BASAL UNITS IN THE MORMOOPIDAE

The species diversity of the mormoopids has been underestimated. Two hypotheses underlie competing assessments of species diversity. The first hypothesis assumes populations of widespread species as part of a continuum of differentiation that appears great at the extremes, but is only slight between adjacent groups (Koopman, 1955). Although some populations are allopatric, it is assumed that gene flow among them exists or occurred until recently. Because morphological intergradation (used to infer gene flow) among insular and continental populations is not observed, the range of (non)resemblance permitted in a given species has been widened (Smith, 1972).

Bayes, parameters used in Bayesian analysis of concatenated data; GTR, general time reversible model; ML, parameters used in maximum likelihood analyses; R -matrix, rate matrix parameter (with respect to G-T transversion); α, shape parameter, I, proportion of invariant sites; –2logΛ, 2[log L 1 − log L 2], where L 1 = likelihood without clock and L 2 = likelihood with clock. Parameters obtained from Bayesian analyses are followed by the 95% confidence interval (in parentheses)

.

The assessment of sequence variation among putative units (subspecies) within the mormoopids (Table 2) revealed multiple instances of characters that appear to be fixed in cytochrome b. These molecular data support a second hypothesis: gene flow between insular and continental populations appears to have ceased even before fully recognized biological species (e.g. P. gymnonotus and davyi; Fig. 3B) evolved into separate lineages. The subtle morphological differences dismissed under a presumption of gene flow provide evidence for the isolation and independent evolution of separate lineages in widespread species such as parnellii, davyi, and personatus . Because sampling sizes for molecular markers were small, these differences alone cannot provide species limits. In some instances, molecular character differences, high sequence divergence among presumed conspecifics (bottom two tiers of Fig. 2), distributional ranges that encompass broad areas separated by water and land barriers (Table 1, Fig. 1), and taxonomic limits based on morphological variation (Smith, 1972) coincide and strengthen the hypothesis of evolutionary independence. These criteria (Tables 1 and 2) apply to named island populations of P. parnellii: parnellii (Gray, 1843), pusillus (Allen, 1917), and portoricensis (Miller, 1902); the continental P. parnellii ranging from Mexico to Guyana currently classified in the subspecies mexicanus (Miller, 1902), mesoamericanus (Smith, 1972), and rubiginosus (Wagner, 1843); the currently recognized subspecies of P. davyi: davyi (Gray, 1838) and fulvus (Thomas, 1892); and subspecies of P. personatus: personatus (Wagner, 1843) and psilotis (Dobson, 1878) . Each of these populations should be considered as a species, named using the subspecies taxonomy. The name Pteronotus rubiginosus (Wagner, 1843) precedes mexicanus and mesoamericanus, and applies to the continental bats in the P. parnellii lineage as described above (note that the status of fuscus and paraguanensis was not evaluated; Table 1). Both cytochrome b (Table 2) and Rag 2 (Lewis Oritt et al., 2001) showed differentiation in Mexican and Central American populations of P. psilotis . Further sampling is necessary to determine if characters are fixed because taxonomic conclusions derived from single molecular exemplars would be suspect.

*Length difference arises from constraining ( parnellii Caribbean, parnellii Mexico, and Central America), and not from relationships between quadridens and macleayii, and their sister taxon. †Differences pertain the constraint [( quadridens, macleayii), personatus] only. MP, maximum parsimony (number of extra steps and significance in Templeton (1983) test; PB, parametric bootstrap (maximum likelihood); SH, Shimodaira & Hasegawa (1999) test (maximum likelihood).

Mitochondrial cytochrome b from samples of Pteronotus rubiginosus and P. personatus from northern South America west of Guyana is distinct from that sampled east of Guyana (Suriname and/or French Guiana; Table 2, Fig. 1). These character differences and attendant levels of sequence divergence had not been anticipated in the morphological study of Smith (1972). French Guianan specimens of P. parnellii can also be distinguished from those from the remainder of the range by their larger size (Simmons & Voss, 1998). In Venezuela, P. paraguanensis appears to have become isolated as a result of breaks in the humid forest (Gutiérrez, 2004). This mechanism might explain the differentiation observed, but greater geographical and character sampling is needed to investigate these (possibly) cryptic species, and test the possibility that accelerated rates of sequence evolution have led to this pattern (although this is unlikely, see Table 4).

In P. quadridens discontinuous variation in cytochrome b occurs between Cuba and Jamaica, and Hispaniola and Puerto Rico, rather than coinciding with the subspecies taxonomy that separates Cuban from other Greater Antillean bats (Tables 1, 2). These taxa are not elevated to species here, despite the possible geographical isolation by ocean barriers, because sampling was sparse, the molecular differentiation does not match subspecies boundaries based on morphology, and no differences were detected in Rag 2. For the purpose of estimating ancestral areas, each terminal that appears with a name in Figure 3 was treated as a separate taxon.