Computational protocol: Transoceanic origin of microendemic and flightless New Caledonian weevils

Similar protocols

Protocol publication

[…] We used IQ-TREE 1.5 [] to infer the phylogenetic relationships among Trigonopterus weevils in a maximum-likelihood (ML) framework. The concatenated dataset was partitioned according to the results of a PartitionFinder v. 1.1.1 [] analysis. The partitioning schemes and corresponding models of substitutions were searched using the greedy algorithm and the raxml set of models in PartitionFinder. The likelihoods of the different models of substitution were compared using the Akaike information criterion corrected (AICc). We performed 1000 ultrafast bootstrap replicates [] and 1000 SH-aLRT single branch test replicates to investigate nodal support across the topology. [...] To infer the divergence time estimates of the group, we first tested the hypothesis of the molecular clock in MEGA6 [] by comparing the ML value of the IQ-TREE topology with and without the molecular clock constraints under the Tamura–Nei model. The null hypothesis of an equal evolutionary rate throughout the tree was rejected at a 5% significance level (p < 0.001). Therefore, we used a Bayesian relaxed clock approach as implemented in BEAST v. 1.8.2 [].‘Cryptorhynchinae sensu latu’ are defined by a rostral furrow which was attained by several unrelated groups, while the monophyletic Cryptorhynchinae sensu strictu currently cannot be diagnosed based on single morphological characters []. To date, there is no fossil that could be related to Trigonopterus and even the assignment of published records to Cryptorhynchinae s.s. is doubtful. Thus, we used two types of calibration strategies to obtain absolute divergence time estimates for the origin and further diversification of New Caledonian Trigonopterus weevils.In our first strategy of calibration (CS1), we relied upon a biogeographical calibration point to enforce maximum ages in the phylogeny. In the absence of alternative reliable external calibration points (e.g. fossils), biogeographical calibrations can be very useful and provide meaningful insights []. Here, we used the complete submergence of Java ca 10 Ma [,] to enforce a maximum age constraint on the stem of Javanese species and/or clades in the topology. The maximum constraints were enforced in BEAUti 1.8.2 [] using normal distributions with a mean of 10 and an s.d. of 1. The distributions were truncated to fit the interval (5–15 Ma). The use of a normal prior distribution allows considering the possibility that Javanese lineages are slightly older than the age of the re-emergence of Java ca 10 Ma. The concatenated dataset was partitioned by gene and each fragment was assigned a different uncorrelated lognormal relaxed clock model except for the mitochondrial fragments that were assigned a unique clock model.In our second strategy of calibration (CS2), we used published rates of nucleotide substitution to constrain the ucld.mean (arithmetic mean of branch rates) of gene fragments sequenced in this study. A recent study used multiple fossil constraints to calculate estimates of substitution rates in different gene fragments sequenced for Carabus ground beetles []. This is one of the most recent, fossil-based studies providing ucld.mean estimates and credibility intervals for different gene fragments in flightless beetles. This study recovered a ucld.mean of 0.0134 substitutions per site per million year and per lineage (credibility interval CI = 0.0108–0.0162 subst/site/Myr/lin). A different study [] conducted on flightless darkling beetles (Coleoptera, Tenebrionidae) recovered comparable rates for the mitochondrial clock using a well-documented biogeographic calibration point (ucld.mean = 0.0119 subst/site/Myr/lin; CI = 0.0108–0.0140 subst/site/Myr/lin). Considering the assumption that mitochondrial genes are following a same clock model, we constrained the mitochondrial ucld.mean in our dataset (CO1 and 16S combined), with a uniform prior distribution encompassing the CI (0.0108–0.0162 subst/site/Myr/lin), calculated for mitochondrial genes of Carabus beetles []. We assumed a different clock for each nuclear gene fragment, and the ucld.mean estimates of these were left unconstrained.In both analyses, the tree model was set to a Yule model, and we used the IQ-TREE topology as a fixed input by editing the .xml file. The runs consisted of 30 million generations sampled every 1000 generations for CS1, and of 50 million generations sampled every 5000 generations for CS2. Convergence of the runs was assessed by checking the likelihood trace of each parameter and its effective sample size in Tracer v. 1.6 (http://BEAST.bio.ed.ac.uk/Tracer). A value of ESS >200 was acknowledged as a good indicator of convergence. The maximum credibility consensus tree, median ages and their 95% highest posterior density (HPD) were generated afterwards under TreeAnnotator v. 1.8.2 []. […]

Pipeline specifications

Software tools IQ-TREE, PartitionFinder, RAxML, MEGA, BEAST
Application Phylogenetics
Organisms cellular organisms