Computational protocol: Is geography an accurate predictor of evolutionary history in the millipede family Xystodesmidae?

Similar protocols

Protocol publication

[…] Sequence chromatograms were edited in the Mesquite module Chromaseq (Version 1.2) implementing phred and phrap for nucleotide base-calling, trimming, and quality control (; ). Sequences for the six genes were exported as individual FASTA files to PRANK (Version 140110) for multiple sequence alignment (). In PRANK, the default iterative process of constructing multiple guide trees was used, and -F option was included for the mitochondrial ribosomal genes where there was a large number of insertions/deletions. Aligned sequences were partitioned by gene, codon position, and exon/intron boundaries, and concatenated in Mesquite. PartitionFinder (Version 1.1.1) was used to compare alternative partitioning schemes and nucleotide substitution models using the Bayesian Information Criterion (BIC) model selection method and models of evolution for MrBayes (). MrBayes (Version 3.2.2) was used for phylogenetic inference with the best-fit partitioning model (). We ran two hot and two cold independent MCMC chains simultaneously, retaining two of the chains for an initial 1,000,000 generations each. Each chain was sampled every 100 generations, with a one-quarter generation burn-in to reach a posterior distribution of trees and a consensus topology. The standard deviation of split frequencies (SDSF) convergence statistic was monitored during the analysis and then checked upon completion to ensure convergence between the independent chains (). Upon conclusion of the run, parameters were averaged and the consensus topology generated from the posterior distribution of trees with a consensus type of all compatible groupings. To assess varying resolutions of gene histories, single-gene trees were independently estimated in MrBayes. Nucleotide frequencies and base composition were assessed in PAUP* (). [...] Ninety xystodesmid millipede species-group taxa were examined for 68 qualitative morphological characters (). Morphological traits were examined and character states scored using a Leica M125 stereomicroscope (Leica Microsystems, Wetzlar, Germany). Gonopods from the left side of the 7th body ring were dissected following , and photographed in alcohol with a Canon 6D dSLR camera and a 65 mm MP-E macro lens mounted on a Passport II Portable Digital Imaging System (Visionary Digital, Charlottesville, VA, USA). Gonopods were photographed at 6–20 focal planes and stacked in the program Helicon Focus (Helicon, Kharkiv, Ukraine). Gonopods were illustrated by tracing photographs of specimens in Adobe Illustrator CS6 (Adobe, San Jose, CA, USA). The matrix was comprised of 46 binary and 22 multi-state characters, of which 40 were male gonopodal characteristics, 23 were from the exoskeleton and five were from female cyphopods (; ; ). The morphological dataset was assembled and characters scored in Mesquite. MrBayes was used to infer a phylogeny based on the morphological data partition using likelihood-based models of character state change. Tree topologies were evaluated based on the Markov k (Mk) model () with and without Γ-distributed rates of character change, and with the character coding set to “variable”. We ran the analysis as with the molecular partition, monitoring the SDSF diagnostic to ensure convergence and averaging the parameters and generating a consensus tree with all compatible groupings. Alternative models with and without Γ distributed rate variation were compared with Bayes factors to determine the best-fit to the data. [...] To assess homoplasy of morphological characters and therefore evaluate utility of the features as diagnostic characters for taxonomy, character states were mapped onto the molecular phylogeny using the posterior probability mapping program SIMMAP 1.5.2b21072010 (). SIMMAP takes into account the uncertainty in modeling evolutionary change, using parameters such as branch length, rates of evolutionary change and topology (; ). To map morphological characters onto the molecular tree, we used a morphological/standard model with a bias parameter and a beta distribution prior (alpha = 1, k = 31) for binary characters and an empirical prior for multistate characters (). We ran a Markov chain Monte Carlo analysis with 100,000 generations with a 10% burn in for each morphological character to determine the best-fit rate parameter alpha and beta priors (k = 90, T rate = 1). To measure homoplasy, the number of transformations per character were estimated using SIMMAP and then used to calculate the consistency index (CI) for each character, calculated: CI = expected number of state changes/observed number of state changes. The degree of homoplasy for a morphological character (HI = 1–CI) decreases as the CI approaches one. SIMMAP was also used to calculate the dwell time for each character state, which is the proportion of time that a character spent in a specific state ().Because male genitalia are highly variable anatomically and central to the taxonomy of the family, we evaluated the spatial distribution of homoplasy on the gonopods. Male gonopods are composed of six podomeres that have fused during gonopod development, and only three vestigial podomeres, the coxa, prefemur, and tibiotarsus, (the latter of which is a fusion of the other four segments) remain distinguishable. We divided the gonopod into these three regions and summed the CI in each region (dividing each region’s sum by the total of all three regions to standardize) to assess if CI, and convergence, is evenly spread among the podomeres. […]

Pipeline specifications