Mate desertion and reproductive effort in the snail kite

Trivers' (1972) mate desertion model was tested by relating measures of reproductive effort for snail kites, Rostrhamus sociabilis, to the sex of the deserting parent. Deserters incubated, brooded and chased potential predators as frequently as did non-deserting mates. Only provisioning of nestlings differed significantly between deserters and their mates, correctly predicting which individual deserted at all nests. Reduced feeding by deserters became established within the first 2 weeks after hatching. Cluster analysis sorted three of four nests deserted by males into one cluster and the other male-deserted nest, which was deserted later in the nesting cycle than all others, clustered with non-deserting nests. Daily energy expenditure was measured throughout the nesting cycle to estimate reproductive effort: at three of four nests, deserters expended less energy than their mates prior to desertion, but in one case, cumulative energy investments did not predict which mate should have deserted. Differences in reproductive effort between mates may have predicted which mate deserted because (1) cumulative reproductive effort is a good indicator of the costs and benefits of desertion, and (2) deserters may assess the potential for deserting by testing their mate's abilities to care for the young. The influence of other factors affecting the costs and benefits of mate desertion decisions by snail kite parents is also discussed. Trivers (1972) recognized that when the disparity of reproductive effort between the sexes was large, sexual selection could result in different strategies by the sexes to optimize the lifetime expenditure of reproductive effort. A mating system may then evolve as a function of the relative and temporal disparity in reproductive effort invested by the sexes. Trivers' (1972) hypothesis predicts that the sex whose cumulative investment is exceeded by its mate should be more tempted to desert because the deserter loses less than its partner if no young are raised. However, desertion decisions should be based on the expected costs and benefits (in terms of future reproductive success) of desertion to the deserter, regardless of past investments unless they affect future reproductive efforts (Dawkins & Carlisle 1976; Boucher 1977). Thus, Trivers' hypothesis can be restated as: an individual investing less reproductive effort than its mate should desert because it has less to lose if the current bout fails, since in future bouts a deserter expects to expend less reproductive effort than its mate to accrue the * Present address: Department of Zoological Research, National Zoological Park, Smithsonian Institution, Washington, D.C. 20008, U.S.A. (to which correspondence should be addressed) and Department of Wildlife and Range Sciences, School of Forest Resources and Conservation, University of Florida, Gainseville, Florida 32611, U.S.A. same reproductive benefit (reach the same point as when it deserted in the present bout). The tendency to desert, however, will depend not only on the differences in reproductive effort between mates, but also on other factors affecting the costs and benefits of desertion (Boucher 1977; Maynard Smith 1977; Alexander & Borgia 1979): (1) the potential for increased or decreased reproductive success by deserting, (2) confidence in paternity or maternity, and (3) age as it relates to both reproductive value (Fisher 1930) and experience. Studies testing Trivers' mate desertion model (sometimes denoted as parental investment theory) are rare. The conditions favouring mate desertion have been examined using game theory and graphical models (Maynard Smith 1977, 1982; Grafen & Sibly 1978), but it is not clear from field studies whether animals make decisions based on past investments rather than future benefits (Robertson & Biermann 1979; Weatherhead 1979; Andersson et al. 1980; Dawkins & Brockmann 1980; Coleman et al. 1985). For several reasons no studies have convincingly related reproductive effort to specific mating systems. First, mate desertion, if it occurs, is usually sex-specific (Kleiman 1977; Ridley 1978; Perrone & Zaret 1979; Wells 1981; Oring 1982); since only one sex deserts, the sex of the deserter can rarely be manipulated experimentally. Second, the distribution of reproductive effort between the