A Tale of Two Phylogenies: Comparative Analyses of Ecological Interactions
A Tale of Two Phylogenies: Comparative Analyses of Ecological Interactions
复制标题
DOI：
10.1086/674445
复制
发表时间：
2014-02-01
影响因子：
2.9
通讯作者：
Nakagawa, Shinichi
中科院分区：
文献类型：
--
作者：
Hadfield, Jarrod D.;Krasnov, Boris R.;Nakagawa, Shinichi
涉及生态相互作用的特征的进化，如捕食者-猎物、宿主-寄生虫和植物-传粉者的相互作用，可能是由双方的系统发育历史塑造的。我们开发了广义线性混合效应模型（GLMM）来估计双方的系统发育历史对性状进化的影响，既可以单独考虑，也可以考虑两种历史如何相互作用。利用121种哺乳动物中206种跳蚤的发生率和丰度数据，我们证明了我们的方法，并将其与以前用于检测宿主-寄生虫共同进化的方法进行了比较。在大空间尺度上，系统发育互作效应显著，表明亲缘寄生物种更容易在亲缘宿主物种上找到。在较小的空间尺度上，当采样努力不受控制时，系统发育对宿主所寄生寄生虫种类的数量和类型的影响占主导地位。我们继续表明，在这些额外的系统发育效应存在的情况下，以前的方法在测试系统发育相互作用时具有非常高的I型错误率。我们的GLMM方法代表了一种稳健可靠的方法来量化由生态相互作用决定或定义的性状的系统发育效应，并且具有比现有排列测试更容易在更广泛的背景下扩展和解释的优势。
The evolution of traits involved in ecological interactions such as predator-prey, host-parasite, and plant-pollinator interactions, are likely to be shaped by the phylogenetic history of both parties. We develop generalized linear mixed-effects models (GLMM) that estimate the effect of both parties'' phylogenetic history on trait evolution, both in isolation but also in terms of how the two histories interact. Using data on the incidence and abundance of 206 flea species on 121 mammal species, we illustrate our method and compare it to previously used methods for detecting host-parasite coevolution. At large spatial scales we find that the phylogenetic interaction effect was substantial, indicating that related parasite species were more likely to be found on related host species. At smaller spatial scales, and when sampling effort was not controlled for, phylogenetic effects on the number and types of parasite species harbored by hosts were found to dominate. We go on to show that in situations where these additional phylogenetic effects exist, then previous methods have very high Type I error rates when testing for the phylogenetic interaction. Our GLMM method represents a robust and reliable approach to quantify the phylogenetic effects of traits determined by, or defined by, ecological interactions and has the advantage that it can easily be extended and interpreted in a broader context than existing permutation tests.