A long-term study of reproductive performance in tree swallows: the influence of age and senescence on output

Authors

  • Raleigh J. Robertson,

    Corresponding author
    1. Department of Biology, Queen’s University, Kingston, Ontario K7L 3N6 Canada
      Dr Wallace Rendell, #3–1859 Fulton Street, San Francisco, CA 94117 USA. wbr_berkeley@hotmail.com
    Search for more papers by this author
  • Wallace B. Rendell

    1. Department of Biology, Queen’s University, Kingston, Ontario K7L 3N6 Canada
    Search for more papers by this author
    • *

      Present address: Wallace Rendell, Hastings Natural History Reservation, Museum of Vertebrate Zoology, University of California at Berkeley, 3101 Valley Life Sciences Building, Berkeley, CA 94720–3160 USA. Tel.: (415) 221–7400. E-mail: wbr_berkeley@hotmail.com


Dr Wallace Rendell, #3–1859 Fulton Street, San Francisco, CA 94117 USA. wbr_berkeley@hotmail.com

Summary

  • 1We describe age-related reproductive performance and recapture rates of tree swallows (Tachycineta bicolor Vieillot) based on a 25-year study of a nestbox population in south-eastern Ontario, Canada (1975–99).
  • 2Performance improved from first-time breeders to intermediate-aged birds. Nest initiation advanced, and clutch size increased in both sexes. In females the number of hatchlings and fledglings increased, and the proportion of nests failing completely declined. Performance declined in females after ‘middle-age’, in the number of young fledged, and the proportion of young fledged relative to initial clutch and brood size. Also, the proportion of nests that failed completely increased in the oldest birds. Males showed similar patterns.
  • 3An index of performance incorporating clutch size, hatching and fledging efficiency, and two measures of total nest failure increased to, then declined after, 4 years of age in females and 3 years in males. The relationship between this index and age was best predicted by quadratic regression.
  • 4We found no support for three of four hypotheses to explain improvement in performance with age. Recapture rates declined after age 4Y in males, but remained unchanged in females until age 7Y +, while output decreased in both sexes (Residual Reproductive Value). Birds breeding repeatedly did not perform better during their first attempt compared to birds that bred only once (Selection). Birds with varied breeding experience did not differ in their performance within age-groups (Breeding Experience). We did find support for the Breeding Age hypothesis; in females with no breeding experience, there was a successive advance in laying and increase in clutch size from 2 to 4 years of age.
  • 5Improved performance may be due to skills acquired with age, such as those devoted to feeding and balancing energy demands, which are necessary to prepare and maintain individual condition prior to, and during, breeding. Senescence in performance after ‘middle-age’ may result from accumulated costs of previous breeding effort which have been identified in this species based on research elsewhere.

Introduction

The observation that reproductive performance improves with age in animals has been documented in many taxa (e.g. fishes Hodder 1963; amphibians Salthe 1969; reptiles Tinkle & Ballinger 1972; mammals Clutton-Brock 1984). Much research on this subject has been conducted on birds, encompassing a variety of families, species and life-histories, and researchers have analysed numerous measures of performance (reviewed Sæther 1990). Studies have typically compared two age-groups: young, first-time breeders vs. all older, more-experienced breeders combined. They have revealed that older birds arrive earlier at breeding sites, begin laying, hatching, and fledging offspring earlier, or they may replace lost clutches earlier than younger birds (e.g. Coulson & White 1958; Flint & Sedinger 1992; Forslund & Larsson 1992; Perdeck & Cavé 1992; Blums, Hepp & Mednis 1997; Møller & De Lope 1999). Also, older birds lay larger clutches, or larger eggs with higher nutrient reserves, and may hatch, fledge, and recruit more and larger offspring (e.g. Richdale 1955; Hamer & Furness 1991; Pyle et al. 1991; Komdeur 1996a). Authors have also identified age-related differences in breeding behaviours, associated with differences in breeding performance. Older individuals defend broods more vigorously, or defend larger territories of higher quality (e.g. Reid 1988; Catterall, Kikkawa & Gray 1989). Also, they may be better foragers, or better providers (reviewed in Marchetti & Price 1989; Desrochers 1992a), with consequences for offspring sex-ratio and nestling mortality (Cooch et al. 1997).

The best evidence for the influence of age on breeding performance comes from experimental studies, which held constant covariates of age and output that could confound interpretation of the importance of age. De Steven (1980) enlarged broods for some yearling and older female tree swallows (Tachycineta bicolor Vieillot) while leaving others to rear their natural broods. Nestlings from enlarged broods were lighter than nestlings from control broods within yearlings but not adults, suggesting that young birds could not adequately provide for large broods. Desrochers (1992b) experimented with yearling and older European blackbirds (Turdus merula Linnaeus), wherein some females received early season food supplements prior to laying while others did not. Laying date was unaffected by feeding in older females, but food supplements greatly advanced laying date in yearlings, suggesting that foraging success limits breeding onset in young birds. Wheelwright & Schultz (1994) removed early season clutches from yearling and older female savannah sparrows (Passerculus sandwichensis Gmelin) and tree swallows, thereby inducing nesting for both age-groups under similar ecological conditions. Yearling females of both species laid smaller replacement clutches than older females. In savannah sparrows, yearling females lost more mass during breeding and showed a greater latency to renesting than older females, while in tree swallows day-7 nestlings in replacement clutches of yearling females were lighter, possibly because they were fed at a slower rate. Finally Komdeur (1996a) showed that performance was positively associated with age in the cooperatively breeding Seychelles warbler (Acrocephalus sechellensis Oustalet), although Komdeur (1996b) revealed the importance of breeding and helping experience, not age, on their reproductive success using a translocation experiment. Same-aged birds were transferred between islands to novel environments and high-quality territories. He showed that birds with helping or breeding experience reared young to fledging faster than birds with neither type of experience.

Several hypotheses may explain why improved reproductive performance is associated with age, to which some of the experiments above have hinted. We refer to these as restraint and constraint hypotheses (Curio 1983; Sæther 1990; Desrochers 1992b); the former suggest that young birds might withhold effort during breeding early in their lives, while the latter suggest that there are intrinsic, or behavioural or ecological, barriers that suppress the output of young birds. Restraint hypotheses originate in life-history theory, and include the (i) Cost of Reproduction, and (ii) Residual Reproductive Value hypotheses. In the former, resources available for breeding and self-maintenance are assumed to be limiting, so trade-offs arise because the allocation of resources to any activity or function necessarily reduces the availability of the same resources for distribution to other activities or functions (Cody 1966). Costs of reproduction are expected to take the form of a trade-off between current and future reproduction (Williams 1966a,b); individuals may pay the cost of current breeding effort with reduced future fecundity or survival (e.g. Nur 1984, 1988). Mechanisms responsible for a decline in fecundity or survival may be a loss of condition and inability to re-establish resources necessary for successful breeding during a subsequent opportunity (e.g. Wheelwright & Schultz 1994), or compromised immunocompetence (e.g. Norris & Evans 2000; Hasselquist, Wasson & Winkler 2001). If current breeding effort does reduce future output or survival, then the difference in breeding performance between young and old birds may be due to selection for young birds to withhold effort to improve longevity. Longevity is typically the best predictor of lifetime reproductive success in birds (Newton 1989). Similarly, according to the Residual Reproductive Value Hypothesis, selection should favour individuals that invest in reproduction according to the probabilities of success and survival for a given age (Williams 1966a; Pianka & Parker 1975; Stearns 1976; Charlesworth 1980). Younger birds may withhold effort early in life if the chances of success are small, and if age-specific survival decreases with effort. Also, older birds may increase reproductive effort with age if the probability of survival until the next breeding opportunity decreases with age. Constraint hypotheses include the (i) Selection, (ii) Breeding Experience, and (iii) Breeding Age hypotheses. According to the Selection Hypothesis, improved breeding success with age may be a reflection of the ‘quality’ of breeding individuals if high-quality birds live longer and have higher performance compared to low-quality individuals (Nol & Smith 1987). Thus, higher output in older birds would not necessarily be a reflection of age or experience attained, but merely differential survival due to intrinsic, individual characteristics. The Breeding Experience Hypothesis maintains that reproductive performance may improve due to accumulated experience with tasks unique to breeding such as nest building, nest defence or nestling feeding, while the Breeding Age Hypothesis asserts that performance may improve because of skills acquired with age that are devoted to things indirectly related to reproduction, such as self-maintenance, foraging, predator avoidance, or competitive interactions (Lack 1968; Orians 1969; Perrins 1970). Distinguishing which of the above hypotheses best explains breeding improvement with age can be difficult because they are not necessarily mutually exclusive. For the restraint hypotheses, reproductive effort, its associated costs, and an individual’s future ‘value’ are all intertwined. Likewise for the constraint hypotheses where it is intuitive that breeding experience and age are highly correlated.

Despite the wealth of studies concerned with age-related breeding performance, relatively few have made attempts to test explicitly predictions of the various hypotheses, and those that have have yielded mixed support for each of the hypotheses. In support of the Residual Reproductive Hypothesis Hamer & Furness (1991) found that great skuas (Catharacta skua Brünnich) increased reproductive effort with age, based on higher aggression rates and greater foraging time in older birds. However, other studies have found either partial support for this hypothesis (e.g. bison Bison bisonLinnaeus, Green 1990), or none at all. For example Røskaft, Espmark & Järvi (1983) concluded that young male rooks (Corvus frugilegus Linnaeus) do not withhold reproductive effort compared to older males because young males fed mates more frequently, and were away from their territories more often (see also Nol & Smith 1987; Reid 1988; Desrochers 1992b; Wheelwright & Schultz 1994; Komdeur 1996b). Nol & Smith (1987) concluded that the Selection Hypothesis best explained the relationship between age and performance in female song sparrows (Melospiza melodia Wilson) because females that bred in their first year fledged fewer offspring per attempt than females that bred again in the population. Four subsequent studies on other species failed to find such an effect (Forslund & Larsson 1992; Wheelwright & Schultz 1994; Komdeur 1996b; Blums et al. 1997). Komdeur’s (1996b) translocation experiment described above provided strong evidence that breeding experience, either in the form of helping or actual breeding, significantly affected success in Seychelles warblers because age was standardized among individuals in the study. Nol & Smith (1987), however, managed to partition breeding experience from breeding age and found some support for the effect of age, not experience, on success in song sparrows. The Breeding Experience and Age hypotheses have rarely been distinguished. Four previous studies have either been unable to distinguish among them, or they have found support for both hypotheses (Pyle et al. 1991; Forslund & Larsson 1992; Wheelwright & Schultz 1994; Blums et al. 1997).

While these hypotheses may account for increases in breeding performance with age, they do not address the observation that reproductive performance declines in older individuals, probably due to senescence. Senescence is defined as the ‘decline in an individual’s age-specific fitness components due to physiological deterioration’ (Rose 1991). Hypotheses about the evolution of senescence include optimality and mutational explanations (Partridge & Barton 1993). According to the former, senescence is part of an evolved, optimal genotype whose fitness is maximized by increasing reproductive effort earlier in life, but due to the cost of reproduction increased reproductive effort early in life results in a decline in potential and ability later (Williams 1957; Bell & Koufopanou 1986; Stearns 1989). In the latter, because selection should be weaker on mutations with deleterious effects when expressed late in life compared to those expressed early in life, late-acting mutations may accumulate in a genotype over generations, and contribute to a decline in condition of individuals as they age (Medawar 1952). Partridge & Barton (1993) point out that both explanations for the evolution of senescence probably apply. Deleterious mutations have been identified that have a pleiotropic effect on fecundity and survival (Van Voorhies 1992), and antagonistic pleiotropy of fecundity and longevity has been identified by selection experiments with Drosophila (Rose & Charlesworth 1981ab). In wild vertebrates, senescence in performance with age has been identified infrequently, primarily because of the need for long-term study of marked individuals (e.g. Perrins & Moss 1974; Rockwell et al. 1993; Newton & Rothery 1997). Fewer still are empirical field studies where senescence in output has been experimentally induced. Using brood manipulations, Gustafsson & Pärt (1990) identified a trade-off between breeding effort early in life and senescence in age-related fecundity in the collared flycatcher (Ficedula albicollis Temminck); females with enlarged broods early in life laid smaller clutches in subsequent years compared to similarly aged females with control and reduced broods. Though it may be argued that long-term, observational studies have drawbacks because they fail to control for numerous variables associated with both age and reproductive output (e.g. habitat quality), they may certainly compliment experimental studies and aid in hypothesis testing.

Here, we describe age-related nest initiation and reproductive output in tree swallows, a secondary cavity-nesting, aerial insectivore. Age-related breeding performance has been described before in this species, but typically only in comparisons between two age-groups of one sex: first-time and older breeding females (e.g. De Steven 1978, 1980; Stutchbury & Robertson 1988; Robertson & Rendell 1990; Wiggins 1990; Wheelwright & Schultz 1994; Lombardo et al. 1995; Lozano & Handford 1995). Instead, we describe the breeding performance of females and males, aged 2Y to 7Y +, using cross-sectional and longitudinal data collected over a 25-year period in south-eastern Ontario, Canada. We provide evidence that reproductive performance improves to, then declines after, 4–5 years of age in both sexes, and we test predictions of the above hypotheses that may explain the observed improvement in breeding performance from first-time to middle-aged breeders. By distinguishing between the Breeding Experience and Age Hypotheses, we show that age, not experience, affects reproductive performance, particularly in first-time breeders.

STUDY SPECIES

Tree swallows (Hirundinidae; ≈ 20 g) nest in cavities in trees but are incapable of excavating their own nest-site. As a result, they are highly territorial birds, returning to our breeding sites in late March to acquire nestboxes while snow may still be on the ground. Apparent nest-site limitation leads to ‘floating’ birds which frequently visit occupied territories throughout the breeding season (Lombardo 1986; Stutchbury & Robertson 1985, 1987a). Defence of nest-sites from floaters, and interspecific competitors, can lead to injury and even death, while displacement of an individual in a pair, or nest usurpation, are not uncommon events up until the nestling period, in both natural and nestbox populations (Leffelaar & Robertson 1985; Rendell & Robertson 1989; Robertson & Rendell 1990; Rendell 1993). Tree swallows are single-brooded, although they will renest if one or more earlier nesting attempts have failed (Rooneem & Robertson 1997). They lay 3–8 eggs, with 5- and 6-egg clutches most common, and 8-egg clutches extremely rare (Dunn & Robertson 1992; Dunn et al. 2000). Extra-pair paternity is very common in this species in both natural and nestbox populations; as many as 50% of nests within a study site may have one extra-pair nestling, and in some cases no young in a brood are genetically related to the social father (e.g. Lifjeld et al. 1993; Barber, Robertson & Boag 1996; Kempenaers et al. 1999). Despite this latter observation, males feed nestlings as often as females throughout the nestling period, despite a probable reduction in confidence of paternity (Leffelaar & Robertson 1986; Whittingham, Dunn & Robertson 1993). Egg-dumping by females occurs very rarely (Lombardo 1988; Barber, Robertson & Boag 1996). Life expectancy for males and females is only 2–3 years (Butler 1988), although the oldest known birds in the wild have reached 12 years of age (Hussell 1982; W.B. Rendell & R.J. Robertson, unpublished data). For more information about the natural history of tree swallows see Robertson, Stutchbury & Cohen (1992).

Methods

HABITAT AND NESTBOX CHARACTERISTICS

This study was conducted on research tracts of the Queen’s University Biological Station (QUBS; 44°34′ N, 76°20′ W), south-eastern Ontario, Canada, during March–August 1975–99. The surrounding habitat is secondary, mixed-deciduous forest, farm fields, secondary roads, and numerous small ponds and lakes. Nestboxes were situated solitarily in fields and along roadsides, or in rows and columns (i.e. grids) in hayfields, close to water (i.e. within 500 m), except for one grid situated entirely over water. Topographically, each hayfield was similar to the next; mostly flat, with the occasional gentle rise. With few exceptions, all boxes in a grid were visible from every other one. Variations on the distribution of boxes in grids can be seen in figures in Robertson & Gibbs (1982) and Rendell & Robertson (1990). The number of available boxes in our population ranged from 77 (1975) to a high of 284 (1995), spread out over a distance of ≈ 30 km in this latter year. Box densities varied considerably in our grids until 1985 (5–17 boxes ha−1), due largely to experimentation on territoriality in tree swallows (Robertson & Gibbs 1982), but they have been relatively constant since then (≈ 7 boxes ha−1), as have been the densities of solitary boxes since 1975 (i.e. ≈ 3 boxes ha−1). The densities of boxes in our grids are similar to those of natural cavities in beaver ponds used for nesting by tree swallows (Rendell & Robertson 1989; Robertson & Rendell 1990). On average, the ratio of boxes situated in grids to those situated solitarily was 2·5 : 1 (n = 25 years), and the ratio of densities of boxes in grids to solitary sites was 7·0 : 1 (n = 25 years). Box characteristics were largely uniform. Entrances faced approximately east (Rendell & Robertson 1994). In grids, boxes were 28–40 m from each other; solitary ones were often more than 100 m from another box. They were constructed of plywood and identified by alpha-numeric codes. Their design underwent cosmetic changes in 1983, however, the internal volume (≈ 3900 cm3) and floor area (≈ 180 cm2) were similar throughout the study. Grid boxes were mounted on aluminium or steel posts, while many solitary boxes were attached to fences. After each breeding season, we removed old nests from each box.

After severe predation by raccoons (Procyon lotor Linnaeus) at our grids in 1987 (Robertson & Rendell 1990), we added aluminium predator cones to box posts. Predator cones were also added to selected solitary boxes during some experiments. However, not all boxes in the population have had this protection from terrestrial predators. Typically, predation on eggs (by mice, Peromyscus spp.) or nestlings (by raccoons, or black-rat snakes, Elaphe obsoleta Say) was uncommon (≈ 5% annually).

CAPTURE, HANDLING, SEXING AND AGEING OF INDIVIDUALS

We captured tree swallows using mist nets, inside boxes using box traps (Stutchbury & Robertson 1986), or simply by hand (e.g. incubating females or feeding adults). All birds were banded with Canadian Wildlife Service aluminium bands and, prior to release, adults were marked at positions along the wing and tail with unique combinations of non-toxic acrylic paints. These markings allowed identification of birds throughout the breeding season without the need for recapture.

We sexed tree swallows using a variety of methods including: (i) presence of a large or vascular brood patch (females only) or distended cloacal protuberance (males only); (ii) wing length (≈ 113 mm, female; ≈ 122 mm, male); (iii) nest-building behaviour (only females build the grass nest matrix); (iv) copulation behaviour; and (v) plumage characteristics (see Hussell 1983; Stutchbury & Robertson 1987b). Tree swallows are one of only two North American passerines with a sub-adult female plumage (i.e. second-year, 2Y Hussell 1983). In 2Y females, plumage colour varies widely from ≈ 100% dull sandy-brown to ≈ 90% irridescent blue-green. Also, females who are completely blue-green often have 1–5 mm of brown feathering at the base of the upper mandible (i.e. forehead). Males at 2Y immediately adopt the blue-green irridescent plumage of adults, and never have a brown forehead. Known-sex individuals were given a red plastic leg band if female, blue if male. Males were more difficult to catch than females, and banding effort directed toward males varied from year to year; thus the sample sizes for males are much smaller than those for females. To age tree swallows we used plumage characteristics or our banding records. Exact ages (e.g. 1Y = nestling, 2Y, 3Y, etc.) were known only for those birds originally banded in our population as nestlings or sub-adult females. For other birds we estimated age. Known females scored as ≈ 91% blue-green plumage when banded for the first time were declared to be after-second-year (A2Y), while known males banded for the first time were after-hatch-year (A1Y). Plumage was scored for females by estimating the proportions of body regions covered by blue-green plumage. An overall plumage score (PLUM) was calculated for each female according to the equation (Stutchbury & Robertson 1987a):

inline image

where H, E, M, and R are the body regions, head, epaulette, mantle, and rump, respectively. The numbers in the equation are the percentage areas of the dorsal side of a female that each body region represents. Any new female with ≈ 90% blue-green plumage was declared a sub-adult. This ageing method differs from Hussell (1983) and Stutchbury & Robertson (1987a); they concluded that females with 51–90% blue-green plumage could not be assigned accurately to 2Y or A2Y age groups. However, our records show that 93% (25/27) of females born in our population and recaptured subsequently with plumage scores between 51 and 90% are, in fact, 2Y birds, so we concluded that our age-assignments for females based on plumage are reasonably accurate.

Return rates of young and both sexes of adults were determined through recapture in subsequent years. We chose a 16-year period (1982–97) for analysis. This period was chosen for several reasons. First, 1982 was the first year that we captured and banded greater than 100 adult females, as we have done every year since then. Second, after seven field seasons 1982 was the first year when we could expect to have females spanning the range of adult age groups (i.e. 2Y to 7Y & A6Y) which incorporate over 95% of females breeding in our population in any given year. Finally, 1997 was the last year chosen because the 2-year period following it has allowed us to fairly estimate returns of adult birds banded and handled that year. Return rates of young per adult were calculated within age-groups by simply adding up all the recruits and dividing by the number of adults within a sex-age combination. Return rates of adults were calculated within age-groups annually as the proportion of adults returning from the total number of individuals in an age-group the previous year.

COLLECTION OF REPRODUCTIVE DATA

Annually, we visited nestboxes every 1–3 days from late April until early July, and at each box we recorded: first-egg date, clutch size, hatch date, number of hatchlings, fledging date and fledgling number. Typically, nestlings were banded on nestling-day (nd) 14, where hatch day = nd 0. After nd 14, we visited nests to determine how many young remained, but we no longer handled young to prevent premature fledging. Banded young that we found dead outside a box soon after a brood fledged were not counted as fledged. Using the clutch, hatch and fledge data above for each nest, we calculated: hatch success (i.e. number of hatchlings/clutch size), brood success (i.e. number of fledglings/number of hatchlings), and fledging success (i.e. number of fledglings/clutch size). We calculated these variables only for nests where at least one egg hatched, or one nestling fledged. Two other variables, nest failure and brood failure, were calculated within year-, sex- and age-groups as the proportion of nests which failed completely between first-egg and hatch, and during the nestling period, respectively.

DATA SCREENING

We selected an individual’s reproductive data for our analyses based on several criteria. First, we analysed only those data from the first breeding attempt by an individual in any given year. Second, we used only those individuals whose clutch size was 3–7 eggs. Clutches of 1–2 eggs were considered incomplete as no female incubated such clutches; clutches of 8 or more eggs in our population were typically the product of two females laying in the same cup (see Quinney 1983). Finally, many behavioural and ecological experiments have been performed at our population that have influenced aspects of individuals’ total reproductive output (e.g. infanticide Robertson & Stutchbury 1988; cavity size effects on clutch size: Rendell & Robertson 1993; cost of reproduction: Rendell 1999). Where experimental activities directly influenced a variable(s), we excluded data from analyses. So, from approximately 3600 nest records, our samples for aged individuals in both sexes number only in the 10 s or 100 s. We applied the same restrictions to data in tests of the Selection, Breeding Experience and Breeding Age hypotheses as those used in the analysis of overall performance with age.

STATISTICAL ANALYSIS

The need to exclude many data points prevented us from accumulating a sample of males or females with consecutive, annual nest attempts sufficient for longitudinal analysis, so this is largely a cross-sectional study. Where we have sufficient longitudinal data for analyses the results are presented and discussed. The lack of longitudinal data is not so problematic, as two previous studies on geese suggest that mean age-specific patterns of reproductive output reflect accurately patterns of change in reproductive output with changes in age within individuals (Forslund & Larsson 1992; Rockwell et al. 1993).

We grouped females into six age categories (i.e. 2Y, 3Y, 4Y, 5Y, 6Y, and 7Y +), and males into five (i.e. 2Y, A1Y, 3Y, 4Y, and 5Y +). For both sexes, 2Y includes only those birds known to be in their second-year (i.e. the first year of breeding). For males, A1Y includes adults whose age is not known; they could be 2Y or older. For both sexes, age-group 3Y to the oldest age-group include both known age birds, and birds whose minimum age is known; for example, 3Y also includes A2Y individuals, 4Y includes A3Y individuals, etc. Also, for females, 7Y + includes all individuals greater than after-6-years of age, while in males, 5Y + includes all individuals greater than after-4-years of age.

Due to significant, annual, population-wide variation in first-egg date and clutch size, and because clutch size declines throughout the breeding season in tree swallows (Stutchbury & Robertson 1988; Winkler & Allen 1996), before combining data across years we standardized our raw data by calculating z-scores (Zar 1999; p. 72), using the equation:

inline image

where YS= standardized datum, YO= observed datum, and MEANYEAR, MONTH and SDYEAR, MONTH are a variable’s within-year, within-month mean and standard deviation. Each year in our population tree swallows breed in May (i.e. ‘early’ nesters) and June (i.e. ‘late’ nesters). These standardized values are unitless and, when graphed, present age-specific reproductive output in a relative, not absolute, manner. For longitudinal data, values were standardized in a similar manner, but using an individual’s lifetime mean and standard deviation for a variable. Also, despite the effect of habitat on clutch size in this species (Hussell & Quinney 1987), we combined data across aquatic and terrestrial habitats because the age distributions of females were not different between the two habitat-types (all years combined, Contingency Table, G = 9·44, d.f. = 5, P = 0·10).

We analysed for differences in breeding phenology and reproductive output between age groups, within sexes, using resampling statistics (Resampling Stats vs. 4·0 Simon 1995). The standardized values for a variable within each age group were sampled with replacement 1000 times to generate 95% confidence intervals about the means of a variable’s standardized values; where confidence limits do not overlap between age groups, the means are significantly different. Longitudinal data were analysed using one-way anova, with differences between ages identified with Tukey-Kramer post hoc tests (P < 0·05).

We analysed the reproductive performance data of tree swallows in another manner as well. After Rockwell et al. (1993), we calculated a composite measure of reproductive performance for tree swallows by multiplying several breeding output variables together, using parameters that Rockwell et al. (1993; Fig. 1, p. 324) refer to as state variables and transition probabilities. Our sequence of breeding components are analagous to theirs but differ because we did not measure the transition probability ‘egg survival’ (Fig. 1). The result was a variable, expected brood size at fledge (E[BSF]), calculated within each year-, sex-, and age-group (e.g. 1975, female, 2Ys; 1975, female, 3Ys; etc.), according to the equation:

Figure 1.

Breeding components of tree swallows used for calculation of ‘expected brood size at fledging’ (E[BSF]). After Rockwell et al. 1993, state variables (boxes) include clutch size, number of hatchlings and E(BSF), while transition probabilities (ovals) include proportion of nests that failed completely between first-egg and hatching (i.e. nest failure), proportion of nests that failed completely during the nestling period (i.e. brood failure), and hatching (i.e. number hatchlings/clutch size) and brood success (i.e. number fledglings/number of hatchlings).

inline image

where CS = mean clutch size, NF = absolute proportion of nests experiencing total nest failure between first-egg and hatching (e.g. due to abandonment, predation), HC = mean proportion hatch success, BF = proportion of nests experiencing total brood failure during the nestling period, and FH = mean proportion fledging success. We used both resampling techniques, and regression, to analyse E(BSF) across age. Linear and quadratic regressions were performed in JMP 3·2·2 for Macintosh (SAS 1997). Rockwell et al. (1993), citing Welsh, Peterson & Altmann (1988), noted that as long as the variables used in calculating the composite measure were not correlated E(BSF) could be considered an unbiased estimate of reproductive performance. Pairwise correlations revealed that none of the five variables were correlated within either sex (Females, ranges, r = 0·07–0·23, N = 102; Males, r = −0·10–0·27, N = 65; all P > Bonferroni-corrected α = 0·005).

Results

PAIRING, AGE AND EXPERIENCE

Older females tended to pair with older males, and younger females with younger males (Table 1). We chose pairs for this test where both members of a pair had been aged, and each male and female was included only once in the test. This relationship did not hold when the breeding experience of individuals was substituted for age; that is, females with no breeding experience were no more likely to pair with mates with no experience as mates with considerable experience (Table 1). The same goes for males. For this latter analysis, we categorized individuals of both sexes into three groups according to their breeding experience: (i) considerable (i.e. have made ≥ two nesting attempts in our population), (ii) some (one), and (iii) no experience. We assumed that individuals found breeding for the first time in our population had no prior experience, independent of their actual or estimated age. This is a reasonable assumption as our banding effort and population coverage has been consistent throughout the study, and tree swallows show considerable breeding-site philopatry (Rendell & Robertson unpublished data).

Table 1.  Frequency of pairing by females and males according to age and breeding experience of mate. Values are numbers of pairs, with percentage of total number of pairs in brackets. See Methods for ages of birds. Breeding experience groups are: (i) considerable (i.e. have made ≥ 2 nesting attempts in our population); (ii) some (one); and (iii) no experience. Age influenced pairing in tree swallows (chi-square test, χ4,2762 = 12·01, P = 0·02, n = 282 pairs), but breeding experience did not (χ4,2642 = 2·05, P = 0·73, n = 270 pairs). For analysis, cells were combined in 2 × 2 table to meet requirements of chi-square tests (Zar 1999)
Male ageFemale ageMale experienceFemale experience
2Y3Y & A2Y4Y + & A3Y +NoneSomeConsiderable
2Y 7 (2·5)13 (4·6) 2 (0·7)None80 (29·6)13 (4·8)3 (1·1)
3Y & A2Y15 (5·3)88 (31·2)34 (12·1)Some93 (34·4)22 (8·1)6 (2·2)
4Y + & A3Y +16 (5·7)66 (23·4)41 (14·5)Considerable40 (14·8)11 (4·1)2 (0·7)

BREEDING PERFORMANCE AND AGE

Nest initiation

Older females began laying earlier than younger females, and typically earlier than the population average each year (Fig. 2). In absolute terms, females breeding for the first time (i.e. 2Y) laid approximately 3–4·5 days later than experienced females, while females in their second breeding season (i.e. 3Y) began laying approximately 1 day later than older females. After age 4Y, first-egg date was similar among older females, and there was no apparent trend towards later nesting. For males, a similar pattern was evident; mean standardized values of first-egg date shifted from positive (i.e. later laying) to negative (i.e. earlier laying) as birds got older, significantly so between A1Y and 5Y + males (Fig. 2).

Figure 2.

Standardized first-egg date (mean ± 95% CI) by age for female and male tree swallows 1975–99. Values are mean z-scores relative to population-wide annual averages, and confidence intervals were generated using resampling with replacement. See Methods for details of standardization calculations, resampling procedure, and age groups. Differences between age groups occur where 95% CIs do not overlap, and are marked accordingly with letters. Sample sizes of individuals are shown. Grey line represents population-wide mean z-score. First-egg date advanced with age in females and males.

Our longitudinal data provide some support for the results of the cross-sectional data above (Fig. 3). First-egg date for females and males tended to advance with age, significantly so for females who bred in three consecutive years beginning at age A2Y (Fig. 3b).

Figure 3.

Standardized first-egg date (mean ± SE) by age within individual female and male tree swallows 1975–99: (a) females that bred three consecutive times starting at age 2Y; (b) females that bred three consecutive times starting at age A2Y; (c) females that bred four consecutive times starting at age A2Y; and (d) males which bred three consecutive times starting at age A1Y. Values are mean z-scores relative to an individual’s lifetime average. See Methods for details of standardization calculations and age groups. Differences between ages were identified with one-way anova and Tukey-Kramer post hoc tests (P < 0·05), and are marked as such with letters. Sample sizes of individuals are shown. Grey line represents population-wide mean z-score. Generally, first-egg date advanced with age in both sexes, although significantly so only in females who began breeding at age A2Y.

Reproductive output

Reproductive output varied significantly with age in females. First, 2Y females laid clutches that were approximately 0·5 eggs, in absolute terms, smaller than all older females (Fig. 4). Third-year females laid clutches close to the population average, which were significantly smaller than clutches of 4Y and 5Y females by approximately 0·15 eggs. After their fifth year, clutch size declined slightly for 6Y and older females back toward the population average. Second, 2Y females had roughly 0·5 fewer hatchlings than 3Y and 4Y females, and the mean standardized number of hatchlings for all older breeders was at or above the population average (Fig. 4). Third, 2Y females fledged approximately 0·5 fewer nestlings than 3Y and 4Y females (Fig. 4). Interestingly, 5Y and older females fledged considerably fewer young than 3Y and 4Y females.

Figure 4.

Standardized clutch size, and numbers of hatchlings and fledglings (mean ± 95% CI) by age for female and male tree swallows 1975–99. For further details see Fig. 2. Generally, all three variables increased from first-time breeders to middle-aged birds, then tended to decline in the oldest birds, in both sexes.

Our longitudinal data for clutch size provided only qualitative support for the results of the cross-sectional data above. Clutch size for the youngest females tended to increase with age then decrease in the oldest birds, but the relationship was not statistically significant, nor was it for males (one-way anovas, all P≥ 0·26; time of first breeding, age-groups, and sample sizes as in Fig. 3). Data screening resulted in insufficient data for longitudinal analyses of any other reproductive variables.

Older females were no more ‘efficient’ than younger ones with respect to their hatch, brood, or fledging success; all three variables showed only a slight, non-significant improvement from 2 to 4 years of age (Fig. 5). In fact, mean standardized brood and fledging success for females 5Y and older were considerably lower than those for younger females aged 2Y–4Y. To investigate further the decline in number of young fledged, and brood and fledging success in females past middle age, we combined data within middle-aged (i.e. aged 3Y and 4Y) and older females (i.e. aged 5Y +), then applied resampling analysis. The decline in number of young fledged, and brood success, past middle age was not significant, but fledging success was significantly lower in older compared to middle-aged females (Fig. 6).

Figure 5.

Standardized hatch, brood and fledge success (mean ± 95% CI) by age for female and male tree swallows 1975–99. For further details see Fig. 2. Qualitatively, all three variables increased from first-time breeders to middle-aged birds, then tended to decline in the oldest birds, in both sexes, but there were no significant differences with age.

Figure 6.

Standardized number of fledglings, brood and fledge success (mean ± 95% CI) by age for female tree swallows 1975–99. We combined data within middle-aged (i.e. aged 3Y and 4Y) and older females (i.e. aged 5Y +) in this reanalysis. For further details see Fig. 2. Qualitatively, all three variables increased from first-time breeders to middle-aged birds, then tended to decline in the oldest birds. The number of fledglings increased significantly between first-time breeders and middle-aged birds, and fledging success declined significantly between middle-aged birds and the oldest breeders.

For males, considerable overlap in the confidence intervals for each of the output and success variables precluded identifying significant differences between age groups (Figs 4 and 5). Except for clutch size and hatch success, however, mean standardized values for the output and success variables appear to increase until age 4Y, then decline, qualitatively similar to the results for females.

Breeding failure

The proportion of clutches that failed completely prior to hatch, and during the nestling stage, differed with age in females, but not males (Fig. 7). For females, total nest failure became significantly less frequent from 3Y to 6Y of age, and total brood failure declined significantly from 2Y to 5Y birds. There was a trend toward fewer breeding failures in middle-aged males as well, and for both sexes, means of standardized values for both variables were higher in the oldest age-groups.

Figure 7.

Standardized proportion of breeding attempts resulting in nest and brood failure (mean ± 95% CI) by age for female and male tree swallows 1975–99. For further details see Fig. 2, except that sample sizes here refer to the number of years during which data were collected for each age group. Qualitatively, total breeding failure decreased with age, significantly so in females, but increased in older birds.

Expected brood size at fledging

A plot of expected brood size at fledge (E[BSF]), our composite measure of tree swallow breeding output, suggests that reproductive output increased until, then declined after, age 4Y in females and age 3Y in males (Fig. 8). In females, E(BSF) increased significantly from age 2Y to 3–5Y, and was at its highest in 4Y birds, according to the distribution of 95% CIs generated for each age-group by resampling. In males, there was a similar trend with 3Y birds showing the highest E(BSF), significantly so compared to 2Y males. When we regressed E(BSF) on age, significant variance in E(BSF) for females and males was explained by a quadratic relationship (Fig. 8). Linear regression explained a significant, but smaller, proportion of the variance compared to quadratic regression in females (F1,100 = 4·86, P = 0·03, r2 = 0·05), and linear regression did not explain a significant amount of variation in E(BSF) on age for males (Males: F1,63 = 0·78, P = 0·38).

Figure 8.

Standardized expected brood size at fledge (E[BSF], mean ± 95% CI) by age for female and male tree swallows 1975–99. See Methods for details of E(BSF) calculations, and for further graph details see Fig. 2. Sample sizes refer to the number of years during which data were collected for each age group. Plotted for each sex are quadratic regression lines predicting E(BSF) (Females: E[BSF] = −1·36 + 0·89*AGE –0·12*AGE2, F2,99 = 8·59, P = 0·0004, r2 = 0·15; Males: E[BSF] =−1·92 + 1·38*AGE – 0·21*AGE2, F2,62 = 5·49, P = 0·006, r2 = 0·15). Expected brood size at fledge increased significantly in both sexes until middle age, then declined.

Offspring recapture

Mean recapture rates of offspring in our breeding population tended to increase from first-time breeders to intermediate-aged birds, then decrease in the oldest breeders, for both sexes; however, this relationship was not significant for either sex (Fig. 9). In females, 4Y and 6Y birds had the highest mean offspring recapture rate; for males, 3Y birds had the highest mean offspring recapture rate, as one might expect given the E(BSF) results for both sexes (Fig. 8). The lowest mean offspring recapture rates were found for 2Y, 5Y and 7Y + females, and for 2Y and 5Y + males.

Figure 9.

Standardized number of offspring subsequently recaptured in our population (mean ± SE) according to the age of female and male tree swallow parents 1982–97. Recapture rates for offspring reared by middle-aged parents tended to be higher, but this relationship was not significant for either sex (one-way anovas: Females, F5,1446 = 1·07, P = 0·38; Males, F4,717 = 0·45, P = 0·77).

HYPOTHESIS TESTING

Residual Reproductive Value Hypothesis

Individuals may increase their reproductive output with age, and advance their laying date, if the probability of surviving to another breeding opportunity declines with age. Declining age-specific return rates, coinciding with increasing output, would provide support for this hypothesis; no change or increases in age-specific return rates, and declines in output with age, would not. Our breeding performance data above indicate that while output increased, and nest initiation advanced, until an intermediate age, output appeared to remain the same or even decline in older birds, and nest initiation remained similar after ‘middle-age’. Also, survival to another breeding opportunity appeared not to change after age 3Y in females and males, although there was a slight, non-significant decrease in male recapture rates in older birds (Fig. 10).

Figure 10.

Standardized proportion of female and male tree swallows recaptured in years after first-handling according to their age when they were first banded 1982–97. For further details see Fig. 2. In males and females, first-time breeders had a significantly lower recapture rate, at least compared to 3Y and 4Y birds.

Selection hypothesis

This hypothesis predicts that birds that breed more than once should perform better than birds that breed only once when compared within their first breeding opportunities. We analysed the z-scores of four performance variables to test this prediction: first-egg date, clutch size, and hatch and brood success. We found no support for this prediction in either sex for any of the four variables, neither within ages for each sex (one-tailed t-tests, all P≥ 0·015, Bonferroni-corrected α = 0·006), nor within sexes when all ages were combined (Table 2). Thus, birds known to breed only once in our population nested as early, and had similar output, compared to birds known to breed repeatedly in our population during their first breeding event.

Table 2.  Test results for predictions of the Selection Hypothesis in female and male tree swallows. Values are mean z-scores ± SE and sample sizes within first-breeding attempts for birds that bred only once in our population (i.e. Return? = No) compared to those that bred more than once (i.e. Return? = Yes). Birds that bred in more than 2 years in our population did not lay earlier, or have higher breeding output, compared to birds that bred only once, in both sexes
VariableSexReturn?t-test
NoYestd.f.P
1st-egg dateFemale 0·07 ± 0·03, 962 0·06 ± 0·08, 1620·1211220·90
 Male 0·05 ± 0·04, 536−0·08 ± 0·10, 951·22 6290·22
Clutch sizeFemale−0·03 ± 0·03, 947−0·09 ± 0·08, 1640·6311090·53
 Male−0·05 ± 0·04, 515−0·01 ± 0·10, 940·38 6070·70
Hatch successFemale−0·03 ± 0·04, 728 0·07 ± 0·09, 1411·10 8670·27
 Male 0·06 ± 0·04, 422 0·10 ± 0·10, 800·43 5000·66
Brood successFemale−0·06 ± 0·05, 434−0·16 ± 0·11, 990·80 5310·42
 Male 0·04 ± 0·05, 285−0·23 ± 0·11, 571·58 3400·11

Breeding Experience Hypothesis

This hypothesis maintains that individuals with more breeding experience should have higher reproductive output, and earlier nest initiation, than those with less experience when compared within age-groups. To test these predictions we used the three ‘breeding experience’-groups described in the analysis of pairing and age above. We expected that birds with no, some, and considerable experience should start egg-laying progressively earlier, and increase output, respectively. Again, we analysed the z-scores of four performance variables to test these predictions: first-egg date, clutch size, and hatch and brood success. In females within each of four age-groups (i.e. 2Y to 5Y +), and in males within each of three age-groups (i.e. 2Y to 4Y +, A1Y males not analysed), prior breeding experience had no effect on date of nest initiation or output (see Table 3 for results for females aged 4Y; otherwise, t-tests for birds aged 2Y [i.e. experience groups = none vs. some], one-way anovas [all other ages], all P≥ 0·02, Bonferroni-corrected α = 0·01). That is, for example, a bird aged 4Y with no known breeding experience in our population nested as early, and had similar output, compared to a bird aged 4Y with considerable experience.

Table 3.  Test results for predictions of the Breeding Experience Hypothesis in female tree swallows. Values are mean z-scores within experience categories for 4-year-old females. Experience categories are: Considerable (i.e. have made ≥ 2 nesting attempts), Some (one attempt only), and None (no known experience). Standard errors and sample sizes are below each mean. Same-aged females with some, or considerable experience, did not lay earlier, or have higher output, compared to birds with none
VariableExperienceanova
NoneSomeConsiderableFd.f.P
1st-egg date−0·21 0·13−0·443·242, 710·05
  0·21, 16 0·17, 23 0·14, 35   
Clutch size 0·10 0·19 0·320·442, 700·64
  0·21, 16 0·18, 22 0·14, 35   
Hatch success−0·34−0·39−0·220·092, 540·91
  0·38, 12 0·33, 16 0·25, 29   
Brood success 0·13−0·22 0·040·322, 330·73
  0·36, 9 0·30, 13   0·29, 14   

Age Hypothesis

According to this hypothesis, older individuals should outperform younger ones when compared within experience groups. As above, we analysed the z-scores of four performance variables to test these predictions: first-egg date, clutch size, and hatch and brood success. Our results for females with no known breeding experience support this hypothesis (Table 4). Those aged 3Y, 4Y, and 5Y + began laying earlier than 2Y-old, first-time breeders. Also, 4Y-old females laid larger clutches than 3Y females, and both laid more eggs than 2Y females. Females 5Y and older laid larger clutches than first-time breeders, but could not be distinguished from females aged 3Y and 4Y. Age did not affect hatch and brood success (Table 4), and within the two other experience groups, there were no significant differences between different-aged females (t-tests for birds with considerable experience [i.e. age groups = 3Y vs. 4Y + ], one-way anovas for birds with some experience [ages 2Y, 3Y, 4Y + ], all P≥ 0·05, Bonferroni-corrected α = 0·02). Age did not influence the breeding performance of males within experience-groups (t-tests and one-way anovas, all P≥ 0·05). Thus, the Age Hypothesis, but neither the Experience, Selection, nor Residual Reproductive Value Hypotheses, may explain the improvement in breeding performance of tree swallows up until ‘middle-age’ in females.

Table 4.  Test results for predictions of the Breeding Age Hypothesis in female tree swallows. Values are mean z-scores within age categories for females with no known breeding experience. Standard errors and sample sizes are below each mean. Letters denote significant differences between means (one-way anova with Tukey–Kramer HSD post hoc comparison, P < 0·05). Older females breeding for the first time laid earlier and larger clutches than younger females
VariableAGEanova
2Y3Y4Y5Y +Fd.f.P
1st-egg date 0·61 a−0·07 b−0·34 b−0·31 b40·043, 1212< 0·0001
  0·06, 270 0·03, 873 0·13, 52 0·20, 21   
Clutch size−0·44 a 0·05 b 0·44 c 0·20 bc22·403, 1198< 0·0001
  0·06, 264 0·03, 866 0·14, 51 0·21, 21   
Hatch success−0·09 0·01 0·01−0·06 0·543, 950  0·65
  0·07, 200 0·04, 697 0·16, 40 0·25, 17   
Brood success−0·20−0·03−0·22−0·27 1·063, 534  0·37
  0·09, 126 0·05, 377 0·22, 23 0·30, 12   

Discussion

Breeding performance improved within the first few years of life for tree swallows. Nest initiation advanced and clutch size, number of hatchlings, and number of fledglings all increased in consecutive years from females breeding for the first time to those 4 or 5 years of age, based on our cross-sectional data (Figs 2 and 4), and to a lesser extent our longitudinal data (Fig. 3). Also, our measures of breeding efficiency followed a similar pattern (Figs 5 and 6), and the proportion of nests that failed completely declined with age until 5–6 years of age (Fig. 7), as in barnacle (Branta leucopsis Bechstein) and lesser snow geese (Chen caerulescens Linnaeus; Forslund & Larsson 1992; Rockwell et al. 1993). A similar pattern of improvement was seen for males in most of these variables until approximately the same age. The best evidence for improvements in performance come from our first-egg and clutch size data where first-egg date advanced, and clutch size increased, with age in a significant stepwise fashion, as has been shown in barnacle geese (Forslund & Larsson 1992); that is, both variables improved between 2Y and 3Y females, and 3Y and 4Y females. This is relevant because larger clutches increase potential annual output for individuals, and recruitment of juveniles is typically highest for young from early nests (e.g. great tits Parus major Linnaeus, Perrins 1970; collared flycatcher Gustafsson 1989; barn swallows Hirundo rustica Linnaeus Møller 1994). Our results extend the known range of ages over which performance improves in tree swallows. Previous studies have shown differences in first-egg dates and reproductive output in comparisons of 2Y females, and a group, A2Y, into which all older females were combined (De Steven 1978, 1980; Stutchbury & Robertson 1988; Wiggins 1990; Wheelwright & Schultz 1994; Lombardo et al. 1995). Similar to our results Wheelwright & Schultz (1994) also found increases in mean clutch size, and advances in hatching date (as a substitute for first-egg date), until the same age of 4Y in female tree swallows, although these authors did not find significant differences for either variable. Considered together, the results suggest that tree swallows aged approximately 4Y may contribute the most young to the population in subsequent years, a point supported to some extent by our measure of expected brood size at fledging (Fig. 8) and by our offspring recapture data (Fig. 9). Other studies of similarly sized, hole-nesting birds (great tit Perrins & McCleery 1985; collared flycatcher Gustafsson & Pärt 1990), and a hirundine (barn swallow Møller & De Lope 1999), have found output to be highest for birds of ≈ 3–4 years of age as well.

What accounts for the improvement in breeding performance with age in this species? We found no support for three possible hypotheses mentioned above. In tree swallows, reproductive output apparently declines after 4 years of age, while survivorship stays unchanged after 3 years of age in females, and possibly males (Fig. 10). The combination of these two observations does not support the Residual Reproductive Value Hypothesis, which predicts that while survivorship declines with advancing age breeding effort should increase. Other studies have detected declines in survivorship in species with age (e.g. Newton & Rothery 1997 and refs therein), but this has coincided typically with a decline in breeding output, not an increase. Also Røskaft et al. (1983) found that younger rooks show greater foraging effort than old rooks, but most studies have failed to find evidence for differences in effort between age-groups (Nol & Smith 1987; Reid 1988; Desrochers 1992b; Wheelwright & Schultz 1994; Komdeur 1996b). With respect to the Selection Hypothesis, improvements in performance with age in tree swallows do not reflect differential survival between relatively low and high quality individuals, a conclusion reached for the same species by Wheelwright & Schultz (1994; see also Forslund & Larsson 1992; Komdeur 1996b). Differences in quality between individuals in a population may arise due to the ‘silver spoon’ effect (Grafen 1988), as in Columbian ground squirrels (Spermophilus columbianus Ord) where a series of positive associations between female age, size, condition, and early breeding all positively affect litter mass and size, and survival of mothers and young to year-1 (Dobson, Risch & Murie 1999). In our study, first-time breeders (i.e. 2Y birds) showed lower output (Figs 2–6) and lower recapture rates in both sexes (Fig. 10). However, unlike song sparrows (Nol & Smith 1987), birds which bred repeatedly in our population did not begin laying earlier, or show higher output, the first time they nested compared to those birds that only bred once (Table 2). Finally, with respect to the Breeding Experience Hypothesis, male and female tree swallows with considerable breeding experience did not lay earlier, or lay more eggs, nor experience higher reproductive output, within age-groups compared to birds with little or no breeding experience (Table 3; see also Nol & Smith 1987; cf. Komdeur 1996b; Blums et al. 1997). Lozano & Handford (1995) found no differences in the abilities of 2Y (i.e. first-time breeders) and A2Y female tree swallows to feed young, and we found no significant differences in incubation efficiency or ability to rear young to fledge between first-time breeders and older females (Figs 5 and 6). In both of these comparisons, females with no breeding experience performed as well as those with considerable experience.

Instead, our results suggest that the reproductive performance of younger tree swallows is constrained by age. Older females, particularly 4Y-olds, with no known breeding experience, laid earlier, larger clutches compared to 2Y and 3Y females (Table 4). Similarly Nol & Smith (1987) and Pyle et al. (1991) found that age enhanced reproductive performance independently of breeding experience in song sparrows and western gulls (Larus occidentalis Audubon), respectively. Perhaps birds acquire skills as they age that are necessary to prepare and maintain their condition prior to, and during, breeding. Presumably these are not direct breeding-skills, such as incubation or nestling-care skills, but ones devoted to self-maintenance, such as feeding. Reid (1988) attributed the improvement in output with age in glaucous-winged gulls (L. glaucescens Naumann) to what he referred as ‘better coordination of nesting activities’ (p. 1454). With no evidence of increased breeding effort in older birds, female western gulls managed to acquire greater mass after incubation during the nestling period, and older birds spent more time resting and less time in aggressive behaviours than young breeders. Also, other studies have shown that older birds are better foragers than younger ones in a variety of species (reviewed in Marchetti & Price 1989; see also Desrochers 1992a; Cooch et al. 1997). Where authors have found that foraging skills do not differ between older and younger birds, the relative lack of difficulty of the mode of foraging is usually cited as the reason (i.e. ground vs. aerial foraging; Hannon & Smith 1984; Desrochers 1992a). Lozano & Handford (1995) found no difference in nestling-feeding abilities between 2Y and A2Y female tree swallows, and this may seem contrary to the idea that foraging skills are lacking in young tree swallows. In other words, if nestling feeding skills are not different between these female age-groups, why would self-feeding skills be any different? This is not contradictory, however, if one distinguishes nestling- and self-maintenance as two distinct components. Perhaps tree swallows improve in their ability to balance these two competing pressures with age, allowing them to establish and maintain a higher level of ‘conditioning’ throughout the breeding season. In support of this assertion Lozano & Handford (1995) concluded that ‘blue’ (i.e. A2Y) females were in better condition than ‘brown’ (i.e. 2Y) females during breeding; A2Y females were heavier, and had higher fat and protein compositions. Also, mass and protein were significantly reduced in young females compared to older ones as the breeding season progressed, suggesting that young females could not maintain the condition they began with at the start of breeding. Other studies have also shown that adult breeders may be in better condition before and after breeding compared to young breeders (e.g. American coots Fulica americana Gmelin, Alisauskas & Ankney 1987; bison, Green 1990; savannah sparrows, Wheelwright & Schultz 1994).

Lower output of young birds could also result, in part, from being paired with relatively young mates, and studies on other species have found this to be the case (e.g. Pyle et al. 1991; Komdeur 1996b). In our study, there was a weak association between young females mating with young males, although this was not found to be related to their levels of breeding experience.

We found evidence that reproductive performance declines with age after the age of four to five in female tree swallows, and after 3 years in males, similar to another hirundine (barn swallow, Møller & De Lope 1999). For females, clutch, hatch and fledge all increased to 4 years of age then declined, considerably so for 5- and 6-year-olds in their number of fledglings and overall fledging success (Figs 4–6). Further evidence for a decline in output is that E(BSF) for females was best predicted by a quadratic, not linear, regression (Fig. 8), as shown in other studies of age-related breeding output in female birds (e.gs. great skuas, Hamer & Furness 1991; lesser snow geese, Rockwell et al. 1993; sparrowhawks Accipiter nisus Linnaeus, Newton & Rothery 1997). We attribute this decline in breeding performance with age to senescence. Senescence could arise due to accumulated costs of previous breeding effort, accumulated late-expression mutations in the genome, or both (Partridge & Barton 1993); these are not necessarily mutually exclusive hypotheses. It is beyond the scope of this empirical field study to address the possibilities or relevance of mutation accumulation in our birds, but we do have evidence that female tree swallows express costs of reproduction. Elsewhere, we showed that females laid smaller clutches in years after having been forced to raise experimentally enlarged broods (Rendell 1999), and several other studies have found evidence indicating costs to breeding and brood size in this species (De Steven 1980; Wiggins 1990; Wheelwright, Leary & Fitzgerald 1991; Winkler & Allen 1995; Murphy et al. 2000l; Hasselquist et al. 2001). Gustafsson & Sutherland (1988) also found a reduction in future fecundity during brood manipulation experiments on collared flycatchers and they, too, attributed an observed decline in reproductive output with age in this species to senescence, or ‘innate deterioration’ (p. 280), brought on by costs incurred during reproduction (Gustafsson & Pärt 1990). Newton & Rothery (1997) reported senescence in reproductive output and survival after ‘middle-age’ (i.e. 4–6 years) in sparrowhawks based on a long-term, non-experimental study. Without elaborating on costs of reproduction per se, they attributed the decline in performance to ‘general wear and tear’ that accumulates with time, and presumably breeding effort, leading to a decline in ‘efficiency and social status’ (p. 1005; see also Abrams 1991). Likewise for Møller & De Lope’s (1999) study of senescence in barn swallows, where age brought on deterioration in a suite of variables, including fledging success, survival and physical characteristics such as wing and tail length, the latter character being a target of sexual selection (Møller 1989). We do not know if ageing results in a deterioration of physical characteristics in tree swallows. Rendell (1999) found that three physical characteristics of female tree swallows (i.e. body mass, wing & tarsus length) were highly repeatable between years suggesting that any physical deterioration is limited or unmeasurable, but these females were tested only over a limited age-range (i.e. 3 years) and most were fairly young birds; thus we may not have reached the age where senescence begins to have conspicuous phenotypic effects. Rockwell et al. (1993) not only suggested that a decline in the physical abilities of female lesser snow geese could explain senescence in breeding performance in this species, but that a deterioration of the surrounding feeding environment on their nesting grounds over 15 years also had a significant influence. We have no evidence to suggest that variation in local food availability could have affected our results, but this should not matter because we standardized all our absolute measures of individual output on within-year means to remove the possibility of annual variation in output biasing our analyses.

Another possible explanation for the pattern of decline in breeding output with age in females could be that older age-classes include birds with little breeding experience, if reduced breeding experience results in enhanced longevity. Long-term studies of small passerines have found evidence that first-breeding attempts by some adults do not occur until 4 years of age (e.g. blue tit P. caeruleus Ord, Dhondt 1989; pied flycatcher F. hypoleuca Pallas, Sternberg 1989; this study). Curio (1983) suggested that some young birds might forego breeding to acquire skills necessary for successful breeding. Also, it is possible that some individuals might start to nest, but fail repeatedly early in their attempts over successive years, leaving them with advanced age, but little experience. However, the best predictor of lifetime reproductive success in birds is breeding lifespan (Newton 1989 and refs therein; see also Gustafsson & Pärt 1990; Komdeur 1996a), so it is unlikely that selection favours individuals that forego breeding for any significant period of time early in life, especially in relatively small, short-lived passerines (see also Perrins & Moss 1974).

To some extent the relationship between performance and age was weaker in male compared to female tree swallows, as in other species (barnacle geese, Forslund & Larsson 1992; barn swallows, Møller & De Lope 1999). The small sample size of 2-year-old males resulted in considerable variance within that age-group, and therefore great overlap in the confidence intervals of all age-groups when considered together. But the patterns of change were similar overall (Figs 2–7), and E(BSF) was best described by a quadratic regression (Fig. 8) as it was in females, suggesting that males experience an improvement and decline in performance with age. Any analysis of male age and breeding output is complicated by paternity issues in tree swallows because, while maternity is unequivocal because egg dumping is rare in this species (cf. Lombardo 1988; Barber, Robertson & Boag 1996), paternity can fluctuate widely between broods. Thus, coarse estimates of reproductive output for males do not necessarily reflect true breeding success. Several studies in our population have shown that some males may raise entire broods that are genetically unrelated to them (e.g. Lifjeld et al. 1993; Barber, Robertson & Boag 1996; Kempenaers et al. 1999). Second, although molecular studies of paternity in our population have not revealed that male age is related to the frequency of extra-pair copulations or fertilizations (Dunn et al. 1994; Kempenaers et al. 1999), male age is positively associated with extra-pair success in other passerines (e.g. yellow warbler Dendroica petechia Linnaeus, Yezerinac, Weatherhead & Boag 1995 and refs therein). We know of no reason why males cannot acquire the necessary skills for self-maintenance and condition that improve performance in females. With respect to breeding decline, if a male’s overall breeding effort during the breeding season is similar to female effort, independent of the activities to which energy is devoted, then one might predict males to show reduced performance at around the same age as females. Male and female tree swallows provision young at a similar rate throughout the nestling period (Leffelaar & Robertson 1986; Rendell 1999). Also Williams (1988) measured field metabolism in male and female tree swallows during the nestling rearing period only, and found that males and females expired CO2 at comparable rates, suggesting that there is no sex-bias in energy expenditure, at least during the nestling rearing period. Further, of four species of small, secondary cavity-nesting passerines for which sex-specific average survival rates were calculated in Newton (1989; i.e. blue and great tits, pied and collared flycatchers), all four species showed no sex-specific bias in survivorship. Likewise Butler (1988) did not report such a finding in tree swallows, and our results give no indication of a sex-bias in mortality (Fig. 10).

In conclusion, our study has refined our understanding of age-related breeding performance in tree swallows showing that, reproductively speaking, not all adult swallows are alike. Females and males show an improvement from their first possible breeding opportunity over 3 years until ‘middle age’ at about 4 years old, followed by evidence of a decline in the number of young they can expect to fledge. Similar to previous studies (Sæther 1990; Wheelwright & Schultz 1994), we attribute the change in performance with age to both restraint and constraint hypotheses. Improvement in breeding performance may arise from skills acquired with age which keep adults in better condition, both prior to and during each breeding season, while breeding decline may arise due to senescence. Costs of reproduction probably contribute to both the improvement in breeding performance, if there is selection on young birds to withhold reproductive effort, and to senescence.

Acknowledgements

Throughout this study, many enthusiastic and capable people, too many to name, have contributed significantly to data collection in the field and data organization in the laboratory. To them, we offer our gratitude; without their efforts, this research would not have been possible. We also thank landowners in and around the Lake Opinicon, Newboro, and Forfar areas for kindly allowing us access to their property. Frank Phelan, Floyd Connor and all the QUBS staff provided excellent logistical support. Karen Holder, Erica Nol, Laurene Ratcliffe, Bruce Tufts, Harry McCaughey, Brian Cummings, Michael Adams, Ken Norris and two anonymous referees provided constructive criticism of this study and manuscript. During preparation of the manuscript, Tom Smith and Todd Schafer provided desk space and computer facilities for W.B.R. at the Centre for Tropical Research, San Francisco State University. This research was funded by operating grants to R.J.R. from the Natural Sciences and Engineering Research Council (NSERC) of Canada. W.B.R. was supported by a NSERC Postdoctoral Fellowship, NSERC Postgraduate Scholarship, Ontario Graduate Scholarship, Queen’s University Travel and Graduate Awards, and a Queen’s University Thesis Bursary.

Ancillary

Advertisement