Anim. Behav., 1997, 53, 423–427 COMMENTARIES Studying paternity and paternal care: pitfalls and problems BART KEMPENAERS* & BEN C. SHELDON† *Konrad Lorenz Institute for Comparative Ethology, Vienna †Institute of Cell, Animal & Population Biology, University of Edinburgh (Received 15 August 1996; initial acceptance 6 September 1996; final acceptance 12 November 1996; MS. number: -1086) Theoretical treatments of the relationship between paternity (or certainty of paternity) and paternal care predict, under certain conditions, that indi- vidual males should invest less in paternal care if their paternity is reduced (Whittingham et al. 1992; Westneat & Sherman 1993). Several recent papers have addressed this question using information about males’ genetic paternity from unmanipulated breeding attempts together with measures of those males’ contributions to some components of parental care (Table I). The gen- eral picture that emerges from these studies is that there is no positive relationship between paternity and paternal effort (although in some studies the sample size was small, and statistical power lim- ited). However, we argue here that it is inappro- priate to use data of this kind to test models relating paternity to paternal care. We discuss the sort of data that can provide a better test. Models of optimal paternal care in response to variation in paternity are based on the effects of trade-offs between current reproductive effort and future reproductive success (Westneat & Sargent 1996). They ask, in effect, what the optimal response in terms of parental effort would be for an individual male responding to a change in paternity. Individuals may lie at different pos- itions along a trade-off between current and future reproduction, and may also experience qualita- tively different trade-offs because of (for ex- ample) differences in resource availability (van Noordwijk & de Jong 1986). As a consequence, one cannot assume that all males within a population will be subject to the same rules relat- ing paternity to paternal effort. We illustrate this argument and its consequences by considering three different cases that are likely to obscure relationships between paternity and paternal care at the population level. These can be summarized as resulting from (1) different optimal levels of care for different individuals, (2) the interdepend- ence of paternal and maternal care and (3) differ- ing trade-offs between paternal care and mating opportunities faced by different individuals. The points that we make here are simple, have general applicability to many problems in behavioural ecology, and similar points have been made many times before in different contexts (for example in discussions of the pros and cons of non- experimental studies of the costs of reproduction: Partridge & Harvey 1985; Lessells 1991). Never- theless, it is our impression that they are not given adequate consideration in most of the studies referred to in Table I. Experimental studies of birds have demon- strated that paternal care may be costly, in terms of reduced survival, breeding or mating oppor- tunities (reviewed in Clutton-Brock 1991). As a consequence, allocation of resources to paternal care is expected to be individually optimized, as with other life-history traits such as clutch size. For example, males in poor condition might reduce their parental care, if the cost of a single ‘unit’ of parental care was greater for males in poor condition. An increasing number of studies have related parentage in natural populations to the relative magnitude of condition-dependent secondary sexual characters (Smith et al. 1991; Hasselquist et al. 1996; Sundberg & Dixon 1996), and there are good reasons to expect such re- lationships to be rather common (Møller 1992). Correspondence: B. C. Sheldon, Department of Zool- ogy, Uppsala University, Villava ¨gen 9 S-752 36 Uppsala, Sweden (email: ben.sheldon@zoologi.uu.se). Bart Kempenaers is at KLIVV, Savoyenstrasse 1a, A-1160 Vienna, Austria. 0003–3472/97/020423+05 $25.00/0/ar960377  1997 The Association for the Study of Animal Behaviour 423