Hoplopleura ileile, Musser, Guy G., Durden, Lance A., Holden, Mary Ellen & Light, Jessica E., 2010

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

Hoplopleura ileile , new species

HOLOTYPE: Female ex female Hyosciurus ileile (AMNH 224620; Rodentia , Sciuridae , Nannosciurinae, Nannosciurini, Hyosciurina) collected by G.G. Musser at 3150 ft (960 m) at Sungai Sadaunta (1 ° 239S, 119 ° 589E; see gazetteer and fig. 35), Propinsi Sulawesi Tengah, Indonesia on 1 November 1974. The holotype is deposited in USNM.

REFERRED SPECIMENS: The following specimens are all from Hyosciurus ileile hosts collected in the central core of Sulawesi by G.G., Musser : an additional two nymphs from the same individual as the holotype ; two females ex male H. ileile (AMNH 224618) at 2900 ft (884 m) also at Sungai Sadaunta on 21 October 1974; one female ex female H. ileile (AMNH 226499) at 1100 ft (335 m) at Tolai, Sungai Tolewonu (1 ° 049S, 120 ° 279E) on 31 January 1976; four nymphs ex male H. ileile (AMNH 223540) at 4800 ft (1463 m) on Gunung Kanino (1 ° 179S, 120 ° 089E) on 20 November 1973. Paratypes are deposited in BMNH and LAD .

DISTRIBUTION: Hoplopleura ileile is known from voucher hosts collected at only three localities in the northern portion of Sulawesi’s central core. Samples of the ground squirrel are from the northern part of the west-central region in the central core and one locality on the northern peninsula (see fig. 35 and table 4).

ETYMOLOGY: Hoplopleura ileile has been named for the specific epithet of its host.

DIAGNOSIS: The female of Hoplopleura ileile can be distinguished from all other species of Hoplopleura by a combination of the following characters: (1) the arrangement of the dorsal and ventral abdominal plates and the rows of DCAS and VCAS between these plates; (2) the shape of the thoracic sternal plate; (3) the shape and setation of the paratergal plates; and (4) the shape and setation of the subgenital plate. The female of Hoplopleura ileile differs from the female of Hoplopleura heinrichi in several characters including the following: (1) the shape of the thoracic sternal plate, which is elongate in Hoplopleura heinrichi (fig. 57B) but more diamond-shaped with a blunt anterior apex and with distinct lateral apices in Hoplopleura ileile (fig. 58B); (2) the shape and setation of the paratergal plates, especially the longer apical setae on plate II of Hoplopleura heinrichi (figs. 57C and 58C); (3) the significant reduction in the number and width of the dorsal and ventral abdominal plates in Hoplopleura ileile when compared to Hoplopleura heinrichi ; (4) the presence of rows of DCAS and VCAS between rows of TeAS and StAS in Hoplopleura ileile but not in Hoplopleura heinrichi (except for a few long setae situated anteriorly on the venter of Hoplopleura heinrichi ); (5) the larger number of dagger-shaped abdominal setae in Hoplopleura heinrichi ; (6) the difference in the shapes of the subgenital plates, especially the deeper lateral indentations in Hoplopleura ileile .

DESCRIPTION: Female (fig. 58A–D). Length of holotype 1.05 mm (mean for series 1.07, range 1.05–1.09, N 5 4). Head, thorax, and abdomen fairly well sclerotized.

Head. Longer than wide with broadly rounded anterior apex; 2 SuHS, 4 DMHS (displaced medially in some specimens as in fig. 58A), 1 DAnCHS, 1 DPoCHS, 1 SpAtHS, 1 DPaHS, 2DAnHS, 1 ApHS, 1–2 AnMHS, 1–2 OrS, and 1 VPHS on each side; DPHS moderate in length, just extending to level of thoracic spiracle with 1 DAcHS medial to DPHS. Antenna 5-segmented with basal segment wider than second segment, about as wide as long.

Thorax. Broader than long; thoracic sternal plate (fig. 58B) diamond shaped but with blunt anterior apex, moderately rounded posterior apex, and bifid lateral apices; mesothoracic spiracle moderate in size (0.020 mm in diameter) with 1 DMsS medial to spiracle; DPTS moderate in length (0.125 mm in paratypes, broken in holotype), extending to paratergal plate II. Legs. Forelegs small with narrow acuminate claw; midlegs and hindlegs progressively larger than forelegs with proportionately larger acuminate claws; mid coxae subtriangular.

Abdomen. Wider than thorax with one plate ventrally on each of segments 1–3 and 5–6; one to two narrow plates dorsally on segment 3 and two narrow plates dorsally on segments 4–6; rows of VCAS between StAS; rows of TeAS between pairs of dorsal plates as in figure 58A; first ventral plate without setae, second ventral plate broad, articulating with corresponding paratergal plate, and with 8–10 StAS, including 3–4 (usually three) clustered laterally on each side; third ventral plate also articulating with corresponding paratergal plate but with 7 StAS (including two long, lateral setae on each side); remaining two ventral plates with 4–5 and 5–7 StAS, respectively; rows of 5–8 VCAS present between ventral plates as shown in figure 58A; first row of DCAS with two setae, second and third rows each with 4 DCAS, and other rows with 5–8 DCAS between dorsal plates as shown in figure 57A; 2–8 TeAS on each dorsal plate; few VLAS and DLAS present; some posterior VLAS and DLAS dagger shaped. Paratergal plates (fig. 58C) present on segments 1–8: plate I small and lacking setae; plate II with two short apical setae of subequal length; plates III–VI each with two apical setae of intermediate length; plates VII and VIII each with two long apical setae; plates II–VIII subtriangular; plates III–VII each with moderate-size spiracle.

Genitalia (fig. 58D). Subgenital plate with deep lateral indentation on each side, lightly sclerotized posteriorly; two long setae on each anterior arm of subgenital plate, and 3– 4 small central setae between these; gonopod VIII with three setae with lateral seta longer than medial and central setae; gonopod IX with one long stout seta.

REMARKS: The presence of two species of sucking lice on the two host species, Hyosciurus heinrichi and Hyosciurus ileile , augments the morphological and morphometric data that provide support for the phenetic and taxonomic separation of the hosts. As a group, lice associated with mammals typically phyletically track their hosts ( Kim, 1985, 1988; Lyal, 1986, 1987; Grimaldi and Engel, 2005; Light and Hafner, 2007; Smith et al., 2008) so that host speciation often (in about 80% of known cases) also results in louse speciation (Fahrenholz’s Rule). Because Hoplopleura heinrichi and Hoplopleura ileile are distinct taxa and provide important inferences for the taxonomy of their hosts, we felt that both lice should be described in this paper despite the availability of only females and nymphs for the latter species. Nevertheless, two third-instar nymphs of Hoplopleura ileile that were preserved just prior to molting to adult males were available and it was from these pharate males (clearly visible beneath the nymphal cuticle) that we assessed male characters for cladistic analyses.

With the species of Hoplopleura from the Sulawesi squirrels named and described, the next step is to discuss them in three contexts. First, we provide results of our analysis of the phylogenetic relationships among the Sulawesi louse species and other species of Hoplopleura on Indomalayan squirrels. Second, we place the Sulawesi Hoplopleura within a portrayal of global sciurid-Anoplura associations. Finally, we use the information presented within the frameworks of the first and second contexts to describe zoogeographic patterns formed by the genera of Anoplura infesting Indomalayan squirrel faunas. The following three sections cover these subjects.

PHYLOGENETIC RELATIONSHIPS AMONG SULAWESI SQUIRREL LICE

Six of the eight new species of lice described in this report form a compact group of geographically and host-defined ectoparasites that are well suited for phylogenetic analysis based on cladistic principles. The 43 characters examined and analyzed for plesiomorphic or apomorphic states are listed in table 52. Characters were obtained for males and females of all the six species of lice. The male characters for Hoplopleura ileile were assessed from pharate specimens visible beneath the cuticle of third-instar nymphs preserved immediately before their molt to adult males. Samples of Hoplopleura leucomus and H. alstoni could not be included in the analysis because they did not contain adult or pharate males. Appendix 1 is a data matrix for the 43 characters examined with 0 denoting plesiomorphy and 1, 2, and 3 reflecting successive apomorphic states.

In addition to the six species of lice parasitic on the endemic Sulawesi squirrels, we included morphological characters of lice from four other species of squirrel hosts and four louse species parasitizing endemic Sulawesi murid rodents. Of the squirrel lice, Hoplopleura trispinosa parasitizes the North American flying squirrels, Glaucomys volans and G. sabrinus , and was employed as an outgroup in the analysis. Hoplopleura distorta from Tamiops swinhoei , H. thurmanae from Tamiops macclellandi , and H. erismata from species of Callosciurus represented lice parasitizing species of Indomalayan squirrels found outside Sulawesi (see table 54). We restricted our sample of Hoplopleura to those species parasitizing species of squirrels in Nannosciurini (of Nannosciurinae ), which excluded other Indomalayan squirrel species

TABLE 52 Morphological Characters Assessed for Cladistic Analysis of Sulawesi Squirrel Lice and Related Lice a,b (0 5 plesiomorphic state; 1, 2, and 3 5 successive apomorphic states).

CHARACTERS:

1. Total body length of male more than 1.2 mm (0), 0.8–1.2 mm (1), or less than 0.8 mm (2).

Head

2. Anterior apex of head angular/rounded (0), or broadly flattened (1).

3. Antennae 5-segmented (0) or 4-segmented (1).

4. First antennal segment without (0), or with (1) spinelike seta.

5. Postantennal angle rounded (0), or angular (1).

6. Postantennal angle directed laterally (0), or anterolaterally (1).

7. Postantennal angle lightly (0), or heavily (1) sclerotized.

8. 3 SHS (0), 2 SHS (1), or 1 SHS (2) on each side.

9. Female with 4 DMHS (0), 3 DMHS (1), 2 DMHS (2), or 0 DMHS (3).

10. DPHS situated on lateral lobe (0), or not (1).

11. DPHS long (0), or short (1).

Thorax

12. DPTS long (0), short (1), or absent (2).

13. Mesothoracic spiracle small (0), or large (1).

14. Anterior and posterior processes of thoracic sternal plate dissimilar (0), or similar (1).

15. Posterior apex of thoracic sternal plate elongate (0), broadly rounded (1), or doubled (2).

16. Bladderlike outgrowths absent (0), or present (1) on legs.

Abdomen

17. Abdomen with long (0), or short (1) setae.

18. Second sternite short or absent (0), extended laterally to articulate with corresponding paratergal plate and entire (1), or enlarged and divided medially (2).

19. Sternal abdominal setae uniform (0), or some modified into spiniform or lanceolate forms (1).

20. Tergal abdominal setae uniform (0), or some modified into spiniform or lanceolate forms (1).

21. First TeAS long (0), or short (1).

22. Lateral StAS of segment 3 distinctly enlarged (0), slightly enlarged (1), or not enlarged (2) compared to adjacent setae.

23. Lateral StAS of segment 3 tripled (0), or doubled (1) on each side.

24. Average number of paired DLAS in female more than 20 (0), 8–20 (1), or less than 8 (2).

25. Abdominal sternites on segments 4–7 in male well-developed (0), or poorly developed (1).

26. Abdominal tergites on segments 4–7 in male well-developed (0), poorly developed (1), or absent (2).

27. Abdominal tergite on segment 3 in male without (0), or with (1) distal lobe.

28. Distal end of female abdomen without (0), or with (1) distinct protuberances each bearing an elongate seta to collectively form a ‘‘comblike’’ structure.

Paratergal plates

29. Spiracles associated with 6 pairs (0), 5 pairs (1), or 3 pairs (2) of paratergal plates.

30. Diameter of spiracles IV–VI small (0), or large (1).

31. Paratergal plate II with 2 long apical setae (0), with 1 long and 1 short apical seta (1), with 2 short apical setae (2), or with 1 seta (3).

32. Paratergal plate II not appreciably elongate (0), or greatly elongate as a ‘‘winglike’’ appendage (1).

33. Paratergal plates III–VI with both apical setae of similar lengths (0), or of contrasting lengths (1).

34. Paratergal plates VII and VIII of female symmetrical (0), or asymmetrical (1).

Female genitalia

35. Female gonopod VIII with long (0), or short (1) setae.

36. Female subgenital plate well sclerotized (0), or poorly sclerotized (1).

37. Female subgenital plate without (0), or with (1) deep lateral indentation on each side.

Male genitalia

38. Margins of male subgenital plate smooth (0), or scalloped (1).

39. Male basal apodeme longer than (0), or about equal in length (1) to parameres.

40. Parameres uniformly (0), or differentially (1) sclerotized.

TABLE 52 (Continued)

CHARACTERS:

41. Medial margins of parameres smooth (0), or with a spinelike projection (1).

42. Pseudopenis extending well beyond the apices of the parameres (0), or not (1).

43. Lateral margins of pseudopenis smooth (0), with fine crenulations (1), or with a spinelike projection (2).

a Some uninformative characters were informative when different outgroup taxa were included in the matrix (data available upon request).

b Synopsis of species of Hoplopleura from which morphological traits were surveyed and employed in the data matrix and cladogram (all known species of Hoplopleura that parasitize species of squirrels in the Nannosciurini are included in the analysis): (1) H. trispinosa : outgroup; parasitizes species of Glaucomys in North America. (2) H. distorta : parasitizes Tamiops swinhoei in China. (3) H. erismata : parasitizes species of Callosciurus on mainland Southeast Asia ( Myanmar, India, Nepal, peninsular Malaysia, China, and Thailand). (4) H. thurmanae : parasitizes species of Tamiops and Menetes berdmorei in China and Thailand. (5) H. abstrusus : parasitizes Prosciurillus abstrusus in Sulawesi. (6) H. heinrichi : parasitizes Hyosciurus heinrichi in Sulawesi. (7) H. ileile : parasitizes Hyosciurus ileile in Sulawesi. (8) H. murinus : parasitizes Prosciurillus murinus in Sulawesi. (9) H. rubrisciuri : parasitizes Rubrisciurus rubriventer in Sulawesi. (10) H. topapuensis : parasitizes Prosciurillus topapuensis in Sulawesi. (11) H. chrysocomi parasitizes Bunomys chrysocomus in Sulawesi. (12) H. musseri parasitizes Maxomys musschenboekii in Sulawesi. (13) H. sembeli parasitizes Maxomys hellwaldii in Sulawesi. (14) H. traubi parasitizes Maxomys wattsi in Sulawesi.

in other tribes (Indian Funambulus in Funambulini of Nannosciurinae , for example, and Chinese Sciurotamias in Marmotini of Xerinae ).

Because most representatives of Hoplopleura parasitize murid rodents ( Durden and Musser, 1994) and secondarily infest sciurids (representing host switching), we also employed in the analysis characters from four species of Hoplopleura recorded only from endemic Sulawesi rats ( Muridae ). Hoplopleura chrysocomi is specific to its host, Bunomys chrysocomus (Durden, 1990) ; H. sembeli is mainly a parasite of Maxomys hellwaldii but a few specimens have been collected from Bunomys fratrorum and Rattus hoffmanni (Durden, 1990) ; H. musseri is host-specific to Maxomys musschenbroekii (Durden, 1990) ; H. traubi paratisizes Maxomys wattsi ( Durden and Musser, 1991) . Including these four murid lice and the six squirrel lice, all endemic to Sulawesi, would help test the cladistic relationship of the squirrel parasites, whether they were recovered as a monophyletic cluster or instead intermingled with the murid-associated species of lice.

Of the 43 characters examined, 32 were potentially parsimony informative. MP analysis produced two equally parsimonious trees (length, 101; consistency index [CI], 0.426; retention index [RI], 0.525; rescaled consistency index [RC], 0.223; fig. 59). Bayesian analyses yielded a similar, although generally less resolved, tree (fig. 59; Bayesian topology available upon request). Bootstrap and Bremer support values and Bayesian posterior probabilities were generally small (fig. 59). The MP analysis resulted in a monophyletic clade of Sulawesi squirrel lice (Bayesian posterior probability 5 0.60, Bremer support 5 2; fig. 59) distinct from a monophyletic clade of Sulawesi murid lice (Bayesian posterior probability 5 0.68, Bremer support 5 2; fig. 59)

Reconciliation analyses using TreeMap 2.0b did not support significant phylogenetic congruence between squirrel lice and their hosts for all combinations of host species relationships (P. 0.3). There are two phylogenetic incongruities within the squirrel-louse cladogram. One is the position of Hoplopleura rubrisciurus , which parasitizes the tree squirrel Rubrisciurus rubriventer , nested in the clade containing H. heinrichi and H. ileile , the lice infesting the two species of ground squirrels in Hyosciurus . Among the hosts, Rubrisciurus is cladistically linked to the tree squirrels in Prosciurillus in a clade separate from Hyosciurus ( Mercer and Roth, 2003) . The other phylogenetic incongruity is the association of H. murinus (infesting Prosciurillus murinus ) with H. topapuensis (the louse on P. topapuensis ). Among species of Prosciurillus , P. murinus is most closely related to P. abstrusus and the two together comprise the P. murinus group, while P. topapuensis is a member of the larger-bodied P. leucomus group (which includes P. leucomus , P. alstoni , P. weberi , and P. rosenbergii ).

But ‘‘Cladograms of hosts and their symbionts are never perfectly congruent …,’’ and how ‘‘… the cladograms match to each other is the result of several ‘sorting events’: cospeciation, host colonization/shift, intrahost speciation, extinction, or any combination of these’’ ( Grimaldi and Engel, 2005: 278). Such is the imperfect match between cladograms of Sulawesi squirrels and their parasitic lice. Degrees of incongruence are also illustrated in results of the cospeciation studies involving lice and their bird hosts (seabirds in Procelariiformes and Sphenisciformes [ Paterson et al., 2000] and swiftlets in Collocallinae [ Page et al., 1998]); and even between chewing lice and their pocket gopher hosts ( Geomyidae ; Hafner et al., 1994; Hafner and Page, 1995), which is probably the ‘‘—most intensively studied and well-known example of lousehost cospeciation …’’ ( Grimaldi and Engel, 2005: 278).

Our results derive from morphological characters possessed by the sucking lice and gene sequences sampled from the hosts. The lice-host cospeciation studies for birds and pocket gophers cited previously employed gene sequences from the bird and rodent lice and from their hosts. Certainly the future step in testing the cospeciation of lice and Sulawesi squirrel hosts and the integrity of our cladogram is to employ sequences from mitochondrial and nuclear genes. At a different level, the significant cladistic pattern shown here describes the six species of Hoplopleura from their squirrel hosts as a monophyletic group separate from the other species of Hoplopleura employed (that parasitize non-Sulawesi Indomalayan squirrels and endemic Sulawesi murid rodents), which is congruent with the monophyly of Rubrisciurus , Prosciurillus , and Hyosciurus relative to the separate clades recovered in other Indomalayan nannosciurine genera ( Mercer and Roth, 2003).

A SYNOPSIS OF GLOBAL SCIURID-ANOPLURA ASSOCIATIONS

The new taxa described in this paper represent a significant addition to the sucking louse fauna known to parasitize sciurid rodents and it is appropriate to consider briefly the host associations and zoogeography of currently recognized genera of Anoplura known to parasitize this group of mammals. The hosts and geographical distributions of all 11 genera of Anoplura that are known to parasitize sciurids are listed in table 53. This figure is higher than the number of anopluran genera known to parasitize any other single family of mammals ( Durden and Musser, 1994; Kim, 2006), with the next highest figures being six genera ( Hoplopleura , Cuyana , Eulinognathus , Galeophthirus , Lagidiophthirus , and Polyplax ) associated with rodents in Muroidea (as interpreted by Musser and Carleton, 2005) and Hystricognathi (see Woods and Kilpatrick, 2005), three genera ( Hoplopleura , Pterophthirus , and Ctenophthirus ) associated with rodents in Echimyidae (infraorder Hystricognathi ), and three genera ( Haematopinus , Linognathus , and Solenopotes ) associated with each of the Bovidae and Cervidae .

All five genera ( Atopophthirus , Enderleinellus , Microphthirus , Phthirunculus , and Werneckia ) belonging to the anopluran family Enderleinellidae (representing representing 56 species, including two undescribed species of Enderleinellus —one from Microsciurus flaviventer in Peru, the other from Tamiops maritimus in Laos) are exclusively parasites of sciurids (table 53; Ferris, 1919; Werneck, 1948; Kim and Ludwig, 1978; Chin, 1979; Kim, 1966, 1985, 1988, 2006). There appears to have been significant parallel evolution between the Enderleinellidae and the Sciuridae , with close phyletic host tracking shown by these lice at both the generic and species levels ( Kim, 1988, 2006). As a group, the Enderleinellidae is widely distributed within the Ethiopian, Nearctic, Neotropical, Oriental, and Palearctic regions as parasites of squirrels in the tribes Nannosciurini , Funambulini , Protoxerini , Sciurini , and Pteryomyini. However, most of this wide distribution can be attributed to the various species of Enderleinellus because

TABLE 53 Genera of Sucking Lice Parasitic on Sciurid Rodents

Atopophthirus and the monotypic Phthirunculus are confined to Oriental flying squirrels, the monotypic Microphthirus parasitizes Nearctic flying squirrels, and the five known species of Werneckia occur only on Afrotropical Funambulini and Protoxerini ( Durden and Musser, 1994; see also table 53).

Two genera of hoplopleurid sucking lice, Hoplopleura and Paradoxophthirus , parasitize sciurids (table 53). Most representatives of the numerically large genus Hoplopleura parasitize murid rodents ( Durden and Musser, 1994) although secondary infestations (representing host switching) such as those of the sciurids listed in table 54, have evidently occurred on a few occasions during the evolution of the Hoplopleuridae . Infestations of sciurids by characteristic species of Hoplopleura reflect scenarios where strict phyletic tracking of murid hosts failed to occur and host switching took place instead; other examples of host switching have been discussed in connection with trichodectid chewing lice and their hosts by Lyal (1986, 1987), Hafner and Nadler (1988, 1990), and Page (1990), and for argasid ticks by Klompen et al. (1994). Data included in table 54 suggest that representatives of the genus Hoplopleura switched from murids to sciurids as hosts on several separate occasions with present-day descendants of these switches now represent- ed by two species ( H. sciuricola and H. trispinosa ) on North American tree squirrels and flying squirrels (subfamily Sciurinae ); two species ( H. arboricola and H. erratica ) on Nearctic species of chipmunks, and one ( H. xiei ) on a Chinese ground squirrel (subfamily Xerinae ; 14 species ( H. abstrusus , H. distorta , H. erismata , H. funambuli , H. heinrichi , H. ileile , H. karachiensis , H. maniculata , H. murinus , H. rubrisciuri , H. thurmanae , H. topapuensis , H. leucomus , and H. alstoni ) associated with nannosciurine squirrels in the Indomalayan region. While the three species

TABLE 54 Described Species of Hoplopleura Parasitic on Sciurid Rodents a of Hoplopleura associated with Funambulus parasitize closely related (congeneric) squirrels, hosts of the North American and Indomalayan species of Hoplopleura are less intimately related, and more than one host switch from murids to sciurids may have been responsible for present host-louse associations in each region.

Paradoxophthirus emarginata (Ferris) , the sole representative of its genus, also was formerly placed in the genus Hoplopleura ( Ferris, 1922; Durden and Musser, 1994). However, this louse has several unusual morphological features (including some that are shared with the polyplacid genus Neohaematopinus ) that led Chin (1989) to erect a new genus for it. Paradoxophthirus emarginata is exclusively parasitic on the Chinese ground squirrel Sciurotamias davidianus , a member of the tribe Marmotini of subfamily Xerinae that also embraces Nearctic and Palearctic marmots and chipmunks, North American and Eurasian ground squirrels, and North American prairie dogs (see Thorington and Hoffmann, 2005, and our section on classification of squirrels).

Four genera of polyplacid sucking lice parasitize sciurids (table 53). Two of these genera, Johnsonpthirus and Linognathoides , exclusively parasitize sciurids ( Kim and Adler, 1982; Durden, 1991 a, 1991b; Durden and Musser, 1994; Kim, 2006). The five described species of Johnsonpthirus parasitize African squirrels belonging to Paraxerus , Funisciurus , and Heliosciurus in Protoxerini of Xerinae ( Kim and Adler, 1982) . Species of Linognathoides are parasites of squirrels in Xerinae : Asian, European, and North American Marmotini (marmots and prairie dogs) and African Xerini ( Kim and Adler, 1982) .

Neohaematopinus is another polyplacid louse genus that primarily parasitizes sciurids with 30 of the 32 described species exclusively associated with rodents belonging to this family ( Durden, 1991a; Kim, 2006). The two exceptions, N. inornatus and N. neotomae , parasitize species of woodrats ( Neotoma ) in North and Central America ( Kim et al., 1986; Durden and Musser, 1994) and reflect host switching from sciurid to neotomine rodents. Neohaematopinus is mainly a genus of the northern land masses with no native species in Africa (or Antarctica or Australia where suitable hosts would be unavailable anyway) and only two species with ranges that extend partially into the northern Neotropics ( Johnson, 1972; Durden, 1991a). However, this louse genus is widespread across North America and Eurasia and it has successfully parasitized numerous species of Indomalayan nannosciurine squirrels ( Johnson, 1959, 1964; Durden and Musser, 1994). Some other species of Neohaematopinus are associated with squirrels in Sciurini and Pteromyini ( Sciurinae ).

Although the genus Polyplax principally parasitizes murid rodents ( Durden and Musser, 1994), one species, P. bureschi Touleshkov , infests the common suslik ( Spermophilus citellus ) in Europe ( Touleshkov, 1957). This louse-squirrel association apparently represents another case of host switching from the more normal murine-tracking path of Polyplax onto a sciurid.

ZOOGEOGRAPHY OF ANOPLURA PARASITIZING INDOMALAYAN SCIURIDS

The geographical distributions and numbers of species present for the seven genera of sciurid-infesting Anoplura in eight Southeast Asian countries, regions, islands, or islandgroups traveling southwest from China and Thailand through peninsular Malaysia and then eastwards through the Greater Sunda Islands to Sulawesi and the Philippines are listed in table 55. This approximates the hypothesized migration route that ancestral sciurids traversed during their historical colonization of the Sunda Shelf, Sulawesi, and the Philippines from mainland Southeast Asia ( Traub, 1972; Durden and Traub, 1990). Sulawesi and the southern Philippines represent the easterly most extensions of the sciurid colonizations east of the Sunda Shelf. We stress that many areas in the Indomalayan region have been inadequately sampled for sucking lice and forthcoming descriptions of new taxa will necessitate emendations to some of the numbers given in table 55. However, current data clearly show that the three mainland regions all have more known genera and species of sciurid-infesting Anoplura than any of the five islands or islandgroups listed. It appears that progressively fewer sucking lice accompanied ancestral sciurids in their colonizations deeper onto the Sunda Shelf, Sulawesi and the Philippines. A similar situation has been described for ceratophyllid fleas parasitizing sciurids and for leptopsyllid fleas infesting murine rodents in the Indomalayan region ( Traub, 1983; Durden and Traub, 1990).

The larger number of anopluran genera and species reported from mainland Southeast Asia, particularly China, is also partly related to the greater diversity of potential

TABLE 55 The Known Louse Species in Seven Genera Parasitic on Sciurid Rodents in the Indomalayan Region sciurid hosts in these areas ( Chin, 1979, 1985, 1989, 1991, 1994; Corbet and Hill, 1992; Thorington and Hoffmann, 2005). For example, marmots are parasitized by sucking lice belonging to Linognathoides (and Enderleinellus in some other regions) but in continental Asia, marmots do not range farther south than northern China ( Chin, 1985; Thorington and Hoffmann, 2005; Hoffman and Smith, 2008). Similarly, Sciurotamias davidianus , the sole ground squirrel host of the monotypic anopluran Paradoxophthirus , is confined to China ( Moore and Tate, 1965; Corbet and Hill, 1992; Hoffman and Smith, 2008). Also, Chin (1979, 1985, 1989, 1991, 1994) has intensively surveyed the Chinese sucking louse fauna, while Johnson (1959, 1964) reported on the anopluran parasites of rodents, including sciurids, in Thailand, peninsular Malaysia, and parts of Borneo. Except for a few descriptive papers ( Ferris, 1919, 1923; Johnson, 1959, 1964; Kuhn and Ludwig, 1965; Durden, 1991a), little has previously been published on any Anoplura from sciurids that are native to the islands on the Sunda Shelf and those lying to the east. Nevertheless, progressively fewer genera and species of sciurid-infesting Anoplura evidently accompanied their hosts in their southeastern colonization of the Indomalayan region, Sulawesi, and the Philippines. An alternative scenario would be that some of these louse genera did initially accompany squirrels in their southeastern colonizations but that these lice were eventually lost from these hosts. We have no

TABLE 56

Summary of Ectoparasite Records for the Species of Rubrisciurus , Hyosciurus , and Prosciurillus

(All the lice are described in the present report; published sources for fleas, ticks, and mites are referenced in the ECTOPARASITES subheading in species accounts of the squirrels.)

proof for either phenomenon but, because external factors such as ambient temperature, relative humidity and host habitat, would presumably have little effect on the survival of permanent ectoparasites like sucking lice, we cannot speculate why lice would be lost from a particular host based on these factors alone. The diminution across China to the Sunda Shelf from seven genera of sucking lice to just one ( Hoplopleura ) on Sulawesi roughly parallels the decrease in squirrel taxa, from 22 genera (including flying squirrels) in three subfamilies and four tribes to the three endemic Sulawesian genera contained in one subtribe.

Three sciurid-infesting anopluran genera that occur on mainland Southeast Asia apparently have not colonized any of the Sunda Islands, Sulawesi, or the Philippines. As outlined above, two of these genera, Linognathoides and Paradoxophthirus , are specific parasites of sciurid taxa that similarly have been unable to colonize these islands. However, the third louse genus, Atopophthirus , parasitizes Petaurista elegans ( Kim, 1977) , a flying squirrel that also inhabits the Greater Sunda Islands ( Corbet and Hill, 1992; Thorington and Hoffmann, 2005) but from which this louse has not been collected in these insular locations.

Except for Sumatra (a very poorly worked island for Anoplura), the sciurid-specific louse genus, Enderleinellus , successfully accompanied squirrel hosts as far east as Borneo and Java ( Johnson, 1964). However, Enderleinellus is unknown from Philippine or Sulawesian squirrels and therefore appears to have either failed to accompany the nannosciurines in their eastern-most colonizations east of the Sunda Shelf or to have been lost from them if this louse genus did initially accompany squirrels onto these islands. This apparent absence of members of Enderleinellus could reflect undercollecting for Philippine squirrels, but our meticulous searches of numerous Sulawesian squirrel pelts during this study, make us confident that Enderleinellus does not parasitize native squirrels on that island.

Hoplopleura shows an intriguing geographical distribution on Indomalayan squirrels (table 55). Species of this louse genus parasitize squirrels on mainland Southeast Asia but are currently known only from Philippine (one undescribed species known to us) and Sulawesian (eight new species described in this paper) squirrels within the broader Indomalayan region. If the ancestral squirrels reached the Philippines and Sulawesi via the Greater Sunda Islands, as seems likely, species of Hoplopleura would also be expected to parasitize squirrels on Borneo and/or Java, the principal Sundaic islands from which the ancestral squirrels presumably invaded the Philippines and Sulawesi. It is possible that Hoplopleura became extinct on Bornean and/or Javanese nannosciurine squirrels (perhaps because of competition from other genera of sucking lice) but it seems more likely that squirrel-infesting species of Hoplopleura await discovery on one or both of these islands. The sucking louse fauna of Bornean rodents has been fairly well documented by Johnson (1964), but surprisingly (because of its large human population with the associated removal of pristine forest habitats) the sucking louse fauna of native Javanese mammals is poorly known.

Nevertheless, just one invasion of Sulawesi by a representative of Hoplopleura (or its immediate ancestor) could feasibly account for the present diversity of this louse genus on the Sulawesian squirrel fauna. Theoretically, an ancestral lineage of Hoplopleura could have accompanied an ancestral squirrel lineage in its colonization of Sulawesi and then phyletically tracked this lineage as it evolved and radiated into the currently recognized endemic squirrel species in Rubrisciurus , Prosciurillus , and Hyosciurus . This scenario is consistent with results of molecular analyses by Mercer and Roth (2003: 1570) who indicated ‘‘a single lineage of squirrels crossed Wallace’s Line to give rise to the three genera … and multiple species of squirrels on Sulawesi.’’ That ancestor (along with its obligate Hoplopleura ) may have dispersed across the Makassar Strait from the Sunda Shelf to Sulawesi sometime between 11.4 to 10.5 million years ago ( Mercer and Roth, 2003: 1571). Because neither Enderleinellus nor Neohaematopinus apparently accompanied colonizing squirrels into Sulawesi, host tracking and co-speciation by Hoplopleura or its ancestor of different endemic Sulawesian Hyosciurina squirrels may have been easier to accomplish in the absence of potential competition from members of Enderleinellus and Neohaematopinus .

The anopluran Neohaematopinus accompanied the ancestral nannosciurines throughout their ramifications into the Indomalayan region except for Sulawesi and most of the Philippines (the single known species from Sumatra and the absence of known species from Java probably reflect undercollecting). Apparently, Neohaematopinus did not accompany the nannosciurines to Sulawesi or to the Philippines proper (the only Philippine records are from Palawan, an island that lies on the Sunda shelf and that is zoogeographically more closely allied to Borneo than to other Philippine islands [ Heaney, 1986]).

Results from analyses of future collections of sucking lice from Indomalayan sciurids will be used to test and refine some of the hypotheses presented here.

In the previous sections, we document information covering aspects of Sulawesi’s endemic species of squirrels and their Hoplopleura parasites. We use these data as a painter’s tool, a brush to render on a broad canvas a picture of Sulawesi squirrel systematics and their louse associates that is different from the usual image derived from contents in current published archives. While our depiction is intellectually and visually satisfying in its informational substance and zoogeographic texture, the picture is incomplete. Gaps remain. Portions of the canvas are without paint, other parts are fuzzy: hues and tones are diluted, outlines and shapes are undefined. Some spots will remain empty because the forests and squirrels, the source of the paint, have disappeared. Still, a nearly complete rendering could be accomplished, and the possible sources of the pigments needed for a final version are itemized in the next segment.

FUTURE INQUIRIES

Our analyses of data derived from museum specimens have allowed us to define eight species of tree squirrels in Rubrisciurus and Prosciurillus and two species of ground squirrels in Hyosciurus that are endemic to mainland Sulawesi, its offshore islands, and the Sangihe Archipelago. All species are diurnal. They exhibit an impressive range in body size (table 3) along with color pattern of pelage. From ecological observations and trapping data for certain species on the Sulawesi mainland, we know of their attachment to tall old-growth forests. Some species occur only in lowland tropical evergreen rainforest habitats, others are restricted to montane formations, and a few range from lowlands to montane environments (table 4). Within any locality, each species can be found on well-drained forested ridges, in valleys, and forests bordering streams and rivers. Rubrisciurus rubriventer and Prosciurillus murinus occur throughout most of Sulawesi, other species of Prosciurillus have restricted ranges on the mainland, and P. rosenbergii is the only squirrel known from islands in the Sangihe Archipelago (tables 3, 4). On the mainland, several species may occur in the same forest, but other distributions form parapatric or allopatric patterns (table 6). Understory habitats are occupied by Rubrisciurus rubriventer and Prosciurillus murinus (and probably P. abstrusus ), the high canopy layer and emergent crowns are the foraging and nesting sites of members in the Prosciurillus leucomus group of species, and the two species of Hyosciurus are terrestrial. Fruits, seeds, and arthropods comprise the diets of those species for which this information is available (table 57). One or two young comprise the litters; Rubrisciurus has two pairs of teats, species in the other genera have three pairs (table 58).

Of the 10 species of squirrels, eight is each a host to a unique species of sucking lice in Hoplopleura . The lice comprise a monophyletic group relative to species of Hoplopleura parasitizing non-Sulawesi Indomalayan squirrels and endemic Sulawesi murid rodents (fig. 59). Their unique lineage mirrors that of the three genera of squirrels, which are monophyletic relative to other Indomalayan genera in the Nannosciurinae , one of the three subfamilies constituting Sciuridae .

Our portrayal of the endemic Sulawesi squirrels and their lice parasites should be viewed as a beginning, a first attempt towards understanding the actual regional and altitudinal distributions of the species, their ecologies, and their parasites. The results outlined here provide hypotheses that can be tested by analyses of data gathered in future inquiries. Researchers need to draw information from new biological surveys in particular regions of Sulawesi and employ not only the kinds of data analyzed here but additional sources of information, particularly that derived from focused observation of squirrel activities, and that obtained from genetic samples. In the following paragraphs we elaborate on the subjects that need to be addressed by fresh research efforts.

THE SQUIRRELS

Prosciurillus leucomus : The northern peninsula of Sulawesi is the range we describe for Prosciurillus leucomus . It is drawn from suitably large samples collected on the northeastern end but fewer and smaller series from elsewhere on the peninsula. Size of nape patches and their contrast with the back-

TABLE 57 Summary of Fruit, Seeds, and Arthropods Eaten by Species of Rubrisciurus , Prosciurillus , and Hyosciurus a TABLE 58 Summary of Reproductive Information for Species of Rubrisciurus , Prosciurillus , and Hyosciurus ground fur, along with coloration of fur covering underparts of head and body, vary noticeably among present samples, but not cranial and dental dimensions. East of the Gorontalo region the squirrels have bright, white, and expansive nape patches, and rich reddish orange underparts; west of Gorontalo the specimens examined exhibit the range from necks with large and conspicuous patches to diluted markings or none at all, and the underparts range from deep reddish orange to buffy orange. Female lice gathered from one specimen in the northeast and from one at the base of the northern peninsula in the west show slight differences in lengths of apical setae on some of the paratergal plates, but we consider this minor variation to be within the morphological limits of a single louse species because morphology of the subgenital plate and its associated setae of the lice from both squirrel specimens are highly distinctive compared with the morphologies found in the other Sulawesi species of squirrels. For the squirrels, the evidence from nape markings and color saturation of the underparts suggests some past, likely partial, suppression of gene flow between eastern and western populations, a pattern seen in macaques ( Fooden, 1969) and some murid rodents (Musser’s research). But more and larger samples of squirrels and their lice parasites are required from that region extending from the Gorontalo area west to the base of the peninsula so the variation in pelage coloration and patterning can be more completely described and its significance understood.

Prosciurillus rosenbergii : Specimens collected in the late 1800s and now stored in museums remain the only source of information about Prosciurillus rosenbergii in the Sangihe Archipelago. We know nothing of its altitudinal distributions on the various islands or anything about its ecology. We assume it is a species frequenting the upper canopy, as Musser observed for mainland P. topapuensis and P. alstoni , but because P. rosenbergii is the only squirrel on the four islands in Kepulauan Sangihe, it may be active at all levels of the forest, from the ground to the upper canopy layer. We need new inventories of indigenous mammals inhabiting the Sangihe islands, especially those living on Pulau Sangihe itself. If intact primary forest is still present, the squirrels should also be there; unfortunately, most of the original forests on Pulau Sangihe were long ago converted to plantations of coconut and nutmeg, and by 1920 most of the northern islands had lost their original forest cover (see the account of P. rosenbergii ).

The specimens stored in museums consist of skins, some accompanied by damaged skulls. We could not assess variation among insular populations using data from cranial and dental measurements. This obstacle could be overcome by sampling hairs or sections of skins from the old specimens from which DNA might be extracted and gene sequences subsequently analyzed. The results would provide a picture of genetic variation among the island populations, at least among the populations that were present at the times of collection, and perhaps inform us about degree of genetic isolation among the populations of squirrels from each island.

Prosciurillus weberi : A similar level of ignorance attends Prosciurillus weberi . We know of only eight specimens representing this species, all collected between 1889 and 1940 from the environs of Masamba and Palopo in the southern part of Sulawesi’s central core. We strongly suspect that the actual range of P. weberi is more extensive than is indicated by the small sample stored in museum collections, possibly extending throughout the lowlands at the southern end of the western mountain block, south into the southwestern peninsula, and north along the western coastal lowlands of Sulawesi’s central core (see the account of P. weberi ). Its ecology is likely similar to that of P. alstoni , which occurs to the east of P. weberi ’s range, but this is a guess in the absence of data. Future research should endeavor to illuminate the actual range of P. weberi as well as its ecological characteristics.

Allopatry among species in the Prosciurillus leucomus group: On mainland Sulawesi, P. leucomus , P. alstoni , P. weberi , and P. topapuensis have allopatric ranges relative to each other (see the map in figure 11). In the accounts of species we described the closest distances between collection localities for different sets of species’. Areas in between these ranges have yet to be sampled and we remain ignorant of the finer distributional relationships among the different species we have defined here, and their possible genetic interactions. Is there narrow sympatry along edges of the geographic ranges or does each abut in some parapatric pattern? Does each species retain its phenetic and genetic integrity at the range boundaries or is there gene exchange and some level of hybridization as has been documented among some of the very morphologically distinct species of Sulawesi macaques (Watanabe and Matsumara, 1991; Watanabe et al., 1991 a, 1991b; Froehlich and Supriatna, 1996; Bynum et al., 1997; Groves, 2001; Evans et al., 2001, 2003)?

Rising and falling sea levels accompanied by tectonic uplift and subsidence, along with volcanism, have shaped Sulawesi through the Pleistocene, fragmenting it into an archipelago, and later reuniting the islands into a solid landmass ( Fooden, 1969). Such past geological transformations are reflected in the present distributions of some animals endemic to Sulawesi’s forests. In this context, the six (or seven, depending on the research- er) species of Sulawesi macaques are iconic, their species’ formations tied to population isolation and genetic uniqueness on Archipelago Sulawesi with renewed population contacts after the island’s restoration to a single body of land ( Fooden, 1969). Sulawesi macaques have been the focus of intense observational, behavioral, morphometric and morphological, developmental, endoparasitic, and genetic inquiries ( Albrecht, 1978; Watanabe and Matsumura, 1991; Watanabe et al., 1991 a, 1991b; Suryobroto, 1992; Muroyama and Thierry, 1998; Gotoh et al., 2001; Groves, 1980 b, 2001; see also the references in Evans et al., 2003). Distinct populations of the toad Bufo celebensis show geographic distributions concordant with those of Macaca (Evans et al., 2003) . The ranges we sketch here for mainland members of the Prosciurillus leucomus group of species provide another potential set of distributional patterns that can be compared with those of macaques and toads. But to carefully test any congruence will first require collections of new material from regions of Sulawesi that are now unsampled for the P. leucomus assemblage of species, and from places between currently defined boundaries of each species.

Genetic relationships among species in the Prosciurillus leucomus group: We cannot currently describe the pattern of phylogenetic affinities among the species. We considered a cladistic analysis employing cranial and dental traits but rejected this approach as likely unproductive because the species are so similar except for color patterns of the fur and geographic distributions. Speculating about the degree of genetic relationship based on the allopatric pattern of the species’ ranges is frustrating and could be misleading. For example, it seems reasonable to assume that P. rosenbergii in the Sangihe Archipelago would be more closely related to P. leucomus on the northern peninsula of Sulawesi, but the former is strikingly different in fur coloration from the latter, and from the other mainland species in the P. leucomus cluster. It is not far-fetched to look on P. rosenbergii as an early offshoot from the ancestor of the group, and not especially closely related to any particular species.

Prosciurillus topapuensis is known only from the west-central region of Sulawesi’s central core. In its range and color pattern, the species seems as isolated from surrounding species as is P. rosenbergii in the northern archipelago. Prosciurillus topapuensis does exhibit black ear tufts, which is characteristic of P. leucomus to the north and P. weberi to the southeast, and its altitudinal range embraces tropical lowland evergreen and montane rain-forest habitats, as does P. leucomus ; by contrast, P. weberi and the eastern P. alstoni have been collected only in lowland forests. Do these observations have significance in unraveling the degree of phylogenetic affinity between P. topapuensis and any of the other species? Viewing another trait, P. leucomus , P. alstoni , and P. weberi share brightly pigmented venters (red to orange hues). Does this similarity signify a closer relationship among those three species than to P. topapuensis with its dark gray underparts?

We believe that analysis of data from gene sequences would be the most productive approach to recover phylogenetic relationships among species in the P. leucomus group. Results from such an endeavor may also identify the oldest member of the group and possibly inform the timing and geographic sequence of past isolation events on the island.

The Prosciurillus murinus group: Uniting the small-bodied P. murinus with P. abstrusus in opposition to the larger-bodied species in the P. leucomus group conforms strictly to morphological criteria. Prosciurillus abstrusus is typically physically larger than P. murinus and the outer surfaces of its ears are covered with short, white fur; otherwise, the two species are closely similar in pelage coloration and cranial conformation. A patch of fur covering the medial surface of each ear that contrasts with the head and neck, and that may or may not form tufts, is typical of Sulawesi’s tree squirrels Rubrisciurus rubriventer and most of the species in the Prosciurillus leucomus group but not P. murinus . It is possible that P. abstrusus is a closer genetic relative of P. leucomus and the other members in that group than of P. murinus . On the other hand, swatches of fur on the ears that contrast with the head may be primitive for Sulawesian tree squirrels, a trait simply lost in P. murinus . While we strongly advocate the hypothesis that P. abstrusus is a montane representative of P. murinus that is genetically and geographically isolated on Pegunungan Mekongga, this assessment should be tested by obtaining results from analyses of gene sequences.

Except that P. abstrusus was collected in montane forest habitats, other aspects of its ecology are unknown, as is its altitudinal distributional in relation to P. murinus . Do the species exhibit similar ecologies? Does P. abstrusus descend into the upper reaches of tropical lowland evergreen rain forest and P. murinus penetrate montane habitats or are their ranges parapatric with P. abstrusus restricted to montane forest and P. murinus to lowland forest habitats? A careful survey of these species along a transect from lowlands at the base of Pegunungan Mekongga onto the mountain range would answer the questions.

We have a reasonable sample of skins representing P. murinus from the southern highlands of the southwestern peninsula, but could not compare covariation in cranial and dental variables for these specimens with other geographic samples. The skulls are missing, damaged, or juvenile. Measurements of length of maxillary tooth row average slightly larger than is typical for samples from regions outside the southwestern peninsula, suggesting that peninsular squirrels may average larger in body size. Study of fresh, intact specimens would help resolve the morphometric relationship between the peninsular population and those to the north.

Hyosciurus : We document the boundaries of two species of ground squirrels here, but our samples of the lowland form, Hyosciurus ileile , are small and from only one place on the northern peninsula and two areas in the northern part of the central core (see fig. 35). The actual geographic distribution of this species is unknown, but we suspect its range to be much broader than is indicated by present samples. If Musser’s experience with the species is usual, it is uncommon in lowland and middle-altitude habitats, at least in central Sulawesi, and would require targeted surveys to determine the boundaries of its true distribution over the island.

Hyosciurus heinrichi , on the other hand, appears to be plentiful in its mossy forest environs. Voucher specimens reflect a distribution confined to the western mountain block of Sulawesi’s central core, a range concordant with the tree squirrel Prosciurillus topapuensis along with shrews, tarsiers, and a cluster of murid rodents (table 25). However, surveys are needed in other mountainous regions of Sulawesi —those on the east-central peninsula, for example—before this highland region can be confidentally recognized as an endemic haven for H. heinrichi , as well as the other mammal species listed in table 25.

ECOLOGY

Nests: Other than the anecdotal observations concerning nests recorded for Rubrisciurus rubriventer , we provide scanty information covering nest sites. Musser did not detect leaf nests in tree crowns during the time he worked in Sulawesi. He suspects that members of Prosciurillus use tree holes and spaces in the accumulated debris contained by emergent strangler figs as nesting sites. The two species of ground squirrels may utilize burrows but may also construct nests inside decaying tree trunks lying on the ground. Whether a particular nest site is used faithfully during the life of a squirrel or different sites are utilized depending upon food availability, population density of predators, and season is unknown.

Diet: We provide dietary information for some species, but the survey of foods is incomplete. Not all the remains of fruits and seeds found in stomachs could be identified. Only segments of the year (or fruiting cycles, which may be less or more than 12 months in the tropics) are represented in our data (table 57). Dietary components need to be studied over a period that includes fruiting cycles (and their overlap or congruence among tree species) and seasonal fluctuations (rainy versus dry seasons). The foods we report here for a particular species collected during particular time segments may change over a longer period. For example, the montane ground squirrel Hyosciurus heinrichi consumes acorns and arthropods. Production of Lithocarpus may be high some years but low during others. During times of low yield does the squirrel consume primarily arthropods or does it turn to other kinds of fruit to supplement its diet? This is but one of the questions to be answered in any future study of Sulawesi’s squirrels that focuses on diet.

Reproductive biology: Present reproductive information is meager. We can suggest litter size for the species collected by Musser (based on embryos and placental scars), but not for the others (table 58). Other questions require answers: How many litters are produced during a reproductive season? Is litter size fixed within a particular species of squirrel or does it change throughout the reproductive season depending on availability of food and nesting sites? These and other questions could be answered by comprehensive ecological studies of the different squirrel species endemic to Sulawesi.

GEOGRAPHIC REGIONS

The west-central region or western mountain block in Sulawesi’s central core: This is the landscape of coastal lowlands, foothills, high peaks (some reaching 3000 m or more), and interior valleys latitudinally extending from the Palu area in the north (about 00 ° 539S) to the southern margin of Pegunungan Latimojong in the south (about 04 ° S), and longitudinally lying between the western margin of the island bordering the Makassar Strait and Danau Poso in the east (120 ° 379E). The mountains that form most of this region, the most extensive highland landscape on the island, are the product of Pliocene orogeny related to the late Miocene–early Pliocene collision of the Banggai-Sula Australian microcontinental fragment with eastern Sulawesi (Bergman et al., 1966; Wilson and Moss, 1999; Charlton, 2000; Calvert and Hall, 2006). Prior to this orogenic period ‘‘throughout the Early Miocene and, in places, until the Middle or Late Miocene, carbonates and mudstones were deposited on a shallow-marine continental margin’’ in western Sulawesi (Calvert and Hall, 2006: 366), and until the Pliocene western Sulawesi may have ‘‘been little more than islands’’ ( Hall, 2001: 49).

Throughout this western mountain block our attention focuses on uplands above 100 m where a suite of mammals have been collected only from this region. Three species of shrews ( Crocidura ), at least two and possibly three tarsiers ( Tarsius ), a tree squirrel ( Prosciurillus topapuensis ), one ground squirrel ( Hyosciurus heinrichi ), and 14 species of murid rodents mark the western highland complex as a center of mammalian endemism. A few species are found in tropical lowland evergreen rain forests covering lower altitudes on foothills and interior valleys, but most occur at higher altitudes in montane forests (table 25).

Turning to the squirrels, we have at hand samples of Prosciurillus topapuensis and Hyosciurus heinrichi from a very few places in this western mountain block, not enough samples to reliably discern geographic patterns of population variations and their significance (see accounts of the two species). New material from unsampled montane localities from throughout the highlands would be welcome, especially samples from which molecular data can be extracted to better determine the extent of genetic continuity or discontinuity among populations.

Approximately 350 m is the lowest record for Prosciurillus topapuensis along Musser’s transect in the Danau Lindu area. The lowest altitudinal limit for the species throughout the rest of the west-central region is unknown. The few other samples are from montane forests. How low does P. topapuensis descend along the margins of the mountain block adjacent to the distributions of the lowland P. alstoni and P. weberi? Does P. topapuensis inhabit the coastal forests 100 m and below along the fringe of the Makassar Strait or is it excluded by a different species in the P. leucomus group (see above)?

Except for the Danau Lindu area and adjacent mountains, the altitudinal distributional pattern of Hyosciurus heinrichi and H. ileile is unknown throughout most of the western mountain block. Does H. ileile occur throughout lower altitudes in the region in a parapatric pattern similar to that present on Gunung Kanino, or is it absent from some places where the lower altitudes are occupied by H. heinrichi ?

The eastern central core and east-central peninsula: The distribution of squirrels in the eastern portion of Sulawesi’s central core and the east-central peninsula is currently a mystery. We suspect that surveys throughout that region would document the presence of Rubrisciurus rubriventer and Prosciurillus murinus , for these two species are recorded from the western part of the central core and the other arms of Sulawesi (see the distribution maps in figures 5 and 30). The type series of tonkeanus is the only sample of Prosciurillus alstoni we know about from the east-central peninsula, but the species probably ranges throughout the region in tropical lowland evergreen rain forest. If it does, we need to know the extent of geographic variation in expression and color of ear tufts along with other pelage traits.

Hyosciurus ileile , the lowland ground squirrel, may also be encountered in the east-central peninsula but the montane H. heinrichi of the western mountain block in the central core may not. Very little is known about the montane rodent fauna in the east-central arm of Sulawesi. To date, the only highland area sampled for small mammals is Gunung Tambusisi (01 ° 389S, 121 ° 239E) near the western end of the peninsula. Two rats, Bunomys prolatus and Maxomys wattsi , are recorded from montane habitats on the mountain (Musser, 1991). Neither species occurs on any other sampled mountain range elsewhere on Sulawesi, suggesting that the east-central peninsular mountains may harbor a unique assemblage of rodent species (and possibly representatives of other mammalian orders) restricted to mossy forest environments. That group may contain distinct species of squirrels, possibly a species of Hyosciurus , with traits different from those characterizing H. heinrichi . If H. heinrichi or a close montane relative does not occur in those peninsular highlands, possibly the montane forests are occupied by H. ileile ranging up from the lowlands.

A member of the Prosciurillus leucomus group, other than P. alstoni , might also occur in the uplands of the east-central peninsula. Outside Gunung Tambusisi, the places to expect endemic montane squirrel species would be in the other highland regions of the east-central arm: Gunung Tokala to the east of Gunung Tambusisi, Gunung Katopasa to the north, and Gunung Bulutumpu farther out on the peninsula (see the map in Whitten et al., 1987: 498). Pegunungan Pompangeo to the west of Gunung Tokala and northeast of Danau Poso is another potential montane region that should be surveyed to determine if a montane member of the P. leucomus cluster occurs there, and if H. heinrichi or another kind of ground squirrel is also present. These mountains and stretches of highlands connecting them are isolated from the western mountain block in the central core of the island.

The contention that the east-central peninsula is home to a species of macaque distinct from M. tonkeana ( Froehlich et al., 1998) , the species found in central Sulawesi, is another indication of the possible surprises awaiting future surveyors of the east-central arm.

Results of surveys for small mammals in the east-central peninsula will also be important for testing the endemic stature of the west-central region. Currently, 23 species of small mammals have been found in the western mountain block and nowhere else on Sulawesi (see above). Finding none, or only a few of these species in the east-central peninsula would certainly strengthen the position of the west-central region as an area of significant endemism in the evolutionary history of Sulawesi’s mammals.

Southeastern peninsular islands: Kabaena, Muna, and Buton, three large islands off the southern coast of the southeastern peninsula, have been inadequately surveyed for their squirrel populations. We examined two old specimens said to come from Buton. Characteristics of their skulls and fur coloration and patterning closely resemble our samples of Prosciurillus alstoni from the mainland (see that account). The population on Pulau Kabaena is represented by only Schwarz’s (1911) published description of ‘‘ Sciurus elbertae ’’ and two skulls, a very young adult and a juvenile (see the account of P. alstoni ). Although we ally the Buton and Kabaena samples with mainland P. alstoni , this postulated kinship needs to be tested by study of fresh and larger samples. We have not seen any squirrels from Pulau Muna. Rubrisciurus rubriventer and Prosciurillus murinus , which inhabit mainland lowland forests, should also occur on each of the three islands; if so, they have yet to be collected there.

Western coastal lowlands of the central core: The coastal lowlands fringing the western margins of the western mountain block in Sulawesi’s central core, from the Palu area south to the northern sector of the southwestern peninsula, is another area that requires biological surveys for small mammals. The tree squirrels Rubrisciurus rubriventer and Prosciurillus murinus should be expected there, but we are uncertain which species in the Prosciurillus leucomus group would be encountered. Prosciurillus weberi is currently represented by a few specimens collected from a small area in the southern part of the central core of Sulawesi (see the map in figure 11) and may range north along the western coastal lowland forests. On the other hand, P. topapuensis , endemic to the western mountain block, may extend into low altitudes and occupy forests all along the coast (see those species’ accounts).

GENETIC ANALYSES

Genetic analyses: New samples of all 10 of the species of squirrels we document here should include tissues from which DNA can be extracted. Results from phylogenetic analyses of gene sequences could be used to test the following. (1) Whether geographic samples of Rubrisciurus rubriventer and Prosciurillus murinus represent single species that range throughout most of Sulawesi. (2) The genetic integrity of the five species we define in the Prosciurillus leucomus group, the presence or absence of hybridization along their range borders, and phylogenetic affinities among them. (3) Whether a single species is represented by current samples we define as P. leucomus from the northern peninsula, with east-west geographic variation in pelage traits. (4) The pattern of genetic variation among insular populations of P. rosenbergii in Kepulauan Sangihe. (5) The supposition that voucher specimens of Hyosciurus ileile from the northern peninsula and central core are drawn from populations representing the same species, but one exhibiting geographic variation possibly concordant with distributions on the northern peninsula and central core. (6) Whether our samples of H. heinrichi represent a single species that exhibits morphometric variation in cranial and dental traits among montane populations.

Fresh samples of squirrels may not be necessary for analyses of gene sequences. We have identified voucher specimens by catalog numbers and the institutions where they are stored. Hairs or bits of dry tissue from those specimens might yield viable DNA to be employed in analyses to substantiate or modify our template of species-diversity.

THE LICE

Sucking lice ( Hoplopleura ): Samples of two of the Sulawesi tree squirrels, Prosciurillus weberi and P. rosenbergii , did not yield sucking lice in our survey. Fresh specimens of these two species are needed to determine if each supports its own distinctive species of louse, which would fit with the pattern of host-parasite relationship characterizing the other eight endemic species of Sulawesi squirrels. Two other squirrel hosts, P. leucomus and P. alstoni , did yield samples of Hoplopleura , and while they were sufficient to indicate that each represents a separate species unique to their hosts, the samples contained only females and nymphs. Because male lice were absent from these samples, their morphological traits could not be incorporated into our phylogenetic analyses of Sulawesi Hoplopleura . Additional samples containing male lice will be required to augment descriptions of the two species, and to provide characters needed to determine if they are part of the monophyletic clade containing the other species of lice parasitizing the endemic squirrels.

Just as the diversity of squirrel species we describe here requires testing using molecular data, so does our interpretation of anatomical differences among samples of sucking lice. Analyses of mitochrondrial and nuclear gene sequences would provide insight into the accuracy of species-diversity documented here as well as the pattern of phylogenetic relationships of the lice recovered from our cladistic analysis based on external anatomy. Some Sulawesian squirrel study skins in various museums undoubtedly still have (desiccated) louse specimens attached to them; these louse specimens could be rehydrated and, in some cases with the appropriate techniques, viable DNA may still be harvested from them.

FOREST COVER

We are not unaware that the future surveys we suggest may be impossible to implement because of past forest destruction and continuing transformation of primary forests into anthropogenic landscapes in some regions of Sulawesi. Opportunity to learn about the squirrel fauna in the southern portion of the southwestern peninsular lowlands, for example, disappeared long ago with the conversion of primary habitats to agricultural crops. Skeletal fragments from archaeological surveys constitute the primary evidence for the composition of species that once occurred there. However, intact forests remain elsewhere on the island, even outside of protected natural reserves and national parks (see the map in Supriatna et al., 1992: y37; and map 8d in MacKinnon, 1997).

EPILOGUE

By about 10 million years ago, ‘‘The Makassar Strait was still fairly wide but there is at that time the first good evidence for the emergence of land in much of Sulawesi’’ ( Hall, 2001: 52; also see Calvert and Hall, 2006), ‘‘although west Sulawesi may itself have been little more than islands until the Pliocene’’ ( Hall, 2001: 49). Between 11.4 and 10.5 million years ago a single ancestral squirrel lineage, likely accompanied by its unique species of sucking louse, crossed the Makassar Strait from the Sunda Shelf to this early Sulawesi Archipelago; from it evolved eight species of tree squirrels and two species of ground squirrels in three monophyletic clusters (genera) that now reside on the island and nowhere else. Habitats in the forest understory are occupied by the largest of the endemic squirrels, Rubrisciurus rubriventer , and the smallest, Prosciurillus murinus . Five species of Prosciurillus live in the upper forest tree canopy, and two species of Hyosciurus are terrestrial, inhabiting lowland and montane forests. The extant products of this radiation, although unique to Sulawesi, comprise a range of physical size and general ecology similar to that seen in parts of the nannosciurine squirrel fauna inhabiting the forests of Borneo, across the Makassar Strait west of Sulawesi.

The Sulawesi red squirrel is the largest of the island’s endemic suite. It has no actual counterpart on Borneo, but its occurrence in the forest understory and on the ground, along with a diet of fruit, seeds, and insects, recalls the Bornean population of Sundasciurus hippurus . Although physically smaller than Rubrisciurus (260–365 g versus 500– 860 g), the general conformation of its skull resembles that of Rubrisciurus , including procumbent incisors. Sundasciurus hippurus is active in the understory where on Borneo it is ‘‘Most often seen in small trees but sometimes travels on the ground’’ ( Payne et al., 1985: 239), and on the Malay Peninsula the squirrel is ‘‘Largely restricted to tall forest, ranging from the under-canopy to the ground’’ ( Medway, 1969: 59). Diet consists of pulpy fruit, seeds, earthworms, and insects ( Davis, 1962: 79; Payne et al., 1985: 239).

Rheithrosciurus macrotis , a Bornean endemic, and one of the giants of the species in the squirrel fauna recorded from that island (1170–1280 g), is also found mostly on the ground or low on tree trunks, but is strikingly different from Rubrisciurus in color pattern of the fur and cranial morphology (see the description of R. macrotis in Payne et al., 1985: 243; see Moore, 1959, for cranial traits), and a member of a different evolutionary lineage ( Sciurinae rather than Nannosciurinae ; see Mercer and Roth, 2003, and our classification section).

The two species in the Prosciurillus murinus group, P. murinus and P. abstrusus , are the smallest in body size among Sulawesi’s indigenous squirrels (length of head and body 5 102–150 mm). Their physical stature and vertical distribution in the forests (at least for P. murinus for which there are ecological data) are comparable to the Bornean populations of Callosciurus adamsi and C. orestes , along with Sundasciurus lowii and S. tenuis (collective lengths of head and body 5 131–183 mm); all are small-bodied, subcanopy species and include fruits, seeds, and arthropods in their diets ( Davis, 1962: 81; Payne et al., 1985).

Prosciurillus leucomus and its allies in the P. leucomus group inhabit the upper levels of primary forest on mainland Sulawesi; their canopy counterpart on Borneo is Callosciurus prevostii , although it is larger in body size than the Sulawesi species (250–500 g versus 135–210 g); C. prevostii consumes a variety of soft fruits along with insects, dietary components similar to those species in the P. leucomus group (see Davis, 1962, and Payne et al., 1985, for accounts of Bornean populations).

The two species of Sulawesi ground squirrels recall two species in two genera living on Borneo. Hyosciurus heinrichi is reminiscent of Borneo’s population of the shrew-faced ground squirrel Rhinosciurus laticaudatus . Both have very long muzzles and are about the same body size (length of head and body 5 195–233 mm for R. laticaudatus [ Payne et al., 1985: 242], 195– 240 mm for H. heinrichi ). Both species eat insects, but H. heinrichi includes Lithocarpus and other fruits in its diet while R. laticaudatus is more specialized, feeding exclusively on insects and earthworms ( Davis, 1962: 81; Medway, 1969: 62). Hyosciurus ileile , with a shorter muzzle, is physically much larger than the Bornean endemic ground squirrel Dremomys everetti (293–520 g versus 75– 185 g), but resembles that species in general conformation of the skull (see Moore and Tate, 1965: 261, for a drawing of the skull of the Chinese long-nosed ground squirrel Dremomys pernyi , which has a cranial conformation closely similar to that of the Bornean D. everetti ); both consume fruit and insects ( D. everetti also eats earthworms; Payne et al., 1985: 242).

Three groups of squirrels found on Borneo have no counterparts on Sulawesi. The giant squirrel Ratufa affinis (875–1500 g) is a member of one of the the oldest of the phylogenetic lineages within the Sciuridae ( Mercer and Roth, 2003; Steppan et al., 2004). On Borneo, R. affinis lives in tall forest, coming to the ground only to cross gaps in the forest canopy ( Payne et al., 1985: 233). Species of pygmy squirrels in Nannosciurus (length of head and body 5 62–82 mm) and Exilisciurus (length of head and body 5 62–93 mm) are the smallest in body size of the nonvolant Bornean squirrels; nothing like them is part of the modern squirrel fauna on Sulawesi. Finally, 8 genera and 14 species of flying squirrels inhabit Bornean forests ( Payne et al., 1985); if an ancient ancestral flying squirrel lineage crossed from the Sunda Shelf to the early Sulawesi archipelago, it is not represented in Sulawesi’s modern cluster of endemic sciurids.

The suite of squirrel species is just one component of the unique modern assemblage of native mammals living in the old-growth forested landscapes of Sulawesi and its nearby islands. They are joined by phalangers (two genera and at least five species), shrews (nine species of Crocidura ), several species of bats and tarsiers, macaques (six or seven species of Macaca ), rats and mice (at least 46 species in Crunomys , Echiothrix , Tateomys , Melasmothrix , Sommeromys , Margaretamys , Lenomys , Eropeplus , Haeromys , Maxomys , Bunomys , Paruromys , Taeromys , and Rattus ), anoas (two species), pigs (Sus celebensis and one or more species of Babyrousa ), and the top viverrid carnivore Macrogalidia . All are the modern products of a faunal evolution in isolation; ancestral lineages for some groups—squirrels, for example—likely date to late Miocene, at a time when the beginnings of Sulawesi likely existed as an archipelago and was already separated from the Sunda Shelf margin by a deep strait.

The man thought about the island’s unique mammals as he looked over the lake valley from his camp on a high ridge (fig. 60). Strident barking honks from three knobbed hornbills were answered by a squirrel’s scolding chatter as all squabbled over ripe figs high in the crown of a strangler fig near his tent

Forests clothing the hills and mountains stepping away from lake-side to horizon had day by day and month by month yielded some of their secrets. He remembered an Indonesian saying: ‘‘ Ada udang dibelakang setiap batu ’’; there is a crayfish behind every rock, meaning everything is not as it seems— what appears only as a rock in a stream is really the fortress of a crayfish. Across the horizon, there seemed to exist only undulating ridges of different heights mantled by tropical rain forest, but hidden beneath the canopy resides an assemblage of mammals unique to Sulawesi, and in the unseen rocky foundation beneath the forest floor is preserved a record of the island’s tectonic origin.

A lone hornbill left the fig and flew high over the lake, its noisy barking honks and loud whooshing produced by the slow beating of its wings resonated in the tropical air. The bird was quietly replaced by a second squirrel arriving at the fig from an adjacent magnolia, its presence marked only by the arcing branch and rustle of leaves in the canopy.

Encountering the squirrels had been especially satisfying to the man: the chattering tree squirrels in the upper canopy; the aggressive little brown squirrels announcing their positions low on tree trunks; the silent, wary red squirrel bounding across leaf litter; and the elusive ground squirrels, glimpsed here and there as flashes of dark brown in the deep forest.

Thick gray clouds layered the valley, resting on dark green mountaintops as if anchored by the tops of trees emerging above the canopy. Earlier just a sigh through the crown of the strangler fig, the breeze turned into gusty wind, its intensity reflected in the clouds moving rapidly across the valley. The gray clouds were soon pushed to the horizon by low, black pillows portending rain. With the first few drops came calls from squirrels and forest birds announcing the impending downpour. Sound of hard rain on leaves quickly silenced squirrel chatter and birdsong. The remaining hornbills left the strangler fig, honking and barking across the valley as they whooshed through the dark and wet sky to disappear into the brooding gloom.