Cytochrome b phylogeny and molecular species delimitation.
The cytb gene trees produced identical topologies using both the BI and ML reconstruction methods (Fig. 2). The Aporodon major clade recovered the R. mexicanus (R. brevirostris , R. mexicanus , R. darienensis , R. garichensis, R. gracilis, and R. spectabilis ) and R. tenuirostris(represented here by R. microdon , R. tenuirostris , andR. cherrii ) species groups. Specimens from Costa Rica identified as R. sp. by Miller and Engstrom (2008) were recovered as closely related to R. garichensis . R. mexicanus specimens split into three clades that were well-supported in the BI method, but weakly supported by ML. Clade I included individuals with geographical distribution in the Sierra Madre Oriental and northern Oaxaca, Mexico. This clade was sister to the R. mexicanus and R. tenuirostris species groups (pP = 1). Clade II corresponded to specimens distributed in Mexico, Guatemala, and El Salvador, and was recovered as the sister group to R. brevirostris from Costa Rica (pP = 1). The remaining individuals of R. mexicanus were grouped into a clade III with distribution in Colombia and Ecuador and sister toR. darienensis from Panama (pP = 1).
The results of the species delimitation methods were not congruent with each other (Fig. 2). For the R. mexicanus group, the mPTP method identified 15 putative species while 11 were delimited by the bGMYC. The multi-locus method STACEY suggested 10 species for both datasets (cytb + Fgb-I7 and cytb + IRBP), excluding species that were not successfully sequenced. The three methods supported the recognition of R. darienensis , R . sp. from Volcan Poas, Alajuela, Costa Rica, and R. garichensis (including the R . sp . individual from La Carpentera, Cartago, Costa Rica), as species. In addition, R. gracilis from Yucatán and R. spectabilis were considered the same species, excluding specimens of R. gracilisdistributed in El Salvador, which were delimited as a distinct species. For R. brevirostris , each delimitation method proposed different numbers of possible species (from 1 to 3), although bGMYC and K2P genetic distances were congruent in delimiting only one species.
Within the R. mexicanus species complex, the R. mexicanusclade I was supported at the species level by all three delimitation methods. Within the R. mexicanus clade I, STACEY (cytb + Fgb-I7) demarcated four possible species, while the other methods delimited only two. Following the recognition of two species, one of them is formed of samples from Mexico and Guatemala (R. mexicanus clade IIA), whereas the other corresponded to individuals from El Salvador, which were considered a distinct species under all the methods (R. mexicanus clade IIB). Within clade III, the mPTP identified 5 species, whereas bGMYC 2, and STACEY 3 (cytb + Fgb-I7). In this clade, the consensus between methods demarcated two putative species: R. mexicanus clade IIIA and R. mexicanus clade IIIB. The cytb genetic distance values between the species delimitation consensus ranged from 4.96 to 18.18, with the highest value generally betweenR. mexicanus clade I and the other identified species. The cytb genetic distance values between delimited taxa in the R. mexicanus species group are shown in Table 2 and Appendix 2.