deemed a distinct species and “P. grandis closely related to P. petaurista rather than to P. philippensis or P. albiventer” (Yu et al., 2006). As for P. a. chayuensis, the name has not been published with any diagnostic characteristics or designated type specimen, and thus, according to Article 16. 4 of the International Code of Zoological Nomenclature (4th Edition), it is invalid. That is to say, until now, only P. albiventer nigra has been described as a valid subspecies in China.

In this study, we implemented a typical morphometric analysis, including skull morphometrics and comparison of pelage characteristics, to study the geographic variation of P. albiventer in southwestern China. We also discussed the relationships between the differentiation of P. albiventer and its environment in southwestern China.

Materials and Methods

Data collection

The specimens used in this study are preserved in the Kunming Natural History Museum of Zoology, Kunming Institute of Zoology (KIZ), Chinese Academy of Sciences (CAS) (Kunming, China), and the Institute of Zoology (IOZ), CAS (Beijing, China). The numbers and collection localities of the specimens examined are listed in the Appendix.

A total of 45 specimens were studied, including 14 males, 12 females, and 19 specimens without recorded sex. Twenty-two specimens had intact skulls, which could be used for the statistical analyses. All specimens were considered as adults due to their erupted molars and M2 crown with longitudinal worn dentine link (Lu et al., 1987). In total, 15 cranial measurements were taken with a digital caliper with the greatest possible accuracy (0.01 mm), including: greatest length of skull (GLS), condylobasal length (CBL), basal length (BL), occipitonasal length (ONL), palatal length (PL), length of palatal bridge (PBL), length of incisive formina (LIF), breadth of incisive formina (BIF), length of upper tooth row (LUTR), length of upper molars (LUM), maximum upper molars breadth (GUMB), heigth of coronoid process of mandible (HCPM), length of lower molar row (LLMR), lower tooth row (LTR), mandibular height (MH) (Fig. 2).

In addition, four external measurements were taken from the specimen records directly, including: head and body length (HB), tail length (TL), hind foot length (HFL), and ear length (EL). Because these measurements were taken by different collectors, they were not included in the multivariate analyses or coefficients of differences (CDs).

Data analysis

The 15 cranial measurements were first log-transformed, with sexual dimorphism analysis (for the 21 intact skull samples with recorded sex) then performed. Without assuming a prior group, overall similarities and differences of the 15 cranial measurements among the samples were assessed through principal component analyses (PCA), and lastly the groups identified via PCA were assigned with names. The CDs (Mayr, 1969) between the groups were calculated using the equation: CD = (Mb − Ma) / (SDa + SDb), where Mb is the mean of population b, Ma is the mean of population a, SDa is the standard deviation of population a, and SDb is the standard deviation of population b. The PCA was performed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA).

Pelage comparisons

According to the PCA and CD analysis results, pelage characteristics of each group were described in detail, and based on significant differences between groups, a key for each was determined.

Results

Sexual dimorphism and principal component analyses

As noted above, 21 intact skull samples (10 males and 11 females) had recorded sex data. Tests of equality of group means between the male and female groups indicated no significant differences for any of the 15 cranial variables (Table I).

Since sexual dimorphism was not observed for the 15 cranial variables, we used them to conduct PCA, resulting in eigenvalues for the first three principal components of 9.52, 1.98, and 1.06, respectively, which accounted for 83.71% of total variance. Most measured characteristics had high positive loadings on the first principal component, suggesting that this component (63.46% of the total variance) represented size variation among samples. The second principal component (13.18% of variance) was strongly correlated with LUM and LLMR (factor loadings > 0.80), whereas the third principal component (7.07% of variance) was correlated primarily with BIF (factor loadings > 0.80) (Table II). Figure 3 shows the plots of P. albiventer samples on principal component factors 1 vs. 2 and 2 vs. 3, respectively.

Variable codes are given in the text and Fig. 2, significant difference level (p < 0.05).

Figure 3 indicates that the samples formed two different groups, one composed of samples of P. a. nigra from western and northwestern Yunnan, including Bijiang, Gongshan, Lianghe, Lushui, Tengchong, Yingjiang, Weixi, and Zhongdian, and the other composed of samples from Muzong, Chayu and Tibet. We calculated the CDs and compared pelage characteristics between them.

Variable codes are given in the text and Fig. 2, the extraction method used was principal component analysis, and the rotation method was Varimax with Kaiser Normalization.

Coefficients of differences (CDs)

The CD comparisons for the 15 cranial variables between P. a. nigra and the Muzong samples showed that the CDs of LUM and LLMR were larger than 1.28 (Table III).

Comparison of pelage characteristics

The specimens from Muzong (Chayu, Tibet) could be distinguished from P. a. nigra by the following pelage characteristics: 1, no or few creamy-white guard hairs on the back, compared with creamy-white guard hairs on the waist and buttock areas, but those of P. a. nigra distribute forward on the neck and head areas; 2, base of the ear area mainly rufous-brown (or grey-brown occasionally), compared with white (or light grey-brown); 3, back of the feet brown black, with a rufous-brown shade on the back of the front feet, compared with deep black brown; 4, tail not as black, but with a rufous-brown shade (Fig. 4).

Italicized values denote measurements greater than 1.28, variable codes are given in the text and Fig. 2.

Description of new subspecies

Petaurista albiventer muzongensis Li and Feng, subsp. nov.

Holotype:

IOZ 27021, ♂, adult, collected 7 August 1973, from Muzong, Chayu, Tibet, China, elevation 2,300 m.

Paratype:

IOZ 27020, ♀, adult, collected 7 August 1973, from Muzong, Chayu, Tibet, China, elevation 2,300 m.

Specimens examined:

1 ♂♂ and two (no recorded sex) pelage specimens without skulls from Xizang.

Etymology:

This new subspecies is named according to the type locality.

Diagnosis:

Few or no creamy-white guard hairs on the back, base of the ear area mainly rufous-brown (or grey-brown occasionally), back of the front and hind feet brown black, with a rufous-brown shade on the back of the front feet, tail not as black as that of P. a. nigra, but with a rufous-brown shade.

Description:

P. a. muzongensis subsp. nov. exhibits several distinct and apparently stable characteristics: (1) general dorsal color chestnut-red, deeper on the forebody and narrower on the waist and buttock areas, changing from chestnut-red to reddish-brown from the back center to the flank, with few to no creamy-white guard hairs on the back; (2) underparts ochraceous buff (hair base yellow-white, and tip yellow-brown), center area deeper, anus area grey-brown, throat white or white-grey, chin brown with a deep brown spot on its center area; (3) muzzle and sides mainly brown-black, but with a reddish shade, crown and nape nearly the same as the back, but the forehead deeper, cheeks rufous-brown, base of the ear area mainly rufous-brown (or grey-brown occasionally), orbits blackish-brown, and ear brown without tufts; (4) flanks reddish-brown, and back and belly clearly distinguished; (5) back of the front and hind feet brown-black, but with a rufous-brown shade on the back of the front feet, pads developed; (6) tail brown-black, with a rufous-brown shade.

Synonyms:

Pteromys albiventer albiventer, Gray, Illustr. Zool., pl. xviii; Petaurista petaurista nigra Wang, 1981, New Mammals from the Gaoligong Mountains (1), Acta Theriologica Sinica, Vol. 1, No. 2, 169-170.

Measurements

see Table IV for external and skull of type specimens of Petaurista albiventer muzongensis subsp. nov.

Variable codes are given in the text and Fig. 2, HB, head and body length; TL, tail length; HFL, hind foot length; EL, ear length.

External and cranial measurements of P. a. muzongensis subsp. nov. and P. a. nigra are listed in Table V, and it shows most measurement features overlap between the two subspecies with exceptions of LUM and LLMR. Figure 4 indicates the type specimens of P. a. muzongensis subsp. nov. and P. a. nigra.

Variable codes are given in the text and Fig. 2. Values are mean ± std. dev. with the range given in parentheses. HB, head and body length; TL, tail length; HFL, hind foot length; EL, ear length.

Key to the subspecies of P. albiventer

1. 1. Creamy-white guard hairs on waist and buttock areas, and distributed forward on the neck and head areas, with a white (or light grey-brown) area under the base of the ear----------------------------------------P. a. nigra
2. 2. Few or no creamy-white guard hairs on the back, base of the ear area mainly rufous-brown (or grey-brown occasionally)-------------P. a. muzongensis subsp. nov.

Figure 5 shows the geographic localities of the samples examined in the study.

DiSCUSSION

As noted above, it is unquestionable that P. albiventer should be a valid Petaurista species. In the present study, we found that the specimens stored at KIZ named as P. a. yunanensis should be classified as P. albiventer due to their dorsal pelage, which is a deep reddish-chestnut with creamy-white guard hairs scattered on half the back, but predominately absent from the shoulders and head, whereas “yunanensis” shows “numerous white-tipped hairs giving a frosted effect” (Ellerman, 1961). In their study, Yu et al. (2006) stated that P. yunanensis (tissues used in molecular study were all collected from KIZ, and thus were actually P. albiventer) was a sister group of P. albiventer (samples from Pakistan), with differences of only 6.5% in the pairwise comparison of the complete cytochrome b gene sequences. However, their discriminant function analysis and PCA morphology results (cranial measurements from specimens deposited in the American Museum of Natural History, National Museum of Natural History, and Chinese Institute of Zoology, not the KIZ specimens used in their molecular analysis) showed that P. yunanensis and P. philippensis could not be separated. Thus, their research not only demonstrated that the specimens named as P. a. yunanensis in KIZ should be a subspecies of P. albiventer apart from those of Pakistan, but also supported the results of Corbet and Hill (1992) and Thorington and Hoffmann (2005) that P. yunanensis should be a synonym of P. philippensis. Under the same circumstances, Li et al. (2012) also used the KIZ “P. a. yunanensis” specimens (which should be P. albiventer) in their cranial morphometric study of four giant flying squirrels, and demonstrated that P. albiventer (rather than P. yunanensis) should be a valid species.

The PCA results in the present study showed that the specimens were clustered into two distinct geographic groups: P. a. muzongensis subsp. nov. and P. a. nigra (Fig. 3). Furthermore, CD analyses indicated that two CDs among the 15 cranial measurements between P. a. muzongensis subsp. nov. and P. a. nigra exceeded 1.28 (Table III). According to Mayr (1969), CD values on subspecific differentiation should be equal to or larger than 1.28; thus, we concluded that P. a. muzongensis subsp. nov. should be a new valid subspecies of P. albiventer. Furthermore, pelage comparison between P. a. muzongensis subsp. nov. and P. a. nigra further sustained their valid subspecies status (Fig. 4), and classification as allopatric populations (Fig. 5). In conclusion, we strongly suggest that P. a. muzongensis subsp. nov. should be considered as a valid subspecies of P. albiventer.

Unfortunately, the type locality of P. albiventer is in Nepal (Gray, 1834), and we do not have the type specimen to compare with P. a. muzongensis subsp. nov. Based on their cyt b differences (Yu et al., 2006), however, we believe that P. a. nigra should be a different subspecies from P. a. albiventer (6.5% differences in cyt b gene sequences between them). Further studies are required to determine the relationship between P. a. muzongensis subsp. nov. and P. a. albiventer, especially in regards to cranial morphometrics and pelage characteristics.

As shown in Figure 5, the distribution of specimens examined in this study exhibited the highest species diversity not only in China (Chen, 2002), but also the world (Myers et al., 2000). The collision of the Indian and Eurasian continental plates led to the high mountains and deep gorges in the area, with these complex environments expediting intra-specific geographic variation. From the view of biogeography, P. albiventer inhabits northern Pakistan eastward to southwestern China, stretching across the whole southern Himalayan mountains. However, determining how has it adapted to such complex environments and what mechanisms are responsible for subspecies differentiation still require clarification. Further studies, especially molecular data analyses, should be performed to elucidate the relationships between subspecies differentiation, local adaptation, and geographical evolution.

Conclusion

It is concluded from the current study that P. a. muzongensis subsp. nov. is a valid subspecies of P. albiventer.

ACKNOWLEDGEMENTS

This work was conducted at the Kunming Institute of Zoology, Chinese Academy of Sciences (CAS). Special thanks to Dr. Chen Jun of the Institute of Zoology, CAS, for assistance with specimen examination. This study was supported by the National Natural Science Foundation of China (30970332, 31093430), National Special Fund on Basic Research of Science and Technology of China (2014FY110100) and Funds of KFDZ-SW-208-01, KFDZ-SW-208 and 2005DKA21402.

Conflict of interest statement

Authors declared that there is no conflict of interest.

Supplementary Material