INTRODUCTION

The lowland faunas of the mesic Chocó and the xeric Tumbes regions come into contact west of the Andes in Ecuador (Chapman, 1926; Albuja-V. et al., 1980; Cracraft, 1985; Best and Kessler, 1995). This faunal turnover corresponds to a rapid and relatively uniform decrease in precipitation from north to south due primarily to the influence of the cold Humboldt Current in the south (Chapman, 1926; Lynch and Duellman, 1997; Cerón et al., 1999). Evergreen forests of western Ecuador belong to the Chocó, whereas deciduous (“dry”) forests and xeric, nonforested areas to the south comprise the Tumbes region (Cracraft, 1985). The distributions of many mammalian groups typical of lowland Neotropical rainforests terminate west of the Andes in this region (Emmons, 1997; Tirira-S., 1999). Furthermore, several clades present in rainforests west of the Andes (trans-Andean) in Central America and western Colombia encounter their southernmost distributions in western Ecuador and are not found in Amazonian forests to the east (e.g., Diplomys, Heteromys, Hoplomys, and Tylomys; Hershkovitz, 1972; Emmons, 1997). We here consider Heteromys, which shows a unique biogeographic history and has been used in several recent geographic studies (Anderson, 1999; Anderson and Soriano, 1999; Anderson et al., 2002a, 2002b).

The family Heteromyidae originated in North America, and three extant subfamilies are recognized: Dipodomyinae (kangaroo rats and kangaroo mice), Perognathinae (silky pocket mice), and Heteromyinae (spiny pocket mice). While the Dipodomyinae and Perognathinae have long fossil records in North America (Wood, 1935), members of the subfamily Heteromyinae show only Recent and Pleistocene records (Wahlert, 1993, removed the extinct genera Proheteromys, Diprionomys, and Peridiomys from the Heteromyinae). Despite this lack of fossil sequence, heteromyines comprise a well-defined monophyletic group distinct from either of the two other living subfamilies (Hafner, 1981; Hafner and Hafner, 1983; Wahlert, 1991).

Two genera exist in the subfamily Heteromyinae (Heteromys and Liomys), and their constituent species range latitudinally from extreme southern Texas (United States) to Ecuador. Within Heteromys, six (Williams et al., 1993) or seven (Patton, 1993) species are currently recognized, although more probably exist (Handley, 1976; Rogers, 1990; Anderson, 1999). Members of the genus Heteromys constitute the only heteromyids to have dispersed to South America (Hershkovitz, 1972; Simpson, 1980), and the timing and number of colonization events remain controversial (Anderson and Soriano, 1999; Anderson et al., 2002a).

In South America, Heteromys is restricted to the northwestern reaches of the continent, where two species are known to be widely distributed and one has a marginal range. Heteromys anomalus is found in deciduous and moderately mesic evergreen tropical forests across the Caribbean coast and northern foothills of Colombia and Venezuela, as well as on the islands of Margarita, Trinidad, and Tobago (Handley, 1976; Williams et al., 1993; Anderson, 1999) and from relatively xeric regions of the upper Magdalena Valley in Colombia (Hernández-Camacho, 1956; Anderson, 1999). In contrast, H. australis inhabits very mesic rainforests in the Pacific (Chocoan) lowlands of Ecuador, Colombia, and eastern Panamá as well as mesic montane forests eastward in the Colombian Andes. It ranges from sea level to circa 2500 m in sufficiently wet areas (Anderson, 1999) and has also been reported from the Cordillera de Mérida in western Venezuela (Anderson and Soriano, 1999). In addition to these two widely distributed species, a member of the H. desmarestianus complex inhabits high elevations of the Serranía del Darién in extreme northwestern Colombia (provisionally referred to as H. d. crassirostris by Anderson, 1999).

Thus, recent works have considered Heteromys australis the only spiny pocket mouse present in Ecuador (Albuja-V., 1991; Patton, 1993; Williams et al., 1993; Emmons, 1997; Anderson, 1999; Eisenberg and Redford, 1999; Nowak, 1999; Tirira-S., 1999). Thomas (1901) described H. australis from San Javier, in the coastal lowlands of Esmeraldas province in northwestern Ecuador near the Colombian border. In the next 100 years, the only specimens of the genus reported from the country derived from the type locality and two additional sites in the northwestern part of the country (Bulim and Río Palenque Science Center; Anderson, 1999), and one locality in the central-western region (Jauneche; Albuja-V., 1992). Here we examine a much larger number of specimens from Ecuador and southwestern Colombia in order to reevaluate the morphological variation, distributional extent, and taxonomic status of the spiny pocket mice present in this region.

MATERIALS AND METHODS

Museum Specimens

We examined 58 specimens of Heteromys from Ecuador, including most known from that country (and all known to us in museums in Canada, Ecuador, and the United States). We compared the Ecuadorian specimens with available material of H. australis from southwestern Colombia and with representative samples of other recognized species in the genus, including the majority of extant holotypes. Locality information not provided by the collector appears in brackets. Where original elevation was reported in feet, we provide that datum as well as the metric equivalent to the nearest whole number. Localities in Ecuador and southwestern Colombia are detailed in the Gazetteer (appendix 1). Specimens examined (see appendix 2) are housed in the following museum collections (* denotes museums holding specimens from Ecuador): Academy of Natural Sciences, Philadelphia (ANSP); American Museum of Natural History, New York (AMNH*); Departamento de Ciencias Biológicas, Escuela Politécnica Nacional, Quito (EPN*); Field Museum, Chicago (FMNH*); Instituto de Ciencias Naturales, Universidad Nacional de Colombia, Bogotá (ICN); Museo de Zoología, Pontificia Universidad Católica del Ecuador, Quito (QCAZ*); Museo Ecuatoriano de Ciencias Naturales, Quito (MECN*); Museum of Comparative Zoology, Harvard University, Cambridge (MCZ); Royal Ontario Museum, Toronto (ROM*); United States National Museum of Natural History, Washington, D.C. (USNM*); Universidad del Valle, Cali (UV); University of Kansas Natural History Museum, Lawrence (KU*); University of Michigan Museum of Zoology, Ann Arbor (UMMZ); and University of Wisconsin Zoological Museum, Madison (UWZM*).

We also searched for specimens of the genus from Ecuador in the following museums, but did not find any records: California Academy of Sciences; Carnegie Museum of Natural History; Florida Museum of Natural History, University of Florida; Louisiana State University Museum of Natural Science; Michigan State University Museum; Museum of Texas Tech University; Museum of Vertebrate Zoology, University of California, Berkeley; Natural History Museum of Los Angeles County; Texas Cooperative Wildlife Collection, Texas A&M University; and University of New Mexico Museum of Southwestern Biology.

We examined external and cranial morphological characters, making comparisons among specimens of approximately the same age. Specimens were assigned to age classes based on patterns of toothwear and molt, following Rogers and Schmidly (1982). Morphological nomenclature follows Wahlert (1985) and Anderson (1999). To assess the habitat requirements of the species in Ecuador, we used available published reports and attempted to contact all living collectors.

Measurements

We here define and illustrate 11 standard cranial measurements for Heteromys (fig. 1). These linear distances generally follow Genoways (1973), but we add parietal breadth, modify nasal length, substitute squamosal breadth for mastoid breadth, and follow Rogers and Schmidly (1982) in replacing depth of braincase with skull depth. Also, we rename greatest length of skull and alter the wording of Genoways (1973) to unify a telegraphic style. This current suite of measurements better reflects the morphology of the genus Heteromys relative to its sister genus Liomys, for which the measurements of Genoways (1973) were originally designed. The nasal length and mastoid breadth of Anderson (1999) and Anderson and Soriano (1999) represent the nasal length and squamosal breadth presented here, not those of Genoways (1973).

We took these measurements on all available intact skulls to the nearest 0.01 mm with digital calipers (except specimens from the ICN and UV collections, which were measured to the nearest 0.1 mm). Only adult specimens (age classes 4–6) were used for statistical analyses (see below). External measurements and mass were copied from specimen tags, but measurements taken only to the nearest centimeter or half centimeter on two series collected from Nariño, Colombia, in 1912 were excluded. Because some specimens lacked hind foot measurements and others were taken by a variety of collectors (some of which did not include the claws), we verified hind foot measurements with digital calipers for Ecuadorian specimens with complete pedal elements in study skins or fluid-preserved specimens.

Statistics

Specimens from Ecuador and southwestern Colombia were used for quantitative comparisons. Using species assignments based on our qualitative morphological examinations, we tested for sexual dimorphism in cranial measurements within a species in a series of univariate two-tailed t-tests of adult specimens in age class 4; based on those results, the sexes were pooled in all further analyses. For each species, we calculated descriptive statistics of adult specimens in age class 4 for: external measurements, mass, cranial measurements, and two derived ratios (tail length/head-and-body length and interparietal width/zygomatic breadth). All localities for each species were pooled because sample sizes were too small at individual sites. Additionally, we conducted two-tailed t-tests comparing means of the two species for each measurement or ratio.

We also conducted a principal components analysis (PCA) based on the variance-covariance matrix of log_e-transformed cranial measurements of adult specimens in age classes 4–6. We interpreted the multivariate axes by examination of both loadings and coefficients (elements) of the unit eigenvector corresponding to each component. Loadings are Pearson product-moment correlations between specimen scores on an axis, and the respective log_e-transformed measurements. We tested for differences between the species for scores on the first three principal components, by two-tailed t-tests. All statistical analyses were performed in Minitab (1998, version 12.1), and probabilities were compared to an α of 0.05 for hypothesis testing.

SYSTEMATICS

Our analyses indicate that two species of Heteromys are present in Ecuador. One species includes topotypes of H. australis and corresponds to the morphological diagnosis of that species (Anderson, 1999). The other species displays proportions and measurements distinct from those of H. australis as well as differences from all other recognized species of the genus. As no available name exists for it, we describe this taxon as:

Heteromys teleus, new species Figure 2

Holotype:

AMNH 64694, adult male; skin and skull in excellent condition; left and right rami of mandible separated. Collected on 2 June 1923 from Ecuador: Provincia Guayas: Cerro Manglar Alto, western slope by G.H.H. Tate; original number 1536. Tate's field camp was located at 1500 ft [457 m], but his notes and field catalog indicate that the holotype was taken at 2000 ft [610 m].

Paratypes:

We designate the following specimens (adult skins and skulls in good condition) from a variety of museum collections as paratypes. See also appendix 2 for other specimens not included in the type series (ICZN, 1999: article 72.4.6; recommendation 72B). ECUADOR (10): Cotopaxi: San Francisco de las Pampas, Reserva Bosque Nublado Otonga [= Bosque Integral Otonga], 1900–2000 m, QCAZ 1788. Guayas: Cerro Manglar Alto, eastern slope, Camp I, 1200 ft [366 m] (camp), specimen from 1400 ft [427 m], AMNH 64676; Cerro Manglar Alto, western slope, Camp II, 1500 ft [457 m] (camp), specimens from 2000 ft [610 m], AMNH 64681, 64682, 64686, 64689. Los Ríos: Río Palenque Biological Station [= Río Palenque Science Center], 220 m, KU 149133, 149135; Río Palenque Science Center, Santo Domingo, 47 km S of (by road), USNM 528573; Cantón Vinces, Parroquia Palenque, Jauneche, Estación Biológica Pedro Franco Dávila, 50 m, EPN 91.3038.

Etymology:

Latinized from the Greek teleos, meaning “having reached its end, finished, complete” (Brown 1956: 785) in reference to the species' distribution at the southern end of the Chocó, representing the southernmost known distribution of the genus as well as of the superfamily Geomyoidea.

Distribution:

Known only from central-western Ecuador (fig. 3) from near the Río Esmeraldas and Río Guayllabamba (circa 0° 20′N) south to the Cordillera de Chongón-Colonche (southern extension of the Cordillera de la Costa; circa 1°47′S). Present from near sea level on the coastal plain to circa 2000 m on the western slopes of the Andes. Specimens from the Cordillera de Chongón-Colonche may be isolated from other populations, as this low cordillera lies in a matrix of drier vegetation (see Chapman, 1926; Foster, 1992b; Cerón et al., 1999; Sierra et al., 1999b). Note that some localities are documented only by specimens provisionally referred to the species (without examined cranial material) as listed in appendices 1 and 2.

Diagnosis:

A species of spiny pocket mouse with adults showing the following combination of characters (fig. 2): mesopterygoid fossa V-shaped, with long, thin hamular processes of pterygoids; optic foramen small, with exterior margin formed by strong bar of bone; p4 with 3 lophs; skull large (ONL circa 35–38 mm in age class 4), robust, and wide (table 1); rostrum extremely wide, especially basal half; interparietal narrow and laterally rounded; molars distinctly wide; plantar surface of hind feet naked; dorsal pelage dark gray or blackish; no lateral ochraceous band present; pelage always harsh and spiny; ears small to medium in size; body size large, with hind foot 35 mm or greater in adults; tail (occasionally) approximately equal to or (more often) slightly longer than head-and-body length.

Description:

Dorsal pelage spiny and dark gray or blackish, sharply contrasting with soft, pure-white pelage of venter (see illustration in Jarrín-V., 2001); ears blackish and small to medium in size (table 1); tail usually longer than head-and-body length (table 1) and moderately bicolored, with some dark hairs present on ventral surface, especially distally; hairs on dorsal surface of hind feet white, or white with moderate infusion of black hairs; hairs at base of claws of hind feet white; inner side of forearms white; hind feet large (table 1), with naked plantar surface; skull large (table 1) and robust (fig. 2); posterior half of premaxilla extremely wide, forming a distinctive shelf along dorsolateral border of rostrum; anterior roots of zygomatic arches robust; interorbital constriction moderately wide; braincase moderately wide with strong parietal crests, but not inflated; interparietal narrow (table 1) and rounded laterally; anterodorsal lobe of periotic capsule well developed, causing distinct undulation in posterior margin of squamosal (see schematic drawing in Anderson, 1999: 619); incisive foramina wide and bowed; molars large and wide; permanent upper premolar (P4) with short fold in anterior border of posterior loph; mesopterygoid fossa V-shaped, with hamular processes of pterygoids long and thin; optic foramen small, with exterior margin formed by strong bar of bone; jugular foramen (anterolateral to occipital condyle) large; p4 with 3 lophs.

Dental Formula:

(incisors 1/1, canines 0/0, premolars 1/1; molars 3/3) × 2 = total 20.

Measurements of the Holotype (mm):

Total length 283, tail length 154, hind foot 37, ONL 36.44, ZB 18.68, RL 15.87, NL 15.30, IOC 8.59, SB 15.95, MTR 6.05, IW 8.01, IL 5.23, PB 12.61, SD 11.33.

Comparisons:

The dark slaty-gray pelage of Heteromys teleus differentiates it from several species of the genus that have dorsal pelage strongly infused with brown or cinnamon (ochraceous) hairs, causing a grizzled effect. Among these, H. gaumeri and members of the H. desmarestianus complex (including H. goldmani; see Rogers, 1990, and Williams et al., 1993) inhabit Central America far from the known distribution of H. teleus. Specimens of H. gaumeri and many populations of the H. desmarestianus complex possess a lateral ochraceous band that is absent in H. teleus. Furthermore, H. gaumeri is much smaller than the new species (see Engstrom et al., 1987). Members of most populations of H. anomalus (from Colombia, Venezuela, and Trinidad and Tobago) have pale, especially grizzled pelage, in stark contrast to H. teleus. Heteromys teleus can be separated from all populations of H. anomalus by the narrower rostrum, relatively longer (usually sharply bicolored) tail, and larger, rounded ears of H. anomalus, as well as by differences in cranial proportions (relatively wider skull in H. teleus and narrower interorbit in H. anomalus).

When compared with other species of Heteromys that have dark pelage, H. teleus is easily distinguished by its extremely wide rostrum. Heteromys oresterus (endemic to the Cordillera de Talamanca in Costa Rica) has a characteristically long, thin rostrum. Heteromys nelsoni (distributed in southern México and western Guatemala) is even larger than the new species (Williams et al., 1993), with a more elongated skull (including rostrum), larger ears, dusky inner forearms, and soft dorsal pelage. The only other species of Heteromys with a similarly wide rostrum is a member of the H. desmarestianus complex present in the Darién region of eastern Panamá and northwestern Colombia (provisionally referred to as H. d. crassirostris in Anderson, 1999). That taxon, however, is one of the smallest of the genus (ONL 32.2–35.1 mm and hind foot length 29–34 mm in age classes 4–5; Anderson, 1999), has barely perceptible temporal fossae (erroneously named masseter-temporal fossae by Anderson, 1999), straight posterior margin of the squamosal, tapered incisive foramina, chocolate-brown dorsal pelage, and a relatively shorter tail.

Heteromys australis, which has a geographic range that probably comes into contact with that of H. teleus in northwestern Ecuador (or did historically; see appendices 1 and 2), merits closer comparisons. The two species are virtually indistinguishable externally (see below), but differ in cranial size and proportions. No sexual dimorphism was detected for any cranial measurement within either species (P = 0.18–0.98). In contrast, all raw measurements were significantly different between the two species (P ranges from < 0.0001 to 0.026 for significant comparisons; table 1) except for parietal breadth (P = 0.71). Generally, H. teleus is larger, but the interparietal width averages greater in H. australis. No overlap occurs between specimens (of any age class) of the two species in observed hind foot measurements (range = 35–40 mm for H. teleus, 29–34 mm for H. australis), or for the derived ratio of interparietal width/zygomatic breadth (range = 0.423–0.520 for H. teleus, 0.555–0.689 for H. australis).

Multivariate analyses support the morphological distinctiveness of Heteromys teleus from H. australis. The two species separate clearly along the first principal component of the PCA undertaken on adult specimens (fig. 4), with no overlap between the two. The coefficients and loadings indicate that PC I represents a contrast between overall size and interparietal width (table 2). All measurements but interparietal width load negatively; thus, specimens with low scores on PC I (left portion of fig. 4) have relatively large values for all measurements except interparietal width. In contrast, specimens with high scores (right portion of fig. 4) hold small values for most measurements, but large ones for interparietal width. Species means were highly significantly different on PC I (P < 0.0001), but not on PC II or PC III (P = 0.24–0.97).

In sum, adults of Heteromys teleus show overall larger body size than H. australis and differ in cranial proportions. The hind foot is longer (nonoverlapping measurements regardless of age class) and the zygomatic breadth wider (nonoverlapping measurements for adults in age class 4; table 1). The rostrum is strikingly wider in H. teleus, both relatively and absolutely (although not measured), and the anterior roots of the zygomatic arches are more robust (fig. 2). In addition, the jugular foramina anterolateral to the occipital condyles are larger in H. teleus, and the molars are wider. Finally, regardless of age class, the interparietal is laterally rounded and relatively narrower in the new species, in contrast to the interparietal of H. australis, which is wide and flattened-diamond shaped (at least in this geographic region; figs. 2 and 5).

Both species possess dark gray dorsal pelage, but whereas the dorsal pelage of Heteromys teleus is always harsh and spiny, that of H. australis varies with altitude (spiny in lowlands, softer in highlands; see Anderson, 1999). The length of the hind foot represents the only reliable external character we know of for distinguishing the two species. Even the single available specimen of H. teleus in age class 1 has a larger hind foot (35 mm) than does any specimen of H. australis from Ecuador of any age class (the largest known specimens have a hind foot measuring 34 mm). Nevertheless, definitive identifications require collection of museum voucher specimens, given the difficulty of taking accurate hind foot measurements on live animals and the paucity of available specimens of the genus from Ecuadorian localities in museum collections.

Natural History:

In contrast to some populations of Heteromys currently referred to H. desmarestianus in Central America (e.g., Fleming, 1970, 1974a, 1974b, 1977a, 1977b; Fleming and Brown, 1975; Vandermeer, 1979; Coates-Estrada and Estrada, 1988; Quintero and Sánchez-Cordero, 1989; Martínez-Gallardo and Sánchez-Cordero, 1993; Sánchez-Cordero, 1993; Sánchez-Cordero and Fleming, 1993; Brewer, 2001) or to H. anomalus in Venezuela (Pirlot, 1963; Rood, 1963; Rood and Test, 1968; Valdez et al., 1985), very little ecological information is available for the two species of Heteromys present in Ecuador (but see González-M. and Alberico [1993] for information on H. australis in western Colombia and Anderson [1999] for natural history information on H. australis gleaned from specimen tags and field notes). G.H.H. Tate offered in his field notes of 17 May 1923 at Cerro Manglar Alto this tantalizing report on the species that we now recognize as H. teleus: the Heteromys “use distinct runways, these are well worn and frequently placed about a foot below the top of the steep bank of the quebrada [= stream]. Every specimen but one [at eastern-slope camp] taken in quebrada, the other about 50 yards above in a cornfield. One specimen had a large cockroach in one cheek-pouch. Nocturnal.” On 21 May 1923 at the western camp, he added that the species “is probably distributed in the small watercourses throughout the region. It may be a water rat, but it is certainly a bank-rat, being almost always taken along the banks of the quebradas.” Another specimen had small seeds in its pouches.

DISCUSSION

Ecogeography

Within Ecuador, Heteromys australis inhabits areas wetter than the known localities of H. teleus. Whereas H. australis is restricted to the extreme northwestern portion of the country in very wet forests of the lowland Chocó and on the western versant of the Andes, H. teleus occupies slightly drier (but still evergreen) forests in the southernmost extension of the Chocó, also on the Pacific coastal plain and western slopes of the Andes (fig. 3; see also below and appendix 1). Both species inhabit a wide range of altitudes, from near sea level to approximately 2000 m (H. teleus) or to 1400 m (H. australis) in Ecuador, although the latter species is known to range up to circa 2500 m in parts of Colombia (Anderson, 1999). Lowland localities of H. australis in Ecuador are restricted to areas receiving more than 3000 mm of annual precipitation, but the two piedmont localities of that species lie in regions of between 2000 and 3000 mm of precipitation (IGM, 1995b). Such a shift in tolerances corresponds to lower potential evapotranspiration at higher (and thus cooler) elevations (Holdridge et al., 1971) and agrees with patterns for H. australis throughout the rest of its range (Anderson et al., 2002b). In contrast, northern localities of H. teleus hug the 3000-mm isopleth of annual precipitation, and southern ones enter regions receiving 1300–2000 mm of rainfall or even less (800–1300 mm for Cerro Manglar Alto; IGM, 1995b). However, the vegetation on the summits and high western slopes of this last site is nourished by “horizontal” precipitation in the form of mist and fog coming off the Pacific, making such areas functionally much more mesic than suggested by their direct precipitation (G.H.H. Tate field notes, 1923; see also Foster, 1992b, and Best and Kessler, 1995, for descriptions of similar sites).

This difference in absolute precipitation between collection localities of the two species correlates with seasonality of rainfall. Regions of Ecuador with over approximately 3000 mm in rainfall receive that precipitation relatively uniformly throughout the year (;le 4 dry months; Sierra et al., 1999b) and are categorized as “húmedo tropical” in a broad scheme of climates and potential vegetation (IGM, 1995a). All localities of Heteromys australis in Ecuador correspond to this category. In contrast, areas receiving less rainfall in western Ecuador experience a strong dry season (IGM, 1995a, 1995b; Lynch and Duellman, 1997; see also Sierra et al., 1999a, 1999b). All but one locality of H. teleus fall into or very near areas of the “tropical monzón” category, with a pronounced dry season from June to November (IGM, 1995a). The one exception is Cerro Manglar Alto, which has a more stable climate due to coastal mists and fog. Hence, the macrogeographic distributions of these species in western Ecuador appear to correspond to climatic factors.

Current faunal and floral classifications of the region do not distinguish between areas corresponding to the respective distributions of Heteromys australis and H. teleus. Together, their distributions generally correspond to the “tropical noroccidental” and western “subtropical” zoogeographic regions of Albuja-V. et al. (1980; see also Albuja-V., 1999), the Colombian-Pacific Fauna of Chapman (1926), and the “humid tropical,” “humid subtropical,” and “subhumid tropical” regimes of Lynch and Duellman (1997). However, the individual distributions of the two species do not coincide with demarcations within the former systems, or with categories in a recent classification of vegetation types in the country (Sierra et al., 1999b; see also supporting documentation in Sierra et al., 1999a). While Sierra et al. (1999b) recognized 46 vegetational formations, they did not make a distinction between the very wet, relatively unseasonal areas of evergreen forest where H. australis is found and the slightly drier, often highly seasonal (but still evergreen) regions characteristic of H. teleus. Both species have been collected in (or within 3 km of, given potential error in our geographic coordinates) regions classified as “bosque siempreverde de tierras bajas,” “bosque siempreverde piemontano,” and “bosque simpreverde montano bajo.” In addition, H. teleus also is documented from areas of “bosque de neblina montano” (Bosque Integral Otonga, the highest record for the species in the country) and “bosque de neblina montano bajo” (Cerro Manglar Alto, which is aberrantly mesic due to fog and mist off the ocean). This lack of predictive information regarding the individual species' distributions in these phytogeographic, zoogeographic, or bioclimatic classification systems highlights the value of an independent modeling approach for each species in faunal mapping projects (Peterson et al., 1999, 2002) rather than a vegetation-surrogate method.

Conservation Status

Data provided by collectors indicate that both Heteromys teleus and H. australis have been captured in disturbed as well as primary forests (appendix 1). The disturbed forests vary in the forms of low-intensity human use, but they include systems that contain small cultivated plots, that are occasionally entered by free-ranging livestock, or from which wood has been selectively extracted by local inhabitants. No Ecuadorian collection locality of either species derives from a truly secondary site (recovering from being clearcut recently), deciduous forest, or open, nonforested area.

Six of the seven localities where Heteromys teleus has been collected now correspond to landscapes where most of the natural vegetation has been removed, and the other one (Bosque Integral Otonga) lies within 3 km of nonnatural vegetation (Sierra, 1999; our geographic coordinates have accuracy of approximately 3 km). Most of the original vegetation on the coastal plain of central-western Ecuador has been converted to agriculture, especially plantations of bananas and oil palms (Parker and Carr, 1992; Best and Kessler, 1995). Recent collections of H. teleus are from very small reserves containing relictual forest patches within the agricultural matrix (see appendices 1 and 2). As is typical elsewhere in the Neotropics, inventories of small mammals in Ecuador in past few decades have focused on areas of remaining natural or seminatural vegetation, and sampling effort for H. teleus in agricultural areas is inadequate. However, species of the genus Heteromys characteristically inhabit closed-canopy forests (Sánchez-Cordero and Fleming, 1993; Anderson, 1999), although H. anomalus also occurs in low-intensity or “subsistence” agricultural areas but not in high-intensity, large-scale industrial croplands (Anderson, 1999; Utrera et al., 2000). Hence, while the presence of H. teleus in areas of agricultural plantations in western Ecuador is unlikely, further collecting efforts are needed to establish the status of populations of native mammals there (emulating Cartaya and Aguilera, 1985; Martino and Aguilera, 1993; Utrera et al., 2000).

The long-term survival of Heteromys teleus in small, isolated forest remnants is unlikely (Harrison, 1991). Some larger tracts of forest remain along the western slope of the Andes (Sierra, 1999), but only one locality of the new species is from such high elevations (Bosque Integral Otonga). Among habitats that we consider typical for the species, the largest areas of intact forest probably lie in the Reserva Ecológica Mache-Chindul, which holds forest from near sea level to circa 700 m in the northern portion of the coastal cordillera (Sierra, 1999), but H. teleus has not yet been collected there. The other large protected area within the species' likely range is the Parque Nacional Machalilla, in the Cordillera de Chongón-Colonche (southern portion of the Cordillera de la Costa). However, much of this national park is deciduous forest not suitable for the species, and the park has been heavily logged and grazed by goats and cattle (L. Emmons, in litt. January 2002). In addition, all national parks and other governmentally protected areas in Ecuador face encroachment and various human uses, and the future of currently intact habitat at Mache-Chindul cannot be taken for granted. Like most rodents, no species of Heteromys was evaluated for the Libro rojo de los mamíferos del Ecuador (Tirira-S., 2001), but the limited range suggested by known localities indicates that H. teleus should be evaluated for potential listing.

In contrast, relatively large expanses of unseasonal wet forests remain in extreme northwestern Ecuador, in the range of Heteromys australis (Sierra, 1999). Because of extreme rainfall, these regions are not as suitable for agriculture, but timber is currently being exploited for large-scale industrial use (Parker and Carr, 1992; see also Best and Kessler, 1995). Seven of the 12 Ecuadorian localities of H. australis currently hold natural vegetation (Sierra, 1999), although they are within 3 km of areas converted to agriculture. The others lie in areas now deforested (Sierra, 1999), but which held forests at the dates of collection no more than 15 years ago (1987–1996; see appendix 1). The Reserva Cotacachi-Cayapas protects extensive premontane forests in northwestern Ecuador, but no large reserve exists in western Ecuador in unseasonal wet lowland rainforest. While expanses of suitable habitat remain in the Pacific lowlands of western Colombia (IAvH and Etter-R., 1998), little is protected in national parks (Garcés-Guerrero and De la Zerda-Lerner, 1994). Overall, the Andean distribution of H. australis has become highly fragmented by deforestation in the past 50 years, but its lowland distribution appears more intact (Anderson, 1999).

Biogeography of Heteromys in South America

The occurrence of Heteromys only west of the Andes apparently results from the northern origin of the group and contingent, historical factors related to its subsequent colonization of South America (Hershkovitz, 1972; Anderson, 1999). The existence of 19 collection localities west of the Andes in Ecuador, with none known from the eastern slopes or Amazonian lowlands, makes the presence of any species of Heteromys east of the Andes in Ecuador unlikely. Albuja-V. (1999) presented a relatively uniform distribution of collection localities for bats east and west of the Andes in the country. Thus, because patterns of fieldwork for small rodents in the northern Neotropics are generally similar to those for bats, the genus' absence east of the Andes is not likely an artifact of biased sampling. Predictive models of the potential geographic distribution of H. australis indicated suitable environmental conditions for the species along both eastern and western flanks of the Andes in Ecuador (Anderson et al., 2002a). Heteromys australis is not known from the eastern slopes of the Andes in Colombia either, although the species has been recorded for the headwaters of the Río Magdalena near the crest of the Cordillera Oriental in southern Colombia. Its evident failure to cross this low pass, in seemingly suitable habitat, to colonize the eastern versant of the Andes of Colombia and Ecuador remains mysterious (Chapman, 1926: 44; Anderson, 1999; Anderson et al., 2002a).

Heteromys classically has been considered one of the latest North American groups to disperse to South America, with only a minimal level of differentiation on that continent (Hershkovitz, 1972; Simpson, 1980; Marshall et al., 1982; Webb, 1997). While pocket gophers (Geomyidae), which represent the sister group to heteromyids, have dispersed only to extreme northwestern South America west of the Río Atrato/Río San Juan lowlands or Bolívar Geosyncline (Alberico, 1990), heteromyids are distributed much more extensively (Handley, 1976; Anderson, 1999). Although the timing of their entry into South America is controversial, recent alpha-systematic (species-level) and geographic-modeling research has shown that Heteromys distributions in South America are more complex than previously recognized (Anderson, 1999; Anderson and Soriano, 1999; Anderson et al., 2002a). The present discovery of a new species endemic to western Ecuador adds a new level of complexity to the emerging understanding of the group's differentiation in South America. Evolutionary relationships among species of the genus remain unresolved (Rogers, 1990), but a phylogenetic study of the subfamily Heteromyinae using morphological characters is underway (Anderson, unpubl.). Furthermore, other undescribed species of Heteromys exist in South America (Handley, 1976; Anderson, 1999). Thus, an overall account of heteromyid diversity and evolution in South America is not yet possible.

Faunal Relationships and Endemism

Under various names, the faunal turnover in western Ecuador between Chocoan rainforest elements and Tumbesan species characteristic of southern xeric areas has been long established (Chapman, 1926; Albuja-V. et al., 1980; Cracraft, 1985). Many mammalian species characteristic of trans-Andean rainforests reach their southernmost geographic distribution in Ecuador west of the Andes (e.g., primates Alouatta palliata and Cebus capuchinus; rodents Microsciurus mimulus and Oryzomys bolivaris; and the bat Rhinophylla alethina; Emmons, 1997; Tirira-S., 1999). Most such trans-Andean rainforest species are closely related to Amazonian species in the same genus, and—at least following current taxonomy—many other rainforest species are found on both sides of the Andes (Emmons, 1997; Tirira-S., 1999). In contrast, Heteromys and Tylomys (naked-tailed climbing rats) represent the only mammalian clades of North American origin to reach their southern termination in western Ecuador (Hershkovitz, 1972; Tirira-S., 1999; note that Reithrodontomys, harvest mice, and Cryptotis, shrews, also represent diverse North American clades with marginal ranges in South America, but there they have Andean distributions that terminate in montane Ecuador and Peru, respectively; Hooper, 1952; Vivar et al., 1997).

Further, Heteromys teleus holds a unique position among mammals endemic to Ecuador. The overwhelming majority of Ecuadorian endemics are rodents, and H. teleus is the only one of them found on the western coastal plain. Among the 27 terrestrial mammals listed as endemic to the country by Tirira-S. (1999), most are montane (13; found only over 1000 m) or from the Galápagos islands (8). Two others (Proechimys gularis and Oryzomys tatei; rodents) are Amazonian, one is of unknown locality within the country (Bolomys punctulatus, a rodent; see Voss, 1991), and the remainder (3) are from west of the Andes. Of the three western species (all bats), one remains undescribed (Sturnira sp. A of Tirira-S., 1999), and one has since been reported for southwestern Colombia (Balantiopteryx infusca; Alberico et al., 2000; see also McCarthy et al., 2000). The last (Cabreramops [= Molossops] aequatorianus) is a rare, high-flying bat only known from two localities in central-western Ecuador (Ibáñez, 1980; Tirira-S., 1999: 138). That species' distribution and evolutionary history remain enigmatic, but it appears to inhabit areas drier than those typical of H. teleus, including deciduous forests where H. teleus is not known. Albuja-V. (1999) also mentioned undescribed species of the chiropteran genera Lonchophylla and Sturnira from extreme northwestern regions of the country, areas wetter than those inhabited by H. teleus. Hence, H. teleus may represent the only mammalian species restricted to seasonal evergreen forests of the southern Chocó. However, Emmons and Albuja (1992) mentioned the possible species-level status of a brocket deer (currently Mazama americana fuscata) and a white-fronted capuchin (currently Cebus albifrons aequatorialis) in central-western Ecuador. The taxonomic status of those forms has not yet been resolved, but they do show distributional patterns similar to that of H. teleus.

We know of no formalized faunal division within Ecuador corresponding to the distributions of Heteromys australis and H. teleus, but several studies hint that may be the case. Several authors have divided the Chocó into northern and southern faunal regions for mammals and herpetofauna, with the division present at approximately Buenaventura in central-western Colombia (e.g., Lynch, 1979; Hernández-Camacho et al., 1992). This division is much less marked than that recognized for birds between the Chocó and the Nechí region north of the Andes in Colombia (Cracraft, 1985). In contrast to these northern divisions for faunal groups, Gentry (1992: 57) was the first to recognize western Ecuador south of the town of Esmeraldas as a “unique and distinctive floristic region for wet and moist forest … rather than as the tail-end of the Chocó flora.” Although not formalized as a faunal division, distributional maps for rainforest lizards and snakes suggest a similar pattern (Dixon, 1979). Furthermore, Almendáriz and Carr (1992) state that 42% of the species of herpetofauna that they collected in western Ecuador were endemic to that region, including some species characteristic of moist forests.

Nevertheless, as mentioned by Emmons and Albuja (1992), Hernández-Camacho et al. (1992), Voss and Emmons (1996), and Cadena et al. (1998), vast regions of the Chocó remain unsampled or inadequately sampled, and taxonomic understanding of small mammals (and most other taxonomic groups) of this and other adjacent regions remains woefully incomplete. Thus, fine faunal divisions and analyses of endemism remain premature and dependent upon scattered and incomplete collection localities, as well as inadequate alpha-taxonomic work. Future inventories and comprehensive taxonomic revisions will undoubtedly bring to light currently unrecognized species and refine our understanding of the true distributional patterns of small mammals in the region, which in most cases can now only be imperfectly approximated.