Grabowski Mark
Research Centre for Evolutionary Anthropology and Palaeocology, School of Biological and Environmental Sciences, Liverpool John Moores University, James Parson Building, 3 Byrom Street, Liverpool L3 3AF, UK.
Department of Biosciences, CEES, University of Oslo, Blinderen, PB 1066, 0316 Oslo, Norway.
Syst Biol. 2024 Nov 29;73(6):1038-1050. doi: 10.1093/sysbio/syae044.
Relationships among species in the tree of life can complicate comparative methods and testing adaptive hypotheses. Models based on the Ornstein-Uhlenbeck process permit hypotheses about adaptation to be tested by allowing traits to either evolve toward fixed adaptive optima (e.g., regimes or niches) or track continuously changing optima that can be influenced by other traits. These models allow estimation of the effects of both adaptation and phylogenetic inertia-resistance to adaptation due to any source-on trait evolution, an approach known as the "adaptation-inertia" framework. However, previous applications of this framework, and most approaches suggested to deal with the issue of species non-independence, are based on a maximum likelihood approach, and thus it is difficult to include information based on prior biological knowledge in the analysis, which can affect resulting inferences. Here, I present Blouch, (Bayesian Linear Ornstein-Uhlenbeck Models for Comparative Hypotheses), which fits allometric and adaptive models of continuous trait evolution in a Bayesian framework based on fixed or continuous predictors and incorporates measurement error. I first briefly discuss the models implemented in Blouch, and then the new applications for these models provided by a Bayesian framework. This includes the advantages of assigning biologically meaningful priors when compared to non-Bayesian approaches, allowing for varying effects (intercepts and slopes), and multilevel modeling. Validations on simulated data show good performance in recovering the true evolutionary parameters for all models. To demonstrate the workflow of Blouch on an empirical dataset, I test the hypothesis that the relatively larger antlers of larger-bodied deer are the result of more intense sexual selection that comes along with their tendency to live in larger breeding groups. While results show that larger-bodied deer that live in larger breeding groups have relatively larger antlers, deer living in the smallest groups appear to have a different and steeper scaling pattern of antler size to body size than other groups. These results are contrary to previous findings and may argue that a different type of sexual selection or other selective pressures govern optimum antler size in the smallest breeding groups.
生命之树中物种间的关系会使比较方法和检验适应性假说变得复杂。基于奥恩斯坦 - 乌伦贝克过程的模型允许通过使性状朝着固定的适应性最优值（例如，生态位或小生境）进化或追踪可受其他性状影响的不断变化的最优值来检验关于适应性的假说。这些模型能够估计适应性和系统发育惯性（由于任何来源导致的对适应性的抗性）对性状进化的影响，这种方法被称为“适应性 - 惯性”框架。然而，该框架先前的应用以及大多数为处理物种非独立性问题而提出的方法都基于最大似然法，因此在分析中难以纳入基于先前生物学知识的信息，这可能会影响最终的推断。在此，我介绍了Blouch（用于比较假说的贝叶斯线性奥恩斯坦 - 乌伦贝克模型），它在基于固定或连续预测变量的贝叶斯框架中拟合连续性状进化的异速生长和适应性模型，并纳入了测量误差。我首先简要讨论Blouch中实现的模型，然后介绍贝叶斯框架为这些模型提供的新应用。这包括与非贝叶斯方法相比赋予生物学上有意义的先验的优势、允许不同的效应（截距和斜率）以及多水平建模。对模拟数据的验证表明，所有模型在恢复真实进化参数方面表现良好。为了在一个实证数据集上展示Blouch的工作流程，我检验了这样一个假说：体型较大的鹿的鹿角相对较大是由于它们倾向于生活在较大的繁殖群体中而伴随的更强烈的性选择的结果。虽然结果表明生活在较大繁殖群体中的体型较大的鹿有相对较大的鹿角，但生活在最小群体中的鹿似乎与其他群体相比，鹿角大小与体型的缩放模式不同且更陡峭。这些结果与先前的发现相反，可能表明在最小的繁殖群体中，不同类型的性选择或其他选择压力决定了最优的鹿角大小。
