Bullimus is 1 of 15 genera of murid rodents endemic to the oceanic portion of the Philippines (i.e., excluding Palawan and other islands on the Sunda Shelf). Two species have been recognized previously: B. bagobus, widespread on Mindanao and on islands that were connected to it during Pleistocene periods of lower sea level and B. luzonicus, occurring only on Luzon. Recent surveys have revealed a 3rd species endemic to the small island of Camiguin, located just north of Mindanao and isolated from that large island by deep water. Multivariate analyses support the recognition of these 3 species. The new species of Bullimus from Camiguin is distinguishable from congeners by its smaller size, soft and uniformly dark pelage, inflated braincase, shorter palate and incisive foramina, more divergent molar toothrows, and other cranial and dental features. The patterns of variation and island distribution in Bullimus are consistent with our understanding of regional geography; the 3 species occur on separate islands or island groups that remained isolated during periods of lower sea level. The discovery of a species endemic to Camiguin reflects the fact that this small island is close enough to Mindanao to have received some nonvolant mammals by overwater dispersal, but its isolation has been sufficient to promote speciation.
The murine genus Bullimus includes large-bodied rats native to the Philippines that have relatively short, bicolored tails, long hind feet, elongate heads, and distinctive morphology of the auditory bullae (Musser and Heaney 1992). As originally described by Mearns (1905), Bullimus was only vaguely defined and was considered a subgenus of Rattus until Musser (1982) and Musser and Newcomb (1983) reestablished its generic status. The genus was rediagnosed by Musser and Heaney (1992) who noted that the full extent of individual and insular variation remained to be investigated but provisionally recognized 2 species: Bullimus bagobusMearns (1905) and B. luzonicus (Thomas 1895). The former occurs on Mindanao and on several adjacent islands that were connected to it during the Pleistocene. The latter is known only from Luzon Island (Heaney et al. 1998).
Heaney (1984) reported the possible occurrence of B. bagobus on Camiguin, a small (265 km2) volcanic island located 8 km north of Mindanao. This record was based on the skull of a juvenile animal collected in the late 1960s that had been mismatched with the skin of a juvenile Rattus tanezumi (Heaney 1984—specimens identified and annotated by G. G. Musser). Recent field surveys since have produced a large series of Bullimus from Camiguin that represent a new species endemic to that island (Heaney et al. 1998). In this paper we describe the Camiguin Bullimus, and to place the new species in proper context, we also evaluate morphological variation within the genus.
Materials and Methods
Specimens examined in this study are housed at the Delaware Museum of Natural History, Greenville (DMNH); Florida Museum of Natural History, University of Florida, Gainesville (UF); The Field Museum, Chicago (FMNH); United States National Museum of Natural History, Washington, D.C. (USNM); and Mindanao State University—Illigan Institute of Technology, Illigan City, Philippines (uncataloged specimens designated by field numbers of Tabaranza, BRT). Material examined included specimens prepared as museum skins with skulls, skins with incomplete skeletons, complete skeletons, skulls only, and formalin-fixed specimens stored in 70% ethyl alcohol (some with skulls subsequently removed and cleaned). The following samples were examined: Bullimus bagobus—BOHOL ISLAND: Bohol Province, Sierra Bullones, Sandayoong, 1,500 feet (about 460 m) elev. (FMNH 87479, 87511); 1 km S, 1 km E Bilar, 320 m elev. (USNM 459889). DINAGAT ISLAND: Surigao del Norte Province, Loreto Municipality, Balitbiton (DMNH 4585–4588); Kambinlio (DMNH 4583); Paragua (DMNH 4584); Plaridel Municipality, Albor (DMNH 4589). LEYTE ISLAND: Leyte Province, Baybay Municipality, Mt. Pangasugan, 300–500 m elev. (USNM 458785, 458788, 458789, 459890); Burauen Municipality, Mt. Lobi Range, Bario Buri (DMNH 4091, 4093); Bario Tambis (DMNH 4087–4089, 4092, 4098, 4099, 4104–4108, 4781, 4782). MARIPIPI ISLAND: Leyte Province, 2 km N, 3 km W Maripipi town, 600–800 m elev. (USNM 458791, 458792, 459926). MINDANAO ISLAND: Agusan del Norte Province, Mt. Hilong-hilong (DMNH 5845, 5850–5852, 5954–5956, 5859, 5861, 5862, 5868, 5869); Bukidnon Province, Sumilao Municipality (FMNH 167388–167390); Davao del Sur Province, east slope Mt. McKinley, 2,800 feet (about 850 m) elev. (FMNH 56205); east slope Mt. Apo, Todaya, 2,500 feet (about 750 m) elev. (FMNH 61481, 61482); Lanao del Norte Province, Mahayahay, Mainit, Iligin City (DMNH 5894); Misamis Occidental Province, Mt. Malindang, Duminagat (FMNH 87483–87485); Surigao del Sur Province, Sibajay, Lanuza (DMNH 5871–5880, 5882–5886, 5899); Zamboanga del Norte Province, Katipunan, Sigayan (FMNH 67807). SIARGAO ISLAND: Surigao del Norte Province, Osmeña, Dapa (DMNH 4382). Bullimus luzonicus—LUZON ISLAND: Benguet Province, Mt. Data, 7,500 feet (about 2,300 m) elev. (FMNH 62298–62303, 62348, 62349); Kalinga Province, Balbalan Municipality, Balbalasang, Mapga, 1,200 m (FMNH 169127); Magdallao, 1,600 m (FMNH 167310); Am-licao, 1,900 m elev. (FMNH 169128, 169129); Camarines Sur Province, Caramoan Municipality, Kasine Mtn. (UF 30109); Cabangon Mtn. near Tina'go (UF 30110–30112). Bullimus n. sp.—Holotype and referred specimens from CAMIGUIN ISLAND (see below).
Specimens were assigned to age categories as defined by Musser and Heaney (1992) based on relative body size, pelage characteristics, and molar tooth wear. Adult specimens of both sexes were pooled in statistical analyses. Descriptive terminology for external features of the head and limbs follows Brown (1971) and Brown and Yalden (1973). Terminology for cranial and dental features follows Musser and Heaney (1992). Scanning electron micrographs of molar teeth were made from uncoated specimens.
Values for external dimensions of total length, length of tail (LT), length of hind foot, including claws (LHF), length of ear from notch (LE), and weight (WT) were taken from field notes of collectors or from specimen tags. Hind foot length was measured directly on some specimens. Length of head and body (LHB) was determined by subtracting length of tail from total length. The number of scale rings per centimeter (TSR) was counted at a point on the tail one-third the distance from the base, lengths of over-fur (LOF) and guard hairs (LGH) were determined from measurements taken middorsally (Musser and Heaney 1992).
On adult specimens, 25 cranial, mandibular, and dental measurements were made to the nearest 0.1 mm using dial calipers (Musser 1979, 1982; Musser and Heaney 1992): occipitonasal length (ONL), interorbital breadth (IB), zygomatic breadth (ZB), breadth of braincase (BBC), height of braincase (HBC), length of nasal bones (LN), length of rostrum (LR), breadth of rostrum (BR), breadth of zygomatic plate (BZP), length of diastema (LD), palatal length (PL), postpalatal length (PPL), length of incisive foramina (LIF), breadth across incisive foramina (BIF), length of bony palate (LBP), breadth of palate at M1 (PBM1), breadth of palate at M3 (PBM3), breadth of mesopterygoid fossa (BMF), length of auditory bulla (LB), alveolar length of maxillary molar row (ALM1–3), crown breadth of M1 (BM1), breadth of upper incisor (BI), depth of upper incisor (DI), length of mandible (LM), height of mandible (HM). Molar measurements only were taken on subadults with erupted permanent dentition. Proportional relationships among taxa are illustrated by ratio diagrams comparing log10-transformed means ±2 SE of cranial measurements (Musser and Heaney 1992).
Statistical analyses were performed using SYSTAT 10 for Windows (SPSS, Inc. 2000). Descriptive statistics (mean, standard deviation, and observed range) were calculated for sample groups. Multivariate analyses were conducted on log10-transformed cranial measurements from adult specimens. A subset of 21 measurements was used to increase sample sizes by including some specimens with minimally damaged crania. Multivariate analysis of variance was used to test for significant differences between taxa. Phenetic variation and group distinctiveness were assessed through principal components analysis (using a correlation matrix) and discriminant function analysis (using backward stepwise estimation and jackknifed classification).
External, cranial, and dental measurements for population samples of Bullimus from throughout the Philippines indicate that apart from differences in overall size, the 3 putative species share the same general body form and have similar limb and tail proportions (Table 1; Appendices I and II). B. bagobus, on an average the largest in body size of the 3 species, exhibits geographic variation in size, with the largest individuals from north central, western, and southern Mindanao and the smallest from northeast Mindanao and from a series of smaller islands north of Mindanao (Fig. 1; Appendix I). The new species from Camiguin Island is smallest in most dimensions, and B. luzonicus is intermediate in size (Table 1; Appendix II). Differences in pelage thickness and coloration and in skin pigmentation are addressed in the separate species accounts.
A principal components analysis was conducted on 21 cranial and dental measurements from 43 adults. The first 4 components had eigenvalues of 11.95, 1.88, 1.47, and 1.10, respectively, accounting for more than 78% of the total variance (56.9, 9.0, 7.0, and 5.2, respectively). High positive loadings of most characters on the 1st axis indicate that this component was related principally to size. The 2nd component separated individuals based on interorbital breadth, alveolar length of maxillary molar row, and crown breadth of M1 (loadings > 0.50). The 3rd component separated individuals on the basis of breadth of palate at M3 and breadth of mesopteryoid fossa (loadings > 0.55) and the 4th on breadth of braincase, length of bony palate, and length of auditory bulla (loadings > 0.43). In a bivariate plot of components 1 and 2 (Fig. 2A), the B. luzonicus and Camiguin Island samples overlap broadly, but both are well separated from B. bagobus along the 1st axis. A plot of components 3 and 4 (Fig. 2B) separates the Camiguin and B. luzonicus samples, but both overlap B. bagobus. Multivariate analysis of variance revealed highly significant differences between the 3 taxa (Wilks' lambda = 0.007, F = 10.26, d.f. = 42, 40, P < 0.001).
To further assess variation, we conducted discriminant function analyses of cranial measurement data. In a 1st analysis, specimens were grouped into 8 population samples from 6 islands (Fig. 1): Camiguin (n = 4), northern Luzon (n = 3), southern Luzon (n = 3), southern and northeastern Mindanao (n = 19), western Mindanao (n = 3), Dinagat (n = 4), Leyte (n = 5), and Maripipi (n = 2). Highly significant differences were detected between population samples (Wilks' lambda = 0.001, F= 6.17, d.f. = 77, 157, P < 0.001). The first 3 canonical variates had, respectively, eigenvalues of 13.63, 5.72, and 4.81, canonical correlations of 0.97, 0.92, and 0.91, and accounted for 50.2%, 21.0%, and 17.8% of the total variance. The 1st axis was most strongly weighted by length of auditory bulla (1.47), breadth of palate at M1 (1.36), breadth across incisive foramina (−0.51), length of incisive foramina (−0.60), and breadth of mesopterygoid fossa (−0.84). The 2nd axis was weighted strongly by palatal length (3.26), breadth across incisive foramina (1.10), interorbital breadth (−0.65), breadth of mesopterygoid fossa (−0.71), length of bony palate (−1.00), length of mandible (−1.28), and length of incisive foramina (−1.66). A posteriori jackknifed classification correctly identified all specimens from Camiguin and from northern Luzon. One specimen from southern Luzon was classified with northern Luzon. The 2 sample groups from Mindanao and those from 3 smaller islands north of Mindanao (Dinagat, Leyte, and Maripipi) had misclassification rates between 50% and 80%, but in every instance, misclassified specimens were placed with samples from other islands within this same group. The 2 Luzon samples overlap but are separate from the samples from Mindanao and the associated smaller islands, all of which overlap. The Camiguin sample is well separated from the other groups. These relationships are illustrated in a plot of the group centroids and 95% confidence ellipses onto the first 2 canonical variates (Fig. 2C).
A 2nd discriminant function analysis grouping specimens into 3 taxa also revealed highly significant differences (Wilks' lambda = 0.022, F = 20.56, d.f. = 18, 64, P < 0.001). The 2 canonical variates had, respectively, eigenvalues of 10.57 and 2.98, canonical correlations of 0.96 and 0.87, and accounted for 78% and 22% of the total variance. The 1st variate was weighted by length of auditory bulla (1.59), breadth of palate at M1 (1.37), breadth of zygomatic plate (−0.52), breadth across incisive foramina (−0.73), and breadth of mesopterygoid fossa (−0.80). The 2nd variate was defined by length of incisive foramina (1.38), crown breadth of M1 (0.55), length of diastema (−0.75), and interorbital breadth (−0.86). A posteriori jackknifed classification correctly allocated all 43 specimens. Projection of individual specimen scores onto the first 2 variates shows 3 distinct clusters (Fig. 2D).
On the basis of these results, we recognize 3 species of Bullimus, as summarized in the following accounts.
Genus BullimusMearns, 1905
Bullimus bagobusMearns (1905:450).
A genus of Muridae in the subfamily Murinae (Musser and Carleton 1993) distinguished by the following combination of characters: (1) tail substantially shorter than the combined head and body length; (2) hind feet long and moderately wide, plantar pads small, including a small hypothenar; (3) females with 4 pairs of mammae, 1 postaxial, 1 abdominal, and 2 inguinal; (4) skull with elongated rostrum, wide braincase, moderate ridges on superorbital margins, weak postorbital and temporal ridges, short occiput, and relatively small interparietal; (5) wide zygomatic plate and deep zygomatic notch, posterior margin of the plate even with the middle of the 1st molar; (6) squamosal intact, without a subsquamosal fenestra; (7) no lateral alisphenoid strut, foramen ovale accessorius coalesced with masticatory and buccinator foramina; (8) a large stapedial foramen and a groove on the posterolateral margin of the pterygoid plate for the infraorbital branch of the stapedial artery; (9) auditory bulla inflated medially and ventrally but concave on lower lateral surface, the dorsal lateral surface with a large ectotympanic shield causing the external auditory meatus to be oriented caudad, posterior lamina of ectotympanic forming a wide flange concealing a portion of the incus; (10) bony palate wide and moderately long, palatine grooves well defined, posterior palatine foramina at a level near the gap between the 2nd and 3rd molars; (11) upper and lower molars strongly hypsodont and large relative to the palatal area; (12) posterior cingulum absent on upper molars, cusp t3 small on 1st upper molar, usually absent from 2nd and 3rd upper molars; (13) anteroconid on 1st lower molar composed of prominent anterolabial and anterolingual cusps and no anterocentral cusp (Musser and Heaney 1992).
Content and distribution
Bullimus includes 3 species confined to the oceanic portion of the Philippine Archipelago (i.e., excluding Palawan and smaller islands on the Sunda Shelf).
Bullimus bagobus Mearns, 1905Large Mindanao forest rat; Bagobo rat
1905. Bullimus bagobus Mearns, Proceedings of the United States National Museum 28:450.
1946. Rattus bagobus barkeri Johnson, Journal of the Washington Academy of Science 36:318.
1952. Rattus rabori Sanborn, Fieldiana: Zoology 33:130.
Restricted to the Mindanao faunal region where it is widespread (Fig. 1). Records from Bohol, Dinagat, Leyte, Maripipi, Mindanao, and Siargao islands, at elevations ranging from 200 to 1,800 m (Heaney et al. 1998).
Table 1, Appendix I.
The brief, original description of this species (and the genus) was based on a single specimen from Mt. Apo, Davao Province, southern Mindanao. Additional specimens from the type region were described by Sanborn (1952). Specimens from Samar and the adjacent islet of Calicoan were described as a separate subspecies, barkeri (Johnson 1946). Sanborn (1952) described rabori as a separate species from Zamboanga, western Mindanao. There is some geographic variation in body size; the largest specimens, (designated as rabori) are from western Mindanao and the smallest from northeastern Mindanao and the series of smaller islands north of Mindanao (Appendix I—Musser 1982:89, figure 57). However, discriminant function analysis of cranial measurements from adult specimens reveals broad overlap among the sample groups from these areas (Fig. 2C). Pelage coloration and skin pigmentation, features thought to distinguish barkeri and rabori (Johnson 1946; Sanborn 1952), vary among individuals within local or regional samples and therefore do not appear to be diagnostic. From our analysis, there is not sufficient evidence to recognize either barkeri or rabori as separate entities.
Bullimus luzonicus (Thomas, 1895)Large Luzon forest rat
1895. Mus luzonicus Thomas, Annals and Magazine of Natural History, Series 6, 16:1.
Known only from scattered localities on Luzon Island (Kalinga, Benguet, Aurora, and Camarines Sur provinces), at elevations ranging from 200 to 2,400 m (Heaney et al. 1998, and collection records of Field Museum of Natural History).
Table 1, Appendix II.
This species was described by Thomas (1895, 1898), based on a small sample from Mt. Data in the Central Cordillera of northern Luzon. Sanborn (1952) and Rabor (1955) briefly commented on additional specimens from this region. Available specimens from northern and southern Luzon are comparable in size (Table 1; Appendix II); in discriminant function analysis of cranial measurements, the 2 sample groups overlap (Fig. 2C). B. luzonicus is smaller than B. bagobus in most external and cranial dimensions (Table 1; Appendices I and II). The 2 species share similar cranial proportions although the auditory bulla is less inflated and the rostrum is relatively shorter in B. luzonicus (Figs. 3A and 4). Discriminant function analyses clearly separate the 2 taxa (Fig. 2D), with heavy weighting on length of auditory bulla and breadth of palate at M1. Both B. bagobus and B. luzonicus have pelage that is relatively thin and harsh textured, with a strong contrast between the dark dorsum and the pale venter. Additional comparisons are made in the following account of the new species from Camiguin Island.
Bullimus gamay, new species
Adult female, FMNH 154882, collected 20 March 1995 (field number 5383 of L. R. Heaney). Specimen initially fixed in formalin, body preserved in 70% ethyl alcohol with skull subsequently removed and cleaned. Tissue samples (liver and muscle) are deposited at FMNH. There is an incision in the abdomen, and the right hamular process of the skull is broken, but otherwise the specimen is in excellent condition.
Mt. Timpoong, 2 km N, 6.5 km W Mahinog, Camiguin Province, Camiguin Island, Philippines, 1,275 m elev., 9°11′N, 124°43′E.
Referred specimens and localities
In addition to the holotype, 21 other examples of B. gamay are known. Nineteen specimens collected in 1994 and 1995 were prepared as complete skeletons or fixed in formalin and preserved in 70% ethyl alcohol. In May 1994, specimens were collected from Barangay Kital-is, Sagay Municipality, 6.5 km W Mahinog, 900 m, 9°9′N, 124°43′E (BRT 90, 97, 116, 156), 0.5 km N, 6.5 km W Mahinog, 1,200 m, 9°9′N, 124°43′E (FMNH 167885, BRT 178), and 1 km N, 7.5 km W Mahinog, 1,400 m, 9°9′N, 124°43′E (FMNH 167884, 167886, 167887). In March 1995, additional specimens were taken at the type locality (FMNH 154821–154823, FMNH 154877–154881, 154883) and at 2.5 km N, 6.5 km W Mahinog, 1,475 m, 9°11′N, 124°43′E (FMNH 154884).
Two additional specimens of B. gamay, housed at DMNH, are part of a series of mammals collected on Camiguin Island in 1968 and 1969 by D. S. Rabor and associates from Mindanao State University. They consist of crania and mandibles that mistakenly were associated with skins of other taxa and subsequently identified by G. G. Musser as B. bagobus. Heaney (1984) reported one of these specimens, a juvenile with partially erupted molars that is mismatched with the skin of a juvenile R. tanezumi (DMNH 4124). The 2nd specimen, mismatched with the skin of a female R. everetti (DMNH 4173), is a subadult with fully erupted molars. There are no locality data directly associated with the skulls, but they may have been collected from the same locality as the mismatched skins, both of which were taken 17–18 June 1968 at Kasang-sangan (or “Casangsangan”), Catarman Municipality. Heaney (1984) gives an elevation range of 1,000–2,000 feet (300–600 m) for this locality.
Bullimus gamay is known only from Camiguin, a small (265 km2) island located in the Bohol Sea 8 km north of north-central Mindanao (Fig. 1). The species has been collected in forest habitat at elevations between 900 and 1,475 m (Heaney and Tabaranza 1997; Heaney et al. 1998). It may occur (or have occurred formerly) in forested habitat at lower elevations.
Table 1, Appendix II.
In Cebuano, the dominant language of the central Philippines, gamay (pronounced “gah-my” with the accent on the 2nd syllable) means little or small sized (Cabonce 1983). It refers to the relatively small body size of the new species. “Camiguin forest rat” is proposed as the English common name.
A species of Bullimus grouped with other members of the genus and distinguished from other Philippine murines by the combination of characters enumerated in the diagnosis of the genus. The following characters distinguish the new species from congeners (with characteristics of the latter in parentheses): (1) body size smaller (other species generally larger); (2) pelage soft and thick, uniformly dark reddish brown, venter only slightly paler than dorsum (pelage thin and coarse, sharply contrasting dark dorsal and pale ventral color); (3) skull with a relatively broad interorbit and inflated braincase (interorbit narrow, braincase less inflated); (4) anterior zygomatic spine inflected medially (spine projecting more directly forward); (5) incisive foramina relatively short, not extending to the anterior margins of the 1st molars and substantially shorter than the molar toothrows (foramina longer, projecting beyond the anterior margin of 1st molar and as long as or longer than toothrows); (6) posterior margin of bony palate located anterior to the posterior margins of the 3rd molars (palate extending beyond the posterior margins of the molars); (7) anterior portion of palate substantially narrower than posterior portion so that molar toothrows are divergent (palatal breadth more uniform, molar toothrows more nearly parallel); (8) mesopterygoid fossa wide (fossa narrow); (9) upper and lower 3rd molars smaller relative to other molars (3rd molars relatively large).
Description and comparisons
Bullimus gamay (Fig. 5) is smaller than other members of the genus, as reflected in both external and cranial measurements (Table 1; Appendices I and II). Proportional differences in cranial and dental measurements of B. gamay and those of a sample of B. bagobus from northeastern Mindanao Island are illustrated in a ratio diagram (Fig. 3B).
The new species has thick, soft pelage. Individual over-fur hairs on the middorsum are up to 20 mm long and are dark gray for three-fourths of length from base with reddish brown tips. Dorsal guard hairs are up to 30 mm long and uniformly dark gray or black. Dorsal coloration is dark reddish brown, darkest middorsally and on the rump, and paler laterally because of fewer guard hairs and a scattering of silvery over-fur hairs. There is no sharp color transition between the dorsum and venter, but the underparts are slightly paler reddish brown with silver highlights because of interspersed silvery hairs. Ventral color is darkest on the chest and palest in the urogenital region. Some specimens have a small white pectoral patch. Relatively dense reddish brown fur extends along the lateral surfaces of the limbs to the wrists and ankles. Along the medial surfaces of the limbs, the hair is paler and sparser. General body coloration of B. gamay differs from B. bagobus or B. luzonicus, which have dark dorsal pelage that sharply contrasts with underparts that range from yellowish gray to white.
The snout and sides of the face are clothed in dark gray hairs. The lips and rhinarium are unpigmented with a scattering of short, pale gray hairs. There is no eye ring. Mystacial and superciliary vibrissae are relatively long (extending back beyond ears) and are dark brownish gray, except for the mystacials closest to the mouth, which are silver. Ear pinnae are similar to those of other species of Bullimus, but are substantially smaller (Table 1). They have pale brown pigmentation (pale gray in preserved specimens) and are nearly naked on both surfaces, except for a scattering of very short gray hairs.
As in other species of Bullimus, the tail is substantially shorter than the combined lengths of the head and body (Table 1). The basal portion (one-half to two-thirds) is covered with brownish gray scales that are uniformly distributed on the dorsal and ventral surfaces in some specimens or with a slightly mottled pattern of unpigmented scales on the ventral surface in others. The distal portion is covered entirely with unpigmented scales. Scales are moderate sized (8 rows per cm in adults), and each has 3 very short hairs (<1 mm) that are silvery gray on the pigmented zone and translucent white in unpigmented areas. Tail pigmentation and scale pattern of B. bagobus and B. luzonicus are similar, but in both species the scales are smaller than in B. gamay (Table 1), and in B. bagobus the pigmented regions are darker gray or black.
Relative size, shape, pigment patterns, and pilosity of the feet are similar among all species of Bullimus. Front feet of B. gamay are large relative to the body size and have strong digits. The hind feet are long and moderately wide, with elongate digits. The pollex bears a flat, pale colored nail. All other digits have pale claws that are long, thick, and relatively straight. On dorsal surfaces of both the fore and hind feet, pale brown pigmentation (gray in preserved specimens) extends from wrist and ankle across the metapodial surfaces to the basal portion of the digits, which are unpigmented distally. Hairs on the dorsal surface are silver, except for the medial portion of the metapodial region where there are scattered darker hairs. Scattered hairs extend along dorsal surfaces of the digit and form very sparse ungual tufts at the base of each claw. On the palmar surface, there are 3 relatively small interdigital pads and a thenar and hypothenar that are large and nearly equal in size. The base of the hypothenar is dark brown, but otherwise the palmar surface is unpigmented and entirely naked. The plantar surface is naked, with dark brown pigment from the heel to the base of the digits, which are unpigmented distally. Plantar pads are more darkly pigmented in some specimens. They include 4 interdigital pads, a relatively large thenar, and a tiny hypothenar. All are small compared with the entire plantar surface.
As in congeners, there are 8 mammae (1 postaxillary, 1 abdominal, and 2 inguinal pairs). The prominent midventral cutaneous gland found on many adult specimens of the other species is not present on any specimen of B. gamay.
From a dorsal perspective of the cranium (Fig. 6), the new species has a relatively shorter rostrum and broader braincase and interorbit than do congeners (Fig. 4). As in the other species, the zygomatic arches are heavy and the zygomatic notches deep. In contrast, however, the zygomatic spines are inflected medially in B. gamay rather than projecting more directly forward as in the other species. As in congeners, superorbital ridges are well developed, but postorbital and temporal ridges are weak. The interparietal is relatively small, and the occiput is not deep.
In lateral view (Fig. 6), the premaxillaries extend only slightly beyond the anterior faces of the upper incisors. The nasal tips project well beyond the premaxillaries and curve downward slightly. The rostrum is deep, and the dorsal profile is nearly straight from a point just posterior to the nasal tips to the interorbit. In contrast to the other members of the genus (Fig. 4), the braincase of B. gamay is inflated such that its dorsal profile is domed from the interorbit to the occiput. The posterior margin of the occiput is convex rather than straight and projects slightly beyond the occipital condyles. Because of the orientation of the zygomatic spine, the dorsal anterior margin of the zygomatic plate is angled obliquely downward. In other species of Bullimus, dorsal anterior margin of the plate has a more abrupt shoulder because of forward projection of the zygomatic spine. As in other species, the zygomatic plate itself is wide, and the posterior margin is positioned near the middle of the 1st molar.
As in other species of Bullimus, the alisphenoid canal is open laterally, such that the foramen ovale accessorius is coalesced with masticatory and buccinator foramina rather than being separated by a bony strut. The conformation of the squamosal dorsal to the auditory bulla also is uniform within the genus. The postglenoid foramen is not large, there is no subsquamosal fenestra, and the squamosal root of the zygomatic arch is situated low on the side of the braincase, and its posterior process does not reach the lambdoidal crest.
Features of the auditory bulla as diagnostic for the genus (Fig. 7A). The bullar capsule is large, inflated ventrally and medially, and concave on the ventral lateral surface. On the dorsal lateral surface, a broad, shield-like process of the ectotympanic causes the external auditory meatus to open caudad rather than laterally. The posterior lamina of the ectotympanic forms a flange that conceals part of the incus.
A ventral view of the cranium (Fig. 6) reveals incisive foramina that are broad, but relatively short compared with those of congeners (Figs. 3 and 4). In contrast to the other species, the foramina do not extend to the level of the anterior margins of the 1st molars, nor do they exceed the alveolar length of the toothrow (Appendices I and II). The bony palate is relatively long and is marked by shallow palatine grooves. The posterior palatine foramina are located within the grooves at a level even with the space between the 2nd and 3rd molars. In contrast to other species of Bullimus, posterior margin of the palate is located at a level anterior to posterior margins of the 3rd molars, rather than extending caudad beyond the toothrows. Compared with the other species, the anterior palate is substantially narrower than the posterior portion, and the toothrows are more divergent (Figs. 3 and 4).
The general configuration of the pterygoid region of the new species is the same as in other members of the genus; however, the mesopterygoid fossa is wider relative to the skull length, and its walls are breached by large sphenopalatine vacuities (Fig. 3). The pterygoid fossae are deep, and they contain large sphenopterygoid vacuities. At the posterior margin of each pterygoid plate, the middle lacerate foramen separating the plate from the inflated auditory bulla is extremely narrow, and in some specimens the plate and bullar capsule are in partial contact. A groove on the posterolateral margin of the plate describes the path of the interorbital branch of the stapedial artery. This arrangement, together with the large stapedial foramen on the posteromedial margin of the bulla that receives the stapedial artery, represents the common, and presumably primitive, condition for murines (Musser and Heaney 1992).
The general form of the mandible is similar in all species of Bullimus (Figs. 4 and 6). The body of the mandible is robust, being thick and deep relative to its length. The coronoid process is large, and the posterior margin between the condyloid and angular processes is deeply concave. Some distinctive mandibular traits of B. gamay derive from the fact that the lower incisor describes a relatively tight arc: that portion of the mandible anterior to the molar toothrow is relatively shorter and turns dorsad more sharply, and the entire mandible is deeper relative to its length (Fig. 3).
Upper incisors of B. gamay have orange enamel and are ungrooved. They are broad relative to their depth (Appendix II) and are slightly opisthodont relative to the rostrum. Lower incisors have dark yellow enamel, are relatively broad, and have flat anterior faces that wear to sharp chisel-shaped tips. Compared with the other species of Bullimus, the lower incisors describe a tighter arc relative to the size of the mandible. As in the other species, molars are large relative to the palatal area (Figs. 4 and 6), are hypsodont, and have relatively simple cuspidation (Fig. 7). On the 1st upper molar, the posterior cingulum and cusp t7 are absent. Cusp t3 is small on the 1st upper molar and is either absent or greatly reduced on the 2nd and 3rd molars. On the 1st lower molar, the anteroconid consists of distinct anterolabial and anterolingual cusps, with no evidence of an anterocentral cusp. Compared with congeners, B. gamay has upper and lower 3rd molars that are smaller relative to the other teeth (Fig. 7; Musser and Heaney 1992:120, figure 76). This involves a reduction in the size of cusp t1 on the upper molar and reduction or loss of the anterior labial cusplet on the lower.
During the 1994 and 1995 surveys in the Mt. Timpoong region, B. gamay was documented over a broad elevational range. It was most common (1.5% trap success) at the type locality at 1,275 m elevation, an area characterized by steep slopes, thin leaf litter, and shallow, volcanic soil, supporting undisturbed primary montane forest vegetation (Heaney and Tabaranza 1997). It was less common in secondary lowland forest at 900 m elevation, in disturbed lower montane forest at 1,200 m, and at sites in mossy forest habitat at 1,400 and 1,475 m elevations. Specimens were taken in traps placed on the ground, mainly in runways beneath root tangles, under logs, or near large rocks. As in other species of Bullimus, elongate hind feet with relatively small plantar pads and a tail that is substantially shorter than the head–body length suggest terrestrial, as opposed to scansorial, habits (Rabor 1955, Rickart et al. 1993). There was evidence of recent reproduction in mid-March 1995, but not in May 1994 (L. R. Heaney, in litt.).
The pattern of variation seen in Bullimus is best understood in the context of the historical geography of the Philippines. With the exception of Palawan and some associated islands that lie immediately north of Borneo, the Philippine archipelago is oceanic in origin and has never had dry land connections to continental Asia (Hall 1996; Hamilton 1979; Heaney 1985, 1986). Geological evidence therefore suggests that fauna of the main portion of the Philippines originated through overwater colonization (Heaney 1986). Current distribution patterns of species have been shaped by events of the late Pleistocene when sea level was 120 m lower than at present (Fig. 1; Heaney 1985, 1991). The archipelago is divisible into distinct zoogeographic regions that correspond to the limits of late Pleistocene islands (Heaney 1985, 1986).
The 3 species of Bullimus occur in different Pleistocene island groups. B. bagobus is widespread among islands of the Mindanao faunal region (Heaney et al. 1998), occurring throughout Mindanao Island proper, and on several smaller islands which were part of Pleistocene Greater Mindanao during glacial periods when sea level was lower than at present (Fig. 1; Heaney 1985). There are far fewer records of B. luzonicus, but that species apparently is confined to Luzon Island, which represents a separate faunal region (Heaney et al. 1998).
Bullimus gamay is confined to Camiguin Island. Despite its proximity to Mindanao (Fig. 1), Camiguin was not part of Pleistocene Greater Mindanao. It is isolated by a deep water channel (minimum depth 385 m) and therefore remained separate from Mindanao during periods of lower sea level (Heaney 1985, 1991). Despite its small size, Camiguin supports a relatively rich native mammal fauna of 21 species, 19 of which are found on Mindanao (Heaney and Tabaranza 1997, Heaney et al. 1998). However, 2 of 4 native rodents (B. gamay and an undescribed species of Apomys) are island endemics, making Camiguin the smallest Philippine island known to support endemic species of mammals (Heaney et al. 1998). This status reflects the island's geographic position within the archipelago: proximity to a large island with a rich fauna (Mindanao) has allowed some colonization to occur, but isolation by deep water has been sufficient to promote speciation among some nonvolant mammals.
For assistance with field work on Camiguin we thank N. Antoque, E. Batara, N. Batocael, N. Bojo, M. Carmona, A. DeOcampo, M. Jayoma, L. Mostrales, G. Rosell, A. Tabaranza, B. Tabaranza III, and D. Tabaranza. Permits were provided by the Protected Areas and Wildlife Bureau (Philippine Department of Environment and Natural Resources, DENR), with special thanks to A. Alcala, C. Custodio, and M. Mendoza. For the loan of specimens under their care, we are indebted to M. Carleton and H. Kafka (USNM), G. Hess (DMNH), and C. McCaffery and L. Wilkins (UF). W. Newmark assisted with multivariate analysis. Scanning electron micrographs were made by K. Albertine and N. Chandler at the Research Microscopy Facility, University of Utah. We thank G. Musser and an anonymous reviewer for constructive comments on earlier drafts of the manuscript. Fieldwork was supported by the World Environment and Resources Program of the John D. and Catherine T. MacArthur Foundation. Additional support was from the Ellen Thorne Smith Fund and the Barbara Brown Fund for Mammal Research (The Field Museum, Chicago).