Prosciurillus rosenbergii

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

Prosciurillus rosenbergii

We examined all the specimens in RMNH and ZMT, and most from SNSD. Feiler’s (1990) informative report documenting historical material in SNSD collected from the Kepulauan Sangihe did not tie individual specimens by catalog number to different island provenances, but Clara Stefen kindly provided that information for the few SNSD specimens we did not examine. Kepulauan Sangihe, Pulau Sangihe (also spelled

‘‘Sangir’’), 03 ° 339N, 125 ° 309E; BMNH 1876.

10.21.4; FMNH 31846; RMNH 13351–53

(specimens ‘‘ a–c ’’ in Jentink’s [1888: 23]

catalog; paralectotypes of Sciurus Rosenbergii ), RMNH 13354 (specimen ‘‘ d ’’ in Jentink’s [1888: 24] catalog; paralectotype of

Sciurus rosenbergii ) and RMNH specimens

‘‘ m ’’ and ‘‘ n ’’ in Jentink’s (1888: 24) catalog;

SNSD 2551–53, 2559; ZMB 5128, 84973,

92377. Pulau Sangihe, Tabukan: SNSD 499,

500. (Chris Smeenk provided this note: one should be careful here. Although the specimens obtained by von Rosenberg are labeled

‘‘Sanghir’’, in his manuscript notes von

Rosenberg uses that name for the entire archipelago, and it cannot now be ascertained whether his material came indeed, all or part of it, from the main island P. Sangihe.

Jentink’s specimen ‘‘ d ’’ collected by Hoedt seems more reliable in this respect, as Hoedt labeled all his other specimens ‘‘Siao’’. The specimens obtained from the dealer G.A.

Frank again, are labeled ‘‘Sangir’’ without further specification.) Kepulauan Sangihe, Pulau Siau (also spelled

‘‘Siao’’), 02 ° 499N, 125 ° 239E: RMNH

13355–61 (specimens ‘‘ e–k ’’ in Jentink’s

[1888: 24] catalog; paralectotypes of Sciurus

Rosenbergii), RMNH 13362 specimen ‘‘l’’ in

Jentink’s [1888: 24] catalog; lectotype of

Sciurus Rosenbergii ); SNSD 323, 2554,

2555–58, 2560, 2561.

Kepulauan Sangihe, Pulau Tahulandang (also spelled ‘‘Tahoelandang,’’ ‘‘Tagulandang,’’ and ‘‘Tangulandang’’), 02 ° 219N, 125 ° 259E: SNSD B2842 (lectotype of Sciurus tingahi ), B2843 and B2844 (paralectotypes of Sciurus tingahi ).

Kepulauan Sangihe, Pulau Ruang (also spelled ‘‘Roeang’’), Gunung Api, 02 ° 189N, 125 ° 229E for Pulau Ruang: SNSD 2847, 2848, 2850, 2851 (paralectotypes of Sciurus tingahi ).

Prosciurillus leucomus (Müller and Schlegel, 1844)

Sciurus leucomus Müller and Schlegel, 1844: 87 . Sciurus leucomus occidentalis Meyer, 1898: 2 .

LECTOTYPE AND TYPE LOCALITY: The lectotype of Prosciurillus leucomus is an adult male (RMNH 13344, specimen ‘‘ b ’’ in Jentink’s, [1888: 24] catalog) obtained by E.A. Forsten in 1840 from ‘‘ Célèbes, Kéma.’’ It consists of a skin mounted in a live pose and a slightly damaged skull (extracted from the mount after Jentink’s tenure at Leiden). The nasals are missing, there is a large hole in the right frontal bone, one upper molar and one maxillary premolar are missing, and four upper molars and one premaxillary molar are present but have fallen out of their alveoli; the mandible is intact and all teeth are present. Dental and some cranial measurements are listed in table 13.

Specimen ‘‘ b ’’ was one of eight specimens cataloged as types (5 syntypes) by Jentink (1887: 191, 1888: 24). Following Chris Smeenk’s suggestion (in litt., 2008), we select specimen ‘‘ b ’’ as the lectotype, following the rules promulgated in Article 74.1 of the Code ( ICZN, 1999: 82). The other seven become paralectotypes: (1) specimen ‘‘ a ’’ ( Jentink, 1888: 24), RMNH 13343, an adult female, a skin mounted in live pose and extracted skull collected by E.A. Forsten on April 6, 1840, from ‘‘ Célèbes, Ménado’’; (2) skeleton ‘‘ a ’’ ( Jentink, 1887: 191), RMNH 39392, an adult skull and postcranial skeleton, sex unknown, collected by E.A. Forsten in 1841 from ‘‘Célèbes’’; (3) skeleton ‘‘ b ’’ ( Jentink, 1887: 191), RMNH 39393, a juvenile skull and postcranial skeleton, sex unknown, collected by E.A. Forsten in 1841 from ‘‘Célèbes’’; (4) skull ‘‘ c ’’ ( Jentink, 1887: 191), RMNH 39395, an adult damaged cranium with missing dentaries, sex unknown, collected by E.A. Forsten in 1841 from ‘‘Célèbes’’; (5) skull ‘‘ d ’’ ( Jentink, 1887: 191), RMNH 39396, an adult skull with incomplete right dentary, sex unknown, collected by E.A. Forsten in 1841 from ‘‘Célèbes’’; (6) skull ‘‘ e ’’ ( Jentink, 1887: 191), an adult damaged skull, sex unknown, collected by E.A. Forsten in 1841 from ‘‘Célèbes’’; and (7) skull ‘‘ f ’’ ( Jentink, 1887: 191), a damaged skull, sex unknown, collect- ed by E.A. Forsten in 1841 from ‘‘ Célèbes.’’ Chris Smeenk (in his unpublished catalog of types at Leiden) has also included specimen ‘‘ k ’’ ( Jentink, 1888: 25), RMNH 39394, in the type series, an adult male, mounted in live pose, no skull present, collected from ‘‘ Célèbes, Pagowat.’’ Although no further details are given, this can only be the ‘‘ Sciurus ’’ collected by E.A. Forsten from Pagowat (5 Paguat) during November 5–12, 1841, as mentioned in his diary, a copy of which (in an unknown hand) is preserved in the archives of the Leiden Museum (Chris Smeenk, in litt., 2008).

The type locality is Kema, 01 ° 229N, 125 ° 039E, near sea level on the coastal plain of the northeastern peninsula (locality 8 in gazetteer and map in figure 11), Propinsi Sulawesi Utara, Indonesia.

Chris Smeenk provided us with pages dealing with the Sulawesi squirrels from his manuscript version of the ‘‘Type-specimens of recent mammals in the National Museum of Natural History, Leiden,’’ and it is pertinent here to reproduce his remarks regarding the lectotypes and paralectotypes of ‘‘ Sciurus leucomus ’’:

Müller & Schlegel (1844) do not state how many specimens they had before them. The skulls of Jentink’s skins a and b (RMNH 13343 and 13344) were extracted at a later date. Jentink (1887) also lists as types two skeletons and four separate skulls collected by Forsten. Although Müller & Schlegel do not mention skeletal material, they had these specimens available when writing their descriptions, and hence they are included in the type series here.

Forsten arrived in Manado in Northeast Celebes on 22 March 1840 ; on 15 April he made his headquarters at Tondano, from where he explored Minahasa (Manado) District in the northeastern tip of the island. He arrived back in Manado on 24 April 1841, departed from Kema on 14 June, and landed in Ternate on 19 June. On 9 September 1841 he proceeded to Gorontalo in NE Celebes, where he arrived on

TABLE 13

Age, Sex, and External, Cranial, and Dental Measurements (mm) for Holotypes Associated with Prosciurillus topapuensis (includes hirsutus), Prosciurillus weberi , and Prosciurillus leucomus (includes occidentalis)

The taxa were originally described as members of either Sciurus or Callosciurus .

18 September. From there, he travelled further west along the coast as far as Paguat, and returned in Gorontalo on 14 November. He arrived back at Kema on 28 November and worked again in Minahasa district , until he finally left Celebes on 14 April 1842 (Veth, 1875; 98, 107; Van Steenis-Kruseman, 1950: 179; and Forsten’s unpublished diary) .

RMNH 13343 [the lectotype] must be the animal collected on 6 April 1840 near Manado. In Forsten’s Celebes diary, a copy of which (written in an unknown hand) is preserved in the archives of the Leiden Museum, the entry for that day relates that his hunter brought ‘‘Eene Securus [lapsus by the copyist for Sciurus ] welke ik geloof nieuw te zijn, kenbaar aan eene witte plek op de schouderen, …’’ (a Sciurus which I believe to be new, characterized by a white spot on the shoulders). In the same diary, Forsten records having collected ‘‘eene

TABLE 14

Age, Sex, and External, Cranial and Dental Measurements (mm) for Holotypes of Prosciurillus alstoni (includes sarasinorum , mowewensis , and elbertae ) and Prosciurillus rosenbergii (includes tingahi )

The taxa were originally described as members of Sciurus .

Sciurus’’ at Pagowat (Paguat) during the week of 5–12 November 1841. This is almost certainly Jentink’s skin k (RMNH 39394), the pedestal of which reads ‘‘ Sciurus leucomus Forst Pagowat Célebes’’ in C.J. Temminck’s handwriting, though Forsten is not mentioned as the collector. Jentink (1888) must have overlooked Forsten’s diary note and so did not mention this specimen as a type; it is included in the type series here.

EMENDED DIAGNOSIS: Prosciurillus leucomus is the first named of the five species in the P. leucomus group, sharing with those other members moderate body size and tail equal to, or shorter than, length of the head and body. It contrasts with the other four species by the following combination of pelage traits: (1) dorsomedial surface of each ear covered with long black hairs that project beyond the ear rim to form a prominent tuft; (2) inside of ears densely covered with bright ochraceous fur; (3) white, whitish buff, or grayish white patches of variable intensity and size on the nape behind the ears; (4) no black middorsal stripe extending from neck to base of tail; (5) reddish orange, orange-red, or ochraceous underparts; and (6) a geographic distribution restricted to the northern peninsula of Sulawesi where it ranges from coastal lowlands to montane forests.

GEOGRAPHIC AND ALTITUDINAL DISTRI- BUTIONS: Apart from the sample collected on Pulau Lembeh off the coast of the northeastern peninsula, all voucher specimens identified as Prosciurillus leucomus come from the mainland of the northern peninsula of Sulawesi (see gazetteer and map in fig. 11). Collection localities are scattered over the peninsula from the coastal plain near sea level at the northeastern tip (Likupang and Teteamoet, localities 1 and 2 on the map in fig. 11) to Bumbarujaba (locality 25) near the southern margin of the peninsula. Most localities lie between the coastal plain and 1000 m in what still is or once was tropical lowland evergreen rainforest habitats. The site at 1700 m in montane forest on Gunung Ile-Ile locality is the exception and the highest elevation recorded for any sample of the species.

Bumbarujaba is the most southern collection site for P. leucomus and how much farther south it may occur is unknown at present. Approximately 70 km south of Bumbarujaba in the mountains around Danau Lindu where Musser worked are the northernmost records for P. topapuensis ; and Kuala Navusu, northeast of Lindu on the eastern coastal lowlands of the central core, approximately 100 km south of Bumbarujaba, is the most northern collection site for P. alstoni (see gazetteers and map in fig. 11). What kind of squirrel occurs in the landscapes between those two places and the southern end of the northern peninsula is unknown.

In addition to Prosciurillus leucomus , several other Sulawesi endemic mammals are recorded from Pulau Lembeh. There is another squirrel, P. murinus (see that account); a murid rodent, Rattus hoffmanni ( Musser and Holden, 1991) ; a macaque, Macaca nigra ( Fooden, 1969) ; the Celebes wild boar, Sus celebensis ( Groves, 1981) ; and the babirusa, Babyrousa babyrousa ( Groves, 1980a; Grubb, 2005: 637, recognizes several species, listing the sample from Pulau Lembeh as B. celebensis ). The closest points between island and mainland are less than a kilometer apart, and the Lembeh Strait is shallow, less than 50 m deep (Sheet NA 51- 12). During periods of Pleistocene glaciation when sea level in the Indomalayan region dropped by at least 120 m ( Bintanja et al., 2005) and possibly 200 m ( Morley and Flenley, 1987), Lembeh would have been part of the mainland (see the map in Fooden, 1969: 65). With post-glaciation rise in sea level, rodent and other mammal populations on Lembeh were separated from the mainland. This scenario seems the most likely explanation for the occurrence of the three rodent species on the island, although we cannot rule out over-water dispersal. Whatever the process, squirrels in available samples of each species (as well as the other mammals referred to above) are similar in pelage coloration and morphometric traits to those in samples from the mainland, suggesting no significant genetic isolation between island and mainland populations.

DESCRIPTION: Müller and Schlegel (1844: 87) aptly described the diagnostic features of Prosciurillus leucomus (translated from the original Dutch by Chris Smeenk, in litt., 2008; the Dutch text is reproduced in appendix 2):

9.) Sciurus leucomus, Forsten , n. sp. Size and shape as in the two preceding ones [ Sciurus vittatus , a synonym of the Sundaic Callosciurus notatus , and Sciurus nigrovittatus , currently

Callosciurus nigrovittatus ; see Corbet and Hill,

1992; 291–292]. Inside of the ears densely covered with brownish yellow [hairs], back [of the ears] with long black hairs, protruding far above the ears. Colour of the upper parts and outside of the legs olive-brown; the hairs with rusty-yellow rings and partly black tips. Tail checkered with the three colours mentioned. A large white patch behind the ears, on either side of the neck. Lower parts of the body rusty coloured, tending to reddish-brown. Celebes.

Samples from the northern peninsula of Sulawesi portray a squirrel of modest body size with a tail averaging as long as, or only slightly shorter than, length of head and body (length of head and body, 165–188 mm; length of hind foot, 43–47 mm; see table 15). The dense, uniformly thick (12–15 mm) coat

TABLE 15 Descriptive Statistics for Measurements (mm) of Lengths of Head and Body, Tail, Hind Foot, and Ear, and for Weight (g), Derived from Samples of Prosciurillus leucomus , Prosciurillus alstoni , Prosciurillus weberi , and Prosciurillus rosenbergii 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

covering upperparts of head and body is dark brown flecked with black and orange along the midline of the body from forehead to base of tail, but grading into brownish gray spotted with black and pale buff along sides of the body, forelegs, and hind legs. Dorsal surfaces of the front and hind feet resemble the head and back in most specimens; some are slightly darker or buffy gray. Coloration of the body fur results from the interplay of dark gray curly underfur; longer overfur composed of black hairs, each interrupted by a subterminal orange or pale buffy band; and black guard hairs barely projecting beyond the overhair layer.

This dorsal background is interrupted by distinctive color patterning on the head, ears, and neck. Top and sides of the muzzle are ochraceous, and the cheeks are either the same color as the top of the head or a grayish buff that extends to each nape patch. A prominent, dense black tuft covers the medial surface of each ear, projecting 5–10 mm beyond the free dorsal rim of the ear. Behind each ear on the nape is a large and conspicuous white, whitish buff, or whitish

TABLE 16 Descriptive Statistics for Measurements (mm) of Lengths of Head and Body, Tail, Hind Foot, and Ear, and for Weight (g), Derived from Samples of Prosciurillus topapuensis from the West-Central Core of Sulawesi 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

gray patch (20–30 mm long, 15–20 mm wide), its shape resembling an epaulet (pointed toward the posterior margin); and the inner surface of each pinnae is densely clothed with bright ochraceous hairs. The overall visual pattern evolves around an ear with an ochraceous center rimmed by black that is set off against a large whitish postauricular epaulet (see the color plate in fig. 9). While the black tufting is present in all geographic samples of P. leucomus , the nape patches vary in size, color, and frequency of occurrence; these aspects are addressed in the sections covering geographic variation and synonyms.

The short (5–7 mm thick) ventral coat ranges from bright reddish orange through orange-red to ochraceous; orange-red predominates in the samples. The hairs forming the ventral covering are pale gray along their basal half and deeply pigmented along the distal half.

The tail, equal to or slightly shorter than length of head and body (LT/LHB 5 95%– 98%, see table 15), is covered in long hairs, each patterned by alternating black and buffy or orange bands. The overall effect is rings of black and buff with buffy bands outlining margins of the tail and long black hairs forming a black terminal tuft.

Females have six teats, positioned as one postaxillary pair and two inguinal pairs. Litter size has not been recorded for P. leucomus , but one young would be usual if this aspect of reproduction is like its allopatric relatives, P. topapuensis and P. alstoni (table 57).

The skull of P. leucomus closely resembles those of P. alstoni and P. topapuensis , which are illustrated in figures 12–14. Evident in those drawings is the convex dorsal curve of the skull reflecting pronounced cranial flexion, short and wide nasals and rostrum, wide interorbital region, postorbital processes that are even with the anterior curve of braincase, faint and widely spaced temporal ridges that do not coalesce into a sagittal crest, posterior margin of the bony palate situated slightly anterior to backs of the parallel tooth rows, wide pterygoid fossae (each triangular as seen from ventral view), conspicuous posterior maxillary notch in the bony palate between the end of the tooth row and anterior portion of the pterygoid fossa, proodont (procumbent) upper incisors, and small third molar compared with the larger second and third molars and fourth premolar. The dentary has a short coronoid process and deep ramus, and deeply concave (even angularly concave in some specimens) posterior margin between condyloid and angular processes. Cranial and dental measurements are summarized in tables 17 and 19.

COMPARISONS: The geographic distribution of P. leucomus is allopatric to the ranges of P. topapuensis and P. alstoni to the south in the central core of Sulawesi, and comparisons between P. leucomus and each of these two species will be presented in those accounts. No other samples of squirrels from regions on Sulawesi outside of the northern peninsula show the distinctive color patterns seen in P. leucomus (see the diagnostic traits compared in table 12).

Other than a close resemblance in body size, tail proportions, and some cranial dimensions, P. leucomus is also unlike P. rosenbergii , the only species occurring in the Sangihe Archipelago north of the northeastern tip of Sulawesi, and north of the distribution of P. leucomus . It is the the only member of the P. leucomus group with uniform dark chestnut upperparts and brownish gray or buff underparts. No ear tufts, nape patches, or middorsal black stripe mark the dorsal coat (table 12). Unfortunately, we do not have intact skulls from the samples of P. rosenbergii and cannot provide cranial and dental morphometric comparisons between it and P. leucomus .

GEOGRAPHIC VARIATION: Samples from throughout the geographic range of P. leucomus (the northern peninsula of Sulawesi) contain squirrels closely similar in body size (table 15). The noticeable geographic variation concerns coloration of the muzzle and the underparts, along with changes in color and extent of the nape patches. Most specimens from about the middle of the northern peninsula—in the region of Gorontalo (00 ° 319N, 123 ° 039E)—to the east have ochraceous muzzles and cheeks, prominent black ear tufts, large white or whitish buff nape patches, and reddish orange underparts. Squirrels in samples from west of Gorontalo show pale buff or gray muzzles and cheeks (as opposed to bright ochraceous), and exhibit considerable variation in extent and color of the nape patches. These patches may

TABLE 17 Descriptive Statistics for Cranial and Dental Measurements (mm) Derived from Combined Population Samples of Species in the Prosciurillus leucomus Group Mean ± 1 SD and observed range (in parentheses) are listed; size of sample follows the parentheses for P. rosenbergii .

be prominent and white or whitish buff as in the squirrels inhabiting the peninsula east of the Gorontalo region; diluted, consisting of diffuse grayish patchs barely distinguishable from the rest of the neck (a few individuals from east of Gorontalo have diluted nape patches); or not present, the areas behind the ears being the same color as the rest of the neck and upperparts of the head and body. All these specimens retain a prominent black tuft projecting beyond the rim of each ear. Underparts of specimens in the western samples are paler: ochraceous is the dominant hue, a very few specimens exhibit reddish orange venters; reddish orange characterizes most specimens from east of Gorontalo, but some individuals show orange-red or ochraceous underparts. The variation in color of the nape patches and their extent has been the stimuli for recognizing the western peninsular samples as P. leucomus occidentalis ( Meyer, 1898: 2) . We elaborate on this variation and its significance in the section discussing synonyms.

The magnitudes of cranial and dental dimensions measured in our samples are similar (table 19). Mean values for certain dimensions average greater in the western samples than those from the northeastern end of the peninsula but the differences are not significant, as is reflected in figure 15 where specimen scores representing two population samples of P. leucomus —one from the northeast, the other from the western portion of the peninsula (identified in table 1)—are projected onto the first and second principal components extracted from principal-components analysis. The spread of scores along the first axis is primarily influenced by size, the larger (and sometimes older) adults spreading to the right, the smaller (and usually, but not always, younger) adults to the left. Covariation in most cranial variables spread the scores along the first axis with the most influential being breadths of the interorbit and rostrum, and lengths of the nasals, rostrum, diastema, and bony palate (table 18). The western squirrels have, on average, a larger facial skeleton, but the contrast is not significant as reflected in the complete overlap of the ellipses outlining the 95% confidence limits for the specimen scores representing each population sample.

The geographic variation in expression of nape patches within P. leucomus does not appear to have a counterpart in cranial and dental morphometric variation among our samples. However, other than the large series

TABLE 18 Results of Principal-Components Analysis Contrasting Population Samples of Prosciurillus leucomus from the Northern Peninsula Principal components are extracted from a covariance matrix of log-transformed values for 15 cranial and 1 dental variable; see figure 15.

from Bumbulan, we have small numbers of squirrels from elsewhere in the western part of the peninsula, especially from that arc between Bumbulan and Labuan Sore at the base of the peninsula. Large samples are currently available from both west (Bumbulan, for example) and east (the Minahasa region) of Gorontalo, but we were still hampered in our investigation of cranial and dental morphometric variation because so many of the skulls are damaged, the result of sampling by shotgun. Clearly, the results we describe here will have to be tested by analyzing samples from regions now unrepresented by collections, and larger series of specimens containing intact skulls.

Morphology of available samples of sucking lice ( Hoplopleura ) varies slightly in accordance with host populations of P. leucomus but the samples of parasite and host are small. A few female and nymphal lice were found on a squirrel from Rurukan ( leucomus morphology), east of the Gorontalo region, and on one collected at Labuan Sore (occidentalis morphology), to the west of Gorontalo at the base of the northern peninsula. Specimens in each sample of Hoplopleura differ slightly in shapes of the paratergal plates and lengths of their apical setae. However, males of these lice were not available for study (morphology of male genitalia is often the most important distinguishing character for different species of lice) and we consider the variation in the female paratergal plates to lie within the limits of a single species, Hoplopleura leucomus , n. sp. (see accounts describing the lice).

The variation in pelage traits among samples of squirrels collected over the northern peninsula roughly coincides with peninsular distributions of murid rodents and monkeys. Macaca hecki occurs from the base of the northern peninsula to the Gorontalo region, which coincides with the distribution of the color pattern variation in P. leucomus that has been recognized as occidentalis. Macaca nigrescens and M. nigra have allopatric ranges east of the Gorontalo area and extending to the northeastern tip and onto the offshore islands Pulau Lembeh and Pulah Talise ( Fooden, 1969; Groves, 1980b, 2005); combined ranges of these two macaques are congruent with that of the typical form of Prosciurillus leucomus , which occurs east of the Gorontalo region all the way to the northeast tip of the peninsula and on Pulau Lembeh. The murids, Echiothrix leucurus , Bunomys fratrorum , Taeromys taerae , and Rattus xanthurus are also known only from this eastern segment of the peninsula (from Gorontalo eastward); their phylogenetic counterparts ( Echiothrix centrosa ; Bunomys andrewsi , B. penitus , and B. sp.; Taeromys hamatus ; and Rattus marmosurus ) range to the west of Gorontalo into Sulawesi’s central core, or have been recorded only from the central core and the southeastern and southwestern peninsulae (Musser, MS; Musser and Carleton, 2005).

ECOLOGY: We lack first-hand experience with P. leucomus in the wild and are unaware of any published observations. Judged from its close resemblance in body size and tail proportions to populations of squirrels from the central core of Sulawesi ( P. topapuensis and P. alstoni ) that Musser encountered, we suspect that P. leucomus is an habitué of the canopy, descending to the ground between gaps in the canopy and likely for some foraging. Soft fruits and insects probably comprise the diet.

ECTOPARASITES: The host-specific sucking louse Hoplopleura leucomus , n. sp. (described in a following section), is the only ectoparasite recorded from Prosciurillus leucomus (table 56).

SYMPATRY: The range of P. leucomus on mainland Sulawesi overlaps those of P. murinus , Rubrisciurus rubriventer , and the ground squirrel Hyosciurus ileile (table 6). Both P. leucomus and P. murinus have been collected on Pulau Lembeh off the northeastern coast of the northern peninsula (table 4).

SYNONYMS: Only one synonym is attached to Prosciurillus leucomus , which we discuss below.

Sciurus leucomus occidentalis Meyer, 1898: 2 . LECTOTYPE: SNSD B168 (stuffed museum study skin and a skull; see table 13 for measurements), an adult female collected by Riedel in 1875. TYPE LOCALITY: Gorontalo, 00 ° 319N, 123 ° 039E, near sea level on the southern coastal plain of the northern peninsula, Propinsi Sulawesi Utara, Indonesia (locality 15 in gazetteer and map in figure 11).

To understand why we regard SNSD B168 to be the lectotype and Gorontalo to be the type locality, we must begin with Meyer’s (1896) monograph covering the mammals of Sulawesi where he provided (pp. 25–26) an account of ‘‘ Sciurus leucomus ’’ and described the variation in expression of the white patch on the nape just behind each ear (see appendix 8 where the account is reproduced in the original German). Meyer set the context by noting Anderson’s (1879: 252) observations derived from his visits to the museums in Paris and Leiden where he had examined specimens of ‘‘ Sciurus leucomus .’’ According to Anderson,

The series in the Leyden and Paris Museums from the Celebes prove that the white on the side of the neck is not always present, for one specimen shows it disappearing and in another there is no trace of it, and others lead from the one to the other extreme, but when this neck spot is full developed it forms a great violet-white lappet.

Meyer studied 16 specimens from the

Minahasa region (collected at Main, Manado, Lotta, and Amurang; see our gazetteer), two from Pulau Lembeh (locality 3 in our gazetteer), and two collected at Gorontalo (locality 15 in our gazetteer), on the northern peninsula far to the west of the Minahasa region . In his specimens at hand from the Minahasa district of the northeastern peninsula, the nape patches were expansive and white (white mixed with a few black-tipped or buffy-tipped hairs in some of our specimens), vividly contrasting with the large black ear tufts and brown dorsal fur, a pattern characteristic of leucomus from the northeast. The two specimens from Gorontalo, however, exhibited a different pattern. Both were adults as judged by characteristics of the skull and both had prominent black ear tufts, but one (B168) completely lacked the nape patches, and the other (857) exhibited only traces of the pattern: ‘‘ Auf der anderen Seite zeigt von den 2 Exemplaren von Gorontalo das eine (B 168) keine Spur von Weiss, und es ist nach dem Schädel ein ganz altes Individuum, und das andere (857), ebenfalls adult, hat den Fleck nur sehr schwach entwickelt’’ ( Meyer, 1896: 26) .

Meyer also referred to material at Leiden that had been collected in the Gorontalo region. Specimens from Panibi, Modelido, and Limbotto (see our gazetteer) resembled the two Dresden specimens from Gorontalo in showing only faint nape patches or completely lacking them. On the other hand, Meyer admitted that examples from Tulabello, just east of Gorontalo, and Paguat, far to the west of Gorontalo, had prominent, whitish nape patches.

We see a similar range in variation in color and size of nape patches among our specimens stored in the USNM and AMNH. In USNM samples from the northwestern and western portions of the northern peninsula (Sungai Paleleh, Tolitoli, Labuan Sore, and Bumbarujaba; see the gazetteer), the black ear tufts are prominent but the nape patches are reduced in size and intensity as compared to those on squirrels from east of Gorontalo. The variation consists of individuals with modest white or whitish buff nape patches behind each ear (USNM 218074 from Labuan Sore, for example); specimens expressing diffuse patches, usually grayish or buffy white, barely distinguishable from the neck (USNM 218077, another squirrel from Labuan Sore); or a few specimens with no patches—at least the sides of the neck are indistinguishable from the bright brown fur tipped with buff over the head and body (USNM 200271 from Sungai Paleleh).

The variation in two geographic AMNH samples from west of Gorontalo is also illustrative. Twenty specimens come from Bumbulan (see gazetteer). One individual has large white nape patches that match the color and size of the patches characterizing the samples east of Gorontalo. Seven specimens have large, conspicuous nape patches, but they are dull compared with the bright patches seen in the eastern samples—gray or buffy gray instead of white. The patches are inconspicuous on most of the remaining specimens—small buffy areas behind the ears—and three lack the patches.

Sixteen AMNH specimens are from Matinan and Gunung Ile-Ile (see gazetteer), west of Bumbulan. All have prominent black ear tufts, similar in size to those on squirrels from east of Gorontalo. Six have smaller, inconspicuous buffy nape patches that slightly contrast with the rest of the fur covering the neck. Nape patches are conspicuous on squirrels in the remainder of the sample, but the patches are slightly smaller and duller (gray or buffy white) than those characterizing squirrels occurring east of Gorontalo.

In 1898, Meyer, under an account titled ‘‘ Sciurus leucomus occidentalis n. subsp. ’’ (p. 2), referred to his earlier report ( Meyer, 1896) where he had described the whitish nape patches on the two squirrels from Gorontalo as being weakly developed or not present. In that report he had also speculated that the variation in expression of the nape patch seen on the two Gorontalo specimens might represent an intermediate pattern. At one extreme would be the pattern found in squirrels from the northeastern end of the northern peninsula: large, black ear tufts contrasting with expansive white nape patches. The sample of ‘‘ Sciurus tonkeanus ’’ ( Meyer, 1896: 25) from the east-central arm of central Sulawesi, which lacks both black ear tufts and nape patches (ears and neck are the same color as the head and body; see our account of P. alstoni ) would define the opposite extreme. Large black ear tufts combined with nape patches either reduced in size and intensity or absent would be the intermediate pattern, which is characteristic of squirrels inhabiting the northern peninsula in and west of the Gorontalo area.

Additional specimens collected by Paul and Fritz Sarasin and sent to Meyer for identification supported, in Meyer’s view, his supposition. Meyer’s identifications of the specimens collected by the Sarasins are recorded in his 1898 publication where he focused on contrasting variation in expression of the nape patches among the specimens. Five (three adult males, one adult female, and one juvenile female) from the Minahasa region (three from Tomohon, one from Kottabuna on the border of Minahasa, and one from ‘‘Minahasa’’) exhibited the prominent white nape patches typical of leucomus (‘‘spotted’’ [‘‘gefleckte Form’’] as Meyer described them). Two other squirrels, both males, were collected in the northern peninsula west of Gorontalo and lacked the nape patches (‘‘unspotted’’ [‘‘ungefleckte Form’’]). One of these came from between Kottabongon and Bolang Mongondo at about 250 m, the other from the north side of the Matinan (‘‘Matinang’’) range at about 1000 m. After some discussion, Meyer (1898: 3) proclaimed that the ‘‘unspotted’’ form could be separated as a subspecies only from typical leucomus because of the apparent transition between the two forms: ‘‘Die ungefleckte Form lässt sich nur als Subspecies von leucomus abtrennen, da Uebergänge zwischen beiden vorhanden zu sein scheinen. …’’

Some researchers who consulted only Meyer’s 1898 report where he proposed the taxon occidentalis have assumed that Meyer regarded the two ‘‘unspotted’’ specimens collected by the Sarasins to be comparable to cotypes. But, as was the custom at that time, Meyer did not select a type, did not identify types of any kind, and did not designate a type locality. For him ( Meyer, 1898: 3), the range of occidentalis extended from west of the Minahasa region (which is in the northeastern tip of the northern peninsula, the range of typical leucomus ) to the Buol area (near Tolitoli; see gazetteer) in the northwest (‘‘westlich von der Minahasa bis Buol’’). His view was shaped not only by the two ‘‘unspotted’’ examples collected by the Sarasins and listed in the later 1898 publication, but also by the specimens he discussed in his earlier 1896 report: the two Dresden specimens from Gorontalo and the Leiden material collected around the Gorontalo region and farther west.

Without consulting both of Meyer’s reports, any attempt to identify a type specimen and pinpoint the type locality is frustrating and has yielded different results from researchers. Ellerman (1940: 375), for example, listed occidentalis as being simply from ‘‘West Celebes.’’ In 1954, Laurie and Hill (p. 93) listed the type locality as ‘‘Between Kottabangon and Bolang Mongondo, north-western Celebes,’’ which suggests that the first of the two of the Sarasin’s ‘‘unspotted’’ specimens listed by Meyer in 1898 would be the type. This locality designation was followed by Corbet and Hill (1992: 304). Feiler (1999: 407) in his catalog of types housed at Dresden, however, identified the ‘‘holotype’’ of occidentalis as SNSD B168, the squirrel mentioned by Meyer (1896) as being from Gorontalo. But Feiler, confusingly, listed the type locality as ‘‘ Sulawesi, Nordseite der Matinang-Kette, etwa 1000 m hoch,’’ which is the collection locality of the Sarasin’s second ‘‘unspotted’’ squirrel listed in Meyer’s 1898 publication; Feiler does not record the presence of that specimen in Dresden.

Our attempt to examine all the specimens studied by Meyer has also proved frustrating. Some of the Dresden material Meyer discussed in his 1896 report is no longer in that institution (C. Stefen, in litt., 2008), but the Leiden specimens are still present in the Nationaal Natuurhistorisch Museum (C. Smeenk, in litt., 2008). Three of the five ‘‘white-spotted’’ squirrels collected by the Sarasins from the Minahasa region and identified in Meyer’s 1898 publication are in the collection of the Naturhistorisches Museum Basel (L. Costeur, in litt., 2008; see gazetteer), another is in the British Museum (P. Jenkins, in litt., 2008). We also located only one of the Sarasin’s two ‘‘unspotted’’ specimens, the example from the Matinan range (‘‘nordseite der Matinang-Kette, etwa 1000 m hoch’’), a skin and skull, which is also in the collection at Basel (see gazetteer). We have no idea where the other specimen is, the one collected between Kottabangon and Bolang Mongondo; it is not in Basel, Dresden, Leiden, or London.

SNSD B168, one of the two specimens from Gorontalo that Meyer (1896: 26) referred to by numbers and the one that Feiler (1999) regarded as the holotype, remains at Dresden (the other, 857, is no longer listed in the collection inventory; C. Stefen, in litt., 2008). It is currently identified in the inventory of mammal specimens at that institution, and on the skin and skull tags, as the ‘‘type’’ of occidentalis and the collection locality as ‘‘ Gorontalo.’’ Clara Stefen kindly sent us color images of the skin and skull. The slightly stuffed nearly flat skin is intact except for a long tear in the throat, and most of the tail is missing. In ventral view, three pairs of large teats are clearly evident. On the dorsal surface the black ear tufts are prominent but there is no discernable trace of nape patches behind the ears. The occiput and part of the basioccipital region of the skull are missing, the mandible is intact and all upper and lower teeth are present.

SNSD B168 was identified by Feiler (1999: 407) as the holotype of occidentalis, and the Matinan range as the type locality. The specimen (NMB 1198/9543) from the Matinan range, however, is at Basel where it was cataloged in 1942 and is a male; it cannot define the type locality if the Dresden squirrel is indeed the type. However, SNSD B168, a female, is to be regarded as the lectotype (see below). Since Meyer (1896: 26) had explicitly stated that it came from Gorontalo, the type locality changes accordingly. Sometime between the publication of Feiler’s 1999 catalog of types and the present, the skin and skull of SNSD B168, the supposed holotype of occidentalis in Feiler’s catalog, was correctly linked to its provenance, Gorontalo, in the museum inventory.

The identification as ‘‘holotype’’ assigned to SNSD B168 by Feiler (1999: 407) should have been ‘‘lectotype’’ if we interpret correctly Article 74.5 (‘‘ Lectotype designations before 2000,’’ ICZN, 1999: 82): ‘‘In a lectotype designation made before 2000, either the term ‘‘ lectotype,’’ or an exact translation or equivalent expression (e.g. ‘‘the type’’ [in this case, Feiler’s use of the term ‘‘holotype’’]), must have been used. …’’

In light of the above, the other specimens discussed by Meyer (1896: 26, 1898: 2) now become paralectotypes. These include the squirrel at Basel (NMB 1198/9543) collected on the north slopes of the Matinan range, and the following six specimens from Leiden: (1) specimen ‘‘ i ’’ ( Jentink, 1888: 25), RMNH 39399, a mounted skin still containing the skull, male, from Tulabolo, near Gorontalo, northern Sulawesi, collected on May 10, 1864 by C.B.H. Rosenberg; (2) specimen ‘‘ j ’’ ( Jentink, 1888: 25), RMNH 39400, a mounted skin with skull still inside, female, from Tulabolo, near Gorontalo, northern Sulawesi, collected on May 10, 1864, by C.B.H. Rosenberg; (3) specimen ‘‘ k ’’ ( Jentink, 1888: 25), RMNH 39394, a mounted skin without skull, male, from Paguat, northern Sulawesi, collected between November 5–12 by E.A. Forsten (also a paralectotype of P. leucomus ; see that account); (4) specimens ‘‘ l ’’ (the skin, in Jentink, 1888: 25) and ‘‘ h ’’ (the skull, in Jentink, 1887: 191), RMNH 39401, a mounted skin and cranium, male, from Panibi, near Gorontalo, northern Sulawesi, collected between September 7–15, 1863, by C.B.H. von Rosenberg; (5) specimen ‘‘ m ’’ ( Jentink, 1888: 25), RMNH 39402, a mounted skin still containing the skull, female, from Medelido, near Gorontalo, northern Sulawesi, collected between May 9–20, 1863 by C.B.H. von Rosenberg; (6) specimen ‘‘ p ’’ ( Jentink, 1888: 25), RMNH 39403, mounted skin still containing the skull, sex unknown, from Limboto, northern Sulawesi, collected during January, 1876, by S.C.J.W. van Musschenbroek.

Prosciurillus alstoni ( Anderson, 1879)

Sciurus alstoni Anderson, 1879: 252 , pl. xxi (1878 is imprinted on the title page of the book, but the Corrigenda on the following page proclaims ‘‘The sanguine expectation that this work would have been issued during the past year has led to 1878 appearing on the titlepage instead of 1879, the delay having arisen from circumstances over which the author had no control.’’).

Sciurus tonkeanus Meyer, 1896: 25 , pl. X, fig. 4.

Sciurus sarasinorum Meyer, 1898: 1 ( 1899: 21, pl.V).

Sciurus mowewensis Roux, 1910: 519 .

Sciurus elbertae Schwarz, 1911: 639 .

HOLOTYPE AND TYPE LOCALITY: The holotype of Sciurus alstoni (ZSI 9546) consists of a museum study skin and damaged skull and is an unsexed adult that was ‘‘purchased’’ (see Sclater, 1891: 21; Khajuria et al., 1977: 31). Some measurements are listed in table 14.

The type locality is the Malakosa region (northern portion of the central core of Sulawesi), Kuala Navusu, approximately 00 ° 589S, 120 ° 279E, in the coastal lowlands at 30 m, between Tanjung Pandendelisa (jutting into Teluk Tomini) and the mouth of Sungai Sausu to the southeast (locality 1 in gazetteer and identified on the maps in figures 11 and 17), Indonesia, Propinsi Sulawesi Tengah.

Our identification of the holotype of alstoni as a member of the P. leucomus group and selection of a type locality requires explanation, beginning with Anderson’s (1879: 252) description of Sciurus alstoni , which is short but diagnostic (see also fig. 16):

This beautiful species, in the colouring of the upper parts and tail, closely resembles S. lokriah

[5 Dremomys lokriah , endemic to Central

Nepal, Assam, Burma and western Yunnan;

Corbet and Hill, 1992: 298], whilst the under parts differ in being dusky chestnut instead of orange. The peculiarity of the species is the beautiful pure white tufting to the ears, which projects a considerable way backwards, in a pointed manner. The external surface of the tip of the ear is covered with short brown hairs which stand out against the white. The subapical brown, or rather black band of the hairs of the tail, is broad and rather deeply edged with whitish; the tip of the tail is blackish, and the remainder more or less obscurely tinged with black and orange.

The incisors are pale yellow, and narrow; the facial portion of the skull is rather short and moderately pointed, and the nasals are rather broad posteriorly.

Anderson provided measurements for ‘‘Length, muzzle to tail [our Length of Head and Body],’’ as 7.15 in. (5 181.6 mm); ‘‘Length of tail without hair [our Length of Tail],’’ 6.50 in. (5 165.1 mm); and ‘‘Length of tail with hair,’’ 8.30 in. (5 210 mm). We have no comparable standard measurement for length of tail including the terminal tuft, so we measured the tail tufts of seven skins from Sungai Tolewonu and 19 skins from Kuala Navusu and added those values to lengths of tails (5 tail vertebrae) for those specimens, obtaining the range 209–235 mm for the Tolewonu series, and 198–236 mm for the Navusu sample. As to the provenance of the specimen, Anderson wrote: ‘‘The locality from whence this species was obtained has not been accurately ascertained, but it is probably Borneo.’’

Anderson’s alstoni , with a few exceptions, disappeared from published checklists of squirrel species ( Agrawal and Chakraborty, 1979, for example; Thorington and Hoffmann, 2005, who did list an alstoni , proposed by J.J. Allen in 1889, but that alstoni is a synonym of Sciurus nayaritensis , a fox squirrel occurring in northern Mexico and the southwestern United States; see p. 762 in that publication), checklists of rodents focusing on those native to the Indomalayan region ( Chasen, 1940, for example), and publications devoted to mammals of a particular geographic region in the Indomalayan area, especially Borneo and the Malay Peninsula ( Moore and Tate, 1965; Medway, 1969, 1977; Payne et al., 1985; Corbet and Hill, 1992). An early exception is Jentink’s (1883: 118) ‘‘List of the specimens of squirrels in the Leyden Museum’’ where he recorded that Sciurus alstoni was ‘‘Characterized by the beautiful pure white tufting to the ears, which projects a considerable way backwards.’’ Jentink cited Anderson’s description and noted that ‘‘no specimens were in the Leyden Museum.’’ Another is an entry for Sciurus alstoni in Sclater’s (1891: 21) catalog of mammals in the Indian Museum where he noted that ‘‘The type and only specimen known is said to have come from Borneo,’’ indicating that a skin was present in the collection and that the specimen had been ‘‘purchased.’’ In Ellerman’s (1940: 376) compendium on the families and genera of living rodents, he listed alstoni as a species of Callosciurus in a category ‘‘ incertae sedis; not allocated to groups.’’ Another exception is Khajuria et al. (1977: 31) who provided ‘‘An annotated catalogue of the type specimens of mammals in the collections of the Zoological Survey of India [formerly the Indian Museum],’’ and entered information on the holotype of Sciurus alstoni . While Anderson (1879), Sclater (1891), and Ellerman (1940) noted the type locality to probably be ‘‘Borneo,’’ Khajuria et al. (1977: 31) narrowed it to ‘‘ Indonesia, Kalimantan’’ but gave no reasons for doing so.

The most recent and only insightful allocation of Anderson’s alstoni was made by Corbet and Hill (1992: 304) in their systematic review of the mammals of the Indomalayan region. They listed the combination ‘‘? Sciurus alstoni ’’ among the scientific names associated with Prosciurillus leucomus , the provenance as ‘‘locality unknown,’’ and repeated Anderson’s ‘‘probably Borneo.’’

We have not personally studied the holotype of Anderson’s Sciurus alstoni . We contacted Dr. Sujit Chakraborty at the headquarters of the Zoological Survey of India who informed us that the holotype is present in the collection and kindly provided us with values for some cranial measurements. The measurements, along with Anderson’s description, measurements of body and tail, and the superb color plate depicting the squirrel in live pose (reproduced in fig. 16) do not describe any species of squirrel, no matter the genus, that is native to Borneo, as anyone can determine by simply thumbing through the color plates of squirrels (pls. 23–30) and relevant pages of descriptive text (233–243) in Payne et al.’s (1985) field guide to the mammals of Borneo. We know of no other species of tropical Indomalayan squirrel occurring outside of Borneo and west of Sulawesi, or in the Philippines that bears close resemblance to ‘‘ Sciurus alstoni ’’ (see the color plates depicting species of Indomalayan squirrels from the Malay Peninsula [ Medway, 1969], Thailand [ Askins, 1988], and ‘‘Southeast Asia’’ [ Francis, 2008]; the statement is also based on Musser’s study of Indomalayan squirrels stored in North American, Asian, and European museums).

The animal described by Anderson and depicted in his color plate is a member of the Prosciurillus leucomus group and an unambiguous representative of the same species from which the samples collected by Fritz and Paul Sarasin and described as sarasinorum ( Meyer, 1898) and mowewensis ( Roux, 1910) were obtained (see section discussing synonyms). Specimens from Kuala Navusu and Sungai Tolewonu collected by Musser (in AMNH), the sample from Pinedapa obtained by H.C. Raven (in USNM), and the squirrels collected from the southeastern peninsula by G. Heinrich (in AMNH) represent the same species. Values for length of head and body (181.6 mm) and length of tail (165.1 mm) for the holotype of alstoni fall within the range of variation listed in table 15 for these variables in the sample from Kuala Navusu and Sungai Tolewonu; length of tail relative to head and body length (91%) is also comparable, as is length of tail including the tuft (see above). The highlights of Anderson’s description—dusky chestnut underparts, prominent white ear tufts, subapical blackish bands of the tail hairs that are tipped with white, blackish tail tip and the rest of the tail tinged with black and orange—match Musser’s material laid out in front of us from Kuala Navusu and Sungai Tolewonu as well as the two examples of sarasinorum shown in Meyer’s (1899) color plate (see fig. 10). Anderson’s color plate of alstoni (see fig. 16), except for the underparts that appear faded, rendered orange rather than ‘‘dusky chestnut,’’ exhibits the same diagnostic pelage highlights described by him.

Where the holotype of alstoni was collect- ed in Sulawesi is unknown. The specimen was purchased, according to Sclater (1891: 21) and Khajuria et al. (1977: 31), probably during Anderson’s tenure as Superintendent of the Indian Museum, Calcutta (now the Zoological Survey of India, Kolkata), certainly before 1879. It had to have been collected within the range of the Prosciurillus exhibiting the alstoni pelage color pattern. Our voucher specimens characterized by that coloration and patterning come from lowlands and middle altitudes (31–1200 m) in the eastern portion of the central core of Sulawesi and throughout the mainland of the southeastern peninsula (see gazetteer and fig. 11). Forests of lowlands and mountains in the western portion of the central core are occupied by P. topapuensis , the distinctive P. weberi is found in lowlands at the southern sector of the central core west of the range of alstoni , and forests on the northern peninsula are inhabited by P. leucomus (see gazetteers and fig. 11).

There are two early travel routes in this range where the holotype of alstoni could have been collected. One extended from the Luwu region at the north end of Teluk Bone north to Mapane at the southern end of Teluk Tomini. Paul and Fritz Sarasin traveled this route in the late 1800s where they obtained a squirrel at Usu on the southern end and a specimen from Mapane at the northern end. The two specimens became the basis for Meyer’s (1898) description of Sciurus sarasinorum . The other route transects the southeastern peninsula from Kolaka at Teluk Mekongga in the west to Kendari at the margin of Teluk Kendari in the east. The Sarasins traveled here in the early 1900s where they obtained the two specimens later described by Roux (1910) as Sciurus mowewensis . Earlier European travelers may have taken the same routes and possibly acquired Anderson’s specimen along the way but we just do not know. The animal could even have been brought to Borneo by traders.

In the absence of any information about a collection locality for the holotype of alstoni , we turned to the International Code of Zoological Nomenclature (4th ed., 1999) for insight into selecting a type locality. One of the recommendations listed in that document (76A.1.4.) is pertinent here: ‘‘In ascertaining or clarifying a type locality, an author should take into account ‘‘as a last resort, and without prejudice to other clarification, localities within the known range of the taxon or from which specimens referred to the taxon had been taken.’’

Within what is the known range of Prosciurillus alstoni defined by voucher specimens, we select Kuala Navusu (approximately 00 ° 589S, 120 ° 279E) as the type locality for the taxon. This is one of Musser’s camps where during August–November 1975, he obtained 26 specimens of P. alstoni in lowland tropical evergreen rain forest between 30 and 229 m (see gazetteer). The campsite and surrounding region where Musser surveyed was located between the village of Malakosa (behind Tanjung Pandendelisa jutting into Teluk Tomini) and the mouth of Sungai Sausu (fig. 17). Prosciurillus alstoni was common in the forest there and also south of Kuala Navusu at Musser’s camp on the Sungai Tolewonu (approximately 01 ° 049S, 120 ° 279E; fig. 17), one of the tributaries of Sungai Sausu, where during January and February 1976 he collected 16 specimens between 122 and 366 m (see gazetteer). The specimens from both places match Anderson’s (1879) description of ‘‘ Sciurus alstoni ’’ in all aspects of body dimensions and pelage coloration.

EMENDED DIAGNOSIS: Prosciurillus alstoni shares with other members of the P. leucomus group moderate body size and a tail equal to or shorter than length of the head and body (tables 2, 14, 15). It contrasts with the other four species in the group by the following combination of pelage traits (see table 12): (1) dorsomedial surface of each ear covered with long white hairs projecting beyond ear rim to form a prominent tuft on most specimens, ears without tuft and concolorous with head and neck in a few individuals; (2) no bright ochraceous hairs lining inside of the pinnae, which is same color as fur on neck and head; (3) no nape patches behind the ears; (4) no middorsal black stripe extending along back from neck to base of tail; (5) dark reddish brown (approaching chestnut) underparts; and (6) a geographic range restrict- ed to tropical lowland evergreen rainforest habitats on the eastern portion of Sulawesi’s central core, the east-central peninsula, the southeastern peninsula, and islands offshore of that arm.

GEOGRAPHIC AND ALTITUDINAL DISTRI- BUTIONS: Collection localities of P. alstoni are from the eastern portion of Sulawesi’s central core, the southern coastal plain on the east-central peninsula, the southeastern peninsula, and the islands of Buton and Kabaena (see gazetteer and map in figure 11). Collection sites range in elevation from 30 m (Kuala Navusu and Pinedapa) to 1200 m (Sungai Tolewonu), all in lowland tropical evergreen rainforest habitats (fig. 17). The distribution of collection sites indicates that P. alstoni is not sympatric with any other member of the P. leucomus group. The most northern record of P. alstoni is Kuala Navusu (00 ° 589S, 120 ° 279E). Approximately 100 km north of there is Bumbarujabu (00 ° 439S, 120 ° 049E) at the base of the northern peninsula and the southernmost collection locality for P. leucomus (see gazetteer and fig. 11). Between those two places along the eastern coast, assuming the presence of intact forest, the ranges of P. leucomus and P. alstoni either overlap (likely narrowly) or are parapatric. The pattern of contact is unknown because no members of the P. leucomus group have been collected along that coastal strip.

The mountains south of Bumbarujabu and west of Danau Poso in the western half of Sulawesi’s central core—west of the range of P. alstoni —are inhabited by P. topapuensis , another member of the P. leucomus group. The boundaries of each range in that vast region of central Sulawesi between Teluk Tomini in the north and Teluk Bone in the south are unknown because no voucher specimens are available from there (see map in fig. 11). Members of the P. leucomus group have been collected only in the coastal lowlands bordering Teluk Tomini and lowlands around Teluk Bone. In the northern lowlands, samples of P. alstoni are from Mapane (01 ° 269S, 120 ° 409E) on the coast and inland a bit at Pinedapa (01 ° 259S, 120 ° 359E), 31 m. Westward toward the mountains, approximately 20 km from Pinedapa, is Rano Rano (01 ° 309S, 120 ° 289E), 1829 m, the easternmost collection site for P. topapuensis (see gazetteer and fig. 11). In this region, and possibly farther south along flanks of the mountains, P. topapuensis may occur only in montane forest formations (although it descends much lower, down to 350 m, in the western part of its range) and P. alstoni in lowland evergreen rain forests; only results from future surveys will provide a description of the boundary between these two species.

Southern lowlands fringing Teluk Bone have yielded two species in the P. leucomus group. Prosciurillus alstoni has been collected at the base of the southeastern peninsula at Usu (02 ° 369S, 121 ° 069E), and Masamba (02 ° 349S, 120 ° 199E), about 80 km west of Usu, is the easternmost collection site for P. weberi . Whether the two species overlap or their ranges abut in this lowland interval is unknown.

Prosciurillus alstoni may be restricted to habitats in tropical lowland evergreen rain forests. While P. leucomus , its northern neighbor, ranges from coastal lowlands to habitats in montane forest (coast to 1700 m), and P. topapuensis , its western relative, extends from tropical lowland forests at 350 m to upper montane formations at 2800 m, P. alstoni has yet to be collected above 1200 m (table 4). Montane areas within its range, however, have received little or no survey focusing on collections of small mammals with the exception of Pegunungan Mekongga in the southern portion of the southeastern peninsula. Samples of P. alstoni come from Wawo (03 ° 419S, 121 ° 029E) 50 m, on the coast west of that mountain range and Masembo (03 ° 359S, 121 ° 159E) on the lower flanks of the range at 550 m, all collected by G. Heinrich in the early 1930s. At higher elevations in lower and upper montane rainforest formations, Heinrich encountered only Rubrisciurus rubriventer (1400 m) and Prosciurillus abstrusus (1500 and 2000 m). He was an able naturalist and skilled hunter, daily walking the forests shooting birds and squirrels for various museums, and likely would have collected examples of P. alstoni if it occurred at elevations higher than 550 m.

Much remains to be learned about the actual geographic range of P. alstoni . Its distribution on the east-central peninsula, for example, is a mystery; specimens are from only a single locality on the southern coast, but the species probably occurs throughout the lowlands fringing the mountainous backbone of the peninsula. Whether P. alstoni or some other member occurs in the mountains is unknown; it is possible a member of the P. leucomus group is absent from high altitudes on the peninsula, a pattern that would resemble that seen on Pegunungan Mekongga. The nature of the contact zones between the range of P. alstoni and those of P. leucomus to the north and P. topapuensis and P. weberi to the west are enigmas. Surveys in the east-central peninsula and through relevant regions in the central core of Sulawesi are needed to refine the geographic distribution of P. alstoni and its distributional pattern relative to the other three species.

A segment of P. alstoni ’s range matches that of the macaque, Macaca ochreata . That monkey has been recorded from the islands of Buton and Muna, and from the mainland of the southeastern peninsula north to the lakes region (Danau Towuti, Danau Mahatona, and Danau Matana). In landscapes north of the lakes in the east-central peninsula and the central core of Sulawesi, M. ochreata is replaced by M. tonkeana ( Fooden, 1969; Groves, 2001, 2005). Prosciurillus alstoni also occurs on the islands and mainland south of the lakes region, but as opposed to Macaca ochreata , it ranges farther north into the east-central limb and to the northeastern portion of the central core.

DESCRIPTION: Prosciurillus alstoni is similar to P. leucomus in body size, with the tail averaging as long as, or only slightly shorter than, the length of head and body as in that species (length of head and body, 157– 195 mm; length of hind foot, 40–48 mm; length of ear, 15–19 mm; weight, 135–210 g; see table 15). Thickness and color of the fur covering upperparts of head and body are also similar; both possess a uniformly thick (12–15 mm) coat that is dark brown flecked with black and orange along midline of the body from forehead to base of tail, but grading into brownish gray spotted with black and pale buff along sides of the head, body, forelegs, and hind legs. Dorsal surfaces of the front and hind feet may be the same color as the head and back, slightly darker, or slighter paler (buffy gray) in both species. Coloration of the body fur results from the mixture of dark gray curly underfur; longer overfur composed of black hairs, each blackish gray interrupted by a subterminal orange or pale buffy band and black tip; and black guard hairs barely projecting beyond the overhair layer. Cheeks range from buffy to ochraceous.

The dorsal background forming the upperparts of P. alstoni is interrupted by color patterning on the head and ears. Top and sides of the muzzle are ochraceous, a pattern also seen in P. leucomus . Unlike that species, however, P. alstoni lacks any sign of nape patches behind the ears—fur over the neck is indistinguishable in color from that covering the rest of the upperparts. Prosciurillus alstoni also lacks the prominent black ear tufts and ochraceous inner ear lining characteristic of P. leucomus . Instead, the hairs covering most of each ear on P. alstoni are similar in coloration to the head and body, a small black tuft sits at the inner dorsal margin of the ear, and from the medial surface springs a conspicuous white tuft that projects 5–8 mm beyond the dorsal curvature of the ear. The visual pattern evolves around an ear with a dab of black on the inner surface and rimmed by a white crest (see the color plates in figs. 10, 16). The white tufting is variable in expression both within and between samples; it is whitish buff in some specimens and absent from others. This variation is more fully explored in the sections discussing geographic variation and synonyms.

The short (5–7 mm thick) ventral coat of P. alstoni is deep, dark red—bordering on reddish brown or chestnut—on most specimens; some individuals from the southeastern peninsula have reddish orange or orange-red venters. The hairs forming the ventral covering are dark gray along their basal half and deeply pigmented with red or reddish orange along the distal half.

The tail of P. alstoni is equal to or slightly shorter than length of head and body (LT/ LHB 5 89%–99%, see table 15) and is covered in long hairs, each patterned by alternating black and buffy or orange bands. The overall effect is rings of black and buff with buffy bands outlining margins of the tail and long black hairs forming a black terminal tuft. The ventral surface of the tail is reddish (paralleling the venter coloration) bordered by black and buffy margins. Overall coloration of the tail is closely similar to that seen in P. leucomus .

Females have three pairs of teats, one postaxillary pair, and two inguinal pairs. A single embryo was found in each of the few pregnant squirrels examined.

Views of the skull are illustrated in figures 12–14; cranial and dental measurements are summarized in tables 16 and 19.

COMPARISONS: Prosciurillus alstoni requires comparisons with P. leucomus , endemic to the northern peninsula north of the distribution of P. alstoni ; P. weberi , known only by specimens from the lowlands fringing Teluk Bone west of the range of P. alstoni ; and P. topapuensis , inhabiting the mountainous western portion of the central core to the west of P. alstoni .

Prosciurillus alstoni and P. leucomus : Both species are similar in body size and length of tail relative to length of head and body (table 15), in coloration of the fur covering head and body, legs, dorsal surfaces of front and hind feet, and color and patterning over the tail. Both share an ochraceous muzzle. Both show bright underparts but those of P. alstoni are typically more deeply pigmented, dark reddish brown as opposed to orange-red on average in P. leucomus (reddish orange to ochraceous is the chromatic range). Other pelage dissimilarities are striking (table 12): P. alstoni lacks any indication of nape patches behind the ears, a pattern characteristic of most examples of P. leucomus ; and the ears are either adorned with a prominent white tuft or lack such a tuft in which case the ears, including the inner surfaces, are the same color as the head and neck—all specimens of P. leucomus exhibit prominent black ear tufts and bright ochraceous inner ear surfaces beneath the tufts.

The two species are not so different in cranial and dental dimensions (table 17) but significant proportional differences exist, which are reflected in the ordination of specimen scores projected onto first and second principal components extracted from principal-components analysis shown in figure 18. No marked separation among the scores into two discernable clusters exists along the first axis, where the spread is influenced by covariation in all of the dimensions measured except mastoid breadth and height of braincase; breadth of rostrum and lengths of nasals, rostrum, and diastema have somewhat more impact (table 20).

The points separate into two overlapping clusters along the second component, reflecting the relatively slightly smaller skull, shorter tooth rows, but wider rostrum and longer diastema in P. alstoni compared to P. leucomus . These distinctions, while not immediately evident when comparing the descriptive statistics for each dimension in the two samples (tables 17, 19), can be appreciated visually by comparing side-to-side adult skulls with a comparable degree of wear on the teeth. Prosciurillus alstoni generally has a smaller skull and shorter tooth rows than P. leucomus .

TABLE 19

Descriptive Statistics for Cranial and Dental Measurements (mm) Derived from Separate Population Samples of Prosciurillus leucomus and Prosciurillus alstoni

Mean ± 1 SD and observed range (in parentheses) are listed.

Although the morphometric distinctions between the two species are subtle, the contrasts in color pattern are not. While not every example of P. alstoni has white ear tufts, none exhibits black ear tufts, and not one of the specimens we examined has nape patches behind the ears. These traits, along with their average darker red underparts and proportionately smaller skull and tooth rows, set P. alstoni apart from P. leucomus .

Prosciurillus alstoni and P. weberi : Our sample of P. weberi consists of eight individuals collected at two localities, with only five yielding a full set of cranial and dental measurements. If the two collection sites, in lowlands (100 m or less) fringing the northwestern margin of Teluk Bone, are representative of the altitude of P. weberi —namely tropical lowland evergreen rainforest habitats—then both P. weberi and P. alstoni share this lowland habitat affinity. The northern peninsular P. leucomus occurs through a range of forest formations from the coastal lowlands to montane habitats, and the west-central P. topapuensis inhabits mountain forests but in places ranges into lowlands.

Measurements of head and body, tail, hind feet, and ear from specimens of P. weberi are scanty but do indicate a close resemblance to P. alstoni , and other members of the P. leucomus group, in body size (tables 15, 16, and 17). Prosciurillus alstoni also closely resembles P. weberi in coloration of fur covering upperparts and underparts, as well as color and patterning on the tail. The striking differences between the two species involve markings on the back and ears. Although the background color of fur over the back is similar in both, that of P. weberi is broken by a ‘‘broad black band along the spine of the back, running from the neck, increasing in broadness in the middle of the back and diminishing towards the root of the tail,’’ as Jentink (1890: 116) described it. This dorsal pattern marks every specimen of P. weberi we studied but is absent from all examples of P. alstoni examined. Black tufts adorn the ears of P. weberi but comparable tufts in P. alstoni are white or not present (table 12). Jentink (1890: 116) noted that in one of the specimens of P. weberi he examined, ‘‘the black earpencils are slightly tipped with white.’’

Our available samples from which we derived morphometric data for analyses are unsatisfactorily uneven: 63 skulls of P. alstoni but only five from P. weberi . Results we present here must be tested by employing a larger sample of P. weberi . Mean values for most cranial dimensions in the sample of P. weberi are smaller than comparable dimensions in the larger sample of P. alstoni (table 17). These distinctions are transformed visually in the ordination showing specimen scores projected onto the first and second principal components extracted from principal-components analysis in figure 19. There the large cloud of scores representing P. alstoni overlaps the smaller group of points for P. weberi . Covariation in all cranial and dental dimensions except height of braincase contributed to the spread of scores along the first principal component (table 21). The center of the larger constellation of points lies to the right of the smaller cluster, reflecting the overall average greater cranial and dental dimensions of P. alstoni compared to P. weberi .

The scatter along the second principal component results in separation of the specimen scores for P. weberi from the larger cloud formed by the sample of P. alstoni , and is a measure of distinctions in cranial and dental proportions. Highlight- ed is the relatively narrower skull and interorbital region of P. weberi compared with P. alstoni , its relatively shorter nasals and rostrum, lower braincase, smaller bullae, and shorter tooth row, but wider rostrum and longer diastema and postpalatal

TABLE 20 Results of Principal-Components Analysis Contrasting Population Samples of Prosciurillus alstoni with Those of Prosciurillus leucomus Principal components are extracted from a covariance matrix of log-transformed values for 15 cranial and 1 dental variable; see figure 18.

region (table 21). Mean skull length is not so different in the two species (table 17), but many of the internal dimensions are less for P. weberi relative to those dimensions in skulls of P. alstoni , especially interorbital breadth and toothrow length, and three internal dimensions are of relatively greater magnitude.

Prosciurillus alstoni and P. topapuensis : These two species are dissimilar in altitudinal distributions, coat color and patterning, and cranial proportions. Prosciurillus alstoni inhabits tropical lowland evergreen rain forests and has not been collected above 1200 m within its range, which is east of the distribution for P. topapuensis (see map in fig. 11). That species is endemic to the western mountains and foothills of central Sulawesi and ranges through a variety of old-growth forest habitats, from tropical lowland evergreen rain forest to upper montane rain forest (table 4).

The species resemble each other in physical size with the exception that across geographic samples, P. topapuensis typically has a tail that is absolutely shorter and shorter relative

TABLE 21 Results of Principal-Components Analysis Contrasting Population Samples of Prosciurillus alstoni with the Sample of Prosciurillus weberi Principal components are extracted from a covariance matrix of log-transformed values for 15 cranial and 1 dental variable; see figure 19.

to head and body length, and larger ears (tables 15, 16). Most P. topapuensis have thicker fur, especially samples from high altitudes; the dorsal coat of specimens from low altitudes is inseparable in color from upperparts of P. alstoni , but individuals from higher altitudes show a duller fur: the colored bands are mostly buffy instead of orange and buff. Tails of the two are similar in coloration and pattern of hair bands.

The striking dissimilarities between the two species involve the underparts and ears. All examples of P. topapuensis we examined have dark gray underparts lightly washed with buffy or ochraceous hues, or tinted silver, which contrasts strikingly with the dark reddish to reddish brown venters so characteristic of P. alstoni . Black tufts mark the ears on P. topapuensis . These are not as prominent as seen in some samples of the northern peninsular P. leucomus , and the variation in expression ranges from conspicuous tufts to just a trace of black so the ears are covered with the same color fur as the head and neck. No example of P. topapuensis has white ear tufts, which is usual in P. alstoni (table 12).

Skulls of the two species are similar in overall size and in many internal cranial dimensions but differ in dimensional proportions. The correspondence in size is indicated by the mean values listed in table 17 and reflected in the graph of specimen scores projected onto first and second principal components extracted from principal-components analysis shown in figure 20. No separation of scores into discrete clusters is evident along the first component where all but two variables load positively (height of braincase and length of tooth row are the exceptions) and contribute to the the spread of points. Fewer variables load strongly along the second principal component but they point to significant differences in shape between the two species (table 22). Compared with P. topapuensis , the skull of P. alstoni is relatively wider (as indexed by zygomatic, interorbital, and mastoid breadths) and the orbit and ectotympanic bullae relatively longer; the nasals, diastema, and tooth row are relatively shorter, and the rostrum narrower.

GEOGRAPHIC VARIATION: Specimens of P. alstoni generally look alike no matter where they were obtained throughout the geographic range of the species as indicated by our samples at hand (tables 14, 19). The notable variation we see among adults involves the ear region. Ears of the nine specimens in the type series of tonkeanus from Tonkean on the east-central peninsula of Sulawesi (see gazetteer and fig. 11) have only short buffy hairs fringing their margins (clearly rendered in Meyer’s, 1896, plate [reproduced in fig. 9] and evident in the color images of the type series given to us by Clara Stefen) and lack the characteristic white ear tufts seen in samples elsewhere within the range of P. alstoni ; backs of the ears are covered with dark buffy hairs, which reach the rim of the pinna but do not project beyond it to form a tuft in most specimens, and form a short tuft in others. Otherwise, color of the upperparts, underparts, and tail of individuals in the Tonkean series match our samples from the central core of the island and the southeastern peninsula. No other samples of the squirrel have been collected from the east-central peninsula so we do not know whether the absence of white tufts is fixed in the population there, is individually variable at other localities, or if presence or absence of the white tufts shows a geographic pattern over the peninsula.

Our samples of P. alstoni elsewhere within its range exhibit variation in frequency of ear tufts and slight variation in color. In the central core of the range, three of the 24 skins from Kuala Navusu have short, inconspicuous whitish buff or buffy white tufts. Backs of the ears are covered with buffy white hairs that project 1–2 mm beyond the ear rims. The remaining specimens support a white coat over the medial surface of each ear, and the white hairs project from 3 mm to 8 mm beyond the ear rim. Individuals with the longer tufts match Anderson’s color plate of alstoni (fig. 16) and Meyer’s portrait of sarasinorum (fig. 10). One of the 12 skins from Sungai Tolewonu exhibits an inconspicuous, short buffy white tuft—nearly a fringe—but the others sport white tufts, the range in length comparable to that of the sample from Kuala Navusu. Of the 23 specimens from the southeastern peninsula (Wawo, Masembo, and Lalolei), one lacks tufts, back of the ears are dark buff with the hairs forming a low buffy fringe on the rim of the pinna (closely similar to the type series of tonkeanus ); backs of the ears are buffy in seven individuals, with the hairs projecting 2 mm beyond to form an inconspicuous, short buffy tuft or high fringe; 18 specimens have prominent white tufts, the range in length comparable to that of the samples from Kuala Navusu and Sungai Tolewonu.

Two adult skins and skulls (MZB 6250 and 6251) from Pulau Buton are insular examples of P. alstoni (if Buton is the actual provenance) and were discussed by Sody (1949: 107) under ‘‘ Callosciurus tonkeanus .’’ Musser examined and compared the pair with AMNH specimens of P. alstoni collected at several places (Wawo, Masambo, and Lalolei) on the southeastern peninsula. In size and configuration of the skull, body size and proportions of appendages, and general coloration of fur covering body and tail, the two squirrels from Buton are similar to samples of P. alstoni from the mainland of the southeastern peninsula, and differ only in expression of the ear tufts. Instead of showing the tall and conspicuous tufting seen in most P. alstoni , the two individuals from Buton have less prominent tufts and they are white tinged with buff, matching some specimens from the central core and southeastern peninsular mainland described above. AMNH 101327, for example, from Lalolei, on the mainland, has short, whitish buff ear tufts; in this aspect and overall pelage color and patterning, the specimen is indistinguishable from the two Buton squirrels—although collected from different places at different times, all three squirrels could have been part of the same litter.

Multivariate analyses of cranial and dental variables among our geographic samples (identified in table 1) of P. alstoni drawn from the eastern section of Sulawesi’s central core and the southeastern peninsula do not identify geographic components of the variation in these variables. The scatter of specimen scores projected onto first and second principal components extracted from principal-components analysis shown in figure 21 forms one large cloud in which are intermingled the points representing squirrels in the three geographic population samples. In addition to not revealing significant internal structure that correlates with geo-

TABLE 22 Results of Principal-Components Analysis Contrasting Population Samples of Prosciurillus alstoni with Those of Prosciurillus topapuensis Principal components are extracted from a covariance matrix of log-transformed values for 15 cranial and 1 dental variable; see figure 20.

graphic origin of the samples, the ordination also expresses individual variation within an age class, and variation associated with age since the range from young adults to old adults comprise each sample. Covariation in most cranial variables contributed to the spread of scores along the first axis (size), fewer load heavily on the second component (shape; table 23). Missing from this analysis is the sample from Tonkean, which consists of skins only.

ECOLOGY: Prosciurillus alstoni inhabits the upper canopy of primary tropical lowland evergreen rain forests, traveling through this layer along the pathways formed by the interlaced branches and woody vines. At gaps in the canopy squirrels descend to the ground where they bound along tree trunks and limbs lying on the forest floor and cross streams and rivers on bridges formed by branches of live trees arching over the water as well as rotting tree and palm trunks and tree limbs connecting opposite stream terraces. Most of the specimens Musser collected at the Kuala Navusu and Sungai Tolewonu areas were trapped on limbs of live trees growing over water or on rotting trunks and treefalls bridging streams (see summary of habitats at trapping sites in table 24, and photographs of forest in figs. 8, 22, and 23).

While trapping records indicate P. alstoni to be common in the region where Musser worked, he rarely saw the squirrels; they were quiet and wary compared with the P. topapuensis (see that account) he encountered. Occasionally squirrels were seen traveling through the upper canopy and sometimes dashing along tops of rotting treefalls on the ground in forested ridgetops, hillsides, and stream terraces. Calls heard were similar to those vocalizations of P. topapuensis . So too were the factors eliciting alarm calls: approaching rain, human sounds, and diurnal avian predators.

Contents of somachs indicate that soft fruits and insects comprise the diet (table 24). Remains of figs were the most common fruit found in stomachs. In addition to remains of fruit, several stomachs contained a variety of insects, most of which would be encountered in tree crowns (table 57). Macrolepidopteran larvae (caterpillars) feed on foliage and, judged by the large numbers in stomachs, were actively gleaned from leaf surfaces by the squirrels. Larvae of buprestid beetles live beneath the bark of standing trees and are dug out. Along with fruit, arboreal termitid termites were in two stomachs, one with about two dozen workers and soldiers, but the other packed with these insects. The squirrels had located and scratched open a nest adhering to a limb in the tree crown either before or after eating fruit. Adult beetles and beetle larvae other than buprestids, as well as the occasional ant are also picked off surfaces of leaves and bark. The few ant pupae found in stomachs were likely dug out of a nest.

ECTOPARASITES: The only record is the sucking louse Hoplopleura alstoni , n. sp. (table 56), that we describe in a following section; this species of louse is unique to its host.

SYMPATRY: The range of P. alstoni on the central core and southeastern peninsula of mainland Sulawesi overlaps distributions of P. murinus , Rubrisciurus rubriventer , and the ground squirrel, Hyosciurus ileile (table 6). In the lowlands behind Kuala Navusu and Sungai Tolewonu at the northern section of the central core of Sulawesi, Musser trapped P. alstoni in the same traplines where he encountered the other three species, and sometimes in the same traps (table 24).

SYNONYMS: Four scientific names represent samples of P. alstoni . Information covering their type specimens and reasons behind their allocations is summarized below.

Sciurus tonkeanus Meyer, 1896: 25 , plate X, figure 4. LECTOTYPE: SNSD B3178 (skin mounted in a live pose, without a skull), an unsexed adult purchased from Charles Cursham in 1889. TYPE LOCALITY: Indonesia, Propinsi Sulawesi Tengah, east-central peninsula of Sulawesi, Tonkean (near Sinorang), 01 ° 249S, 122 ° 309E, on the coastal plain, 0–100 m (locality 5 in gazetteer and on map in figure 11).

Feiler (1999: 407) identified SNSD B3178 as the ‘‘holotype’’ and eight additional specimens as ‘‘ paratypes.’’ Meyer (1896: 25) had before him an unspecified number of flat skins without skulls and postcranial skeletons (‘‘platte Felle ohne Knochen’’) when he named and diagnosed tonkeanus and did not designate a holotype, the usual practice at the time. Feiler’s indication therefore has made SNSD B3178 the lectotype of Sciurus tonkeanus Meyer, 1896 , the other eight specimens being paralectotypes, designations in agreement with Article 74.5 in the Code ( ICZN, 1999: 82–83): ‘‘In a lectotype designation made before 2000, either the term ‘ lectotype,’ or an exact translation or equiv-

TABLE 23 Results of Principal-Components Analysis Contrasting Population Samples of Prosciurillus alstoni Principal components are extracted from a covariance matrix of log-transformed values for 15 cranial and 1 dental variable; see figure 21.

alent expression (e.g. ‘the type’), must have been used or the author must have unambiguously selected a particular syntype to act as the unique name-bearing type of the taxon. …’’ By using the term ‘‘holotype’’ (an ‘‘equivalent expression (e.g. ‘the type’),’’ Feiler has ‘‘unambiguously selected a particular syntype to act as the unique name-bearing type of the taxon.’’

The eight paralectotypes (SNSD 3179–86) are unsexed and without skulls. All are unstuffed pelts that ‘‘seem to have dried in the position they just happened to be in and were not prepared ‘properly’ … the ears are all flat or broken and the tails bent or lost.’’ (C. Stefen, in litt., 2008).

Clara Stefen kindly sent us color images of the lectotype and paralectotypes, and based upon them, along with Meyer’s (1896) description and color plate (fig. 9), we can write that tonkeanus closely resembles the holotypes of sarasinorum (fig. 10) and alstoni (fig. 16) in coloration of the fur covering body and tail, differing only in color of the ears and expression of tufts (see appendix 3 for Meyer’s description in the original German). The white covering on the medial surface of each ear that extends beyond the margin of the pinna to form a conspicuous white tuft that is so characteristic of most examples of P. alstoni is absent from the sample of tonkeanus . Instead, outer and inner surfaces of the ears are dark brown or buffy, and these hairs project slightly beyond the margin of the pinna to form a short and inconspicuous fringe in most specimens, and a short tuft in others. We don’t know if color and size of the ear fringe seen in the sample from Tonkean are characteristic of populations inhabiting all forests on the east-central peninsula, for no other samples have been collected in that region. We do know that the tonkeanus ear pattern is present at a low frequency in our samples from the central core and southeastern peninsula (see discussion in the section covering geographic variation); these specimens closely resemble the condition in the lectotype and paralectotypes of tonkeanus , as shown in Meyer’s plate (fig. 9) and the color images provided by Clara Stefen.

More samples of tree squirrels from the east-central arm of Sulawesi, especially in the region near the central core, are required to assess the actual range of variation in color and size of the ear tufting in those populations. Although we are currently ignorant of the variation in cranial and dental dimensions among populations on the east-central arm, we suspect the ranges of those dimensions would not be significantly different than those we have uncovered here for our samples from the central core and southeastern peninsula (table 19), but only through study of new material from the east-central peninsula can our assumption be tested. Until then we regard the sample of tonkeanus to represent a population of Prosciurillus alstoni in which the ears have buffy fringes and lack white ear tufts, at least in the type series, the only sample currently available for study.

Sciurus sarasinorum Meyer, 1898: 1 ( 1899: 21, pl. V). LECTOTYPE: SNSD B3826 (skin and skull; measurements are listed in table 14), an adult male collected during February 1895, by Fritz and Paul Sarasin. TYPE LOCALITY: Indonesia, Propinsi Sulawesi Tengah (northern portion of the central core), Mapane, 01 ° 269S, 120 ° 409E, near sea level at the southern margin of Teluk Tomini (locality 4 in gazetteer and on map in figure 11).

Clara Stefen sent us color photographs of the skin and skull of the specimen collected at Mapane—and we also have in front of us prints of the skull derived from negatives

TABLE 24 Summary of Habitat at Trapping Sites, Stomach Contents, and Other Relevant Information for Specimens of Prosciurillus alstoni Collected by Musser in Central Sulawesi, 1975–1976 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). The two collection localities (Kuala Navusu and Sungai Tolewonu) are in tropical lowland evergreen 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 and (ASE) Elevation numbers (m) Date Trap site and other information

Kuala Navusu 31 Aug. 1975 On trunk (diameter 5 in.) of tree that had fallen across stream,

226076 (3080) connecting the stream terrace on one side to the opposite bank and

226079 (3105) bordering forest, trunk 5 ft above surface of water; just up from

226080 (3170) camp. ASE 3105 caught during early morning in September 1975;

226081 (3265) ASE 3170 and ASE 3265 also caught during September. A Prosciurillus murinus was trapped on same spot on a different day. Stomach of ASE 3080: distended with freshly consumed mixture of pink and tan fruit pulp consisting mostly of fig seeds and rind; also a few small macrolepidopteran larvae. Stomach of ASE 3170: partially full of fruit pulp, along with fibrous material, probably the outside of a fibrous seed around which the pulp was removed.

226078 (3092) 31 Sept. 1975 On rotting trunk (1 ft diameter) lying on flat, wet ground on forested terrace below camp.

226077 (3093) 31 Sept. 1975 On rotting trunk (1.5 in. diameter) lying in rattan and shrub understory on wet bank of stream below camp. Caught during early morning.

226086 (3228) 40 Sept. 1975 On trunk (6 in. diameter) lying across main stream; trunk is at water level and usually inundated with every rise in stream after rains. Stomach: full of figs (mostly the tiny fig seeds) and a little bait.

226083 (3168) 40 Sept. 1975 On rotting section of trunk (8 in. diameter) lying across deep canyon containing tributary of Kuala Navusu. Stomach: full of brown fruit mash.

226084 (3276) 40 Sept. 1975 On rotting trunk (8 in. diameter) lying across steep tributary ravine. Rubrisciurus rubriventer was caught at the same spot on a different day. Stomach: full of fruit, some of it figs; also macrolepidopteran caterpillars.

226085 (3308) 40 Sept. 1975 On long, smooth trunk of Livistona palm that had fallen across part of stream terrace, connecting one terrace with that on the other side of the stream; trunk about 10 ft above water surface. A Prosciurillus murinus was trapped at same spot on a different day. Stomach: empty.

226087 (3584) 53 Nov. 1975 On trunk (6 in. diameter) lying across deep ravine dissecting forested hillsides where the trees Artocarpus sericarpus and Myristica sp. are common. Stomach: full of tan fruit mash.

226088 (3335) 122 Oct. 1975 On rotting trunk (5 in. diameter) lying across deep streambed (10 ft

226090 (3365) above rocky bed) and 4 ft above slope on either side of stream (now

226095 (3432) dry). Dense undergrowth and scrubby forest on either side of ravine. Stomach of ASE 3365: partially full of soft fruit mash; a few formicine worker ants, pupae of a larger ant species (probably dug out of a nest), several large beetle larvae, and macrolepidopteran caterpillars. Stomach of ASE 3432: partially full of soft fruit mash.

TABLE 24 (Continued)

Locality, AMNH and (ASE) Elevation numbers (m) Date Trap site and other information

226089 (3367) 122 Oct. 1975 On trunk (10 in. diameter) lying horizontal in understory; from base

226091 (3375) of slope it extends 3 ft above ground and projects over deep (15 ft)

226092 (3397) rocky ravine where crown ends about 10 ft off ground on other side of ravine. Dense understory covering wet and rocky slopes. ASE 3397 caught in early morning. Stomach of ASE 3367: partially full of unidentifiable brown stuff and insects (several badly decomposed small caterpillars, remains of one large adult beetle, some buprestid beetle larvae, a heavily sclerotized beetle larva, and a few other larvae that are probably from a different kind of beetle.

226096 (3595) 122 Nov. 1975 On limb (6 in. diameter) bridging shallow, steep-sided ravine in good

226093 (3638) hillside forest adjacent to main stream in an area of recent tree-falls. Stomach of 3595: partially full of fruit mash.

226094 (3427) 122 Oct. 1975 On slender 40 ft trunk (5 in. diameter) of dead, small tree lying across ravine and enclosed stream; trunk extends from one side of ravine to the other side where it projects about 2 ft off the ground into dense and scrubby understory. The rat Paruromys dominator was caught in same place.

226097 (3366) 137 Oct. 1975 On trunk (8 in. diameter) leaning across hillside at 35 ° from ground into top of scrub in second-growth forest. Stomach: partially full of a little soft fruit mash, but mostly many beetle larvae of Buprestidae , some complete, others chewed into pieces—both large and small instars are represented.

226098 (3383) 137 Oct. 1975 On smooth trunk (8 in. diameter) lying across rocky slope in open

226098 (3468) understory near base of large, clifflike outcrop. Trunk lay 3 ft above ground for most of its length (60 ft). Trapped examples of Prosciurillus murinus on same spot during different days. Stomach of ASE 3383: full of reddish brown fruit mashed with a few macrolepidopteran caterpillar remains. Stomach of ASE 3468: full of tan fruit mash plus bright yellow fruit mash; also tiny white insect larvae that may have infested the fruit.

226100 (3656) 152 Nov. 1975 On large (60 ft long, 3 ft diameter) section of trunk lying on ground alongside shallow and wide part of main stream; good forest with broken canopy. Stomach: distended, half is bright yellow fruit mash that has been found in Rubrisciurus rubriventer and the rat Taeromys celebensis ; the other half is pinkish-tan fruit mash mixed with sclerites and legs of a few small adult beetles.

226101 (3257) 229 Sept. 1975 On rotten trunk (6 in. diameter) lying across damp ravine that is well shaded by rattan and understory trees on steep slope above tributary stream. Caught Prosciurillus murinus at same spot on a different day. The rat Maxomys hellwaldii was taken on ground nearby.

Sungai Tolewonu 122 Jan. 1976 On rotting trunk (20 ft long, 6 in. diameter) lying across wet ravine in

226510 (3855) thick understory about 20 ft from edge of river. Stomach: full of brown fruit mash packed with worker and soldier termites (members of Termitidae ). Squirrel clearly found the nest of arboreal termites. No other kinds of insects present.

226511 (3876) 122 Jan. 1976 On trunk lying across wet ravine 2 ft away from trunk where ASE 3855 was caught yesterday. Both of these tree trunks are wet and rotten, about the same size, and lay at 45 ° across the ravine near its confluence with river. Forest is short, understory is scrubby; tall canopy trees are Paba, Leutu, and others. Stomach: partially full of brown fruit mash; no insects.

TABLE 24 (Continued)

Locality, AMNH and (ASE) Elevation numbers (m) Date Trap site and other information

226512 (3894) 122 Jan. 1976 On rotting branch (8 in. diameter, 15 ft long) that lay over depression on hillside just above river and below camp. In thick understory of shrubs, palm rosettes, saplings, and understory trees; Paba and Uru common canopy trees. Stomach: empty except for a few remains of undigested insects, mostly small fragments of sclerites.

226513 (3889) 152 Jan. 1976 On decaying trunk lying across tributary stream about 4 ft above water surface, bridging the two banks; caught early in morning. Stomach; full of brownish red fruit mash, rubbery pink mash; also containing numerous remains of macrolepidopteran caterpillars.

226514 (4151) 165 Feb. 1976 On large trunk of canopy tree that has fallen across main river

226515 (4228) between confluences of second and third long tributaries. Tree rests from river terrace across to rocky hillsides, is about 45 ft long and 1.5 ft in diameter, and rests 5 ft above water level. Trunk is free of vegetation except for a limb growing parallel to the trunk along its entire length, which has sprouted leaves and provides partial cover for the entire length of trunk. Caught ASE 4151 early in morning. The tree squirrels, Prosciurillus murinus and Rubrisciurus rubriventer , and rat, Maxomys hellwaldii , were taken on same spot. Stomach of ASE 4151: full of brown mash in which are mixed about two dozen termitid termites and remains of at least one small beetle.

226516 (4152) 198 Feb. 1976 On a slim (3 in. diameter) rotting trunk that lay across the left-hand fork of third large tributary; trunk extends from steep hillside bank across to stream terrace, 6 to 3 ft above water; caught early in morning. Stomach: nearly empty: brown mash, similar to that found in ASE 4151, pieces of insect legs.

226517 (4153) 198 Feb. 1976 On horizontal smooth trunk (3 in. diameter) of common understory Eugenia that is growing across the stream bed about 10 ft upstream from where ASE 4152 was caught; good forest, steep hillside on one bank, terrace on the other. Stomach: empty.

226518 (3928) 244 Jan. 1976 On decaying trunk lying across ravine of tributary stream far back and up from confluence with main river; intact hill forest, thick understory. Stomach: empty.

226519 (4053) 290 Jan. 1976 On a smooth trunk (10 in. diameter) 5 ft above water surface. This is one of several trees that have fallen; two other trunks extend from one side to other across top of ravine. Next to stream the forest understory is dense and scrubby; usual broken high canopy in back of ravine edge; caught early this morning. Stomach: nearly empty, bits of fig, mostly numerous pieces of small adult insects (sclerites, legs, antennae, some are from beetles).

226520 (4018) 305 Jan. 1976 On decaying trunk lying across branch of tributary stream in hilly headwaters that are covered with good hillside forest; tall broken canopy, dense tree understory and good ground cover. Stomach: empty.

226521 (4036) 305 Jan. 1976 On long, slender, moss-covered trunk (8 in. diameter, 40 ft long) that rests along one side of ravine, curves across onto other side; good hillside forest and near where the other squirrel, ASE 4018, was trapped yesterday; caught this morning. Stomach: partially full of remains of figs (rind with large seeds on stalks).

TABLE 24 (Continued)

made by G.H.H. Tate in 1951 during his visits to European museums. An unstuffed pelt dried in a distorted shape forms the skin. All appendages (ears, feet, and tail) are intact, and color of the fur is similar to that depicted in Meyer’s color plate (1899; reproduced in fig. 10); the white ear tufts are conspicuous. The color hues of the pelt have retained the vibrant integrity of the live squirrel, even after 113 years. In color of upperparts, underparts, tail, ears, and extent of white ear tufts, the skin of SNSD B3826 is inseparable from H.C. Raven’s sample collected in the 1920s from Pinedapa, southeast of Mapane, Musser’s specimens collected in the 1970s from the Malakosa region northwest of Mapane (see map in fig. 17) and from G. Heinrich’s samples collected in the 1930s from the southeastern peninsula. Except that the hues are paler in Anderson’s (1879) color plate of ‘‘ Sciurus alstoni ’’ (see fig. 16), in particular the underside being much less brightly colored, that portrait bears a striking resemblance to the squirrels from Mapane and Usu portrayed as sarasinorum in Meyer’s color plate (see fig. 10), a population identity reinforced by their shared white and prominent ear tufts.

The skull of SNSD B3826 is intact except for a slight separation between the interparietal and occipital region, and a hole in the bony palate and one on top of the skull behind the nasals, which traces the path of a shotgun pellet. The left dentary is whole but the posterior ramus of the right dentary is in fragments, the anterior portion undamaged. Values from cranial and dental dimensions of the holotype fall within the range of variation of our samples that share similar fur coloration and patterning. In the ordination of specimen scores for population samples (identified in table 1) from the Malakosa region, Pinedapa, and the southeastern peninsula projected onto the first and second principal components extracted from principal-components analysis, the score for the holotype of sarasinorum clusters with scores representing specimens from those three geographic regions (fig. 21).

Meyer (1898: 1) based his description of sarasinorum on two specimens collected by the Sarasins and, as was the custom then, did not designate either as the holotype. One was collected at Usu (‘‘Oesoe’’ on old maps; also spelled ‘‘Ussu’’), 02 ° 369S, 121 ° 069E, near sea level at the northern end of Teluk Bone on February 18, 1896; the other was obtained farther north at Mapane on the south coast of Teluk Tomini in February 1895. Both specimens are represented by live poses in Meyer’s (1899, pl. 5) color plate (reproduced here in fig. 10). Meyer (1899: 21) noted that the bottom pose represents the individual from Usu and was rendered natural size, while the top pose, which is one-half natural size, was drawn from the animal collected at Mapane. Both individuals have prominent white ear tufts, and are closely similar in color and patterning of the fur over body and tail.

Identifying the type locality and type specimen for sarasinorum is puzzling for anyone checking the published literature up to the late 1990s. Ellerman’s (1940: 375) classic compendium on families and genera of living rodents noted only ‘‘Central Celebes’’ as the source of the two specimens. More than a decade later, in their checklist of land mammals of New Guinea, Celebes, and adjacent islands, Laurie and Hill (1954: 93) simply referred to both locations in their indication of a type locality: ‘‘Ussu, northern end of Gulf of Boni, and Mapane, southern end of Gulf of Tomini, Celebes.’’ By the early 1990s, Corbet and Hill (1992: 304) had stated the type locality to be ‘‘Ussu, N end of Gulf of Boni, Sulawesi,’’ which would imply that they regarded the specimen from Usu to be the type.

Neither Usu nor the specimen collected there in 1896 were listed in Feiler’s (1999: 407) accounts of type specimens housed in the collection of the Staatliches Museum für Tierkunde at Dresden. He identified as ‘‘holotype’’ the specimen from Mapane, and that place as the type locality. No reference was made to a specimen from Usu. Perplexed, we inquired of Clara Stefen if the squirrel from Usu was still in the collection at Dresden, and she (in litt., 2008) wrote ‘‘I can’t find any hint of it; it’s not in the collection, but [I] did not even see an entry on the relevant systematic index card which seems odd to me, as they [the cards] probably were started under the directorship of Meyer.’’

We subsequently located the squirrel from Usu in the collection at the Naturhistorisches Museum Basel. In the mammalogy archives at AMNH we found a photograph of a skull from Usu taken by George H.H. Tate when he visited Basel in 1951. Our contact at Basel, Loïc Costeur, informed us that the specimen is listed in the museum’s database; he subsequently found it in the collection, and sent us color images of the skin and skull. The adult skin (NMB 1199), prepared as a stuffed museum study specimen, is intact except for the head where large patches of skin are missing on the right and left side. The animal was clearly shot in the head and this damage is reflected in the skull (NMB 8080). The anterior half is intact and all teeth are present but all that remains of the posterior half is the roof of the braincase and cracked basicranial region still retaining the auditory bullae; the mandible is complete. The specimen was collected by Fritz and Paul Sarasin, and is one of the two studied by Meyer (1898, 1899). The specimen from

The only species from the southeastern peninsula heretofore known has already been spoken of above [the previous description covered ‘‘ Sciurus sarasinorum ’’]. In the new Sarasinian material there were two skins from two neighboring localities from about the central part of this region. These skins are very closely related to Sciurus sarasinorum though slightly different. The body size is the same as is that of the tail; the head is somewhat shorter.

The two skins exhibit the same colouration. The hairs of the back are yellowish brown with black annulations; the brown colour is darker in the median zone than it is on the sides where it has a more grayish tone. I have observed the same in Sciurus leucomus . The tufts on the ears are very well developed; they are white with a distinct yellowish tint. The inside of the ears is equipped with fine fox red hairs. A yellowish brown ring of fine short hairs surrounds the eyes. The sides of the snout to almost under the eyes are brick red. In Sciurus sarasinorum these parts are of the same colour as the rest of the head. The surface of the belly, the underside of the limbs and a stripe in the anal region reaching to the first fifth of the tail are brick red. On the upper sides of the feet this coloration is somewhat lighter and mixed with black. The tail displays about the same colouring as that of Sciurus sarasinorum ; it is not on the whole, however, as black, but is more brown because the brown rings are broader The subterminal portion of the hair is white; the hairs on the tuft are also mostly tipped whitish. [Roux listed measurements for the two specimens; see appendix 5]

It is most likely that the southern peninsula also harbours species of Sciurus that belong to the Leucomus group. So far, however, they are not known.

Roux basically redescribed sarasinorum , which was detected by Moore (1958: 2): ‘‘Inspection of the Archbold material [at AMNH], Roux’s (1910, p. 519) description, and Meyer’s (1899, p. 5) colored plate of sarasinorum shows that Roux’s squirrel is an obvious synonym of sarasinorum . ’’ This attribution is reinforced by our inspection of the color images of the lectotype and paralectotype (these designations were made by someone at Basel and entered into their unpublished catalog of type specimens, according to Loïc Costeur, who examined the specimens for us; Roux had not identified either specimen as the holotype) sent to us by Loïc Costeur. Following this manuscript catalog, we hereby formally designate NMB 1626/4243 the lectotype of Sciurus mowewensis Roux, 1910 , the other specimen (NMB 1627/4244) being the paralectotype. Each skin is prepared as a dry and stretched pelt; color of the fur is still bright and unaltered. The pattern of color on the body and tail, along with their prominent white ear tufts, identifies the specimens as s arasinorum, and in turn as examples of the older name, alstoni , and not something uniquely distinct.

Values for cranial and dental dimensions measured fall within the range of variation for our samples of P. alstoni from the Malakosa region, Pinedapa, and the southeastern peninsula (table 19). Except for missing second molars and a missing third premolar from the holotype, both skulls are intact, as can be seen from the photographs taken by G.H.H. Tate in 1951 (stored in the mammalogy archives at AMNH). Loïc Costeur measured both skulls for us. His values, along with those from the holotype of sarasinorum and our samples from central Sulawesi and the southwestern arm, became the subjects of principal-components analysis, which resulted in the ordination depicted in figure 21. The scores for the types of mowewensis cluster with points representing specimens from Pinedapa and the Malakosa region; the score for the holotype of sarasinorum is nearby. Scores denoting the AMNH specimens from the southeastern peninsula are scattered throughout the cluster derived from the samples from the Malakosa region and Pinedapa. There is no evidence from cranial and dental dimensions or proportions derived from the two skulls of mowewensis indicating they are anything other than another sample of sarasinorum , which identifies a population for which P. alstoni is the oldest name.

Sciurus elbertae Schwarz, 1911: 639 . HOLOTYPE: SMF 721 (a skin and skull were described by Schwarz; the skull is present in the Senckenberg Museum but the skin is missing [G. Storch, in litt., 2008]; measurements are listed in table 14), an unsexed adult collected by J. Elbert (original number 227). TYPE LOCALITY: Indonesia, Propinsi Sulawesi Tenggara (southeastern peninsula and offshore islands of Sulawesi), Eempuhu, Pulau Kabaena (05 ° 159S, 121 ° 559E), off the coast of the southeast peninsula (see map in figure 11). Schwarz identified the type locality as ‘‘Eempuhu, East Kabaëna. ’’ No elevation is attached to the specimen. Highlands constitute most of the island with topographic relief ranging from sea level to 1570 m.

Schwarz (1911: 639) examined two specimens, each consisting of a skin and skull, and regarded them to be ‘‘A pale-coloured member of the Sciurus leucomus- group.’’ Judged from his description, the fur of elbertae is somewhat paler than the rich and darker hues typical of sarasinorum from the mainland. No reference to ear tufts were made, only that the general color of the upperparts was brownish (between ‘‘olive’’ and ‘‘raw umber’’) ‘‘with a creamy-buff tinge on head.’’ We have not seen a ‘‘creamy-buff tinge’’ on the head of any specimen of P. alstoni , or any other specimen in the P. leucomus group. Unfortunately, both skins are missing from the collection at Senckenberg so we cannot determine if Schwarz was actually describing ‘‘creamy-buff’’ ear tufts, which would make sense. Both specimens from adjacent Pulau Buton have whitish buff ear tufts (see discussion of geographic variation).

The skull of the holotype (SMF 721) is complete except for broken basicranial and occipital regions; all premolars and molars are fully erupted and show moderate wear. The skull of the paratype (SMF 4878), slightly damaged, is that of a juvenile (G. Storch, in litt., 2008). Values for the cranial and dental dimensions that could be measured (by G. Storch) are listed in table 14. Some dimensions are smaller in the skull of elbertae compared with samples of species in the P. leucomus group (contrast tables 13 and 19); values for interorbital breadth, and lengths of nasals, rostrum, diastema, and bony palate, for example, are less than the range of values for these variables in the sample of P. alstoni . Length of PM4–M3 (7.3 mm) is within the range of variation for most of our samples of the P. leucomus group, which suggests to us that although all teeth are erupted and moderately worn, the skull of SMF 721 may be from a very young adult. Currently, we cannot assess the significance of the mensural differences described here without a larger sample from Pulau Kabaena containing a range of ages, from very young to old adults.

Schwarz compared elbertae with the skull of what he identified as ‘‘ Sciurus mowewensis ’’ collected on Pulau Buton. Among the distinctions he noted (primarily some smaller dimensions of elbertae ) was the presence of a distinct medial spine extending from the posterior margin of the bony palate in elbertae and its absence in the Buton specimen. Among our samples of P. alstoni (of which mowewensis is a synonym), the posterior margin of the bony palate exhibits three variants: a straight edge, a margin interrupted by a median bony nubbin, or ‘‘a distinct median spine’’ marking the edge.

On our copy of the published description where Schwarz indicated elbertae to be a member of the ‘‘ Sciurus leucomus -group,’’ G.H.H. Tate had crossed out ‘‘ leucomus ’’ and scribbled ‘‘ nigrovittatus or notatus ’’ in the margin. We had also considered these possible identifications but rejected them. Ranges of both species cover primarily peninsular Thailand and Malaya, and islands on the Sunda Shelf. Callosciurus nigrovittatus is recorded from peninsular Thailand, the Malay Peninsula, Sumatra and Java, and the small islands of Tioman and Saddle on the Sunda Shelf ( Corbet and Hill, 1992: 292). This squirrel is larger-bodied than Schwarz’s elbertae . Schwarz, for example, recorded length of the ‘‘upper tooth row’’ (PM3–M3) as 7.9 mm, which is well outside the observed range of values (minimum, 8.3–9.2 mm; maximum, 9.0– 9.7 mm; N 5 93) given by Sody (1949: 95) for 17 samples from Java, the island supporting populations of C. nigrovittatus closest to Sulawesi. Callosciurus nigrovitattus has gray underparts and prominent white and buffy parallel stripes on sides of the body (see color pl. 7 in Medway, 1969, and pl. III in Askins, 1988), ventral coloration and body patterning unlike elbertae or any other kind of squirrel known from Sulawesi.

Callosciurus notatus has a broader geographic range than C. nigrovitattus , and is found on peninsular Thailand and Malaya; the larger Sundanese islands of Sumatra, Java, and Borneo; many smaller islands on the Sunda Shelf; east to Pulau Lombok in the Lesser Sunda Islands; and on Pulau Salayar, just south of the southwestern peninsula of Sulawesi ( Corbet and Hill, 1992: 291). Sody (1949: 82) published observed ranges for length of upper tooth row derived from 17 samples of C. notatus from Java (minimum, 8.4–9.6; maximum, 8.6–10.3 mm; N 5 92) and for five specimens from Pulau Salayar, off the southern coast of Sulawesi’s southwestern peninsula (minimum, 9.3 mm; maximum, 9.7 mm), illustrating a range much greater than the value of 7.9 mm for the holotype of elbertae . Callosciurus notatus resembles C. nigrovitattus in having parallel buffy and white stripes on each side of the body; the populations on Java have either gray or buffy underparts and specimens from Pulau Salayar have dull buffy venters (see the color plates in Medway, 1969, and Askins, 1988; pl. 25 in Payne et al., 1985; Corbet and Hill, 1991: 292). The two examples of elbertae are not examples of either C. nigovitattus or C. notatus .

Prosciurillus weberi ( Jentink, 1890)

Sciurus weberi Jentink, 1890: 115 , pl. VIII, pl. X figs. 1–3.

LECTOTYPE AND TYPE LOCALITY: The lectotype of Prosciurillus weberi is an adult female (RMNH 13342) collected by Max Weber in February of 1889. A mounted skin and accompanying skeleton comprise the specimen. The skull is intact. Measurements are listed in table 13. Jentink (1890: 115) had ‘‘ three adult skins, two skeletons and one skull ’’ at hand when he described weberi , and with the aid of Chris Smeenk at Leiden, Wim Bergmans at Amsterdam, and Paula Jenkins at London, we have located all the elements that are the basis for Jentink’s description. Only RMNH 13342 remains at Leiden, and following Chris Smeenk’s suggestion (in litt., 2008), we select it as lectotype, following Article 74.1 of the Code ( ICZN, 1999: 82). One of Weber’s specimens is now in the collection at the Natural History Museum in London (BMNH 94.7.4.6, a skin only), which becomes a paralectotype. Two of Weber’s other specimens are in the collection at the Zoological Museum, Amsterdam: ZMA 11.327, a mounted skeleton (including skull) and ZMA 11.328, a mounted skin and skull; both are paralectotypes. (Chris Smeenk wrote us that in 2010, the Amsterdam collections will be moved to Leiden and become incorporated into the Leiden collections; the catalogue numbers will remain unchanged and preceeded by ZMA.)

The type locality is Palopo (‘‘near Palopo’’ is indicated on some tags), 03 ° 019S /120 ° 139E, 0–100 m (locality 2 in the gazetteer and figure 11) in the southern lowlands of the central core of Sulawesi, Propinsi Sulawesi Selatan, Indonesia. Jentink (1890: 115) gave ‘‘ Luwu near Palopo, central Celebes’ ’ as the type locality ; Luwu is the administrative district containing Palopo.

EMENDED DIAGNOSIS: As in most of the other members of the Prosciurillus leucomus group, P. weberi has a moderate-size body and a tail equal to or shorter than length of the head and body. It contrasts with the other four species by the following combination of pelage traits (see table 12): (1) dorsomedial surface of each ear covered with long black hairs projecting beyond ear rim to form a prominent tuft; (2) no bright ochraceous hairs lining inside of the pinnae, which is the same color as the fur on neck and head; (3) no nape patches behind the ears; (4) prominent middorsal black stripe extending along back from neck to base of tail (5) reddish orange underparts; and (6) currently known only from lowlands fringing the northwestern margin of Teluk Bone.

GEOGRAPHIC AND ALTITUDINAL DISTRI- BUTIONS: Records of P. weberi are represent- ed only by the voucher specimens collected between 1889 and 1940 near Masamba and Palopo in lowlands fringing the central mountains to the west and Teluk Bone to the east in the southeastern margin of Sulawesi’s central core (see gazetteer and map in figure 11). These sites are no higher than 100 m.

The known range of P. weberi is allopatric to those of two other members in the P. leucomus group. To the east, P. alstoni occurs primarily in lowland forest habitats in the eastern portion of the central core, the east-central peninsula, the southeastern peninsula, and has been collected from two islands (see map in fig. 11). Usu, in the coastal lowlands fringing the southwestern end of Teluk Bone, is the collection locality for P. alstoni closest to Masamba, which is about 80 km to the west of Usu. See the account of P. alstoni for an expanded discussion contrasting the geographic ranges of the two species.

Prosciurillus topapuensis inhabits forests in the mountains to the west of Masamba and Palopo (see map in fig. 11). No specimens have been collected lower than 350 m and no collection localities are close to the range of P. weberi . Surveys in the foothills west of Masamba and Palopo, provided that intact forest is present, are required to determine the distributional relationships between the two species. Here P. weberi may be restricted to lowland tropical evergreen rain forest and P. topapuensis to montane forest habitats.

The actual geographic distribution of P. weberi is unknown. If it is restricted to tropical lowland rain-forest habitats, the species may occur all along the eastern coast of the central core, westward through the northern part of the Tempe Depression along the margin of the mountains, and north along the western coast possibly to the Palu area—the range would fringe the west-central foothill and mountain distribution of P. topapuensis . No samples of squirrels are available from these lowlands and no place along those coasts or in the Tempe Depression has been carefully surveyed for small mammals. Much of the southern part of this lowland fringe has been deforested (see maps in Whitten et al., 1987: 102, and Supriatna et al., 1992: 37), so we may never know what member of the P. leucomus group inhabited the orginal forest cover.

In the past, P. weberi may have ranged across the Tempe Depression south throughout lowlands of the southern part of the southwestern peninsula. No modern specimens of the P. leucomus group have been collected in that region, but at least one once occurred there. A subfossil mandibular fragment collected from a cave in the Maros region attests to the former presence of a tree squirrel, possibly representing P. weberi (see the account of Prosciurillus sp. ). Tree squirrels may still exist in the small patches of lowland forest remaining on the southern part of the peninsula, but most of the region is densely populated agricultural land (see the maps in Whitten et al., 1987: 102, and Supriatna et al., 1992: 37).

Our experience with nonvolant endemic Sulawesi mammal distributions indicates that the range of P. weberi is unusual. As currently documented by voucher specimens, its range is not congruent with any other endemic mammal, a pattern contrasting with many of the distributions shared by species in other mammal groups. This discordance suggests to us that either the range of P. weberi will prove to be greater than what is now represented by specimens (as we outlined above) or that it once occurred on the southwestern peninsula but is now extinct (and possibly represented by the subfossil from the Maros region).

A young adult Prosciurillus weberi in MZB (6255) bears a skin tag indicating it was collected in 1908 from ‘‘Menado’’ (see also Sody, 1949: 107), but was certainly mislabeled. Prosciurillus leucomus is the only member of the P. leucomus group that has ever been collected from the Manado region (see gazetteer for P. leucomus ).

DESCRIPTION: ‘‘This beautiful Squirrel,’’ wrote Jentink (1890: 115–116) of weberi ,

belongs to a group of middle sized species, consisting of Sciurus leucomus , rosenbergii and others having no stripes or bands on the sides of the body and more or less prominently pencilled ears. It is distinguished from all the hitherto known East-Indian-Squirrels by having a rather broad black band along the spine of the back, running from the neck, increasing in broadness in the middle of the back and diminishing towards the root of the tail. The ears are adorned with rather long black hairs which form a kind of small pencil. It is to be observed that in one of the three type-specimens the black earpencils are slightly tipped with white. For the rest the upperparts and sides of the head and body and the legs are covered with very soft hairs showing a reddish-black tinge, occasioned by being each black hair ringed with reddish; the underside of head, the breast, belly and inside of legs have the black hairs largely tipped with red, so that the named parts present a fine red hue. The tail shows undistinct rings; upperpart of tail with hairs ringed with red and black ending in white tips; towards the tip of the tail the red and white disappear so that the tip of that organ is black; as the tail is distichous it is evident, that on the underside the red tinge prevails. The whiskers are black; they reach as far backwards as the end of the earpencils.

In its physical size and proportions of feet and tail, P. weberi closely resembles P. alstoni (table 15). Dark brown upperparts sprinkled with orange and black, along with dark reddish or reddish orange underparts, are also common to both species. The combination of prominent black ear tufts and black middorsal stripe, however, is unique to P. weberi , and is illustrated in Jentink’s (1890) color plate of weberi posed on a tree limb. Views of the skull are also provided by Jentink. Number and position of teats match the pattern seen in all members of the P. leucomus group (one postaxillary pair, two inguinal pairs). Conformation of the skull is closely similar to that illustrated for P. alstoni and P. topapuensis (figs. 12–14); cranial and dental measurements are summarized in table 17. See the account of P. alstoni for additional information.

COMPARISONS: Prosciurillus weberi is allopatric to the geographic ranges of P. alstoni and P. topapuensis and requires close comparisons with those species. Contrasts are described in those two accounts.

GEOGRAPHIC VARIATION: Because the few specimens of P. weberi are from the same small geographic area, we cannot assess the nature of geographic variation in this species.

ECOLOGY: Other than knowing that the collection localities were in lowland tropical evergreen rain forest, we have no other information about the biology of P. weberi . We suspect that its habits are similar to those of P. topapuensis and P. alstoni , which inhabit the upper canopy and eat mostly soft fruits, seeds, and insects.

ECTOPARASITES: No records.

SYMPATRY: Only Rubrisciurus rubriventer has been collected at the same place as P. weberi (see gazetteer and table 6), but P. murinus probably inhabited the same forests at the time the specimens of R. rubriventer and P. weberi were collected.

SYNONYMS: None.

Prosciurillus topapuensis ( Roux, 1910)

Sciurus topapuensis Roux, 1910: 518 . Callosciurus leucomus hirsutus Hayman, 1945:

576.

HOLOTYPE AND TYPE LOCALITY: The holotype of Prosciurillus topapuensis is an adult female (NMB 1628/4245) collected by Paul and Fritz Sarasin on September 16, 1902. A dry skin and accompanying skull comprise the holotype. The skin is an intact flat pelt. Most of the skull is present; the right zygomatic arch, most of the basioccipital and the occipital condyles are missing, as are the left third and fourth premolars and third molar. The mandible is complete except for missing coronoid processes, and all teeth are missing from the right dentary. Measurements are listed in table 13.

The type locality is Gunung Topapu (approximately 02 ° S, 120 ° 159E; estimated from a copy of the original map used by Paul and Fritz Sarasin), 1550 m (locality 16 in gazetteer and map in figure 11). This highland is contained in the mountains west of Teleboi in the Bada region of the west-central region of Sulawesi, the north and western portion of Propinsi Sulawesi Selatan, Indonesia .

Our identification of the holotype of topapuensis as an example of the P. leucomus group requires explanation. A single specimen is the basis of Roux’s (1910: 518) description of Sciurus topapuensis and his account is short (translated from the original in German; the German text is reproduced in appendix 5):

This animal most closely resembles Sciurus tonkeanus Meyer but displays certain differences. The general colouring of the head, the back and the tail is the same, the ventral side being different. Instead of the common brick red colour these have a grayish Yellow tinge, the base of the hairs being grey while the tips are more yellowish.

The ears are quite distinctly tufted, the tufts being of the same colour as the head. The yellowish tone is somewhat brighter on the sides of the body and on the limbs than it is on the back. The tail hairs have three broad yellowish brown rings of which the two basal ones are slightly darker coloured than the third subterminal. Most of the hairs of the tail tuft are black with a broad yellow terminal ring.

Measurements: Total length, 30; head length, 5; body 10 cm; tail with tuft, 15, tuft, 5.5 cm; hind foot c.u., 4.1; front foot c.u., 2.5; ear with tuft, 1.3 cm.

To judge from the skull the animal is fullgrown. There is no difference between this skull and that of Sciurus leucomus . The dimensions of this form are somewhat under normal for the species of the Leucomus group. Furthermore, the ventral side is brick red in all the other species. Only in this species is it grey. I am not able to determine whether this is ascribable to seasonal variation or not.

Roux compared the skin with ‘‘ Sciurus tonkeanus ’’ [5 Prosciurillus alstoni ], noting close resemblance between the two in color of the fur over the head, back, and tail, but a sharp difference in coloration of the underparts. Instead of being ‘‘brick red’’ as in tonkeanus , the venter was described as grayish yellow, the base of the hairs being gray and the tips yellow. Later in the account, Roux referred to the underparts as gray. He also noted the presence of distinct ear tufts the same color as the head, and that the yellowish tone seen on the back was brighter along sides of the body and on the limbs.

Roux’s references to ‘‘yellow’’ (‘‘gelb’’) and ‘‘yellowish’’ (‘‘gelblich’’) actually refer to slightly different buffy tones. On all specimens we assign to P. topapuensis (see gazetteer), the dorsal fur is a rich brown with buff highlights intermixed with black, a reflection of the black and orange or buff banding on the overhairs. Sides of the body and dorsal surfaces of the limbs are slightly paler (grayer) because the buffy bands of the hairs are much paler than those on the back, ranging from pale buff to cream. The coat covering underparts of these specimens is dark gray speckled or washed with pale buff or cream—the hairs are gray for most of their lengths and tipped with pale buff or cream, a combination reflecting Roux’s ‘‘yellowish buff.’’ Overall, the underparts appear dark gray flecked with buff, as Roux noted (a few specimens have dark gray underparts tinted with silver—a dark grayish white), and contrast sharply with the reddish venters common to all the other species in the P. leucomus group. Most specimens we examined exhibit black ear tufts ranging in expression from prominent to less conspicuous; a few individuals sprout tufts containing fewer black hairs with the result that the ears appear nearly inseparable from color of the head, the pattern Roux described for the single specimen of topapuensis he examined.

Coloration of the fur on the skin of the holotype of topapuensis has altered through the years in storage. Loïc Costeur kindly sent us a color image of the skin, which is preserved flat, unstuffed. The black bands on hairs of the back are now dark brown, the buffy bands yellow (the color of straw); the remnants of the underparts are paler than upperparts but yellowish with only a hint of the original gray.

Although the skull of the holotype is damaged, Loïc Costeur was able to measure all but four of the cranial and dental dimensions we investigated (table 26) and the values were included in our multivariate analyses. In the ordination showing specimen scores projected onto first and second principal components in figure 26, the score representing the holotype of topapuensis clusters with those for specimens from Sungai Miu and Sungai Sadaunta; falling nearby are the scores representing specimens from Gunung Kanino, Besoa, and Rano Rano.

EMENDED DIAGNOSIS: Prosciurillus topapuensis shares with other members of the P. leucomus group a body moderate in size (tables 15, 16), but contrasts with the other four species by the following combination of pelage traits (see table 12): (1) tail averages shorter relative to length of head and body; (2) dorsomedial surface of each ear covered with long black hairs projecting beyond ear rim to form a conspicuous black tuft on most specimens; (3) no bright ochraceous hairs lining inside of the pinnae, which is the same color as the fur on neck and head; (4) no nape patches behind the ears; (5) no middorsal black stripe extending along the back from neck to base of tail (6) fur over upperparts thicker and subdued in tone on squirrels from high altitudes; (7) dark gray underparts lightly brushed with silver, pale buff, or ochraceous tones; and (8) a geographic range restricted to the western mountainous portion of Sulawesi’s central core, where the species ranges in altitude from tropical lowland evergreen rain forests covering foothills to montane rain-forest formations.

GEOGRAPHIC AND ALTITUDINAL DISTRI- BUTIONS: Voucher specimens identified as P. topapuensis describe a range concordant with valleys, foothills and mountains covering the western portion of the central core of Sulawesi, from Danau Lindu and surrounding highlands in the north (about 01 ° 199S, 120 ° 009E) to Pegunungan Latimojong (03 ° 309S, 121 ° 239E) in the south—the west-central region (or western mountain block) of Sulawesi’s central core—and an altitudinal range from 350 m in tropical lowland evergreen rain forest to 2800 m in upper montane forest habitats (see gazetteer for P. topapuensis and map in fig. 11). Although collection sites are spotty throughout this extensive region, P. topapuensis likely occurs everywhere in suitable forested habitats. Whether it inhabits forests along the western coastal lowlands (between about 1 ° and 3 ° 309S) is unknown, for no specimens of squirrels have been collected there. This region has never enjoyed surveys documenting the diversity of small mammal species. If P. topapuensis does descend to altitudes along the coastal plain just above sea level, it would be in forests covering these western lowlands bordering the Makassar Strait. On the other hand, possibly a different member of the P. leucomus groups inhabits those coastal forests. A good candidate is P. weberi , which is currently known from only two places east of the mountains in the lowlands fringing Teluk Bone. Conceivably, the species could be found in the lowlands flanking the southern margins of the central mountains and all along the western coast in lowland forests fringing the western flanks of the mountainous central region.

We cannot, however, dismiss a speculation that P. topapuensis may inhabit only foothills and mountains but not along the coast. Along Musser’s transect in the northern portion of Sulawesi’s southern core, he collected different species of squirrels from 290 m to the summit of Gunung Nokilalaki above 2200 m. The lowest elevation was along Sungai Oha Kecil, a tributary of the Sungai Miu and downriver from the camp at 350 m, which is the lowest record for P. topapuensis . Musser trapped along the Oha Kecil and in the adjacent hillsides employing long traplines that included many Conibear traps. He caught only the large-bodied Rubrisciurus rubriventer and the much smaller-bodied Prosciurillus murinus (see those respective gazetteers). No examples of P. topapuensis were encountered or even seen in the forest until the Sungai Miu camp and higher along the Sungai Sadaunta. We need more inventories of tree squirrels derived from transects extending into coastal lowlands elsewhere along the western coastal flanks of the west-central mountain block.

No specimens of squirrels in the P. leucomus group are available from the highlands between the Danau Lindu region and approximately 80 km to the north in the basal portion of the northern peninsula at Bumbarujabu (00 ° 439S, 120 ° 049E). The sample from Bumbarujabu is P. leucomus , exhibiting a very different pelage pattern and found only on the northern peninsula (see section on comparisons). We do not know where the geographic range of P. topapuensis ends in this unsurveyed interval and that of P. leucomus begins.

The forested lowlands east of the mountainous range of P. topapuensis , from south of Parigi along the fringing eastern coast to the eastern portion of the central core, are inhabited by populations of P. alstoni , another member of the P. leucomus group characterized by a color pattern strikingly different than seen in P. topapuensis (see comparisons). In the Mapane region at the southern margin of Teluk Tomini, the distance between collection sites for P. alstoni at Mapane and Pinedapa and to the west in the mountains at Rano Rano where P. topapuensis was encountered is only about 20 airline km. Other collection sites for each species are much farther apart—where the western range boundary for P. alstoni lies relative to the eastern boundary for P. topapuensis is not known (see the account of P. alstoni for details). The eastern flanks of the mountainous central core—generally west of a line extending from the Mapane region in the north through the west side of Danau Poso to the western foothills in the south in the areas west of Masamba, Palopo, and the eastern flanks of Pegunungan Latimojong may delimit the eastern range boundary for P. topapuensis , at least where suitable forest habitats persist and in the past when forest cover was less altered than at present.

In the Masamba and Palopo lowlands west of the central mountains, the indigenous member of the P. leucomus group is P. weberi , not P. alstoni (see that account). In this area the mountainous range of P. topapuensis sits adjacent to that of lowland P. weberi , which has been collected only around Masamba (02 ° 349S, 120 ° 199E) and Palopo (03 ° 019S, 120 ° 139E).

The southern flanks of Pegunungan Latimojong and the adjacent highlands to the west likely mark the southern boundary of the range of P. topapuensis . This end of the west-central mountainous core of Sulawesi is separated from the southern portion of the southwestern peninsula by the Tempe Depression, a lowland trough containing flood lakes over which a river may have flowed in the past. One of the lakes, Danau Tempe, ‘‘… appears to be a remnant of an ancient strait that formerly separated the southern arm of the Toraja highlands [Pegunungan Latimojong and highlands to the north] from southern Sulawesi’’ ( Whitten et al., 1987: 259). No modern specimens of the P. leucomus group have been collected south of the Tempe Depression, but a member once occurred there and that population is represented by a subfossil fragment (see account of Prosciurillus sp. ).

Available voucher specimens of Prosciurillus topapuensis identify it as part of the mammalian fauna endemic to the west-central region in Sulawesi’s central core. This is the area of foothills, interior valleys, and mountain peaks lying west of Danau Poso, and extending from the Palu area in the north to Pegunungan Latimojong in the south. In addition to P. topapuensis , samples of three species of shrews ( Crocidura ), one or two primates ( Tarsius ), a ground squirrel ( Hyosciurus ), and 13 species of murid rodents have been collected so far only from the west-central foothills, valleys, and mountains (table 25).

DESCRIPTION: Prosciurillus topapuensis has dark brown upperparts, black ear tufts, and dark gray underparts washed with buff or silver. The squirrel is similar to P. leucomus and P. alstoni in body size and most physical proportions (length of head and body, 151–190 mm; length of hind foot, 40–49 mm; length of ear, 15–21 mm; weight, 130–210 g; see tables 15, 16). Thickness and color of the fur covering upperparts of head and body are also similar; all three species possess a uniformly thick coat (12–15 mm) that is dark brown flecked with black and orange along midline of the body from forehead to base of tail, but grading into a slightly paler brown spotted with black and pale buff along sides of the head, body, forelegs, and hind legs. Dorsal surfaces of the front and hind feet may be the same color as the head and back, slightly darker, or slighter paler (buffy gray). Coloration of the body fur results from the mixture of dark gray curly underfur; longer overfur composed of hairs each of which is blackish gray with a subterminal orange or buffy band and black tip; and black guard hairs barely projecting beyond the overhair layer (about 5 mm). Animals from high altitudes in montane forests tend to be slightly paler; the subterminal bands on the overhairs are buff instead

TABLE 25 Mammal Species Currently Recorded Only from the West-Central Region in Sulawesi’s Central Core a of orange, and the dorsal coat slightly thicker.

The dorsal background forming upperparts of P. topapuensis is interrupted by a buffy ring encircling each eye, a buffy muzzle and chin, and black ear tufts. Glossy black hairs cover the medial surface of each pinna and extend beyond its margin to form a short (4–8 mm) black tuft, closely similar in configuration to the tufts in the samples of P. leucomus from the northern peninsula. Within the sample, the tufts vary in their expression, and are so reduced in one specimen (a few black hairs) as to be inconspicuous—this squirrel has a uniform dark brown back with buff and black highlights. Prosciurillus topapuensis lacks any sign of nape patches behind the ears— fur covering top and sides of the neck is indistinguishable in color from the rest of the upperparts—or a black middorsal stripe.

Fur covering the underparts from chin to base of the tail is moderately thick (up to 10 mm) and very dark gray with silver highlights (dark grayish white) or washed with pale buffy or ochraceous hues. The overhairs are dark gray for two-thirds of their lengths and unpigmented at the tips or retain pale buffy or ochraceous bands. Dark grayish white venters are infrequent in our sample of P. topapuensis , and even those specimens with primarily silver-tinted venters exhibit small buffy infusions on the chest. The ventral coloration hardly contrasts with that of the upperparts.

The tail of P. topapuensis is shorter than the length of the head and body (LT/LHB 5 84%–96%, see table 16) and is covered in long hairs, each patterned by alternating black and buffy or orange bands. The overall effect is rings of black and buff with buffy and short black bands outlining margins of the tail and long black hairs forming a black terminal tuft. The ventral surface of the tail resembles the upperparts, dark brown with orange and black highlights bordered by black and buffy margins. Overall coloration of the tail is similar to that seen in P. leucomus , P. alstoni , and P. weberi , but somewhat browner.

As with the other members of the P. leucomus group, female P. topapuensis have three pairs of teats, one abdominal pair and two inguinal pairs. One female with enlarged teats exhibited a placental scar in the right uterine horn; other females examined, although adult, showed no reproductive activity.

Views of the skull are provided in figures 12–14; cranial and dental measurements are summarized in tables 17 and 26.

COMPARISONS: Prosciurillus topapuensis requires comparisons with P. leucomus , which ranges to the north throughout the northern peninsula, and P. alstoni and P. weberi , both occurring in lowlands adjacent to the mountainous distribution of P. topapuensis . See the account of P. alstoni for contrasts between it and P. topapuensis ; here we describe differences between P. topapuensis and the other two species.

Prosciurillus topapuensis and P. leucomus : Body size is comparable in the two species, but the tail averages shorter relative to length of head and body in P. topapuensis (tables 15, 16). Representatives of P. topapuensis from lower altitudes show a dorsal coat similar in thickness and coloration to that of P. leucomus , but individuals from upper montane forests have a thicker, slightly duller coat, the orange and bright buffy bands of the hairs giving the lowland squirrels their speckling against the brown background are slightly paler in the montane animals. Both species have black ear tufts, but those of P. topapuensis vary in size among the samples, ranging from being prominent (like those in samples of P. leucomus ) to showing only traces of black on the ears of a very few individuals; every specimen of P. leucomus has prominent black tufts. The inner surface of each ear at the base of the tuft is the same color as the rest of the head and not bright ochraceous as seen in P. leucomus . Fur behind the ears on the neck of P. topapuensis is the same color as the head and back, which is strikingly unlike that region in most specimens of P. leucomus where the neck is marked by conspicuous whitish nape patches. Finally, all examples we studied of P. topapuensis show a dark gray coat highlight- ed with silver or hues of buff and ochraceous covering the underparts; the ventral fur of P. leucomus ranges from reddish orange to ochraceous (table 12). Specimens of P. topapuensis appear monochromatic (brownish gray everywhere), those of P. leucomus are bicolor (dark brown upperparts, reddish underparts).

As with all members of the P. leucomus group, skulls of P. topapuensis and P. leucomus closely resemble one another in general size and shape but contrast in average size and proportions of some dimensions. Except for length of nasals, breadth of rostrum, and height of braincase, cranial and dental dimensions average less in the sample of P. topapuensis compared with P. leucomus (table 17). That difference can also be visualized by results of multivariate analyses illustrated in an ordination showing specimen scores projected onto the first and second principal components extracted from principal-components analysis (fig. 24). The group of scores representing geographic samples of P. leucomus overlaps with the cloud of points identifying specimens of P. topapuensis along the first component (an estimate of size) but is pushed slightly farther to the right by the moderate to high positive loadings of all variables except height of braincase (table 27). Fewer variables strongly influence the distribution of scores along the second axis, which reflects shape distinctions. Compared with P. topapuensis , the northern peninsular species P. leucomus has a relatively wider interorbit, mastoid region, and bony palate, and longer orbit and auditory bulla, but relatively shorter nasals, diastema, and tooth row, and narrower rostrum. These are subtle shape differences but some of them, especially the narrower interorbit and shorter orbit of P. topapuensis , are visually apparent when skulls of each species are compared side-by-side.

Prosciurillus topapuensis and P. weberi : These two species are similar in body size and contrast in relative tail length along with

TABLE 26 Descriptive Statistics for Cranial and Dental Measurements (mm) Derived from Population Samples of Prosciurillus topapuensis Mean ± 1 SD and observed range (in parentheses) are listed.

fur texture in much the same way as seen between P. topapuensis and P. leucomus . The former has a shorter tail relative to length of head and body, duller upperparts, and dark grayish buff or grayish white underparts instead of the reddish orange covering typical of P. weberi . Both species lack nape patches, both exhibit black ear tufts, but the back of P. topapuensis is unmarked by a black middorsal stripe, which is so conspicuous on all examples of P. weberi .

The ordination of specimen scores project- ed onto the first and second principal components extracted from principal-components analysis in figure 25 summarizes our multivariate analysis of the cranial and dental variables we employed. No separation of the scores representing specimens of each species is apparent along the first axis, a general estimate of size, partly because the sample of P. weberi is so small compared with that of P. topapuensis , and partly because of the breadth of individual and geographic variation in that species as indicated by the wide scatter of points along the first component, which is influenced by the moderate to high positive loadings of most variables (table 28). The few variables primarily contributing to the spread of scores that form two nearly separate constellations along the second axis reflect contrasts in shape. Prosciurillus weberi has a relatively broader rostrum compared with P. topapuensis and relatively shorter nasals and tooth rows. This analysis needs to be revisited employing a much larger sample of P. weberi than is currently available.

GEOGRAPHIC VARIATION: Every example of P. topapuensis from the west-central mountain block is immediately recognizable by its short and black ear tufts, unpatterned neck and back, and very dark grayish white or grayish buff underparts. Variation in color patterning over the fur covering head and body appears to be altitudinal and individual rather than regional. Examples from high altitudes, usually in upper montane forests, have slightly thicker and duller upperparts than those taken in the foothills and lower places. Extent of the ear tufts range from being conspicuous on most squirrels to nearly hidden in the fur surrounding the ears in a very few specimens, and underparts range from dark gray speckled or washed with buff or ochraceous to dark gray lightly sprinkled with white (appearing as silvery highlights); these expressions in ear tufts and tone of underparts occur within the same sample as well as among population samples and do not appear to be correlated with altitude.

Body size may vary with altitude. Along Musser’s transect, the only line along which samples have been collected that extends from tropical evergreen rain forests in the lowlands to montane rain-forest habitats in adjacent highlands, squirrels from Sungai Miu and Sungai Sadaunta, in lowland evergreen rain forest between 350 and 915 m, average slightly greater in many body and cranial dimensions than do those collected from higher altitudes along the transect, and from mountains east and south of the region where Musser worked (tables 16, 26).

Results of multivariate analyses of cranial and dental variables as presented in the ordination of specimen scores projected onto the first and second principal components extracted from principal-components analysis in figure 26A reflect the altitudinal variation in size and also identifies a regional component. Covariation in nearly all variables contributes to the spread of scores along the first axis (table 29), reflecting increasing size from left to right, with points for most geographic montane samples intermingled, and scores representing samples from Sungai Miu and Sungai Sadaunta (350–915 m) anchoring the right sector of the constellation. Scores representing squirrels collected between 1400 and 1500 m on nearby Gunung Kanino are sprinkled through the cloud from the right to the center.

Scores for specimens from Gunung Lehio (crosses in fig. 26A), just south and east of Musser’s transect, are scattered in the con-

TABLE 27 Results of Principal-Components Analysis Contrasting Population Samples of Prosciurillus topapuensis with Those of Prosciurillus leucomus Principal components are extracted from a covariance matrix of log-transformed values for 15 cranial and 1 dental variable; see figure 24.

stellation from left to right along the first component and also tend to lay along the outskirts of the larger cluster from the perspective of the second axis. This alignment suggests the sample from Gunung Lehio to have a relatively wider rostrum and bony palate, longer tooth row, and shorter diastema as compared to specimens in the other geographic samples (table 29).

We can easily describe the results diagrammed in the principal-components ordination but assessing the significance of the pattern is difficult without additional samples. Certainly variation in size of skull concomitant with age reflects one component of the broad spread of scores along the first axis—samples are mixed, containing the range from young adults to old adults. Comparing larger samples, each composed of the same relative age, may better resolve the true pattern of altitudinal and regional variation. Musser’s small samples from low altitudes are highlighted in the cluster of scores, suggesting that populations in lowlands and foothills may average larger in body size than those living at higher altitudes

TABLE 28 Results of Principal-Components Analysis Contrasting Population Samples of Prosciurillus topapuensis with the Sample of Prosciurillus weberi Principal components are extracted from a covariance matrix of log-transformed values for 15 cranial and 1 dental variable; see figure 25.

in montane forest habitats. This aspect of variation gleaned from study of material at hand needs to be tested by larger samples collected along transects from different regions in the mountainous western section of Sulawesi’s central core. Finally, some populations of P. topapuensis may be partially isolated from one another by large rivers or other kinds of geographic barriers, which might be revealed by a greater density of sampling over the geographic range that we have described.

We were tantalized by the average differences in body size between the two samples of squirrels, one from 350–915 m and the other from higher altitudes, because among other groups of mammals, especially murid rodents, there is a pattern of closely related species replacing each other with altitude generally concordant with forest type in the western mountain block of Sulawesi’s central core (table 31). One set of shrews ( Crocidura ), bats ( Thoopterus ), tarsiers ( Tarsius ), and squirrels ( Hyosciurus ), and three sets of murid rodents ( Bunomys , Haeromys , and Margaretamys ) collected along Musser’s transect, from the lowlands in the Sungai Miu region to the summit of Gunung Nokilalaki in montane forests provide examples. The lowland species in each of these sets is very distinctive compared with their respective montane relatives; the dissimilarity is expressed in body size, fur coloration, and marked contrasts in cranial conformations along with cranial and dental dimensions.

No such marked morphological differences contrast squirrels in the samples of P. topapuensis from Sungai Miu and Sungai Sadaunta compared with those collected on the slopes of Gunung Kanino or even higher in other parts of the western mountain block. Specimens in each set of samples are indistinguishable in fur coloration and expression of black ear tufts, those from the lower altitudes simply average slightly larger in body size and have slightly brighter upperparts. Apparently the same species ranges from foothills to higher in the mountains, which parallels the altitudinal distributions of two species of Crocidura , one macaque, two other tree squirrels, and six species of murid rodents (table 31).

ECOLOGY: Prosciurillus topapuensis inhabits the upper canopy of primary forest. In all the different altitudinal forest formations, the squirrels mostly remain in crowns and woody vines forming the upper canopy. They are active in the mornings and afternoons during sunny days, but might be active all day long on cloudy days. They travel through the upper canopy, but come to the ground occasionally, especially in places where the canopy is broken. When on the ground, the squirrels run on trunks and limbs lying on the forest floor and regularly cross streams and rivers along bridges formed by branches of live trees arching over the water as well as rotting tree trunks and limbs and palm trunks connecting opposite stream terraces. Most of Musser’s specimens were trapped on these kinds of river and stream crossings (habitats at trap sites are summarized in table 32).

Musser encountered P. topapuensis all along the transect line, from Sungai Miu at 350 m to the summit of Gunung Nokilalaki at 2300 m (see photograph of forest on Gunung Kanino in fig. 27), from lowland evergreen rain forest to upper montane rain forest, and through a range of ambient temperatures (table 2) and habitats—forested stream terraces, hillsides, and ridgetops. Although wary and quiet in the forests around Tomado at 1000 m, and usually seen only in tree crowns forming the high canopy, the squirrels were aggressive and loud elsewhere along the transect line. At higher altitudes along the Sungai Tokararu and slopes of Gunung Kanino (1150–1500 m), for example, P. topapuensis was common, aggressively vocal, and unwary—Musser saw squirrels there every day. On several different days he would be walking along the trapline and every time he stopped to rebait a trap a squirrel would begin its loud, resonant, scolding chatter, always positioned just out of reach above his head. The squirrel remained quiet as long as Musser kept walking, but once he stopped the scolding began again until he left the area. On one occasion, a squirrel was in the crown of an understory tree isolated from nearby trees. To get out of the tree, the squirrel had to come to the ground but would not do so while Musser was there. It would scamper out to the end of a branch, stop and stare at Musser, then run to a different branch, all the time its tail twitching spastically with each raspy chuck. On another day, a squirrel left the upper canopy descending to the trunk where it stopped, splayed against the bark, head down, and tail flipping in agitated undulations with each loud chatter. It scold- ed for one or two minutes, then would run back up into the crown and disappear from sight. In addition to human presence, the squirrels were very sensitive to human voices, the sounds of trees falling in the forest, or even a limb hitting the ground, and the sights of hawks and eagles—all elicited a cacophony of strident loud chatter.

The sound of rain approaching in the forest also prompted alarm calls from squirrels and some birds. After the clouds build up and the sky darkens, and just before the torrent begins there is a period of absolute silence in the forest, shattered only by birdsong and a squirrel’s resonant chucks. The squirrels call individually, not in chorus, and cease as the first drops splatter against the leaves, signaling the more intense approaching rain, which passes through the forest like a moving wet curtain. Several times Musser saw a squirrel on a large woody vine in the upper canopy calling in a long series of chattering chucks, then dart into a hole in the nearest tree just before the first drops were felt.

Musser looked for leaf nests but never found them. At one camp he passed a tall strangler fig on the way to check traps. The host tree had long since decayed and the enclosing trunk was filled with debris and leaf litter, held tightly by the criss-cross patterns of ropy trunk. Each day he would see a pair of squirrels in the crown, running over the upper trunk, limbs, and along the huge woody vines encircling the trunk. Periodically they would disappear into a space between the latticework, and then reemerge. One evening he watched them chase each other, feed on figs, and eventually disappear out of sight inside the trunk. Tall strangler figs are common in lowland rain forest and offer numerous sites for nests within the debrisfilled hollows. Likely, cavities in other kinds

TABLE 30 Results of Principal-Components Analysis of Population Samples of Prosciurillus topapuensis Fewer variables are used than listed in table 29 to accommodate the damaged holotypes of topapuensis and hirsutus. Principal components are extracted from a covariance matrix of log-transformed values for 10 cranial and 1 dental variable; see figure 26B.

of trees are also used for nest sites by P. topapuensis .

Judged by contents of stomachs, the diet of P. topapuensis consists of soft fruits, especially figs, some seeds, and arboreal insects (table 32). All the different fruits were thoroughly masticated, and only figs could be positively identified (pieces of rind and endosperm with tiny seeds attached). The insects consisted of small adult beetles (represented by fragments of wings, elytra, legs, and sclerites), macrolepidopteran caterpillars (whole or chewed into pieces) gleaned from surfaces of leaves and bark, and cockroaches (indicated by legs, antennal segments, wings, and masticated bodies). Beetles were uncommon in stomachs, but the caterpillars must be a favorite prey. Remains were found in many stomachs, and of these one (AMNH 225501) was distended with large (20–30 mm long) macrolepidopteran caterpillars, many of them still in one piece. The stomach of another squirrel (AMNH 225504) was packed with smaller (10–15 mm long) macrolepidopteran larvae, whole and in pieces. Both squirrels were caught on Gunung Kanino above 1400 m (table 32). Cockroaches are in the trees and on the ground. Soft fruits and insects also

TABLE 31

Tropical Lowland Evergreen Rainforest Species, Their Montane Forest Replacements, and the Species Found in Both Lowland and Montane Forests in the West-Central Region of Sulawesi’s Central Core a comprise the diet of the genetic and geographic relative P. alstoni , but the range of insect groups encountered in stomachs of that species was greater than that recovered from stomachs of P. topapuensis (table 24).

ECTOPARASITES: Prosciurillus topapuensis is parasitized by a host-specific sucking louse, Hoplopleura topapuensis , n. sp. (described in a following section), as well as two species of flea (table 56). Sigmactenus sulawesiensis

TABLE 32

Summary of Habitat at Trapping Sites, Stomach Contents, and Other Relevant Information for Specimens of Prosciurillus topapuensis

Collected by Musser in Central Sulawesi, 1973–1976 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). Three of the collection localities (Sungai Miu, Sungai Sadaunta, and Tomado) are in tropical lowland evergreen rain forest; lower montane rain forest describes the places on Gunung Kanino, and tropical upper montane rain forest the site on Gunung Nokilalaki. With one exception, 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 and other information

Sungai Miu 350 Mar. 1974 On top of rotting trunk lying in forest understory next to small

224042 (1510) stream. Stomach: empty except for the few remains of a small adult beetle.

Sungai Sadaunta 675 Feb. 1974 5 ft above ground on dead tree limb projecting through

224041 (1413) understory of primary streamside forest. Stomach: partially filled with grayish red remains of large-seeded fig; pieces of small adult beetles (sclerites with tissue attached).

225581 (2059) 762 Sept. 1974 On wet, smooth and rotting trunk lying across stream near camp in streamside forest. A Prosciurillus murinus and a rat, Maxomys musschenbroekii , were taken in same spot. Stomach: full of reddish mash from a fig with large seeds; a small ant, and bits of sclerites from small adult beetle.

224582 (2037) 824 Sept. 1974 On wet, decomposing Pigafetta palm trunk lying across Sungai Sadaunta about 200 ft upstream from where Rubrisciurus ASE 2033 was trapped (see table 10). A Prosciurillus murinus was caught on same spot on a different day. Stomach: full of reddish mash of fig remains, with small seeds; no insects.

224583 (2194) 824 Oct. 1974 On rotten, wet wanga ( Pigafetta filaris ) palm trunk lying across Sungai Sadaunta in undisturbed streamside forest. Prosciurillus murinus and the rat, Rattus hoffmanni were taken at same spot. Stomach: nearly empty, remains of a soft fruit; no insects.

224584 (2119) 869 Oct. 1974 On half-shredded, rotting, and slick Pigafetta palm trunk lying across Sungai Sadaunta, extending from a low terrace (muddy, rocky, and eroded) on one side of stream across to the opposite rocky bank. A rat, Bunomys sp. , was caught in same trap. A few feet upstream another rat, Paruromys dominator , was caught on a different trunk bridging the stream. Prosciurillus murinus was caught at same spot as ASE 2119 on a different day. Stomach: partially full of same reddish fig pulp and seeds found in other stomachs.

224585 (2223) 869 Oct. 1974 On large rotting trunk lying across stream connecting one forested terrace to the opposite stream terrace. Several examples of the rats Paruromys dominator and Rattus hoffmanni were trapped at same spot. Stomach: full of finely chewed fruit, some of it figs, along with remains of small macrolepidopteran caterpillars and a few small adult beetles.

224586 (2281) 915 Oct. 1974 On large, rotting trunk covered with moss, epiphytes, ferns, shrubs, and vines lying across stream, leaning from the bank on one side of stream to terrace on other side. Stomach: partially full of reddish brown mash from soft fruit.

TABLE 32 (Continued)

Locality, AMNH Elevation and (ASE) numbers (m) Date Trap site and other information

224587 (2325) 915 Nov. 1974 On limb of understory tree growing across ravine containing main

224588 (2345) 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 the limb to be used regularly; trap set about 7 ft from stream level. Examples of three kinds of squirrels (the ground squirrel, Hyosciurus ileile ; the tree squirrels, Rubrisciurus rubriventer and Prosciurillus murinus ) and an arboreal rat ( Rattus marmosurus ) were taken in the same spot. So three kinds of tree squirrels, one ground squirrel, and at least one species of arboreal murid used the same living pathway as a bridge over the stream.

226835 (4320) 915 Mar. 1976 Caught about 6:00 a.m. at same spot on limb of understory tree where Prosciurillus topapuensis ASE 2325 and ASE 2345 and other squirrels and rats were trapped in 1974. Stomach: distended with remains of tiny-seeded figs with remains of a few macrolepidopteran caterpillars.

Tomado 1000 Aug. 1973 On rotting trunk lying across narrow stream in high, primary

226932 (491) forest. Stomach: partially full of orangish tan soft fruit mash containing masticated remains of a few macrolepidopteran caterpillars, a cockroach, and a small adult beetle.

Gunung Kanino 1402 Feb. 1975 On wet, rotting limb lying across small stream (Sungai Salubeka)

225501 (2545) bordered by dense forest understory in steep canyon. A rat, Paruromys dominator , was also trapped here. Stomach: packed with large (20–30 mm long) macrolepidopteran caterpillars mixed with a bit of blackish debris clinging to everything; no other kind of insects.

225502 (2547) 1402 Feb. 1975 On large rotting, moss-covered trunk lying across Sungai Salubeka. Stomach empty.

225504 (2487) 1418 Feb. 1975 On rotting, wet, moss-covered tree limb bridging the Sungai Salubeka. Stomach: packed with a mass of small (10–15 mm long) macrolepidopteran caterpillars, whole and in pieces, mixed with brownish black debris, similar to ASE 2545 except larvae are much smaller.

225503 (2466) 1418 Jan. 1975 On limb growing across Sungai Salubeka in canyon. Stomach: half full with dark brown and tan remains of fruit, along with chewed pieces of several macrolepidopteran caterpillars, and bits of a small adult beetle.

223532 (829) 1463 Nov. 1973 On rotting trunk lying along ground in forest. Stomach of 829: full

223533 (831) of orange-brown soft fruit mash (same stuff as in 831 and other squirrels from Gunung Kanino) and remains of two macrolepidopteran caterpillars. Stomach of ASE 831: distended with orange-brown soft fruit mash mixed with the masticated remains of a few macrolepidopteran caterpillars, and several cockroaches.

223534 (845) 1463 Nov. 1973 On wet, rotting tree trunk lying on ground in forest. Stomach: nearly empty except for a few bits of soft fruit (mostly rind) and remains of two macrolepidopteran larvae (caterpillars).

TABLE 32 (Continued)

( Siphonaptera , Leptopsyllidae ) is recorded from a voucher host collected at Tomado, and the same species of flea also parasitizes five species of endemic Sulawesian murid rodents: Bunomys fratrorum , B. penitus , Eropeplus canus , Maxomys musschenbroekii , and Paruromys dominator ( Durden and Beaucournu, 2000) . Murids are typical hosts of Sigmactenus spp. fleas (Durden and Traub, 1990) so the record of S. sulawesiensis from P. topapuensis is likely an atypical host association ( Durden and Beaucournu, 2000). Farhangia sedecimdentata ( Pygiopsyllidae ) was found on voucher squirrels collected at Sungai Tokararu and Gunung Kanino, and to date has not been discovered parasitizing other rodents ( Mardon and Durden, 2003). The squirrel hosts were initially identified as Prosciurillus leucomus . Farhangia spp. are considered to be ‘‘nest fleas’’ of Asian (Bornean and Sulawesian) tree squirrels ( Traub, 1980); F. sedecimdentata is therefore probably more common in nests of P. topapuensis .

SYMPATRY: The range of P. topapuensis in the central core of mainland Sulawesi overlaps those of the tree squirrels P. murinus , Rubrisciurus rubriventer , and both species of ground squirrels, Hyosciurus (table 6). Along his transect in the northern portion of central Sulawesi, Musser caught P. topapuensis in the same traplines along with examples of those four species, and sometimes in the same traps (see table 32).

SYNONYMS: Only one scientific name is a synonym of P. topapuensis : reasons behind its allocation are presented below.

Callosciurus leucomus hirsutus Hayman 1945: 576 . HOLOTYPE: BMNH 40.691g (stuffed museum study skin and slightly damaged skull; see table 13 for measurements), an adult male collected by W.J.C. Frost in 1938 or 1939. TYPE LOCALITY: Tamalanti, a ‘‘Plantation between Rantekaroa [02 ° 509S, 119 ° 509E] and Koelawi [01 ° 279S, 119 ° 599E]’’ ( Laurie and Hill, 1954: 156), 3300 ft (1006 m), in the western portion of the central core of the island, Propinsi Sulawesi Selatan, Indonesia.

Frost collected six tree squirrels from Tamalanti, and Hayman (1945: 577) considered them to represent a member of the ‘‘peculiar leucomus group’’ and diagnosed Callosciurus leucomus hirsutus this way:

Slightly smaller, according to available skull measurements, than typical leucomus ; general colour above, a uniform warm grizzled mixture of amber- brown and black, below, pale buffish mixed with grey, the buff colouring most clear and intense on throat and chest. The underside very distinct from the clear brick red of leucomus and other related forms, but sharing this distinction with Callosciurus topapuensis Roux (the nearest form geographically). Ears clothed above with black hairs, forming a distinct tuft contrasting with the colour of head and body. Tail short, of same general hue as body, but with long terminal pencil or tuft of all black hairs. Coat very full and shaggy.

‘‘The shaggy coat is particularly long and full in [hirsutus]’’ wrote Hayman (1945: 578), ‘‘and in conjunction with the tufted ears and particularly long terminal tail tuft suggested the subspecific name.’’ While Hayman appreciated the unique coloration of the fur characterizing hirsutus, he did not consider it to be diagnostic of a separate species; ‘‘ Callosciurus leucomus hirsutus is strikingly different from C. leucomus leucomus of N. E. Celebes in colour, but is here treated as a subspecies to indicate relationship, in view of its obvious affinities as shown by the skull’’ ( Hayman, 1945: 578). Hayman acknowledged that among the named forms of Callosciurus leucomus (mentioning leucomus and occidentalis but not alstoni , tonkeanus , sarasinorum , mowewensis , or elbertae ), hirsutus resembled only Roux’s (1910) Callosciurus topapuensis , which he knew only from Roux’s published description, writing ‘‘But topapuensis is said to have ear tufts coloured as the head, and to have the sides of the body, and the limbs a lighter, more yellowish, tone than the back, two characters which definitely do not apply to the Tamalanti form.’’ But these tonal aspects of Roux’s description do apply to all six of Hayman’s hirsutus. In these and all examples of the other species in the Prosciurillus leucomus group, sides of the body and dorsal surfaces of the limbs and feet are slightly paler than the back due to the pale buff or cream bands on the overhairs in the fur covering these regions, as we explained above. The contrast in color of the ear tufts, from concolorous with the head, as in the holotype of topapuensis , to black as described by Hayman for hirsutus reflects extremes in the range of variation we see within our sample from several highland regions in the central core of Sulawesi.

Some dimensions of the skull from the holotype of hirsutus (kindly measured for us by Paula Jenkins) are slightly smaller than those for the holotype of topapuensis (table 13). This distinction is reflected in the ordination of specimen scores projected onto first and second principal components derived from principal-components analysis that is shown in figure 26B where the score for hirsutus falls among scores representing specimens from Rano Rano, Gunung Lehio, Besoa, and Pegunungan Latimojong. Mean values for cranial dimensions average smaller in these highland samples than in the material from Sungai Miu and Sungai Sadaunta, which is closer to the holotype of topapuensis in size of skull. Again, the two holotypes represent variation within a single species, this time in cranial dimensions, with the skull of hirsutus nesting with the smaller specimens and that of topapuensis with the larger individuals (fig. 26B, table 30).

Prosciurillus rosenbergii ( Jentink, 1879)

Sciurus Rosenbergii Jentink, 1879: 37 . Sciurus tingahi Meyer, 1896: 27 .

LECTOTYPE AND TYPE LOCALITY: The lectotype of Prosciurillus rosenbergii is a young adult female (RMNH 13362, specimen ‘‘l’’ in Jentink’s [1888: 24] catalog) obtained by D. Hoedt on October 27, 1865, from ‘‘ Siao,’’ Kepulauan Sangihe. It consists of a skin mounted in a live pose and an incomplete skull (extracted from the mount after Jentink’s tenure at Leiden). Values from dental and some cranial measurements are listed in table 14.

Specimen ‘‘l’’ was one of 12 specimens cataloged as ‘‘types’’ (5 syntypes) by Jentink (1887: 190–191, 1888: 23–24). All the skulls are badly damaged or smashed. At the suggestion of Chris Smeenk (in litt., 2008), we select specimen ‘‘l’’ as lectotype following the procedure stipulated in Article 74.1 of the Code; ICZN, 1999: 82). Of the 12 syntypes, specimen ‘‘l’’ ‘‘has a beautiful skin and the second best skull’’ (C. Smeenk, in litt., 2008). The remaining 11 specimens become paralectotypes: (1) specimen ‘‘ a ’’ ( Jentink,1888: 23), RMNH 13351, a mounted skin and incomplete skull, male, from ‘‘Sanghi’’ (5 Pulau Sangihe or Kepulauan Sangihe), collected November 7, 1864, by collectors for C.B.H. von Rosenberg; (2) specimen ‘‘ b ’’ ( Jentink, 1888: 23), RMNH 13352, a mounted skin and cranial fragments, female, from ‘‘Sanghi’’ (5 Pulau Sangihe or Kepulauan Sangihe), collected November 2, 1864, by collectors for C.B.H. von Rosenberg; (3) specimen ‘‘ c ’’ ( Jentink, 1888: 23), RMNH 13353, a mounted skin and incomplete skull, female, from ‘‘Sanghi’’ (5 Pulau Sangihe or Kepulauan Sangihe), collected November 10, 1864, by collectors for C.B.H. von Rosenberg; (4) specimens ‘‘ a ’’ (the skull, in Jentink’s osteological catalog, 1887: 190) and ‘‘ d ’’ (the skin, in Jentink, 1888: 24), RMNH 13354, a mounted skin and incomplete skull, male, from ‘‘Sanghi’’ (5 Pulau Sangihe) collected January 16, 1866, by D. Hoedt; (5) specimen ‘‘ e,’’ RMNH 13355, a mounted skin and incomplete skull, male, from ‘‘Siao’’ (5 Pulau Siau) collected October 20, 1865, by D. Hoedt; (6) specimens ‘‘ b ’’ (the skull, in Jentink, 1887: 190) and ‘‘ f ’’ (the skin, in Jentink, 1888: 24), RMNH 13356, a mounted skin and incomplete skull, male, from ‘‘Siao’’ (5 Pulau Siau) collected October 23, 1865 by D. Hoedt; (7); specimen ‘‘ g ’’ ( Jentink, 1888: 24), RMNH 13357, a mounted skin and fragments of a skull, male, from ‘‘Siao’’ (5 Pulau Siau) collected October 22, 1865 by D. Hoedt; (8) specimens ‘‘ c ’’ (the skull, in Jentink, 1887: 190) and ‘‘ h ’’ (the skin, in Jentink, 1888: 24), RMNH 13358, a mounted skin with incomplete skull, male, from ‘‘Siao’’ (5 Pulau Siau) collected October 24, 1865, by D. Hoedt; (9) specimen ‘‘ i ’’ ( Jentink, 1888: 24), RMNH 13359, a mounted skin along with skull fragments, male, from ‘‘Siao’’ (5 Pulau Siau) collected October 24, 1865, by D. Hoedt; (10) specimens ‘‘ d ’’ (the skull, in Jentink, 1887: 192) and ‘‘ j ’’ (the skin, in Jentink, 1888: 24), RMNH 13360, mounted skin and damaged skull, male, from ‘‘Siao’’ (5 Pulau Siau) collected November 9, 1865, by D. Hoedt; (11) specimens ‘‘ e ’’ (the skull, in Jentink, 1887: 192) and ‘‘ k ’’ (the skin, in Jentink, 1888: 24), RMNH 13361, a mounted skin and partial skull, female, from ‘‘Siao’’ (5 Pulau Siau) collected October 23, 1865, by D. Hoedt. Skulls ‘‘ a–e,’’ were removed during Jentink’s tenure; the others were extracted later.

The type locality is Kepulauan Sangihe, Pulau Siau (also spelled ‘‘Siao’’), 02 ° 499N, 125 ° 239E (see gazetteer and map in figure 28), Propinsi Sulawesi Utara, Indonesia. No details are recorded concerning the precise collection locality on Pulau Siau.

Chris Smeenk (in his manuscript pages of the ‘‘Type-specimens of recent mammals in the National Museum of Natural History, Leiden’’; sent to us in 2009) notes that ‘‘ Jentink (1888) omitted the locality for the specimens collected by von Rosenberg. The latter did not visit the Sangir islands himself, but had sent local hunters to collect on the islands during the period September–November 1864; see his notes preserved in the archives of the Leiden Museum. It is not recorded on which island the material was collected.’’

EMENDED DIAGNOSIS: Prosciurillus rosenbergii is comparable in body size and length of tail relative to head and body length to other species in the P. leucomus group (tables 14, 15). It is unique among members of that cluster by having dark brown or chestnut brown upperparts without contrasting chromatic patterns (no patches on the nape behind the ears, no black middorsal stripe) or adornments (no ear tufts), dark brownish gray or brownish buff underparts (no specimen with reddish-orange to ochraceous or dark gray ventral fur) and blackish tail.

GEOGRAPHIC AND ALTITUDINAL DISTRI- BUTIONS: Prosciurillus rosenbergii is endemic to Kepulauan Sangihe where samples come from Pulau Sangihe, the largest island in the archipelago; the smaller Pulau Siau, south of Sangihe; and the small islands of Tahulandang and nearby Ruang (see table 4 and fig. 28). All the specimens were collected during the latter 1800s and none have information indicating elevations at which they were caught. We do not know the present status of the species in the archipelago but would not be surprised if most populations are extinct. Whitten et al. (1987: 49) write that ‘‘Virtually all of [Pulau] Sangihe has been converted to coconut and nutmeg plantations or else is covered in patches of secondary forest from abandoned gardens,’’ and that ‘‘Sangihe and other northern islands were largely deforested by 1920’’ (p. 91).

Two other species of nonvolant mammals are endemic to the Sangihe Archipelago, both recorded only from Pulau Sangihe. One is the primate, Tarsius sangirensis , which is very different in its morphology compared with samples of tarsiers from mainland Sulawesi and adjacent islands ( Feiler, 1990; Groves, 2005). The other is a distinctive species of murid rodent related to those in the Rattus xanthurus group on mainland Sulawesi and nearby Pulau Peleng (Musser, MS; Musser and Carleton, 2005: 1462).

The islands forming Kepulauan Sangihe are mostly volcanos ringed by coastal lowlands scattered along the wide volcanic Sangihe arc platform, most of which is submerged about 1000 m beneath sea level (Sheets NA 51-4 and 51-12). Volcanic activity along the arc has been present since middle Miocene times ( Hall, 2002: 382). Pulau Biaro, the southernmost island in the archipelago, is separated from Pulau Talise and Pulau Bangka, lying just off the Sulawesi mainland in shallow water, by sea at least 1000 m deep. There is no current evidence of any past Pleistocene or earlier land connection between the archipelago and mainland; presumably the ancestors of the few nonvolant mammals restricted to the archipelago reached the islands by some form of overwater dispersal.

DESCRIPTION: The highlights of Jentink’s (1879: 37) description portray a squirrel characterized by coloration of fur strikingly different from any members of the P. leucomus group occurring on mainland Sulawesi and closely adjacent offshore islands.

General tint of the upperparts of the body and outside of legs rusty brown; the colour of the underparts is much brighter. Hairs of the back with a broad black ring towards the tip. Tail darker than the back, each long hair is embellished with a broad black ring and a very long black tip. In-and outside of the ears closely covered with hairs which do not form a pencil.

Whiskers black and not projecting beyond the tips of the ears. Toes and fingers and also the fore and hind-feet closely covered with hairs on the upper parts, the hairs being here dusky near the base with silvery tips.

Prosciurillus rosenbergii is comparable to squirrels in samples of P. leucomus , the member of the P. leucomus group that is closest geographicallyto P. rosenbergii , in size of body, appendages, and length of tail relative to head and body length (table 15), but strikingly different in coloration of pelage and without any of the color patterns seen in the other members of the P. leucomus complex. The entire upperparts—head, neck, back, ears, and dorsal surfaces of the feet—of specimens from the two northern islands, Sangihe and Siau, are rich chestnut-brown (Jentink’s ‘‘rusty brown’’), a background due to the mix of dark gray underfur, longer overhairs that are dark gray for most of their length and tipped with subterminal reddish brown and black bands, and the glossy dark guard hairs. The dorsal coat is unmarked by ear tufts, nape patches, or a middorsal black stripe. The underparts range from dark brownish gray to brownish buff and hardly contrast with color of the fur over dorsal surfaces of the head and body; hairs are dark gray basally and tipped with brown or brownish buff bands. Squirrels in samples from the southern island, Tahulandang, and its nearby smaller companion, Ruang, are slightly paler, having brown upperparts and brownish buff underparts. In all samples, the tail conspicuously contrasts with the head and body because it is blackish with rusty highlights everywhere, an effect resulting from the long black hairs with subterminal dark brown bands and long black tips (as Jentink observed, ‘‘each long hair is embellished with a broad black ring and a very long black tip.’’ See Feiler’s (1990: 89) black and white images of P. rosenbergii and Meyer’s [1896] color plate, which is reproduced in figure 9.

Females exhibit three pairs of teats, one postaxillary pair and two inguinal pairs.

Although all the skulls we examined are damaged in some way, the intact portions are closely similar to comparable cranial sections in skulls of other species in the P. leucomus group; cranial and dental measurements for some dimensions are summarized in table 17.

COMPARISONS: Brown or chestnut-brown upperparts, slightly paler underparts, and a blackish tail define the color pattern peculiar to P. rosenbergii . All other species in the P. leucomus group have brown upperparts flecked with orange and buff, and the underparts are reddish orange to ochraceous ( P. leucomus ), reddish brown to reddish orange ( P. alstoni and P. weberi ), or dark gray with a silver or buffy wash ( P. topapuensis ). The ears are the same color as the head and neck and lack tufts (conspicuous black tufts in P. leucomus , P. topapuensis , and P. weberi ; white tufts in P. alstoni ), no white, buffy or gray patches adorn sides of the neck behind the ears (nape patches are typical in P. leucomus ) and no middorsal black stripe marks the back (diagnostic for P. weberi ). The tail is blackish flecked with rust except at the tip (the tail is the same color as upperparts in the other species and only the long tuft at the end is black). Because all the skulls available for measuring are severely damaged, we cannot provide comparative morphometric analyses other than to report that mean values of those dimensions that could be measured are comparable to those derived from samples of other species in the P. leucomus group (table 17).

GEOGRAPHIC VARIATION: Other than the slight difference in color saturation of the pelage in samples from the two northern islands in the Archipelago as opposed to the southern Pulau Tahulandang and neighboring Pulau Ruang, we can provide no information on aspects of other kinds of variation that might be associated with island location. Insular variation in cranial and dental dimensions among samples from the four islands cannot be assessed because all skulls are incomplete.

ECOLOGY: No information is available. In its morphology, P. rosenbergii is closely related to other members of the P. leucomus group, and we suspect the species to inhabit primarily the upper canopy as Musser observed for P. alstoni and P. topapuensis ; only direct field observations will provide this kind of information.

ECTOPARASITES: No records. Durden combed the fur in a large sample of P. rosenbergii but could not locate any attached sucking lice.

SYMPATRY: No other species of squirrel has ever been collected on any island in Kepulauan Sangihe.

SYNONYMS: The only other scientific name attached to Prosciurillus rosenbergii is discussed below.

Sciurus tingahi Meyer, 1896: 27 . LECTOTYPE: SNSD B2842 (stuffed museum study skin and damaged skull; see table 14 for measurements), an adult male purchased from Charles Cursham in 1894. TYPE LOCALITY: Kepulauan Sangihe , Pulau Tahulandang , 02 ° 219N, 125 ° 259E, Propinsi Sulawesi Utara, Indonesia (see gazetteer and map in figure 28). No details are recorded about the precise collection locality on Pulau Tahulandang.

Feiler (1999: 408–409) identified SNSD B2842 as the ‘‘holotype’’ and four other specimens as paratypes. Meyer (1896: 27) based his description of tingahi on several specimens without identifying a holotype so the specimen Feiler set aside as holotype is a lectotype and the others are paralectotypes (see Article 74.5 of the Code [ ICZN, 1999: 82–83], and our accounts of P. leucomus and P. alstoni ).

Although described as a separate species related to Prosciurillus rosenbergii from the northern Pulau Sangihe and Pulau Siau, we consider Meyer’s sample of tingahi to represent populations of P. rosenbergii on the southern Pulau Tahulandang and closely adjacent Pulau Ruang. Specimens of tingahi have slightly paler pelage than those from Pulau Sangihe and Pulau Siau, brown upperparts and buffy brown underparts as opposed to dark chestnut brown upperparts and dark brownish gray or buff venters in squirrels from the two northern islands. Meyer’s tingahi is portrayed in his color plate reproduced in figure 9.