Hyosciurus heinrichi Archbold and Tate, 1935

Musser, Guy G., Durden, Lance A., Holden, Mary Ellen & Light, Jessica E., 2010, Systematic Review of Endemic Sulawesi Squirrels (Rodentia, Sciuridae), with Descriptions of New Species of Associated Sucking Lice (Insecta, Anoplura), and Phylogenetic and Zoogeographic Assessments of Sciurid Lice, Bulletin of the American Museum of Natural History 2010 (339), pp. 1-260 : 1-260

Hyosciurus heinrichi Archbold and Tate, 1935

HOLOTYPE AND TYPE LOCALITY: The holotype of Hyosciurus heinrichi is an adult male (AMNH 101310) collected by Gerd Heinrich on July 19, 1930. A stuffed museum study skin and accompanying skull comprise the holotype. The skin is intact as is the cranium and the mandible; all teeth are present (the skull, occlusal views of upper and lower cheek teeth, left dentary and right front foot are illustrated in Archbold and Tate, 1935: 3). Measurements are listed in table 42.

The type locality is Pegunungan Latimojong (‘‘Latimojong Mts.’’), 2300 m, in the northern section of the southwestern peninsula (see the maps and discussions in Heinrich [1932] and Stresemann [1940], and locality 15 in the gazetteer and map in figure 35), 03 ° 309S, 121 ° 239E, Propinsi Sulawesi Selatan, Indonesia.

EMENDED DIAGNOSIS: A ground squirrel distinguished from species of endemic tree squirrels in Rubrisciurus and Prosciurillus by the traits enumerated in the generic diagnosis. Morphologically similar to Hyosciurus ileile but differs by having: (1) a much longer muzzle; (2) darker upperparts; (3) dark brownish gray underparts marked by a narrow midventral white swath with irregular borders that is broken up into white segments in some individuals; (4) slightly longer and thinner claws relative to size of feet; (5) larger ears but average shorter body, tail, and hind foot, as well as less mass; (6) much longer and slimmer rostrum; the tube formed by nasals and premaxillaries projecting farther beyond front faces of incisors; (7) longer nasals relative to length of frontals; (8) anterolateral rim of anterior opening of infraorbital canal slanted; (9) dorsal process of jugal lower, producing a more gracile zygomatic arch; (10) pterygoid fossae much narrower; (11) foramen ovale about twice as large; (12) posterior border of bony palate without a median projection in most specimens; (13) body of dentary slightly lower, posterior border between condyloid and angular processes more deeply concave. Hyosciurus heinrichi also differs from H. ileile in absolute means and proportions of the cranial and dental dimensions described in the section covering comparisons.

GEOGRAPHIC AND ALTITUDINAL DISTRI- BUTIONS: The latitudinal and longitudinal boundaries of H. heinrichi on Sulawesi are unknown, but the species may occur only in the mountainous west-central core of the island if the collection localities of voucher specimens reflect its real distribution (see gazetteer and map in figure 35). Samples come from the northern area (Gunung Kanino and Gunung Nokilalaki), the central region (Pegunungan Takolekaju and Tamalanti), and the southern highlands (Pegunungan Latimojong) of the western mountain block. The distribution of H. heinrichi in this western mountainous region of Sulawesi’s central core is concordant with a suite of mammals—three species of shrews, at least two and possibly three tarsiers, a tree squirrel, and 14 species of murid rodents— that are so far known only from forest habitats in this region (table 25).

Elevational collection records that we judge to be reliable range from 1479 m in lower montane rain forest to 2287 in upper montane rain forest. This bracket contains elevations recorded by Musser on Gunung

TABLE 42

Age, Sex, and External, Cranial and Dental Measurements (mm) for Holotypes of Hyosciurus ileile and Hyosciurus heinrichi

Kanino and Gunung Nokilalaki, and those made by Heinrich on Pegunungan Latimojong. W.J.C. Frost took specimens between 4000 and 4300 ft (1220–1312 m) on Pegunungan Takolekaju (Molengraff range), which is probably at the lower limit of lower montane rain forest. The three specimens collected by Frost from Tamalanti have no elevations recorded on the skin labels. Laurie and Hill (1954: 156) claimed that most of Frost’s material from Tamalanti was obtained between 3300 and 3800 ft (1006– 1159 m), but this range is within tropical lowland evergreen rain forest and too low for H. heinrichi , judged from Musser’s experience with the species along his transect.

DESCRIPTION: Hyosciurus heinrichi is a moderately large, long-nosed, dark brown, and short-tailed ground squirrel (length of head and body, 195–240 mm; length of hind foot, 51–61 mm; weight, 228–370 g; extremes from table 43). Among the species of squirrels endemic to Sulawesi, H. heinrichi is slighter smaller than H. ileile and Rubrisciurus rubriventer in physical size but much larger than members of the Prosciurillus leucomus and P. murinus groups (table 3).

Chocolate brown fur flecked with buff and black covers the upperparts from the head to the base of the tail, including the ears, which lack tufts. The mix of dark gray wooly underhairs tipped with pale buff, longer overhairs that are dark gray or blackish along most of their lengths with a subterminal ochraceous or pale buffy band and black tip, and glistening black guard hairs combine to produce the lustrous dark brown cast to the pelage, which is also soft and dense (15– 18 mm thick over the back). On the top of the neck just behind the ears the fur is reversed and then parted along the midline exposing the skin along a 30–40 mm strip. Each eye is surrounded by a hairless dark buffy ring. Dorsal surfaces of the front and hind feet are black to blackish gray, the palmar and plantar surfaces are black. Claws on the front digits are long, slim, and only slightly recurved (see the illustration in Archbold and Tate, 1935), 30%–35% of the length of the front foot (15%–20% for the much shorter and deeply recurved claws of the species of Rubrisciurus and Prosciurillus ), and almost as long as the digits. Claws on digits of the hind feet are also slim and gently arcuate but are shorter (12%) relative to length of the foot.

Fur covering the underparts is white and dark grizzled brown in various patterns. Among the specimens from Gunung Kanino, Gunung Nokilalaki, Pegunungan Takolekaju, and Pegunungan Latimojong, the fullest expression is exhibited by five specimens: a pure white strip with ragged margins extends from the chin to the base of the tail and down inside each thigh (hairs are unpigmented for their entire lengths; the strip may appear cream on dry skins but is white on freshly caught squirrels). This midventral white swath is bordered by dark brownish gray ventral fur speckled with pale buff (hairs are very dark gray for most of their lengths and tipped with buff). The ventral region in the remainder of the sample of H. heinrichi can be sorted into five categories that roughly illustrate the nature of variation in venter pattern. (1) A white midventral strip extends from chin to groin, has uneven margins, and is broken up by a small dark brown midventral spot on the lower abdomen. (2) A white strip is interrupted by an elongate midventral dark brown patch on lower abdomen, and the grizzled brown of the abdominal margins protrudes into the white strip as round blobs on the upper abdomen and chest. (3) The white portion is reduced to a solid strip from the chin to upper abdomen where it splits into two white threads that bound a large midventral grizzled brown area, then rejoin to form a solid white expanse at the base of the tail. (4) The entire underparts are dark grizzled brown with a patch of white covering the chin, neck and chest, and another covering the inguinal region—the two white areas are separated by a expanse of grizzled brown; some specimens show a diffuse, wispy thread of white, interrupted in places, extending between the two solid white areas, and usually present only on one side. (5) Solid white covers the chin, throat, and chest, grizzled brown overlays the upper abdomen, and white covers the lower abdomen and inguinal region; the two solid white regions are wispily connected by a few white hairs along the midventral line through the brown area. From a solid white midventral strip (wide but not extending to lateral margins of the underparts) with uneven margins, the range of variation reflects a fracturing of that strip into white and brown areas by incursions to different degrees of the dark brown lateral margins of the venter fur.

TABLE 43 Descriptive Statistics for Measurements (mm) of Lengths of Head and Body, Tail, Hind Foot, and Ear, and for Weight (g) Derived from Samples of Hyosciurus heinrichi and Hyosciurus ileile a Mean ± 1 SD, observed range (in parentheses), and size of sample are listed. Mean values were used to compute LT/LHB. Specimens measured are listed below.b

Slightly less than half the length of head and body (LT/LHB 5 45%–47%), the tail is very dark brown with buffy and black highlights, coloration similar to the upperparts of the head and body.

Females have three pairs of teats: one abdominal pair (situated in the middle of the abdomen) and two inguinal pairs. The number matches that in all the species of Prosciurillus , but the most anterior pair is postaxillary in position in Prosciurillus , not in the middle of the abdomen. One or two young comprise a litter. No embryos were found in the few mature females examined, but one specimen had a placental scar in one uterine horn, and two squirrels had one scar in each horn.

Views of the elongate skull are illustrated in figure 36. Note the protracted rostrum and its prominent projection beyond the incisors, the moderately strong temporal ridges meeting to form a short sagittal crest, the short orbit, relatively low jugal component of the zygoma, postorbital process even with anterior surface of the braincase, position of the posterior margin of the bony palate relative to backs of the parallel tooth rows, very narrow ptergoid fossae bounded by a slight lateral ridge, spaceous foramen ovale, posterior maxillary foramen at each posterolateral margin of the bony palate, incisors emerging from the rostrum at a right angle (orthodont configuration), relatively smaller third upper molar, and slender but sturdy dentary. Cranial and dental measurements are summarized in table 44.

COMPARISONS: Hyosciurus heinrichi needs to be compared with the only other species in the genus, H. ileile . Both species share a similar physical build, but H. heinrichi is more gracile and less stocky than H. ileile , with a much longer muzzle. Typically, H. heinrichi has larger ears; a shorter body, tail, and hind foot; and weighs less (table 43).

The external dimensional contrasts are multivariately summarized in the ordination of scores projected onto the first and second principal components extracted from principal-components analysis illustrated in figure 38. The points form two clusters that barely overlap along the second axis but do so to a greater extent along the first component. The very high positive score for length of tail along with the lower and positive scores for length of head and body, and length of hind foot, contribute to the distribution of scores along the first axis (table 46), pushing the bulk of the scores representing specimens of H. ileile to the right of the H. heinrichi constellation, a reflection of the average greater lengths of these dimensions shown by H. ileile (table 43). Length of ear loads heaviest along the second component followed by length of head and body, and then length of hind foot. Clearly reflected here are the relatively much larger ears of H. heinrichi compared with H. ileile , and its relatively shorter body and hind feet.

Differences also exist between the two species in pelage coloration. Fur covering upperparts of head and body is dark brown flecked with buff and black in both species, but the coat of H. heinrichi is thicker, 15– 18 mm as opposed to 10–12 mm in H. ileile , and appears slightly darker due to the longer, darker hairs. Fur on the neck just behind the ears is reversed followed by a part in both species, but the median part is 30–40 mm long in H. heinrichi and only 10 mm long in H. ileile . Coloration of the front and hind feet are similar in the two species although the palmar and plantar surfaces of H. heinrichi tend to be darker, black rather than slate gray as in H. ileile .

Color pattern of the ventral fur differs. It is dark grayish black speckled with pale buff on Hyosciurus heinrichi and marked by a midventral pure white region with irregular borders that is variable in coverage of the underparts. In its fullest extent, the white swath is moderately wide, but never covers the entire width of the underparts, and reaches from the chin to the base of the tail and inner surfaces of the thighs. Its weakest expression is a pattern where dark brown ventral fur covers the upper abdomen separating a white swath on the chin, throat, and chest from the white covering the lower abdomen and inguinal region (a detailed exposition of the range is provided in the section on description).

The entire venter, from the chin to base of the tail and inner surfaces of the thighs is cream-colored in H. ileile , not pure white as in H. heinrichi (Musser recorded this chromatic distinction in the field as seen on freshly caught animals). The cream swath covers nearly the entire venter and is bounded by even borders. On each specimen from the central core of Sulawesi and the two from Gunung Ile-Ile, the cream underparts are interrupted by a small midventral brown area. At one end of the variation of spotting in the samples from Musser’s transect is a specimen that displays a small (5 X 10 mm) midventral spot between the thighs. At the other end is a squirrel with a thin midventral brown strip reaching from the abdomen to base of the tail. Of the rest of the specimens, one has two midventral brown spots on the lower abdomen, and the rest show a single brown blob (10 X 20 mm) in the middle of the lower abdomen. Among the two squirrels from Gunung Ile-Ile, the holotype has a narrow patch of midventral brown hairs on the lower abdomen (not obvious because the patch was sewn into the seam after the skin was stuffed), and a round brown incursion in the right postaxillary region. Lateral borders of the cream portion are uneven because of the uneven way the skin was sewn. The other specimen has a tuft of brown hairs in the inguinal region (partially concealed in the sewn seam). The range in patterning of the venter that is seen in H. heinrichi is not present in H. ileile .

Both species have long claws but those of H. heinrichi are slightly slimmer (more gracile) and longer relative to length of foot (30%–35% of front foot length in H. heinrichi , 25%–28% in H. ileile ).

The majority of differences between the two species expressed in the skull reflect in H. heinrichi a greater specialization associated with the elongated rostrum and a less specialized conformation in H. ileile . The protracted muzzle of H. heinrichi compared to the shorter version typical of H. ileile extends to the cranial architecture (figs. 36, 37). Mean values of dimensions from the back of the skull—its breadth (indexed by MB), postpalatal region (PPL), breadth of the bony palate (BBP), and length of the tooth row (CLPMM)—are not significantly different in the two species (tables 44, 45). On the other hand, the mean values of dimensions describing the facial skeleton—lengths of the rostrum (LR), nasals (LN), diastema (LD), bony palate (LBP), and projection of the bony nasal tube beyond the upper incisors (LPT)—are much greater in H. heinrichi than in the sample of H. ileile , and the rostrum is not as deep (HR), reflecting the absolutely longer, slimmer, and more protracted rostrum of H. heinrichi , and its much longer extension beyond front faces of the incisors compared with the shorter and broader anterior half of the skull in H. ileile . Paralleling the longer rostrum of H. heinrichi is its longer skull (ONL, CBL). The skull of H. heinrichi is also typically narrower (narrower zygomatic breadth, ZB), with a narrower interorbital region (IB), shorter orbit (LO), shallower braincase (HBC), and much smaller ectotympanic bullae (LB) compared with those dimensions in H. ileile .

Qualitative contrasts involving the nasals, jugal, opening of the infraorbital foramina, bony palate, pterygoid fossa, foramen ovale, and dentaries also distinguish the two species. Nasals of H. heinrichi are about as long as the frontals, those of H. ileile are shorter than the frontals (83%–86% the length of the frontals). The jugal of H. heinrichi is low and the entire arch is gracile, while the jugal of H. ileile has a higher dorsal process and the arch appears more robust (not shown in the drawings as clearly as is seen on specimens). We did not measure jugal height but the contrast between the two species is evident when skulls are compared side-by-side. The lateral lip of the infraorbital canal is typically at a slant in H. heinrichi but vertical in H. ileile . Most specimens of H. heinrichi are without a conspicuous midventral posterior projection on the posterior margin of the bony palate (fig. 36); a very few display a midventral nubbin on the bony margin or a slightly larger bump. All the examples of H. ileile except the specimen portrayed in figure 37 have a prominent midventral process at the posterior margin of the bony palate. The pterygoid fossa in H. heinrichi is narrow, its lateral margin defined by an inconspicuous low ridge. Hyosciurus ileile has a much wider fossa, its surface triangular from ventral perspective, and marked laterally by a higher ridge; the shape, relative surface area of the fossa, and height of the bordering ridge form a configuration closely similar to that seen in the species of

TABLE 44 Descriptive Statistics for Cranial and Dental Measurements (mm) Derived from Population Samples of Hyosciurus heinrichi Mean ± 1 SD and observed range (in parentheses) are listed.

Prosciurillus (figs. 12–14). Each foramen ovale is spacious in H. heinrichi , about twice the size of comparable foramina in H. ileile . Finally, H. heinrichi has a slightly slimmer dentary as indicated by its lower ramus, and the posterior border between condyloid and angular processes is more deeply concave than the configuration in H. ileile with its somewhat deeper ramus and shallower posterior border.

Two widely spaced clusters contained in the ordination of specimen scores projected onto the first and second principal components extracted from principal components analysis in figure 38 mirror the cranial differences described above. The position of the H. ileile cluster to the left and the H. heinrichi group to the right along the first axis is influenced mostly by the high and positive loadings for lengths of nasals, rostrum, and

TABLE 45 Descriptive Statistics for Cranial and Dental Measurements (mm) Derived from Population Samples of Hyosciurus ileile Mean ± 1 SD and observed range (in parentheses) are listed.

diastema along with extent of the rostral projection beyond the incisors (table 47); the moderate but still positive loadings for lengths of skull and bony palate; and the moderate to high negative values for interorbital breadth, heights of rostrum and braincase, and lengths of orbit and bulla. Loadings for breadths of the braincase (mastoid breadth) and bony palate as well as lengths of the postpalatal region and tooth rows have a marginal effect on producing the two clusters along the first principal component. The pattern of scores in this principal-components ordination is a multivariate analytical summary of the cranial differences between the two species of Hyosciurus that is also expressed by mean values of cranial dimensions (tables 44, 45) and can be seen in the illustrations of skulls (figs. 36, 37): H. heinrichi has an appreciably longer and more slender rostrum, narrower interorbital region, shorter orbit, shallower braincase, and smaller ectotympanic bullae compared with the cranial dimensions in H. ileile ; skulls of H. heinrichi appear gracile, thin, and long, those of H. ileile appear stocky and short by comparison.

The bony mandibular architecture is similar in the two species (figs. 36, 37). The dentaries of each species are about the same length, but the body of the ramus is not as deep in H. heinrichi as it is in H. ileile , a distinction shown in the figures of skulls (we did not measure the mandibles).

Among the traits Tate and Archbold (1936: 1) used to separate H. ileile from H. heinrichi was the smaller third premolar (‘‘a mere spicule’’) and first molar of H. ileile and its narrower pterygoid fossa (5 mesopterygoid fossa). Of the two specimens in the type series from Gunung Ile-Ile, the holotype (AMNH 101308) has small, thin, and peg-like third premolars, and the other (AMNH 196507) has a small right third premolar, but lacks the left third premolar. Small to robust third premolars describes the range of

TABLE 46 Results of Principal-Components Analysis Contrasting Population Samples of Hyosciurus ileile with Those of Hyosciurus heinrichi from Central Sulawesi Principal components are extracted from a covariance matrix of log-transformed values for lengths of head and body, tail, hind foot, and ear; see figure 38.

variation in the series of H. ileile from the central core of Sulawesi, and a similar range in size occurs in our large sample of H. heinrichi . We detected no appreciable difference between the two species in size of the first molar or breadth of the mesopterygoid region; Corbet and Hill (1992) also did not appreciate the differences noted by Tate and Archbold.

GEOGRAPHIC VARIATION: Judged by head-and-body length, the specimens of H. heinrichi from Pegunungan Takolekaju average slightly larger than the series from Pegunungan Latimojong, Gunung Kanino, and Gunung Nokilalaki; however, absolute length of tail and tail length relative to body length, along with length of hind foot, are closely similar among all four geographic samples (table 43).

Color of the fur is comparable among the samples. All contain squirrels with chocolate brown upperparts and blackish feet. Underparts are white and blackish gray configured into a range of patterns as we described previously.

Mean values of cranial and dental dimensions reveal a picture of geographic variation that is summarized by the descriptive statistics listed in table 44 and results of multivariate analyses presented as specimen scores projected onto the first and second principal components extracted from principal-components analysis (fig. 39). Covariation in most variables influenced the spread of scores along the first axis (table 48), which reflects the greater size of certain dimensions in the sample from Pegunungan Takolekaju compared to the samples from Pegunungan Latimojong and from Musser’s transect along Gunung Kanino and adjacent Gunung Nokilalaki. Mean values for some of the variables—length of skull and tooth row, breadth of the interorbital region, height of rostrum, and breadth of the bony palate— exceed those in the sample from Gunung Kanino and Gunung Nokilalaki (table 44).

Scores representing the four specimens from Pegunungan Latimojong fall between those from Musser’s transect and the series from Pegunungan Takolekaju. The ellipses that outline 95% confidence limits for scores of specimens in each geographic sample indicate the relative homogeneity of the samples from Gunung Kanino and Gunung Nokilalaki (the former is a high ridge attached to the latter and unbroken montane forests cover both highlands), the separation of the Takolekaju sample from the Kanino- Nokilalaki specimens, and the intermediate position of the Latimojong series. This is somewhat surprising because Pegunungan Takolekaju lies between the highlands along Musser’s transect and Pegunungan Latimojong, which is at the southern end of the mountains forming the western part of Sulawesi’s central core (see map in fig. 35). Our analysis reveals variation in certain cranial dimensions among the three basic geographic samples—Gunung Kanino + Gunung Nokilalaki, Pegunungan Latimojong, and Pegunungan Takolekaju—with those from Musser’s transect exhibiting the smaller skulls and those from Pegunungan Takolekaju the largest. What is the significance of these results? Does the variation in cranial and dental measurements among present samples simply reflect geographic variation due to differing intensities of gene flow among populations, or point to the Takolekaju sample as representing a population genetically isolated—or nearly so—from the other montane populations of ground squirrels? We do not know. Within a wide-ranging montane species like H. heinrichi , we would expect some morphometric variation from place to place, especially among highlands supporting montane forest habitats that are separated by major river valleys or lowland evergreen rain-forest habitats. Uncovering existing patterns of genetic continuity or partial interruption among populations of H. heinrichi will depend on analyzing variation in external, cranial, and dental measurements among larger samples from more locations in the western mountain block south of Musser’s transect. Compared with the large series from Gunung Kanino and Gunung Nokilalaki, there are only a few specimens in each of the samples from Pegunungan Takolekaju and Pegunungan Latimojong, and there are no specimens from montane regions between any of present collection sites. Fewer specimens per collection site would be required for extracting DNA samples, but broad geographic coverage would still be required.

In addition to the four geographic samples discussed above, a fifth, Tamalanti (see gazetteer), is closer to Gunung Kanino and Gunung Nokilalaki than to the other two collection sites; yielded three specimens that we did not measure because the skulls were incomplete or their condition otherwise unsatisfactory. During one visit to the British Museum, Musser compared the three skulls with his material and found them to be inseparable in overall size.

ECOLOGY: Hyosciurus heinrichi is terrestrial, diurnal, and inhabits cool and damp montane forests along streams, on hillsides,

TABLE 47 Results of Principal-Components Analysis Contrasting Population Samples of Hyosciurus ileile with Those of Hyosciurus heinrichi Principal components are extracted from a covariance matrix of log-transformed values for 18 cranial and 1 dental variable; see figure 39.

and on ridgetops (see table 2 where ambient temperature regimes are listed for Gunung Nokilalaki and Gunung Kanino). Musser worked only in primary forest on those highlands, so we do not know if populations of this ground squirrel can successfully inhabit second-growth forest or scrub habitats. On Gunung Nokilalaki, Musser caught the squirrels with Conibear traps placed in wide and prominent runways along the side of cut banks on steep slopes and along rock faces; beneath rotting, moss-covered tree trunks; under wet, mossy tangled trunks and limbs of rotting tree-falls; and among the mossy roots of rotting stumps and living trees (figs. 41, 42). Some of the runways ended in freshly dug burrows. In the vicinity of Gunung Kanino, Musser trapped the squirrels on tree trunks and limbs lying across the Sungai Tokararu (see trapping summary in table 49).

The furtive and quiet H. heinrichi , although common, is rarely seen in its dark and wet mossy forest habitat. Musser was fortunate to observe three individuals on the summit of Gunung Nokilalaki at 10:00 A.M. on a sunny day in April for about half an hour. At that spot the understory was open with scanty ground cover compared to the denser forest downslope where the thick understory made it nearly impossible to see the ground squirrels. Each of the squirrels was in the open, each had dashed from a runway, stopped in the clearing on a rock or raised point of ground, then dashed off again to disappear along or beneath a rotting tree trunk or stump and reappear again in the general vicinity. When running they appeared as long blackish streaks. At times each would stop, sit on its haunches, and twitch its tail much as all ground squirrels do, but never vocalize, then streak off and eventually disappear from sight. This was an exceptional encounter because squirrels were sometimes heard rustling through the undergrowth, but rarely seen.

A year later, on another April morning, Musser watched an individual for 10–15 minutes in a different area of forest on the summit of Nokilalaki. He stood about 10 ft from the squirrel, which was sitting on a rotting tree trunk covered with thick moss. From a distance, the squirrel resembled a round, dark brown ball of fur with a long face and big eyes. At first it responded to Musser’s presence by leaping to the ground and scampering away for a few feet, but then returned to the top of the trunk to sit, occasionally grooming and biting ectoparasite-infested areas, scratching at other places in its fur, and sometimes chirping and sniffing about on the tree trunk. The squirrel kept its tail pressed closely to its back while sitting, jerking it along the back with every chirp. When the squirrel emitted a long series of calls, its tail would extend out behind the body; afterword, the tail reverted to its position against the squirrel’s back. Often the squirrel would jump to the ground, scamper beneath the trunk where it remained for a minute or so looking over several old acorn husks, groom itself, and vocalize again. Once, after emitting a series of chirps it slowly approached Musser with nervous, jerky movements, stopping within 3 ft of him, and twitching its tail up and down before quickly scampering back to the trunk. Finally, Musser moved closer to the trunk, and the squirrel streaked away over the wet ground chirping continuously as it went, holding its tail straight behind its body.

The squirrel’s voice was distinct from that of any species of the tree squirrels Prosciurillus and Rubriventer that Musser heard vocalize. As it sat, the ground squirrel uttered a string of single chirps, sounding more like a bird and not at all like a squirrel. When the squirrel became nervous or frightened, the calls increased in frequency and and the chirps had the slight suggestion of a chuck. Once or twice it became very agitated and chattered as most squirrels do, but the sounds were still soft, without the timber or resonance of the tree squirrels. Its bird-like vocalizations were deceptive, and as Musser worked daily in the forest, he did not know if the chirps heard most of the mornings issued from ground squirrels or birds.

Judged from contents of stomachs, the diet of Hyosciurus heinrichi consists of arthropods and some fruits, primarily acorns (fruit from species of Lithocarpus ); see summary of stomach samples in tables 49 and 57. Some stomachs contained only arthropods, others arthropods and acorn mash, and a few were filled only with acorn mash. Geophilomorph and Scolopendromorph centipedes, the occasional ponerine and formicine ants and ant pupae, long-legged cursorial beetle larvae, several other kinds of beetle larvae with short legs, adult beetles, cockroaches, and certain kinds of fly larvae ( Diptera ) comprise an arthropod suite common to leaf litter (D. Grimaldi, personal commun., 2009). Another group of dipteran larvae found in stomachs,

TABLE 48 Results of Principal-Components Analysis Contrasting Population Samples of Hyosciurus heinrichi Principal components are extracted from a covariance matrix of log-transformed values for 18 cranial and 1 dental variable; see figure 40.

the Sciaroidea, live just below the ground surface and in wet moss; the female fly oviposits in the wet ground or moss below the surface and the eggs hatch into a ball of larvae. With its elongate muzzle and very long claws, the ground squirrel locates prey, then scratches and pokes through the leaf pack and moss (small segments of moss were found in stomachs) to pounce on any arthropods encountered. Exposing the subsurface soil environment with its long claws would allow the squirrel access to the larval Sciaroidea dipterans. In stomachs, this suite of leaf litter and subsurface arthropods were often contained in a clay-like ball separate from acorn mash. No stomach examined contained the large beetle larvae, usually legless, that live in rotting wood; such kinds of larvae were found in many stomachs of Rubrisciurus rubriventer .

The cranial architecture common to specimens of H. heinrichi exhibits adaptations well suited for procuring the dietary components found in contents of stomachs. A very long nasal region likely contains increased nasal epithelium that would enhance olfactory acuity. To a human, odors from wet soil, decaying leaf litter and rotting treefalls, and soaked moss saturate the air in mossy forest, and must be even more concentrated to a squirrel, which has better olfactory ability. Detecting dipteran larvae in wet ground and moss, and other arthropods within soaked and compressed leaf litter on the surface, probably requires heightened olfactory acuity that would cut through the prevailing ambient forest odors.

Opening acorns, a tough woody fruit, requires strong incisors along with forceful contributions from masseter and temporal muscles, and high premolars and molars affording surfaces for the mechanical breakdown of nuts over the squirrels’ lifetime. Prominent temporal ridges meet in the midline near the occiput to form a short sagittal crest on the skull of H. heinrichi , the jugal component of each zygomatic arch is high, and the dentary elongate but robust; all provide firm origins and insertions for the array of masticatory muscles. This myological complex is joined by narrow but strong incisors, and large, high cheek teeth (fig. 36). Except for a shorter rostrum, the architecture is basically similar in H. ileile (fig. 37). The morphology contrasts sharply with that peculiar to the Indomalayan shrew-faced squirrel, Rhinosciurus laticaudatus , which has weak incisors that are very reduced in size relative to bulk of the skull, low cheek teeth that quickly wear to occlusal basins, slight temporal ridging extending to the occiput without forming a sagittal crest, low jugal, and delicate dentary (see illustration of skull in Corbet and Hill, 1992). Only arthropods and earthworms are eaten by this tropical ground squirrel ( Davis, 1962; Medway, 1969; Payne et al., 1985), and that diet is reflected in the structure of its teeth and skull.

The small-bodied terrestrial shrew rat Melasmothrix naso occupies the same wet, mossy montane forest habitat as Hyosciurus heinrichi . With its long muzzle and claws, dark chestnut fur, short tail, and diurnal activity, the murid resembles a small version of the ground squirrel. Melasmothrix naso digs into the wet moss and ground for Sciaroidea dipteran larvae, but also eats earthworms ( Musser, 1982). No annelids were found in stomachs of the ground squirrel.

Stashes of empty acorn hulls are scattered through the montane forest beneath tree roots, rotting tree trunks lying on the ground, and tree holes just above ground. Many of the piles probably resulted from ground squirrel activities. Most of the empty hulls were in halves, some were mostly intact except for a small hole large enough to allow extraction of the seed. The ripe seed consists of a firm, white or gray endosperm enclosed by a thin brown seed coat. The squirrels eat the entire seed and it shows up in stomach samples as a semi-dry granular whitish or pale grayish mass (the endosperm) riddled with brown flecks (the seed coat).

Lithocarpus havilandii is the common oak in the mossy forest on Gunung Nokilalaki where older trees contribute to the canopy and younger trees and saplings are scattered throughout the understory. During December of 1973, acorns rained from the trees and littered the ground where they became forage for native pigs (Sus celebensis ) and rodents. At lower altitudes in lower montane forest on Gunung Kanino, L. glutinosus and L. elegans

TABLE 49

Summary of Habitat at Trapping Sites, Stomach Contents, and Other Relevant Information for Specimens of Hyosciurus heinrichi Collected by Musser in Central Sulawesi, 1973–1975

Collection locality, specimen number, elevation, and month and year of collection are included. Descriptions of the trapping sites and contents of stomachs, slightly edited, are from Musser’s field journals (in mammalogy archives at AMNH). Collection localities on Gunung Kanino are in lower montane rain forest, those on Gunung Nokilalaki are in upper montane rain forest. All squirrels were caught during the day in Conibear traps (rats taken in the same traps were caught during the night). Unless noted differently, trapping sites were in primary forest formations.

Locality, AMNH Elevation and (ASE) numbers (m) Date Trap site, other species trapped there, and stomach contents

Gunung Kanino 1479 May 1975 On wet, decaying trunk (10 in. diameter) covered with thin layer

225483 (3048) of moss, lying 2 ft above water surface, crossing the Sungai Tokararu from one stream terrace to the other. Stomach: distended with acorn mash; no arthropod remains.

225484 (3053) 1479 May 1975 On decaying trunk (10 in. diameter), densely covered with wet moss, lying across Sungai Tokararu extending from a low terrace on one side of the stream a few inches above water level and crossing over stream to a higher terrace, 3 ft above water; no trails in moss, which is 1.5 in. thick and covers the entire trunk; also supports dense growth of epiphytes. This squirrel caught about 100 ft downstream from Hyosciurus ASE 3048, little change in altitude through this stretch of stream. Stomach: partially full, a bit of acorn mash and the rest remains of at least one large (2 in. long) chewed adult beetle.

225485 (3016) 1537 May 1975 On large rotting trunk (1 ft diameter) lying from stream terrace of the Tokararu across stream to opposite bank; terrace 3 ft high, thinly covered with wet moss, many ferns, young gingers, small plants, and young Pandanus . Stomach: full of acorn mash and arthropods (several small pomerine and formicine ants, several chewed geophilomorph and scolopendromorph centipedes, a small adult beetle, and remains of at least one cockroach—all leaf litter inhabitants).

225486 (3041) 1555 May 1975 On mossy trunk (4 in. diameter) growing horizontally across Sungai Tokararu about 1 ft above water level. Stomach: partially full, acorn mash, several whole and chewed geophilomorph centipedes, many chopped cursorial beetle larvae, a few dipteran larvae, and remains of small adult beetles.

225487 (3006) 1570 May 1975 On rotting trunk lying across Sungai Tokararu and 2 ft above water surface. Here the stream is about 4 ft wide, its cut bank is 8 ft wide and 2 ft high, and its terrace flat and 30 ft from slope to slope. Trunk (10–12 in. diameter) lies across terraces, is densely covered with thick moss (1 in. deep), from which are growing gingers, ferns, an oak-leaf fern, small shrubs, and tiny plants; no path is worn in moss. Terrace is wet and muddy, similar to that along the Sungai Salubeka where we were trapping Hyosciurus ileile . Stomach: distended with mostly acorn mash; mixed with it are chewed remains of several cursorial beetle larvae (with long legs) and another kind of beetle larvae (with short legs), fragments of small adult beetles, piece of geophilomorph centipede; the beetle larvae are the same kind as found in stomachs of Hyosciurus heinrichi caught at higher altitudes.

Gunung Nokilalaki 1738 Dec. 1973 In prominent, wide runway beneath long, decaying moss-covered

223541 (980) trunk decomposing on wet and steep forested slope; caught at 8:20 a.m.

TABLE 49 (Continued)

Locality, AMNH Elevation and (ASE) numbers (m) Date Trap site, other species trapped there, and stomach contents

223790 (1057) 1768 Dec. 1973 In wide, damp and roofed runway beneath cut bank on steep

223965 (1070) slope just above camp. Runway is a tunnellike path extending

225465 (2563) along side of bank and leading into large alcoves eroded out from beneath tree roots. ASE 2563 was caught in February, its stomach was empty. Caught the rat Bunomys penitus in same spot as the three Hyosciurus .

225465 (2857) 1982 April 1975 In damp runway beneath section of rotting trunk covered with moss and shrubbery lying on side of ravine. Steep slope; trunk nearly concealed by dense undergrowth of shrubs and ferns.

225466 (2618) 2073 Mar. 1975 In damp runway (5 in. wide) alongside and partially under large decaying, wet, and moss-covered trunk lying on muddy slope between ridgetop and terrace; shrubby ground cover, dense rattan; always cool and wet here, sometimes flowing water just to one side of trap site. Stomach: full of acorn mash in which are mixed pieces of cursorial beetle larvae (with long legs), three other kinds of small beetle larva (with short legs), a few Sciaroidea dipteran larvae, a geophilomorph centipede, and bits of adult beetles.

225467 (2823) 2073 April 1975 In well-used runway beneath long, rotten and moss-covered trunk laying downslope about 20 ft from top of ridge; caught another Hyosciurus heinrichi on other side of ridge, also the rats, Paruromys dominator and Rattus hoffmanni . Stomach: partially filled with mostly acorn mash with fragments of at least one small adult beetle.

225468 (2888) 2091 April 1975 In runway alongside decaying, wet and moss-covered large and long section of trunk (2 ft diameter) lying in open ridge forest on steep slope 20 ft from top of ridge near where Bunomys penitus was caught. Scanty undercover. Stomach: full of mostly acorn mash mixed with a many chewed cursorial beetle larvae (with long legs) and a few of the short-legged beetle larvae, and fragments of adult beetles.

225469 (2593) 2134 Mar. 1975 On muddy ground next to face of boulder on wet slope; caught the rats Bunomys penitus and Melasmothrix naso in same spot.

Stomach: partially full of some acorn mash, small geophilomorph centipede, a few cursorial beetle larvae (long, strong legs), another kind of beetle larvae (with tiny short legs), remains of adult beetles, and diptera larvae (Sciaroidea).

225470 (2898) 2195 April 1975 Trapped in wet runway alongside rotting, moss-covered section of trunk lying on side of muddy and rocky ravine. Caught the rat, Rattus hoffmanni , in same spot. Stomach: partially full of some acorn mash; abundant chewed pieces of cursorial beetle larvae (with long legs), a few examples of another kind of beetle larva (with short legs), some Sciaroidea dipteran larvae, and fragments from a few adult beetles.

223542 (1049) 2226 Dec. 1974 At base of moss-covered wet rock on steep slope below summit.

223548 (1143)

223544 (1069) 2226 Dec. 1974 In wet runway along base of wet, moss-covered rock under tree roots on nearly vertical slope below summit. Another Hyosciurus heinrichi was caught just across the trail in similar kind of runway.

TABLE 49 (Continued)

Locality, AMNH Elevation and (ASE) numbers (m) Date Trap site, other species trapped there, and stomach contents

225472 (2838) 2256 April 1975 On ground in forest near where ASE 2256 was trapped. Stomach: distended, about half consisting of acorn mash and a lemon-tan fruit (smells fruity, not acorns), rest a muddy purple matrix packed with remains of at least two dozen large cursorial beetle larvae (more than seen in other stomachs).

225473 (2973) 2256 April 1975 In damp runway beneath roots of tall, old, live canopy oak growing on edge of ridge. Caught the rats, Rattus hoffmanni and Melasmothrix naso , in same area. Stomach: distended, 2/3 consists of acorn mash; rest is the muddy-purple paste containing two kinds of cursorial beetle larvae, remains of at least two kinds of adult beetles, and two large dipteran larvae (definitely associated with the purple mash and not the tan acorn mash).

223551 (1254) 2271 Dec. 1974 In wide, wet runway beneath rotting moss-covered trunk in wet forest.

225482 (2858) 2287 April 1975 In wide (8 in.), damp runway beneath long, rotting, moss-covered trunk (1.5 ft diameter) laying on muddy floor of short forest next to cement marker on summit. Stomach: full of acorn mash, no insect remains.

225474 (2619) 2287 Mar. 1975 In mossy, 6 in. wide runway beneath rotting, moss-covered trunk that is part of a large tangle of limbs, trunks, and branches from an old tree-fall lying amid thickets of ferns, sedges, gingers, and shrubs on steep slope below ridge summit. Stomach: filled mostly with acorn mash plus chewed remains of a few cursorial beetle larvae and adult beetles.

223546 (1113) 2287 Dec. 1974 Beneath a huge rotting trunk that forms the bottom of a tree-fall on summit; caught around 9:15 a.m.

223547 (1123) 2287 Dec. 1974 In damp runway beneath decaying, wet moss-covered trunk lying on ground at summit.

223549 (1175) 2287 Dec. 1974 In wet runway beneath a jumble of roots and trunks, all covered with thick and wet moss, lying on vertical side of summit in moss forest. The rat, Rattus hoffmanni was trapped in same spot.

226550 (1186) 2287 Dec. 1974 Beneath rotten, wet and moss-covered trunk lying on muddy ground at summit.

225475 (2630) 2287 Mar. 1975 In a classic runway (5–8 in. wide, tamped down, looking like a rat road) beneath a long, rotten, wet, and moss-covered trunk lying along a low hill just on edge of rolling, partially dissected terrain of ridge that forms summit. Runway extends entire length of trunk (30 ft), then through a pile of very rotten wood, and out into open where it can no longer be detected. The shrew rat, Tateomys macrocercus , was caught in same spot. Stomach: partially full of brown acorn mash, fragments of a few cursorial beetle larvae, and many Sciaroidea dipteran larvae.

225476 (2659) 2287 Mar. 1975 In wet runway (not mossy) beneath rotting, moss-covered, and wet trunk in forest on rolling edge of summit near cement marker. Stomach: full of mostly acorn mash discolored purple, abundant chewed remains of cursorial beetle larvae, and pieces of a small adult beetle.

225477 (2699) 2287 Mar. 1975 In wet runway running underneath a long rotten and thickly moss-covered trunk on side of the ridge a few feet below summit. Stomach: full of acorn mash; no insect remains.

TABLE 49 (Continued)

replace the upper montane L. havilandii . In some places they are scattered through the forest on hillsides and ridgetops; in other areas they form large groves.

The fruit segment of the diet described here for H. heinrichi represents records for only part of a year. What fruits the squirrels may eat during other months, during the short dry season on the mountain, and at times when acorn production is unusually low, is unknown. Judged from stomach samples (table 49), fruit from oaks is preferred, but some unidentifiable fruit remains were present in a few stomachs. Fruit of chestnuts, Castanopsis acuminatissima , for example, may also be eaten. Groves of chestnuts intermingled with oaks and Calophylum are common on the upper slopes of Gunung Kanino and lower reaches of Gunung Nokilalaki to about 1800 m. Musser’s observations contribute a degree of ecological information for H. heinrichi , but there is much more to learn about the ecology of this squirrel.

ECTOPARASITES: The sucking louse Hoplopleura heinrichi , n. sp. (see description in following section), uniquely parasitizes Hyosciurus heinrichi . The only other ectoparasite records consist of immature (larvae and/or nymphs) stages of hard ticks (Acari, Ixodoidea) belonging to the genera Amblyomma and Haemaphysalis ( Durden et al., 2008; also see table 56 and ectoparasite account for Rubrisciurus rubriventer ).

SYMPATRY: The range of H. heinrichi on the central core of mainland Sulawesi overlaps those of Prosciurillus topapuensis , P. murinus , and Rubrisciurus rubriventer (table 6). Along his transect in the northern portion of central Sulawesi, Musser trapped H. heinrichi in the same traplines that yielded the species of Prosciurillus and Rubrisciurus , and sometimes in the same traps (table 49).

Voucher specimens from collection sites along Musser’s transect around Gunung Kanino indicate that in Sulawesi’s central core Hyosciurus heinrichi and H. ileile are regionally sympatric but locally parapatric in an elevational pattern where the lower border of the range of the montane species abuts the upper margin of the range for the lowland species. Musser never encountered the two species in the same trapline. The spatial relationship is roughly portrayed on the inset map showing the distribution of collection sites for each species (see fig. 35). From the southwestern side of Gunung Nokilalaki extends a high ridge that is bisected in the middle by a stream, the Sungai Tokararu (labeled on the map in fig. 4); the local people refer to the ridge on the west of the stream as Gunung Tokararu and that to the east as Gunung Kanino (both ridges are labeled on the inset map in fig. 35). The Tokararu originates below the summit of Nokilalaki, flows into a deep ravine along the northwestern margin of Gunung Kanino to pass through the cleft between the two ridges, and eventually empties into the waters of Danau Lindu. Musser worked on the top and upper sides of the Kanino ridge, on up to the summit of Nokilalaki on the eastern side of the Sungai Tokararu, and along the Tokararu stream at lower altitudes where it flowed alongside the Kanino ridge.

He collected H. heinrichi from the summit of Nokilalaki downslope east of the Tokararu to 1738 m, and along the Sungai Tokararu at the northwestern base of the Kanino ridge from 1570 m down to 1479 m (see gazetteer). He encountered H. ileile on the southeastern slope of the Kanino ridge just below the ridge-top and on the northwestern side just below the top of the ridge in a ravine containing the Sungai Salubeka, which is a small tributary of the Sungai Tokararu (and is drawn on the inset map in fig. 35 between localities 7 and 8, and labeled on the map in fig. 4); squirrels were trapped at 1512, m, 1463 m, and 1402 m, and encountered again along the Sungai Sadaunta below Danau Lindu between 884 m and 960 m.

In the Kanino area, the range of each species appears to reflect a parapatric pattern. Hyosciurus heinrichi was caught in the stream valley of the Tokararu at the base of the Kanino ridge and H. ileile was taken on the ridge itself just below the top (see tables 49 and 51 for habitats at each collection site). Parapatry may describe the ranges of the two species throughout the western mountains of the central core but no data are currently available to test this speculation. Both are similar in physical size, and both consume fruit, primarily acorns, and arthropods (see tables 43, 49, 51, and 57) and it seems unlikely that they would be found together at the same site. While samples of H. heinrichi have been obtained in the mountains south of Musser’s transect (see gazetteer and fig. 35), no examples of H. ileile have ever been collected there, which may indicate its true absence but is more likely an artifact of trapping effort. Musser found H. heinrichi to be common in montane habitats and H. ileile to be uncommon wherever it occurred; short trapping periods (a few days) would produce the former but longer durations (months) and targeted trapping days would be required to sample the latter.

SYNONYMS: None.

Hyosciurus ileile Tate and Archbold, 1936

Hyosciurus heinrichi ileile Tate and Archbold, 1936: 1 .

HOLOTYPE AND TYPE LOCALITY: The holotype of Hyosciurus ileile is a young adult female (AMNH 101308) collected by Gerd Heinrich (collector number 130) on November 21, 1930. A stuffed museum study skin and accompanying skull comprise the holotype. The skin is intact, as is the cranium with mandibles; all teeth are present. Measurements are listed in table 42.

The type locality is Gunung Ile-Ile, 00 ° 589N, 121 ° 489E (part of the larger Pegunungan Peleleh forming the mountainous backbone of the northwestern portion of the northern peninsula), 1700 m (see the maps and discussions in Heinrich [1932] and Stresemann [1940], and locality 1 in the gazetteer and map in figure 35), the extreme northwestern part of Propinsi Sulawesi Tengah, Indonesia.

EMENDED DIAGNOSIS: Hyosciurus ileile is most similar to H. heinrichi in morphological attributes, but differs in (1) its somewhat brighter upperparts; (2) typically mostly solid cream underparts; (3) shorter and blunter muzzle; (4) smaller ears; (5) heavier body and longer hind feet; (6) slightly shorter claws relative to size of feet; (7) shorter and more robust rostrum; (8) shorter projection beyond incisor faces, and shorter nasals that are not as long as the frontals; (9) higher jugal that forms a more sturdy zygomatic arch; (10) higher and wider skull on average; and (11) different cranial proportions, as described in the account of H. heinrichi (see section detailing comparisons).

GEOGRAPHIC AND ALTITUDINAL DISTRI- BUTIONS: The actual distribution of Hyosciurus ileile on Sulawesi is unknown. The few localities from which voucher specimens have been obtained include one montane area in the northwestern portion of the northern peninsula (Gunung Ile-Ile) and several places in the northern portion of the island’s central core in tropical lowland evergreen and lower montane habitats (see gazetteer and map in figure 35). The species certainly has a broad- er distribution than present samples indicate, and should be sought elsewhere in the central core of the island, and on both the east-central and southeastern peninsulae. Its range on the northern peninsula most likely extends beyond Gunung Ile-Ile; we would expect to find the squirrel in landscapes throughout the western curve of the peninsula, from the Gorontalo region west to the base of the peninsula.

The range of H. ileile may not extend from Gorontalo east to the northeastern tip of the peninsula. This region was transformed into one or more islands during warm periods in the Pleistocene when sea level was considerably higher than at present ( Fooden, 1969; Bintanja et al., 2005). Two species of macaques ( M. nigra and M. nigrescens ) and four murid species ( Echiothrix leucura , Taeromys taerae , Bunomys fratrorum , and Rattus xanthurus ) occur east of Gorontalo and have never been recorded from the northern peninsula west of the Gorontalo area ( Fooden, 1969; Groves, 2005; Musser, MS). All these mammals have close phylogenetic relatives elsewhere on Sulawesi and their phenetic and genetic distinctiveness likely evolved from ancestral populations that were isolated on the northeastern Pleistocene islands. After Archipelago Sulawesi was transformed into its present-day configuration, some mammal species formerly occurring only in the western section of the northern peninsula and central part of the island may have gradually occupied habitats east of Gorontalo. Two species of murid rodents provide examples. Bunomys chyrsocomus now occurs in forests throughout most of Sulawesi, including the northern peninsula where it is sympatric with Bunomys fratrorum in the region west of Gorontalo. Rattus marmosurus , a member of the Rattus xanthurus group ( Musser and Carleton, 2005), is common in the central core of Sulawesi and also ranges through the northern peninsula where it can be found in the same habitat and altitude as the closely related Rattus xanthurus in the northeast. At least three species of shrews ( Crocidura nigripes , C. rhoditis , and C. elongate ), two tree squirrels ( Rubrisciurus rubriventer and Prosciurillus murinus ), and seven other species of murids ( Maxomys musschenbroekii , Maxomys hellwaldii , Lenomys meyeri , Margaretamys beccarii , Haeromys minahasae , Paruromys dominator , and Rattus hoffmanni ) occur in the central core of Sulawesi and on the northern peninsula east of Gorontalo. There is nothing in the morphology of specimens in samples of these species to suggest genetic isolation between populations in the central core and those on the eastern portion of the northern peninsula (this report; Musser, MS) and they may represent expansion from central Sulawesi into the northeastern peninsula after drops in sea level that exposed a connecting land mass covered in suitable forest habitats extending from the central core to the northeastern tip of intact Sulawesi. In contrast with these species, Hyosciurus ileile may have never dispersed east of the Gorontalo area. From the early days of Dutch occupation until the present, the northeastern Minahasa district has probably been the focus of more mammal collecting than any other region of Sulawesi, yet no samples of the ground squirrel have ever been obtained.

Gunung Ile-Ile at 1700 m is the highest site from which H. ileile has been collected and is in montane forest. In the central core of the island, 168 m to 1512 m are the elevational extremes documented by voucher material, and those samples came from a region of continuous forest cover extending from tropical lowland evergreen rain forest to lower montane rain forest.

The presence of H. ileile on the northern peninsula in montane forest at 1700 m, nearly 200 m higher than recorded in Sulawesi’s central core, may reflect the absence of H. heinrichi from the peninsula. Present samples of H. heinrichi document its distribution in montane forest habitats, not in tropical lowland rain-forest habitats, and its parapatry with H. ileile at the lower boundary of its elevational range and the upper boundary limit of H. ileile . Both are ground squirrels, both are similar in body size and proportions, both exhibit similar pelage coloration, and both feed on Lithocarpus fruit and arthropods. This possible competitive altitudinal pattern should be the subject of careful ecological inquiries both on the northern peninsula and in the central core of the island.

DESCRIPTION: Hyosciurus ileile is of moderate size with a short tail (length of head and body, 215–250 mm; length of tail, 70– 125 mm; length of hind foot, 57–60 mm; weight, 293–520 g; LT/LHB 5 49%; extremes for sample from central Sulawesi extracted from table 43). It resembles H. heinrichi in physical build, but is larger and chunky in appearance with a shorter muzzle that appears blunt compared to the long face of H. heinrichi (photographs of a live H. ileile reproduced in figs. 43, 44). Upperparts show more brown tints, underparts are cream.

As in H. heinrichi , three pairs of teats are typical of H. ileile , one abdominal pair and two inguinal pairs. One female contained two embryos in a left uterine horn.

Views of the skull are provided in figure 37; compare these with those of H. heinrichi on the opposite page in figure 36. Cranial and dental measurements are summarized in table 45.

COMPARISONS: Hyosciurus ileile has been compared with H. heinrichi in the account of the latter.

GEOGRAPHIC VARIATION: We have two sets of samples for H. ileile . One consists of two very young adults (the type series) from Gunung Ile-Ile on the northern peninsula. The cheek teeth of both are slightly worn, and the presphenoid-basisphenoid and basisphenoid-basioccipital sutures are open. The other is composed of three samples from the northern part of Sulawesi’s central core: three from lowlands in the Sungai Tolewonu region , three from Sungai Sadaunta, and six from Gunung Kanino (five were measured) ; very young adults, young adults, and adults are contained in these series. Specimens in both sets of samples resemble one another in body size and pelage coloration, except that the tails and hind feet of the two specimens from Gunung Ile-Ile average slightly shorter than those variables in the three samples from the central core of the island (table 43).

Some cranial dimensions are of comparable size in the two sets of samples (length of maxillary molar row, for example; table 45), but others are not. The two squirrels from Gunung Ile-Ile have a slightly shorter facial skeleton and basicranial region than do most of the specimens from Musser’s transect. This contrast is reflected in mean values for dimensions related to the muzzle (length of nasals, LN; length of rostrum, LR; projection of nasals and premaxillary beyond the incisors, LPT; and height of rostrum, HR) and back of the skull (postpalatal length, PPL) that are listed in table 45, and the scatter of specimen scores projected onto the first and second principal components extracted from principal-components analysis displayed in figure 45. Covariation in nearly all cranial variables contributes at some intensity to the spread of scores along the first axis, an estimate of increasing size from left to right; especially forceful are the variables associated with length of the muzzle and back of the skull (table 50). Scores for the two squirrels from Gunung Ile-Ile lie on the left, and all but one of the scores representing specimens from the central core are scattered to the right. Of the five scores representing specimens from Gunung Kanino, a single very young adult (AMNH 225549: showing slight wear on the premolars and molars, and open presphenoid-basisphenoid and basisphenoid-basioccipital sutures) clusters with the two points for the squirrels from Gunung Ile-Ile.

Impressive variation in skull size exists within a particular age group, at least in those samples collected along Musser’s transect (Sungai Tolewonu, Sungai Sadaunta, and Gunung Kanino). For example, of the 10 skulls measured for the principal-components analysis, five are adults and five are young adults (all have slight wear on the occlusal surfaces of the cheek teeth and the basioccipital-basisphenoid suture and basisphenoid-presphenoid suture are completely open). The largest of the young adults has a skull length of 62.4 mm (AMNH 226497), the smallest is 57.0 mm (AMNH 225459). Skulls from the two squirrels collected on the northern peninsula on Gunung Ile-Ile are closely similar to AMNH 225459 in overall size and general conformation.

To reduce the effect of the adult specimens in the principal-components analysis, we eliminated them and employed only the two young adults from Gunung Ile-Ile and the five young adults from Musser’s transect. The pattern of scores within the ordination defined by first and second principal components (not illustrated) is similar to the spread shown in figure 45 that was derived from young adults, adults, and old adults.

Whether the distribution of specimen scores derived from the available material reflects regional variation in cranial dimensions between samples from the northern peninsula and those from Sulawesi’s central core, variation concomitant with age, or different gene pools is unclear. We have only two young squirrels from one montane locality at 1700 m on the northern peninsula, and 12 specimens from the central core collected through a range from 168 m to 1512 m. We know nothing about the characteristics of Hyosciurus that may inhabit forests between these sampled regions. Is the montane population of H. ileile on Gunung Ile-Ile characterized by small physical build or are we just seeing the very young adult end of the age spectrum? Is H. ileile restricted to montane habitats on the northern peninsula or does it range into tropical lowland evergreen forests on that peninsula? If H. ileile does inhabit lowland forests on the peninsula, are the phenetic and genetic attributes of these lowland squirrels the same as those in Sulawesi’s central core? Or would genetic discontinuity exist between the lowland populations somewhere near the base of the northeastern part of the peninsula and the central core, a break coinciding with that between the tree squirrels Prosciurillus leucomus (peninsular distribution) and P. topapuensis (central core) and macaques M. hecki (peninsular) and M. tonkeana (central core)? Additional and larger samples from both the northern peninsula and the central core are required to answer these questions that bear on unraveling the possible significance of the morphometric variation we document among the present geographic samples.

ECOLOGY: Hyosciurus ileile is diurnal and terrestrial. In the northern part of central Sulawesi, the ground squirrel occurs throughout a range of forest habitats, from the warm and humid lowland tropical evergreen rain forests (environments similar to those shown in figs. 8, 22, and 23) to the cooler lower montane forests (fig. 27; see table 2 for ambient temperatures recorded at various camps through this range of habitats). The squirrels were trapped in streamside habitats and in forests on hills and ridgetops (see table 51 for a summary of habitats at trapping sites). Musser worked in the forest nearly every day at camps scattered from the lowlands to higher altitudes, but outside of those in traps saw the ground squirrel only once, near Tomado. He trapped 12 H. ileile at localities far apart along his transect; by contrast, twice as many H. heinrichi were encountered and within a relatively much smaller area on the transect line. Similar trapping regimes were employed everywhere along the transect.

That is why the following particular capture was a pleasant surprise. Musser recorded observations of an adult male H. ileile livetrapped on February 14, 1976, along the Sungai Tolewonu at 152 m. This handsome, beautiful squirrel was at first very nervous when caged, but soon grew accustomed to the enclosure, bumping and biting the sides only if unduly disturbed. He was active during the day, vocalizing in the morning and late afternoon; at night he slept curled on his side on a platform.

This individual’s vocalization was unlike that of H. heinrichi and consisted of two different calls. When disturbed, or when he grabbed a beetle larva from Musser’s fingers, he would growl ‘‘Errr, errr, grrr’’ in short bursts from deep within his throat; the tree squirrel Rubrisciurus rubriventer growls in a similar way. The only other type of call Musser heard was usually made in mornings and afternoons: a high-pitched squeal, ‘‘eeee, eeee,’’ each squeal lasting 3–5 seconds. A variant was a drawn out single ‘‘eeee----e’’ that started high and dropped to silence, and a double ‘‘ee, ee’’ uttered quickly. All these kinds of squeals resembled the whine of an unoiled machine.

The squirrel quickly learned to drink water through a spout from a bottle, which he used throughout the day. Musser provided him with bait (a ground mixture of rolled oats, raisins, peanut butter, and bacon), nutmeg and ginger fruit, other kinds of fruit that are regularly eaten by Rubrisciurus rubriventer , small adult carabid beetles, small lizards, and earthworms. All were ignored. The only items accepted were large beetle larvae that Musser excavated from rotting trunks and limbs lying on the forest floor, and acorns (fruit from Lithocarpus ). The squirrel would grab the larva from Musser’s hand, grasp it between his front feet, and proceed to eat the head, then open the thorax and consume the internal organs, but discard the soft abdomen. Adult beetles found in the same rotting wood as the larvae were offered to the squirrel but were ignored.

Musser cut open acorns from the two species of oak growing in the Tolewonu area to determine which species the squirrel preferred. Lithocarpus celebicus , the more common of the two, is a tall canopy tree, or sometimes an emergent, occurring on flat, damp river terraces and hillsides above the river, and here and there on higher slopes below ridgetops. This species was seen only in tropical lowland evergreen rain forest between 30 and 1000 m along Musser’s transect. The other oak, Lithocarpus glutinosus , was less frequently encountered, growing on hillsides but apparently absent from wetter areas along stream terraces. Musser identified this species along his transect all the way up to Gunung Kanino. Most of the opened acorns from L. celebicus contained pulp and a small seed. The squirrel grabbed each opened section and quickly ate the seed, then gouged out and ate some of the pulp, but left most of it. Then he was given acorns from L. glutinosus , each of which was packed with a solid nutty seed covered with a very thin pale brown skin, with no pulp at all. The squirrel quickly took them and ate all the contents leaving only the hard husks. He clearly recognized the fruit and ate it voraciously, holding the acorn with his front feet and using the lower incisors to scoop out the contents. The squirrel sat on his haunches as he ate the seed, with his tail curled against his back.

The squirrel was released a few days after capture; he jumped from cage to ground, poked about the camp for a few minutes, and finally fled into the forest uttering a low and guttural series of chucks as he disappeared from sight into the undergrowth.

Hyosciurus ileile , like its montane relative H. heinrichi , is a consumer of fruits and arthropods (tables 51, 57). Acorns ( Lithocarpus fruits) and figs seem to be preferred, but other kinds of fruits with large seeds are also taken—the variety of fruits consumed is greater than that of H. heinrichi , at least judged by our samples. Remains of pill bugs (Crustacea, Isopoda), Geophilomorph and Scolopendromorph centipedes, scorpions, orthopterans, occasional ants, ant pupae, adult beetles, and large, long-legged cursorial beetle larvae were found in stomachs, pointing to leaf litter habitats on the ground surface as the source of prey.

The question of why H. ileile seems to be uncommon within its range compared to H. heinrichi needs to be addressed in future ecological studies. Relative abundance of oaks in tropical lowland evergreen rain forest as opposed to montane forests may be one of several factors responsible. Musser, in the process of collecting botanical samples, observed that oaks in the lowlands were widely scattered throughout the forest, usually single trees here and there, occasionally in small groves of three to four trees.

TABLE 50 Results of Principal-Components Analysis Contrasting Population Samples of Hyosciurus ileile Principal components are extracted from a covariance matrix of log-transformed values for 18 cranial and 1 dental variable; see figure 45.

Although L. celebicus was seen more frequently than L. glutinosus , neither of the two was common.

Oaks are more abundant in lower and upper montane forests and at least three species are present. Between 1400 and 1700 m on ridgetops and hillsides of Gunung Kanino and the lower slopes of Gunung Nokilalaki, L. glutinosus and L. elegans occur with chestnuts ( Castanopsis acuminatissima ) to form extensive groves in which some other trees are intermingled (Calophylum, for example). Those two oaks are replaced by Lithocarpus havilandii on higher slopes and the summit of Gunung Nokilalaki where it comprises one of the most common components of the forest, even forming extensive groves in places.

Both species of ground squirrels consume oak fruit and arthropods found in leaf litter (see table 57). Within the context of the distribution of oaks described above, population size of H. ileile in lowland forests may be constrained by the lower abundance of oaks compared to their higher abundance in montane forest habitats. These observations represent unquantified anecdotal information, but provide a hypothesis that could be tested by results from careful ecological inquiries.

Gunung Ile-Ile or elsewhere in Pegunungan Peleleh, which forms the mountainous backbone of the northwestern portion of the northern peninsula, would be an ideal place for a transect study extending from the lowlands into the mountain forests. Hyosciurus ileile has been taken in a high montane habitat; presumably it also occurs at lower altitudes down through tropical lowland evergreen rain forest, and present data indicate H. heinrichi is absent from the northern peninsula. In the absence of H. heinrichi , is the population size of H. ileile greater in the montane habitats and reduced in lowland forests? And along such a transect line, does the relative abundance of oak species resemble that observed by Musser in the central core of the island?

ECTOPARASITES: In addition to a host-specific sucking louse, Hoplopleura ileile , n. sp., this ground squirrel is parasitized by immature stages (larvae and/or nymphs) of ticks (Acari, Ixodidae ) belonging to the genera Dermacentor and Haemaphysalis ( Durden et al., 2008; also see table 56 and the ectoparasites account for Rubrisciurus rubriventer ). Further, ectoparasitic laelapid mites were recovered from pelts AMNH 223540 and 225460, both from Gunung Kanino in central Sulawesi at elevations of 1402 and 1463 m in the 1970s.

SYMPATRY: The range of H. ileile on the northern peninsula of Sulawesi overlaps the ranges of Prosciurillus leucomus , P. murinus , and Rubrisciurus rubriventer (table 6). On Sulawesi’s central core, the range overlaps those of P. topapuensis , P. murinus , and R. rubrisciurus (table 6). Along his transect in the northern portion of central Sulawesi, Musser trapped H. ileile in the same traplines, and sometimes in the same traps, as P. topapuensis , P. murinus , and R. rubrisciurus (table 51). See the account of H. heinrichi for the topographic relationships between the geographic and altitudinal distributions of H. ileile and H. heinrichi .

SYNONYMS: None.

We leave our accounting of Sulawesi’s endemic species of squirrels and turn to

TABLE 51

Summary of Habitat, at Trapping Sites, Stomach Contents, and Other Relevant Information for Specimens of Hyosciurus ileile Collected by Musser in Central Sulawesi, 1974–1976

Collection locality, specimen numbers, elevation, and month and year of collection are included. Descriptions of the trapping sites and contents of stomachs, slightly edited, are from Musser’s field journals (in mammalogy archives at AMNH). Two collection localities (Sungai Sadaunta and Sungai Tolewonu) are in tropical lowland evergreen rain forest. Sites on Gunung Kanino are in lower montane rain forest. All squirrels were caught during the day in Conibear traps (rats taken in the same traps were caught during the night). Unless noted differently, trapping sites were in primary forest formations.

Locality, AMNH Elevation and (ASE) numbers (m) Date Trap site, other species trapped there, and stomach contents

Sungai Sadaunta 884 Oct. 1974 On decaying, wet Pigafetta palm trunk lying across stream in

224618 (2225) intact streamside forest. Three rats were trapped on same spot during different nights: two Echiothrix centrosa and one Taeromys callitrichus . Stomach: full of two kinds of figs (skin, rind, large and tiny seeds), brown acorn mash, hard pieces resembling a calyx, and soft green pieces, along with a few cut pieces of large beetle larvae, several young instar beetle larvae (with legs), and remains of an adult beetle.

224619 (2348) 915 Nov. 1974 On limb of understory tree growing across ravine containing main upper tributary of the Sungai Sadaunta; base of the trunk is at edge of stream, with the trunk leaning over the water and its upper branches reclining on the opposite terrace. The main connecting limb (3–5 ft in diameter) on which the trap was placed is mossy, and the moss is trampled, indicating frequent use; trap set about 7 ft from stream level. Examples of three kinds of tree squirrels ( Rubrisciurus rubriventer , Prosciurillus topapuensis , and P. murinus ) and an arboreal rat ( Rattus marmosurus ) were taken in the same spot. So three tree squirrels, one ground squirrel, and at least one species of arboreal murid used the same living pathway as a bridge over the stream. Stomach: full, with hard seeds covered by tissue (resembling seeds from the tall ginger but smaller and capsule shaped), remains of a cockroach and adult beetle, and several large long-legged cursorial beetle larvae (large pinkish orange larval abdomens, most without head and thorax, and part of thorax with legs of one larva).

224620 (2293) 960 Nov. 1974 On rotting trunk lying across high (3 ft) banks of main upper tributary of Sungai Sadaunta. Trunk covered with moss, ferns, and shrubs at either end, but clear in middle. Here sides of the canyon are steep, the ravine heavily eroded and cluttered with rotting Pigafetta palm fronds— these palms are common. Rattan is thick in understory of hillside and terrace forest above. Trees seem short, forest smaller than at 2500 ft. Caught during early morning. Stomach: full, mostly remains of large long-legged cursorial coleopteran, fragments of adult beetles (including wings), and some unidentifiable fruit mash.

TABLE 51 (Continued)

Locality, AMNH Elevation and (ASE) numbers (m) Date Trap site, other species trapped there, and stomach contents

Guning Kanino 1402 Feb. 1974 On understory tree growing from one bank of Sungai Salubeka

225459 (2531) out over stream to other bank, main trunk about 4 ft above water.

225460 (2532) Tree is densely leaved; branches, twigs, and leaves nearly hidden

225461 (2537) by hanging, wet moss. Trunk is covered with moss that is beaten down along the top as though used often by animals to cross the stream. ASE 2531 caught during the day between 10:00 and 12:00; cloudy and drizzly during that time. Stomach of ASE 2531: full of part purple, fibrous fruit with hard red seeds; large chunks of a white seed; small pieces of yellowish green mash; pieces of brown material (probably from acorns), a few remains from small adult beetles. Stomach of ASE 2532: full, mostly dry and tan acorn mash with chewed remains of a few long-legged cursorial coleopteran larvae, several adult beetles, a scorpion, and a geophilomorph centipede. Stomach of ASE 2537: filled with acorn mash, large oblong red seeds (same as in ASE 2531), a small cursorial beetle larva, and abundant remains of adult beetles.

225462 (2536) 1402 Feb. 1974 On slender (2–3 in. diameter) understory tree growing across Sungai Salubeka, about 2 ft above the stream, well downstream in canyon. Stomach: full of purple and tan mixture of acorn mash, skin, and large tough seeds of another fruit, chewed remains of several adult beetles, and a grasshopper or katydid ( Orthoptera ).

223540 (961) 1463 Nov. 1973 In prominent runway on steep slope; caught about 9:00 a.m.

225463 (2459) 1512 Jan. 1974 In prominent damp runway alongside decaying, moss-covered trunk lying on terrace below ridgetop and above deep ravine. Dry oak-chestnut forest (several rotting trunks lying on ground are all chestnuts). Stomach: full, mostly acorn mash and remains of arthropods (an ant, an orthopteran, one large adult beetle, a few chewed long-legged cursorial beetle larvae, and a leach).

Sungai Tolewonu 168 Jan. 1976 On rotting trunk lying across second large tributary; part of an old

226497 (4061) tree-fall that is covered with the climbing gingerlike monocot, shrubs, ferns and palm rosettes; about 6 ft above surface of water; caught between 8 and 10:30 a.m. Rubrisciurus rubriventer , and the rat, Maxomys hellwaldii , were taken on same spot. Stomach: half full with mostly figs (chunky pieces of white pulp mixed with tiny amber seeds on stalks); remains of one cursorial coleopteran larva and one adult beetle.

226498 (4073) 290 Jan. 1976 On wet, moss-covered decaying section of trunk that has broken into two pieces and lay in a V across headwater branch of second long tributary. Old and rotten tree-falls are everywhere; many trees had succumbed to old landslides on very steep slopes. Away from these disturbed spots, the steep slopes are covered with intact hill forest; the understory is dense and scrubby; scattered taller trees form a broken canopy. Pometia pinnata is common near the stream; tall Palaquium obovatum and Mussaendopsis beccariana are scattered over the steep slopes and contribute to the top canopy. Squirrel trapped about 7:00 a.m. The rat, Paruromys dominator , was taken in same spot. Stomach: partially full of a bit of pink and dark gray fruit mash (the gray is remains of figs, endosperm, and tiny seeds) and arthropod remains (an adult beetle, pill bug [Crustacea, Isopoda, Oniscoidea], remains of a large scolopendromorph centipede, a scorpion, and a few ant pupae).

TABLE 51 (Continued)

describing samples that represent two species indigenous not to Sulawesi but to forested islands, peninsulae, and mainland in the Indomalayan region west of Sulawesi. Their occurrence on the island represents anthropogenic introductions, likely as pets. We did not survey the skins of these specimens for sucking lice.

THE NONENDEMIC SQUIRRELS REPORTED FROM SULAWESI

Four scientific names associated with Sulawesian squirrels— microtis , erythromelas , erythrogenys , and schlegelii —identify two species native to the Sunda Shelf and not part of the endemic Sulawesian sciurid fauna. Here we briefly discuss the identities of each.

Callosciurus notatus ( Boddaert, 1785)

Sciurus Notatus Boddaert, 1785: 119 . Sciurus (Rheithrosciurus) microtis Jentink, 1879:

41.

LECTOTYPE AND TYPE LOCALITY: The lectotype of Sciurus microtis is an adult female (RMNH 13349, specimen ‘‘ e ’’ in Jentink’s, [1888: 29] catalog) obtained by J.E. Teysmann in 1877. Jentink (1888: 29) incorrectly indicated 1887 to be the year the specimens were collected. Chris Smeenk writes in his unpublished catalog of Leiden mammalian types: ‘‘Teysmann worked on Salayar from 16 November to 11 December 1877 ( Van Steenis-Kruseman, 1950: 524). The year 1878 given by Jentink (1887, 1888) thus refers to the year when the material was received in Leiden.’’ He explained (in litt., 2008) that ‘‘ Jentink (or his predecessors, in this case Schlegel) very often confused the years of collection and reception.’’ The lectotype consists of a skin mounted in a live pose and a skull (extracted from the mount after Jentink’s tenure at Leiden). The skin is overstuffed, leaving an open seam on the head, and the tail is missing. The skull lacks part of the braincase but is in the best condition and the most complete of all the six syntypes.

Specimen ‘‘ e ’’ was one of six specimens cataloged as types (5 syntypes) by Jentink (1888: 29). Following Chris Smeenk’s suggestion (in litt., 2008), we select specimen ‘‘ e ’’ as the lectotype, following the rules promulgated in Article 74.1 of the Code ( ICZN, 1999: 82). The other five, all collected by J.E. Teysmann in November–December, 1877, become paralectotypes: (1) specimen ‘‘ a ’’ ( Jentink, 1887: 193, 1888: 29), RMNH 13345, an adult male, skin mounted in live pose and extracted incomplete skull; (2) specimen ‘‘ b ’’ ( Jentink, 1887: 193, 1888: 29), RMNH 13346, an adult male, skin mounted in live pose and extracted partial skull; (3) specimen ‘‘ c ’’ ( Jentink, 1887: 193, 1888: 29), RMNH 13347, an adult male, skin mounted in live pose and extracted incomplete skull; (4) specimen ‘‘ d ’’ ( Jentink, 1888: 29), RMNH 13348, an adult female, skin mounted in live pose and extracted incomplete skull; (5) specimen ‘‘ f ’’ ( Jentink, 1888: 29), RMNH 13350, an adult female, skin mounted in live pose and extracted incomplete skull.

The type locality is Pulau Salayar (06 ° 059S, 120 ° 309E) off the southern coast of the southwestern peninsula (see one of the distribution maps for any species of the endemic squirrels), Propinsi Sulawesi Selatan, Indonesia.

REMARKS: In addition to the lectotype and five paralectotypes, we have also examined the six specimens discussed by Sody (1949: 92): MZB 6105–6110.

In his description, Jentink (1879: 41) noted that microtis was similar to ‘‘ Sciurus nigrovittatus ’’ (5 Callosciurus nigrovittatus ) in its fur markings, and described the species as follows:

General colour tawny blackish. On the back the hairs are black with two tawny rings; a few ones are entirely black. On the upperpart of the head and outside of legs, feet and hands the hairs have only one tawny ring. On each side of the body is a tawny lateral streak, consisting of black hairs with long tawny tips. The sides of the body present a darker colour than the upperparts of the back, each hair being here black, with a very minute tawny tip. Chin,

throat, chest belly and inside of legs with a beautiful reddish tint, produced by the long reddish tawny coloured tips of the brownish black hairs. Tail shorter than head and body.

The hairs are here black with three tawny rings;

underparts of the tail near the root and circumference of anus more reddish. A circle around the eyes feeble reddish colored. In-and outside of the very short ears closely covered with short reddish brown hairs. Whiskers and bristles on the cheeks long and black. The well arched claws black with white points. Cutting-

teeth orange; upper ones longitudinally grooved.

Major checklists of mammals have identified Jentink’s microtis as a population of Callosciurus notatus occurring on Pulau Salayar ( Ellerman, 1940; Laurie and Hill, 1954; Corbet and Hill, 1992; Hoffmann et al., 1993; Thorington and Hoffmann, 2005). The species is of medium body size with dark brown upperparts, buff and black stripes on sides of the body and underparts ranging from gray to chestnut (see measurements and the illustrations in Medway, 1969, and Payne et al., 1985). Callosciurus notatus has been documented from peninsular Thailand, the Malay Peninsula; the larger islands of Sumatra, Java, Bali, and Borneo, and some smaller islands on the Sunda Shelf; and Lombok and Salayar to the east of the Shelf (Kitchener et al., 1980; Corbet and Hill, 1992: 291). At least 77 scientific names, originally identifying separate species, or subspecies of notatus , have been attached to C. notatus , but as Corbet and Hill (1992: 291) noted,

In spite of the enormous number of named forms, geographical variation is slight compared with that in most other widespread species of Callosciurus . The dorsal pelage and tail are very uniform with only a little local darkening, and most variation concerns the colour of the ventral pelage (pale grey to dark chestnut), the clarity of the lateral dark and light stripes, and the extent, or absence, of reddish brown hairs in the tail-tip.

Corbet and Hill wrote of the dull lateral stripes of squirrels in the Salayar sample, and their dull buff venters, which closely resemble some squirrels in samples from Java. The variation in color of underparts of the specimens we examined range from grayish buff to pale orange, and do match samples from Java we studied, as does the intensity of the lateral stripes.

A careful and comprehensive analysis of geographic variation in pelage coloration, morphometric traits, and molecular characteristics for C. notatus has yet to be realized. Results of such a study may identifiy discrete populations that represent different species, possibly assemble related geographic populations, and hopefully point to the geographic origin of the populations on Lombok and Salayar. The occurrence of C. notatus on those two islands east of the Sunda Shelf may represent natural dispersal, but introduction by humans seems more likely. Chris Smeenk (in litt., 2009) writes us that,

For ages, there has been an intensive trade,

particularly by people from Madura, Celebes

(Makassar and the Minahasa) and Ternate,

with all parts of the Archipelago. Naturally,

Java (e.g., Surabaya) has always been a highly important and profitable destination. Salayar being at the south end of Celebes and relatively close to Madura and Surabaya, this would have been a very likely place where Javan pets could have landed. The population of Lombok too, must have been introduced, possibly from Bali.

That the species could be an anthropogenic introduction to Salayar is suggested by its habitat and the apparent absence of nearly all endemic Sulawesi mammal species from the island. Today little of the original semievergreen rain forest remains on Salayar, and most of the island consists of cleared or cultivated land (see map 8d in MacKinnon, 1997). We have no information about the environment on Salayar where the specimens were collected, but elsewhere within its range C. notatus is abundant in gardens, plantations, and secondary forests. On Borneo the squirrel ‘‘…can live and breed entirely in monoculture plantations’’ ( Payne et al., 1985: 236). Davis (1962: 78) wrote of Bornean populations: ‘‘This species does not appear to be common in the interior of the primary forest, but is abundant around the edges, in clearings, and in secondary growth. It is often seen in towns and cities, and does considerable damage to coconut and rubber plantations.’’ In our experience, the endemic species of Rubrisciurus , Prosciurillus , and Hyosciurus have been encountered only in primary forest habitats. No sample of any species in those three genera has ever been collected on Pulau Salayar, and of all the species of mammals endemic to Sulawesi, only a tarsier has been recorded from the island ( Groves, 2005: 128).

Although the origin of the Salayar population is unresolved, we do know that C. notatus is a member of the Sundaic mammalian fauna, that it has never been recorded from the mainland of Sulawesi or any other offshore island than Pulau Salayar, and that it is phylogenetically not closely related to the endemic Sulawesian species of Rubrisciurus , Prosciurillus , and Hyosciurus .

Callosciurus prevostii (Desmarest, 1822)

Sciurus erythromelas Temminck, 1853: 248 .

Sciurus erythrogenys Schlegel, 1863: 29 , pl. 2, fig. 3; not Waterhouse (1942 [1943]).

Sciurus Schlegelii Gray, 1867: 278 (5 Macroxus Schlegelii Gray, 1867: 278 ).

The three scientific names are attached to two specimens that differ conspicuously in coloration of fur, but likely represent the same species: Callosciurus prevostii . We explain our reasons for this allocation in the following paragraphs.

The lectotype of Sciurus erythromelas is an adult male (RMNH 13378, specimen ‘‘ bb ’’ in Jentink’s [1888: 27] catalog) obtained by E.A. Forsten sometime betwen 1840 and 1842. The lectotype consists of a skin mounted in a live pose and a skull (extracted from the mount after Jentink’s tenure at Leiden). The skin is in very good condition except for slight damage to the mouth caused when the skull was extracted. The occiput, one ectotympanic bulla, and one zygomatic arch are missing from an otherwise complete skull. Specimen ‘‘ bb ’’ was one of two specimens cataloged as types (5 syntypes) by Jentink (1888: 27). Following Chris Smeenk’s suggestion (‘‘It is perfect for a lectotype,’’ in litt., 2009), we select specimen ‘‘ bb ’’ as the lectotype, following the rules promulgated in Article 74.1 of the Code ( ICZN, 1999: 82). The type locality is Kema (01 ° 229N, 125 ° 039E; locality 8 in gazetteer for Prosciurillus leucomus and map in fig. 11), near sea level on the coastal plain of the northeastern end of the northern peninsula, Propinsi Sulawesi Utara, Indonesia. RMNH 13378 is also the holotype of Sciurus erythrogenys Schlegel, 1863 and Macroxus schlegelii Gray, 1867 (see the discussion to follow).

The other specimen, also collected by E.A. Forsten between 1840 and 1842, but from Manado (01 ° 309N, 124 ° 509E; locality 4 in gazetteer for Prosciurillus leucomus and map in fig. 11) on the coastal plain near sea level, is the paralectotype of Sciurus erythromelas : specimen‘‘ cc ’’ ( Jentink, 1888: 27), RMNH 13379, an adult male. The skin, mounted in a live pose with the skull still in the skin, is in good condition, and retains its bright color pattern.

Chris Smeenk is preparing a catalog of the ‘‘Type-specimens of recent mammals in the National Museum of Natural History, Leiden’’ and gave us permission to quote his account for Sciurus erythromelas , which is the best place to begin our discussion of the type specimens. Chris writes that:

Temminck (1853) does not state how many specimens he had before him; he writes: ‘‘Il est heureux que nous ayons pu recevoir cet écureuil nouveau dans son pelage des noces ou à l’époque du rut, en même temps que nous sont parvenus, de la même contrée, des individus tués dans la saison opposée de l’année; obtenus séparément et capturés dans des localités différentes,’’ so one may conclude that he had more than two skins [‘‘ Reading the French text carefully, one must conclude that he had several specimens before him of different coloration and from various localities, but only two are left, both collected by Forsten’ ’; C. Smeenk, in litt., 2009]. The provenance is specified (p. 249) as ‘‘ Les patries septentrionales de Célèbes, Gorontalo et Kema. ’’ There are now two specimens in the Leiden Museum, from Manado and Kema, none from Gorontalo ; no collection date is recorded, so it is not clear why Temminck concluded that the differences in coloration were seasonal. The animal from Manado (RMNH 13379) is also described and illustrated by Schlegel (1863: 28–29, pl. 2 fig. 2), who redescribes the specimen from Kema (RMNH 13378) as Sciurus erythrogenys . …

Forsten collected in NE Celebes between March 1840 and April 1842. … His diary, a copy of which (in an unknown hand) is preserved in the archives of the Leiden Museum, sheds no light on the exact provenance and collection dates of these animals. The squirrels described here must have been introduced from Borneo (Musser, 1987: 80), perhaps as pets ; the coastal towns of Manado, Kema and Gorontalo have always been centres of trade.

Ten years after Temminck’s account was published (see appendix 6 for Temminck’s description in the original French), Schlegel’s [1863: 29] description of Sciurus erythrogenys appeared. Schlegel regarded the squirrels from NE Celebes described by Temminck (1853: 248) as Sciurus erythromelas to represent two distinct species. He restricted the name S. erythromelas to the specimen from Manado (RMNH 13379) and described the squirrel collected at Kema (RMNH 13378) as S. erythrogenys , named for its reddish cheeks (the description in the original French is provided in appendix 7).

Gray (1867: 278) later renamed the same specimen: ‘‘ Sc. schlegelii (with cheeks red), Sc. erythrogenys, Schlegel , l. c. t. 2. f. 3 (1863), not Waterhouse’’ because erythrogenys was proposed earlier by Waterhouse (1842 [1843]) for the African red-cheeked rope squirrel Funisciurus leucogenys ( Waterhouse, 1842) . As Thorington and Hoffmann (2005: 792) explained, ‘‘ Waterhouse (1842 [1843]) renamed this species [ F. leucogenys ] erythrogenys , ‘redcheeked’ in an attempt to replace the inappropriate name leucogenys , ‘white-cheeked.’ This is an unjustified emendation.’’

Gray proposed schlegelii within his account of Macroxus atrocapillus (5 Sciurus atricapillus Schlegel, 1863 ; incorrectly quoted by Gray as ‘‘ atrocapilla ’’; atrocapillus is currently treated as either a synonym of Callosciurus prevostii or a valid subspecies, C. p. atrocapillus ; Payne et al., 1985; Corbet and Hill, 1992; Thorington and Hoffmann, 2005). Gray’s wording, ‘‘ Sc. Schlegelii (with cheeks red), Sc. erythrogenys, Schlegel , l. c. t. 2. f. 3 (1863), not Waterhouse,’’ was meant to explicitly identify schlegelii as a replacement name for Schlegel’s erythrogenys within Sciurus , the genus in which Schlegel described erythrogenys . Gray used Macroxus for species of Oriental squirrels, not Sciurus , so the correct combination for his replacement name is Macroxus schlegelii (we are grateful to Chris Smeenk and Al Gardner for their insight into this nomenclatural issue).

RMNH 13378 is the holotype of both Sciurus erythrogenys Schlegel, 1863 , and Macroxus schlegelii Gray, 1867 , as well as the lectotype of Sciurus erythromelas Temminck, 1853 (see above).

The two specimens are differently colored. The upperparts of the head and body as well as the tail of the lectotype of Sciurus erythromelas (RMNH 13378 from Kema) is dark brown speckled with buff and black, including the outside of the limbs. Sides of the body are marked by a whitish gray stripe above a black stripe. The underparts are reddish orange, which extends to the inside of the front and hind limb but not to the feet, which are black, and the cheeks and throat are orange-red.

The paralectotype of Sciurus erythromelas (RMNH 13379 from Manado) is appreciably darker. The head and back, back of the thighs, and tail are blackish. A single pale stripe (brownish black speckled with gray) marks each side of the body (no black stripe). The area around the mouth, cheeks, shoul- der, upper arms, and front of the thighs are brownish black speckled with buff (appears grizzled). The underparts from chin to base of the tail, including most of the forearms (the blackish upperparts extend along the outside of the forearms as a narrow blackish stripe towards the front feet) are dark reddish orange; sides of the feet and parts of the dorsal surfaces are the same color (although tops of the hind feet are washed with brownish black).

To us, the color patterns of the two types are reminiscent of Callosciurus prevostii , a Sundaic endemic occurring on peninsular Thailand and Malaya, the larger Sundaic islands of Sumatra and Borneo, and smaller islands on the Sunda Shelf, but not Java ( Corbet and Hill, 1992; Thorington and Hoffmann, 2005). The species is medium in body size (length of head and body, 200– 267 mm; length of tail, 202–273 mm; length of hind foot, 45–61 mm; weight, 250–500 g; see Medway, 1969; Payne et al., 1985), physically smaller than the Sulawesian Rubrisciurus rubriventer , but larger than species in the Prosciurillus leucomus group (see table 3). On Borneo, tall and secondary forests are the habitat of C. prevostii , where it stays mostly in the upper canopy, descending to the ground when necessary to cross gaps in the canopy; it will also enter gardens and plantations next to tall forest ( Davis, 1962; Payne et al., 1985: 234). In parts of the Malay Peninsula, the squirrel is locally common in oil plantations and adjacent forest, and is ‘‘popular as a pet because of its bright coloration’’ ( Medway, 1969: 59).

Probably no other species of Callosciurus is as geographically variable in color patterns of the fur, not only among peninsular and insular landmasses but also within large islands, especially so among populations on Borneo ( Corbet and Hill, 1992; Musser’s study of specimens representing Indomalayan species). Payne et al. (1985; pl. 24) provide color renditions of the primary chromatic patterns found among Bornean populations and their general distribution on the island. Some populations exhibit intergradation in these patterns, other populations isolated by rivers apparently do not.

In his list of squirrel specimens in Leiden, and his catalog of mammals in that institution, Jentink (1883: 132, 1888: 27, respectively) listed the types of erythromelas and erythrogenys (including schlegelii ) under ‘‘ Sciurus prevostii .’’ Later checklists arranged erythromelas and schlegelii each as a subspecies of Callosciurus prevostii with both occurring on Sulawesi ( Ellerman, 1940: 367; Laurie and Hill, 1954: 92) or associated with Bornean C. prevostii but unassigned as to subspecies status (Hoffmann et al., 1993: 423; Thorington and Hoffmann, 2005: 279); none of these checklists published after Jentink’s catalog listed Schlegel’s erythrogenys .

We agree with Jentink’s (1888: 27) identification of the lectotype and paralectotype of erythromelas as examples of Callosciurus prevostii . The population of C. prevostii living in northern Borneo in Sabah and northeastern Sarawak, C. p. pluto, is black with bright reddish orange underparts, and some specimens show a weak whitish body stripe (see Payne et al., 1985, pl. 24); the coloration closely resembles the paralectotype of Sciurus erythromelas (RMNH 13379) from Manado; the minor differences could easily be ascribed to individual variation.

The brighter lectotype of Sciurus erythromelas could be a color variant of Bornean C. prevostii . It does not exactly match any of the color examples illustrated by Payne et al. (1985, pl. 24), but its color pattern does fall somewhere between that illustrated for C. p. atricapillus from eastern Sarawak (possessing dark hind feet and limbs) and that for C. p. caroli from farther to the northeast in Sarawak (with orange-red throat and cheeks). Corbet and Hill (1992: 290–291) listed S. erythromelas (along with S. schlegelii ) as a synonym, with question, of Callosciurus baluensis , an endemic of the north Bornean mountains in Sabah and Sarawak, with an altitudinal range of 300 to 2000 m: ‘‘The type of erythromelas (Leiden Museum) matches Bornean C. baluensis rather closely in pelage but has more orange-brown on the cheeks.’’ Musser examined the holotype of baluensis , and other specimens, including the holotype of baramensis, the other name associated with C. baluensis . None of the examples of C. baluensis has the orange-red cheeks and black feet seen in the lectotype of Sciurus erythromelas .

For now we prefer to associate erythromelas (along with erythrogenys and schlegelii ) with Bornean C. prevostii rather than C. baluensis . This identity can be tested in two ways. Although the skull of the lectotype of erythromelas is damaged, enough of its dimensions could be measured and the values compared with those obtained from samples of C. baluensis and geographic samples of C. prevostii from Borneo in a multivariate analysis (principal-components analysis, for example). Results from analyses of DNA sequences extracted from hairs or bits of dry skin from the lectotype of erythromelas and samples of C. baluensis and C. prevostii would also be revealing. If in the future, the lectotype of Sciurus erythromelas can be unambiguously identified as an example of C. baluensis , that latter name will have to be changed to Callosciurus erythromelas , which would then be a species with its indigenous range in northern Borneo, but its type locality outside of that range in northeastern Sulawesi.

We do not know how many specimens (and their collection localities) Temminck (1853) originally used for his description of Sciurus erythromelas . We do know that two remain in the collection at Leiden, that each is different in color pattern, that no other specimen resembling either one has ever been collected on Sulawesi since Forsten’s activities between 1840 and 1842, and that the two types do not represent any species of squirrel truly endemic to Sulawesi. Both specimens are likely examples of geographic variation described for the populations of Callosciurus prevostii on Borneo. We suspect the material obtained by Forsten to have originally been transported as pets from Borneo to the Minahasa region (in which the coastal towns of Kema and Manado are located).

With this last account, we leave the squirrels and turn to the eight species of sucking lice parasitizing eight of the 10 species of Sulawesi’s endemic squirrels. All samples of lice represent new species, and their descriptions are contained in the following segment of our report.

THE SPECIES OF SUCKING LICE FROM THE SQUIRREL HOSTS

Seven subheadings contain information covering descriptions of each of the eight new species of sucking lice, all of which belong to the genus Hoplopleura (Holopleuridae): (1) holotype of the species of Hoplopleura and type locality of the squirrel host; (2) referred specimens of lice to a new species; (3) geographic distribution on Sulawesi; (4) etymology; (5) diagnosis; (6) description; (7) relevant remarks. Refer to the preceding sections for classification of the squirrel hosts as well as morphological characteristics of each host species, its geographic and altitudinal distributions on Sulawesi, its ecology, and other information.