Data from: The challenges of resolving a rapid, recent radiation: empirical and simulated phylogenomics of Philippine shrews

Phylogenetic relationships in recent, rapid radiations can be difficult to resolve due to incomplete lineage sorting and reliance on genetic markers that evolve slowly relative to the rate of speciation. By incorporating hundreds to thousands of unlinked loci, phylogenomic analyses have the potential to mitigate these difficulties. Here, we attempt to resolve phylogenetic relationships among eight shrew species (genus Crocidura) from the Philippines, a phylogenetic problem that has proven intractable with small (< 10 loci) data sets. We sequenced hundreds of ultraconserved elements and whole mitochondrial genomes in these species and estimated phylogenies using concatenation, summary coalescent, and hierarchical coalescent methods. The concatenated approach recovered a maximally supported and fully resolved tree. In contrast, the coalescent-based approaches produced similar topologies, but each had several poorly supported nodes. Using simulations, we demonstrate that the concatenated tree could be positively misleading. Our simulations also show that the tree shape we tend to infer, which involves a series of short internal branches, is difficult to resolve, even if substitution models are known and multiple individuals per species are sampled. As such, the low support we obtained for backbone relationships in our coalescent-based inferences reflects a real and appropriate lack of certainty. Our results illuminate the challenges of estimating a bifurcating tree in a rapid and recent radiation, providing a rare empirical example of a nearly simultaneous series of speciation events in a terrestrial animal lineage as it spreads across an oceanic archipelago.Online Appendix 1 - SNP Analysis PipelineText file containing commands and software used to obtain SNPs from our UCE reads.Online Appendix 2 - Locus InfoExcel file containing information on each UCE locus used in this study. This information includes: locus length, number of informative sites, number of SNPs, and best fitting model. There are two sheets: one listing all of the loci included in our final dataset (1,112 loci) and another listing only the loci that had phylogenetically informative sites (919 loci).Online Appendix 3 - BPP ResultsTheta and tau estimates, and corresponding ESS values, from eight BP&P runs corresponding to eight subsets of the 919-locus dataset. These values were used to construct a species tree that was used as the "true" tree from which we simulated character data.Figure S1 SmilogramFigure S1. Scatter plot of the frequency of variant bases relative to the position of those sites in their alignmentsFigure S2 Concatenated Full MrBayesFigure S2. Non-condensed trees from concatenated analyses. Condensed trees are shown in Fig. 3 in the main text. Numbers at nodes indicate Bayesian posterior probabilities.Figure S3 Empirical StarBEASTFigure S3. All 19 empirical *BEAST subsets. Subsets 1–18 are based on random subsets of 50 loci each; subset 19 includes the remaining 19 loci not included in the other subsets. Numbers at nodes denote Bayesian PPs. Black circles at nodes indicate PPs ≥ 0.95.Figure S4 Sim-Matching StarBEASTFigure S4. All 10 Sim-Matching *BEAST subsets. Each species tree is based on analysis of a different subset of 50 simulated loci. Numbers at nodes denote Bayesian PPs.Figure S5 Sim-Multi-Individual StarBEASTFigure S5. All 10 Sim-Multi-Individual *BEAST subsets. Each species tree is based on analysis of a different subset of 50 simulated loci. Numbers at nodes denote Bayesian PPs.Figure S6 Sim-3x-Rate StarBEASTFigure S6. All 10 Sim-3x-Rate *BEAST subsets. Each species tree is based on analysis of a different subset of 50 simulated loci. Numbers at nodes denote Bayesian PPs.1112 Empirical UCE Nexus FilesAll 1,112 UCE nexus alignments.new-incomplete-50percent-all-species-padded.zip500 Sim-Matching simulated alignmentsFasta files for Sim-Matching simulation500-sim-matching-fasta-files.zip500 Sim-Multi-Individual simulated alignmentsFasta files for Sim-Multi-Individual simulation500-sim-multi-individual-fasta-files.zip500 Sim-3x-Rate simulated alignmentsFasta files for Sim-3x-Rate simulation500-sim-3x-rate-fasta-files.zipSNP DataNexus file containing data for 1,170 SNPs genotyped for all 19 individuals.SNP-data.nexAll 919 Empirical Gene TreesAll 919 PhyML gene trees inferred using CloudForest. After collapsing short branches into polytomies, these trees were used to infer the species tree topologies used in MulRF and ASTRAL.genetrees.treAll Empirical Bootstrap Gene Trees100 bootstrap replicates for 919 loci. Bootstrap replicates were made in CloudForest. Each of the 100 bootstrap sets were used to infer species trees. These pseudoreplicates were used to add nodal support metrics to empirical MulRF and ASTRAL trees.100-bootreps.tree.zipWhole Mitochondrial Genomes and MrBayes BlockNexus file containing sequences for 19 whole mitochondrial genomes and a MrBayes block with MrBayes settings.MrBayes-WMG-Input.ne