Computational protocol: Ecological opportunity may facilitate diversification in Palearctic freshwater organisms: a case study on hydrobiid gastropods

Similar protocols

Protocol publication

[…] Bi-directional sequences were aligned in BIOEDIT v.7.5.3 [] and compiled to gene-specific datasets. Alignments of the 16S and 28S fragments were conducted using the MAFFT multiple alignment program [] with default settings for gap penalties (Gap opening penalty (GOP) = 1.53). Sequences of the protein-coding COI gene were unambiguously aligned in BIOEDIT.The COI, 16S, and 28S data partitions were analyzed using Bayesian and coalescent methods, with and without outgroups, respectively (Additional file : Table S1). Prior to the phylogenetic analyses, we identified gene-specific nucleotide substitution models in JMODELTEST v.2.1.4 [] under the corrected Akaike’s information criterion (AICc; [–]). The selected models were HKY [] +I (invariable sites) +G (rate variation among sites) for the COI partition, K80 [] +G for the 16S partition, and TIM3 [] +I +G for the 28S partition in the presence of outgroups. Models TPM3uf [] +I +G, K80 +G, and TrN [] +I +G were selected for the COI, 16S, and 28S partitions, respectively, in the absence of outgroups.To infer species relationships and assignments, we first conducted different Bayesian Inference (BI) analyses with the individual and concatenated datasets (including outgroups) in MRBAYES v.3.1.2 [, ] and the best-fit nucleotide substitution models through 2 independent runs of 4 Metropolis-coupled chains with 5 million generations each and a sampling frequency of 1000. After ensuring stationary of the chains (i.e., standard deviation of split frequencies below 0.01), we discarded the initial 10% of the trees as burn-in. The majority-rule consensus tree obtained in the combined analysis is depicted in Additional file .We then generated an ultrametric species tree without outgroups in the program *BEAST [] with the best-fit nucleotide substitution models inferred above. As these models are not available in the *BEAST interface BEAUti v.1.8.0 by default [], they were specified manually. Because of the lack of a robust fossil record in the subfamily Pseudamnicolinae, the subsequent phylogenetic analysis was done using relative divergence time (i.e., all branches evolving with a rate of 1 substitution per site per unit of time). A birth-death model [] was selected as prior topology, which is appropriate for species-level phylogenies. We ran 75 million generations, sampling every 2,000th tree. Resulting log files were checked in TRACER v.1.6 [] in order to ensure that the posterior distribution of the parameters reached stationary (effective sample size, ESSs, above 200). The final species tree (i.e., the maximum clade-credibility tree, MCC tree) was identified in TreeAnnotator v.1.8.0, with the initial 10% of the topologies discarded as burn-in and displayed using FigTree v.1.3.1 []. [...] As explorative analysis, we first visualized the overall trend of lineage diversification in each genus by a lineage-through-time (LTT) plot of the MCC tree. We included the effect of phylogenetic uncertainty by calculating the 95% confidence interval of the LTT based on 1000 random post-burn-in trees of the *BEAST posterior distribution using the phytools 0.4-56 package [] for the R 3.2 statistical environment []. Although these plots are frequently utilized to show general diversification trends, they do not inform about differences in evolutionary rates []. Therefore, a Bayesian analysis of Macroevolutionary Mixtures (BAMM; []) was performed utilizing the program BAMM v.2.5.0 [], which models speciation and extinction trends over time. Because of the relatively low number of species and inherent limitations in inferring extinction rates from phylogenies [], we here tested only for differences in speciation rates between clades. We accounted for an incomplete taxon sampling, which may bias rate inference [], by including an analytical correction ratio of 0.4 for the genus Pseudamnicola. We sampled every 1000th out of 1 million BAMM generations, excluded the first 10% as burn-in, and inferred lineage specific speciation rates by averaging over the remaining posterior distribution of 900 generations. As suggested by Shi and Rabosky [], the Bayes factor serves as measure of evidence for the number of rate shifts and integrates phylogenetic uncertainty. Accordingly, we subjected 1000 random post-burn-in trees of the *BEAST posterior to individual BAMM analysis, combined the BAMM posterior distributions into one pseudo-posterior, and calculated the Bayes factor to investigate whether there is statistical evidence for rate shifts among clades.Recently, Moore et al. [] indicated that the BAMM approach cannot detect rate shifts in completely extinct clades and is overly prior-sensitive (but see Rabosky et al. [] for a reply). However, we did not infer extinction rates. Moreover, our analyses of the influence of elevation and environmental temperature on speciation rates confirmed the BAMM results (see below). […]

Pipeline specifications

Software tools BioEdit, MAFFT, jModelTest, MrBayes, BEAST, FigTree, Phytools
Application Phylogenetics
Diseases Pulmonary Fibrosis, Intracranial Hypertension