The Himalayan uplift and evolution of aquatic biodiversity across Asia: Snowtrout (Cyprininae: Schizothorax) as a test case

to improve. Support for the resulting tree, inferred using the selected model, was explored using 1,000,000 ultrafast bootstrap approximations (UFBOOT2; [43]). We also assessed branch support using the SH-ALRT test (Shimodaira-Hasegawa approximate Likelihood Ratio Test, based on a null distribution derived from bootstrap replicates of varying sizes). Values �80% indicated strong support for clades, whereas those for UFBOOT were �95% (IQ-TREE v.2 Man- ual: http://www.iqtree.org). To employ a Bayesian approach, we first derived an input XML file using a relaxed molecu- lar clock with an uncorrelated lognormal rate distribution and a Yule process prior (graphical user interface (i.e., BEAUTY) within the phylogenetic software suite BEAST (v2.61; [44]). Each analysis was run for 100 million generations, with samples taken every 1,000th iteration, fol- lowing a burn-in of 20%. Effective sample size (ESS) was verified as >200 across all parameters with MCMC chain convergence visually inspected using TRACER [45]. Trees from three inde- pendent runs were then combined (via LOGCOMBINER) as a maximum clade credibility (MCC; TREEANNOTATOR [44]), with nodal support representing posterior probabilities. The resulting topology was visualized using FIGTREE (v.1.4.3; [46]). We statistically compared topologies from ML and BA analyses with respect to the original sequence alignment (IQ-TREE2 v2; [43]). We then examined the support displayed by each regarding RELL approximation (10,000 replicates), as well as: (a) Raw change in log-likeli-

hoods; (b) Bootstrap proportions [47]; (c) Kishino-Hasegawa test [48]; (d) Shimodaira-Hase- gawa test [49]; (e) Approximately Unbiased test [50]; and (f) Expected Likelihood weights [51]. Developing a TIMETREE For molecular dating, we employed a rapid, high-performance ML-based model (RELTIME, MEGA X; [33]) that incorporates a relative rate framework (RRF) within which the evolution- ary rates of sister lineages are contrasted against a minimum rate change, as diagnosed between evolutionary lineages and their respective descendants. The computational speed and accuracy of the model are appropriate for the analysis of large data sets (>100 ingroup sequences; [52], with procedural steps for RELTIME previously documented [53]). The evolu- tionary model for our ML-based phylogenetic tree was also implemented in RELTIME. We estimated divergence times based on the earliest fossil evidence for Schizothoracini (e.g., Paleoschizothorax qaidamensis; [54], as gauged in the Qaidam Basin [55]). This placed initial divergence at the onset of the Oligocene (33 Ma) (see also [56]). Phylogeographic analyses A haplotype network provides a visualization of genealogical relationships based on individual mutations separating haplotypes, and as such, provides insights into genetic diversity within populations or closely related species. It is beneficial in that it effectively separates those indi- viduals with similar haplotypes th generally depicted within a phylogenetic tree as unresolved polytomies (i.e., ’combs’). However, it differs from a strictly bifurcating phylogenetic visualiza-

tions in that each haplotype (depicted as a circle) can potentially be connected to multiple, rather than just a single, node. A second difference is that a comparative phylogenetic tree represents a platform from which genetic variation can be interpreted within the context of geologic history, as supported by statistical and/or Bayesian analyses [57]. In addition, a phylogenetic tree is rooted from ancestral to more derived, with geographically co-distributed species (and their shared lineage histories) transcending. In short, relationships can be well-depicted within a species using a haplotype network, but often much less so for distinct species, given the increase in genetic variability and the lack of an evolutionary trajectory. PLOS ONE Himalayan uplift and aquatic evolution PLOS ONE | https://doi.org/10.1371/journal.pone.0289736 October 24, 2023 9 / 27