Geographic and ecological variation in clutch size of Tree Swallows

Auk, The, Jan 2000 by Dunn, Peter O, Thusius, Kevin J, Kimber, Kevin, Winkler, David W

ABSTRACT.-In many birds, populations breeding at higher latitudes have larger clutch sizes. It has been hypothesized that this relationship results from the greater availability of food at higher latitudes. We examined geographic variation in clutch size of Tree Swallows (Tachycitieta bicolor) in relation to latitude, longitude, laying date, breeding density, elevation, and summer actual evapotranspiration (AE). AE is a measure of terrestrial primary productivity, and we assumed that it was correlated with resource abundance. Clutch size data were compiled from nest records throughout the breeding range in the United States and Canada (n = 7,459). Summer AE estimates were made from long-term climatological records, and breeding density was estimated from Breeding Bird Surveys. All variables were averaged for each I' block of latitude and longitude (n = 85 latilong blocks). We found a strong positive relationship between clutch size and latitude. Absolute levels of resource abundance (summer AE) were not related to clutch size but were related inversely to latitude and longitude. After controlling for the effects of confounding variables, clutch size was related positively to latitude and relative resource abundance (summer AE after controlling for breeding density). Our results are consistent with Ashmole's hypothesis that relative, rather than absolute, levels of resource abundance determine at least some of the variation in clutch size. Received 21 December 1998, accepted 20 July 1999.

ONE OF THE MOST consistent patterns of variation in avian reproduction is an increase in clutch size with latitude. Numerous studies since the 1940s have documented this relationship in the Northern and Southern hemispheres (Moreau 1944; Lack 1947, 1948). Hesse (1937) and Lack (1947, 1948) suggested that in altricial birds, this pattern resulted from increased foraging time afforded by longer day length at higher latitudes. Alternatively, Ashmole (1961, 1963) suggested that clutch size depends on the relative level of resource availability during the breeding season. Thus, although temperate areas have fewer absolute resources than tropical areas, Ashmole hypothesized that populations at higher latitudes would have larger clutches because their breeding densities are much lower relative to the abundance of resources. Hence, Ashmole's hypothesis is based on the seasonality of resources. If population size is limited by periods of resource scarcity during the nonbreeding season, then highly seasonal environments will have more resources per capita available for breeding. As a consequence, clutch sizes should be larger in environments with greater seasonal variation in resources.

Ricklefs (1980) assumed that food supply was the main limiting resource for bird populations and examined Ashmole's hypothesis using the ratio of summer to winter actual evapotranspiration (AE) as an estimate of seasonality. Actual evapotranspiration is moisture transfer from the earth's surface to the atmosphere and is a measure of terrestrial primary productivity (Rosenzweig 1968). Support for Ashmole's hypothesis has been found in analyses of passerine assemblages worldwide (Ricklefs 1980) and in woodpeckers (Koenig 1984, 1986), but not in swallows (Moller 1984) or House Wrens (Troglodytes aedon; Young 1994). One potential problem with testing this hypothesis for migratory species is estimating the ratio of summer to winter AE. In migratory species, one must know the location of the wintering grounds to estimate the winter AE experienced by a particular breeding population, and in many cases this information is unknown. A more direct test of Ashmole's hypothesis involves estimating relative resource abundance using summer AE and breeding density, rather than inferring relative resource abundance from seasonality. Ricklefs (1980) and Koenig (1984, 1986) have performed the only tests of Ashmole's hypothesis using direct estimates of breeding density to calculate relative resource abundance. Thus, the generality of Ashmole's hypothesis remains unresolved.

In this study, we examined geographic variation in the clutch size of Tree Swallows (Tachycineta bicolor). The Tree Swallow is a socially monogamous aerial insectivore that uses secondary cavities as nest sites, although it readily accepts nest boxes (Robertson et al. 1992). Tree Swallows typically raise a single brood per season and breed throughout most of the United States and Canada north to tree line. Clutch size typically varies from three to eight eggs (Paynter 1954, Dunn and Hannon 1992) and is positively correlated with food abundance during laying (Hussell and Quinney 1985, Dunn and Hannon 1992, Winkler and Allen 1995). Large variation in clutch size, wide geographic range, single-broodedness, and willingness to use nest boxes makes Tree Swallows an excellent choice for studying large-scale patterns of variation in clutch size.

METHODS

We examined 40 years of data (1952 to 1992) from more than 21,000 nest record cards supplied primarily by the North American Nest Record Program (Cornell University, 1970 to 1992), the British Columbia Nest Records Scheme (BC Wildlife Branch, 1952 to 1991), the Ellis Bird Farm (Lacombe, Alberta, 1981 to 1986), the Maritimes Nest Records Scheme (Canadian Wildlife Service, 1962 to 1990), the Prairie Nest Records Scheme (Manitoba Museum of Man and Nature, 1958 to 1986), and the Quebec Nest Records Scheme (National Museum of Canada, 1959 to 1988). These records span most of the known breeding range of Tree Swallows across the United States and Canada (Fig. 1). All records used in this study came from nest boxes, which have larger mean clutch sizes than natural cavities (Robertson and Rendell 1990). We used only records with at least two visits to the nest box during May or June (the normal laying period) in which eggs or nestlings were recorded on both visits. We did not use any records where it appeared that our estimates may have been biased by egg dumping or brood reduction after hatching. We assumed that any potential biases such as observer error (see McNair 1987), type of nest box, and maternal age effects (DeSteven 1978) were similar with regard to our predictor variables (e.g. latitude, elevation, and evapotranspiration). We also included date of clutch initiation (laying date) in the analysis because it has been suggested that geographic variation in clutch size is caused by variation in the timing of laying (Hendricks 1997), and clutch size is affected strongly by laying date in Tree Swallows (Stutchbury and Roberston 1988, Winkler and Allen 1996). Laying date was estimated only at nests in which the number of eggs differed between the two (or more) visits. Laying date was estimated by backdating one egg per day from the date of the first record of eggs in the nest. Clutches with fewer than three eggs (n = 43) were dismissed because these numbers likely were due to abandonment. Clutches with more than eight eggs (n = 26) also were not used owing to the likelihood of egg dumping or of two females nesting in the same box (Robertson et al. 1992). Varying numbers of samples were collected from the same general location (Fig. 1). To reduce the problem of pseudoreplication, we estimated the mean of each variable from all nests within 1' blocks of latitude and longitude (latilong). Analyses were restricted to latilong blocks with at least 10 nest records. Our final sample size was based on 7,459 Tree Swallow clutches in 85 latilong blocks 88 clutches per block; Fig. 1).