Comparisons of stegodon and elephantid abundances in the Late Pleistocene of southern China

H. Saegusa Museum of Nature and Human Activities, Sanda, Hyogo, Japan saegusa @nat-museum.sanda.hyogo.jp

SUMMARY: In the Pleistocene fauna of southern China, Elephas coexists with Stegodon whose molar struc- ture is more primitive and which presumably fed on more browses than the former. I surveyed the Chinese literature of ca. 180 localities in southern China where Stegodon and Elephas occur separately or co-occur, to learn if stegodon outnumber elephants in the late Pleistocene fauna of China. I found that stegodon localities significantly outnumber those with elephants. The above results for south China seem to support the Generalist vs. Specialist hypothesis proposed by Shoshani (1992). However, because of the complexities in feeding mechanisms in Stegodon and Elephas, the definitions of the terms “grazer” and “browser” do not nec- essary help to distinguish between generalist and specialist taxa. Detailed reconstruction of feeding habits of stegodons is necessary for the evaluation of the present results.

1. INTRODUCTION ber derived, specialist taxon in diverse habitat, if the animals compared share same body size Elephas maximus is the only species of the and close phylogenetic relationship. Proboscidea living in present day East Asia, but Census study of the late Pleistocene pro- Elephas coexisted with extinct genus Stegodon boscideans is worth to be conducted not only in throughout Pleistocene in southern China North America but also in other area of the (Takahashi & Namatsu 2000). Youngest world, for instance East Asia, where stegodons remains of Stegodon were found from a and elephants coexisted during Pleistocene. Neolithic site (Xiaohe Cave) in Yunnan Specialist vs. generalist hypothesis predicts Province, which is dated at 4100 bp (Ma & that stegodons must outnumber elephants dur- Tang 1992). Thus, extinction of the stegodons ing late Pleistocene in China, because ste- is a very recent event which occurred during godons are primitive compared to the elephants Holocene. in molar structure and they were living in Sympatry of two proboscidean species has forested, diverse habitat. Thus, prime motiva- been known also from the late Pleistocene of tion of this study is to test this prediction. North America, where primitive taxon However, as noted later, the results obtained American mastodons coexisted with derived through such census studies must have implica- taxon mammoths. Shoshani (1989) examined tion not only on generalist vs. specialist hypo- the number of proboscidean fossils found from thesis but also on the of the late Pleistocene Michigan and found that American mastodons extinction of megafauna in East Asia. outnumber mammoths in forested area (diverse habitat), and vice versa in open habitat. He 2. LITERATURE SURVEY argued that relative abundances of pro- boscideans observed in the late Pleistocene of 2.1 Tong-Liang case North America fits his generalist vs. specialist, hypothesis (Shoshani 1992). According to this The objective of this study is learn if ste- hypothesis, primitive, generalist taxon outnum- godons had outnumbered elephants during late 345 The World of Elephants - International Congress, Rome 2001

Pleistocene in China, where forested environ- sis, the description of the stratigraphy of the ment has been predominated. I have examined locality in the literature is examined. More than published data of southern Chinese cave fauna half of the reports are found not detailed in the in attempt to see the relative abundance of ste- description of the stratigraphy. Such published godons and elephants. data are omitted together with obviously con- The best snapshot of population densities of taminated assemblage from the data set. elephants and stegodons during the late Late middle and late Pleistocene assem- Pleistocene is available from a Paleolithic site blages are selected using radiometric dating called Tong-Liang in Sichuan, China, where data and an index fossil. In South China, 14C large number of proboscidean fossils are and uranium series methods have been done obtained from a fluvial bed dated at 21550±310 only on the site from where important hominid bp (Zhang et al. 1982). Pollens and plants fossils or Paleolithic tools have been excavated. macrofossils obtained from the bed suggest the Thus, limited number of the site will be vegetation was similar to present-day subtropi- remained in the data set if only the radiometric cal damp forest of Sichuan. In this habitat, dating is used for the selection of the localities. Stegodon outnumbered Elephas. Dominance of Other means to determine the geological age of Stegodon over Elephas observed in Tong-Liang the assemblages is necessary. Crocuta crocuta is impressive but a piece of data from only one ultima (Matsumoto 1915) is employed here as locality is not sufficient to show the general ten- an index fossil, to include more assemblages in dency of stegodons and elephants densities in our analysis, in addition to the radiometric the late Pleistocene of China. dates. Recent reexamination of several U-series dating of oldest localities of Crocuta crocuta ultima in South China suggests that those local- 2.2 Other localities in southern China ities are geologically much older (ca. 240 ka) than previously thought (Shen & Jin 1991). I examined about 130 papers on localities The last record of this species was reported where Stegodon and Elephas occur separately from in Shengxian Cave at Lishui, Jiangsu or co-occur, to learn if the tendency seen in (11,200±1000 bp.) (Ma & Tang 1992). Thus, Tong-liang is generally the same throughout the possible geological ages of assemblages the late Pleistocene of China. Ideally, census analyzed here are scattered between ca. 240 ka study must be based on the number of speci- and ca. 10 ka. mens obtained from study area, but, except for The assemblages formed under closed habi- Tong-liang, most of Chinese papers do not doc- tat are selected because concern here is whether ument the number of specimens found in local- forested environment is more favored by ities. Instead of the number of specimens, that Elephas or Stegodon. In order to exclude the of assemblages is counted here. If assemblages fossil assemblage formed under temperate or from a locality are clearly distinguished bed by steppe vegetation from the data set, the assem- bed in literatures, they are counted separately. blages containing genus Equus, which is Taxonomic status of Chinese Pleistocene regarded here as an indicator of open habitat, proboscideans must be clarified, prior to the were omitted from the data set. Consequently, examination of the proboscidean localities in the list contains only the assemblages from South China. Ideally, taxa compared must be woodland and forest environment. Another list identified at specific level. However, most of that includes assemblage presumably came the reports on excavation in China are without from densely forested area is also made. the description of the specimens and the specif- Ailuropoda, Pongo and Hylobates are consid- ic status of fossil taxa is not always clear. Thus, ered here as indicators of densely forested area we employed only the generic name in our since they are found today in such area (Corbet analysis. & Hill 1992). Using those three taxa as indica- In order to select reliable data for the analy- tors of denser forest, second list was made. 346 Comparisons of stegodon and elephantid abundances in the late Pleistocene of southern China

3. RESULTS Likewise, in southern Chinese case, primitive taxa stegodons outnumbered derived taxa Two lists obtained through above procedure Elephas in diverse forest habitat. Thus both were subjected to a statistical test. McNemar's cases in North America and China seem to sup- test was carried out here because same assem- port the generalist vs. specialist hypothesis. blages are counted two times in order to obtain However, there remain some problems con- the ratio between number of assemblages con- cerning the distinction between generalist and taining Elephas and those containing Stegodon. specialist in extinct taxa. Shoshani (1989, The null hypothesis is that the frequencies of 1992) suggested that Mammuthus is a special- the assemblages containing Stegodon and those ized grazer whereas Mammut is a primitive containing Elephas are equal in both lists. I browser. However, there still remain uncertain- employed binomial distribution for the calcula- ties concerning feeding habit of American tion of the one-tailed probability of the data mastodons. American mastodons have tradi- because when samples are small, an “exact” tionally been considered to be typical browsers, binomial test is recommended (Sokal & Rholf because of their primitive molar structure. 1995). Number of assemblages containing Following this traditional view, Haynes (1991) Stegodon is significantly greater than that con- considered American mastodons selective feed- taining Elephas in both lists. This result may er who travel between clumps of dicotyledo- suggest that subtropical and tropical forests in nous plants. However, Gobetz & Bozarth south China during the period between ca. 230 (2001) suggested that mammutid fed on grass ka to ca. 10 ka was more favored by Stegodon as well as browse based on the analysis of opal than by Elephas. phytoliths in tooth calculus of Mammut molars. Intestinal contents and coprolite from 4. DISCUSSION mastodont remains also suggest a mixture of browsing and grazing habits (Lepper et al. 4.1 Implication of the Generalist vs. Specialist 1991, Harington et al. 1993). Thus, those hypothesis recently obtained data suggest that American mastodons were mixed feeder. Until detailed census study of Pleistocene The above arguments suggest exact diet of proboscideans from China has been done the American mastodons is still hard to be together with additional radiometric studies, it determined and the primitive nature of molar may be premature to draw conclusion on the structure alone can not specify specialist or relative densities of stegodons and elephants. generalist nature of the extinct animal. Nevertheless, the above result for south China Obviously, we still need some means other than in general, and for the Paleolithic site Tong- molar structure to estimate food selection even Liang in Sichuan in particular seems to suggest in the case of well studied American mastodon, that stegodons outnumbered elephants during much more in the case of stegodons. Structural late Pleistocene and the former taxon did not difference between stegodons and elephants in decline until the very end of Pleistocene at least molar is less than that between mastodons and in forested areas of southern China. This result mammoths, and what is even worse, there has has implications for both generalist vs. special- been nearly no report on plant material associ- ist hypothesis and extinction pattern of ste- ated with stegodons remains from southern godons in East Asia. Chinese cave fauna. Until we obtain reliable According to Shoshani (1989, 1992), data picture of feeding habits of stegodons, it seems from Mammuthus vs. Mammut in North to be quite premature to consider that southern America supports generalist vs. specialist Chinese case fits generalist vs. specialist hypothesis. Primitive taxa Mammut outnum- hypothesis. bered Mammuthus in diverse habitat.

347 The World of Elephants - International Congress, Rome 2001

4.2 Implication on extinction pattern 6. REFERENCES of stegodons Corbet, G.B. & Hill, J.E. 1992. The Mammals Present result cast doubt on my previous idea of the Indomalayan Region: A Systematic that most of stegodons already declined before Review. Natural History Museum Publica- the late Pleistocene (Saegusa 1996). My previ- tions, Oxford University Press, New York. ous idea on the extinction pattern of stegodons Gobetz, K.E. & Bozarth, S.R. 2001. Impli- is provably very biased because when I pre- cations for Late Pleistocene mastodon diet pared a chronological range chart of stegodons, from opal phytoliths in tooth calculus. I employed only the fossil records accompa- Quaternary Research, 55: 115-122. nied with radiometric data and the magne- Harington, D.R., Grant, D.R., & Mott R.J. tostratigraphy of continuous geological sec- 1993. The Hillsborough, New Brunswick, tions. Late survival of stegodons into the very mastodon and comments on other end of the late Pleistocene may not be specific Pleistocene mastodon fossil from Nova to southern China but also in other areas in Scotia. Canadian Journal of Earth Science Asia. Recent radiometric study done on 30: 1242-1253. Ngangdong, central Java (Swisher et al. 1996) Haynes, G. 1991. Mammoth, Mastodonts, & suggests that stegodons still survived until the Elephants. Biology, Behavior, and the Fossil very end of Pleistocene in Java. Therefore, Record. New York: Cambridge University chronological range of stegodons and its abun- Press. dance during Pleistocene must be revised not Lepper, B.T., Frolking, T.A., Fisher, D.C., only in China but also India or Indonesia, based Goldstein, G., Sanger, J.E., Wymer, D.A., on published data, which are not used in my Oden III, J.G. & Hooge, P.E. 1991. previous revision of stegodons. Intestinal contents of a Late Pleistocene mastodont from midcontinental North 5. CONCLUSION America. Quaternary Research, 36: 120- 125 Based on literature survey on ca. 180 locali- Ma, A. & Tang, H. 1992. On discovery and sig- ties in southern China, a data set of stegodons nificance of a Holocene Ailuropoda- and elephants is constructed, to learn if Stegodon fauna from Jinhua, Zhejinag. stegodon outnumber elephants in the late Vertebrata PalAsiatica 30: 295-312. Pleistocene fauna of China. McNemar’s test on Saegusa, H. 1996. Stegodontidae: Evolutionary this data set suggests that stegodon outnumber relationships. In J. Shoshani & P. Tassy elephants in the late Pleistocene fauna of (eds.), The Proboscidea: Evolution and China. This results seem to support the Palaeoecology of Elephants and their Generalist vs. Specialist hypothesis proposed Relatives: 178-190. Oxford: Oxford Uni- by Shoshani (1992). Note, however, that versty Press. because of the complexities in feeding mecha- Shen, G. & Jin, L. 1991. U-series age of Yanhui nisms in Stegodon and Elephas, the definitions Cave, the site of Tongzi Man. Acta of the terms “grazer” and “browser” do not nec- Anthropologica Sinica, 10: 65-72. essary help to distinguish between generalist Shoshani, J. 1989. A report on the Shelton and specialist taxa. Detailed reconstruction of Mastodon Site and a discussion of the feeding habits of stegodons is necessary for the numbers of mastodons and mammoths in evaluation of the present results. Present result Michigan. Michigan Acadecian, 21: 115- also suggests late survival of stegodons into the 132. very end of the late Pleistocene may not be spe- Shoshani, J. 1992. Ecology and taxonomy: a cific, to southern China but also in other areas new approach, using Heterohyrax vs. Pro- in Asia. cavia as a case study. Israel Journal of Zoology, 38: 399-416. 348 Comparisons of stegodon and elephantid abundances in the late Pleistocene of southern China

Sokal, R.R. & Rholf, F.J. 1995. Biometry. Takahashi, K., Namatsu, K. 2000. Origin of the 3rd ed. New York: W.H. Freeman and Japanese Proboscidea in the Plio-Pleisto- Company. cene. Earth Science (Chikyu Kagaku), 54: Swisher, C.C., Rink, W.J., Anton, S.C., 257-267. Schwarcz, H.P., Curtis, G.H., Suprijo, A. & Zhang, S., Wu, Y., Yu, Q., Li, X. & Yang, X. Wadiasmoro 1996. Latest Homo erectus of 1982. Discussion of natural environment of Java: Potential Contemporaneity with Palaeolithic site of Tong-Liang. (Investiga- Homo sapiens in Southeast Asia. Science tion of the natural environment of the Tong- 274: 1870-1874. liang Paleolithic site). Vertebrata Palasia- tica, 20: 165-179.

349