Consequences of a female-biased sex-ratio in a socially monogamous bird: female-female pairs in the Roseate Tern Sterna dougallii

In the socially monogamous gulls and terns, female-biased sex ratios are so metimes revealed by the occurrence of 'supernormal clutches', which are usu ally attended by female-female pairs or other multi-female associations. We studied these phenomena in the endangered Roseate Tern Sterna dougallii at Bird Island, USA, from 1970 to 1995. DNA-techniques were used to sex breed ing adults in 1992-94. Supernormal clutches (with three or four eggs) have comprised 1-7% of all Roseate Tern clutches at Bird Island since at least 1 970, probably increasing in frequency since 1980. Supernormal clutches were spatially clustered; most were laid late in the peak period of nesting dur ing each season. More than 80% of supernormal clutches and at least 7% of n ormal clutches were attended by multi-female associations; most of these we re female-female pairs, with a few trios (male + two females, or three fema les) and one quartet (four females). More than half of the multi-female ass ociations attended normal clutches. Some female-female pairs were maintaine d for up to five years. The age-distribution of females mated to females di d not differ significantly from that of females mated to males. Females mat ed together usually laid eggs synchronously (+/-2 days). Such females laid fewer eggs than females mated to males (means 1.20 versus 1.73), and had lo wer fertility and hatching success (about 46% versus 98%); they were less s uccessful in raising young from eggs that did hatch (means 58% versus 73%), but this difference was not significant. Their overall breeding success wa s much lower (about 0.34 fledglings per female versus 1.35). The sex-ratio of breeders was about 127 females to 100 males; about 20% of breeding femal es did not have male mates. Female Roseate Terns that do not obtain male ma tes appear to be of low phenotypic 'quality' - based on late laying, small clutches and small eggs. Our data support the hypothesis that such females have a higher fitness if they mate with each other and raise a few young th an if they do not breed at all.