Data from: The species versus subspecies conundrum: quantitative delimitation from integrating multiple data types within a single Bayesian approach in Hercules beetles

With the recent attention and focus on quantitative methods for species delimitation, an overlooked but equally important issue regards what has actually been delimited. This study investigates the apparent arbitrariness of some taxonomic distinctions, and in particular how species and subspecies are assigned. Specifically, we use a recently developed Bayesian model-based approach to show that in the Hercules beetles (genus Dynastes) there is no statistical difference in the probability that putative taxa represent different species, irrespective of whether they were given species or subspecies designations. By considering multiple data types, as opposed to relying exclusively on genetic data alone, we also show that both previously recognized species and subspecies represent a variety of points along the speciation spectrum (i.e., previously recognized species are not systematically further along the continuum than subspecies). For example, based on evolutionary models of divergence, some taxa are statistically distinguishable on more than one axis of differentiation (e.g., along both phenotypic and genetic dimensions), whereas other taxa can only be delimited statistically from a single data type. Because both phenotypic and genetic data are analyzed in a common Bayesian framework, our study provides a framework for investigating whether disagreements in species boundaries among data types reflect (i) actual discordance with the actual history of lineage splitting, or instead (ii) differences among data types in the amount of time required for differentiation to become apparent among the delimited taxa. We discuss what the answers to these questions imply about what characters are used to delimit species, as well as the diverse processes involved in the origin and maintenance of species boundaries. With this in mind, we then reflect more generally on how quantitative methods for species delimitation are used to assign taxonomic status.

随着近期物种界定（species delimitation）定量方法受到广泛关注与深入研究，一个常被忽视却同等重要的问题在于：我们实际所界定的究竟是什么。本研究探讨了部分分类学区分的看似任意性，尤其是物种与亚种的划定方式。具体而言，我们借助新近开发的基于贝叶斯模型的分析方法，证实了在犀金龟属（Dynastes）中，无论类群被赋予物种还是亚种的分类地位，其作为不同物种的概率均无统计学差异。相较于仅依赖遗传数据，我们纳入多类数据开展分析，结果显示：已被识别的物种与亚种均代表了物种形成光谱上的不同节点——即已被识别的物种并未系统性地比亚种更靠近物种形成连续体的末端。例如，基于分化演化模型，部分类群可在多个分化维度上实现统计学区分（如同时结合表型与遗传维度），而另一些类群仅能通过单一数据类型完成统计学界定。由于我们将表型与遗传数据置于统一的贝叶斯框架下进行分析，本研究为探讨两类核心问题提供了分析范式：一是不同数据类型间的物种边界分歧是否反映了（i）与谱系分化实际历史不符的真实冲突，抑或是（ii）不同数据类型在使界定类群的分化特征显现所需的时间上存在差异。我们讨论了这些问题的答案对于划定物种所依据的性状选择，以及参与物种边界形成与维持的多样过程有何启示。基于此，我们进一步从更宏观的层面反思：物种界定的定量方法应如何用于划定类群的分类学地位。