Department of Biology and M.L. Bean Museum, Brigham Young University, Provo, UT 84602, USA.
Mol Phylogenet Evol. 2013 May;67(2):494-508. doi: 10.1016/j.ympev.2013.02.019. Epub 2013 Feb 27.
One of the main challenges in analyzing multi-locus phylogenomic data is to find an optimal data partitioning strategy to account for variable evolutionary histories of different loci for any given dataset. Although a number of studies have addressed the issue of data partitioning in a Bayesian phylogenetic framework, such studies in a maximum likelihood framework are comparatively lacking. Furthermore, a rigorous statistical exploration of possible data partitioning schemes has not been applied to mitochondrial genome (mtgenome) data, which provide a complex, but manageable platform for addressing various challenges in analyzing phylogenomic data. In this study, we investigate the issue of data partitioning in the maximum likelihood framework in the context of the mitochondrial phylogenomics of an orthopteran superfamily Acridoidea (Orthoptera: Caelifera). The present study analyzes 34 terminals representing all 8 superfamilies within Caelifera, which includes newly sequenced partial or complete mtgenomes for 11 families. Using a new partition-selection method implemented in the software PartitionFinder, we compare a large number of data partitioning schemes in an attempt to identify the most effective method of analyzing the mtgenome data. We find that the best-fit partitioning scheme selected by PartitionFinder is superior to any a priori schemes commonly utilized in mitochondrial phylogenomics. We also show that over-partitioning is often detrimental to phylogenetic reconstruction. A comparative analysis of mtgenome structures finds that the tRNA gene rearrangement between cytochrome c oxidase subunit II and ATP synthase protein 8 does not occur in the most basal caeliferan lineage Tridactyloidea, suggesting that this gene rearrangement must have evolved at least in the common ancestor of Tetrigoidea and Acridomorpha. We find that mtgenome data contain sufficient phylogenetic information to broadly resolve the relationships across Acridomorpha and Acridoidea.
多基因座系统发育分析的主要挑战之一是为给定数据集的不同基因座找到最佳的数据分区策略，以解释其不同的进化历史。尽管许多研究已经在贝叶斯系统发育框架中解决了数据分区问题，但在最大似然框架中，这种研究相对较少。此外，尚未对线粒体基因组（mtgenome）数据应用严格的统计探索可能的数据分区方案，该方案为解决分析基因组数据中的各种挑战提供了一个复杂但可管理的平台。在本研究中，我们在直翅目（Orthoptera：Caelifera）线粒体系统发育学的背景下，在最大似然框架中研究了数据分区问题。本研究分析了代表直翅目 8 个超科的 34 个末端，其中包括 11 个科的新测序部分或完整 mtgenome。使用软件 PartitionFinder 中实现的新分区选择方法，我们比较了大量数据分区方案，试图确定分析 mtgenome 数据的最有效方法。我们发现，PartitionFinder 选择的最佳分区方案优于线粒体系统发育学中常用的任何先验方案。我们还表明，过度分区通常对系统发育重建有害。对 mtgenome 结构的比较分析发现，细胞色素 c 氧化酶亚基 II 和 ATP 合酶蛋白 8 之间的 tRNA 基因重排在最基础的直翅目（Tridactyloidea）中没有发生，这表明该基因重排至少在 Tetrigoidea 和 Acridomorpha 的共同祖先中发生了。我们发现 mtgenome 数据包含足够的系统发育信息，可以广泛解决 Acridomorpha 和 Acridoidea 之间的关系。
