LAUSR

Conspecific animals in the wild use limited resources—such as food, nest site, and mating opportunities—very differently (Shuster and Wade 2003), which has been widely considered a consequence of phenotypic diversity (Shuster 2010). For instance, breeding individuals that carry superior phenotypes generally have higher resource holding potential (RHP) and are therefore more likely to amass more resource and realize higher reproductive success than those with lower RHP (Hurd 2006; Kelly 2008). Due to the dominance of superior phenotypes’ carriers in resource competition, it is quite natural to take for that superior phenotypic traits will expand rapidly in a population. However, such a phenomenon rarely occurs in nature (Shuster 2010), implying that there must be some negative evolutionary forces that antagonize the dominance of superior phenotypes. In many animal species, adverse social or environmental factors (such as antisocial behaviors directed by conspecific competitors at each other) can reduce the reproductive output of breeding individuals via a negative effect on offspring’s survival (Ridley 2007). Therefore, it can be hypothesized that antisocial competitive behaviors may play a key role in preventing superior phenotypic traits from expanding excessively. We tested this hypothesis in the azure-winged magpie Cyanopica cyanus that breeds on the Tibetan Plateau, by investigating their conspecific nest-raiding behaviors (i.e., an individual stealing either eggs or chicks from the nests of other breeding individuals). Limited by the shrub-nesting character of breeding individuals and the cluster distribution of shrubs on the alpine meadow, the Tibetan population of azurewinged magpie was divided into a string of separate colonies. In each colony, breeders constructed their nests in a highly clumped pattern (20–180 nests ha�1; Ren et al. 2016). Under this circumstance, breeding individuals compete over nest site and mating opportunities intensely (Gao et al. 2021) and conspecific nestraiding events occur frequently (Ren et al. 2016). Since conspecific nest-raiding can reduce the reproductive output of victim individuals, if it is directed more at dominant breeding individuals that can amass more resources and produce more offspring than subordinate ones, the hypothesis will be supported that antisocial competitive behaviors act as a reverse evolutionary force of superior phenotypes. Otherwise, it will provide no evidence for this hypothesis in the Tibetan population of azure-winged magpie. Based on 10-year data about the breeding ecology and adult behaviors in the Tibetan azure-winged magpies (material and methods are provided in the Supplementary Data), we addressed: (1) which nests in the Tibetan population were more likely to be raided by conspecific individuals and (2) the effect of conspecific nest-raiding on the reproductive success of raided individuals. In total, we monitored 608 azure-winged magpie nests that successfully fledged at least one offspring. More than one-third of these nests (207 nests) were raided at least once by conspecific individuals (Figure 1A). Approximately 8% (189 of 244) of conspecific nest-raiding events occurred at the first half of the breeding season. Behavioral data extracted from adults’ provisioning videos showed that the majority of conspecific nestraiding (19 of 21 cases) was carried out by later-breeding individuals. The age of raiders (2.216 0.43, n1⁄414) did not differ significantly from that of females (2.6461.08; t1⁄41.31, df1⁄413, P1⁄40.21) and males in the raided nests (2.576 1.02; t1⁄41.10, df1⁄413, P1⁄40.29); however, their nest commencement dates (12.056 4.27 d, n1⁄421 nests) were significantly later than that of raided nests (7.576 3.85 d; t1⁄47.56, df1⁄420, P<0.001). Comparisons in offspring numbers between raided and un-raided nests showed that raided nests produced significantly more eggs at the beginning of reproduction (6.166 1.12, n1⁄4207 nests) than un-raided nests (5.416 1.12, n1⁄4401 nests; t 1⁄4 8.50, df1⁄4606, P<0.001), however, their fledgling number (4.146 1.36, n1⁄4207 nests) was significantly lower than that of unraided nests (4.406 1.39, n1⁄4401 nests; t 1⁄4 2.50, df1⁄4606, P1⁄40.013; Figure 1B). The results of fitting generalized linear mixed models indicated that both the probability and the times of a nest being raided by conspecific individuals were quite closely related to clutch size but not highly correlated to nest commencement date