Abstract
► We report a rarely achieved male-removal experiment to test alternative theories for multiple-stallion harem defence in horses (Equus caballus). ► The removal of subordinate stallions from feral horse harems did not result in loss of mares. ► Harem defence and harassment by dominant stallions decreased, not increased, when the subordinate was absent. ► Subordinates are more likely to be reproductive parasites, than partners in an alliance with the dominant stallion.
Relationships between males defending the same harem are described as cooperative or competitive and explained by mutualism, reciprocal altruism, and reproductive concessions or limited control between unequal contestants. These alternate hypotheses can be tested by removing males from harems. Some feral horse (Equus caballus) harems are defended by more than one and up to five stallions. We temporarily removed the subordinate stallion from two of six multi-stallion bands (a harem and its stallions) for three weeks during the breeding season. We monitored harems for changes in composition, and measured and compared rates of (i) intra- and inter-band stallion aggression, (ii) stallion aggression towards mares, and (iii) stallion-mare proximity before subordinate stallion removal and after his return with rates during his absence. Harems were successfully defended during the subordinate's absence and stallion-mare aggression was substantially reduced. Dominant stallions did not require assistance in harem defence, and heightened harassment of mares is directly attributable to subordinate stallion's residence, not characteristics of the dominant stallion or mares. Cooperative hypotheses do not appear to explain multi-stallion bands but the experiments in this study should be replicated further. The limited control hypothesis (e.g., mate-parasitism) appears better supported but we outline its limitations too.