Abstract: A sequence of fluvio-lacustrine and paludine deposits with thickness over 150 meters occurs in the Nihewan Basin, Hebei, China(Yuan et al.,1996; Min et al.,2006).A series of ~30m(="90 feet") thick red homogeneous clay exposured near the Sanggan River Gorge (=Shixia) was originally considered as the weathering matter of the base rocks(Barbour et al.,1926). Due to the absence of fossils, its age once was speculated to belong to the Pontian (Teilhard and Piveteau,1930) or the Late Miocene Baodean(Black et al.,1933). However, a sequence of ~30 m thick red gravel-bearing clay at the bottom of the section of Dahonggou in Shixia was named as"Dahonggou Formation"by Chen(1988). At the same time, the~12 m and～7 m thick red clays in the lower part of the Hongya Nangou and Pump Station, respectively, were also included into this Pliocene formation.A series of 20~30 m thick redish or yellow-redish alternating sandy clay and gravel beds are exposed in the Luanshigedagou near Hongya village. Yielding Hipparion and Chilotherium, this deposit was considered to belong to the"Hipparion Red Clay"with a Pliocene age(Huang et al.,1974). The sequence of~45 m thick red clay containing sandy gravel and lens of calcareous nodules in the Huabaogou near Xiyaozitou village was divided into the upper"Yuxian Formation"and the lower"Huliuhe Formation"by the composition of large fossil mammals. These two formations were respectively correlated to be equal in age to the Pliocene Jingle Formation and the Late Miocene Bahe Formation(Wang,1982). In view of the illegibility of the boundary and the lack of significant differences in the fossil faunas between these two formations, Zhang et al.(2003) incorporated them both in the Late Pliocene Yuxian Formation.A series of fluvio-lacustrine and paludine deposits exposed on the both sides of the Huliu River were named "Daodi Formation"and considered of Late Pliocene age(Du et al.,1988). The original sections used for erecting the"Daodi Fm." are located in the Laowogou near Daodi, the Nangou near Hongya, the Jiangjungou near Xiyaozitou, the Hougou near Qijiazhuang, the Yuanzigou near Yuanzi, the Xiaoshuigou near Qianjiashawa, the Lianjiegou near Beimajuan, the Niutoushan(="Pulu") near Pulu and the Danangou near Dongyaozitou. Zhang et al.(2003) figured that these short sections all belong to the"Yuxian Fm.", so they suggested that"Daodi Fm."should be abolished and attributed to the"Yuxian Fm.", whereas Cai et al.(2004) retained both"Daodi Fm."and "Yuxian Fm.". The red clay stratum from the first layer at the bottom of the Laowogou section is stillunnamed due to the uniformity of the grain size, the absence of gravels and aquatic animal remains. Based on the limited fossil mammals, the age of this stratum was thought to be the late Middle or early Late Pliocene(Zhang et al.,2003; Cai et al.,2004). The Pliocene/Pleistocene boundaries, on the other hand, have already been established in the Laowogou, Donggou, Taiergou and Niutoushan sequences (Cai et al.,2004; Zheng et al.,2006; Min et al.,2006; Cai et al.,2007). Judging from the known information, the lower part of the exposed strata in the Nihewan Basin should be attributed to the Late Pliocene Yuxian Formation or some unnamed red clays.These outcrops exist in the Sanggan River Gorge area and on the both sides of the lower reaches of the Huliu River with lithology of aeolian clay, fluvio-lacustrine red clay contained gravels and paludine sandy clay. On the both sides of the lower reaches of the Huliu River, there are 12 localities and sections, in which one or more layers of fossil mammals have been found(Fig.1). The purpose of this paper is to update a faunal list of these mammals based on the review of specimens and their localities, and also an attempt to correlate the other sections to the Laowogou section by comparing their mammalian compositions and lithological strata. Finally,a Late Pliocene biostratigraphic sequence and the environmental changes in the Nihewan Basin are discussed. A new lithological unit, the Dingshanyanchi(Dingshan Salt Lake) Formation, is established based on a set of reddish silts that is distributed in the area of Dingshan Salt Lake in the southern area of the Ulungur River, northern Junggar Basin, Xinjiang(Fig.1). At the type lo-cality, the Dingshanyanchi Formation measures 46.2 m in thickness, overlays the Middle Mio-cene Halamagai Formation, and is capped by a set of 5 m thick conglomerates of unknown age (Fig.2). Along the escarpment, the contact between the Halamagai and Dingshanyanchi formations varies from place to place. In some areas,a hiatus is distinct between the two units, whereas in others they appear continuous. We tentatively consider that a disconformity exists between the two formations and that the hiatus, if any, does not represent a major sedimentary gap. Within the basal beds of the Dingshanyanchi Formation,2.3 m above the base of the formation,a fossil assemblage was collected, consisting of two dozens of species, mainly small mammals. Of the 22 small mammals,12 are in common with the Tunggur fauna of Nei Mongol (Inner Mongolia): Alloptox gobiensis, Desmalolagus sp., Heterosminthus orientalis, Protalactaga major,P. grabaui, Democricetodon lindsayi,D. tongi, Megacricetodon sinensis,M. pusillus, Plesiodipus leei, Miodyromys sp., and Keramidomys fahlbuschi. Among identifiable larger mammals, Anchitherium and Hemicyon cf.H. stehlini are also present in Tunggur fauna. Faunal correlation suggests that the lower beds containing the fossil assemblage are middle to late Middle Miocene in age and are correlative to European MN7+8 or part of the Chinese Tunggurian. In addition to mammal fossils,8 gastropod species were collected from the same beds, of which 6 are terrestrial species and are common taxa in the Chinese loess. The lithology and faunas indicate the aeolian origin of these red beds, or at least some of them. The upper beds of Dingshanyanchi Formation, at the level of 12.4 m below the top of the formation, yield sparse and fragmentary fossils. The only identifiable specimens belong to Hipparion(Plesiohipparion) houfenense. This species, along with tooth fragments of rhinocerotids and proboscidean from the same level, suggests a Hipparion fauna with the age of Late Miocene Baodean or Pliocene. Fossils from the two levels indicate that the Dingshanyanchi Formationspans a time interval from the middle Middle Miocene to Late Miocene, or even Pliocene. However, the thickness of the formation appears thin in contrast to the time interval inferred from the fossils. Previously, the Kekemaideng Formation overlying the Halamagai Formation was considered the youngest Tertiary record in the Ulungur River area. The Kekemaideng fauna consists of mainly large mammals that are correlative to those of Tunggur fauna. We interpret that the Kekemaideng Formation, which consists of primarily conglomerates, is laterally equivalent to the lower portion of the Dingshanyanchi Formation and represents a different facies that has a more restricted geographic distribution. The macrofauna from the Kekemaideng Formation may well be attributable to its taphonomy. The age of the Halamagai fauna has remained uncertain. Different views treat it as fromMN5 to MN7+8, respectively. Given the superpositional relationship of the Dingshanyanchi and Halamagai formations, the Dingshanyanchi faunas provide an age constraint of being no later than MN6 for the Halamagai fauna. The age of the Halamagai fauna, which consists of fossils primarily from the lower beds of the formation, is probably equivalent to MN6, possibly plus part of MN5. The superpositional relationships of formations from Dingshan Salt Lake and Tieersihabahe sections allow us to compose a lithological sequence in the Ulungur River area, including four units: Tieersihabahe, Suosuoquan, Halamagai and Dingshanyanchi formations(Fig.3). The Tieersihabahe section ranges from the Late Oligocene to Middle Miocene(Meng et al.,2006), whereas the addition of the Dingshanyanchi Formation extends the sequence into the Late Mio-cene or Pliocene. Magnetostratigraphic work has been done at the Tieersihabahe section, but is pending at the Dingshanyanchi Formation. With the available litho-, bio-and magnetostratigraphic data, we provide a preliminary division and correlation for the composite sequence inquestion (Fig.3). The work cited in Fig.3 is only for convenience of correlation; we note that they differ in several aspects from each other and from other studies on the Chinese Neogene. The oldest portion of this sequence is represented by the Late Oligocene(Tabenbulukian) Tieersihabahe Formation and the lowest beds of the Suosuoquan Formation. Three biozones have been recognized within the Tabenbulukian sediments and are collectively correlative to the European PM28-30. This suggests the possibility to divide the Chinese Paleogene Land Mammal Ages into biochronologic subunits. The base of Xiejian Stage,a chronostratigraphic unit of Chinese terrestrial Neogene, is 7.25 m above the base of the Suosuoquan Formation and is correlative to the base of Miocene and Neogene. The recognition of the boundary is based on cormelation of magnetic polarity obtained from the formation with the base of C6Cn.2n, not on biological evidence, because the Xiejian Stage(Age) has not been appropriately defined by any biological marker. Similarly, the base of Shanwangian is identified at the level 56.25 m above the base of the Suosuoquan Formation in the same section, with the geomagnetic polarity age of 20.43 Ma. The Suosuoquan Formation at the Tieersihabahe section contains beds with a time interval equivalent to MN1-3, but beds of age equivalent to MN4 and part of MN5 appear missing.This suggests that hiatus between the Suosuoquan and Halamagai formations in the Tieersihaba-he section represents a gap of ca 1 Ma. However,a section in the vicinity of Tieersihabahe lo-cality that has a continuous transition from the Suosuoquan Formation to the Halamagai Formation and a newly discovered fauna correlative to those of MN4 or 5 suggest that sediments deposited during the time of MN4-5 are probably present in the Ulungur River region. The Halamagai Formation and the lower part of the Dingshanyanchi Formation probably span an interval from part of MN5 to MN7+8, or late Shangwangian to Tunggurian, as discussed above. The upper part of the Dingshanyanchi Formation is biostratigraphically less constrained. Based on the fossils and the continuous sequence observed in the field, it possibly represents sediments laid down during the time equivalent to MN9-13. If this is the case, the depositional rate of these beds must be very low. The lithology and faunas from the Late Oligocene-Miocene strata in the Ulungur River area are significant in several aspects. Given the paleomagnetic and biostratigraphic calibrations available and interpreted, the Late Oligocene faunas are coincident with the Late Oligocene warming, although the faunal structure remains similar to those of the Oligocene in the Mongolian Plateau, which had been established during the Mongolian Remoulding at the Eocene-Oligocene transition. The transition from the Oligocene to Miocene does not show significant turnover of mammals at the generic level. Many Oligocene genera survived into the Early Miocene. However, at the species level, there is a considerable turmover at the transition.The fluvial sediments and highly diverse fauna of the Halamagai Formation are in sharp contrast to those below and above. The fauna contains 49 species from almost all Neogene mammalian groups, including primates, proboscideans, and bats. Large numbers of specimens were from ungulates, carnivoreans and small mammals. Similar Middle Miocene faunas have also been known in other areas of northern China. The fauna convincingly indicate a warm, wet environment with flourishing vegetation, probably including patches of forests. Because of its coincidence with the Mid-Miocene Climatic Optimum, we entertain a causal relationship of the Halamagai fauna with the global climate. In contrast, the Miocene Qin' an section(Guo et al.,2002) at the east rim of the Tibet Plateau displays a different record, in which there are no significant lithological and faunal changes during the Middle Miocene. The lithology and faunas of the Suosuoquan and Dingshanyanchi formations suggest aeolian origin of the sediments, more so for the latter. If this proves to be true, the source areas must be different from the loess deposits that have been laid down in central China, and the formationof the Dingshanyanchi Formation is probably related to the global cooling starting from the late Middle Miocene.

摘要： 在新疆准噶尔盆地北缘乌伦古河流域建立了一个新近纪岩石地层单位：顶山盐池组。该组平行不整合于哈拉玛盖组之上，是一套以红色泥质粉砂岩为主体的细碎屑岩堆积，厚约50m，含有上下两个化石带。上化石带时代为中新世晚期或上新世；下化石带为中中新世中晚期，与通古尔期默尔根小哺乳动物群大体相当。根据岩性和化石组合，推测顶山盐池组含有风成沉积物，或主要为风成堆积。其下部地层与流域中可可买登组为同期异相沉积。通过下伏的哈拉玛盖组，顶山盐池组可以和附近地区的铁尔斯哈巴合剖面衔接，构成乌伦古河流域一个从晚渐新世到中新世晚期或上新世大体连续的地层序列。这个序列有可能包含了塔本布鲁克期至保德期的沉积物和动物群。我们将这个序列与中国以及欧洲的相同时段生物年代序列进行了初步的对比，并讨论了有关的生物年代和年代地层问题。以哈拉玛盖动物群为代表的中中新世动物群的繁盛与它们的地理分布和中中新世全球气候最佳期的温暖气候可能相关。