All-species analysis
Of the 17 recognized species that we sampled in this study, the following provided the first complete cytb sequences from Costa Rica: Didelphis marsupialis, Nyctomys sumichrasti, Oryzomys couesi, Peromyscus nudipes, Philander melanurus, Proechimys semispinosus, Scotinomys teguina, Sigmodon hirsutus. In those instances where complete cytb sequences were already available for Costa Rica, our data often provided new locations within the country.
Combining our new data with previously published sequences in INSDC, we generated a cytb phylogeny using IQ-TREE (Fig. S1). The general features of the phylogeny were as expected, with placentals and marsupials each forming monophyletic groups sister to each other, although the support for the placental grouping is weak. Within the placentals, the Eulipotyphla and the three families of rodents (Cricetidae, Echimyidae, Heteromyidae: Table 1) all form well-supported monophyletic groups. The relative positioning of these four groups is unresolved, so that the Rodentia do not form a monophyletic group. This may be a consequence of long-branch attraction (Bergsten 2005).
Nearly all the 41 named species analysed form monophyletic groups within the tree. Nevertheless, some are paraphyletic, or have an otherwise unclear relationship, namely: Peromyscus nudipes/nicaraguae, Scotinomys xerampelinus and Transandinomys talamancae. Heteromys sp shows features that sets it apart from named species, forming its own clade. Interpreting the results with Reithrodontomys sp is more complex. Sequences classified as Reithrodontomys sp mostly occupy a large independent clade. However, there is one sequence that is positioned separately with the Reithrodontomys brevirostris sequence as its closest relative.
The following list considers each species in turn, with reference to location data in Tables S1 and S2 and Fig. 1, and referring to the phylogenetic structure in Fig. S1 (with highlighted branches in Fig. 2):
Caluromys derbianus (Fig. 2D): There are no new sequences and no phylogenetic structure with the sequences from Costa Rica, Panamá and Ecuador.
Cryptotis (3 species) (Fig. 2C): These three species are represented by four sequences, clustered together. We contributed a new sequence of C. nigrescens from BCNP in the province of Heredia, while the previous sequence was also from Costa Rica but from a different province (Puntarenas). They cluster together with long branch lengths.
Didelphis marsupialis (Fig. 2D): Our three sequences from Costa Rica, from close to the border with Panamá, add to previous sequences from Panamá, Brazil and Mexico. There is little structure but our sequences cluster most closely with those from Panamá and those from Mexico are most distinctive.
Didelphis virginiana (Fig. 2D): There are no new sequences, and minor differentiation between the sequences from Mexico and the United States.
Diplomys labilis (Fig. 2C): There are no new sequences, just one existing sequence from Panamá.
Handleyomys alfaroi (Fig. 2A): Our new sequence from Costa Rica (LAIP-PT, Puntarenas) clusters with another sequence from Puntarenas and a sequence from Panamá. The sequences from Honduras, Guatemala and Nicaragua are marginally separate, but there is very little discernible subdivision.
Heteromys desmarestianus (Fig. 2C): Our six sequences from coastal central Costa Rica (MANP, Puntarenas) cluster closely with a previous sequence from Puntarenas and a sequence from Cartago (inland central Costa Rica). A sequence from another more northern Costa Rican province (Alajuela) is more distantly related. More distantly related still are two sequences from Honduras (one was ours), which cluster together, and a sequence from Mexico. Thus, there is some subdivision in H. desmarestianus over a wide geographic area.
Heteromys nubicolens (Fig. 2C): There are no new sequences, and the two sequences previously obtained (from Guanacaste and Puntarenas in Costa Rica) are very similar.
Heteromys oresterus (Fig. 2C): There are no new sequences, and the five sequences previously obtained (from San José and Cartago in central inland Costa Rica) are very similar.
Heteromys salvini (Fig. 2C): Our two sequences from SRNP (Guanacaste) in Costa Rica group very closely with previous sequences from Guanacaste and Puntarenas and a sequence from Honduras.
Heteromys sp (Fig. 2C): This provisional designation by Rogers and González (2010) was based on a distinct lineage located in Costa Rica (found in Alajuela and Limón). We added another sequence from Limón (AFRS). All sequences are very closely related. On the basis of all the cytb data available for this form and in the context of our whole phylogenetic tree, Heteromys sp most likely represents a separate species, having a similar level of distinctiveness as other recognized species (both considering Heteromys and other genera in our phylogenetic tree).
Hoplomys gymnurus (Fig. 2C): There are no new sequences, just one existing sequence from Panamá.
Marmosa alstoni (Fig. 2D): There are no new sequences, just two identical existing sequences from Panamá.
Marmosa mexicana (Fig. 2D): There are no new sequences, and most of the five sequences previously obtained from Guatemala and Mexico are very similar to each other. One sequence from Guatemala (HM106344) stands out as distinctive.
Melanomys chrysomelas (Fig. 2A): The four previous sequences from Nicaragua and Costa Rica (Heredia) are closely related to each other, and one of our four new sequences is closely related to those (SM3690). However, the other three new sequences from the same locality (MANP, Puntarenas) form a somewhat distinct lineage.
Metachirus nudicaudatus (Fig. 2D): There are no new sequences, and the five sequences from four countries (Ecuador, Guyana, Panamá, Peru) are distinctive from each other – the sequence from Guyana particularly so. The two sequences from Panamá do form a clade though.
Nephelomys devius (Fig. 2A): All three sequences are from Costa Rica and our sequence (from LAIP-VS, Limón) and previous sequences from Puntarenas and Cartago are extremely similar.
Nyctomys sumichrasti (Fig. 2A): The five existing sequences from three different countries (El Salvador, Guatamala, Mexico) are generally distinctive from each other except the two from El Salvador that are very similar. The two sequences from Mexico are particularly divergent. Our two new sequences from Costa Rica, although both coming from the same northern inland locality (S. Verde, Heredia) are different from each other and from the other sequences.
Oecomys trinitatis (Fig. 2A): There are no new sequences, just one existing sequence from Peru.
Oligoryzomys costaricensis (Fig. 2A): There are no new sequences, just two identical existing sequences from Panamá.
Oryzomys couesi (Fig. 2A): Our new sequence from Costa Rica (LAIP-PT, Puntarenas) is very similar to existing sequences from Guatemala, Honduras and Nicaragua.
Ototylomys phyllotis (Fig. 2A): There are no new sequences, just one existing sequence from Honduras.
Peromyscus nicaraguae/nudipes (Fig. 2B): We contributed 18 new sequences from Costa Rica (BCNP, LAIP) and one from Honduras (SM3205). They were all labelled P. nudipes but in fact they fall into one of two distinct clusters of tightly related individuals. One of these clusters, which includes our sequence from Honduras and all our sequences from BCNP in Heredia (and one from LAIP-PT, Puntarenas), also includes five previous sequences from Costa Rica (Puntarenas), Honduras and Nicaragua. Four of these previous sequences have been named P. nicaraguae, and it appears likely that the sequences in this cluster should all be classified as P. nicaraguae. The other clade with ten of our sequences from LAIP-VS (Limón) and a previous sequence from Panamá are likely appropriately named P. nudipes.
Philander melanurus (Fig. 2D): Our new sequences from Costa Rica (LAIP-EA in Puntarenas, CERS and AFRS in Limón) cluster very closely with previous sequences from Colombia and Panamá.
Proechimys semispinosus (Fig. 2C): Our new sequences from Costa Rica (MANP and Sansi in Puntarenas and S. Verde in Heredia) cluster with a sequence from Colombia. We have contributed eight new sequences from Costa Rica and there appears to be substantial variation among those.
Reithrodontomys
The genus-specific analysis is given below.
Rheomys raptor (Fig. 2A): There are no new sequences, just one existing sequence from Costa Rica.
Scotinomys teguina/xerampelinus
The combination of our new sequences from Costa Rica and existing sequences from Honduras (S. teguina) and Costa Rica (S. xerampelinus) indicate that there is more genetic subdivision than just the two named species. This is considered in more detail in González et al. (in prep.).
Sigmodon hirsutus (Fig. 2A): Our new sequences from Costa Rica (SRNP, Guanacaste) cluster very closely with previous sequences from Honduras, Mexico and Nicaragua.
Sigmodontomys alfari (Fig. 2A): There are no new sequences, and minor differentiation between the existing sequences from Panamá and Ecuador.
Transandinomys bolivaris/talamancae (Fig. 2A): There are no new sequences and these two species are represented by five existing sequences. The three T. talamancae sequences from Panamá cluster together. However, the Transandinomys from Ecuador form two further branches with the disposition making T. talamancae paraphyletic.
Zygodontomys brevicauda (Fig. 2A): There are no new sequences, and the five sequences from three countries (Bolivia, Panamá, Venezuela) show a degree of genetic subdivision, though it does not relate in a simple way to geography (e.g. the three sequences from Venezuela are not each other’s closest relatives).
Reithrodontomys analysis
The sequences that we contributed to the phylogeny of the Reithrodontomys genus were Reithrodontomys sp, R. creper and R. sumichrasti (Fig. 3). Our sequences of R. creper and R. sumichrasti are very similar to INSDC sequences of the same species and confirm the well-supported monophyly of those two species within the tree. For R. creper, the existing sequences are from the central northern interior of Costa Rica (Cartago, Heredia). Our new sequences came both from this region (BCNP, Heredia) but also at the interior border with Panamá (LAIP-VS, Limón). All 24 sequences are very similar to each other. The previously obtained sequences of R. sumichrasti come from a broader geographic area (Costa Rica, Guatemala, Honduras, Nicaragua, Panamá) and our two new sequences are identical to the existing sequence from Costa Rica, being from the same general region in the central interior of the country (our sequences: BCNP, in the south of Heredia; the earlier sequence: Cartago).
The five sequences from R. fulvescens are all from INSDC and form a distinct monophyletic grouping at the base of the Reithrodontomys clade. All the sequences are from Mexico and there is considerably more structure within the lineage than in R. creper and R. sumichrasti.
The situation elsewhere within the Reithrodontomys lineage is more complex and our Reithrodontomys sp sequences are on three separate branches. One of those sequences found in three individuals (SM3143, 3149, 3150) from LAIP-EA and LAIP-PT (Puntarenas) is closely related to a published R. brevirostris sequence from Cartago, suggesting that those particular Reithrodontomys sp are R. brevirostris. This is supported by the inclusion of a shorter published R. brevirostris sequence from Alajuela in the same clade (Fig. S2). The addition of another short published sequence of R. dariensis suggests that the R. brevirostris lineage is in a larger clade consisting of R. dariensis, R. gracilis and R. mexicanus lineages as well (Fig. S2). The apparent misplacement of one R. mexicanus sequence (AF108708) clustering with R. dariensis is not surprising because “R. mexicanus” has been a standard “catch-all” designation for Reithrodontomys in Central America until a recent taxonomic reevaluation (Gardner and Carleton 2009).
A similar phenomenon may explain the clade containing two short sequences of R. cherrii together with R. mexicanus sequences all from the same province in Costa Rica (San José) (Fig. S2). That apparent R. cherrii lineage is grouped with R. tenuirostris and R. microdon (showing paraphyly) (Fig. S2). The sequences from these two species are from Mexico and Guatemala.
The final two lineages consisting solely of Reithrodontomys sp in the main analysis are notable for the lack of genetic subdivision despite large numbers of sequences (33 in one lineage, six in the other) (Fig. 3). Again, a short sequence suggests that the larger clade maybe R. garichensis (Fig. S2).