LAUSR

Long-branch attraction (LBA) is a well-known artifact in phylogenetic reconstruction. Sparse taxon sampling and extreme heterogeneity of evolutionary rates among lineages generate propitious situations for LBA, even defying probabilistic methods of phylogenetic inference. A clear example illustrating LBA challenges is the difficulty of reconstructing the deep gastropod phylogeny, particularly using mitochondrial (mt) genomes. Previous studies consistently obtained unorthodox phylogenetic relationships due to the LBA between the mitogenomes of patellogastropods (true limpets, represented only by Lottia digitalis), heterobranchs, and outgroup taxa. Here, we use the reconstruction of the gastropod mitogenomic phylogeny as a case exercise to test the effect of key methodological approaches proposed to counteract LBA, including the selection of slow-evolving representatives, the use of different outgroups, the application of site-heterogeneous evolutionary models, and the removal of fast-evolving sites. In this regard, we sequenced three new patellogastropod mt genomes, which displayed shorter branches than the one of Lottia as well as gene organizations more similar to that of the hypothetical gastropod ancestor. Phylogenetic analyses incorporating the mt genomes of Patella ferruginea, Patella vulgata, and Cellana radiata allowed eliminating the artificial clustering of Patellogastropoda and Heterobranchia that had prevailed in previous studies. Furthermore, the use of site-heterogeneous models with certain combinations of lineages within the outgroup allowed eliminating also the LBA between Heterobranchia and the outgroup, and recovering Apogastropoda (i.e., Caenogastropoda + Heterobranchia). Hence, for the first time, we were able to obtain a mitogenomic phylogeny of gastropods that is congruent with both morphological and nuclear datasets.