In most species, females invest more in their offspring than males. In mammals, this also holds true after the birth of young since only mothers can lactate and fathers are thus more likely to desert and seek additional mating opportunities. These opportunities are largely dependent on resource distribution and the resulting distribution of females. The emerging social and mating systems of different species give rise to different benefits of investment in paternal care. Paternal care should have evolved in relation to the frequency of losing paternities to other males. Hence, in species living in one-male systems with little opportunity for female extra-pair copulations, paternal care should be more common than in those species living in multi-male groups. The guinea pigs (Caviinae) comprise closely related species exhibiting different social organizations and mating systems. The amount of paternal behaviour and offspring-directed aggression was investigated in three species kept under laboratory conditions: the polygynous Cavia aperea, the promiscuous Galea musteloides and a newly described monogamous species (Galea monasteriensis sp. nov.). All animals were kept in groups of one adult male and one adult female with their non-weaned offspring. Cavia aperea as well as G. monasteriensis males showed a high incidence of paternal behaviour, which consisted primarily of playing with offspring and grooming them. In contrast, G. musteloides males never engaged in paternal behaviour but directed more aggressive behaviour towards their offspring. Mating with multiple partners and multiple paternities are common in G. musteloides whereas in the former two species there is a much higher certainty of paternity. We conclude that paternal behaviour in guinea pigs probably co-evolved in alignment with the certainty of paternity that males of a species encounter under natural conditions.

Several hypotheses about changes in adult survival of European storm-petrels Hydrobates pelagicus and predation by yellow-legged gulls Larus michahellis at Benidorm Island (western Mediterranean) during 1993–2003 were assessed. Two approaches were used: analysis of pellets containing remains of predated petrels from two caves, and multistate capture–recapture models using recaptures of breeders together with recoveries only from one of these caves. These models were used to estimate separately adult storm-petrel survival and a minimum probability of being killed by gulls. At least 11–14% of the storm-petrels killed had been ringed as breeders. Local adult survival varied greatly among years, from 0.701 (SE=0.095) to extremely high values (0.951, SE=0.031) for such a small species (0.833 on average for the whole period). Nevertheless, it was not possible to assess whether this variation was influenced by inter-annual differences in gull predation. The peak of predation occurred in June, when maximum numbers of prospecting petrels visit the colonies. Predation was much higher at the cave that received stronger illumination from Benidorm city. Minimum predation probabilities estimated from capture–recapture varied with time from 0.022 (SE=0.011) to 0.040 (SE=0.016), and increased significantly after the powering of light received from the city in 1997. In these years, predation by yellow-legged gulls accounted for a large percentage of mortality of storm-petrels at Benidorm Island (up to 33%). Both methodological approaches confirmed that predation was carried out by specialist gulls, and that neither the size of the gull colony nor its food availability influenced predation on petrels. Management might reduce predation rate by removing gulls that nest in the caves, since these individuals tend to kill most storm-petrels.