In a population of Green Woodhoopoes (Phoeniculus purpureus) in the Rift Valley of Kenya, food abundance and breeding success varied greatly as a result of highly unpredictable patterns and abundance of rainfall. A given pair of woodhoopoes may breed successfully from zero to three times between May and December. We found a female-biased sex ratio in first nests of the year among those female breeders that had few (0-2) helpers. Apart from predation of entire broods, nestling mortality was extremely low (5%), and the bias was not based on differential starvation of male and female chicks. We examined several possible explanations for the female-biased sex ratio, and we suggest that production of female offspring by breeders in small groups is adaptive in that females are smaller and probably less expensive to rear than males. Young female nest helpers contributed significantly more feedings to subsequent broods than did their male siblings of the same age. The bias toward daughters could be effected either via nonrandom segregation of sex chromosomes or by differential mortality of eggs by sex. In several other bird species, larger eggs may give rise to males. This provides a possible means for the female parent to discriminate the sex of offspring before hatching. Received 23 January 1990, accepted 12 May 1990. IN ANIMALS that reproduce sexually, each breeding individual should attempt to maximize its lifetime reproduction. One way parents might enhance their own reproductive advantage is through facultative control of the sex of their offspring. For vertebrates, this is a controversial and poorly understood subject. Although numerous explanations exist for why individuals should control the sex of offspring under specific conditions (Trivers and Willard 1973, Myers 1978, Williams 1979, Burley 1982, Charnov 1982, Clutton-Brock and Albon 1982, Clutton-Brock 1986), evidence for adaptive control of offspring gender in birds is scanty (Howe 1977, Williams 1979, Fiala 1981, Richter 1983, Clutton-Brock 1986, Breitwisch 1989) and usually ambiguous (cf. Ankney 1982 and Cooke and Harmsen 1983, and Blank and Nolan 1983 and Weatherhead 1985). In his review of sex ratio variation in birds, Clutton-Brock (1986: 326) concluded that "Sound evidence for sex ratio variation at hatching is thus scarce." We report data that imply that in a cooperatively breeding bird (the Green Woodhoopoe, Phoeniculus purpureus), a female-biased sex ratio at hatching may occur under certain ecological and social conditions. METHODS AND PERTINENT NATURAL HISTORY We studied the social behavior and ecology of a marked population of Green Woodhoopoes near Lake Naivasha, Kenya, from mid-1975 through January 1982, and in June-July 1984 (Ligon and Ligon 1978, 1983, 1988, 1989a, b). Briefly, social units consist of 2-16 birds that contain no more than a single breeding pair, regardless of flock size. Each group defends a territory year-round, and territory size often is related positively to group size. In each year breeding normally begins in late May or early June, following the "long rains" of March to May, and generally terminates in December or earlier. The number of successful nesting efforts per year appears to be controlled by food availability, which in turn is determined largely by patterns of rainfall. Two broods during the period June to December are common, and if environmental conditions are extremely favorable, three broods can be produced by one pair during this interval. When conditions are poor, breeding activity is limited, and only one or even no nesting attempt may occur. In such years, there is a lot of territory-to-territory variation both in the time of breeding and in the number of nests attempted. Moth larvae make up the vast majority of food items of nestlings and adults during the sixor seven-month nesting season (unpubl. data), and numbers of these caterpillars through the year are related most directly to the amount of precipitation during the putative dry season, January to March. During this period, the moth caterpillars pupate in the ground and survive well only if the soil remains dry. If extensive rain falls during the "dry" season, mortality of the pupae is high, and few adults emerge to give rise to the generation of caterpillars that appears following the long rains. There is a significant inverse relationship between dry-season rainfall and subsequent repro765 The Auk 107: 765-771. October 1990 This content downloaded from 157.55.39.104 on Mon, 20 Jun 2016 06:37:42 UTC All use subject to http://about.jstor.org/terms 766 LIGON AND LIGON [Auk, Vol. 107 ductive success in this population of woodhoopoes (Ligon and Ligon 1989a, b). Nests are placed in tree cavities. We used an adjustable mirror and flashlight to inspect nest cavities after the eggs were laid. After determining clutch size, we generally did not inspect the nest again until the young began to hatch. This was indicated by changes in the behavior of the breeding female and helpers. After one or more eggs had hatched, the breeding female continued to eat large food items brought to her, but carried very small items into the nest cavity. As the eggs hatched, the helpers increased greatly the frequency of feeding visits and the number of minute food items brought to the nest. Unhatched eggs remained in the nest. Clutch size in Green Woodhoopoes varied little; 88% were either threeor four-egg clutches. Despite this conservatism in clutch size variation, one third of the eggs in clutches of all sizes failed to hatch (see Ligon and Ligon 1988: table V). Separation of the data by year revealed a similar pattern: one third of all eggs laid in each year did not hatch (Ligon and Ligon 1988: table VI). Ligon and Ligon (1988) attribute this high rate of hatching failure to inbreeding depres