Computational protocol: The rise of the Himalaya enforced the diversification of SE Asian ferns by altering the monsoon regimes

Similar protocols

Protocol publication

[…] We assembled sequences of four chloroplast genome regions, rbcL, rbcL-atpB intergenic spacer sequences (IGS), rps4 plus rps4-trnS IGS, and trnL-F region including the trnL intron and trnL-trnF IGS using sequences generated in previous studies [,,]. A few additional sequences of expanded sampled species (L. nudus Ching, L. nylamensis Ching et S. K. Wu, L. perrierianus Ching, L. schraderi (Mett.) Faden, L. rotundus Ching and L. vesiculari-peleaceus (Hieron.) Pic. Serm.) were obtained using the protocols published in these studies. In total, the dataset included 65 out of the ca. 70 currently known species of Lepisorus plus 11 species of other lepisoroid fern genera (See Additional file ). The taxonomy follows the most recent treatments [,,-,]. The sampling was designed to cover more than 90% of the species diversity of Lepisorus and represent its geographic range without any obvious bias. All alignment were generated manually in Maclade 4.08 [] and ambiguous regions excluded from further analyses.Initially, the dataset was explored by carefully designed phylogenetic analyses using PAUP 4.0 [] for maximum parsimony, PHYML 3.0 [] for maximum likelihood and MrBayes 3.1.2 [] for Bayesian Inference of phylogeny as described in Wang et al. []. The model of sequence evolution was determined using jModelTest [] and a likelihood ratio test (LHR) was carried out to test for the presence of a molecular clock. No evidence for topological heterogeneity was discovered and all subsequent analyses were carried out with a combined dataset. [...] Bayesian approaches with uncorrelated relaxed clock model were employed to estimate the divergence times of the lepisoroid ferns with focus on the genus Lepisorus. The LHR test results supported the use of the relaxed clock model. All divergence time estimates were carried out using BEAST 1.6.2 ( http://beast/ []. The model selected with jModelTest was implemented, but parameters were estimated simultaneously with the BEAST analyses. Several BEAST analyses were carried out and their results were summarized utilizing TRACER 1.5 ( and TREEANNOTATOR, part of the BEAST package. Markov chains were run for 10,000,000 generations with every 10,000 generations sampled and at least a 10% burn-in phase. The final analyses were performed with a relaxed molecular clock [], birth-death model, and calibration of the split of Paragramma and the remaining lepisoroid ferns [] with a lognormal distribution model with a shift of 19.6 ma. This node age was obtained from divergence time estimate carried out with a comprehensive sampling throughout the tree of ferns []. Unfortunately, this study did not provide confidence intervals for nodes age estimates. The obtained age estimates were consistent with divergence time estimates of Polypodiaceae and the limited fossil record of the Polypodiaceae [,,] (see also discussion). As a further confirmation, we calculated the nucleotide substitution rates for several clades and compared them with previously reported substitution rates of the chloroplast genome [,]. The rates were within the expected range. Shown chronograms were calculated using median clade credibility tree plus 95% confidence intervals. [...] To investigate patterns of diversification through time, we used five approaches. First, we explored evidence for non-constant diversification rate in Lepisorus using the Constant Rate (CR) birth death model test [] as implemented in several programs including Gammastatistics V.10 ( and in R applications ( APE [], GEIGER [] and LASER []. A MCCR test using simulations of phylogenies was employed to account for the impact of missing species [,]. Second, we estimated the cumulative density distribution of branch length []. Length of the branches were detached from the calibrated consensus tree and divided into ten length classes. Frequency number of each class was calculated and plotted as cumulative frequency against branch length classes. Third, the diversification rates were visualized by generating multiple lineage-through-time plots (MLTTP) for 50 randomly selected trees from the tree pool output from BEAST using the R packages APE and GEIGER, plus lineage-through-time plots (LTTP) for the consensus chronogram with a mean node age. Fourth, we used Rabosky’s LASER software [] in R to determine whether the observed pattern of diversification through time fits best to a simple model of a constant diversification rate or to complex models with variable diversification rates. The best fitting rate model, maximum shift points and the threshold of shift points were estimated according to the maximum log-likelihood values using Akaike Information Criterion (AIC) weights and Delta-AIC scores []. Fifth, we visualized changes of the diversification rate by plotting the difference of number of newly appearing species (Δn) against the setting time intervals (Δt) as 0.9 Ma. The value of Δn was then transferred into changes of diversification rate per time interval by considering the number of species at the beginning of the interval. The limitations of the outgroup sampling were considered by exploring the impact of different sampling densities. However, the sister lineage of Lepisorus is considerable less species-rich []. All the results were compared to the suggested periods of changes in the monsoon regimes and major geological events [-,]. […]

Pipeline specifications

Software tools PAUP*, PhyML, MrBayes, jModelTest, BEAST, APE, GEIGER
Application Phylogenetics