf)Rf > Rp + sRf (1) where the left side of the inequality describes reproductive success when extra eggs are laid, and the right side describes cases in which no additional eggs are laid. Rp is the number of recruiting offspring produced from the current year's clutches without extra eggs, ARP is the increase in number of recruits from the current year's clutches due to added eggs, Rf is the number of recruits produced in the future (i.e. all subsequent breeding seasons during a female's lifetime), s is the probability of an adult surviving from one breeding season to the next, and Cf is the proportional reduction in adult survival or future reproduction resulting from laying extra eggs in the current year. From life-history theory (e.g. Stearns 1976), laying extra eggs in the current year is expected to both increase the number of offspring produced in the current year (ARp) and decrease adult survival and/or future reproductive success (c/). To determine the theoretical minimum cost of reproduction one must know how finely a bird can regulate its reproductive effort. If reproductive effort can vary with 105 female age (e.g. Pugesek and Diem 1990), the decision to produce extra eggs in one year will not automatically result in the production of additional eggs in all nesting attempts throughout the bird's life. Future reproduction, Rf, does not directly depend on change in present reproduction ( A P t p ) in this case. Alternatively, high reproductive effort in one year may constrain birds to high reproductive effort (i.e. larger clutches) in all breeding seasons. In this case, Rf in eqn. (1) is affected by the increase in clutch size in the current year. I therefore consider both the possibility that females can vary reproductive effort with age and the possibility that reproductive effort is invariant throughout a lifetime. Reproductive Effort Varies with Female Age I first consider the costs of reproduction that the sparrows could incur to increase clutch size without decreasing lifetime reproductive success, if they could vary effort with age. I assume that a bird can start to produce extra eggs in any year of breeding. Given this assumption, expected future reproduction, Rf, can be estimated directly from data. For the Song Sparrows on Mandarte Island the geometric mean Rf is 0.99, 0.66, and 0.41 breeding offspring for females aged 1, 2, and 3 years, respectively. Equation (1) can be rearranged to solve for the cost of producing the extra eggs (cf): This equation (Williams 1966b presents a slightly different form) indicates that pro-duction of extra eggs is favoured when the cost (in increased mortality or reduced future fecundity of adults) is less than or equal to the ratio of the increase in present reproduction resulting from the added eggs (ARp) to expected future reproductive success if extra eggs were not laid (Rf). Note that the economically viable cost is independent of the adults' probability of survival to the next breeding season. By "economically viable cost", I mean any decrease in females' survival or future repro-ductive success, resulting from an increase in clutch size in the current year, that yields a higher lifetime reproductive success than possible without increasing clutch size. 106 Assuming that observed clutch sizes are optimal, the cost of producing an ex-tra egg, Cf, must be greater than ARp/Rf, because Song Sparrows do not produce the hypothetical extra eggs. ARp can be estimated from the model describing repro-ductive success (Appendix 1), and Rf has been estimated (above), so the theoretical minimum cost of producing additional eggs can be calculated. Figure 5-1 plots the maximum costs of reproduction allowable before the production of extra eggs becomes uneconomical. Figure 5-1 shows the case in which one extra egg is laid in each nest during the year, 2.5 nests are produced by the average female, and clutch sizes (before additional eggs) are 3, 4, and 3 eggs for the first through third clutches respectively. These conditions closely match observed values for Song Sparrows on Mandarte. Be-cause extra eggs are not produced, the actual cost of producing these extra eggs must be above the curves in Figure 5-1. There are two points to note from Figure 5-1. First, older birds should be more likely to produce extra eggs (i.e. a sequence of clutches larger than 3, 4, and 3 eggs) than yearling females. For example, even if 3 a year old female was certain to die at the end of the breeding season, increasing her clutches by 1 egg each would maximize lifetime reproductive success .if nesting started at the very beginning of the breeding season, whereas mortality of yearling females would have to increase by only 40% before their lifetime reproductive success is lowered (Figure 5-1) . Second, the economically viable cost of producing an extra egg is lower for birds that start nesting later in a year, indicating that clutch size should decline through the breeding season if the actual cost of producing extra eggs exceeds the benefit only towards the end of the breeding season. This second effect is due to the lower survival of young born later in the year in the Mandarte population of Song Sparrows (Chapter 2) . An important assumption was made in calculating the theoretical minimum cost of reproduction, above, i.e. future reproductive success (Rf) is not predictable, and therefore I used the average future reproduction is the expected value. The calculations 107 Figure 5-1. The theoretical rninimurn cost (as a proportional increase in female mortality) of adding one extra egg to each nest for Song Sparrows, plotted as a function of the date that nesting is started in the current year. Calculations assume that birds are capable of varying their reproductive effort between years. Separate lines are plotted for females of different ages. Calculations assume that a maximum of 3 clutches are laid each year (with clutch sizes 3, 4, and 3 for successive clutches), with half of the birds laying 2 clutches and the other half 3 clutches. For details of calculations see Appendix 1. 80T 109 above assume that a bird has the option of increasing its reproductive effort at the beginning of every year. However, if a bird expends a high effort on reproduction in one year, that bird may be constrained to increase reproductive effort throughout its entire life. Reproductive effort may be a genetically "hard-wired", and invariant through a bird's life. In the following section, I examine the consequences of reproductive effort being invariant with female age. Reproductive Effort Invariant with Female Age Now, I examine the case in which an increase in reproductive effort in one year results in an increase in reproductive effort in all years. Estimating allowable costs of reproduction in this case requires a modification of eqn. (1), because the increase in clutch size affects not only present but also future reproduction. To simplify calcula-tions, I assume that birds live for a maximum of 3 years, rather than leaving survival open-ended. A maximum age of 3 years is a natural choice, because reproductive success of female sparrows declines after this age, indicating the onset of senescence (Nol and Smith 1987). Also, calculations were not altered much by assuming longer survival: too few birds would live past 3 years of age to make a large difference in the calculations. I also assume that future reproductive success is not predictable from current reproductive success. It is reasonable to assume that future reproductive success is not predictable. Even though the time of initiation of nesting for a female is somewhat repeatable between years, reproductive success is very poorly correlated between years (Table 5-1). Assuming unpredictable future reproductive success, eqn. (3) can be re-written as Rx + ARP + (s - cf)Rf, >R!+ sRf where Rf = i? 2 + 8Rf» Rf, = R2 + ARf + (s - cf)Rf», 110 Rfi - R3 Rfin = R3 + ARf In this set of equations the subscript numbers represent different ages. Again, I assume that birds live for a maximum a three years, and that the benefit for increasing clutch size is the same in all years (i.e. ARf = ARP). The value of ARf that I use is the average value across starting times in the breeding season, because I am assuming that birds future reproductive success will not be correlated with reproductive success in the current year. Reproductive success at the different ages can be measured directly for the Song Sparrows. The above equation simplifies to ARP + ARf(s +s2)- 2scf(R3 + ARf) - cf(R2 + ARf) + c){Rz + ARf) > 0 (3) The theoretical minimum cost is found by solving the equation so that the left side of the equation equals zero. For the Song Sparrows, s = 0.53, R2 = 0.660, -R3 — 0.423 (all geometric mean values, averaged over years), and ARf (= 0.303) and ARp (which varies with date) were estimated from the model in Appendix 1. When these values are used in eqn. (4), the resulting theoretical minimum costs are those given in Figure 5-2. There is only one line describing the economically allowable cost in this case, because the decision is made only once, at the beginning of a bird's life. Note that when future reproductive success cannot be predicted from current reproductive success (and thus the mean future success is used) there is a seasonal decline in the economic advantage of producing extra eggs. More importantly, note that the estimated minimum costs at all ages are fixed very close to the viable cost for yearlings in the case in which reproductive effort varied throughout life (Figure 5-1). The economically viable cost of reproduction is strongly constrained by the cost to yearlings, if reproductive effort is fixed over a bird's lifetime. I l l Table 5-1. Correlations of reproductive parameters for birds at different ages. Correlations of date of initiation of breeding are presented below the diagonal, and correlations of numbers of recruiting offspring are above the diagonal. The table presents correlation coefficients and sample sizes (in brackets). Age 1 Age 2 Age 3 Age 1 — -0.002(105) 0.021 (59) Age 2 0.116(113) — 0.171 (59) Age 3 0.229 (59) 0.469 (59)** — P<0.001 112 Figure 5-2. The theoretical minimum cost (proportional increase in adult mortal-ity) of adding one extra egg to each nest, as a function of date of initiation of the first nest of the year. Calculations assume that reproductive effort cannot vary with age, and that future reproductive success is not predictable (solid line). The dotted line is the theoretical minimum cost for yearlings if reproductive effort can vary (line the same as the solid fine in Figure 5-1, plotted here as a reference). Other details as in Figure 5-1. Maximum Viable Cost 0.20 0.30 0.40 0.50 8X1 114 Summary In summary, the economically viable cost of reproduction is higher for older birds, because of their lower residual reproductive value. Birds nesting later in the season have a lower economical cost for laying extra eggs than birds that begin nesting earlier in the year (Figure 5-1), because of the lower value of offspring born later in a breeding season (Chapter 2). However, these patterns can be manifested only if birds can alter their reproductive effort with age or date of nesting, respectively. The economically viable cost of reproduction is constrained to relatively low levels, if reproductive effort cannot vary with age (Figure 5-2). M U L T I P L E COSTS OF R E P R O D U C T I O N To this point, I have treated reproductive cost as though it were a single item. However, brood manipulation experiments have identified two costs to reproduction: reduced adult survival (e.g. Nur 1988), and reduced future reproduction (e.g. R0skaft 1985). These two effects are generally considered to be the result of energetic or nutritional stress on the parents (Lima 1987). However, Lima (1987) suggested that the cost to adults is not nutritional stress, but an increased susceptibility to predation. Lima (1987) also noted that increased brood size could lead to increased mortality of offspring, by increasing detectability of nests (e.g. Perrins 1979, p. 269). The need to forage more intensively for larger broods should increase the probability of predation on adults. In addition, if birds increase clutch size they may have difficulty feeding the larger brood, and produce smaller young that survive poorly (e.g. Nur 1984b, Smith et al. 1989, Chapter 3). Any or all of the above effects could result from an increase in clutch size. It is necessary to understand how the total cost of reproduction can be parti-tioned between these alternatives. Reduced adult survival (through nutritional stress or increased predation), and reduced adult reproductive success are both encompassed in the variable Cf in equations (1) and (2); proportional reductions in survival and fu-115 ture reproduction are equivalent and add up to c/. Reduction in survival of offspring from the present nesting attempt has not been considered to this point in the paper. Changes in the survival of current offspring must be accounted for, even though a reduced survival of current offspring is not a cost of reproduction. Because life-history theory is concerned with trade-offs between present and future reproduction, we would like to know how large an increase in mortality of present offspring is equivalent to a given decrease in future reproductive success. The reduction in survival of offspring from the current nest that is equivalent to a given reduction in future fecundity can be calculated as follows. Let cp be the reduction in survival of offspring from the current nest associated with an increase in clutch size, and c/ be the cost to future fecundity. When the total cost of reproduction is either entirely a present cost or entirely a future cost, and the net resulting lifetime reproductive success is identical for both costs, then the equivalence of the two costs can be calculated. Taking the case in which clutch size is increased in only one year (e.g. eqn. (1)), equal LRS with present cost or future cost can be represented algebraically as: Rp + ARp - cp(Rp + ARp) + sRf = Rp + ARP + (s - cf)Rf The left side of the equality calculates lifetime reproductive success in the case where the sole cost is a decreased survival of current offspring due to extra eggs. The right side (taken from eqn. (1)) describes the case in which the cost is to future reproductive success. Re-arranging the equation in order to calculate the reduction in survival of current offspring gives: cfRf c p Rp + ARp (4) As Cf = ARp/Rf (eqn. (2)), the formula can be re-arranged: 116 This formula indicates that the theoretical minimum increase in mortality of current offspring resulting from extra eggs being laid is independent of parental age. The maximum economically viable increase in mortality rate of current offspring is roughly 0.23, given the costs to future reproduction calculated above and presented in Figure 5-1. Remember that this value assumes that adult survival and future reproductive success are not affected. Song Sparrows could experience a 23% drop in the survival rate of offspring in the current nesting attempt, and still increase clutch size in the current year by one egg without decreasing lifetime fecundity. The economically viable cp does not vary with date of laying because ARP is in constant proportion to Rp in the model used to calculate reproductive success (Appendix 1). Using the argument presented above, the theoretical minimum increase in mortal-ity of present offspring can be calculated for the case in which production of extra eggs in one year results in production of extra eggs in all years of reproduction (eqn. (3)). The equation for calculating the maximum viable drop in survival of current offspring is _ cf(2s(Rz + ARf) + (R2 + ARf) - cf(Rz + ARf))  Cp~ R!+ ARP + s(R2 +ARf) + s> (Rz + ARf) Given values for the variables outlined above (in the section COST OF P R O D U C I N G E X T R A Y O U N G ) we find that the viable decrease in the survival of current offspring is actually larger for birds that start nesting later in the breeding season (Figure 5-3); on average, a decline of up to 27% in the survival of offspring from the present brood is economically viable. In summary, it is possible to determine the equivalence of several costs of repro-duction, and to determine what drop in survival of current offspring is equivalent to any given reproductive cost. The cause of increased mortality is unimportant to calcu-lations. Thus, a proportional increase in predation is equivalent to the same increase in mortality caused by nutritional stress. 117 Figure 5-3. The theoretical minimum reduction in survival of current offspring (proportional increase in mortality) for an increase in clutch size of 1 egg for each nest, plotted as a function of date of initiation of first nest. Calculations assumed that the reproductive effort does not vary with age, and that future reproductive success is unpredictable. Viable Increase in Offspring Loss 0.250 0.260 0.270 0.280 8X1 119 E S T I M A T E D C O S T S F O R A D D I T I O N A L S P E C I E S It is possible to compare expected costs of reproduction (as calculated above) with observed values from brood-size manipulation experiments, given sufficient knowledge of a population of birds. As an example, I assume that reproductive effort varies with female age. Given this assumption, I found sufficient data in the literature to allow me to estimate theoretical minimum costs of reproduction for three species. The three species are: House Wren, Great Tit , and Collared Flycatcher. Data for House Wrens were found in Finke et al. (1987), and Drilling and Thompson (1988; return rates of adults, used as estimates of survival rates). The necessary data for Great Tits came from Boyce and Perrins (1987; data for calculating A.Rp), Pettifor et al. (1988), McCleery and Perrins (1988; residual reproductive value), and McCleery and Perrins (1989; probabilities of adult female survival). Data on Collared Flycatchers were obtained from Gustafsson and Sutherland (1988), and Gustafsson (1989; data on age-specific reproduction, used to calculate residual reproductive value). The values used in my calculations, and the results of calculations are presented in Table 5-II. The calculated theoretical minimum costs of reproduction differ greatly between the four species, with Great Tits having the lowest and Song Sparrows the highest economically viable costs for reproduction. The observed declines in survival of current offspring (cp) with increased clutch size for Great Tits and Collared Flycatchers were well in excess of those allowable for increased clutch sizes to be economical. However, in Collared Flycatchers, where a decrease in future reproductive success was also detected, the observed cost to future reproduction (c/) alone was insufficient to make additional eggs uneconomical. Thus, the cost of reproduction does not keep Collared Flycatchers from laying larger clutches, even where there is a demonstrated cost of reproduction. My calculations do not help to explain why no evidence of reproductive trade-off was found for House Wrens (Finke et al. 1987). The calculated economically viable costs of reproduction are similar to those for the flycatchers, so 120 Table 5-II. Estimates of reproductive parameters and theoretical minimum costs of reproduction for four species of birds, given an increase in clutch size of 1 egg. Sources of information for House Wren, Collared Flycatcher, and Great Tit are noted in the text. Data for Song Sparrows are yearly averages and do not take into account seasonal variation in reproductive success. Viable costs are calculated under the assumption that reproductive effort can be varied with female age (eqn. 2 and 4 in text). Values for reproductive success for House Wrens axe given as numbers of offspring fledging, while reproductive success of other species are measured in number of offspring surviving to enter the breeding population. Rp ARP Rf maximum acceptable actual Cf maximum acceptable cp actual Cp Age House Wren Collared Flycatcher Great Tit Song Sparrow 1 7.7 0.46 0.72 1.03 2 7.7 0.56 0.74 1.03 3 7.7 0.48 0.80 1.03 1.59 0.11 0.006 0.30 1 10.48 0.91 0.69 0.99 2 11.48 0.82 0.72 0.66 3 12.15 0.79 0.71 0.41 1 0.206 0.43 0.34 0.53 2 0.243 0.43 0.39 0.53 3 0.311 0.43 0.41 0.53 1 0.15 0.12 0.009 0.30 2 0.14 0.13 0.008 0.45 3 0.13 0.14 0.008 0.73 — 0.04 0 — 1 0.17 0.19 0.008 0.23 2 0.17 0.16 0.008 0.23 3 0.17 0.19 0.007 0.23 0 0.72 1.32 121 presumably some trade-off should have been observed. I can presume only that the inability of Finke et al. (1987) to follow survival of fledglings and adults explains their results. C O N C L U S I O N S Variation in Adult Quality This chapter has considered reproductive costs of an idealized, average female Song Sparrow; variation around this average may change the economic viability of a given increase in reproductive effort. A bird may be able to recognize that it is on a poor territory by assessing its own nutritional state. Whether birds are able to determine that their reproductive success will be above or below the population average affects the level of reproductive effort that is economical. The one example of inter-female variability explored in this paper is variation in reproductive success with time of initiation of breeding. Eggs of females starting their first nests later in a season have a lower probability of surviving to produce breeding offspring in this population of Song Sparrows (Chapter 2) and in other species (e.g. Perrins 1966, Newton and Marquiss 1984, Nilsson and Smith 1988). If females can assess whether they are nesting early or late in a season and adjust their clutch sizes accordingly, then the economically viable costs of reproduction are those given in Figures 5-1 - 5-3. If reproductive success in the current year cannot be assessed by birds, then there will be no seasonal variation in the economically viable cost of reproduction. Instead, the economic viability of a given cost is determined by some average reproductive success (results given for Song Sparrows in Table 5-II). A more detailed analysis of the effects of inter-female variation is beyond the scope of this chapter. Detecting Costs of Reproduction Given the relatively large costs of reproduction estimated in this paper, a cost of reproduction should be easily detectable in experiments that manipulated clutch size. Two factors, however, make the detection of costs of reproduction difficult: (1) the 122 cost of producing the extra egg is ignored by this manipulation (e.g. Nur 1984b), and (2) the values that I calculated are for only a single cost of reproduction. However, adult survival, future reproduction, and the survival of the current brood may all vary if clutch size is changed. Reductions in adult survival and future reproduction, and re-duction in survival of current offspring must all be considered in determining whether an increase in clutch size is economical. However, dividing the total cost of reproduc-tion into several components makes the statistical detection of a cost of reproduction difficult. The methods outlined here can be used to estimate combinations of expected costs of reproduction in order to estimate the sample sizes required to detect costs of reproduction in field studies. The effects of changing clutch size on adult survival, adult future fecundity, and survival of offspring together determine the economical acceptability of a given re-productive effort. We must explicitly recognize that the viable cost of reproduction has multiple determinants and use statistical tests that account for multiple costs of reproduction (see Hilborn and Stearns 1982). Thus, any study that looks for a cost of reproduction must measure all three of these variables simultaneously. Future work on reproductive costs should determine the relative magnitudes of the various effects of manipulating clutch size. Considerations for Future Research The results of this paper suggest that flexibility of reproductive effort is an impor-tant constraint on the economic viability of a given cost of reproduction. This paper suggests two promising areas of further research into reproductive strategies. The first is whether birds can "predict" the success of their current season's reproduction. The seasonal declines in viable costs of reproduction (Figures 5-1 and 5-2) depend on a bird "knowing" that early nesting will result in higher reproductive success than will a late start to nesting. Otherwise, birds will determine their reproductive effort based on the average expected reproductive success (Table 5-II). Attempts should be made to see if 123 reproductive effort varies with the timing of initiation of breeding. The second promis-ing area is into whether reproductive effort can vary with parental age. My analyses suggest that there is a clear advantage for birds to increase their reproductive effort with age (Figure 5-1). However, if reproductive effort is fixed over a lifetime, the re-productive effort of older birds would be constrained well below its potential optimum (contrast Figure 5-1 and 5-2). Future research should look for evidence of age-related differences in reproductive effort (e.g. Pugesek and Diem 1990), and differences in the response of different aged adults to manipulations of clutch size. 124 C H A P T E R 6 C O N C L U D I N G D I S C U S S I O N Summary - This chapter has three purposes: first, to summarize the thesis' re-sults relative to the two themes noted in Chapter 1; second, to show the relevance of my work to an understanding of the evolution of life-history of Song Sparrows; and third, to note some areas for further research suggested by my results. With respect to the first theme, inter-annual variation, I con-clude that: (1) although patterns of reproductive success are quantitatively different among years, the qualitative patterns are relatively similar, and (2) although density-dependence explains much of the inter-annual variation in reproductive success, the survival of offspring while in the care of their par-ents appears not to be affected by population density. It appears that parents buffer their nestlings against the environment around them. Addressing the second theme, repeatability of success within parents, there is little evidence that certain parents are always better than others; clutch sizes were repeat-able, but no other measured parameter was. My analyses suggest that the life-history of the Song Sparrows evolved to maximize the lifetime reproduc-tive success of an "average" Song Sparrow; only variation in reproductive effort with date of nesting has potential to alter this basic strategy. I find two general areas for further research particularly interesting: (1) the impor-tance of variation among nestlings within a single brood to determining the most productive size of brood, and (2) the implications of multiple avenues by which the cost of reproduction can be expressed. 125 Each of the four preceding chapters is a discrete unit of research, and provides an adequate summary of its findings within the context of that paper. This conclusion places the pieces in more unifying contexts. First, I will return to the two themes noted in Chapter 1: inter-annual variation in patterns of reproductive success (e.g. intra-seasonal changes in clutch size), and repeatability of reproductive success by individual birds. I summarize my findings in the context of these two themes. Next, I interpret my findings in the context of the evolution of life-history. In addition, I note some implications of my work and suggest further research topics. R E C A P I T U L A T I O N OF T H E M E S Inter-Annual Variation As noted in Chapter 1, one reason to analyze inter-annual variation was to see whether the same qualitative pattern of variation within each year occurs in all years. For example, if females that nest earlier generally have higher reproductive success than later nesting females, does that mean that early nesting results in higher reproductive success in all years, most years, or only a bare majority of years? Previous studies have shown mixed results. Van Noordwijk (1980) found early laying was favored in only 16 of 28 years in populations of Great Tits in the Netherlands. Boyce and Perrins (1987) found that selection favored larger clutch sizes in 22 of 23 years in one population of Great Tits, whereas van Noordwijk (1980) found selection favored larger clutch sizes in only 16 of 28 years in other populations. Schluter and Smith (1986a) and Gibbs and Grant (1987) have shown that morphology can come under opposing selection pressure at different times. Nur (1988) found evidence that increased brood size produced decreased adult survival in only 2 of 3 years in his study. In my study, selection on laying date, parental care, and mass and bill length were relatively consistent. Results in Chapter 2 indicate that females that start nest-ing earlier in a year have (10 of 13 years) higher reproductive success, because their 126 offspring survive better than offspring born later in the year. Thus, there is selection for earlier nesting. Chapter 3 shows that nestlings in higher nutritional condition were more likely to reach independence than nestlings in poorer condition; this result was observed in 7 of 8 years. Assuming that nestling condition reflects parental effort in rearing nestlings, then Chapter 3 indicates selection on the effort parents expend in rearing offspring. Offspring of larger mass with long bills had higher survival in most years (Chapter 4). The consistency of selection is important for two reasons. First, consistent se-lection favours a single phenotype in most years, resulting in a single, successful life-history strategy. Second, and more practically, these results indicate that a relatively short-term study would reveal much of the basic information about Song Sparrow life-history. Thus, Nur's (1988) concern that short-term studies may miss important information about a population is unfounded in this case. My second interest in inter-annual variation was with density-dependent effects on reproduction. Density-dependence explains much of the observed variation in re-productive success of Song Sparrows (Arcese et al. in prep). The size of the Song Sparrow population on Mandarte Island has fluctuated markedly over the course of the study, from a high of 72 breeding females in 1985 to lows of 4 (in 1989) and 9 (in 1980) females. These fluctuations have provided an opportunity to explore the effects of density on reproductive success (Arcese and Smith 1988, Arcese et al. in prep). We know that, as population density increases, both the number of offspring produced and the over-winter survival of offspring decrease. Because Song Sparrows are short-lived, and population densities in consecutive years tend to be correlated, even lifetime reproductive success decreases with increasing density in the year that a bird is born (Hochachka et al. 1989). Variation in body size with population density (Chapter 4). is now added to the list of effects of density. Although body size does not consistently influence reproductive success of birds (Schluter and Smith 1986a, Chapter 4), there is 127 the intriguing potential that density-mediated variation in the size of adult sparrows in some years affects their survival or reproductive success. However, not all facets of reproduction are density-dependent. The average nu-tritional condition of nestlings varied among years (Chapter 3), but the average nu-tritional condition of nestlings was not correlated with population density. We know that the variation was systematic because the years in which nestlings were in the highest nutritional condition were the same years in which nestlings of a given condi-tion had the highest probability of survival (Chapter 3). Years in which parents could produce nestlings in good nutritional condition were also conducive to better survival of offspring, independent of their condition. Thus, although clutch size is affected by population density, parents' ability to rear nestlings to independence (judged by nestling condition and survival of offspring of a given condition) is not. Repeatability Within Females Curiously, although there are clearly differences in intrinsic quality of adult spar-rows or their territories (Nol and Smith 1987, Hochachka et al. 1989), adult sparrows were essentially unable to produce two broods of high quality offspring in succession, even when the two broods are raised in the same year on the same territory. Other studies have found that territory quality affects the probability of nest failure (M0ller 1982), clutch size (Hogstedt 1980, 1981, and even the number of fledglings produced in a lifetime (Hotker 1988). The only repeatable reproductive trait of female Song Spar-rows was clutch size (Chapter 2); the earliest nesting females in any year produced larger first and second clutches than females that started nesting later. However, once a clutch was laid, nothing about reproduction was consistent. Probability of eggs hatching was always lower for second broods (Chapter 2). Nestling condition (Chap-ter 3) and hence survival of nestlings to independence (Chapter 2), and environmental variation in nestling size (Chapter 4) were not correlated between broods of the same female in the same year. Survival of offspring after independence depended on when 128 the young were born (Chapter 2), and was not influenced by the care that offspring received from their parents (Chapter 3). There is little evidence that consistent differ-ences among parents or their territories produce the differences in reproductive success observed among Song Sparrows. E V O L U T I O N OF L I F E - H I S T O R Y This thesis looks at various evolutionary and environmental factors that might influence the reproductive rate in a population of Song Sparrows. For most factors measured, the direction df selection appears to be relatively consistent. It is almost always advantageous to begin nesting as early in the year as possible (Chapter 2). Nestling survival to independence almost invariably increases if nestlings are provided with more food (Chapter 3). Heavier nestlings are heavier when fully grown and have higher probability of surviving their first winter (Chapter 4). Selection may alter not only genetic variation in laying date and parental care directly, but also favour birds choosing high quality territories. We know that under stressful circumstances (i.e. high population density) sparrows lay earlier and produce larger nestlings when supplemental food is provided (Arcese and Smith 1988); thus, ability to assess and acquire territories with abundant food would be advantageous. In fact, ability to assess and acquire high quality territories may be more useful, as several studies suggest that differences among territories are more important contributors to reproductive success than inherent differences among individuals (e.g. Hogstedt 1981, Reznick 1985, Price et al. 1988). However, results in this thesis suggest very few consistent benefits accrued to birds, either through inherent advantage or the holding of better territories. Only early laying and its correlates holds demonstrated advantages: larger clutches and better survival of independent offspring. Aside from correlates of date of nesting, reproductive success of Song Sparrows appears to be unpredictable for a given parent. The previous paragraph notes the potential for genetically and environmentally based variation in reproductive strategy. However, the unpredictability noted above 129 suggests that alternative reproductive strategies do not exist. Instead, the population average reproductive success appears to be the best basis for evolution of the repro-ductive strategy of all birds in the population. The only exception is the potential for reproductive strategy to vary with date of nest initiation. These assumptions were carried into the analyses in Chapter .5. Estimates of the cost of reproduction were based on the assumption that birds maximize the expected lifetime reproductive suc-cess of the average individual in the population, and that this average is unmodified by anything except the laying date of their first nest of the current breeding season. I M P L I C A T I O N S F O R F U R T H E R R E S E A R C H During my work, I identified many areas for further research. Two particular topics stem directly from my results. These are briefly discussed below. Variation Within Broods I have limited my discussion of variation among nest-mates because the small sizes of broods of Song Sparrows (1-4 nestlings) make analyses and interpretation of intra-brood variation difficult. Nevertheless, I feel that this variation can have important effects on the economics of reproduction. The form of variation within broods that I find most intriguing is variation in nestling condition. My reasoning is as follows. First, it is possible that not only the mean, but also variation in nestling quality changes with brood size. If birds increase their clutch size, they may produce offspring of more variable quality than parents of smaller broods. This was observed by Smith et al. 1989 for Great Tits, when brood sizes were manipulated. Second, it is likely that the relationship between nestling condition and nestling survival is concave-down (Smith and Fret well 1974, Chapter 3). Given these two conditions, it is possible for increasing variance in nestling condition to decrease parents' reproductive success, even if the mean nestling condition remains unchanged. The process is illustrated in Figure 6-1. Both panels of the figure show cases in which three nestlings are present, with the same mean nestling condition in both panels. The mean condition is equal to the condition 130 gure 6-1. Effect of variance in nestling condition on parents' reproductive suc-cess. Both panels illustrate hests with 3 nestlings, with mean nestling condi-tion equal in both nests. The concave-down curves represent the relationship between nestling condition and nestling survival. The upper panel shows a case in which variance in nestling condition is low, and the lower panel shows the case where variance is high. Heavy arrows at the right of each panel in-dicate mean survival rate of nestlings. Where variance in nestling condition is higher, mean survival is lower. 131 132 of the intermediate nestling in both cases. A given decrease in condition produces a greater change in nestling survival than an increase in condition of the same magnitude, because the relationship between condition and survival is concave-down. Thus, given the same mean nestling condition, a large variance in nestling condition will result in a lower mean survival of nestlings (large arrow, lower panel), than if there was a small variance in nestling condition (upper panel). Because of the above result, I think that any future studies of parental care should pay close attention to variation in the quality of nestlings within a brood. Other things being equal, adults should minimize the variance in quality of the nestlings that they produce. Adult birds are capable of equalizing nesting condition, as has been demonstrated in captive Budgerigars by Stamps et al. (1985). We need to quantify how important variance minimization is to the reproductive success of adults. Multiple Costs of Reproduction One conclusion from Chapter 5 deserves re-iteration. I believe that most empirical studies of reproductive trade-offs and reproductive cost have been unproductive. Their main goal has been to demonstrate the existence of a cost of reproduction. Whether the cost should be large or small, or what form of trade-off to expect have been largely forgotten. Chapter 5 shows that we can estimate at least the minimum cost of reproduction, given certain assumptions and sufficient knowledge of the about animals and system in question. More interesting from Chapter 5, however, is the notion that various costs are equivalent and can be traded off with each other. The result of increasing brood size could be either lower adult survival, or lower fecundity in the future. Alternatively, adults could maintain themselves, and place the burden of higher brood size on their offspring, by giving less care to each nestling in the enlarged brood. I think that applying ideas of risk-minimization will help us understand which alternative parents take, and why. If adult survival is high and relatively constant, while offspring survival is uncertain, adults should be more likely to sacrifice their 133 offspring then themselves. If parents can assure their offsprings' survival, than a high "cost of reproduction" should be incurred. These arguments have been presented before, in early theoretical papers on life-history theory (reviewed in Stearns 1976). Thus, a re-examination of the theoretical foundations of life-history theory would be useful, now that we have relevant data. It may now be possible to make specific predictions about the outcomes of brood manipulation studies, given sufficient background knowledge of study animals. L I T E R A T U R E C I T E D 134 Alatalo, R .V . , L . Gustafsson, and A. Lundberg. 1984. High frequency of cuckoldry in Pied and Collared Flycatchers. Oikos 42: 41-47. Alatalo, R . V . , L. Gustafsson, and A . Lundberg. 1990. Phenotypic selection on heritable size traits: environmental variance and genetic response. American Naturalist 135: 464-471. Alatalo, R . V . , and A. Lundberg. 1986. Heritability and selection on tarsus length in the Pied Flycatcher (Ficedula hypoleuca). Evolution 40: 574-583. Amundsen, T., and J .N. Stokland. 1990. Egg size and parental quality influence nestling growth in the Shag. Auk 107: 410-413. Arcese, P. 1989. Territory acquisition and loss in male Song Sparrows. Animal Behaviour 37: 45-55. Arcese, P., and J . N . M . Smith. 1985. Phenotypic correlates and ecological con-sequences of dominance in Song Sparrows. Journal of Animal Ecology 54: 817-830. Arcese, P., and J . N . M . Smith. 1988. Effects of population density and supple-mental food on reproduction in Song Sparrows. Journal of Animal Ecology 57: 119-136. Arcese, P., J . N . M . Smith, W . M . Hochachka, C M . Rogers and D. Ludwig. in prep. Stability and regulation of an insular Song Sparrow (Melospiza melodia) population. Askenmo, C. and U . Unger. 1986. How to be double-brooded: Trends and timing of breeding performance in the Rock Pipit. Ornis Scandinavica 17: 237-244. 135 Boag, P.T. 1983. The heritability of external morphology in Darwin's Ground Finches (Geospiza) on Isla Daphne Major, Galapagos. Evolution 37: 877-894. Boag, P.T. 1987. Effects of nestling diet on growth and adult size of Zebra Finches (Poephila guttata). Auk 104: 155-166 Boersma, D., and J.P. Ryder. 1984. Reproductive success and body condition of earlier and later nesting Ring-billed Gulls. Journal of Field Ornithology 54: 374-380. Boyce, M.S. , and C . M . Perrins. 1987. Optimizing Great Tit clutch size in a fluctuating environment. Ecology 68: 142-153. Bumpus, H.C. 1899. The elimination of the unfit as illustrated by the introduced sparrow, Passer domesticus. Biological Lectures, Woods Hole Marine Biology Station 6: 209-226. Chapman, F . M . 1940. The post-glacial history of Zonotrichia capensis. Bulletin of the American Museum of Natural History 77: 381-438. Clutton-Brock, T .H. , editor. 1988. Reproductive Success. University of Chicago Press, Chicago. Clutton-Brock, T .H. , F . E . Guinness, and S.D. Albon. 1982. Red Deer: Behavior and Ecology of Two Sexes. University of Chicago Press. Cooke, F., C.S. Findlay, and R.F . Rockwell. 1984. Recruitment and the timing of reproduction in Lesser Snow Geese (Chen caerulescens caerulescens). Auk 101: 451-458. Coulson, J . C , and J . M . Porter. 1985. Reproductive success of the Kittiwake Rissa tridactyla: The roles of clutch size, chick growth and parental quality. Ibis 127: 450-466. 136 Cronrniller, J.R., and C.F. Thompson. 1981. Sex-ratio adjustment in malnour-ished Red-winged Blackbird broods. Journal of Field Ornithology 52: 65-67. Denno, R.F. , M . J . Raupp, D.W. Tallamy, and C.F. Reichelderfer. 1980. Migration in heterogeneous environments: differences in habitat selection between the wing forms of the dimorphic planthopper, Prokelisia marginata (Homoptera: Delphacidae). Ecology 61: 859-867. de Steven, D. 1980. Clutch size, breeding success, and parental survival in the Tree Swallow (Iridoprocne bicolor). Evolution 34: 278-291. Dhondt, A . A . 1982. Heritability of Blue Tit tarsus length from normal and cross-fostered broods. Evolution 36: 418-419. Dhondt, A . A . , R, Eyckerman, and J . Huble. 1979. Wi l l Great Tits become little tits? Biological Journal of the Linnean Society 11: 289-294. Dhondt, A . A . , and J . N . M . Smith. 1980. Postnuptial molt of the Song Sparrow on Mandarte Island in relation to breeding. Canadian Journal of Zoology 58: 513-520. Dijkstra, C , A . Bult, S. Bijlsma, S. Daan, T. Meijer, and M . Zijlstra. 1990. Brood size manipulations in the Kestrel (Falco tinnunculus): Effects on offspring and parent survival. Journal of Animal Ecology 59: 269-285. Dixon, W.J . , M . B . Brown, L. Engelman, J .W. Frane, M . A . Hi l l , R.I. Jennrich, and J.D. Toporek. 1983. B M D P Statistical Software. University of California Press, Berkeley. Dow, H . , and S. Fredga. 1984. Factors affecting reproductive output of the Goldeneye duck Bucephala clangula. Journal of Animal Ecology 53: 679-692. 137 Drent, R., G.F. van Tets, F. Tompa, and K . Vermeer. 1964. The breeding birds of Mandarte Island, British Columbia. Canadian Field-Naturalist 78: 208-263. Drilling, N . E . and C F . Thompson. 1988. Natal and breeding dispersal in House Wrens (Troglodytes aedon). Auk 105: 480-491. Endler, J .A. 1986. Natural Selection in the Wild . Princeton University Press. Princeton. Falconer, D.S. 1981. Introduction to Quantitative Genetics: Second Edition. Longman Group. London. Findlay, C.S., and F. Cooke. 1982. Breeding synchrony in the Lesser Snow Goose (Anser caerulescens caerulescens). I. Genetic and environmental components of hatch date variability and their effects on hatch synchrony. Evolution 36: 342-351. Finke, M . A . , D.J . Milinkovich, and C F . Thompson. 1987. Evolution of clutch size: An experimental test in the House Wren (Troglodytes aedon). Journal of Animal Ecology 56: 99-114. Garnett, M . C . 1981. Body size, its heritability and influence on juvenile survival among Great Tits, Parus major. Ibis 123: 31-41. Gibbs, H.L. , and P.R. Grant. 1987. Oscillating selection on Darwin's Finches. Nature 327: 511-513. Gilbert, J.J. , and J .K. Waage. 1967. Asplancha, Asplancha - substance, and posterolateral spine length variation of the rotifer Brachionus calyciflorus in a natural environment. Ecology 48: 1027-1021. Gochfeld, M . , and J . Burger. 1984. Age differences in foraging behavior of the American Robin (Turdus migratorius). Behaviour 88: 227-239. 138 Grafen, A . 1988. On the uses of data on lifetime reproductive success. Ch. 28 in Reproductive Success, T. H . Clutton-Brock (ed.). University of Chicago Press, Chicago. Grant, B.R. , and P.R. Grant. 1989. Natural selection in a population of Darwin's Finches. American Naturalist 133: 377-393. Grant, P.R. 1986. Ecology and Evolution of Darwin's Finches. Princeton Univer-sity Press. Princeton, N . J . Gustafsson, L . 1986. Lifetime reproductive success and heritability: empirical support for Fisher's fundamental theorem. American Naturalist 128: 761-764. Gustafsson, L . 1989. Collared Flycatcher, in Lifetime Reproduction in Birds (I. Newton, ed.). Academic Press, London. Gustafsson, L . and W . J . Sutherland. 1988. The costs of reproduction in the Collared Flycatcher Ficedula albicollis. Nature 335: 813-815. Hilborn, R., and S.C. Stearns. 1982. On inference in ecology and evolutionary biology: The problem of multiple causes. Acta Biotheoretica 31: 145-164. Hochachka, W . M . , J . N . M . Smith, and P. Arcese. 1989. Song Sparrow, Pp. 135-152 in Lifetime Reproductive Success in Birds, I. Newton (ed.). Academic Press, London. Hogstedt, G. 1980. Evolution of clutch size in birds: Adaptive variation in relation to territory quality. Science 210: 1148-1150. Hogstedt, G. 1981. Should there be a positive or negative correlation between survival of adults in a bird population and their clutch size? American Nat-uralist 118: 568-571. 139 Horn, H.S., and D.I. Rubenstein. 1984. Behavioural adaptations and life history, in Behavioural Ecology: An Evolutionary Approach, Second Edition (J.R. Krebs and N . B . Davies, eds.). Sinauer, London. Hotker, H . 1988. Lifetime reproductive output of male and female Meadow Pipits Anthus pratensis. Journal of Animal Ecology 57: 109-117. Howe, H.F . 1976. Egg size, hatching asynchrony, sex, and brood reduction in the Common Grackle. Ecology 57: 1195-1207. James, F .C . 1983. The environmental component of morphological differentiation in birds. Science 221: 184-186. James, F . C , and C. NeSmith. 1988. Nongenetic effects in geographic differences among nestling populations of Red-winged Blackbirds. Acta X I X Congressus Internationalis Ornithologici 1424-1433. Klomp, H. 1970. The determination of clutch size in birds: A review. Ardea 58: 1-124. Koenig, W.D. , and R .L . Mumme. 1987. Cooperatively Breeding Acorn Wood-peckers. Princeton University Press, Princeton. Korpimaki, E . 1987. Costs of reproduction and success of manipulated broods un-der varying food conditions in Tengmalm's Owl. Journal of Animal Ecology 57: 1027-1039. Lack, D. 1956. Swifts in a Tower. Methuen & Co, London. Lack, D. 1966. Population Studies of Birds. Oxford University Press, Oxford. Lima, S.L. 1987. Clutch size in birds: A predation perspective. Ecology 68: 1062-1070. 140 Linden, M . , and A .P . M.0ller. 1989. Cost of reproduction and covariation of life history traits in birds. Trends in Ecology and Evolution 4: 367-371. Loman, J. 1977. Factors affecting clutch and brood size in the crow, Corvus comix. Oikos 29: 294-301. Maynard Smith, J. 1978. Optimization theory in evolution. Annual Review of Ecology and Systematics 9: 31-56. McCleery, R . H . , and C M . Perrins. 1988. Lifetime Reproductive Success of the Great Tit , Parus major, in Reproductive Success (T.H. Clutton-Brock, ed.). University of Chicago Press, Chicago. McCleery, R .H . , and C M . Perrins. 1989. Great Tit, Pp. 35-53 in Lifetime Reproductive Success in Birds, I. Newton (ed.). Academic Press, London. M0ller, A .P . 1982. Characteristics of Magpie Pica pica territories of varying du-ration. Ornis Scandinavica 13: 94-100. Murphy, M . T . 1986. Temporal components of reproductive variability in Eastern Kingbirds (Tyrannus tyrannus). Ecology 67: 1483-1492. Newton, I., editor. 1989. Lifetime Reproduction in Birds. Academic Press. London. Newton, I. and M . Marquiss. 1984. Seasonal trends in the breeding performance of Sparrowhawks. Journal of Animal Ecology 53: 809-829. Nilsson, J . -A. and H . G . Smith. 1988. Effects of dispersal date on winter flock establishment and social dominance in Marsh Tits Parus palustris. Journal of Animal Ecology 57: 917-928. Nol, E . and J . N . M . Smith. 1987. Effects of age and breeding experience on sea-sonal reproductive success in the Song Sparrow. Journal of Animal Ecology 56: 301-313. 141 Nur, N . 1984a. The consequences of brood size for breeding Blue Tits I. Adult survival, weight change and the cost of reproduction. Journal of Animal Ecology 53: 479-496. Nur, N . 1984b. The consequences of brood size for breeding Blue Tits II. Nestling weight, offspring survival and optimal brood size. Journal of Animal Ecology 53: 497-517. Nur, N . 1986. Is clutch size variation in the Blue Tit (Parus major) adaptive? A n experimental study. Journal of Animal Ecology 55: 983-999. Nur, N . 1988. The consequences of brood size for breeding Blue Tits III. Measur-ing the cost of reproduction: Survival, fecundity, and differential dispersal. Evolution 42: 351-362. O'Connor, R . J . 1977a. Differential growth and body composition in altricial passerines. Ibis 119: 147-166. O'Connor, R . J . 1977b. Growth strategies in nestling passerines. Living Birds 16: 209-238. O'Donald, P. 1983. The Arctic Skua: A study of the ecology and evolution of a seabird. Cambridge University Press. Partridge, L. 1989. Lifetime reproductive success and life-history evolution, Ch. 25 in Lifetime Reproduction in Birds (I. Newton, ed.). Academic Press, London. Perrins, C M . 1966. Survival of young Manx Shearwaters Puffinus puffinus in relation to their presumed date of hatching. Ibis 108: 132-135. Perrins, C M . 1979. British Tits. Collins, London. Perrins, C M . , and R . H . McCleery. 1985. The effect of age and pair bond on the breeding success of Great Tits Parus major. Ibis 127: 306-315. 142 Pettifor, R .A . , C M . Perrins, and R . H . McCleery. 1988. Individual optimization of clutch size in Great Tits. Nature 336: 160-162. Price, T .D. , and P.T. Boag. 1987. Selection in natural populations of birds, in Avian Genetics (F. Cooke and P.A. Buckley, eds.). Academic Press. Price, T.D. , P.R. Grant, H.L. Gibbs, and P.T. Boag. 1984. Recurrent patterns of natural selection in a population of Darwin's Finches. Nature 309: 787-789. Price, T., M . Kirkpatrick, and S.J. Arnold. 1988. Directional selection and the evolution of breeding date in birds. Science 240: 798-799. Pugesek, B . H . and K . L . Diem. 1990. The relationship between reproduction and survival in known-aged California Gulls. Ecology 71: 811-817. Reznick, D. 1985. Costs of reproduction: A n evaluation of the empirical evidence. Oikos 44: 257-267. Ricklefs, R .E . , and S. Peters. 1981. Parental components of variance in growth rate and body size of nestling European Starlings (Sturnus vulgaris) in eastern Pennsylvania. Auk 98: 39-48. Rogers, C M . , J . N . M . Smith, W . M . Hochachka, A . L . E . V . Cassidy, M . J . Taitt, P. Arcese, and D. Schluter. in prep. Winter survival and limitation of a metapopulation of Song Sparrows. R0skaft, E . 1985. The effect of enlarged brood size on the future reproductive potential of the Rook. Journal of Animal Ecology 54: 255-260. Ross, H.A. , and L A . McLaren. 1981. Lack of differential survival among young Ipswich Sparrows. Auk 98: 495-502. Schluter, D. 1988. Estimating the form of natural selection on a quantitative trait. Evolution 42: 849-861. 143 Schluter, D, and J . N . M . Smith. 1986a. Natural selection on beak and body size in the Song Sparrow. Evolution 40: 221-231. Schluter, D., and J . N . M . Smith. 1986b. Genetic and phenotypic correlations in a natural population of Song Sparrows. Biological Journal of the Linnean Society 29: 23-36. Slagsvold, T. 1982. Criteria for estimating the condition of birds: relationship between fat content and body size dimensions in the Hooded Crow Corvus corone comix. Ornis Scandinavica 13: 141-144. Smith, C.C. , and S.D. Fret well. 1974. The optimal balance between size and number of offspring. American Naturalist 108: 499-506. Smith, H.G. , H. Kallander, and J .-A. Nilsson. 1989. The trade-off between off-spring number and quality in the Great Tit Parus major. Journal of Animal Ecology 58: 383-401. Smith, J . N . M . 1981a. Cowbird parasitism, host fitness, and age of the host female in an island Song Sparrow population. Condor 83: 152-161. Smith, J . N . M . 1981b. Does high fecundity reduce survival in Song Sparrows? Evolution 35: 1112-1148. Smith, J . N . M . 1982. Song Sparrow pair raises four broods in one year. Wilson Bulletin 94: 584-585. Smith, J . N . M . , and P. Arcese. 1988. Effects of supplemental food in growth and adult size in the Song Sparrow. Acta X I X Congressus Internationalis Ornithologici 1416-1423. Smith, J . N . M . , P. Arcese, and D. Schluter. 1986. Song Sparrows grow and shrink with age. Auk 103: 210-212. 144 Smith, J . N . M . , and A . A . Dhondt. 1980. Experimental confirmation of heritable morphological variation in a natural population of Song Sparrows. Evolution 34: 1155-1158. Smith, J . N . M . , and D. Roff. 1980. Temporal spacing of broods, brood size, and parental care in Song Sparrows (Melospiza melodia). Canadian Journal of Zoology 58: 1007-1015. Smith, J . N . M . , Y . Yom-Tov, and R. Moss. 1982. Polygyny, male parental care, and sex ratio in Song Sparrows: A n experimental study. Auk 99: 555-564. Smith, J . N . M . , and R. Zach. 1979. Heritability of some morphological characters in a Song Sparrow population. Evolution 33: 460-467. Sokal, R.R. , and F . J . Rohlf.. 1981. Biometry: Second Edition. W . H . Freeman and Company, San Francisco. Stamps, J .A. 1990. When should avian parents differentially provision sons and daughters? American Naturalist 135: 671-685. Stamps, J. , A . Clark, P. Arrowood, and B. Kus. 1985. Parent-offspring conflict in Budgerigars. Behaviour 94: 1-40. Stearns, S.C. 1976. Life-history tactics: A review of the ideas. Quarterly Review of Biology 51: 3-47. Stearns, S.C. 1989. The evolutionary significance of phenotypic plasticity. Bio-science 39: 436-445. Sullivan, K . A . 1989. Predation and starvation: Age-specific mortality in juvenile juncos (Junco phaenotus). Journal of Animal Ecology 58: 275-286. Tinbergen, J . M . , J .H. van Balen, and H . M . Eck. 1985. Density dependent survival in an isolated Great Tit population: Kluyvers data reanalysed. Ardea 73: 38-48. 145 Tompa, F.S. 1964. Factors determining numbers of Song Sparrows Melospiza melodia (Wilson) on Mandarte Island, B .C . , Canada. Acta Zoologica Fennici 109: 3-73. Van Noordwijk, A . J . 1980. On the Genetical Ecology of the Great Tit (Parus major L.) . PhD thesis, University of Utrecht, Netherlands. Van Noordwijk, A. , and G. de Jong. 1986. Acquisition and allocation of resources: Their influence on variation in life history tactics. American Naturalist 128: 137-142. Van Noordwijk, A . J . , J .H. van Balen, and W. Scharloo. 1981. Genetic variation in the timing of reproduction in the Great Tit. Oecologia (Berlin) 49: 158-166. Van Noordwijk, A . J . , J .H. van Balen, and W. Scharloo. 1988. Heritability of body size in a natural population of the Great Tit (Parus major) and its relation to age and environmental conditions during growth. Genetical Research 51: 149-162. Von Haartman, L . 1966. Clutch-size in the Pied Flycatcher. Preceedings of the X I V International Ornithological Congress: 155-164. Whittow, G.C. 1986. Energy metabolism, Ch. 10 in Avian Physiology: Fourth Edition P.D. Sturkie (ed.). Springer-Verlag, New York. Wilkinson, L. 1988. SYSTAT: The System for Statistics. SYSTAT Inc., Evanston, Illinois. Williams, G.C. 1966a. Adaptation and Natural Selection. Princeton University Press, Princeton, N . J . Williams, G.C. 1966b. Natural selection, the costs of reproduction, and a refine-ment of Lack's principle. American Naturalist 100: 687-690. 146 Woolfenden, G.E. , and J .W. Fitzpatrick 1984. The Florida Scrub Jay: Demogra-phy of a Cooperatively-Breeding Bird. Princeton University Press, Princeton. Wolf, L . , E .D. Ketterson, and V . Nolan Jr. 1988. Paternal influence on growth and survival of Dark-eyed Junco young: Do parental males benefit? Animal Behaviour 36: 1601-1618. Yoshimura, J., and W . M . Sheilds. 1987. Probabilistic optimization of phenotype distributions: A general solution for the effects of uncertainty on natural selection? Evolutionary Ecology 1: 125-138. A P P E N D I X 1 E S T I M A T I N G R E P R O D U C T I V E S U C C E S S 147 Summary - This appendix outlines the method used to calculate expected repro-ductive success (number of offspring recruiting into the breeding population) of Song Sparrows, given clutch size. The results of these calculations are used in Chapter 5, where the cost of increasing clutch size (in increased parental or offspring mortality) resulting from an increase in clutch size is estimated. 148 The model used to calculate expected annual reproductive success (Chapter 5) starts by calculating the reproductive success (in number of offspring recruiting into the breeding population) for a given nesting attempt as Rn — CnPiPr where Rn is the number of recruiting offspring from the nth nest of the year, cn is the clutch size of the nth nest, is the probability of an egg producing an offspring that is independent of parental care, and pr is the probability of an independent offspring recruiting into the breeding population. The probability of survival from egg to independence was estimated using logistic regression ( B L D P L R , Dixon et al. 1983). Survival did not vary with date of nesting, clutch size, or female age, with pi = 0.47. The probability of survival from independence to recruitment did not vary with female age or clutch size, but was affected by date: the later young were born, the poorer was their probability of survival from independence to recruitment (Chapter 2). This decline was estimated using logistic regression (Dixon et al. 1983) as e - 0 . 4 4 - 0 . 3 5 d „ PR ~ I _|_ e -0 .44-0 .35d n (1) where dn is the Julian date of clutch initiation of the nth nest. Because data from all years were combined in the logistic regression analysis, and because mean and variance of laying dates varies among years, the dates were standardized to a mean of 0 and s.d. of 1 within each year for the logistic regression. For a typical year (JulianDate - 128.9) dn 24.3 Because Song Sparrows produce more than one nest in a year, the reproductive success of all nests of a bird must be summed. The time taken between a failed nesting attempt (one from which no young fledged) and the start of the next attempt is shorter than the time between a successful nesting attempt and the start of the next nest (time 149 between nests also depends on the number of young fledged from the previous nest (Smith and Roff 1980), but these differences were ignored to simplify the model). On average there are 37.5 days between the start of a successful nesting attempt and the start of the next attempt, and 20.0 days between the start of a failed nest and the start of the next nest. Because the timing of a nesting attempt determines the probability with which independent offspring recruit into the breeding population (eqn. 1), the different potential sequences of success and failure result in differences in reproductive success even if the same number of successful nests are produced in a year. The probability of nest failure (pf) was a function of clutch size pf =1- 0.69c" The probability of failure of a one-egg clutch, 0.31 (1 — 0.69), does not equal pi, because Pi was calculated from only those nests that produced at least one fledged offspring. On average, each female produced 2.5 nests in a year. For simplicity I assume that this means that half of the females produced 2 nests, and half produced 3 nests. For birds producing 2 nests/year there are 4 possible sequences of success and failure of nesting attempts, and for birds producing 3 nests/year there are 8 possible sequences. For each of these sequences of failed and successful nests I calculated the total number of recruits expected to be produced in a year, given the date of the initiation of the first nest and the calculated intervals between nests maxn ^ CnPiPr n=l Each sequence of successful and failed nests was then weighted by the proportion of total females that would produce this sequence. For example, for 4 egg clutches the probability of success is 0.23 and the probability of failure is 0.77; therefore the proportion of all sequences being "success, fail" would be 0.23 x 0.77 x 0.5 (the 0.5 being because only half of all birds produce 2 clutches in a year). A summation of 150 all weighted values of reproductive success then yields the average expected annual reproductive success of a Song Sparrow, given the date of initiation of its first nest. • Because reproductive success varies with the time of year (due to seasonal varia-tion in pr), annual reproductive success must be calculated for each date of initiation of first nest, producing a curve similar to that in Figure 5-1. The range of dates used in my calculations are Julian date 75 to 150, mid-March to the end of May (>99% of all dates of initiation of first clutches of Song Sparrows fell in this period). 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0098774/manifest

Comment

Related Items