Ecological and Social Factors Affecting Hatchability of Eggs

--Hatchability, defined as the proportion of eggs surviving to the end of incubation that hatch, varies among populations of birds. Here, I examine the effects of a variety of variables on hatchability in a comparative analysis of 155 studies of 113 species. Of the ecological nd geographical variables considered, latitude, diet, and nest type explain a significant amount of the variance in hatchability due to a significant increase in hatchability along a latitudinal gradient and a decrease in hatchability in carnivorous and holenesting species compared to herbivorous and open-nesting species. Of the variables related to sociality, all affect hatchability adversely as they increase along a sociality gradient (defined as the likelihood of increasing frequency of interactions with conspecifics), significantly so in the case of social organization and possibly incubation pattern. In addition, hatchability of eggs in nests of Acorn Woodpecker (Melanerpes formicivorus) groups containing more than one breeding male and/or female is significantly lower than that observed in groups with a single breeding individual of each sex. Although the causes of these trends in hatchability are unclear, these results provide comparative vidence to support he hypothesis that there is a direct, detectable r productive cost o individuals pursuing strategies that are more social. Received 9 March 1981, accepted 11 February 1982. A• egg failing to hatch is a considerable energetic loss to the bird that laid it as well as to those that incubated it. As a result, selection can be expected to fine-tune the processes of fertilization and incubation and the physiology of the eggs themselves to maximize the probability of an embryo successfully forming, developing, and hatching. This process has not been perfected, and all species of birds for which a reasonable sample has been obtained suffer some hatching failure unrelated to predation or abandonment. Hatchability of eggs is known to have a genetic basis in domestic fowl (Moseley and Landauer 1949) and is known to be affected by numerous environmental and nutritional factors (Taylor 1949). However, little is known about what factors, if any, affect interspecific variability in hatchability or of social factors that might influence hatchability within a population. The purpose of this paper is to begin to fill this gap. I examine, in a comparative fashion, the effects of a diverse array of variables on hatchability. Considerable variability is shown • Present address: Hastings Reservation, University of California, Star Route Box 80, Carmel Valley, California 93924 USA. 526 to exist, some of which can be correlated with both social and ecological factors. The correlations of the social variables with hatchability prompted a second analysis comparing hatchability among several subsets of a population of the cooperatively breeding Acorn Woodpecker (Melanerpes formicivorus). Significant variability in hatchability is also shown to be present within this species. MATERIALS AND METHODS Hatchability is defined here as the percentage of eggs surviving to the time of hatching that produce a chick. Thus, eggs lost to predation, abandonment, accidental breakage, or any other unknown factor are excluded. Two types of data were analyzed. First, I compiled hatchability data from the literature in as wide a variety of avian species as was practical. In all, data were extracted from 155 populations of 113 species, including representatives from 13 orders, 42 families, and 92 genera. (Sixteen species were the subject of 2 studies each, seven species were the subject of 3 studies, and a single species was the subject of each of 4, 5, and 6 studies [the American Robin (Turdus migrator/us), the Herring Gull (Larus argentatus), and the Eastern Bluebird (S/alia sial/s), respectively].) All but nine studies were done in the Northern Hemisphere. Nearly all were done at low altitudes (only seven were at an altitude above 1,000 m), and my search was biased toward North American rather than European studies. Otherwise, a The Auk 99: 526-536. July 1982 1982] Hatchability of Eggs 527 wide taxonomic, geographical, and ecological range of species was included. A list of the species used and references consulted is available from the author on request. For each population, the following data were recorded. (1) Year the study was performed: before 1946, 1946-1965, or post-1965. This variable was used to test for a difference in hatchability in years before the use of DDT (pre-1946) compared to years when use was widespread (1946-1965) and years when use was declining or absent (post-1965). (2) Taxon: nonpasserine or passerine. (3) Primary habitat type: aquatic or terrestrial. (4) Diet: primarily granivorous/ herbivorous, insectivorous/primary carnivore, or secondary carnivore. (5) Nest location: ground, trees and shrubs, or cavity. (6) Latitude. (7) Approximate altitude. (8) Mean clutch size of the population. (9) Cube root of mean egg volume. Volume was estimated as length x breadth s x •r/6; the cube root of this value is inversely proportional to the surface to volume ratio and thus directly proportional to an egg's thermal inertia (Kendeigh 1972). (10) Spacing pattern: all-purpose territory, mating and nesting territory used only for some food acquisition, or colonial. (11) Incubation: female only or both male and female. (12) Social organization: primarily monogamous, polygyny common, or cooperative breeder. (13) Number of eggs laid that was known to survive to the time of hatching. (14) Number of these eggs that hatched. Most data were derived from the original source when possible; exceptions were egg size (usually obtained from Murphy 1936, Witherby et al. 19381941, Brown and Amadon 1968, or Harrison 1978) and diet (usually derived from Bent 1919-1968, Witherby et al. 1938-1941, or Martin et al. 1951). Several commonly encountered problems were dealt with as follows. (1) When no clutch size was given for a population, this datum was omitted. (2) Presence of male incubation, when not known from the original source, was often inferred from Bent (1919-1968), Witherby et al. (1938-1941), Kendeigh (1952), Skutch (1957), or Verner and Willson (1969). If information on male incubation from these sources conflicted, this datum was omitted. (3) For cooperative breeders, the dichotomy for incubation was whether only a single female incubated or if more than one individual incubated. (4) The category of cooperative breeding as a type of social organization included all species that regularly breed in groups, regardless of the actual or presumed mating pattern within groups. (5) Because many studies lump all causes of egg mortality, care was taken to exclude studies from which it was not possible both to derive an unambiguous estimate of the number of eggs that survived to hatching and to have reasonable confidence about the subsequent fate of those eggs. Studies done on populations likely to have been affected by environmental contaminants were excluded, as were those with a sample of less than 25 eggs. (6) Few studies provide information on the different causes of hatching failure (e.g. infertility versus embryo mortality or death during hatching); thus, all such losses were lumped together in determining hatchability. Dividing the number of eggs hatching (variable 14, above) by the number surviving to the end of incubation (variable 13, above) yields the proportion of eggs hatching. The distribution of values for this variable from the 155 populations considered was tested with the Kolmogorov-Smirnov goodness of fit test (Siegel 1956) and found to be significantly nonnormal (D = 0.151, P < 0.01). Arcsine transformation [H = arcsine X/•, where p = proportion of eggs hatching (Sokal and Rohlf 1969)] successfully normalized the hatchability distribution by this test. Neither the raw nor transformed data was found to be normally distributed, however, when tested by the more powerful Cramer-von Mises or Anderson-Darling statistics (Stephens 1974). Thus, nonparametric statistics were used except when no appropriate nonparametric test existed, in which case the arcsine transformed data were used. Two-tailed tests were used in order to reflect the absence of a priori assumptions as to the direction of differences to be expected. Because numerous species are represented by two or more studies, I performed an analysis of variance of the transformed hatchability data (1) using sets of three or more studies of the same species as groups and (2) using sets of three or more studies of species within the same genus as groups, excluding genera for which all studies were of the same species. This procedure tests for the effect of considering each study an independent sample by assessing the relative variance in hatchability within studies of the same species compared to studies of species within the same genus (see Krebs and Davies 1981: 40). The mean square of hatchability within the 11 groups of studies of the same species was 26.1, whereas the mean square within the eight groups of species within the same genus was 25.3. Because no increase in variance is shown if species are considered rather than individual studies, there appears to be no reason to lump studies of the same species (Krebs and Davies 1981). As a precaution, however, all analyses discussed in the paper were also performed by using only the study with the greatest sample size for each species; the results of these analyses are discussed whenever they differ from those considering each study independently. The second set of data considered was from the cooperatively breeding Acorn Woodpecker (Melaherpes formicivorus). Data were gathered in the field between 1975 and 1981 at Hastings Natural History Reservation, central coastal California. Acorn Woodpeckers live and breed in permanently territorial family groups of varying size (2-15 individuals) and composition (MacRoberts and MacRoberts 1976, Koenig and Pitelka 1979, Koenig et al. in press). 528 WALTER D. KOElXlIG TABLE 1. Effect of five miscellaneous variables on hatchability. a [Auk, Vol. 99 • percentage Variables hatchability n X 2 P value