Rubrisciurus Ellerman, 1954

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

Rubrisciurus Ellerman, 1954

The early aboriginal human inhabitants of Sulawesi’s forests were likely very familiar with this large-bodied and beautiful red squirrel. However, rubriventer , the type-species of Rubrisciurus , did not come to the attention of European naturalists until the Dutch traveler E.A. Forsten obtained a specimen sometime during 1840 to 1842 and sent it to the museum in Leiden where it was studied, named, and described by Müller and Schlegel in 1844 as a species of Sciurus . The tree squirrels, Prosciurillus leucomus and P. murinus , were named and described (also as species of Sciurus ) in the same publication; the three constitute the earliest formal taxonomic exposition of Sulawesi’s endemic squirrel fauna. From the time Forsten’s squirrel reached European awareness, rubriventer has been treated as an especially distinctive species endemic to Sulawesi. Eventually

TABLE 5 Summary of Island Distributions off the Mainland of Sulawesi and in Kepulauan Sangihe for Species of Prosciurillus Derived from Voucher Specimens a See map in figure 28 for the archipelago distribution of P. rosenbergii and maps in figures 11 and 30 for insular distributions of P. leucomus , P. alstoni , and P. murinus .

transferred from Sciurus to Callosciurus , the morphology of rubriventer so impressed Ellerman (1940: 350) that he wrote ‘‘[the species] might perhaps form a distinct subgenus’’ and arranged it as the only member of a ‘‘ rubriventer Group’’ in his annotated compilation of ‘‘The Families and Genera of Living Rodents.’’

Fourteen years later, Ellerman formalized his impression by appointing rubriventer as the type species for Rubrisciurus , which he considered to be a subgenus of Callosciurus . Large size was his primary diagnostic criterion. He first compared the species to Asian Ratufa and Bornean Rheithrosciurus , both containing species of large body size (see Medway, 1969; Payne et al., 1985), and eventually, but somewhat skeptically, to Sciurus , and finally was ambiguous about the relationship of Rubrisciurus to other squirrel genera:

The generic position of C. rubriventer is not very clear as I do not think the structure of its baculum has been recorded, and Callosciurus essentially only differs from Sciurus in the structure of its baculum and its Indomalayan

(instead of Palaearctic or American) habitat.

But whatever the structure of the baculum in C.

rubriventer, it stands so sharply apart from all other Sciurus and Callosciurus species of Europe and Asia that at least subgeneric rank seems required for it.

By 1959 Moore (1959: 176), using a combination of cranial traits, raised subgenus Rubrisciurus to generic level and placed it in the subtribe Hyosciurina , tribe Callosciurini along with the Sulawesian Prosciurillus and Hyosciurus , and the Sundaic-Philippine Exilisciurus . Corbet and Hill (1992: 305) recognized Rubrisciurus in their treatise on The Mammals of the Indomalayan Region but expressed a note of uncertainty: ‘‘This [ Rubrisciurus ] was described as a subgenus of Callosciurus mainly on the basis of its large size. Its relationship to Callosciurus , Sundasciurus and Prosciurillus is unclear.’’ Earlier, Moore (1959) regarded Sundasciurus to be a member of the subtribe Callosciurina in the Callosciurini and did not consider it to be especially closely related to Rubrisciurus , which he placed in a different subtribe. Moore’s hypothesis of phylogenetic relationships for Rubrisciurus has been vindicated in part by analyses of the nuclear IRBP and mitochondrial 12S and 16S ribosomal DNA used by Mercer and Roth (2003) to recover a squirrel phylogeny that included a monophyletic group containing Rubrisciurus , Prosciurillus , and Hyosciurus nested within a cladistic cluster formed by Callosciurus , Sundasciurus , and their relatives. Within the Sulawesian clade, Rubrisciurus is basal to Prosciurillus and Hyosciurus , and more closely related to the former than to the morphologically

TABLE 6 Summary of Distributional Sympatry over the Mainland of Sulawesi for the Species of Rubrisciurus , Hyosciurus , and Prosciurillus Derived from Voucher Specimens a divergent ground squirrels in the latter. The phylogram of relationships generated by Mercer and Roth excludes the species of pygmy squirrels in Exilisciurus from the Sulawesian clade; it forms a monogeneric clade separate from both the Sulawesian clade and that containing Callosciurus and its allies, which includes Nannosciurus , the other group of Indomalayan pygmy squirrels. The pattern again supports Moore’s (1958) hypothesis that the two genera of pygmy squirrels are not close phylogenetic allies, and acquired their specializations independently, but not the contention of Heaney (1985) who considered them to be closely related, or Corbet and Hill (1992) who regarded them as likely congeneric.

Because Rubrisciurus has never been adequately defined, we offer the following emended diagnosis. Rubrisciurus is distinguished from the other nannosciurine genera by the following combination of characters (see comparative measurements in table 3, compare skulls illustrated in figs. 6, 12–14, 36, and 37): (1) tree squirrel of large body size (table 3; among Indomalayan genera exceed- ed by only the Bornean Rheithrosciurus and Indomalayan Ratufa ); (2) upperparts of head and back rich brown speckled with orange, buff, and black, surrounded by dark reddish fore legs and feet, shoulders, thighs, and hind legs and feet; (3) underparts bright reddish orange; (4) tail reddish brown and shorter than length of head and body, LT/LHB 5 82%; (5) back of ears densely covered with glossy black hairs that project beyond pinnae margins to form prominent black tufts; (6) two pairs of inguinal teats; (7) cranium and mandible large and robust; (8) rostrum moderately long relative to rest of skull, nasals about as long as frontals and longer than width of interorbital region; (9) postorbital processes of frontals well in front of anterior surface of braincase (‘‘well anterior to the posterior margin of the suborbit,’’ Moore, 1959: 176); (10) prominent temporal ridges that converge to form a long sagittal crest one-third to three-fourths the length of the parietals, and anterior to the occiput; (11) anterior opening of the infraorbital canal lies slightly posterior to the premaxillary-maxillary suture and is concealed behind greatly projecting bony flange of the ventral root of the zygoma, the anteroventral corner of the flange considerably thickened and bearing a roughened outer surface for insertion of the superficial masseter; (12) jugal with high dorsal process, giving the zygomatic arch a robust conformation; (13) orbit short (indicated by lacrimal and posterior margin of zygomatic plate even with second upper molar); (14) posterior border of bony palate terminating anterior to backs of tooth rows, at middle of each third molar; (15) descending palatine vein typically transmitted through a notch (posterior maxillary notch) at the posterolateral margin of bony palate just caudad and slightly medial to end of tooth row (in about 10% of all skulls examined the opening is enclosed by bone forming a foramen); (16) pterygoid fossa long, narrow, and deep, its lateral margin formed by a high ridge ( Moore’s [1959] ‘‘ectopterygoid ridge’’); (17) transbullar septa absent from most specimens, when present consists of half a septum, one septum, or one and one-half-septa (see Moore, 1959); (18) upper incisors strongly proodont (procumbent), emerging from the rostrum at an angle appreciably greater than 90 °; (19) maxillary tooth rows parallel; (20) third upper premolar present, fourth premolar and three molars robust, all noticeably wider (lingual-labial length) than long (anterior-posterior distance). Among the other diagnostic traits Moore (1959: 176) listed were trenchant supraorbital notches and strongly convex lip of infraorbital canal; neither is characteristic of the skulls we examined.

Rubrisciurus shares fusion of temporal ridges anterior to the occiput with Hyosciurus (although the ridges fuse to form a very short sagittal crest in Hyosciurus compared with the much longer crest in Rubrisciurus ). Rubrisciurus and Prosciurillus share presence of ear tufts (but not present in all species of Prosciurillus ), posterior maxillary notch, flange concealing anterior opening of infraorbital canal and position of that opening relative to the premaxillary-maxillary suture, deep pterygoid fossa with prominent lateral pterygoid ridge, postion of posterior border of bony palate relative to end of tooth rows, proodont upper incisors, and wider than long fourth premolar and first two molars.

Our study of skins and skulls reaffirms the presence of a single species in Rubrisciurus , the definition of which we document below, along with its geographic and altitudinal distributions on Sulawesi derived from the collection localities of voucher specimens. Rubrisciurus rubriventer also hosts a morphologically very distinctive species of sucking louse (see the descriptions of the new species of Hoplopleura below).

GAZETTEER AND SPECIMENS EXAMINED: Collection localities for the 92 specimens of R. rubriventer studied are listed below. The number preceding each locality keys to a symbol on the map in figure 5.

1. Manado (also spelled ‘‘Menado’’), 01 ° 309N, 124 ° 509E, coastal plain near sea level: RMNH specimens ‘‘ d–g ’’ in Jentink’s (1888: 23) catalog; ZMA 15.933, 19.832, 19.833.

2. Rurukan, 01 ° 219N, 124 ° 529E: AMNH 196509 (500 m), 101324 (1000 m).

3. Gunung Masarang, 01 ° 199N, 124 ° 519E, 3500 ft (1067 m): BMNH 97.1.2.14.

4. Temboan, on Kuala Kalait, ‘‘a new clearing of eight houses and lies from Mt. Sapoetan south, 55 ° west and about six miles from Loboe,’’ wrote Raven in his field journal, 1916: 3 (in Mammal Division Library at USNM), 01 ° 039N, 124 ° 339E (estimated from Raven’s map), 500 m (estimated from Sheet NA 51-12): USNM 217824, 217905.

5. Tomohon, 01 ° 199N, 124 ° 499E, 700–800 m (estimated from Sheet NA 51-12): BMNH 99.10.1.7; NMB 3329, 1201/9538. Meyer (1899: 22) reported specimens from here.

6. Tulabolo (‘‘Toelabello’’ in Jentink’s catalog, 1888: 23), 00 ° 319N, 123 ° 169E, 760 ft (230 m; see Fooden, 1969: 137, for details): RMNH: specimen ‘‘ c ’’ in Jentink’s (1888: 23) catalog. ‘‘ Modélido,’’ someplace in the Tulabolo-Gorontalo area east of Bumbulan: RMNH: specimen ‘‘ b ’’ in Jentink’s (1888: 23) catalog.

7. Bumbulan, 00 ° 299N, 122 ° 049E, coastal plain near sea level: AMNH 153909–11, 153279; MZB 6257.

8. Gunung Ile-Ile, 00 ° 589N, 121 ° 489E, 500 m: AMNH 101321–23, 196508.

9. Valley of Sungai Miu, Sungai Oha Kecil (small tributary on left side of Sungai Miu), 01 ° 229S, 119 ° 579E (near confluence with Sungai Miu; estimated from Sheet SA 50- 8), 290 m: AMNH 224621, 224622.

10. Sungai Oha Kecil, 1300 ft (396 m): AMNH 224623.

11. Valley of Sungai Miu, Sungai Miu (right side), 01 ° 239S, 119 ° 589E (estimated from Sheet SA 50-8), 350 m: AMNH 224055, 224056.

12. Valley of Sungai Miu, Sungai Sadaunta (also spelled ‘‘Sidaonta’’ or ‘‘Sidaunta’’; tributary on right side of Sungai Miu), 01 ° 239S, 119 ° 589E (estimated from Sheet SA 50-8), 675 m: AMNH 224052–054.

13. Sungai Sadaunta: 2500 ft (762 m), AMNH 224624; 2550 ft (777 m), AMNH 224629; 2600 ft (793 m), AMNH 224625.

14. Sungai Sadaunta: 2700 ft (823 m), AMNH 224626.

15. Sungai Sadaunta: 3050 ft (930 m), AMNH 224628, 226839, 226840.

16. Sungai Sadaunta, 3300 ft (1006 m): AMNH 224627.

17. Valley of Danau Lindu , Tomado (a village on western shore of Danau Lindu), 01 ° 199S, 120 ° 039E (estimated from Sheet SA 5-8), 1000 m: AMNH 223023–025, 223466, 223467, 224057 ; USNM 218711.

18. Gunung Kanino, 01 ° 179S, 120 ° 089E (estimated from Sheet SA 50-8), 4180 ft (1274 m): AMNH 223552.

19. Gunung Kanino, 4600 ft (1402 m), AMNH 225488; 4650 ft (1418 m), AMNH 225489, 225490.

20. Gunung Kanino, 4960 ft (1512 m): AMNH 225491.

21. Malakosa, Kuala Navusu, 00 ° 589S, 120 ° 279E (estimated from Sheet SA 51-1): 100 ft (30 m), AMNH 226052–054; 130 ft (40 m), AMNH 226055, 226056; 150 ft (46 m), AMNH 226058; 200 ft (61 m), AMNH 226057; 400 ft (122 m), AMNH 226059; 500 ft (152 m), AMNH 226060, 226061; 800 ft (244 m), AMNH 226062.

22. Tolai, Sungai Tolewonu, 01 ° 049S, 120 ° 279E (estimated from Sheet SA 50- 8): 500 ft (152 m), AMNH 226526, 226528–31; 540 (165 m), AMNH 226527; 550 ft (168 m), AMNH 226532; 750 ft (229 m), AMNH 226533; 850 ft (259 m), AMNH 226534; 1000 ft (305 m), AMNH 226535.

23. Kulawi, 01 ° 279S, 119 ° 599E, 500 m: USNM 218710.

24. Besoa, 01 ° 449S, 120 ° 139E (for ‘‘Besoa District,’’ HOUSND, 1944): USNM 219522. H.C. Raven, who collected the squirrel in 1917, noted that ‘‘Besoa is a large level plain, undoubtedly a former lake bed, surrounded by mountains, which are covered by heavy forests; the tops of most of the mountains are above 2000 meters, the level plain is said to be about 1300 meters, or perhaps more. The plain is perhaps 2 or 3 miles wide by about 3 miles long and most of the area is covered with several varieties of long coarse grasses and reeds; in several places there are wet rice fields and the natives have made a few clearings on the lower slopes of the mountains. In some places the lower slopes are covered with grass’’ ( Riley, 1924: 3). Raven collected animals from at least two places in the Besoa region (Doda and Gunung Taewo). No elevation or precise provenance is associated with USNM 219522.

25. Pegunungan Quarles, Bulu (Gunung) Karua (referred to as ‘‘Rantekaroa, Quarles Mt.’’ on specimen tags), 02 ° 569S, 119 ° 399E, 4000 ft (1220 m): BMNH 40.674.

26. Palopo, 03 ° 009S, 120 ° 129E, coastal plain near sea level: MZB 6258–60.

27. Wawo, on the plain between the coast and western foothills of Pegunungan Mekongga (also spelled ‘‘Mengkoka’’), 03 ° 419S, 121 ° 029E, 50 m: AMNH 101314–16.

28. Pegunungan Mekongga (also spelled ‘‘Mengkoka’’), Tanke Salokko (the highest spot in Pegunungan Mekongga; see the maps and discussion in Heinrich [1932] and Stresemann [1940]), 03 ° 359S, 121 ° 159E (for the Pegunungan): 600 m, AMNH 101313; 1400 m, AMNH 101312.

29. Pegunungan Mekongga, Masembo, which is southeast of Wawo and the highest place in the southern portion of Pegunungan Mekongga (see maps and discussion in Heinrich [1932] and Stresemann [1940]), 03 ° 359S, 121 ° 159E (for the Pegunungan), 550 m: AMNH 101317–19.

30. Lalolei (spelled ‘‘Lalolis’’ on specimen tags and some maps), 03 ° 579S, 122 ° 039E, 300 m: AMNH 101320.

31. Lamoncong (‘‘Lamontjong’’ is the older spelling), in the upper part of the Walanae River Valley, approximately 05 ° S, 119 ° 509E (see map in Mulvaney and Soejono, 1970: 164, and Bulbeck, 1996: 1014), 300–600 m: subfossil fragment of a skull described by Sarasin (1905: 47) from a limestone cave. We have not seen this specimen, but Sarasin described what can only be an example of Rubrisciurus , and he identified it as ‘‘ Sciurus rubriventer Müll. Schl. ?’’

The following specimens were studied but the places where they were caught have not been mapped.

A. ‘‘Célèbes’’ (see Jentink, 1888): RMNH 13341 (specimen ‘‘ a ’’ in Jentink’s [1888: 23] catalog; holotype of Sciurus rubriventer ); see account of type locality. ‘‘Celebes’’: BMNH 40.675, 43.238.

B. Central Core, Tamalanti: BMNH 40.691a. Laurie and Hill (1954: 156) noted that Tamalanti is a ‘‘Plantation between Rantekaroa [02 ° 509S, 119 ° 509E] and Koelawi [01 ° 279S, 119 ° 599E],’’ which was also the only information Musser and Paula Jenkins found when they searched through fieldnotes and other documents at BMNH. No elevation is provided for the specimen, but most collections from Tamalanti came from 3300– 3800 ft (1006–1159 m) according to Paula Jenkins (personal commun.).

C. ‘‘Tannya’’: FMNH 8328. We could not locate this place.

Rubrisciurus rubriventer (Müller and Schlegel, 1844) Sciurus rubriventer Müller and Schlegel, 1844: 86 .

HOLOTYPE AND TYPE LOCALITY: The holotype of Rubrisciurus rubriventer is an adult male (RMNH 13341, specimen ‘‘ a ’’ in Jentink’s [1888: 23] catalog) obtained by E.A. Forsten sometime during 1840–1842. It consists of a skin mounted in a live position; no skull is present. The skin apparently was in bad condition when prepared; most fur on the abdomen, right flank, caudal part of the left flank, and part of the left thigh is missing.

The type locality is ‘‘ Célèbes,’’ as listed in Jentink’s (1888: 23) catalog. Forsten obtained most of his material in the region near the tip of the northeastern peninsula, which is in the administrative district of Minahasa (Kabupaten Minahasa), between 1 ° and 2 ° north latitude, Propinsi Sulawesi Utara, Indonesia. Laurie and Hill (1954: 94) indicated ‘‘Minahasa, north Celebes’’ to be the type locality, which restricted the location to that political unit. However, the type locality should rather be defined as ‘‘NE Celebes’’ (5 NE Sulawesi) because Forsten also worked in the area of Gorontalo and as far west as Paguat (September–November 1841; see Van Steenis-Kruseman [1950: 179] and Forsten’s unpublished diary in the archives of the Leiden Museum); hence, it cannot be exclud- ed that specimens from that area are among Forsten’s material (C. Smeenk, in litt., 2008). For additional information covering Forsten’s travels in northeastern Sulawesi, see the account of Prosciurillus leucomus .

EMENDED DIAGNOSIS: Rubrisciurus is monotypic so the emended generic diagnosis applies to the species, R. rubriventer .

GEOGRAPHIC AND ALTITUDINAL DISTRI- BUTIONS: Voucher specimens reflect a distribution throughout most of the mainland of Sulawesi; no material has been collected from offshore islands (see gazetteer and map in figure 5). Most collection records are in regions that are or were covered with primary tropical lowland evergreen rain forest. A few samples come from mountains at altitudes where tropical lower montane rain forest dominates (Gunung Kanino in the central core, and Pegunungan Mekongga on the southeast peninsula). No specimens have been collected from high elevations in cool and wet upper montane rain forest.

Provenances for modern specimens cover most of Sulawesi; only the east-central peninsular extension of the central core, which has not been surveyed for small mammals in any adequate endeavor, and the southwestern peninsula lack modern records. Unfortunately, except for some forested limestone tracts and higher slopes of the volcano Gunung Lompobatang, the southwestern peninsula has mostly been deforested; the former forest cover has been converted to farms and tree plantations from about 1700 m to the coastal lowlands ( Fraser and Henson, 1996; see the map showing remaining forest tracts in Froehlich and Supriatna, 1996; and map 8d in MacKinnon, 1997). But, Rubrisciurus did occur there in prehistoric times as documented by a subfossil cranial fragment excavated from a rockshelter at Lamoncong, in the southern part of the peninsula northeast of Ujung Pandang ( Sarasin, 1905). Archaeological artifacts from Sulawesi rockshelters are usually associated with the period of Toalean prehistory, 7000– 5500 years B.P. (see references in Bulbeck, 1996). There are numerous prehistoric cave sites in the southern peninsula ( Bulbeck, 1996) but, with the exception of the Lamoncong site, remains of Rubrisciurus have not been identified in the large samples of subfossil material excavated from some of them (as determined by Musser’s study of subfossil cranial and dental remains).

In most other surveyed geographic regions of Sulawesi, Rubrisciurus rubriventer occurs with a member of the Prosciurillus leucomus group, with P. murinus , and with the two species of ground squirrels, Hyosciurus ileile and H. heinrichi (table 6). Hyosciurus has not been recorded from the southwestern peninsula. A member of the P. leucomus group is represented by a subfossil fragment from a cave deposit in the region (see account of Prosciurillus sp. ). Modern samples of P. murinus have been collected from the southwestern peninsula but only above 1100 m on foothills and higher slopes of Gunung Lompobatang (see gazetteer), which is still clothed by some forest: the upper forested slopes of the volcano form an island ‘‘in a sea of densely-populated agricultural land east of Ujung Pandang’’ ( Whitten et al., 1987: 519). So all three species of tree squirrels are now represented by subfossil or modern specimens from the southwestern arm. Murid rodents (rats) occurring with Rubrisciurus in lowland and lower montane forest formations in other geographic regions of Sulawesi ( Taeromys celebensis , Taeromys punicans , Bunomys chrysocomus , Bunomys andrewsi , Rattus hoffmanni , and Paruromys dominator are examples) either live on the southwestern peninsula now or did in the past ( Musser, 1984).

Elevational distributions gleaned from collection data tied to voucher specimens (see gazetteer) for three regions of Sulawesi are summarized here. On the northern peninsula, records extend from the coastal plain to 1067 m on Gunung Masarang. All places are within the elevational distribution of tropical lowland evergreen rainforest habitats. Specimens obtained in the central core of the island bracket habitats from the coastal plain at 30 m (Kuala Navusu) to those in the mountains (Gunung Kanino at 1512 m), and range from tropical lowland evergreen rain forest to tropical lower montane rain forest; Musser did not trap or see R. rubriventer in the forests above approximately 1500 m. Lower montane forest at 1400 m on Tanke Salokko in Pengunungan Mekongga is the highest collection record for R. rubriventer on the southeastern peninsula of Sulawesi; all other records lie within 50 to 500 m in tropical lowland evergreen rainforest environments.

DESCRIPTION: The original description of rubriventer (as a species of Sciurus ) by Müller and Schlegel (1844: 86) is short (translated from the original in Dutch by Chris Smeenk [in litt., 2008]; the Dutch text is reproduced in appendix 2):

5.) Sciurus rubriventer, Forsten , n. sp. Size, shape and colour, generally as in the preceding species [ Sciurus hippurus , now Sundasciurus hippurus , a large-bodied reddish species native to Malaya, Sumatra and Borneo; see Corbet and Hill, 1992, and Payne et al., 1985]; but the ears larger and much more strongly covered with black hairs, which are very long, hence protrude far above the ears. The brownish-red of the lower parts extends not only over the inner, but also over the outer sides of the legs. Hairs of the tail black, toward the tips with broad, faintly reddish-brown rings. Celebes.

Rubrisciurus rubriventer is the largestbodied of the species of tree squirrels endemic to Sulawesi (length of head and body, 250– 305 mm; length of hind foot, 67–77 mm; weight, 500–860 g; see tables 2, 7) and one of the largest of the tree squirrels and ground squirrels that are native to the Indomalayan region. It is exceeded in body size only by species of Ratufa (weight 5 875–1620 g for Malayan and Bornean R. bicolor and R. affinis ; see Medway, 1969: 56, and Payne et al., 1985: 233) and the Bornean endemic, Rheithrosciurus macrotis (weight 5 1170– 1280 g; see Payne et al., 1985: 243).

The color pattern, resplendent in its richly pigmented red and orange hues, of the large-bodied and long-legged R. rubrisciurus is best described this way: a reddish brown tail; reddish underparts, forelegs and feet, shoulders, thighs, hind legs and feet; brownish head and back speckled by buff, orange and black; ears highlighted by prominent glossy black tufts. The short fur (6–8 mm thick) covering the entire underparts, from chin to base of tail, is a rich, bright, reddish orange, which darkens to reddish maroon over dorsal surfaces of the front legs and feet, shoulders, along sides of the body and thighs, and dorsal surfaces of hind legs and feet. The top and sides of the head, and rest of the dorsum from the head to base of the tail, are covered by a thick (15–20 mm), rich brown coat flecked with buff, orange, and black (a

TABLE 7 Descriptive Statistics for Measurements (mm) of Lengths of Head and Body, Tail, Hind Foot, and Ear, and for Weight (g), Derived from Samples of Rubrisciurus rubriventer 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

combination provided by the overhairs with dark gray bases, subterminal buff or orange bands, and black tips, along with black guard hairs). There is a dark crescent above each eye, and the eye is circled by a wide buffy ring. The ears conspicuously contrast with the head, neck, back, and rump because the medial surface of each pinna is covered by soft black hairs that converge to form a tuft 5–10 mm long projecting beyond the contour of the dorsal pinna margin, and rimming the posterior margin (top and back edges of the pinna are defined by the black hairs, but the long tuft is confined to the dorsal margin). Color and extent of this tufting is closely similar to that present in P. leucomus . The tail of R. rubriventer is shorter than length of the head and body (LT/LHB 5 82%), both dorsal and ventral surfaces are dark reddish brown (approaching maroon in some individuals), without the darker or paler borders formed by distal banding patterns of the hairs that are common to species of Prosciurillus .

The juvenile coat exhibits a color pattern similar to that of adults but the fur is thinner (hairs are finer), shorter, and the coloration typically duller, the red tones muted instead of being rich and bright.

Females have four teats arranged in two inguinal pairs. A single embryo was found in each of the few females of Rubrisciurus rubriventer caught by Musser.

Views of the large, stocky skull of R. rubriventer are illustrated in figure 6. Note the robust and relatively long rostrum, prominent anterior projection of the zygomatic root concealing the anterior opening of the infraorbital canal, high dorsal process of the jugal component of the zygoma, short orbit, position of the postorbital process well in front of the anterior margin of the braincase, prominent temporal ridges that unite to form a strong sagittal crest, posterior margin of the bony palate situated anterior to the backs of the parallel tooth rows, proodont (procumbent) upper incisors, wide fourth premolar and molars, and sturdy dentary with its high ramus.

Another trait bears inspection. At each posterolateral margin of the bony palate just caudal and slightly medial to the end

TABLE 8 Descriptive Statistics for Cranial and Dental Measurements (mm) Derived from Separate Population Samples of Rubrisciurus rubriventer Mean ± 1 SD and observed range (in parentheses) are listed.

of the tooth row is a large notch or enclosed foramen (the posterior maxillary notch or foramen) that transmits the descending palatine vein ( Wahlert, 1974). The vein passes through a notch in most specimens of R. rubriventer (ends of the notch are connected by cartilage, which is usually removed when the skull is cleaned), which is also typical in all the species of Prosciurillus . In a few examples of R. rubriventer , the opening is enclosed by bone forming a foramen (as shown in fig. 6), which is the typical configuration in both species of Hyosciurus .

Some of the other cranial traits seen in R. rubriventer are common to species of Prosciurillus , Hyosciurus , and most, but not all, other sciurids. Some examples are the proximal borders of the nasals, which lie anterior to the bordering premaxillary-maxillary sutures; moderate to prominent postorbital processes of the frontals; posterior border of the ventral zygomatic root at about the level of the second upper premolar; short and typically narrow incisive foramina contained entirely in the premaxillaries; a bony palate with a smooth ventral surface and penetrated by small postpalatal foramina at about the level of the second upper molars; wide mesopterygoid fossa relative to skull size, its roof entirely osseous or showing slitlike sphenopalatine vacuities; relatively narrow pterygoid fossae bounded by a high lateral ridge; well developed hamular processes; moderately large ectotympanic bullae tightly attached to the basicranium; spacious postglenoid foramen just dorsad of the ectotympanic capsule; moderate-size coronoid process relative to overall outline of the dentary, large condyloid and angular processes, and deeply concave posterior border of the dentary.

Cranial and dental measurements for the population samples of R. rubriventer are summarized in table 8.

COMPARISONS: Members of the Prosciurillus leucomus and P. murinus groups as well as the species of Hyosciurus are all much smaller in body size than is Rubrisciurus rubriventer (table 3), none displays its bold and bright reddish, orange, brownish, and black patterning, and all females show three pairs of teats (see the appropriate accounts of species).

GEOGRAPHIC VARIATION: No additional scientific names have been attached to geographic samples of R. rubriventer , most likely a reflection of its generally uniform attributes associated with body size, pelage coloration and patterning, and cranial and dental dimensions among samples from the northern peninsula of Sulawesi, its central core, and southeastern arm. Sody (1949: 108) came close to naming a subspecies when he discussed the samples at MZB that he studied: a single specimen from Bumbulan on the northern peninsula and three from Palopo in the southern sector of the central core of Sulawesi (see gazetteer and fig. 5). Sody wrote that,

It is only after much hesitation that I decided not to describe the Palopo animals as a new race. They are differing very strikingly from our Bumbulan specimen. Both forms have the same fulvous underside, arms and hands, legs and feet. But in the Bumbulan specimen the colour of the upperside is a mixture of black and fulvous-buff, in the Palopo series a mixture of brownish and buff. The general difference in colour is very large. And perhaps there also is a difference in the measurements. The reason why I do not separate the Palopo series, is, that it consists of old specimens (perhaps discoloured?), whereas the Bumbulan specimen is from formaline.

Musser examined all four squirrels. The color contrast noted by Sody reflects primarily biological age of the specimens. The squirrel from Bumbulan is an adult covered in a full, richly pigmented adult coat. Of the three from Palopo, two are young and were captured as they were molting from juvenile into adult pelage, and the third is a juvenile completely covered in juvenile fur. Juveniles have thinner and shorter pelage showing slightly duller reddish hues along sides of the body, and duller fur over the head and back (the ‘‘mixture of brownish and buff’’) in some individuals, a conspicuous contrast with the thicker adult coat with its brilliant, darker hues.

Even among adults within a single geographic sample, or even in a sample from the same collection site, the fur over the head and back varies in color intensity among individuals. Some squirrels show a dark background speckled with buff (the buffy bands of the hairs are narrow, the black bands wider), but others have brighter fur in which the buffy speckling predominates (the buffy bands are wider, the black narrower); the brilliant reddish hues of the appendages, shoulders and thighs, and underparts are constant within and among population samples. Wear is also reflected in samples. Some older adults with worn pelage have darker upperparts because the fur has worn past the buffy or ochraceous distal bands and the remaining black bands of the hairs provide the prevalent tone.

All samples of R. rubriventer , from wherever they were collected on Sulawesi, exhibit similar pelage coloration and patterning over the body, whether adults or juveniles.

The possible difference in measurements suggested by Sody (1949: 108) can also be attributed to age. The adult from Bumbulan has a larger skull (greatest length 5 68.5 mm) compared with the smaller skulls of the three younger squirrels from Palopo (greatest lengths are 64.0, 64.6, and 65.3 mm). The length of the skull of the Bumbulan animal falls within the range of variation we record within each of our geographic samples (see ONL in table 8). That measurement compatibility reflects the general geographic uniformity of coat color and pattern among our samples of adult squirrels as well as overall size of skull and length of maxillary tooth rows. This cranial and dental dimensional variation within and among geographic samples is summarized in figure 7 where specimen scores representing five population samples of R. rubriventer (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 generally older adults spreading to the right, the smaller and usually younger adults towards the left. Covariation in most cranial variables—particularly length and breadth of rostrum, length of nasals, and lengths of diastema and bony palate—are the most influential in pushing scores along the first principal component; zygomatic breadth, mastoid breadth, and length of maxillary tooth row (PM4–M3) provide negligible impact (table 9).

No significant variation among measured cranial and dental dimensions that reflects altitudinal variation, particularly between lowland tropical evergreen rainforest habitats and lower montane forest, is apparent among our samples. Two sets of specimens provide examples. Scores shown in figure 7 representing specimens collected along Musser’s transect in the central core of Sulawesi from Sungai Oha Kecil at 290 m to Gunung Kanino at 1418 m are identified by filled circles; slim arrows point to scores for the four squirrels collected in lower montane forest between 1274 and 1418 m on Gunung Kanino. Not only do most congregate within the cloud of points representing specimens from tropical lowland evergreen rain forest between Sungai Oha Kecil at 290 m and Tomado at 1000 m, but also within lowland samples from Kuala Navusu and Sungai Tolewonu (30–305 m; identified by asterisks) to the east of Gunung Kanino, and Masembo at 550 m (filled squares) in the southeastern peninsula. The other set originates from the southeastern peninsula and their specimen scores are indicated by filled squares. The square identified as Wawo at 50 m in tropical lowland evergreen rain forest rests next to the square for the specimen from Tanke Salokko at 1400 m in lower montane forest.

No discrete clouds of scores are present in the principal-components ordination that represents separate geographic regions. There are no unique constellations, for example, identifying samples from the northern peninsula, from the central core, or from the southeastern peninsula; all points derived from geographic samples are intermixed. Because each large sample ranges in age from young adults to old adults and includes males and females, the distribution of scores is also a reflection of age, possibly secondary sexual variation, and individual variation within an age class that is unrelated to age.

Four points at the right margin of the cloud in figure 7 require discussion. AMNH 224623, which is an especially large adult from Sungai Oha Kecil, is responsible for the score located farthest to the right in the ordination; most scores for other individuals from that collection site are nested within the primary cloud. Just to the left of AMNH 224623 is another point for a large adult from Sungai Oha Kecil, and that score is aligned with two scores identifying our only measured skulls from the northeastern limb of the northern peninsula. These two are also large adults, but we do not know if they are typical of the population in the northeast. We record more than a dozen specimens from that region (see gazetteer) but some are juveniles, some consist only of skins, and others have damaged skulls. A larger sample of adults with intact skulls needs to be studied to assess the range of cranial and dental morphometric variation within the northeastern peninsular population.

Our analysis of geographic variation in cranial and dental dimensions among samples available to us is rough. The results presented here should be tested by comparing larger samples of comparable age, supplemented by molecular data. What we convey here is the absence of significant differences in color pattern of the pelage, along with cranial and dental dimensions, among available samples of R. rubriventer from the different major geographic regions of Sulawesi. This large red squirrel looks the same wherever it is encountered on Sulawesi, a startling contrast to the variation associated with fur color and patterns among geographic samples of the Prosciurillus leucomus complex (see those accounts).

ECOLOGY: Rubrisciurus rubriventer is diurnal, arboreal, and terrestrial. It nests in the forest understory and forages in the understory canopy and on the ground; Musser never saw any in the canopy layer or higher in the crowns of emergent trees. When squirrels were encountered, they were traveling through crowns forming the understory canopy or along the ground. Of the two squirrels briefly seen by Durden in the northeastern peninsula, one was in an understory tree and the other was on the ground. Musser trapped all his specimens on rotting tree trunks (‘‘boles’’ to the botanists) lying on the forest floor, on tree or palm trunks lying across streams, on limbs of understory trees, and on the ground in scrub adjacent to tall forest (see summary of habitats at trapping sites in table 10, and the forest in fig. 8). Except for the specimen trapped in scrub, which was near tall forest, all sightings and

TABLE 9 Results of Principal-Components Analysis Contrasting Population Samples of Rubrisciurus rubriventer Principal components are extracted from a covariance matrix of log-transformed values for 15 cranial and 1 dental variable; see figure 7.

trapping records of R. rubriventer were in primary forest habitats on stream terraces, hillsides, and ridgetops.

From trapping records, the species is far more common than sightings would suggest— Rubrisciurus is quiet, wary, and easily startled. Although he spent every day in the forest checking traps and collecting botanical samples, Musser seldom saw Rubrisciurus ; when he did, it was usually only as a reddish streak disappearing into the undergrowth or through the understory crowns. Once he saw two near camp at the Sungai Tokararu, one knocked over a live trap, the other scampered along a trunk on the ground, jumped off and poked around, then dashed into the undergrowth.

Two other sightings illustrate the squirrels’ understory habits. On a hillside above the Sungai Tokararu, Musser had rebaited a rat trap and afterwards was sitting quietly on the dry leaf cover listening to sounds in the forest. He heard something upslope and after a few seconds a Rubrisciurus passed about 5 ft from him, progressing downhill in slow bounds alternating with an ambling walk. It held its tail out behind its back in a low arc. The squirrel did not react to Musser’s presence, ambled by, looked around, then bounded down the slope over the litter and through the undergrowth until it was out of sight.

Musser enjoyed another rare encounter, this time in oak-chestnut forest on Gunung Kanino at 1500 m. He was sitting on a tree trunk and heard a squirrel move the dry leaf litter along the ridge, ambling and bounding along past him about 5 ft away. It bounced to a nearby chestnut, climbed over the roots and up one of the trunks to about 4 ft from the ground, straddled the trunk with its head down and its tail stretched out behind, looked downslope, uttered a few low growls, jumped to the ground, then bounded over the ridge and disappeared into the forest.

Rubrisciurus nests in tree cavities not far from the ground. Usma, one of Musser’s helpers, described how he had once helped locate a nest. The villagers where he lived regularly used dogs to find the squirrels in the forest and then chase them to their nest. During one hunt, Usma located a nest in an old but still growing pohon torode ( Pterospermum celebicum ), with a trunk about 2.5 ft in diameter near the base. The base was hollow, forming an inverted cone-shaped cavity inside the trunk extending from the roots up for about 10 ft. The cavity was accessed by the squirrel through a hole about 4 ft from the ground. The original entrance to the cavity had been gnawed around the margins to form a larger opening. Wedged in the cavity between the opening and the roots was a large globular nest, about 1 ft in diameter with a single entrance, constructed entirely of the long, black sturdy fibers from a sugar palm ( Arenga pinnata ). The conformation of this nest, its construction material, and its placement, according to Usma and the few others we spoke to who were familiar with the squirrel’s habits, were typical—large globular nests not far from the ground in cavities in the trunks of large trees.

In tropical lowland evergreen forest between the coast and about 1000 m, Pterospermum celebicum is common. Older trees contribute to the canopy; younger trees and saplings are familiar components of the understory, as well as in more open places in the forest. Musser never saw a tree with a basal diameter greater than 3 ft, and most of those that large had already died and fallen to the ground. It grows along stream terraces, hillsides, and ridgetops. The sugar palm is found in the same forests all the way up to the lower limits of lower montane forest. Probably any large canopy-forming species that develop spacious cavities within the basal portion of the trunk could be used as a nest site. Strangler fig emergents in which the host tree has nearly or completely decayed are also likely candidates for protected nesting sites.

The diet of Rubrisciurus rubriventer consists of fruits, seeds, and insects. Remains of a variety of fruits and seeds were found in stomachs (tables 10, 57). We could identify figs (stomachs would be distended with chopped figs, pulpy endosperm, and small seeds); tan mash, possibly from pangi fruit; seeds from the monocot vine, Rhaphidophora sp. ; fruit (the pulp) from trees in the Sapotaceae ; seeds from the understory tree pohon pangi ( Pangium edule ); fruit from pohon dongi ( Dillenia serrata ), another understory tree; seeds of the ginger, Etlingera celebica ; seeds from the magnolia pohon uru ( Elmerrillia ovalis ); and fruit of Pandanus spp. Caches of acorns from Lithocarpus could be seen in places throughout the forest but we never found anything in the stomachs of Rubrisciurus resembling acorn mash, and Musser never saw the red squirrel eating acorns that had fallen to the ground or were still attached to branches. Still, Lithocarpus fruit cannot be discounted as part of the diet.

Four kinds of fruit are regularly consumed by Rubrisciurus . Many species of fig form components of the forest habitats within the altitudinal range of the squirrel. Several species are limited to the understory, especially along banks of streams where ripe figs fall from the trees and litter the ground. Fruiting tall canopy and emergent species of strangler and other figs attract birds, canopy tree squirrels ( Prosciurillus ), and macaques during the day and fruit bats ( Pteropidae ) at night. The fruits not eaten in the crown of the tree eventually drop to the ground where they are accessible to Rubrisciurus , nocturnal rats, and other terrestrial mammals.

Pangium edule is common in the understory of forests along river terraces and hillsides in tropical lowland evergreen rain forest from the coastal lowlands to about 1100 m. The fruit is large (about 6 in. long and 3–5 in. wide) and woody, containing large and very hard seeds. Near camp on the Sungai Oha Kecil, Musser watched a Rubrisciurus climb a pangi, jump to a branch bearing fruit, bite the stem so the fruit crashed to the ground, then ran down the tree to disappear into the undergrowth because the squirrel had seen Musser. Remains of these large opened seeds are scattered over the stream terraces; here and there are caches beneath tree roots and rocks. The sound of the loud gnawing on the tough seed by the red squirrel travels through the forest, and often provides the only indication of the squirrel’s presence.

The ginger is scattered through the lowland forests where it is most commonly seen in openings where secondary growth begins to cover old treefalls. When mature, the leafy stalks may be 12 ft high and the fruit forms huge clumps close to the ground—easily accessible to Rubrisciurus . Rhaphidophora sp. , the monocot vine, is common in lowland rain forest. The fruit (5 X 1 in.) is composed of small seeds and an outside hard polygonal portion attached to a filamentous white woody base. Along streamside forest near Tomado, Musser startled a Rubrisciurus feeding on fruit from this vine, and partially eaten fruits were found at many places along the transect line.

With few exceptions, a predictable suite of insects was found in stomachs of many red squirrels: numerous remains of soldier and worker termites (head capsules, often still attached to partially digested bodies, bodies and legs, and sometimes intact termites), all representing Rhinotermitidae (Isoptera) ; legless larvae of beetles, Coleoptera (usually without head capsules, apparently the squirrel removes the head and ingests only the thorax and abdomen of the larva), which are common but not as abundant as the termites; and pupae of carpenter ants, occasionally with an adult ant ( Formicidae , Camponotus ). Some stomachs were packed with this combination of insect remains, always with the termites being the most abundant; other stomachs were distended with remains of the insect suite mixed with fruits and seeds. Rhinotermitid termites, legless beetle larvae and those with tiny and short legs, and pupae

TABLE 10

Summary of Habitat at Trapping Sites, Stomach Contents, and other Relevant Information for Specimens of Rubrisciurus rubriventer 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). Six of the collection localities (Sungai Oha Kecil, Sungai Miu, Sungai Sadaunta, Tomado, Kuala Navusu, and Sungai Tolewonu) are in tropical lowland evergreen rain forest; lower montane rain forest describes the places on Gunung Kanino. With two exceptions, 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 Oha Kecil 290 Aug. 1974 On small tree trunk over part of stream; caught early in the

224621 (1758) morning.

224622 (1789) 290 Aug. 1974 On trunk of understory tree growing over stream near spot where Rubrisciurus ASE 1758 was caught.

224623 (1893) 396 Aug. 1974 On small, smooth, wet and rotting trunk lying across stream about 1 ft from surface of water; sides of ravine above stream are steep, the terraces high above the stream.

Sungai Miu 350 Apr. 1974 On wet, bare, rotting trunk lying across stream in wet

224055 (1531) streamside forest; caught between 6:00 and 7:00 a.m.

224056 (1534) 350 Apr. 1974 On wet, rotting trunk lying on wet terrace in dense shrubby undergrowth next to stream.

Sungai Sadaunta 675 Feb. 1974 On decaying trunk lying in streamside forest and extending

224052 (1336) across Sungai Sadaunta to the opposite bank; caught about 8:00 a.m.

224041 (1414) 675 Feb. 1974 Five ft above ground on top of huge rotting trunk of giant canopy tree lying in dense understory of shrubs and ferns, crisscrossed by woody vines, in primary forest adjacent to stream. The tree squirrel Prosciurillus murinus and rat Taeromys celebensis were caught in same spot.

224054 (1459) 675 Mar. 1974 On rotting trunk lying in undergrowth of streamside forest.

224624 (1995) 762 Sept. 1974 On moss and fern-covered trunk lying across a dry ravine in hillside forest near camp; caught early in the morning.

Stomach: empty.

224625 (2006) 793 Sept.1974 On small trunk lying across a ravine in tall hillside forest.

Stomach: distended with remains of figs in which are mixed large and hard ginger seeds, numerous remains of worker and soldier rhinotermitid termites (head capsules, bodies, and intact termites), a few legless beetle larvae, an adult carpenter ant and a few carpenter ant pupae.

224626 (2033) 823 Sept. 1974 On wet, mossy trunk extending from one terrace of

Sungai Sadaunta to the opposite terrace and resting just above water level on rocks in the stream; trunk would be inundated in high water. The rats Maxomys hellwaldii and Bunomys sp. were caught in same spot on different nights.

Stomach: empty.

224627 (2350) 1006 Nov. 1974 On wet, moss-covered tree root growing across tributary creek in steep ravine. Caught rats ( Taeromys celebensis , Rattus marmosurus , Paruromys dominator and Bunomys sp. ) and Prosciurillus murinus just downstream on rotten trunk lying across stream.

TABLE 10 (Continued)

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

224628 (2354) 915 Nov. 1974 On limb of understory tree growing across ravine containing main upper tributary of the Sungai Sadaunta; base of the trunk is at edge of stream, with the trunk leaning over the water and its upper branches reclining on the opposite terrace. The main connecting limb (3–5 ft in. diameter) on which the trap was placed is mossy, and the moss is trampled, indicating frequent use; trap set about 7 ft from stream surface. The ground squirrel Hyosciurus ileile , the tree squirrels Prosciurillus topapuensis and P. murinus , and the arboreal rat Rattus marmosurus were taken in the same spot on different days. So three kinds of tree squirrels, one ground squirrel and at least one species of arboreal rat used the same living pathway as a bridge over the stream, the squirrels during the day, the rat at night.

224629 (2383) 777 Nov. 1974 On rotting, moss-covered trunk, partially covered by vines and ferns, lying across Sungai Sadaunta upstream from camp. A Prosciurillus murinus and the rat Maxomys musschenbroeki were trapped in same spot.

226839 (4318) 915 Mar. 1976 Caught between 5:30 and 6:00 a.m. at same spot on limb of understory tree where Rubrisciurus ASE 2354 and other rodents were trapped in 1974. Stomach: empty.

226840 (4388) 930 Mar. 1976 Caught during afternoon on large (2–3 ft diameter, 20 ft long) rotting trunk lying from high terrace, 10 ft above water, down across stream at an angle to other bank. Trunk is densely covered with ferns, monocot shrubs and smaller plants that form good cover for rodents using the trunk as a bridge. The rat Echiothrix centrosa was trapped in same spot. Stomach: partially full, remains of several kinds of fruit (skin, endosperm, seeds) including white, fibrous remains of seeds from the monocot vine Rhaphidophora sp.

Tomado 1000 Aug. 1973 On rotting trunk lying across narrow stream in tall slightly

223466 (357) disturbed primary forest. Stomach: partially full of mash from a soft fruit, ginger seeds and surrounding tissue, remains of fruit from the monocot vine Rhaphidophora sp , and a few remains of rhinotermitid soldier and worker termites.

223024 (358) 1000 Aug. 1973 On rotting trunk lying across stream in tall slightly disturbed forest; Rubrisciurus ASE 357 was trapped nearby.

223467 (378) 1000 Aug. 1973 Live trap on ground in dense shrubs and ferns at edge of meadow and second-growth forest behind rest house.

Stomach: partially full of bait.

223025 (523) 1000 Aug. 1973 On large, rotting moss-covered trunk lying on ground deep in primary forest.

224057 (1608) 1000 May 1974 On rotten trunk lying across stream; scrub covers one bank, the opposite terrace supports tall forest. The rats Maxomys hellwaldii and Rattus hoffmanni were caught on same trunk.

Gunung Kanino 1274 Nov. 1973 On living tree limb 3 ft from ground in dense, scrubby

223552 (814) understory forest; caught in morning.

225488 (2520) 1402 Feb. 1975 On top of rotting trunk lying across narrow ravine; caught early in morning. Stomach: full of red fruit with small black seeds from Pandanus ; still fresh, probably eaten just before caught.

TABLE 10 (Continued)

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

225489 (2533) 1418 Feb. 1975 On very mossy, rotting trunk lying across small tributary of Sungai Salubeka; the smaller stream courses along bottom of shallow ravine bounded by very wet and muddy slopes. The rat Bunomys penitus was taken at same spot. Stomach: remains of a large adult beetle; a bit of tan fruit; most of contents consist of a fibrous, bright lemon yellow fruit with small white seeds emitting a pungent, sweet, citrus-like odor (from a woody vine).

225490 (2462) 1418 Jan. 1975 On a pile of rotting trunk and limbs lying across small stream below camp. Two Prosciurillus murinus were trapped on same spot on different days. Stomach: full of pieces of white, orange and tan fruits; small hard white seeds; larger brown stony seeds; also bits of insect parts.

225491 (2476) 1512 Jan. 1975 Live-trap on ground placed within hollow base of rotten, moss-covered chestnut stump on ridge. The rat Taeromys sp. was taken in the same trap.

Kuala Navusu 31 Sept. 1975 On trunk (8 in. diameter) lying across ravine (5 ft wide)

226052 (3104) cutting down hillside behind camp. Stomach: full of mostly fruit and seeds (fig seeds from large understory fig, pieces of bright yellow fruit, and some triangular hard seeds and surrounding pulp), remains of large legless beetle larvae, and large margarodid scale insects, rhinotermitid worker and soldier termites (head capsules, some attached to partially digested bodies), and a few carpenter ant pupae; mostly the same combination of insects found in other Rubrisciurus .

226053 (3124) 31 Sept. 1975 On a decomposing palm trunk lying across stream near end of trapline. High, dense scrub covers stream banks in hillside forest partially thinned of canopy trees. A Prosciurillus murinus and the rat Maxomys hellwaldii were trapped at same spot. Stomach: partially full of pinkish tan mash with small cuboidal seeds from woody vine fruit, some remains of rhinotermitid termites, and many pupae of carpenter ants.

226054 (3141) 31 Sept. 1975 On rotting trunk (8 in. diameter) lying across stream below camp in old logging area, scrub and some secondary forest on either side of stream; caught another Rubrisciurus on a trunk lying across same stream about 100 ft downstream from where ASE 3141 was caught. Stomach: full of pink fruit pulp, small and hard cuboidal brown seeds; and long yellow seeds with orange tips from the monocot vine Rhaphidophora sp.

226055 (3177) 40 Sept. 1975 On rotting trunk (8 in. diameter) lying across steep tributary

226056 (3229) ravine; caught early in morning. One tree squirrel, Prosciurillus alstoni , was caught at the same spot on the previous day. Stomach of ASE 3177: full of fruit and seeds (brown fruit mash, some small cuboidal seeds, mostly remains of bright yellow fruit from woody vine) and purplish legless larval beetle abdomens. Stomach of ASE 3229: full of bright lemon-yellow fruit pulp (from woody vine).

TABLE 10 (Continued)

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

226058 (3373) 61 Oct. 1975 On trunk (4 in. diameter) of living tree growing horizontally over a ravine; trunk is 5 ft above ravine bottom and extends 10 ft from one side of ravine to the other. A Prosciurillus murinus was taken in same place during a different day. Stomach: partially full of seeds from village ginger, unidentifiable fruit tissue, a few flat oblong seeds, many large and small legless beetle larvae, numerous remains of rhinotermitid worker and soldier termites, pieces of an adult beetle, including wings.

226057 (3555) 46 Nov. 1975 On branch of small, shrubby streamside tree growing across tributary draining west-facing slope, just up a few feet from main stream. Here ravine is 5 ft wide and 6–8 ft deep; good hillside forest.

226059 (3410) 122 Oct. 1975 Caught in morning on trunk (10 in. diameter) growing horizontal in understory; from base of slope the tree extends 3 ft over ground and across deep (15 ft) rocky ravine where crown ends about 10 ft off ground on other side of ravine; thick understory on both sides of ravine in tall hillside forest. Three Prosciurillus alstoni were taken in same trap on previous days. Stomach: nearly full with some tan fruit mash, but otherwise packed with legless coleopteran larval abdomens (30–50 mm long), rhinotermitid termite workers and soldiers, and a few carpenter ant pupae.

226060 (3617) 152 Nov. 1975 On long section of rotting trunk (10 in. diameter, 50 ft long) lying on sloping terrace above headwaters of main stream. Trunk lays about 2 ft off ground for most of its length. Intact tall hillside forest. Stomach: distended with Sapotaceae fruit; squirrel eats the pulp and green outer skin, but not the seed; could not find any trace of seeds, only pulp and skin. There is a large Nantu dropping fruit on east terrace and a tall Madhuca on west terrace, likely one of these is the source of fruit.

226061 (3655) 152 Nov. 1975 On rotting limb (8 in. wide, 8 ft long) that is extension of a long and large rotting trunk laying down the slope of the hill above a ravine. Tall primary hillside forest blankets area. Stomach: partially full of mostly one kind of fig (hard tiny black seeds, brown skin, thick rind) mixed with a few large (35–45 mm long) and small (5–10 mm) abdomens from legless beetle larvae, and some rhinotermitid worker and soldier termites.

226062 (3439) 244 Oct. 1975 On slender rotting trunk (diameter 6 in.) lying on steep dry slope. One side rests on ground; about 30 ft away, the other end rests on rotting trunk of a canopy tree that is laying down the slope. Forest is open here and adjacent understory is thin; a very old tree-fall cleared a large section that is only slowly being reforested. Stomach: partially full of bright yellow fruit mash.

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

226532 (4016) old tree fall that is covered with a climbing ginger-like monocot, shrubs, ferns, and palm rosettes; about 6 ft above surface of water; caught in morning. A ground squirrel, Hyosciurus ileile , and a rat, Maxomys hellwaldii , were taken on same spot. Stomach: filled with mostly remains of fig (seeds and rind) mixed with several large and hard red seeds, a scattering of worker and soldier rhinotermitid termites, several legless beetle larva and buprestid beetle larvae, several large margarodid scale insects, and one macrolepidopteran caterpillar.

TABLE 10 (Continued)

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

226534 (4017) 259 Jan. 1976 On wet, rotting, moss-covered trunk (40 ft long, 2 ft diameter) that is splintered into large sections and lying across stream and onto terrace and hillside; hillside forest of dense understory with scattered canopy trees. Stomach: nearly empty, small sections of fig and another kind of fruit mixed with chewed segments of legless coleopteran larval abdomens.

226535 (4039) 305 Jan. 1976 On decomposing trunk lying across branch of ravine in area of headwaters. The rat Paruromys dominator was caught in same place. Stomach: full of mostly pinkish-tan fruit mash and some bait.

226533 (4055) 229 Jan. 1976 On a 10-ft section of rotting trunk (10–12 in. diameter) covered with moss and resting across half of streambed 3 ft above water surface. Streambed is about 20 ft wide here between steep forested slopes and just at the top of the section of stream that is all rock. Caught about 8:00 a.m. Stomach: empty.

226526 (4154) 152 Feb. 1976 On rotting trunk of canopy tree (2 ft diameter, 50 ft long)

226528 (4189) bridging the main river about midway between first and

226539 (4220) second large tributaries. Trunk rests 4–7 ft above water level

226530 (4257) and is covered with thick moss on two sides and supports a few

226531 (4259) small shrubs, ferns, and palm seedlings scattered along the top of its length—not enough that provides any decent cover. Along the middle of trunk is a moss-free runway, about 5 in. wide that extends entire length (apparently a path frequently used by rodents to cross the river). On one side of the river is terrace forest, on the other side is steep, rocky moss-covered rocky slope that gives way above to hill forest on steep slopes. A Prosciurillus murinus and the rat Paruromys dominator were trapped at same spot. Rubrisciurus ASE 4220 was trapped in morning. Stomach of ASE 4154: empty. Stomach of ASE 4189: filled with same fruit mash as found in Rubrisciurus ASE 4158, along with small black seeds (probably fig), tiny brown seeds, large and hard oblong orange seeds, white fibrous seeds from the monocot vine Rhaphidophora sp. , but no ginger seeds; a few small pieces of legless beetle larvae, numerous remains of rhinotermitid termite workers and soldiers, and a few carpenter ant pupae—same insect mixture found in other Rubrisciurus stomachs. Stomach of ASE 4220: half full with seeds from village-type ginger mixed with numerous rhinotermitid worker and soldier termites. Stomach of ASE 4257: half full with purple and tan fruit mash (with tiny hard black seeds and some large hard flat, oblong seeds), but most of contents consists of rhinotermitid termite workers and solders, and a few legless larval beetle abdomens. Stomach of ASE 4259: partially filled with hard brown seeds from village-type ginger in which the fruit is borne on stalks near ground— seeds are smaller than village ginger; also watery pulp in stomach is yellowish green.

TABLE 10 (Continued)

of carpenter ants are part of a community found only in dead and rotting wood, especially pieces partially covered by leaf litter and soil (D. Grimaldi, personal commun., 2009). Musser often encountered deeply excavated sections of wet and rotting tree trunks and limbs lying on the forest floor as well as the erect bases of rotting tree and palm trunks. He dug into several undisturbed sections and found adult Coleoptera , legless beetle larvae, and sometimes termites. Rubrisciurus rubriventer travels and forages on the ground as well as in the understory canopy; on the ground, it clearly digs into rotting wood and exposes the insects.

The squirrel must also pick insects off leaf litter, tree trunks, and limbs in the understory canopy. One stomach contained a large adult beetle mixed with fruit mash. Another held buprestid beetle larvae, a macrolepidopteran caterpillar, and large margarodid scale insects mixed with the usual combination of termites and legless beetle larvae. Adult beetles are on the ground and in the trees; buprestid larvae are found beneath rotting bark, both on trunks and limbs decaying on the forest floor and on lower limbs and trunks of living trees; the scale insects occur on the surface of the bark, usually on the trunk or in tree crowns, some occur on wood on or near the ground.

The skull of Rubrisciurus is large and sturdy, with strong procumbent incisors, a long rostrum relative to size of the skull, only slight cranial flexion, prominent temporal ridges that combine to form a pronounced sagittal crest reflecting significant temporal musculature, a large and sturdy platform for the insertion of the superficial masseter, and a large, robust mandible (see fig. 6). That combination of traits among Sulawesi’s endemic tree squirrels— Rubrisciurus and species of Prosciurillus (compare the skull of Rubrisciurus in fig. 6 with those of Prosciurillus depicted in figs. 12–14)—is restricted to Rubrisciurus . This morphology is well adapt- ed to gnawing open large and tough woody nuts, such as those from pangi. It is also structurally suitable for excavating rotten tree limbs, trunks, and stumps. The wood is soft and pulpy and no special adaptations to claws would be necessary for the job. But in addition to front claws, the squirrel may depend on its large and strong incisors, especially the procumbent uppers (see fig. 6) to dig into the decaying wood. Also, the rostrum, relatively long compared with the relatively very short rostrum of the other species of Sulawesi tree squirrels, may contain a greater area of nasal epithelium providing the enhanced olfactory acuity necessary for detecting insects in decaying wood. Musser did not see a red squirrel actually digging into decaying parts of old treefalls, but did surprise one squirrel that had been sniffing about a small rotting tree trunk on the ground.

In contrast to Rubrisciurus rubriventer , the other Sulawesi tree squirrels— Prosciurillus alstoni , P. topapuensis , and P. murinus (and the other members of the P. leucomus and P. murinus groups)—have a very short rostrum compared to skull length, strong cranial flection, faint temporal ridging, and no sagittal crests; and also the anteroventral corner of the ventral zygomatic root is relatively thinner and the outer osseous surface only slightly roughened to accommodate the insertion of the superficial masseter (see figs. 12–14). The morphology suggests a diet of softer foods than those described for Rubrisciurus , and that locating certain dietary elements by olfaction may not be as important in these squirrels as in Rubrisciurus . All three species of Prosciurillus introduced above consume soft fruits (figs are an example) as well as insects (table 57). Contents of stomachs held no indication that tough nuts were taken, such as those from pangi and oaks. The insects in the stomachs examined are macrolepidopteran and geometrid caterpillars along with beetles, scale insects, and cockroaches (most of which are gleaned from surfaces of foliage and bark); and buprestid beetle larvae dug from beneath bark (see ecology accounts for the different species). Contents of stomachs also point to arboreal termites in Termitidae as part of their diet. Nests of these insects are fastened to limbs in tree crowns and easily scratched open by the squirrels, causing the termites to pour out of the nest onto the branch to quickly become easy prey. The suite of insects living in rotting wood—terrestrial termites in Rhinotermitidae , legless beetle larvae, and carpenter ant pupae—sought by Rubrisciurus was not found in the any of the stomachs we examined from Prosciurillus .

Rubrisciurus rubriventer ’s large body size (500–800 g) and long legs, its bright red coloration and bushy tail, its wary and quiet nature, its habits of foraging and traveling through tree crowns in the understory layer and along the ground, and nesting in the base of hollow trees in tropical rainforest habitats are a combination seen in two other species, both in a different phylogenetic lineage ( Sciurinae ) than that of Rubrisciurus (Nannosciurinae) , and occurring on another continent. The northern Amazon red squirrel, Sciurus igniventris (500–900 g), and southern Amazon red squirrel, Sciurus spadiceus (600– 650 g) are endemic to tropical rainforest habitats in the Amazon Basin of South America ( Emmons and Feer, 1990: 167– 168). Each is red, large-bodied, long-legged, and bushy-tailed (see pl. 19 in Emmons and Feer, 1990). Both are wary, quiet, and travel in the understory canopy or along the ground; neither one travels through the upper canopy of the forest. Palm nuts along with other kinds of large seeds and fruit form their diets, and both squirrels can be located in the forest by the sounds of their gnawing on the tough palm nuts. Insects, however, have not been recorded as part of their diets (although we are not aware of any study examining contents of stomachs). There is apparently an ecological niche in New and Old World tropical forests centering on understory habitats occupied by diurnal tree squirrels sharing a close similarity in morphology, diet, and other ecological parameters, but with different evolutionary histories rooted in widely separate lineages within Sciuridae .

ECTOPARASITES: In addition to the sucking louse Hoplopleura rubrisciuri , n. sp. (see description in a following section), the flea Medwayella rubrisciurae ( Siphonaptera , Pygiopsyllidae ) has been recorded from vouchers of Rubrisciurus rubriventer collected at Tomado and Sungai Sadaunta in the west-central region of Sulawesi’s central core ( Durden and Beaucournu, 2006). Another species of flea, Macrostylophora theresae ( Ceratophyllidae ), parasitizes three species of endemic Sulawesian murid rodents ( Bunomys fratrorum , Paruromys dominator , and Rattus xanthurus ). However, Durden and Beaucournu (2006: 224) speculated that Rubrisciurus rubriventer might be the true host because other members of this flea genus are ectoparasites of squirrels, especially larger-bodied tree squirrels, throughout Southeast Asia. Rubrisciurus rubriventer is also parasitized by immature stages (larvae and/or nymphs) of hard ticks (Acari, Ixodoidea, Ixodidae ) belonging to four different genera: Amblyomma sp. , Dermacentor sp. , Haemaphysalis sp. , and Ixodes sp. ( Durden et al., 2008). Amblyomma sp. and Haemaphysalis sp. have also been collected from Hyosciurus heinrichi , and Dermacentor sp. and Haemaphysalis sp. from Hyosciurus ileile (table 56). Collectively, in addition to the squirrel hosts, the four tick genera have been collected from a suite of other mammal hosts living in Sulawesi: shrews (the endemic Crocidura sp. and Crocidura elongate , and the commensal Suncus murinus ), pigs (Sus celebensis , an endemic, and S. scrofa , feral domestic), rusa ( Rusa timorensis , nonnative), water buffalo ( Bubalus bubalis , nonnative), humans, domestic dog (nonnative), nine species of endemic murid rodents ( Bunomys chrysocomus and B. fratrorum ; Margaretamys beccari ; Echiothrix centrosa ; Maxomys hellwaldii , M. musschenbroekii , and M. wattsi ; Paruromys dominator ; Taeromys sp. ; Rattus hoffmanni , R. xanthurus , and R. marmosurus ), and four nonnative rats ( Mus musculus ; Rattus tanezumi , R. argentiventer , and R. exulans ; Durden et al., 2008). Ectoparasitic laelapid mites were recovered from R. rubriventer pelt AMNH 101316 which was collected at Wawo, Pegunungan Mekongga, in southeastern Sulawesi at an elevation of 50 m in 1932.

SYMPATRY: The range of R. rubriventer on mainland Sulawesi overlaps that of all species of Prosciurillus , except for P. abstrusus , as well as both species of Hyosciurus (table 6). Along his transect in the northern portion of central Sulawesi, Musser trapped R. rubriventer in the same traplines, and sometimes in the same traps, as P. topapuensis , P. alstoni , P. murinus , and H. ileile (table 10).

SYNONYMS: None.