Data from: Taxonomy and phylogeny can yield comparable results in comparative palaeontological analyses

Many extinct taxa with extensive fossil records and mature taxonomic classifications have not yet been the subject of formal phylogenetic analysis. Here we test whether the taxonomies available for such groups represent useful (i.e., non-misleading) substitutes for trees derived from matrix-based phylogenetic analyses. We collected data for 52 animal clades that included fossil representatives, and for which a recent cladogram and pre-cladistic taxonomy were available. We quantified the difference between the time-scaled phylogenies implied by taxonomies and cladograms using the matching cluster distance metric. We simulated phenotypic trait values and used them to estimate a series of commonly used, phylogenetically explicit measures (phylogenetic signal [Blomberg's K], phylogenetic generalised least squares [PGLS], mode of evolution [Brownian versus Ornstein-Uhlenbeck] and phylogenetic clustering of extinction [Fritz and Purvis' D]) in order to determine the degree to which they co-varied on taxonomic and cladistic trees. With respect to topology taxonomies are good approximations of the underlying evolutionary relationships as recorded in inferred cladograms. Detection of phylogenetic clustering of extinction could not be properly assessed. For all other evolutionary analyses, results from taxonomy-based phylogenies co-varied with those from cladogram-based phylogenies, but individual comparisons could be misleading. The relative length of terminal branches (influenced by stratigraphy and sampling rate) is a key control on the shared information between, and therefore the relative performance of, taxonomy and cladogram-based phylogenies. Collectively these results suggest that under particular circumstances and after careful consideration some taxonomies, or composite trees that incorporate taxonomic information, could be used in place of a formal analytical solution, but workers must be cautious. This opens certain parts of a previously inaccessible section of the fossil record to interrogation within an explicit comparative framework, which will help to test many classical macroevolutionary hypotheses formulated for groups that currently lack formal phylogenetic estimates.

诸多拥有丰富化石记录与成熟分类学界定的已灭绝分类群，尚未接受正式的系统发育分析。本研究旨在检验：针对此类类群的现有分类体系，能否作为基于矩阵的系统发育分析所构建系统发育树的有效（即无误导性）替代方案。我们收集了52个包含化石代表类群的动物演化支的数据，这些类群均有最新的支序图以及支序分类前的分类体系可供使用。我们采用匹配聚类距离度量，量化了分类体系与支序图所暗示的时间标度系统发育树之间的差异。我们模拟了表型性状值，并利用其估算了一系列常用的系统发育显性分析指标：包括系统发育信号（phylogenetic signal，Blomberg's K）、系统发育广义最小二乘法（phylogenetic generalised least squares, PGLS）、演化模式（布朗运动与奥恩斯坦-乌伦贝克模型）以及灭绝事件的系统发育聚类（Fritz and Purvis' D），以此判断这些指标在分类体系构建的系统发育树与支序图构建的系统发育树上的协变程度。就拓扑结构而言，分类体系能够较好地近似支序图所推断的深层演化关系。但灭绝事件的系统发育聚类无法得到有效评估。其余所有演化分析中，基于分类体系的系统发育树所得结果与基于支序图的系统发育树结果呈协变关系，但单次比较仍可能存在误导性。终端分支的相对长度（受地层学与采样速率影响）是影响分类体系与基于支序图的系统发育树间共享信息的关键调控因素，进而决定二者的相对表现。综合来看，本研究结果表明：在特定条件下并经过审慎考量后，部分分类体系或整合了分类学信息的复合系统发育树，可替代正式的系统发育分析结果，但研究者需保持谨慎。此举将此前难以触及的部分化石记录，纳入了可开展明确比较研究的分析框架之中，有助于检验诸多针对当前尚无正式系统发育估算的类群所提出的经典宏观演化假说。