Table S1. Information on alignments of each species to the X. maculatus reference genome.

Table S2. Large discordant regions identified by the AU test.

Table S3. Large discordant regions identified by PhyML_multi.

Table S4. D-statistic and jackknife standard error for each scaffold (0--149).

Table S5. Values of Patterson's D-statistic for regions supporting a discordant topology (as identified by an AU P-value > 0.9 for an alternate topology) suggest that the D-statistic and AU test give consistent results.

Figure S1. Performance of PhyML_multi on test sequences with known breakpoints between topologies with discordant segments ranging from 2 kb to 10 kb in size.

Figure S2. (A) Size distribution of erroneous discordant regions detected by PhyML_multi in 900 simulated sequences, excluding breakpoints detected in the last 10 kb of the simulated alignment. This suggests that most erroneous discordant regions detected by PhyML_multi are <5 kb. (B) Number of incorrectly identified breakpoints in 100 simulations of each size class of discordant segments. Sequences in which few discordant segments are detected also have few instances of detection of erroneous breakpoints.

Figure S3. AU P-value for support of the discordant topology in 1000 simulations of (A) windows containing support only for the concordant topology, (B) windows contain support only for the discordant topology, and (C) windows containing some support for both topologies.

Figure S4. AU P-value support for the discordant topology in 1000 simulations with varying numbers of contiguous base pairs supporting that topology in 5 kb (A) and 10 kb (B) windows.

Figure S5. Three speciation models simulated with msHOT: (A) allopatric speciation, (B) speciation via admixture, and (C) allopatric speciation with limited gene flow.

Figure S6. Schematic of simulation strategy with msHOT.

Figure S7. Percent divergence (from X. maculatus) and polymorphism in X. birchmanni at different coverage thresholds suggests that polymorphism and divergence estimates are not coverage dependent.

Figure S8. Based on AU tests, the three topologies (X. clemenciae, X. hellerii), (X. maculatus, X. hellerii), and (X. maculatus, X. clemenciae) are all supported in some regions in Scaffolds 0-149, but the regions that support them vary in size.

Figure S9. Size distribution of regions supporting a close grouping of X. hellerii and X. maculatus (sampled from Scaffolds 0-149) based on analysis with PhyML_multi.

Figure S10. Expected size of discordant regions due to ILS based on 100 simulations of 100 kb regions using ms.

Figure S11. D-statistic and 97.5% confidence intervals (calculated from 1000 nonparametric bootstraps of the simulated ABBA and BABA sites) of 1 Mb sequences generated through 100 simulations of allopatric speciation.

EVO_12009_sm_FiguresS1-S11.docx4091KSupporting info item
EVO_12009_sm_Supplementary_TableS2.xlsx16KSupporting info item
EVO_12009_sm_Supplementary_TableS3.xlsx17KSupporting info item
EVO_12009_sm_Supplementary_TableS4.xlsx16KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.