Data from: Effectiveness of phylogenomic data and coalescent species-tree methods for resolving difficult nodes in the phylogeny of advanced snakes (Serpentes: Caenophidia)

Next-generation genomic sequencing promises to quickly and cheaply resolve remaining contentious nodes in the Tree of Life, and facilitates species-tree estimation while taking into account stochastic genealogical discordance among loci. Recent methods for estimating species trees bypass full likelihood-based estimates of the multi-species coalescent, and approximate the true species-tree using simpler summary metrics. These methods converge on the true species-tree with sufficient genomic sampling, even in the anomaly zone. However, no studies have yet evaluated their efficacy on a large-scale phylogenomic dataset, and compared them to previous concatenation strategies. Here, we generate such a dataset for Caenophidian snakes, a group with >2500 species that contains several rapid radiations that were poorly resolved with fewer loci. We generate sequence data for 333 single-copy nuclear loci with ∼100% coverage (∼0% missing data) for 31 major lineages. We estimate phylogenies using neighbor joining, maximum parsimony, maximum likelihood, and three summary species-tree approaches (NJst, STAR, and MP-EST). All methods yield similar resolution and support for most nodes. However, not all methods support monophyly of Caenophidia, with Acrochordidae placed as the sister taxon to Pythonidae in some analyses. Thus, phylogenomic species-tree estimation may occasionally disagree with well-supported relationships from concatenated analyses of small numbers of nuclear or mitochondrial genes, a consideration for future studies. In contrast for at least two diverse, rapid radiations (Lamprophiidae and Colubridae), phylogenomic data and species-tree inference do little to improve resolution and support. Thus, certain nodes may lack strong signal, and larger datasets and more sophisticated analyses may still fail to resolve them.下一代基因组测序（next-generation genomic sequencing）有望快速且低成本地解决生命之树（Tree of Life）中尚存争议的演化分支，并在考量位点间随机基因谱系冲突（stochastic genealogical discordance）的同时辅助物种树估计。近期提出的物种树估计方法绕过了基于全似然的多物种溯祖（multi-species coalescent）的全似然估计，转而通过更为简洁的汇总指标近似真实物种树。这类方法在具备足够基因组采样量的前提下可收敛至真实物种树，即便处于异常区（anomaly zone）。然而目前尚无研究在大规模系统发育基因组数据集（phylogenomic dataset）上评估这些方法的效能，并将其与此前的串联分析（concatenation）策略进行对比。本研究针对新蛇类（Caenophidian snakes）构建了此类数据集：该类群包含超过2500个物种，涵盖了数个仅凭较少位点难以实现有效解析的快速辐射演化类群（rapid radiations）。我们为31个主要演化支生成了333个单拷贝核基因座（single-copy nuclear loci）的序列数据，覆盖度约为100%（缺失数据占比约0%）。随后分别采用邻接法（neighbor joining, NJ）、最大简约法（maximum parsimony, MP）、最大似然法（maximum likelihood, ML）以及三种汇总型物种树方法（summary species-tree approaches，NJst、STAR与MP-EST）开展系统发育树构建。所有方法在多数演化分支上均得到了一致的解析度与支持度。但并非所有方法均支持新蛇类的单系性（monophyly），部分分析结果显示筒蛇科（Acrochordidae）被置于蟒科（Pythonidae）的姊妹群位置。由此可见，系统发育基因组学的物种树估计有时可能与基于少量核基因或线粒体基因的串联分析所得到的强支持关系相悖，这一结论值得未来研究重点关注。相较之下，对于至少两个物种多样且经历快速辐射演化的类群（Lamprophiidae科与游蛇科Colubridae），系统发育基因组数据（phylogenomic data）与物种树推断几乎未提升解析度与支持度。这表明部分演化分支可能缺乏足够的强系统发育信号，即便采用更大规模的数据集与更为复杂的分析方法，仍可能无法实现有效解析。