INTER- AND INTRASPECIFIC RELATIONSHIPS BETWEEN EGG SIZE AND CLUTCH SIZE IN WATERFOWL FRANK C. ROI-IWER • Departmentof Biology,University of Pennsylvania,Philadelphia, Pennsylvania 19104 USA ABSTRACT.--Lack(1967, 1968a)proposed that clutch size of waterfowl and other birds with self-feedingyoung was limited by females'ability to produceeggs. Lack supported this egg- productionhypothesis by showing a stronginverse relationship between egg size and clutch size within and among speciesof waterfowl. A reanalysisusing updated data and more appropriatestatistics failed to confirm Lack's results.Grouping all the waterfowl produced a weak (r= = 0.13) inverserelationship between relative egg size and relative clutch size. This relationshipwas due mainly to a handful of ducksthat nest on oceanicislands. Analyses by tribes showed that relative egg size and relative clutch size were inversely related in only 2 of the 8 major tribes of waterfowl. Finally, intraspecificanalyses failed to reveal a trade-off between egg size and clutch size in Blue-winged Teal (Anasdiscors) and Northern Shovelers (A. clypeata).Similar intraspecificanalyses for 12 other waterfowl have failed to show the predictedinverse relationshipbetween egg size and clutch size. Theseresults suggest that the widely acceptedegg-production hypothesis may be considerablyoveremphasized. Re- ceived19 December1986, accepted 13 July 1987. A CENTRALquestion concerninglife-history and Canada Geese (Branta canadensis) have adaptationis how many young to have in any shown no relationship between survival of breeding event. Field studies of this problem young and brood size (Rohwer 1985, Lessells have largely dealt with birds. Much of this lit- 1986). Brood size alterations brought about by eraturewas inspiredby the work of David Lack intraspecificnest parasitismalso failed to affect (1947, 1948, 1954a, 1968a). Lack's thesiswas that duckling survival in Wood Ducks (Aix sponsa) clutch size in most birds has evolved to corre- and Common Goldeneyes(Bucephala clangula) spond to the maximum number of young the (Heusmann 1972, Clawson et al. 1979, Rothbart parentscan feed. Lack suggestedthat females 1979, Dow and Fredga 1984;but see Andersson that laid largerthan normalclutches would leave and Eriksson1982). Likewise, clutchesenlarged fewer descendants because the brood would be either experimentally or through intraspecific undernourished and suffer greater nestling or nestparasitism have shown ducksto be capable fledgling mortality. Lack's (1954a, 1968a) con- of hatchinggreatly enlarged clutches with little clusionsthat parents' ability to feed young is or no reduction in the percentageof eggs that more likely to constrainclutch size than their hatch (Leopold 1951, Hori 1969, Morse and ability to lay or incubate eggs are widely ac- Wight 1969, Heusmann 1972, Clawson et al. cepted (Klomp 1970, Ricklefs 1977, HiSgstedt 1979,Eriksson 1979, Dow and Fredga 1984,Roh- 1980). wer 1985). Waterfowl(Anatidae) have highly precocious Waterfowllay largeeggs relative to their body young that leavethe nestshortly after hatching size (Lack 1968a, King 1973, Rahn et al. 1975), and secure their own food. Parental duties con- and they lay large clutches (Johnsgard 1978, sistof leading the brood to feeding areas,warm- Bellrose1980). In many speciesthe total clutch ing chilled young, watching for predators,and, massapproaches the massof the female (Ap- in the larger species,defending the brood from pendix). Such a large commitment to egg nu- predators.With such forms of parental care, it trientssuggests that the productionof eggscould seemsunlikely that survival of young would constrainreproductive output. Lack (1967) pro- be affected by brood size. Manipulations of posedthat "the average clutch of each species brood size in Blue-winged Teal (Anas discors) (of waterfowl) has been evolved in relation to the averageavailability of food for the female • Presentaddress: Appalachian Environmental Lab- around the time of laying, modified by the rel- oratory, Gunter Hall, University of Maryland, Frost- ative sizeof the egg." Lacksuggested that species burg, Maryland 21532 USA. laying eggs that were small relative to their 161 The Auk 105: 161-176.January 1988 162 FRANKC. ROHWER [Auk, Vol. 105 body size would be able to lay many eggs, appropriatefor suchdata because standard regression whereas specieslaying large eggs would lay createsa line of best fit by minimizing only the de- fewer eggs. As a test of this hypothesis Lack viations of the presumed dependent variable from (1967, 1968a)related egg size to clutch size, and the regressionline (Kidwell and Chase1967, Harvey and Mace 1982, Ricker 1984). In relating egg size and concludedthat the two were inverselyrelated. clutch size, neither has logical primacy as the causal Lack'sdescription of a trade-off between egg agent of the variation in the other. For this reason,I size and clutch size has been widely accepted usedprincipal axisanalysis (Sokal and Rohlf 1981) to asstrong support for the hypothesisthat clutch provide a line of relationship between variablesthat size in waterfowl is limited by egg production. demonstrateda significant Pearson correlation. The Lack,however, employed an inappropriatecor- strengthof the principal axis is indexed by the cor- rection for the allometry of egg size to body relation coefficient. size, made relatively arbitrary categoriesof egg Intraspecificanalyses of egg size and clutchsize.--I sizesin his analyses,and was forcedto useques- gathered data on Blue-winged Teal and Northern tionable data for some species(Lack 1968a:ap- Shovelers (Anas clypeata)breeding in southwestern Manitoba in pothole habitat (for a descriptionof the pendix 15). I re-examined the relationship be- study area see Evans et al. 1952). For both speciesI tween egg size and clutch size on inter- and located active nests and weighed eggs (or measured intraspecific levels in an effort to reassessthe length and breadth of incubated eggs)and repeatedly hypothesisthat eggproduction limits clutchsize checkednests to determine the number of eggs laid. in waterfowl. Frequent nest checksduring the egg-layingperiod have never revealed casesof intraspecific nest para- METHODS sitism for either species.I also measured wing, bill, tarsus,keel, and body lengths (bill tip to the end of Interspecificanalyses of eggsize and clutch size.--The the tail) for females that •-ere nest trapped (Weller interspecificanalyses required information on female 1957a) or collected(for energeticsstudies) at their body mass,the massof unincubatedeggs, and clutch nest. Blue-winged Teal data were collectedin 1978- size (Appendix). Becauseegg size and clutch size are 1983 and shoveler data in 1980-1983. presumed to covary negatively, I tried to use sources I used egg massas an index to the cost of an egg. reporting data for the samepopulation in the same Eggmass is a goodindex to the costof an egg in terms years.At the minimum I used data for the samesub- of energy or lean dry content for Blue-winged Teal species.I used female massestaken at the beginning (Rohwer 1986a),Northern Shoveler (unpubl. data on of incubation when such detailed data were available. 213 eggs),and severalother waterfowl (Manning 1978; For somespecies egg masswas calculatedfrom egg Ankney 1980;Birkhead 1984, 1985). Likewise, the pro- dimensionsusing the equation: portion of yolk and the compositionof yolk and al- bumen are similar for the eggs of most waterfowl egg mass= constant.length.breadth 2 (Lack 1968b, Rohwer unpubl. data), suggestingthat egg size is an adequateindex of egg costin compar- (Hoyt 1979).The constantof 0.555 (g/cm 3) was used; isonsbetween species. this was calculatedfrom a variety of waterfowl data (Young1972; Laughlin 1976; Mackenzie and Kear 1976; Riggert 1977;Norman 1982;Summers 1983; Rohwer RESULTS 1986a,unpubl. data) and is unaffectedby egg size. I usedLivezey's (1986) tribal classification.The main Interspecificrelationships between egg size and way this classificationdiffers from others (Delacour clutch size.--In waterfowl, egg mass increases and Mayr 1945;Delacour 1954, 1956, 1959;Johnsgard with bodysize, but egg massas a proportionof 1978; Bellrose 1980; A.O.U. 1983; Scott 1985) is to adult body massdecreases with body size, as is eliminate the tribe of perching ducks (previously: typicalof avian groups(e.g. Rahn et al. 1975). Cairinini) and split the swans and geese into two This relationship is best demonstratedas a log- tribes. The former change had been suggestedpre- viously based on skeletal characters,behavior, and arithmic plot of egg massvs. body mass(Fig. 1; hybridization studies (Johnsgard1960, 1979; Wool- r = 0.92, n = 152, P < 0.0001).The slope of the fenden 1961). Names for North American speciesare principal axis of this log-log plot measuresthe those of the American Ornithologists' Union (1983). exponentin the power functionthat relatesegg Specificand subspecificnomenclature for other wa- mass(E) to female body mass(B): terfowl follow Johnsgard(1978). Severaltimes in this paper I report the relationship E = 0.47Bø.72. (1) betweentwo setsof data eachwith natural variability, such as egg massand female mass.Standard regres- The constant,0.47 (g/g body mass),is the in- sion techniques,though commonly applied, are not tercept from the samelog-log plot (Fig. 1), and January1988] WaterfowlEggs and Clutch Size 163 1000 TABLE1. Statistics for the relationship of log egg massand log female mass. Corre- lation 100 coef- Inter- Tribe ficient n P Slopea ceptb Anatini 0.85 53 <0.0001 0.67