Batesian mimics benefit from their phenotypic similarity to their models. An apparent paradox is the presence of many inaccurate mimics in nature. Many hypotheses have been proposed to explain their existence, including the relaxed selection hypothesis, which postulates that the level of mimetic resemblance avoided by predators is context-dependent. More specifically, predators avoid mimics of lower resemblance that imitate highly noxious models due to the high cost of mistakenly attacking such models. We tested this hypothesis using data on 93 pairs of ant-mimicking spider species and their models. We measured the mimetic resemblance in each spider to its putative model. Then, we estimated the noxiousness of ant models on the basis of their defensive arsenal (i.e. mandibles, spines, sting, and chemical defense). In support of the relaxed selection hypothesis, we found a negative correlation between model noxiousness and mimetic accuracy. The proportion of the variability in mimetic accuracy explained by model noxiousness was not high, indicating that other factors influence the evolution of mimetic accuracy. Finally, this comparative analysis provides evidence of the influence of model noxiousness on mimetic accuracy in a Batesian mimicry system.

贝氏拟态受益于它们与其模型的表型相似性。一个明显的悖论是自然界中存在许多不准确的模拟物。已经提出了许多假设来解释它们的存在，包括宽松选择假设，该假设假设捕食者避免的模仿相似程度取决于上下文。更具体地说，由于错误地攻击此类模型的成本很高，捕食者会避免模仿高度有害的模型的相似度较低的模型。我们使用 93 对模仿蚂蚁的蜘蛛物种及其模型的数据测试了这一假设。我们测量了每只蜘蛛与其假定模型的模拟相似性。然后，我们根据蚂蚁模型的防御武器（即下颌骨、刺、刺和化学防御）估计了它们的毒性。为了支持宽松选择假设，我们发现模型有害性和模拟准确性之间存在负相关。模型毒性解释的拟态精度变异性比例不高，表明其他因素影响拟态精度的演变。最后，这种比较分析提供了模型有害性对贝氏拟态系统中拟态准确性影响的证据。