Samia Diogo S M, Blumstein Daniel T
Departamento de Ecologia, Universidade Federal de Goiás, Goiânia, Goiás, Brazil.
Department of Ecology and Evolutionary Biology, University of California Los Angeles, Los Angeles, California, United States of America.
PLoS One. 2014 Nov 18;9(11):e113134. doi: 10.1371/journal.pone.0113134. eCollection 2014.
Optimal escape theory states that animals should counterbalance the costs and benefits of flight when escaping from a potential predator. However, in apparent contradiction with this well-established optimality model, birds and mammals generally initiate escape soon after beginning to monitor an approaching threat, a phenomena codified as the "Flush Early and Avoid the Rush" (FEAR) hypothesis. Typically, the FEAR hypothesis is tested using correlational statistics and is supported when there is a strong relationship between the distance at which an individual first responds behaviorally to an approaching predator (alert distance, AD), and its flight initiation distance (the distance at which it flees the approaching predator, FID). However, such correlational statistics are both inadequate to analyze relationships constrained by an envelope (such as that in the AD-FID relationship) and are sensitive to outliers with high leverage, which can lead one to erroneous conclusions. To overcome these statistical concerns we develop the phi index (Φ), a distribution-free metric to evaluate the goodness of fit of a 1:1 relationship in a constraint envelope (the prediction of the FEAR hypothesis). Using both simulation and empirical data, we conclude that Φ is superior to traditional correlational analyses because it explicitly tests the FEAR prediction, is robust to outliers, and it controls for the disproportionate influence of observations from large predictor values (caused by the constrained envelope in AD-FID relationship). Importantly, by analyzing the empirical data we corroborate the strong effect that alertness has on flight as stated by the FEAR hypothesis.
最优逃逸理论指出，动物在逃离潜在捕食者时应权衡逃跑的成本和收益。然而，与这一已确立的最优性模型明显矛盾的是，鸟类和哺乳动物通常在开始监测到接近的威胁后不久就开始逃跑，这一现象被归纳为“早逃离，避匆忙”（FEAR）假说。通常，FEAR假说通过相关统计进行检验，当个体首次对接近的捕食者做出行为反应的距离（警戒距离，AD）与其飞行起始距离（它逃离接近捕食者的距离，FID）之间存在强关系时，该假说得到支持。然而，这种相关统计既不足以分析受包络线约束的关系（如AD - FID关系中的情况），又对具有高杠杆作用的异常值敏感，这可能导致得出错误结论。为了克服这些统计问题，我们开发了φ指数（Φ），这是一种无分布度量，用于评估约束包络线中1:1关系的拟合优度（FEAR假说的预测）。通过使用模拟数据和实证数据，我们得出结论，Φ优于传统的相关分析，因为它明确检验了FEAR预测，对异常值具有鲁棒性，并且它控制了来自大预测值的观测值的不成比例影响（由AD - FID关系中的约束包络线引起）。重要的是，通过分析实证数据，我们证实了FEAR假说中所述的警觉性对飞行的强烈影响。
