Journal of Mammalogy, 89(2):325–335, 2008 SYSTEMATICS AND BIOGEOGRAPHY OF THE MOZAMBIQUE THICKET RAT, GRAMMOMYS COMETES, IN EASTERN CAPE PROVINCE, SOUTH AFRICA BORIS KRYŠTUFEK,* ROD M. BAXTER, WERNER HABERL, JAN ZIMA, AND ELENA V. BUŽAN Science and Research Centre, University of Primorska, Garibaldijeva 1, SI-6000 Koper, Slovenia (BK, EVB) Department of Zoology, University of Fort Hare, Private Bag X1314, Alice 5700, South Africa (RMB) Hamburgerstrasse 11/17, A-1050 Wien, Austria (WH) Institute of Vertebrate Biology, Academy of Sciences of the Czech Republic, Kvetná 8, CZ-603 65 Brno, Czech Republic (JZ) Taxonomy of thicket rats (Grammomys) is highly provisional and the genus is in a critical need of a thorough revision. We compared G. cometes from Eastern Cape Province (n ¼ 150) with G. ibeanus, G. macmillani, and the southern African G. dolichurus, applying analyses of a partial cytochrome-b (Cytb) sequence (375 base pairs), karyotypes, and cranial morphology. Genetically, G. cometes appeared to be very close to G. dolichurus (mean sequence divergence of 3.4% 6 0.8% SE), whereas G. ibeanus and G. macmillani were separated by a mean sequence divergence of 5.4% 6 1.2%. Nucleotide diversity among haplotypes was higher in G. dolichurus (p ¼ 0.0080 6 0.0010 SD) than in G. cometes (p ¼ 0.0040 6 0.0009). G. cometes and G. dolichurus showed the same diploid chromosome number (2N ¼ 52) of mostly acrocentric autosomes. None of the karyotypes reported so far for various Grammomys species match the chromosomal sets we found in Eastern Cape Province. Discriminant function analysis on 5 cranial measurements that are not affected by age variation was successful in separating G. cometes and G. ibeanus, but G. dolichurus appeared very similar to the former. In spite of their close genetic and morphological proximity, G. cometes and G. dolichurus tend toward ecological segregation and behave as distinct biological species. G. cometes is endemic to the southern African subregion and the 4 Eastern Cape Province localities are possibly isolates. Specimens were caught in the Afromontane forest above 1,000 m elevation and the lowland riverine forests dominated by Combretum caffra. Key words: chromosomes, cytochrome-b gene, Grammomys, morphology, systematics Thicket rats of the genus Grammomys are endemic to subSaharan Africa (Musser and Carleton 2005) and in the past have frequently been reported under the generic name Thamnomys (de Graaff 1981; Delany 1975; Meester et al. 1964; Smithers and Lobão Tello 1976). Twelve species are currently recognized (Musser and Carleton 2005); this number is highly provisional and the genus is in a critical need of a thorough taxonomic revision (Bronner et al. 2003; Skinner and Chimimba 2005). Two species have been reported uniformly for southern Africa (de Graaff 1981; Meester et al. 1986; Skinner and Chimimba 2005; Skinner and Smithers 1990): G. dolichurus and G. cometes. Until quite recently, both species were treated as being widely distributed in sub-Saharan Africa (Hutterer and Dieterlen 1984; Misonne 1974); the larger forms of the genus were frequently pooled under G. cometes and the smaller under G. dolichurus (Misonne 1974). Although G. dolichurus is still considered to range from the Cape region to southern Ethiopia, G. cometes is currently restricted to south of the Zambezi River and is endemic to the southern African subregion (Musser and Carleton 2005). Further north, G. cometes is replaced by G. ibeanus, a species that previously was assigned either a specific rank (Ellerman 1941; Musser and Carleton 1993) or was synonymized with G. cometes (Hutterer and Dieterlen 1984; Misonne 1974). Although G. dolichurus is widespread along the eastern coast of southern Africa as far south as Port Elizabeth, G. cometes is mainly restricted to the northeastern portion of the subregion (de Graaff 1981; Musser and Carleton 2005; Skinner and Chimimba 2005; Skinner and Smithers 1990). The known occurrence of G. cometes in Eastern Cape Province is based on a single specimen from Pirie Forest near King William’s Town (Musser and Carleton 1993), extending the previously known * Correspondent: boris.krystufek@zrs.upr.si Ó 2008 American Society of Mammalogists www.mammalogy.org 325 326 Vol. 89, No. 2 JOURNAL OF MAMMALOGY TABLE 1.—Specimens of Grammomys used in this study. Abbreviations: Cytb—cytochrome b; RSA—Republic of South Africa; GFRR—Great Fish River Reserve; m—male; f—female. For a complete list of specimens, see Appendix I. Species G. ibeanus G. cometes G. dolichurus Country Locality Kenya Tanzania Malawi Mozambique RSA RSA RSA RSA RSA RSA RSA See Appendix I See Appendix I Nyika Plateau Inhambane Kosi Bay Pirie Forest Hobbiton Fort Fordyce GFRR See Appendix I Port St. Johns southern border of the geographic range in KwaZulu-Natal (Taylor 1998) approximately 500 km further southwestward. However, recent sources (Mills and Hes 1997; Skinner and Chimimba 2005; Stuart and Stuart 1997) missed this evidence of range extension. Grammomys abounds with cryptic diversity and recent nominal species are based primarily on their karyotypes (Hutterer and Dieterlen 1984; Musser and Carleton 2005). Diagnostic characters separating the 2 southern African species also are vague, overlap, or even gradate in some areas (Meester et al. 1986; Taylor 1998). Information on chromosomal variation is limited and incomplete, but G. dolichurus is known to be polytypic (Dippenaar et al. 1983). Consequently, the identification of specimens is difficult. The morphological definition of G. ibeanus is unsatisfactory (Musser and Carleton 2005) and de Graaff (1981), in his detailed account on southern African rodents, provided no measurements for G. cometes because the paucity of material. We report in this paper on morphological, chromosomal, and molecular properties of G. cometes from Eastern Cape Province. Next, we compare data sets on G. cometes with those on G. dolichurus from the subregion and on G. ibeanus. In doing so, we aim to define G. cometes as accurately and by as many data sets as possible, to reassess characters allowing its separation from the sympatric G. dolichurus, and to assess its taxonomic relations with G. ibeanus. Furthermore, we tested the monophyly of Grammomys and phylogenetic relationships of the genus with 8 other African murine genera using the partial sequence of the cytochrome-b (Cytb) gene. Our results are largely based on our own small mammal surveys conducted between 2002 and 2005 in various forest types in Eastern Cape Province. Additional information was derived from voucher specimens in museum collections. MATERIALS AND METHODS Thicket rats were trapped between 2002 and 2005 in 5 localities: Hobbiton on Hogsback, 328339S, 268579E; Fort Fordyce Nature Reserve, 328419S, 268289E; Great Fish River Reserve, 338049–338099S, 268379–268499E; Silaka Nature Reserve near Port St. Johns, 318349S, 298269E; and the estuary Cytb Karyotype 2 4 4 1 6 m/2 f 3 m/1 f 3 m/2 f 5 3 m/1 f Morphology 19 25 30 3 1 1 112 27 10 3 6 of the Mzimvubu River near Port St. Johns, 318409S, 298359E. Trapping was performed using aluminum folding Sherman traps (23  8  9 cm; H. B. Sherman Traps, Inc., Tallahassee, Florida) and polyvinyl chloride live traps (Willan 1979) baited with rolled oats mixed with sunflower oil. Traps were set in lines with stations 20 m apart. Four traps were set at each station within a radius of ,5 m. Two traps were placed on the ground and 2 .1 m above ground level and checked twice daily. The term ‘‘trap night’’ is used to describe a trap that was set for a 24-h period. All field procedures involving handling of animals in this study were in compliance with guidelines approved by the American Society of Mammalogists (Gannon et al. 2007). Captured animals were sacrificed and the following external measurements were taken: W—body mass (g), HB—head and body length (mm from snout to anus), TL—length of tail (mm from anus to tail tip excluding terminal hair), HF—length of hind foot (mm without claws), E—ear length (mm). Voucher specimens (carded skins, skulls, postcranial skeletons, and alcoholic material) have been deposited in ZFMK and UPK (see Appendix I for collection acronyms). Among numerous voucher specimens of Grammomys from throughout the range of the genus and examined in BMNH, SMF, and ZFMK, 85 were used for comparison with material collected in Eastern Cape Province (see Appendix I for list of localities and collections). External measurements, date of collection, and localities were recorded from specimen tags. In order to avoid taxonomic confusion regarding the scope of G. dolichurus, we considered only southern African material of this species. Cytochrome-b sequence.— Sixteen Grammomys specimens (Table 1) were analyzed for variation in the mitochondrial Cytb gene. Three types of tissue were used: samples of dry skin, tissue samples from complete specimens in 70% ethanol, and tissue samples from muscles removed from freshly collected specimens and stored in 96% ethanol. Total DNA from tissue preserved in ethanol was extracted using QiaAmp DNA Blood and the Tissue Mini purification kit (Qiagen, Valencia, California) and that from museum skins was extracted on a KingFisher apparatus using the genomic DNA purification kit for King Fisher (Thermo Fisher Scientific, April 2008 KRYŠTUFEK ET AL.—GRAMMOMYS COMETES IN EASTERN CAPE Waltham, Massachusetts). A 375-base pair (bp) Cytb fragment was amplified using ‘‘universal’’ primers L14771 and H15149 (Irwin et al. 1991). Because of the degraded nature of the DNA isolated from the museum skins and alcohol-preserved specimens, amplification of larger fragments of the Cytb gene using existing primers was unsuccessful in these samples. Amplification of DNA fragments was performed on a BIORAD thermocycler (Bio-Rad Laboratories, Hercules, California) using a 50-ll reaction volume containing 2.5 mM MgCl2, 0.5 lM forward and reverse primer, 0.25 mM deoxynucleoside triphosphates, and 1 unit of Bioline Taq (Bioline UK Ltd., London, United Kingdom) in the supplied ammonium buffer. Cycling conditions included an initial step of 958C for 15 min, followed by 35 cycles of denaturation (1 min at 948C), primer annealing (1 min at 558C), and DNA extension (1 min at 728C). Polymerase chain reaction products and negative controls were checked on a 1.5% agarose gel. Double-stranded polymerase chain reaction products were purified with Wizard SV Gel and the PCR Clean-UP System (Promega, Madison, Wisconsin). Sequencing was performed on ABI PRISM 3130 Genetic Analyzer using BigDye Terminators chemistry (Applied Biosystems, Foster City, California). The program CodonCode Aligner (Ewing et al. 1998) was used to align forward and reverse sequences. The resulting consensus sequences for each individual were aligned using Clustal W (Thompson et al. 1997) in combination with Bioedit (Hall 2004). Nucleotide and amino acid composition was analyzed using the program Mega 3.0 (Kumar et al. 2004). The total number of base frequencies in each position was estimated with the program DAMBE 4.2.13 (Xia 2000; Xia and Xie 2001). Nucleotide diversity (p) was estimated using DnaSP (Rozas and Rozas 1999). We assessed phylogenetic relationships among Grammomys haplotypes (Aethomys chrysophilus as outgroup—Ducroz et al. 2001; Jansa et al. 2006) and among 8 African Murinae genera (Otomys irroratus, subfamily Otomyinae, as outgroup). Additional sequences were downloaded from GenBank (Appendix II). The hierarchical likelihood ratio test and the Akaike information criterion implemented from the program Modeltest 3.06 (Posada and Crandall 1998) were used to identify the most appropriate model of DNA substitution for the data. Under both algorithms the general time reversible model plus invariant sites and gamma distribution of variable sites (GTRþIþG— Rodrı́guez et al. 1990) were chosen to assess phylogenetic relations within the Murinae (gamma-distributed shape parameter [a] ¼ 2.05, proportion of invariable sites [I] ¼ 0.54), and the Tamura-Nei model (TrNþG) was chosen for assessing relations among Grammomys haplotypes (a ¼ 0.26). These 2 models were implemented in the Bayesian analysis (Mau et al. 1999; Rannala and Yang 1996; Yang and Rannala 1997) to reconstruct evolutionary trees (program MrBayes—Huelsenbeck and Ronquist 2001), and to calculate maximum-likelihood pairwise genetic distances (PAUP 4.0b10—Swofford 2002). Phylogenetic trees were obtained using 4 Markov chain Monte Carlo chains running simultaneously for 200,000 generations (among haplotypes) and for 1 million generations 327 (among murine genera) with the resulting trees sampled at every 10th generation. Karyotypes.— Twenty-one specimens of 2 southern African species were karyotyped (Table 1) using the preparation of in vivo bone marrow chromosomes (Robbins and Baker 1978). The slides were stained conventionally by Giemsa and Cbanded by the modified technique of Sumner (1972). Morphology.— Only specimens of adults (molars at least moderately worn) were used in morphometric comparisons. Ansell (1974) did not find significant sexual dimorphism in Zambian G. ibeanus; consequently we ignored this factor. Twelve linear measurements were scored from each skull using a vernier caliper to the nearest 0.1 mm. Definitions and acronyms are as follows: CbL—condylobasal length, RoL— length of rostrum, MxT—maxillary toothrow length (on crowns), DiL—length of diastema, FiL—length of incisive foramen, ZyB—zygomatic breadth, BcB—braincase breadth, IoC—interorbital constriction, RoB—breadth of rostrum across molars, BcH1—braincase height, BcH2—braincase height across bullae, BuL—length of bullae. Discriminant function analysis was used to assess overall phenetic similarity among a priori defined groups. Statistical tests were run in Statistica 5.5 (StatSoft, Tulsa, Oklahoma). RESULTS AND DISCUSSION Cytochrome-b sequence.— A total of 8 haplotypes was identified among the 16 Grammomys partial Cytb sequences (375 bp) and an additional haplotype was downloaded from GeneBank (Appendix II). Altogether, 68 sites (18.1%) were polymorphic, with a total of 69 mutations within which 60 sites (16.0%) were parsimony informative. No stop-codon insertions or deletions were present in the alignment. As expected under neutral evolution (Martin and Palumbi 1993) the majority of polymorphic sites were at 3rd positions (49 variable sites, 72.0% of all variable sites), followed by 1st positions (14 variable sites, 20.6% of all variable sites), and 2nd positions (5 variable sites, 7.4% of all variable sites). The Bayesian analysis on Grammomys haplotypes reached stationarity at around 200,000 generations (20,000 saved trees), so that the last 18,000 trees were used to compute a 50% majority-rule consensus tree. The 4 independent analyses converged on similar log-likelihood values, and the mean lnL score for the posterior distribution of these trees was 1,046.05. The majority consensus tree rooted with A. chrysophilus (Ducroz et al. 2001; Jansa et al. 2006) was unresolved at the basal branching (Fig. 1). Grammomys haplotypes fell into 2 major clades separated by a mean sequence divergence of 17.0% 6 2.3% SE (TrN genetic distance). The 1st clade included southern African samples and the 2nd comprised a sample from Malawi and a specimen from Tanzania. The southern African clade was further divided into 2 lineages separated by a mean sequence divergence of 3.4% 6 0.8%. These 2 clades, with a high bootstrap support (100%), are interpreted as representing 2 distinct species, G. cometes and G. dolichurus, respectively. Two geographic samples from north of the Zambezi River were represented by 3 haplotypes 328 JOURNAL OF MAMMALOGY FIG. 1.—Fifty percent majority-rule consensus trees. Above is phylogenetic tree of 18,000 trees from a Bayesian analysis of 9 cytochrome-b (Cytb) haplotypes of Grammomys rooted with Aethomys chrysophilus. The bottom consensus tree of 80,000 trees from a Bayesian analysis of 9 Cytb haplotypes of Grammomys and 10 other species from the subfamily Murinae is rooted with Otomys irroratus. Numbers above branches represent posterior probability values (.0.50). Vol. 89, No. 2 that fell into 2 clusters. These 2 groups were separated by a mean sequence divergence of 5.4% 6 1.2%, but support for branching was low (59%). Two closely related haplotypes from Malawi are ascribed to G. ibeanus, whereas Jansa et al. (2006) identified the Tanzanian haplotype as G. macmillani. Earlier allocations of the latter to G. dolichurus (Castiglia et al. 2003; Ducroz et al. 2001) must be erroneous, considering that the type locality of this species is in South Africa. Nucleotide diversity was highest between 2 haplotypes of G. dolichurus (p ¼ 0.0080 6 0.0010 SD), lowest between haplotypes of G. ibeanus (p ¼ 0.0026 6 0.0006), and intermediate among haplotypes of G. cometes (p ¼ 0.0040 6 0.0009). Contrary to this, haplotype diversity was higher in G. cometes (hd ¼ 0.750 6 0.122 SD) than in G. dolichurus (hd ¼ 0.400 6 0.237). Although such discrepancy might be indicative of a rapid population growth from a small effective population (Avise 2000), samples are by far too small for sound conclusions. To test the monophyly of Grammomys, we ran Bayesian analysis on 9 haplotypes of Grammomys and 10 species of African Murinae, representing 7 genera and 4 divisions (sensu Musser and Carleton 2005). O. irroratus (subfamily Otomyinae) was used as an outgroup. Bayesian analysis became stationary at around 1,000,000 generations (equaling 100,000 trees) so that the last 80,000 trees were used to compute a majority-rule consensus tree. The 4 independent analyses converged on similar log-likelihood values, and the mean lnL score for the posterior distribution of trees was 2,232.33. Hence, the 50% majority consensus tree provided strong support (89%) for the monophyly of Grammomys (Fig. 1). The topology of the clade for Grammomys retained the same clustering pattern of haplotypes as in the above analysis (Fig. 1). Three divisions of Murinae (Aethomys, Arvicanthis, and Hybomys) emerged as a possible sister group to the division for Grammomys and support for the monophyly of this clade was modest (76%). Of note was that the genera in Murinae did not form a monophyletic cluster and that the division for Arvicanthis emerged as paraphyletic because of the sister position of Desmomys to Dasymys (Fig. 1). Karyotypes.— All 4 geographic samples karyotyped from Eastern Cape Province (3 of G. cometes and 1 of G. dolichurus) showed the same diploid number of chromosomes 2N ¼ 52 (Fig. 2). The autosomal complement consisted of 25 pairs of gradually diminishing size. Most of the autosomes were acrocentric. Two pairs were metacentric or submetacentric (numbers 18 and 21), and 2 other pairs were submetacentric or subtelocentric (numbers 6 and 9). The X chromosome was submetacentric and extraordinarily large. Its size comprised approximately 20% of the length of the haploid complement of females. The Y chromosome probably was 1 of the acrocentric or subtelocentric chromosomes. The centromeric regions and the short arms of certain autosomes stained positively in Cbanded preparations. The Y chromosome as well as the long arm and the centromeric area of the X chromosome also were C-positive (Fig. 2). We have not found any consistent differences in the karyotype between the geographical populations or species studied. Differences were indicated in the proportion of April 2008 KRYŠTUFEK ET AL.—GRAMMOMYS COMETES IN EASTERN CAPE acrocentric and subtelocentric autosomes and in the size and centromere position in the Y chromosome. It seems that this variation only represents polymorphism between individuals within populations. Various diploid numbers were reported from different regions and for various species of Grammomys, ranging between 27 and 76 (Table 2). Part of this variation is due to the presence of supernumerary chromosomes (Civitelli et al. 1989). None of these karyotypes match the chromosomal sets we found in Eastern Cape Province. Within southern Africa, Dippenaar et al. (1983) report 2 diploid numbers for G. dolichurus: 2N ¼ 52 from Ngoye Forest and 2N ¼ 44 from Woodbush. The diploid number of G. ibeanus from Nyika Plateau, 2N ¼ 44–48 (Chitaukali et al. 2000), is distinct from the one we found in G. cometes. The diploid number of 52 chromosomes is the most frequently reported and is seemingly widespread across the range of Grammomys. However, the fundamental number of arms may vary. Matthey (1971) recorded dimorphism in the X chromosome that may be related to its large size and presumably high content of heterochromatin. Morphology.— Although a large proportion of specimens of adult G. cometes showed reproductive activity (scrotal testes, embryos, or presence of placental scars), this group was not homogeneous. Further division of the sample of adults from Hobbiton on Hogsback into 4 age groups based on abrasion of the 1st lower molar revealed 7 characters not affected by age: HF, E, MxT, IoC, BcH1, BcH2, and BuL (1-way analysis of variance: F , 2.7, P . 0.05). Details on nongeographic variability will be provided elsewhere. Overall phenetic similarity among samples of Grammomys was thus assessed by discriminant function analysis on 5 cranial variables that were not subjected to age variation. Seven operational taxonomic units were defined at this early stage of analysis: 1—G. ibeanus (Kenya and northern Tanzania), 2—G. ibeanus (southern Tanzania and Malawi), 3—G. dolichurus (Republic 329 FIG. 2.—C-banded karyotype of a male Grammomys cometes from Hobbiton on Hogsback. of South Africa), 4—G. cometes (Mozambique and KwaZuluNatal), 5—G. cometes (Great Fish River Reserve), 6—G. cometes (Hobbiton on Hogsback), and 7—G. cometes (Fort Fordyce). Specimens were classified on the basis of characters by Hutterer and Dieterlen (1984) and the above evidence derived from partial Cytb sequences. Because multivariate statistics require complete data sets and because missing data were not substituted by estimates, 208 of a total 238 specimens were included in subsequent analyses. TABLE 2.—Summary of conventional chromosomal sets reported in the genus Grammomys, except for the G. rutilans (¼ poensis) group, arranged according to descending numbers. Species name is the same as originally reported. Note that G. gazellae is currently considered to be a junior synonym of G. macmillani; G. surdaster is synonymized with G. dolichurus (Hutterer and Dieterlen 1984; Musser and Carleton 2005). Abbreviations: RSA—Republic of South Africa; ECP—Eastern Cape Province; 2N—diploid number; FN—fundamental number of chromosomal arms; FNa—fundamental number of automosomal arms. Species Locality 2N G. gazellae G. gazellae Grammomys G. caniceps G. dolichurus G. surdaster G. buntingi G. cometes G. dolichurus G. dolichurus G. ibeanus G. dolichurus G. minnae Grammomys Central African Republic Central African Republic Somalia Kenya Central Africa Katanga Ivory Coast RSA, Eastern Cape Province RSA, Port St. Johns RSA, Ngoye Forest Malawi, Nyika Plateau RSA, Woodbush Ethiopia Tanzania 6876 5671 5661 56 52 52 52 52 52 52 4448 44 32 27 FN/FNa 82/ /7075 78/ 66/ 66/ 66/ 62/58 62/58 64/ /39 Source Petter and Tranier 1975 Civitelli et al. 1989 Roche et al. 1984 Hutterer and Dieterlen 1984 Matthey 1971 Petter and Tranier 1975 Petter and Tranier 1975 This study This study Dippenaar et al. 1983 Chitaukali et al. 2000 Dippenaar et al. 1983 Hutterer and Dieterlen 1984 Fadda et al. 2001 330 JOURNAL OF MAMMALOGY FIG. 3.—Projection of specimens onto the first 2 discriminant functions derived from a discriminant function analysis on 5 log10transformed cranial variables for 7 a priori defined groups. The percentage of variance explained by a variate is in parentheses. Polygons enclose all the specimens within a group and sample acronyms are plotted on group centroids. Grammomys cometes from Eastern Cape Province (bold line): GFR—Great Fish River Reserve; HH—Hobbiton on Hogsback; FF—Fort Fordyce. Grammomys ibeanus (dotted line): N—northern group; S—southern group. Triangles—G. cometes from Mozambique and KwaZulu-Natal; crosses—G. dolichurus. Insert shows character vectors. See text for character designations and for further definition of groups. Discriminant function analysis (Wilks’ k ¼ 0.218, P , 0.0001) classified 67.3% of specimens into their proper group. The majority of misclassified specimens (52 of 58 misclassified) were placed within geographic samples but from the same species. The proportion of misclassified specimens was highest in G. dolichurus (4 of 6). A projection of discriminant function scores onto the first 2 discriminant functions (DFs), which explained 92.3% of variance in the original data set, separated samples of G. cometes and G. ibeanus along DF1 (Fig. 3). The 2 species primarily differed in molar length (high loadings for G. ibeanus) and in length of the bullae (high loadings for G. cometes). The importance of bullar size in distinguishing these 2 species already has been revealed by Musser and Carleton (2005). All 4 specimens of G. cometes from Mozambique and KwaZulu-Natal matched the Eastern Cape Province conspecifics along DF1 axis but attained higher loadings for DF2. Surprisingly, the entire sample of G. dolichurus appeared morphologically very close to G. cometes from Eastern Cape Province. To improve discrimination between species, we pooled the 2 samples of G. ibeanus and 3 samples of G. cometes from Eastern Cape Province. Discriminant function analysis on 4 operational taxonomic units classified 96.7% of specimens into their proper group (Wilks’ k ¼ 0.275, P , 0.0001) and 16 Vol. 89, No. 2 misclassified specimens (of 17 total) were allocated to a wrong species group. Results were nearly identical to the previous discriminant function analysis on 7 operational taxonomic units (not shown). Another discriminant function analysis on 3 operational taxonomic units (2 pooled samples of G. cometes and G. ibeanus, and 1 of G. dolichurus) did not improve classification results (92.3% of specimens classified properly). Again, G. dolichurus phenetically resembled G. cometes (not shown). The 2 species of Grammomys from the southern African subregion appeared similar, both cranially and dentally. Characters reported in the literature for distinguishing G. dolichurus from G. cometes involve size and color (de Graaff 1981; Skinner and Chimimba 2005; Skinner and Smithers 1990; Smithers and Lobão Tello 1976; Taylor 1998). G. cometes is on average larger but measurements overlap, which makes size (either HB or CbL) of little taxonomic help. This overlap is partly due to heterogeneity in the adult age group as defined in this study. Thus, G. dolichurus does not attain the maximum measurements of G. cometes, and very old specimens of G. cometes with heavily worn molar cusps were outside the range for G. dolichurus: CbL . 27.5 mm, ZyB . 15.0 mm. In addition, G. cometes had longer ears than G. dolichurus, with the cut-off point at approximately 18 mm (Table 3). There were several outliers among the studied G. cometes regarding ear length, but remeasuring this trait on dry skins showed that values , 18 mm were invariably underestimates and thus obviously erroneous. We thus suggest that ear length is of help in distinguishing the 2 species of Grammomys in southern Africa. The fur color (‘‘less gray’’ in G. dolichurus—cf. de Graaff 1981; Taylor 1998) appeared prone to individual variation in our material. The samples of G. cometes from Hobbiton on Hogsback and from Fort Fordyce were of both extremes (paler–darker) and the coloration of specimens of G. dolichurus from Port St. Johns was fairly dark (wood brown). A few specimens of G. cometes from Great Fish River Reserve were clearly paler than G. dolichurus, showing bright russet tints on the back, particularly the posterior. Consequently, we do not suggest coloration be used for morphological identification. Grammomys cometes is reported to ‘‘often but not always’’ have a subaricular tuft of white hairs, whereas G. dolichurus lacks it entirely (Skinner and Chimimba 2005; Skinner and Smithers 1990). We could not score this trait in the majority of BMNH specimens because the tuft is hidden under the dried ears of the prepared skins. Thus, we considered only material collected and processed by ourselves, where the ear was laid forward on 1 side of voucher skins. Presence or absence of the white subauricular tuft was subject to much geographic variation in G. cometes. The white tuft was present in all specimens from Hobbiton on Hogsback (n ¼ 108) but was absent in 37.0% of animals from Fort Fordyce (n ¼ 27). In agreement with published data (Skinner and Smithers 1990), none of 5 G. dolichurus from the Port St. Johns area had such a tuft of white hairs. Thus, this character is only of possible auxiliary value in taxonomic identification and can be misleading because of interpopulation variation, at least in G. April 2008 KRYŠTUFEK ET AL.—GRAMMOMYS COMETES IN EASTERN CAPE 331 TABLE 3.—Summary statistics for external and cranial traits of Grammomys used in this study. Given are samples size (in parentheses), mean 6 SD (upper row), and range (lower row). See Appendix I for further details on geographic origin of samples and text for character acronyms. Note that G. cometes includes material from Eastern Cape Province. Probability level: * P , 0.05, ** P , 0.01, *** P , 0.001, **** P , 0.0001, n.s. P . 0.05. The largest homogeneous sets (in parentheses) were derived from Scheffe’s test. Character HB TL HF E W CbL RoL MxT DiL FiL ZyB BcB IoC RoB BcH1 BcH2 BuL 1, G. ibeanus 2, G. cometes 3, G. dolichurus F-value Homogeneous sets (43) 112.4 6 6.83 100126 (41) 174.5 6 14.66 127203 (45) 23.7 6 1.07 21.026.2 (40) 18.5 6 2.01 16.022.0 (33) 43.8 6 6.69 3558 (61) 28.21 6 0.92 26.330.1 (60) 14.64 6 0.48 13.415.6 (65) 4.72 6 0.16 4.45.1 (60) 7.72 6 0.37 6.88.8 (60) 7.13 6 0.32 6.48.1 (59) 15.05 6 0.52 14.016.1 (55) 12.71 6 0.39 12.013.6 (59) 4.59 6 0.19 4.25.0 (59) 5.76 6 0.23 5.46.3 (56) 8.96 6 0.29 8.49.7 (56) 10.74 6 0.33 9.911.8 (57) 5.31 6 0.24 4.76.1 (143) 121.8 6 6.56 103143 (122) 173.7 6 9.25 147203 (143) 24.8 6 0.80 22.927.3 (141) 19.9 6 0.97 17.922.0 (143) 40.7 6 6.23 2765 (142) 28.11 6 0.92 26.230.1 (144) 14.51 6 0.48 13.115.7 (144) 4.42 6 0.11 4.24.8 (144) 7.66 6 0.37 6.58.6 (144) 7.18 6 0.37 5.88.0 (142) 15.05 6 0.42 14.016.4 (144) 12.70 6 0.32 11.913.4 (144) 4.76 6 0.19 4.05.2 (144) 5.63 6 0.27 5.36.1 (144) 9.20 6 0.28 8.39.8 (144) 11.07 6 0.32 10.211.9 (142) 5.63 6 0.20 5.26.1 (7) 114.1 6 4.34 110122 (5) 171.2 6 3.42 167176 (7) 24.5 6 1.47 22.027.0 (7) 17.2 6 0.35 16.917.7 (4) 33.7 6 2.50 3136.5 (7) 26.67 6 0.66 25.627.2 (7) 13.84 6 0.46 13.214.4 (7) 4.39 6 0.11 4.24.5 (7) 7.18 6 0.33 6.77.7 (7) 6.63 6 0.27 6.37.1 (6) 14.22 6 0.35 13.814.8 (7) 12.24 6 0.23 11.812.5 (7) 4.61 6 0.31 4.15.0 (6) 5.42 6 0.10 5.35.6 (6) 8.87 6 0.28 8.49.2 (6) 10.83 6 0.27 10.311.1 (7) 5.61 6 0.18 5.35.8 37.38**** (1, 3) (2) 0.14n.s. (1, 2, 3) 25.44**** (1, 3) (2, 3) 32.50**** (1, 3) (2) 6.21** (1) (2, 3) 8.62*** (1, 2) (3) 8.87*** (1, 2) 120.61**** (1) (2, 3) 6.61** (1, 2) (3) 8.06*** (1, 2) (3) 9.99**** (1, 2) (3) 6.37** (1, 2) (3) 17.70**** (1, 3) (2, 3) 15.69**** (1) (2) (3) 16.95**** (1, 3) (2) 21.62**** (1, 3) (2) 45.78**** (1) (2, 3) cometes. The taxonomic value of a subauricular white spot has already been questioned in earlier studies (e.g., Ansell 1974). Different counts have been reported for mammary formula in G. cometes (de Graaff 1981). Roberts (1951) gave the number as 6 nipples (2 inguinal pairs in addition to 1 pectoral), Meester et al. (1964) reported 4 (inguinal only), whereas Ellerman (1941) stated that it can be either. Twenty-two females from Fort Fordyce and Hobbitton on Hogsback invariably showed 3 pairs of mammae (1 pectoral and 2 inguinal). Distribution.— Musser and Carleton (2005) report G. cometes endemic to the southern African subregion in Mozambique (Smithers and Lobão Tello 1976), eastern Zimbabwe (Melseter and Umtali districts—Musser and Carleton 2005), southeastern Transvaal (Mpumalanga—Skinner and Smithers 1990), and KwaZulu-Natal (Taylor 1998). No records are available from Lesotho (Lynch 1994) and from Swaziland (Monadjem 1998). Musser and Carleton (2005) report it for Free State but Lynch (1975, 1985) and de Graaff (1981) provide no records. The northern border of the geographic range of G. cometes is on the Zambezi River in Mozambique (Smithers and Lobão Tello 1976), whereas G. ibeanus occurs further north in Malawi (Chitaukali et al. 2002) and in northeastern Zambia (Ansell 1978), but not in Mozambique (Skinner and Smithers 1990; Smithers and Lobão Tello 1976). Until the record from Pirie Forest near King William’s Town (Eastern Cape Province) was published (Musser and Carleton 1993), the southernmost localities were from KwaZulu-Natal (Ngoye Forest and the Royal Natal National Park); the report for the Mpumalunga Province was dubious (Taylor 1998). Our records are from the Amathole forest complex in the Drakensberg Range close to Pirie Forest (Fig. 4). To bridge the gap of approximately 500 km between records in Amathole and KwaZulu-Natal, we sampled in an Afromontane forest fragment located 14 km north and 4.5 km west of Ugie (Witteberg forest complex; elevation 1,500–1,600 m) in February 2005 but failed to catch any thicket rats (1,400 trap nights). Regarding the hitherto reported distribution, and considering possible mistakes in identification of animals, it is not possible at this point to conclude to what extent the Amathole population is an isolate. Clearly, more fieldwork remains to be done in various 332 JOURNAL OF MAMMALOGY FIG. 4.—Map of the eastern part of the southern African subregion showing localities of Grammomys cometes (dots), based on de Graaff (1981), Smithers and Lobão Tello (1976), and Taylor (1998). Absence of G. cometes in the Transkei forest complex 14 km north and 4.5 km west of Ugie is indicated by a triangle. Eastern Cape Province records: 1—Pirie Forest; 2—Great Fish River Reserve; 3—Hobbiton on Hogsback; 4—Fort Fordyce. References: Musser and Carleton (2005): 1; records of BK, WH, and RMB: 2–4. forest fragments along the Great Escarpment mountain ridge, which de Graaff (1981) saw as the western border of the species range (Fig. 4). G. cometes is rare in South Africa and Mugo et al. (1995) report merely 6 localities from the entire country. In Eastern Cape Province we collected G. cometes in 2 different woodland types: the Podocarpus-dominated Afromontane forest above 1,000 m above sea level (Hobbiton on Hogsback and Fort Fordyce) and the lowland riverine forest at 95 m above sea level dominated by Combretum caffra (Great Fish River Reserve). All specimens (n ¼ 156) were trapped at night, which supports earlier statements that G. cometes is nocturnal (Skinner and Chimimba 2005). The majority of rats were Vol. 89, No. 2 collected in traps set on trees: 75% in Hobbiton and 82% in Fort Fordyce; deviation from random was significant in both cases (v2 . 14, P , 0.001). G. cometes appeared to be strictly tied to a closed canopy forest. Pine plantations in Amathole are not a suitable habitat for this species (R. M. Baxter, in litt.). In Great Fish River Reserve we also failed to catch G. cometes in the arid and shrubby Valley Thicket vegetation during 8,406 trap nights that yielded 457 other small mammals. The only report of G. cometes for Eastern Cape Province published so far (Pirie Forest) is based on a single specimen deposited in the American Museum of Natural History (Musser and Carleton 1993). Another voucher specimen from the same locality that was collected in September 1897 (BMNH) has the following note on its tag: ‘‘Very rare in this part.’’ During our summer trapping sessions (December–February 2002 and 2003) in the Amathole forest region, we found G. cometes to be one of the core small mammals in the forest ecosystem, along with Myosorex cafer and Graphiurus murinus. The abundance of G. cometes in our samples of small mammals varied from 17.9% in Fort Fordyce and 18.3% in Hobbiton to 23.4% in Great Fish River Reserve. In spite of this, population densities were fairly low. In Hobbiton on Hogsback we collected 118 specimens of G. cometes in 2,324 trap nights (i.e., 5.1 specimens per 100 trap nights), in Fort Fordyce 27 specimens in 3,228 trap nights (0.8 per 100 trap nights), and in Great Fish River Reserve 11 in 496 trap nights (2.2 per 100 trap nights). Systematics and phylogeny.— In the genus Grammomys, species are loosely defined (Hutterer and Dieterlen 1984; Taylor 1998), thus one can assume a possible bias in the taxonomic interpretation of our results. However, we feel confident that the taxonomic designation of our material is consistent with the current use of the specific names. G. dolichurus is the only thicket rat reported thus far from the Port St. Johns district (de Graaff 1981) and G. cometes was confirmed recently from Pirie Forest near King William’s Town (Musser and Carleton 1993), a forest fragment that is linked directly to Hogsback as part of the Amathole forest (see Fig. 4). Identification of specimens from Mt. Nganda as G. ibeanus accords with the opinion by Chitaukali et al. (2000, 2002). G. macmillani, which was recently reported for the southern African subregion in Zimbabwe (Bronner et al. 2003; Musser and Carleton 1993, 2005), is karyologically very distinct from our material (cf. Table 2). Our nomenclature is thus consistent with that applied by earlier authors and accords also with Musser and Carleton (2005) and Skinner and Chimimba (2005). The 2 southern African thicket rats, G. cometes and G. dolichurus, appeared very similar morphologically and genetically, and also were identical in their karyotypes. Sequence divergence between them (3.4%) is so low that it hardly supports their separation into 2 distinct biological species. Genetic distance values between 2% and 11% are indicative of conspecific populations or valid species (Bradley and Baker 2001) and the lowest reported sequence-divergence value between rodent sister species is 1.2% (Baker and Bradley 2006). In spite of their close genetic and morphological proximity, G. cometes and G. dolichurus behave as distinct biological April 2008 KRYŠTUFEK ET AL.—GRAMMOMYS COMETES IN EASTERN CAPE species. Their ranges partially overlap in southern Africa (Skinner and Smithers 1990). G. cometes usually occurs sympatrically with G. dolichurus (de Graaff 1981) and the 2 species also were found syntopically (Taylor 1998). In addition, they tend toward ecological segregation: G. cometes prefers denser and more developed forest than G. dolichurus, which, in turn, also occurs in open woodland and in shrubland (de Graaff 1981; Taylor 1998). Thus, G. cometes seems restricted to the forest biotic zone, whereas G. dolichurus also occurs in the southern savannah woodland (Rautenbach 1978). Therefore, it is plausible to continue regarding G. cometes and G. dolichurus as 2 separate species. The gene tree, derived from a partial Cytb sequence, represents a minuscule fraction of the genetic history of thicket rats in South Africa. Besides, the history of a gene is not necessarily equal to the history of a species (Avise 2000) and a phylogenetic approach to taxonomy based on single gene loci can be misleading (Avise and Ball 1990). Our results show that in all 3 data sets (molecular, chromosomal, and morphological), G. cometes appears more divergent from G. ibeanus than from G. dolichurus. On the other hand, Tanzanian G. macmillani is genetically linked to G. ibeanus and not to southern African G. dolichurus. Note that Misonne (1974) in his oversimplified taxonomic division of the genus into 2 broadly sympatric and polytypic species (putting aside a very distinct G. rutilans—cf. Hutterer and Dieterlen 1984), the larger G. cometes and the smaller G. dolichurus, synonymized G. ibeanus with the former and G. macmillani with the latter. These 2 size-based species presumably differentiate regionally at the pan-African scale. Our results suggest that the actual evolutionary history of Grammomys might be just the opposite, namely an ancient regionally based divergence followed by a relatively recent speciation that also involved size displacement in young sympatric species pairs. Nucleotide diversity was highest in G. dolichurus and lowest in G. cometes. This difference suggests that G. dolichurus is evolutionarily the older species, which would concur with its putative broad pan-African occurrence (Musser and Carleton 2005). Presuming this, G. cometes likely evolved from G. dolichurus in the southern African subregion. Such a scenario is tenuous because molecular evidence can be interpreted in other ways; for example, by different demographic histories of the 2 species. The degree to which localities where G. cometes was captured in the Amathole forest complex are isolates is not known. Samples from Mozambique and KwaZulu-Natal morphologically differ from those from Eastern Cape Province. Lacking chromosomal and molecular evidence, the possible taxonomic significance of these differences remain unknown and is thus an area for further research. ACKNOWLEDGMENTS Many people helped during fieldwork by providing shelter, logistics, and hospitality, and by granting collecting permits. We thank B. Fike (Great Fish River Reserve); J. Paton, T. Burton, and many others at Hobbiton on Hogsback; T. Pienaar (Fort Fordyce), G. 333 Mphuhlu (Silaka Nature Reserve); and H. Lechmere (Ugie). Access to museum collections was granted by P. Jenkins (BMNH), R. Hutterer (ZFMK), and G. Storch and D. Kock (SMF). Chromosomal research was supported from grant LC 06073 (ME CR). An anonymous referee provided useful comments on an earlier draft. LITERATURE CITED ANSELL, W. F. H. 1974. Some mammals from Zambia and adjacent countries. Puku Supplement 1:1–49. ANSELL, W. F. H. 1978. The mammals of Zambia. National Parks and Wildlife Service, Chilanga, Zambia. AVISE, J. C. 2000. Phylogeography. The history of formation of species. Harvard University Press, Cambridge, Massachusetts. AVISE, J. C., AND R. M. BALL. 1990. Principles of genealogical concordance in species concepts and biological taxonomy. Oxford Survey of Evolutionary Biology 7:45–67. BAKER, R. J., AND R. D. BRADLEY. 2006. Speciation in mammals and the Genetic Species Concept. Journal of Mammalogy 87:643– 662. BRADLEY, R. D., AND R. J. BAKER. 2001. A test of the Genetic Species Concept: cytochrome-b sequences and mammals. Journal of Mammalogy 82:960–973. BRONNER, G. N., ET AL. 2003. A revised systematic checklist of the extant mammals of the southern African subregion. Durban Museum Novitates 28:56–106. CASTIGLIA, R., ET AL. 2003. Chromosomal and molecular characterisation of Aethomys kaiseri from Zambia and Aethomys chrysophilus from Tanzania (Rodentia, Muridae). Hereditas 139:81–89. CHITAUKALI, W. N., H. BURDA, AND D. KOCK. 2000. Rodents of the Nyika Plateau, Malawi. P. 1 in 7th International Conference Rodens et Spatium, July 10–14, 2000. České Budějovice, Czech Republic. CHITAUKALI, W. N., H. BURDA, AND D. KOCK. 2002. On small mammals of the Nyika Plateau, Malawi. Pp. 415–426 in African small mammals (C. Denys, L. Granjon, and A. Poulet, eds.). IRD, Paris, France. CIVITELLI, M. V., P. CONSENTINO, AND E. CAPANNA. 1989. Interand intra-individual chromosome variability in Thamnomys (Grammomys) gazellae (Rodentia, Muridae) B-chromosomes and structural heteromorphisms. Genetica 79:93–105. DE GRAAFF, G. 1981. The rodents of southern Africa. Butterworths, Durban, South Africa. DELANY, M. J. 1975. The rodents of Uganda. British Museum (Natural History), London, United Kingdom. DIPPENAAR, N. J., J. MEESTER, I. L. RAUTENBACH, AND D. A. WOLHUTER. 1983. The status of southern African mammal taxonomy. Koninklijk Museum voor Midden-Afrrika, Zoologische Wetenschappen 237:103–107. DUCROZ, J. F., V. VOLOBOUEV, AND L. GRANJON. 2001. An assessment of the systematics of Arvicanthinae rodents using mitochondrial DNA sequences: evolutionary and biogeographical implications. Journal of Mammalian Evolution 8:173–206. ELLERMAN, J. R. 1941. The families and genera of living rodents. Vol. II. Family Muridae. British Museum (Natural History), London, United Kingdom. EWING, B., L. HILLIER, M. C. WENDL, AND P. GREEN. 1998. Basecalling of automated sequencer traces using phred. I. [Computer software and manual.] Accuracy assessment. Genome Research 8:175–185. FADDA, C., ET AL. 2001. The rodent fauna of Tanzania: a cytotaxonomic report from the Maasai Steppe. Rendiconti Lincei Science Fisiche e Naturali 12:29–49. 334 JOURNAL OF MAMMALOGY GANNON, W. L., R. S. SIKES, AND THE ANIMAL CARE AND USE COMMITTEE OF THE AMERICAN SOCIETY OF MAMMALOGISTS. 2007. Guidelines of the American Society of Mammalogists for the use of wild mammals in research. Journal of Mammalogy 88: 809–823. HALL, T. 2004. Bioedit 7.0.0. North Carolina State University, Raleigh. HUELSENBECK, J. P., AND F. RONQUIST. 2001. MRBAYES: Bayesian inference of phylogeny. Bioinformatics 17:754–755. HUTTERER, R., AND F. DIETERLEN. 1984. Zwei neue Arten der Gattung Grammomys aus Äthiopien und Kenia (Mammalia: Muridae). Stuttgarter Beiträge zur Naturkunde, Serie A (Biologie) 374: 1–18. IRWIN, D. M., T. D. KOCHER, AND A. C. WILSON. 1991. Evolution of the cytochrome b gene of mammals. Journal of Molecular Evolution 32:128–144. JANSA, S. A., F. K. BARKER, AND L. R. HEANEY. 2006. The pattern and timing of diversification of Philippine endemic rodents: evidence from mitochondrial and nuclear gene sequences. Systematic Biology 55:73–88. KUMAR, S., K. TAMURA, AND M. NEI. 2004. Mega3: integrated software for molecular evolutionary genetics analysis and sequence alignment. Briefings in Bioinformatics 5:150–163. LYNCH, C. D. 1975. The distribution of mammals in the Orange Free State, South Africa. Navorsinge van die Nasionale Museum Bloemfontein 3:109–139. LYNCH, C. D. 1985. Mammalian distribution patterns in the Orange Free State. Navorsinge van die Nasionale Museum Bloemfontein 4:473–499. LYNCH, C. D. 1994. The mammals of Lesotho. Navorsinge van die Nasionale Museum Bloemfontein 10:177–241. MARTIN, A. P., AND S. R. PALUMBI. 1993. Protein evolution in different cellular environments: cytochrome b in sharks and mammals. Molecular Biology and Evolution 10:873–891. MATTHEY, R. 1971. Dimorphisme sexuel du chromosome X et caryotype de Grammomys surdaster Th. et Wrough. (Mammalia: Muridae). Bulletin du Museum National d’Histoire Naturelle, Section A, Zoologie 38:183–186. MAU , B., M. A. NEWTON, AND B. LARGET. 1999. Bayesian phylogenetic inference via Markov chain Monte Carlo methods. Biometrics 55:1–12. MEESTER, J. A. J., D. H. S. DAVIS, AND C. G. COETZEE. 1964. An interim classification of southern African mammals. Zoological Society of Southern Africa and S. A. Council for Scientific and Industrial Research, King William’s Town, South Africa. MEESTER, J. A. J., I. L. RAUTENBACH, N. J. DIPPENAAR, AND C. M. BAKER. 1986. Classification of southern African mammals. Transvaal Museum Monograph 5. Transvaal Museum, Pretoria, South Africa MILLS, G., AND L. HES. 1997. The complete book of southern African mammals. Struik Publishers, Cape Town, South Africa. MISONNE, X. 1974. Order Rodentia. Part 6. Pp. 1–39 in The mammals of Africa: an identification manual (J. Meester and W. Setzer, eds.). Smithsonian Institution Press, Washington, D.C. MONADJEM, A. 1998. Mammals of Swaziland. Conservation Trust of Swaziland and Big Game Parks, Mbabane, Swaziland. MUGO, D. N., A. T. LOMBARD, G. N. BRONNER, C. M. GELDERBLOM, AND G. A. BENN. 1995. Distribution and protection of endemic or threatened rodents, lagomorphs and macroscelids in South Africa. South African Journal of Zoology 30:115–126. MUSSER, G. G., AND M. D. CARLETON. 1993. Family Muridae. Pp. 501–755 in Mammal species of the world: a taxonomic and geo- Vol. 89, No. 2 graphic reference (D. E. Wilson and D. M. Reeder, eds.). 2nd ed. Smithsonian Institution Press, Washington, D.C. MUSSER, G. G., AND M. D. CARLETON. 2005. Superfamily Muroidea. Pp. 894–1531 in Mammal species of the world: a taxonomic and geographic reference (D. E. Wilson and D. M. Reeder, eds.). 3rd ed. Johns Hopkins University Press, Baltimore, Maryland. PETTER, F., AND M. TRANIER. 1975. Contribution à l’étude des Thamnomys du groupe dolichurus (Rongeurs, Muridés): systématique et caryologie. Mammalia 39:405–414. POSADA, D., AND K. A. CRANDALL. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14:817–818. RANNALA, B., AND Z. YANG. 1996. Probability distribution of molecular evolutionary trees: a new method of phylogenetic inference. Journal of Molecular Evolution 43:304–311. RAUTENBACH, I. L. 1978. A numerical re-appraisal of the southern African biotic zones. Bulletin of Carnegie Museum of Natural History 6:175–187. ROBBINS, L. W., AND R. J. BAKER. 1978. Karyotypic data for African mammals, with description of an in vivo bone marrow technique. Bulletin of Carnegie Museum of Natural History 6:188–210. ROBERTS, A. 1951. The mammals of South Africa. Trustees of ‘‘The Mammals of South Africa’’ Book Fund, Johannesburg, South Africa. ROCHE, J., E. CAPANNA, M. V. CIVITELLI, AND A. CERASO. 1984. Caryotypes des rongeurs de Somalie. 4. Premiere capture de rongeurs arboricoles de sous-genre Grammomys (genre Thamnomys, Murides) en Republique de Somalie. Monitore Zoologico Italiano 7:259–277. RODRÍGUEZ, F., J. F. OLIVER, A. MARÍN, AND J. R. MEDINA. 1990. The general stochastic model of nucleotide substitutions. Journal of Theoretical Biololgy 142:485–501. ROZAS, J., AND R. ROZAS. 1999. DnaSP version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformatics 15:174–175. SKINNER J. D., AND C. T. CHIMIMBA. 2005. Mammals of the southern African subregion. Cambridge University Press, Cambridge, United Kingdom. SKINNER, J. D., AND R. H. N. SMITHERS. 1990. The mammals of the southern African subregion. University of Pretoria, Pretoria, South Africa. SMITHERS, R. H. N., AND J. L. P. LOBÃO TELLO. 1976. Check list and atlas of the mammals of Moçambique. National Museums and Monuments of Rhodesia, Salisbury, Zimbabwe. STUART, C., AND T. STUART. 1997. Field guide to the mammals of southern Africa. Struik, Cape Town, South Africa. SUMNER, A. T. 1972. A simple technique for demonstrating centromeric heterochromatin. Experimental Cell Research 75:304– 306. SWOFFORD, D. L. 2002. PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4.0b10. Sinauer Associates, Inc., Publishers, Sunderland, Massachusetts. TAYLOR, P. 1998. The smaller mammals of KwaZulu-Natal. University of Natal Press, Pietermaritzburg, South Africa. THOMPSON, J. D., D. G. HIGGINS, AND T. J. GIBSON. 1997. Clustal W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Research 22:4673–4680. WILLAN, K. 1979. Design and field test of a modified small mammal live-trap. South African Journal of Zoology 14:81–84. XIA, X. 2000. Data analysis in molecular biology and evolution. Kluwer Academic Publishers, Boston, Massachusetts. April 2008 KRYŠTUFEK ET AL.—GRAMMOMYS COMETES IN EASTERN CAPE XIA, X., AND Z. XIE. 2001. DAMBE: data analysis in molecular biology and evolution. Journal of Heredity 92:371–373. YANG, Z., AND B. RANNALA. 1997. Bayesian phylogenetic inference using DNA sequences: a Markov chain Monte Carlo method. Molecular Biology and Evolution 14:717–724. Submitted 28 February 2007. Accepted 19 September 2007. Associate Editor was Christian T. Chimimba. 335 1); Kambai Forest Reserve, Usambara Mts., Muheza District, Tanga Region, 335 m (SMF, n ¼ 1); Mbeya, Poroto Mts. (BMNH, n ¼ 3); Irambo Mission, Poroto Mts., Tukuyu District, Mbeya Region, 2,050 m (SMF, n ¼ 3). Malawi: Nyika Plateua (BMNH, n ¼ 13); Mt. Nganda, Nyika National Park, 1,531–2,255 m (SMF, n ¼ 17). Grammomys dolichurus (n ¼ 10).— Republic of South Africa: Ngoye Forest Reserve (BMNH, n ¼ 1); Ngqeleni, 800 m (BMNH, n ¼ 2); KwaZulu-Natal, no locality (BMNH, n ¼ 1); Port St. Johns, 150 m (BMNH, n ¼ 1); Port St. Johns: the estuary of the Mzimvuba River, near sea level (UPK, n ¼ 1); Port St. Johns, Silaca Nature Reserve, below 100 m above sea level (UPK, n ¼ 4). APPENDIX I List of specimens used in this study. Collection acronyms: BMNH—Natural History Museum London; SMF—Senckenberg Forschungsinstitut und Museum, Frankfurt; UPK—University of Primorska, Koper; ZFMK—Zoologiches Forschungsinstitut und Museum A. Koenig, Bonn. Grammomys cometes (n ¼ 154).— Mozambique: Boguno, Inhambane (BMNH, n ¼ 3; includes type of G. cometes). Republic of South Africa: Kosi Bay, KwaZulu-Natal (SMF, n ¼ 1); Pirie Forest, King William’s Town (BMNH, n ¼ 1). Great Fish River Reserve near Grahamstown, 95 m aove sea level (ZFMK, n ¼ 10); Hobbiton on Hogsback, the Amathole Mts., 1,200 m above sea level (ZFMK, n ¼ 112); Fort Fordyce, the Katberg Mts., 1,040 m above sea level (UPK, n ¼ 27). Grammomys ibeanus (n ¼ 74).— Kenya: Mt. Nyiru (BMNH, n ¼ 3; includes type of G. lutosus); Mt. Elgon, 2,740 m (BMNH, n ¼ 4); Uraguess (¼ Gargues) Mts. (BMNH, n ¼ 3); Nanyuki (BMNH, n ¼ 1); Molo (BMNH, n ¼ 1; type of G. ibeanus); Aberdare National Park (BMNH, n ¼ 7). Tanzania: Ngorongoro (crater), 1,700 m, 2,300 m (BMNH, n ¼ 15); Momela Lakes, 30 km NE Arusha (BMNH, n ¼ 1); Lushoto, near Wilhelmastal (BMNH, n ¼ 1); Mazumbai Estate, Lushoto District, Tanga Region, 1,400 m above sea level (BMNH, n ¼ View publication stats APPENDIX II List of species of Muridae whose cytochrome-b sequences were used in this study. GenBank accession numbers are in parentheses. Taxonomy and nomenclature follow Musser and Carleton (2005). Murinae.— Grammomys division: Grammomys macmillani (AF14121). Grammomys cometes (all from Republic of South Africa): Fort Fordyce, haplotypes H1 and H2 (EU275248, EU275249); Great Fish River Reserve, haplotype H3 (EU275250); Hobbiton on Hogsback, haplotype H4 (EU275251). Grammomys dolichurus: Port St. Johns, haplotype H5 (EU275252); Silaca Nature Reserve, haplotype H6 (EU275253). Grammomys ibeanus: Malawi, Mt. Nganda, Nyika National Park, haplotypes H7 and H8 (EU275254, EU275255). Arvicanthis division: Arvicanthis somalicus (neumanni) (AF004574), Arvicanthis niloticus (AF004572), Mylomys dybowskii (AF141212), Lemniscomys macculus (AF141208), Lemniscomys striatus (AF141211), Desmomys harringtoni (AF 141206). Dasymys division: Dasymys incomtus (AF141217), Dasymys rufulus (AF141216). Hybomys division: Hybomys univittatus (AF141219). Aethomys division: Aethomys chrysophilus (AF004587). Otomyinae.— Otomys irroratus (AF141222).