-We investigated the cost of first-time breeding in a population of Lesser Snow Geese (Anser caerulescens caerulescens) nesting at La Perouse Bay, Manitoba, Canada. We estimated local survival and capture probabilities of female geese by capture-recapture analysis. We first found that birds were less likely to be recaptured one year after their first successful breeding than on later occasions. Since only successfully nesting birds are captured, this suggests that first-time breeding affects the ability of nesting the next year. We then show that this effect of first breeding is much more severe for birds nesting at age 2 (the youngest age at which Lesser Snow Geese can breed) than for birds starting to breed at an older age. Finally, we compare the mean expected lifetime reproductive success for birds breeding for the first time as two-year-olds or as three-year-olds, conditionally on their survival until age 4. On average, birds first nesting as two-year-olds produce similar numbers of offspring in a lifetime as birds starting at age 3. Received 28 July 1993, accepted 11 January 1994. ASSESSING THE COSTS of reproduction is one of the key challenges of applying life-history theory to natural populations. There are two major questions: (1) Does present reproductive effort have an adverse effect on future life history (survival or reproductive performance)? (2) Do relative costs of reproduction vary among different segments of the population? Empirical answers to the first question are extremely difficult to obtain in the field. Although life-history theory predicts some genotypic trade-offs between early and late reproduction (Williams 1966), negative phenotypic trade-offs are not necessarily expected, since environmental factors can induce positive covariation among individuals for several components of fitness (Reznick 1985, Van Noordwijk and de Jong 1986). As a consequence, whereas a negative phenotypic covariance among life-history traits is evidence for such trade-offs, results showing no or positive covariance are not evidence against (Nur 1990). Many researchers, therefore, have concluded that experimentation is necessary to investigate the costs of reproduction (e.g. Gustafsson 1990). However, to investigate differences in overall fitness among individuals or groups within a population, there is no substitute for detailed measures of components of fitness under field conditions. Such studies are essential for a bet1 Present address: Department of Biological Sciences, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada. ter understanding of the selective pressures in natural populations. In highly philopatric species, birds often are absent from the nesting area in the year following their first observed successful breeding (Wooller and Coulson 1977, Ollason and Dunnet 1988, Weimerskirch 1990). This suggests that the first breeding experience is more costly than those in subsequent years. If this is the case, we might expect the cost of first-time breeding to vary with the age of first breeding. A second question is whether the cost of first-time breeding is a function of age per se or of the breeding event itself (i.e. do birds fail more frequently after their first successful attempt because they were young, or because they bred successfully relative to birds of same age that did not breed?). Newton (1988) found no differences in lifetime reproductive success among female Sparrowhawks (Accipiter nisus) that had bred for the first time at one, two, or three years of age. He reasoned that, because of the risk of mortality, it is advantageous for the birds to begin reproduction as early in life as possible, even if breeding performance in the early years is poor. He attributed variation in age of first reproduction to variation in quality among individuals, such that only good-quality birds would be able to breed at the youngest ages. He concluded that delayed maturation was a result of environmental conditions (the constraint hypothesis) rather than of life-history trade-offs (the restraint hypothesis; Curio 1983). However, the lifetime-reproductive-success calculations used by Newton (1988) are only valid if resighting probabilities are close to 100%.