The lobopodian Paucipodia inermis from the Lower Cambrian Chengjiang fauna, Yunnan, China ÈM XIAN-GUANG HOU, XIAO-YA MA, JIE ZHAO AND JAN BERGSTRO Hou, X.-G., Ma, X.-Y., Zhao, J. & BergstroÈm, J. 2004 09 15: The lobopodian Paucipodia inermis from the Lower Cambrian Chengjiang fauna, Yunnan, China. Lethaia, Vol. 37, pp. 235±244. Oslo. ISSN 0024-1164. New specimens of Paucipodia inermis Chen, Zhou & RamskoÈld, 1995, are described from the Lower Cambrian Chengjiang LagerstaÈtte in Haikou, Kunming. Details not previously seen in the Chengjiang material appear to be caused by early diagenetic processes. Some features not previously observed in Palaeozoic lobopodians include details of the dermomuscular sac, body cavities, contents of the gut, possible paired ventral nerve ganglia, and a rasping or biting apparatus with teeth. The latter implies a fundamental difference from onychophorans and rules out an ancestral position for Palaeozoic lobopodians. The supposed tail is shown to be the head, and it is shown that this animal possessed nine pairs of lobopods rather than six, as originally stated. The family Paucipodiidae n. fam. is introduced. & Chengjiang fauna, Kunming, LagerstaÈtte, Lobopodians, Lower Cambrian, Paucipodiidae. Xian-Guang Hou [xghou@ynu.edu.cn], Xiao-Ya Ma [dawn_xiaoya@hotmail.com] and Jie Zhao [biofeature@hotmail.com], Yunnan Research Center for Chengjiang Biota, Yunnan University, Kunming 650091, People's Republic of China; Jan BergstroÈm [jan.bergstrom@ nrm.se], Department of Palaeozoology, Swedish Museum of Natural History, P.O. Box 50007, SE-104 05 Stockholm, Sweden; 16th December 2002, revised 17th November 2003. The Chengjiang LagerstaÈtte was named after the locality where it was discovered, Maotianshan (Maotian Hill) in Chengjiang County, eastern Yunnan. The fauna occurs in the lower part of the Yu'anshan Member of the Lower Cambrian Heilinpu (formerly Qiongzhusi) Formation (Zhang & Hou 1985). Correlations within the Lower Cambrian are generally dif®cult because of the rarity of fossils and the provinciality of the fauna. The correlation of the Lower Cambrian in eastern Yunnan indicates that the stratigraphic position of the Chengjiang LagerstaÈtte is in the transitional part between the lower and middle part of the Lower Cambrian (Hou & BergstroÈm 1997, p. 6, ®g. 3). The level is probably equivalent to the Atdabanian in Siberian terms (Hou & BergstroÈm 1997). The Chengjiang faunal community therefore developed during the major biotic radiation known as the Cambrian Explosion, and it represents one of the oldest well-preserved animal assemblages, giving us an insight into this early radiation of multicellular animals. Six species of xenusian lobopodians have been described from the Chengjiang LagerstaÈtte. The morphology is much more varied than within the extant onychophoran lobopodians (Hou & BergstroÈm 1995; BergstroÈm & Hou 2001). Of the six species, Paucipodia inermis was described by Chen et al. (1995) based on four incomplete specimens from Maotian- shan. In March 2001, this site was formally protected as a National Geological Park of China. Further collecting at Maotianshan is therefore not permitted. Our collection includes 20 specimens (9 with counterparts, and 3 on the same slab) of Paucipodia inermis collected since 1999 in Haikou, about 30 km south of Kunming and about 50 km northwest of Maotianshan (Fig. 1). All specimens are from the Anshan section near the village of Mafang. None was found in the nearby section at Ercaicun, but this could be a result of less intensive searches and poorer preservation. The terminology used in this paper follows that of Hou & BergstroÈm (1995). The narrower end of the body (anterior to the ®rst lobopod pair) is here called the `head', and the wider end of the body (posterior to the ninth lobopod pair) is called the `tail'. The plane along which the fossils have split differs between specimens, and the two rock slabs on either side of this surface are referred to as `the lower part' and `the upper part' respectively. Lobopod pairs are numbered L1±L9 (left side of the animal) and R1±R9 (right side) from anterior. All specimens dealt with in this paper are reposited in the Yunnan Research Center for Chengjiang Biota, Yunnan University in Kunming, China. The specimens studied are labelled RCCBYU 10183 to 10189. In the following, they are just referred to as no. 10183, etc. DOI 10.1080/00241160410006555 # 2004 Taylor & Francis 236 X.-G. Hou et al. Fig. 1. Locations of the sections at Haikou, Kunming. Haikou lies at the west shore of Lake Dianchi, about 50 km from Maotianshan, Chengjiang County, the locality at which the Chengjiang fauna was ®rst discovered in 1984. Numbers indicate two important fossil sites: 1, Ercaicun section, located near Ercaicun village, and 2, Anshan section, near Mafang village. These two sections, about 3 km apart, are known from ®nds of Lower Cambrian vertebrates. Preservation Butter®eld (2002, p. 166) stated that ªcalcium phosphate is conspicuously absent in the Chengjiang faunaº. A thin cover of a bluish-whitish mineral that covers many specimens may look like a phosphate, but a search for phosphorous has yielded negative results. Since the rock consists of only quartz and muscovite (Ulf HaÊlenius, see below), it is possible that the mineral cover consists of an aluminosilicate. Such mineralization would explain the occasional preservation of such details as muscle ®bres. The phosphatization and subsequent weathering scenario which Butter®eld (2002); Vannier & Chen (2002) suggested LETHAIA 37 (2004) caused the sediment-like appearance of gut in®ll does not apply to the Chengjiang situation (Hou et al. in prep). Several specimens are laterally compressed. They can be straight (Fig. 2G) or curled up (Fig. 2B, C, E). Others are dorso-ventrally compressed; these are typically straight (Fig. 2A, F). Figure 2A shows only the last two leg pairs symmetrically exposed on both sides of the body, whereas the others are obviously hidden in the matrix under the body. In one specimen (not illustrated) the legs extend below the body, being separated from it by one millimetre of sediment. Some laterally compressed specimens show interesting relationships to bedding. Thus, Figures 2B, C (no. 10185) show a tightly coiled specimen with head and tail crossed and preserved at different levels, with a vertical distance of about one mm. Figure 2D (no. 10186) shows a specimen in which the vertical distance between head and tail is 4±5 mm. A similar relationship is seen in the holotype and paratype (Chen et al. 1995, ®gs 1a±d, g; 2a±b, e), causing the false impression that the complete individual had only six leg pairs and no trunk extension behind the last pair. The degree of compaction has not been calculated for Chengjiang material. However, judgements based on the Burgess Shale have been uncritically applied. Thus, compaction of the Chengjiang sediments has been stated to be 10:1 (Chen et al. 1995, p. 276; Butter®eld 2002, p. 166). However, the Burgess Shale is a ®ssile shale, whereas the Chengjiang sediment is a massive rock dominated by ®ne-grained quartz (Ulf HaÊlenius, see below). Therefore ¯attening of the fossils is to a large degree a result of compaction and disappearance of the organic matter, not just of compaction of the quartzose sediment. If the compaction ratio was 10:1, the original vertical distance in no. 10186 (Fig. 2D) would have been about 30 mm. In fact, the part of the above-mentioned specimen involved in the vertical step to is only about 17±18 mm, which indicates that the compaction rate was only 6:1 or less. Systematic palaeontology Class Xenusia Dzik & Krumbiegel, 1989 Order Archonychophora Hou & BergstroÈm, 1995 Family Paucipodiidae new family Diagnosis. ± Archonychophorans with few segments, ®ne annuli, no dorsal sclerites, and two claws (cf. diagnosis for Order, see Hou & BergstroÈm 1995, p. 6). Included genus. ± Paucipodia Chen, Zhou & RamskoÈld, 1995. LETHAIA 37 (2004) New lobopodian specimens from the Chengjiang fauna 237 Fig. 2. Paucipodia inermis Chen, Zhou & RamskoÈld. & A. RCCBYU 10184, a dorso-ventrally preserved specimen with only the two last lobopod pairs exposed, the others probably concealed in the matrix. See Fig. 4A for explanation. & B. BBCBYU 10185a, dorso-laterally compressed coiled specimen, the lower part, showing the head. See Fig. 4C for explanation. & C. BBCBYU 10185b, upper part of the same, showing the tail. See Fig. 4C for explanation. & D. RCCBYU 10186a, the lower part of an incomplete specimen preserved in different beddings. & E. RCCBYU 10183, a very complete individual showing almost all of the features of Paucipodia. See Fig. 4B for explanation. & F. RCCBYU 10187, a dorso-ventrally preserved incomplete specimen, showing important details of the anterior part. See Fig. 4E, F, G for explanation. & G. RCCBYU 10188, an incomplete laterally preserved specimen, showing some details of the lobopods and claws. See Fig. 4D for explanation. 238 LETHAIA 37 (2004) X.-G. Hou et al. Known range. ± Lower Cambrian. Discussion. ± The Paucipodiidae differ from other known Cambrian xenusians in being devoid of sclerites and of lobopod differentiation, and in having an unusually ®ne annulation and a low number of lobopods. At present, eleven bodily preserved species of Cambrian xenusians have been found, named, and assigned to nine genera: Aysheaia pedunculata Walcott, 1911 (see Whittington 1978); A.? prolata Robison, 1985; Hallucigenia sparsa (Walcott, 1911) (see Conway Morris 1977 and RamskoÈld 1992a, b); Hallucigenia fortis Hou & BergstroÈm, 1995; Xenusion auerwaldae Pompeckj, 1927 (see Dzik & Krumbiegel 1989); Microdictyon sinicum Chen, Hou & Lu, 1989; Luolishania longicruris Hou & Chen, 1989; Onychodictyon ferox Hou, RamskoÈld & BergstroÈm, 1991; Cardiodictyon catenulum Hou, RamskoÈld & BergstroÈm, 1991; Paucipodia inermis Chen, Zhou & RamskoÈld, 1995; and Hadranax augustus Budd & Peel, 1998. Except for Aysheaia and Paucipodia, Cambrian xenusians possessed paired sclerites along the trunk. The systematic position of Paucipodia based on the characters recognized by Hou & BergstroÈm (1995, p.12) lies adjacent to Luolishania among the xenusians. Luolishania and Paucipodia both conform to the diagnosis of the order Archonychophora, and they have serially similar segmentation and undifferentiated lobopods. However, the Luolishaniidae are multi-segmented archonychophorans with one type of segmentally arranged small, rounded sclerites, arranged not in pairs but usually in sets of three in each segment. There were numbers of annuli between each two sclerites decreasing towards both ends of the animal. Paucipodia is a pauci-segmented and pauciclawed archonychophoran with very ®ne annuli best seen when the animal is bent, and it is devoid of dorsal sclerites. Therefore, Paucipodia can not belong to the family Luolishaniidae as de®ned above, and we erect a new family Paucipodiidae for it. Occurrence. ± All material dealt with here is from Anshan section, Mafang village, Haikou, Kunming. Previously described material was collected in Chengjiang. Habit and anteroposterior orientation According to the original description, this species is fairly small (Chen et al. 1995, p. 277±280), whereas some of our specimens are among the largest of all xenusians. This could be taken to indicate a speci®c difference. However, there is no morphological difference between the Chengjiang and Haikou specimens, and the size difference may be because of ontogenetic or some ecological factor. We can correct the primary description of Paucipodia inermis (Chen et al. 1995, p. 276) in two important respects. First, we note that the description was based on incomplete specimens, and only six leg pairs were recognized of the nine actually present in the species. Despite this correction, P. inermis still has the lowest number of lobopods among Cambrian xenusians. The reason why Chen et al. (1995) did not see the whole animal may be that part of it is deposited at another level, just as in no. 10186 (Fig. 2D). Second, RamskoÈld & Chen (1998) discussed the anteroposterior orientation of lobopodians. Unfortunately, their arguments are not applicable to lobopodians with a simple morphology such as P. inermis, in which head and tail are of about the same length, and the legs become shorter towards both ends. If they are correct about the orientation of Microdictyon and the Burgess Shale Hallucigenia, with the narrowest end being an anterior proboscis (see Dzik & Krumbiegel 1989 and Hou & BergstroÈm 1995, p. 7), and if this also applies to P. inermis, they are incorrect in their orientation of the latter. In fact, the ®nd of teeth in the narrow end of two specimens (Figs 2D, F, 3A, B) indicates that this end is anterior. The reddish scar seen by Chen et al. (1995) in the opposite end of the animal is thus the anus (Figs 2A, 3D). Genus Paucipodia Chen, Zhou & RamskoÈld, 1995 Type and only known species. ± Paucipodia inermis Chen, Zhou & RamskoÈld, 1995. Emended diagnosis. ± Paucipodiids with nine pairs of lobopods, each with two distal claws. Head and tail elongated and tapering. Paucipodia inermis Chen, Zhou & RamskoÈld, 1995 Emended diagnosis. ± Same as for the genus, by monotypy. Description The margin of all specimens of this soft-bodied animal is virtually sleek and clear. Only at higher magni®cation is there a slight wavy appearance indicating an annulation. The body is elongated and tapers towards both ends. This animal is one of the largest Cambrian xenusians so far known. A complete specimen is 106.9 mm long and up to 4.3 mm wide (Figs 2A, 4A). A fragmentary, not illustrated specimen is up to 5.2 mm wide, which by extrapolation would corre- LETHAIA 37 (2004) New lobopodian specimens from the Chengjiang fauna 239 Fig. 3. Details of Paucipodia inermis. & A. The head of RCCBYU 10187, showing the teeth and brain; & B. The head of RCCBYU 10186, showing the hard part of the head; & C. One part of the trunk of RCCBYU 10188, showing the annuli and muscle ®bers; & D. The tail with anus in the lower part of RCCBYU 10186b; & E. the three-dimensional gut of RCCBYU 10184; & F. one lobopod of RCCBYU 10188, showing the claws and the central canal. spond to a length of 125 mm. Measurements of the six specimens studied are shown in Table 1. The body extends beyond the lobopods in both ends. The head is much narrower than the tail, and both ends taper towards the termination. The head is 2±3 times longer than wide. This ratio is less variable than the corresponding ratio for the tail. The appendage-bearing part is elongate and tapers slightly towards both ends. It can be divided into lobopod inter-spacings, measured as the distance between the centres of each circular area of lobopod attachment site, and we may name them S1 (the ®rst inter-space: between RL1 and RL2) to S8 (RL8 to RL9) from the head in turn. It is obvious that the lobopod inter-spacings are unequal in length, being largest in the middle. The trunk carries very ®ne annulations, about 45 annulations per millimetre, and the annulations near the head seem to be a little more numerous than in the middle part of the trunk. The trunk lacks dorsal plates. The lobopods are fairly long and taper successively toward the end. Those in the middle of the row are longer and sturdier than those towards the ends. Like the body, the lobopods are ®nely annulated. Each lobopod carries two claws. The claws are seen in most specimens, especially in no. 10188 (Figs 2G, 3F). They show a brown colour that is notably darker than the surrounding matrix. There is a depression at the base of each claw, which may be caused by tissue collapse. The claws are situated just on the distal tip. The claw length appears to be related to the length of the individual lobopod. The claws are commonly seen as straight, but are slightly curved when seen from the side as in the 6th left lobopod in no. 10185a (Fig. 4C). In this lobopod the claws are ¯exed strongly inward, at an angle of almost 90 degrees to the limb axis. In one specimen (Fig. 4B), claws of lobopods L1 and L9 point anteriorly, while the claws of L8 is directed posteriorly, and the claws of L5 are directed oppositely. In one specimen (Figs 2G, 4D), claws of one lobopod are straight, while those of the opposite podomere point to the posterior. The attachment of each lobopod is seen as a virtually circular structure in the body wall, either a depression in the lower part (Fig. 2B) or a convexity in the upper part (Figs 2A±C, 4A, C). This is evidence that the lobopods were round in cross section. Lobopods underneath the lower part are preserved 240 LETHAIA 37 (2004) X.-G. Hou et al. Fig. 4. Camera lucida drawings of Paucipodia inermis Chen, Zhou and RamskoÈld. & A. RCCBYU 10184. & B. RCCBYU 10183. & C. RCCBYU 10185a, b, a compositional drawing. & D. RCCBYU 10188. & E±G. RCCBYU 10187, showing general view, details of the part between RL2± RL3 and the head respectively.Abbreviations: ac = alimentary canal; al = attachment site of left legs; ar = attachment site of right legs; bw = body wall; c = claws; cc = central canal in leg; cl = coelom; h = head; g = gut; ga = ganglion; gw = gut wall; l = left lobopod; nc = nerve cord; r = right lobopod; s = posterior scar; t = tail. not as mounds, but as depressions in the trunk. The reason for this apparent paradox is that the compaction pressure, which acts vertically from above, will force overlying sediment down into the collapsing Table 1. Measurements (in mm) on Paucipodia inermis RCCBYU number Preservation state Length Width L/W ratio 10183 10184 10185 10186 10187 10188 Average Complete Complete Nearly complete Incomplete Partial Partial Complete * The average of RCCBYU 10183-5. 82.2 106.9 78.1 >40.0 >35.0 >38.5 89.1* 4.0 4.3 3.5 3.0 2.9 4.9 3.8 20.6 24.9 22.3 ± ± ± 22.6* lobopod, resulting in a depression at the attachment site. This phenomenon has been described in arthropods by Hou and BergstroÈm (1997, p. 9). It is common in Chengjiang fossils, and is observed in virtually all xenusian specimens (Chen et. al. 1995, p. 280). Some impressions in lateral specimens seem to indicate that the lobopods had elongated insertions into the trunk. As concluded by RamskoÈld (1992a) for Hallucigenia sparsa this is caused by rotation of the trunk. Such rotation has also been noticed in P. inermis (Chen et al. 1995, p. 280). In no. 10188 there is seen an elongated central canal within each lobopod (Fig. 4D), preserved as a dark, longitudinal strand about 1 mm wide, extending along the entire length of the lobopod. The colour of it is LETHAIA 37 (2004) fainter than that of the alimentary canal, and there are some weak traces in other specimens. The end of the distal band seems to connect with the claws, and the band appears to be connected proximally with the alimentary canal. However, this may be an optical illusion. The signi®cance is considered in the discussion. The body surface is moulded in great detail, showing that it is smooth. Annulations are narrow, some 4±5 in a millimetre in a large specimen (Fig. 4A±D). They have low relief, are best visible in inner bends and are not observable everywhere. The dermomuscular sac just inside the integument is clearly visible in some specimens. There is a tight outer layer of circular muscles. For each annulus there are some 4±8 muscle ®bres, which makes some 30 to 40 per millimetre (Fig. 3C). This is particularly seen at the base of lobopod L1 in the counterpart of no. 10186, where there is also a regular width undulation along each ®bre. Occasionally there is a red stain along the ®bres. Staining may be concentrated to one part of each annulus, so that in practice it shows the annulation pattern even when the surface is ¯at. Inside of the sheath of circular muscles there are muscles extending in other directions. As far as can be seen they do not form any muscle layer, but occur as more isolated strands. Similar but much sparser (14±15 ®bers per millimetre) ®bres of circular muscles were described in Kerygmachela kierkegaardi from the Lower Cambrian Sirius Passet of North Greenland (Budd 1999). The head retains interesting structural details. The anterior margin is ¯at in front of the vaulted main part of the head and appears to have been stiff (Fig. 3A, B). A short distance behind it is a hard median structure that gives off a pair of stiff branches, which diverge backwards (Figs 2F, 3A). In two specimens (nos. 10186±87) they form a V to U-shaped ®gure with the open side towards the rear (Fig. 2D, F). This stiff structure is interpreted as a skeletal element of the head, probably related to the mouth apparatus. An unillustrated specimen (no. 10189) has a similar element, and in addition a transverse row of a few teeth. Only two of these were partly preserved, and a third possibly indicated by its basis. The largest preserved tooth was triangular and 0.2 mm long, lacking its tip. The complete length would have been around 0.3 mm. They pointed backwards and were curved towards the dorsal side. Unfortunately the teeth were largely lost after documentation, and only basal parts remain. However, teeth are also seen in no 10187 (Figs 2F, 3A). These are smaller and form an arch with the open side to the rear. Also these teeth point to the rear. The arch is some 0.3 mm across, and it contains about 15 teeth in a complex row. The teeth are some 0.01±0.02 mm wide and 0.05±0.08 mm long. A coloured band extends backward from this arch. It New lobopodian specimens from the Chengjiang fauna 241 contains two rounded cross sections of what may be two larger teeth. There are possibly a few small teeth also at the posterior end of this band. The structures were surrounded and succeeded by a wide opening in the alimentary canal interpreted as a pharynx, or the anterior part of it (Figs 3A, 4E, G). Behind the pharynx, the alimentary canal is well preserved in all specimens, showing a longitudinal band that can be traced along the entire body. Most specimens display three-dimensional parts of the alimentary canal. The anus is situated on the ventral side a short distance in front of the posterior body end (Fig. 3D). In one specimen it is represented by a diffuse red `scar' (Figs 2A, 4A), in another by a fairly distinct brown ring. Chen et al. (1995) described the same structure without identifying it. It is interesting to note the presence of certain structures around the alimentary canal. In some specimens (Figs 2F, 4F, and others not illustrated), we can ®nd that the colour of the body changes regularly from the centre to the sides, forming distinct longitudinal bands. The colours are yellow, orange red, matrixcoloured and violet, respectively, from the midline to the margin. In no. 10187, under the microscope we can see round-paired features fairly regularly distributed along the sides of the alimentary canal. The structures are too small to be seen in Figures 2F or 4F, and they are weakly coloured. The colour can be described as salmon pink. It is fainter that the colour of the alimentary canal and more similar to the colour of the body wall. Occasionally a row of tiny concave spots are seen just along the margin. Interpretation of morphology and anatomy The dermomuscular sac with an outer continuous sheath of circular muscles and a more irregular pattern of other muscles inside has similarities with that in modern onychophorans, which have a continuous sheath of circular muscles, and inside it longitudinal muscles separated into broader bands. Tardigrades do not have any dermomuscular sac, no doubt because of their minute size. A series of structures indicates a certain complexity. Larger and smaller teeth appear to have been arranged around the mouth opening in a way parallel to that in petromyzontid agnathans. The structures appear to have constituted a head skeleton, on which muscles leading to a rasping or biting mouth could have attached. The second skeletal arch is approximately at the same place as a pair of swellings. These are more or 242 X.-G. Hou et al. less surrounded by thin violet lines and possibly indicate the position of a pre-oesophageal brain. A rounded expansion around the tooth arch may be interpreted as a pharynx (Figs 3A, 4G). The alimentary canal may not have been in a strictly ®xed position within the trunk during life (cf. Figs 2A± C, E, 4A±C). There is evidence for a movement toward the concave body side during the ventral ¯exure of the trunk, seemingly because of a slight antero-posterior tension in the alimentary canal. The variable position of the canal may then indicate the lack or poor development of mesentery to hold it in a ®xed position (Chen et. al. 1995, p. 280). The outline of the alimentary canal is unbroken, so there were no diverticula. The width of the canal is not constant along the body. There seems to be a trend that it is widest in the middle part of the body. The longitudinal colour bands along the body show a consecutive pattern (Figs 2F, 4F). We suggest that the yellow colour is the alimentary canal, the narrow bordering orange red the gut wall (and adjoining tissues), and the violet colour the outer body wall composed of ectoderm and mesoderm, respectively. The thinness of the body wall is indicated both by the narrowness of the colour band and, more directly, by the remains seen in different places as a raised margin of the animal. The wide bands between the gut wall and the body wall have the light colour of the matrix. This lightness indicates a large body cavity (pseudocoel, coelom, or a mixture thereof). Bands centrally in the lobopods are also described in Chengjiang Hallucigenia (Chen et. al. 1995, pp. 280± 281) and Microdictyon, where they are well preserved (RamskoÈld & Hou 1991). They are also present in Onychodictyon (Hou et al. 1991) and Cardiodictyon. These structures are clearly common to many, perhaps all, Cambrian xenusians. As concluded by BergstroÈm & Hou (2001), the most likely explanation is that the band represents a body cavity such as the mixocoel as in onychophorans, tardigrades (Eibye-Jacobsen 1996/ 97, p. 214) and arthropods. Both onychophorans and tardigrades possess paired ventral or ventro-lateral nerve cords, with pairs of nerve ganglia arranged along the body in the tardigrades. We suggest that the repeated, paired violetcoloured structures seen in the specimens represent the paired nerve ganglia. In addition, a pair of larger oval structures in the expanded part of the head in no. 10187 (Figs 3A, 4G) may be pre-oesophageal, dorsally expanded brain lobes. Relationships The Cambrian xenusians have certain similarities to LETHAIA 37 (2004) modern onychophorans and tardigrades. The presence in onychophorans and xenusians of a dermomuscular sac with a well-developed circular sheath provides no argument, since this is a character found in many `worm' groups. The lack of an inner sheath of longitudinal muscles in xenusians, but the presence of such a sheath (although incomplete) in onychophorans, is a dissimilarity. The lack of frontal appendages (antennae) in xenusians makes it dif®cult to derive onychophorans directly from xenusians. Even the second appendage in onychophorans, the jaw, may have no counterpart in xenusians. Instead, the arrangement with a transverse arch of teeth is entirely different from the construction in onychophorans with their paired leg-like jaws. The skeletal bars are different from anything in onychophorans. Also, there is no similarity to tardigrades in any of these respects, except for the absence of antennae. Tardigrades, with a terminal to ventral mouth, have one pair of stylets in the mouth. The ring of mouth sclerites in anomalocaridids distinguishes them from Cambrian xenusians. Gut in®ll and feeding habits The alimentary canal of P. inermis is preserved partly completely ¯at, partly three-dimensionally (Fig. 3E). The latter condition has not been described before for any lobopodian. This raises the question of the nature of the ®ll, and what it can say about feeding behaviour. Chemical analyses we have made on the sedimentlike gut contents in Naraoia have shown the presence of less than 1% of elementary phosphorous, if anything. The natural element composition is not distinguishable from that of the surrounding matrix, except that there may be an enrichment of iron. X-ray and clay swelling analyses performed by Ulf HaÊlenius, Swedish Museum of Natural History, Stockholm, show that the sediment is composed of quartz and muscovite. The idea that sediment-like in®ll in guts is weathered calcium phosphorite rather than engulfed sediment (Butter®eld 2002; Vannier & Chen 2002) therefore is incorrect. Grainy, coloured matter, often together with skeletal fragments, in guts of some other fossils, including P. inermis (no. 10186), is probably phosphatized organic contents in the gut. Chen et al. (1995) noted that all their four individuals of Paucipodia inermis were found in close association with Eldonia, and three of our specimens were also entombed with Eldonia. This would indicate that this lobopodian fed on dead, or possibly live, Eldonia bodies. This suggestion gains strength from the fact that similar associations have been observed for different lobopodians. The claws are commonly seen extended in the direction of the lobopod (Fig. LETHAIA 37 (2004) New lobopodian specimens from the Chengjiang fauna 243 Fig. 5. Reconstruction of Paucipodia inermis. The head is to the left. 3F), but in one case they are turned inward. They were thus poorly adapted for walking, and seemingly better adapted for climbing on carcasses or even on live sessile animals. We therefore conclude that P. inermis in all probability was foremost a carrion-feeder, perhaps specialized on consuming Eldonia and its relatives. The long head and the teeth would have given it a mode of feeding on carrion similar to that of myxinoid or petromyzontid vertebrates. However, the occasional occurrence of sediment ®ll, almost unknown in other xenusians, indicates occasional feeding on organic material in sediment. This could have been more or less selective, and could be regarded as sediment-feeding. P. inermis would thus have been somewhat omnivorous, regularly feeding on decaying animals when such were available, and on organic material in sediment at other occasions. This opens the interesting possibility that xenusians, like probably arthropods (Hou & BergstroÈm 1997), arose from generalized sediment-feeders and developed other feeding strategies only within the xenusian group. Paucipodia may therefore be primitive at least in this particular respect. It is possible that sediment-eating in P. inermis may have been connected with the unusual body pro®le, with limb length decreasing forwards from the middle (Fig. 5) rather than increasing as in other xenusians. Conclusion The realization that individual Chengjiang fossils underwent early diagenetic processes that preserved much ®ner details than have previously been reported is a reason to search for such material. Understanding of the real nature of the gut contents may revolutionize our interpretation of feeding in early Cambrian animals, which in turn can be a tool for revealing the systematic position, the ancestry, and the history of limb development. Regarding xenusian lobopodians, the new details make it possible to sharpen the analysis of possible relationships. It is a complication that this analysis shows that they are even more different from modern lobopodians than we had thought, and that an ancestral-descendant relationship between any one of them may be hardly possible although a relationship cannot be excluded. Acknowledgements. ± The study was supported by the National Natural Science Foundation of China (No. 40272017), the Ministry of Science and Technology of China (G2000077702, 2002CB714007 and Pandeng 95-Zhuan-01), the Department of Science and Technology of Yunnan Province (No. 2001D0002R), the Natural Science Foundation of Yunnan Province (No. 2002D0006m) and the Swedish Museum of Natural History. Analyses of the mineral composition of the sediment were kindly performed by Ulf HaÊlenius, and Figure 5 was drawn by Pollyanna von Knorring, both at the Swedish Museum of Natural History. Specimen no 10185 (Fig. 2B, C) was collected by Miss Lucy Siveter when she visited the Anshan section, Haikou in September 1999. Jerzy Dzik and Graham Budd supplied helpful reviews. References BergstroÈm, J. & Hou, X.-G. 2001: Cambrian Onychophora or Xenusians. Zoologischer Anzeiger 240, 237±245. Budd, G. 1998: Stem group arthropods from the Lower Cambrian Sirius Passet fauna of North Greenland. In Fortey, R.A. & Thomas, R.H. (eds): Arthropod Relationships, 125±138. Systematic Association Special Volume Series 55. Chapman and Hall, London. Budd, G. 1999: The morphology and phylogenetic signi®cance of Kerygmachela kierkegaardi Budd (Buen Formation, Lower Cambrian, N Greenland. Transactions of Royal Society of Edinburgh, Earth Sciences 89, 249±290. Butter®eld, N. 2002: Leanchoilia guts and the interpretation of three-dimensional structures in Burgess Shale-type fossils. Paleobiology 28, 155±171. Chen, J.-Y., Hou, X.-G. & Lu, H.-Z. 1989: Early Cambrian netted scale-bearing worm-like sea animal. Acta Palaeontologica Sinica 28, 1±16 (in Chinese with English summary.). Chen, J.-Y., Zhou, G.-Q. & RamskoÈld, L. 1995: A new Early Cambrian onychophoran-like animal, Paucipodia gen. nov., from the Chengjiang fauna, China. Transactions of Royal Society of Edinburgh, Earth Sciences 85, 275±282. Conway Morris, S. 1977: A new metazoan from the Burgess Shale of British Columbia. Palaeontology 20, 623±640. Dzik, J. & Krumbiegel, G. 1989: The oldest `onychophoran' Xenusion: a link connecting phyla? Lethaia 22, 169±181. 244 X.-G. Hou et al. Eibye-Jacobsen, J. 1996/1997: New observations on the embryology of the Tardigrada. Zoologischer Anzeiger 235, 201±216. Hou, X.-G. & BergstroÈm, J. 1995: Cambrian lobopodians ± ancestors of extant onychophorans? Zoological Journal of the Linnean Society 114, 3±19. Hou, X.-G. & BergstroÈm, J. 1997: Arthropods of the Lower Cambrian Chengjiang fauna, southwest China. Fossils and Strata 45, 1±116. Hou, X.-G. & Chen, J.-Y. 1989: Early Cambrian arthropod-annelid intermediate sea animal, Luolishania gen. nov. from Chengjiang, Yunnan. Acta Palaeontologica Sinica 28, 207±213 (in Chinese with English summary.). Hou, X.-G., RamskoÈld, L. & BergstroÈm, J. 1991: Composition and preservation of the Chengjiang fauna ± a Lower Cambrian softbodied biota. Zoologica Scripta 20, 395±411. Pompeckj, J.F. 1927: Ein neues Zeugnis uralten Lebens. PalaÈontologische Zeitschrift 9, 287±313. RamskoÈld, L. 1992a: Homologies in Cambrian Onychophora. Lethaia 25, 443±460. RamskoÈld, L. 1992b: The second lobopod row of Hallucigenia discovered. Lethaia 25, 221±224. RamskoÈld, L. & Chen, J.-Y. 1998: Cambrian lobopodians: Mor- LETHAIA 37 (2004) phology and phylogeny. In Edgecombe, G.D. (ed.): Arthropod fossils and phylogeny, 107±150. Columbia University Press. RamskoÈld, L. & Hou, X.-G. 1991: New Early Cambrian animal and onychophoran af®nities of enigmatic metazoans. Nature 351, 225±228. Robison, R.A. 1985: Af®nities of Aysheaia (Onychophora), with descriptions of a new Cambrian species. Journal of Paleontology 59, 226±235. Vannier, J. & Chen, Y.-J. 2002: Digestive system and feeding mode in Cambrian naraoid arthropods. Lethaia 35, 107±120. Walcott, C.D. 1911: Cambrian geology and paleontology, II. 5. Middle Cambrian Annelids. Smithsonian Miscellaneous Collections 57, 109±144. Whittington, H.B. 1978: The lobopodian animal Aysheaia pedunculata Walcott, Middle Cambrian, Burgess Shale, British Columbia. Philosophical Transactions of the Royal Society of London 284, 165±197. Zhang, W.-T. & Hou, X.-G. 1985: Preliminary notes on the occurrence of the unusual trilobite Naraoiain Asia. Acta Palaeontologica Sinica 24, 591±595 (in Chinese with English Summary).