The relative importance of the different processes that determine the distribution of species and the assembly of communities is a key question in ecology. The distribution of any individual species is affected by a wide range of environmental variables as well as through interactions with other species; the resulting distributions determine the pool of species available to form local communities at fine spatial scales. A challenge in community ecology is that these interactions (e.g. competition, facilitation, etc.) often are not directly measurable. Here, we used hierarchical modelling of species communities (HMSC), a recently developed framework for joint species distribution modelling, to estimate the role of biotic effects alongside environmental factors using latent variables. We investigate the role of these factors determining species distributions in communities of Artiodactyla, Perissodactyla and Proboscidea in the Afrotropics, an area of peak species richness for hoofed mammals. We also calculate pairwise trait dissimilarity between these species, from a mixture of morphological and behavioural traits, and investigate the relationship between dissimilarity and estimated residual co-occurrence in the model. We find that while ungulate distributions appear to be predominantly determined (~ 70%) by climatic variables, such as precipitation, a substantial proportion of the variance in ungulate species distributions (~ 30%) can also be attributed to modelled latent variables that likely represent a combination of dispersal barriers and biotic factors. Although we find only a weak relationship between residual co-occurrence and trait dissimilarity, we suggest that our results may show evidence that biotic factors, likely influenced by historical barriers to species dispersal, are important in determining species communities over a continental area. The HMSC framework can be used to provide insight into factors affecting community assembly at broad scales, and to make more powerful predictions about future species distributions as we enter an era of increasing impacts from anthropogenic change.

中文翻译：

---

使用联合物种分布模型来确定非洲热带有蹄类动物分布的气候和非气候驱动因素


决定物种分布和群落组装的不同过程的相对重要性是生态学中的一个关键问题。任何单个物种的分布都受到广泛的环境变量以及与其他物种的相互作用的影响;由此产生的分布决定了在精细空间尺度上形成局部群落的物种库。社区生态学的一个挑战是这些互动（例如竞争、促进等）通常无法直接测量。在这里，我们使用了物种群落分层模型 （HMSC），这是一个最近开发的联合物种分布模型框架，使用潜在变量来估计生物效应与环境因素的作用。我们研究了这些因素决定物种分布在非洲热带地区的偶蹄目、Perissodactyla 和长鼻群落中的作用，这是有蹄哺乳动物物种丰富度的峰值地区。我们还根据形态学和行为性状的混合物计算这些物种之间的成对性状相异性，并研究相异性与模型中估计的残差共现之间的关系。我们发现，虽然有蹄类动物的分布似乎主要由气候变量（如降水）决定 （~ 70%），但有蹄类动物物种分布的很大一部分方差 （~ 30%） 也可以归因于建模的潜在变量，这些变量可能代表了传播屏障和生物因素的组合。 尽管我们只发现残差共现和性状相异性之间存在微弱的关系，但我们认为我们的结果可能表明，可能受到物种传播历史障碍影响的生物因素在决定大陆地区的物种群落方面很重要。HMSC 框架可用于深入了解在大范围内影响群落组装的因素，并在我们进入人为变化影响日益严重的时代时对未来物种分布做出更有力的预测。