Data quality and the comparative method: the case of pregnancy failure in rodents

Abstract

In mammalian species where infanticide by males is likely, females exhibit counterstrategies to prevent or mitigate the costs of infanticide. One putative mitigation strategy is the “Bruce effect,” in which pregnant or inseminated females exposed to an unfamiliar male experience pregnancy block or failure. Females then mate with the new male, thus shifting investment from a “doomed” pregnancy to a more fruitful one. However, the Bruce effect may be an adaptive response to other factors besides infanticide. For example, if paternal care is necessary for offspring survival, and an unfamiliar male replacing the original mate is unlikely to provide such care to offspring of a litter it did not sire, then a female may terminate a pregnancy to initiate a new one. The infanticide and paternal care hypotheses have not been rigorously tested because comparative data on the Bruce effect across mammals are scarce. We compiled data on the Bruce effect, infanticide, and paternal care from one particularly rich source of information, rodents, but found the data set to be less rich than expected. The Bruce effect, infanticide, and paternal care were common among rodent species, but we found no clear relationship among the traits. However, this was likely due to 1) a bias toward positive results, 2) missing data, and 3) a reliance on studies of captive animals. These are common problems in comparative research, and we outline standards that should be implemented to successfully answer questions of importance in the field.