Data from: Sex-biased dispersal obscures species boundaries in integrative species delimitation approaches

Accurate delimitation of species is crucial for a stable taxonomy, which provides the foundation for the study of evolutionary biology, ecology and essentially all biological disciplines. Several approaches towards impartial and repeatable taxonomic practices are available but all existing methods have potentially unacceptable shortcomings. In particular, problems can arise when the underlying model assumptions are violated, for instance in the presence of reduced gene flow. This is observed in the context of sex-biased dispersal, which is a common but underappreciated feature in many groups of organisms. Previously, simulations have indicated that sex-biased dispersal may lead to erroneous estimations of the true species numbers. However, this phenomenon has never been examined using empirical data. We evaluate the bias introduced by extreme female philopatry on a range of de novo (GMYC, PTP, ABGD, statistical parsimony, trinomial distribution of triplets model [tr2]) and validation (STACEY, iBPP) approaches to species delimitation in the scarab beetle genus Pachypus. Since female philopatry exhibited in this genus in particular can affect mitochondrial gene flow, we compared the results from analyses of single loci, mitochondrial loci, nuclear loci and combined data, as well as the performance of morphometric data as a secondary data source in a fully integrative Bayesian framework. Large overestimation of species numbers was observed across all analyses of combined and mitochondrial DNA datasets, suggesting specimens from nearly every sampling location as separate species. The use of nuclear data resulted in more reasonable estimations of species boundaries, which were largely supported by morphometrics of linear measurements, while geometric morphometrics of body outlines resulted in stronger splitting. Simulations of population divergence with migration, corresponding to the biology of Pachypus, showed that female philopatry strongly increases reciprocal monophyly of mitochondrial markers and may substantially contribute to over-splitting in species delimitation. Robust results recovered using nuclear DNA and morphological data nevertheless enabled us to reach novel conclusions about species boundaries in Pachypus. Our findings suggest that mitochondrial DNA will be less suited to species delimitation in many cases, in particular in the presence of sex-biased dispersal.Supplement figures A1--A28 and Tables A1--A6Online supplementary figures A1 -- A28 and tables A1--A6. Machine readable versions of tables A1, A2, and A4 are provided as separate files in the same repository.Supplement_Figures_A1--A28_and_Tables_A1--A6.pdfTable A1Specimen assignment to the putative species that were delimited in this study, sampling locality (refers to table A2), and GenBank accession numbers.Table_A1.csvTable A2IDs of sampling localities that were used in trees (Fig. 2, Supplementary Figs. A2, A3) and the locality map (Supplementary Fig. A1). Coordinates are given in decimal format (WGS84).Table_A2.csvTable A4Linear measurements of body parts in millimeters. AL = length of antennal club, EWhc = elytron width at humeral callus, EL = elytron length, PWmax = maximum pronotum width, PWb = width of pronotum at base, PL = medial pronotum length, MTL = length of metatibia, MTW = maximum width of metatibia, 'na' = not available.Table_A4.csvTable A7Detailed results of iBPP analyses of guide tree part A. Posterior probabilities are given for all splits in the tree, 5 replicates and their median (means-columns). For each of 9 data sources (including prior sampling), a 3x3 array with results from different prior combinations of theta and tau_0 is given. The 3x3 arrays are arranged as in main text figure 5 and online supplement figures A8--A11.Table_A7.txtTable A8Detailed results of iBPP analyses of the unmodified guide tree part B. Posterior probabilities are given for all splits in the tree, 5 replicates and their median (means-columns). For each of 9 data sources (including prior sampling), a 3x3 array with results from different prior combinations of theta and tau_0 is given. The 3x3 arrays are arranged as in main text figure 5 and online supplement figures A8--A11.Table_A8.txtTable A9Detailed results of iBPP analyses of the geography-informed guide tree part B. Posterior probabilities are given for all splits in the tree, 5 replicates and their median (means-columns). For each of 9 data sources (including prior sampling), a 3x3 array with results from different prior combinations of theta and tau_0 is given. The 3x3 arrays are arranged as in main text figure 5 and online supplement figures A8--A11.Table_A9.txtTable A10Table of parameters, command lines, and results for population genetic simulations with ms and seqgen. Each of the 630 simulation settings was repeated 100 times. See column headers for descriptions.Table_A10.cs