We describe a new avian taxon (Sanshuiornis zhangi gen. et sp. nov.) from Middle Eocene black oil shales in the Huayong Formation of Guangdong Province, south China. The specimen consists of a distal tibiotarsus and a complete foot with tarsometatarsus and pedal digits in articulation. A preliminary phylogenetic analysis does not resolve the affinities of the fossil, but the bones show resemblances to some “ciconiiform” birds. The peculiar hypotarsus morphology, which is block-like and exhibits four cristae, resembles that of the early Eocene Rhynchaeites, which is a stem group representative of the Threskiornithidae. The new Chinese fossil has, however, proportionally longer legs than Rhynchaeites and its phylogenetic affinities probably cannot be resolved without further material.
The fossil record of birds from the Paleogene of China is very scanty (Mayr 2009). To date, six genera and six species have been described. Wanshuina lii and Songzia heidangkouensis, which came from the Paleocene of Anhui and the early Eocene of Hubei respectively, were both referred to the Rallidae (Hou 1990, 1994). A very large species, Zhongyuanus xichuanensis from the early Eocene of Henan, was assigned to the anseriform family Gastornithidae (Hou 1980). Eogrus aeola from the middle Eocene of Inner Mongolia (Wetmore 1934) is a representative of the Eogruidae, which were widespread in the Paleogene of Asia (Kurochkin 1976; Clarke et al. 2005; Mayr 2009). Eociconia sangequanensis from the middle Eocene of Xinjiang (Hou 1989) was considered to belong to the Ciconiidae (storks). Minggangia changgouensis from the late Eocene of Henan (Hou 1982) was referred to the Threskiornithidae (ibises), while Stidham et al. (2005) suggested affinities to Rallidae (rails).
Here we describe a new fossil bird from the black oil shales of the Huayong Formation of the Sanshui Basin in Guangdong, south China (Fig. 1), which, based on morphological comparisons and the result of a phylogenetic analysis, shows closest affinities to some “ciconiiform” birds. The Paleogene strata in the Sanshui Basin are extensive, and numerous geological studies over the past years have made the Sanshui basin one of the best studied nonmarine Paleogene strata and biotas from China (Hou et al. 2007). The deposits represent freshwater lacustrine environments, and abundant plants, invertebrates and some vertebrates have been recovered (Li et al. 2005; Hou et al. 2007). The age of the formation was regarded as either middle or late Eocene according to different authors (Zhang 1999; Hou et al. 2007), but it is most likely middle Eocene (Li et al. 2005; Meemann Chang, personal communication 2011).
Material and methods
The avian material described in this paper is represented by a single incomplete specimen (IVPP V18116) that consists of a distal right tibiotarsus in articulation with a complete right foot, with tarsometatarsus and pedal phalanges in articulation.
The phylogenetic analysis is based on the data set of Mayr and Clarke (2003), with some erroneous scorings modified according to Mayr (2005, 2011) (see Appendix 1: 1). We added one new character to the data matrix, which is defined as: Character 149, second and third phalanx of the fourth toe, not as follows (0), strongly reduced and less than half the length of the fourth phalanx (1), as before but with the proximal phalanx strongly reduced to be nearly equal in length with the second and third phalanx (2). Twenty extant species were scored from materials in the collection of the Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Science (IVPP) (Table 1), and the remaining taxa were scored from Mayr (2005) (see Appendix 1:2, 3).
The phylogenetic analysis was carried out with Paup* 4.0b10 (Swofford 2002), with three characters ordered (characters 55, 71, 91). We performed heuristic searches, with 1000 replicates of random stepwise addition (branch swapping: tree-bisection-reconnection), holding one tree at each step. Branches were collapsed to create polytomies if maximum branch lengths were zero. 500 bootstrap replicates were conducted with the same settings as in the primary search.
Institutional abbreviation.—IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Science, Beijing, China.
Aves Linnaeus, 1758
cf. “Ciconiiformes” Bonaparte, 1984
Genus Sanshuiornis nov.
Type species: Sanshuiornis zhangi sp. nov.; see below.
Etymology: After Sanshui Basin where the holotype specimen was collected.
Diagnosis.—As for the monotypic type species.
Sanshuiornis zhangi gen. et sp. nov.
Etymology: The species name is dedicated to Xianqiu Zhang, a local geologist, who discovered the locality.
Holotype: IVPP V18116, a distal right tibiotarsus in articulation with a complete right foot, with tarsometatarsus and pedal phalanges in articulation.
Type locality: Foshan, Guangdong Province.
Diagnosis.—Medium-sized bird, which differs from all other known avian taxa in the block-like hypotarsus (Fig. 3), which exhibits four cristae and three well-delimited sulci.
The new taxon differs from Rhynchaeites in having the tarsometatarsus much longer, without marked tubercle on mid-section of proximal rim of medial parahypotarsal fossa, first phalanx of first pedal digit proportional longer, and four hypotarsal cristae more prominent. It differs from Eociconia sangequanensis in being proportionally much smaller, and having trochlea metatarsi III in midline of tarsometatarsus. It is distinguished from Eogruidae in that the tarsometatarsus is much shorter, trochlea for second toe not reduced, and digit I well developed.
Anatomical terminology follows Baumel and Witmer (1993). The measurements (Table 2) represent the maximum length along the longitudinal axis; concerning the ungual phalanges, the distance between the extensor tubercle and the tip of the claw was measured; concerning the trochleae, the maximum width is measured.
Tibiotarsus.—The distal end of the right tibiotarsus is preserved in medial view and lacks a portion near the medial condyle. This bone resembles the tibiotarsus of extant Threskiornis aethiopicus (Threskiornithidae, African Sacred Ibis), in both medial and lateral views. The dorso-plantar width of the shaft near the proximal end of the medial condyle is similar to that of the Eocene ibis Rhynchaeites messelensis (SME-ME 3577, see Mayr 2009: fig. 7.8). In medial view, the lateral condyle extends a little further distally than the medial one, but it is shorter in the dorso-plantar direction. A prominent medial epicondyle is located on the upper border of the medial epicondylar depression. The distal rim of the medial condyle is flattened and bears a small notch. Clarke et al. (2005) listed this character as one of six unambiguously optimized synapomorphies supporting a (Psophiidae + Eogruidae + Gruidae) clade, but this character has a wider distribution, and occurs also in, e.g. Threskiornithidae, Podicipedidae, Phoenicopteridae, and some Ciconiidae (Mayr and Clarke 2003).
On the reverse side of the slab, the lateral aspect of the lateral condyle is exposed, and some anatomical details are discernible. The lateral epicondylar depression is shallow and less developed compared with the corresponding depression of the medial condyle. The lateral epicondylar depression appears to be divided into two small depressions by a small crest. The distal rim of the lateral condyle has a subcircular outline, while the outline of the medial condyle is elliptical. A subcircular lateral condyle is also found in Rhynchaeites messelensis (Mayr 2009: fig. 7.8), Minggangia changgouensis (Hou 1982), some extant Threskiornithidae (Min Wang, personal observation), and Ibidopsis hordwelliensis (Harrison and Walker 1976: pl. 6E), but is absent in Eogrus aeola (Wetmore 1934: fig. 5) and extant Gruidae. Stidham et al. (2005) indicated that this primitive character would be consistent with a transfer of Minggangia changgouensis and Ibidopsis hordwelliensis to Rallidae (Harrison and Walker 1976), but the character clearly has a wider distribution (it is, for example, present in extant Threskiornithidae) and the rallid affinities of these two species are questionable (Mayr 2009). The medial and lateral condyles are separated by a shallow ovate depression, and the intercondylar sulcus is deeper than the trochlea for the tibial cartilage. Because the tibiotarsus is preserved in latero-medial view, the cranial aspect of the shaft is hardly exposed, and related anatomical features are uncertain. Clearly there is a sulcus between the medial and lateral condyles, presumably the extensor sulcus, but a supratendinal pons cannot be discerned.
Tarsometatarsus.—The tarsometatarsus is preserved in dorsal view, but on the reverse side of the slab, the plantar surfaces of the proximal and distal ends are exposed (Fig. 5). The shaft is almost straight. Its medio-lateral width narrows slowly from the proximal end to three quarters of the length of the shaft, where it is narrowest, and then expands to the distal end. The narrowest portion has a width of 5 mm, which is nearly half the width of the proximal end. The tarsometatarsus is nearly two times the length of that of Rhynchaeites messelensis and is proportionally more slender. In both taxa the bone is proportionally shorter than in extant Threskiornithidae. Compared with Eogrus aeola (Wetmore 1934), the tarsometatarsus of Sanshuiornis zhangi is much sturdier.
On the proximal end of the shaft, the intercotylar eminentia is low, medio-laterally broad, and directed slightly dorsoproximally. The medial side of the intercotylar eminentia slopes gently to the medial cotyla, while the lateral side slopes abruptly to the lateral cotyla. In dorsal view, the lateral cotyla is located a little more distally than the medial cotyla. The dorsal rim of the lateral cotyla is flattened medio-laterally, and is longer than the dorsal rim of the medial cotyla. The medial cotyla is somewhat curved medially away from the shaft. The dorsal rim of the medial cotyla is lower than the plantar one, and its lateral side abuts the relatively high intercotylar eminentia, so the medial cotylar facet projects dorso-proximally in dorsal view. The lateral cotyla appears not as deep as the medial one, and its facet runs in parallel to the shaft. The above morphologies are similar to those of Rhynchaeites messelensis and extant Threskiornithidae, but different from extant Ciconiidae or Gruidae, in which the intercotylar eminentia does not project abruptly.
The morphology of the hypotarsus resembles Rhynchaeites messelensis (see Mayr and Bertelli 2011: fig. 4) but still shows some clear differences. It lacks any bony canals, but is composed of four prominent cristae, which separate three sulci. If these four cristae are numbered from 1 to 4 starting medially, crista 1 is the widest. The remaining three cristae seem to be situated on the same platform, and diverge from a distal common basis. Cristae 2 and 3 have the same length and width. It is uncertain whether crista 4 is crushed or not, as the preserved section is extremely short. The sulcus between cristae 1 and 2 is markedly deep, and probably represents the sulcus for the tendon of musculus flexor digitorum longus. The two sulci between cristae 2 and 4 are shallower, and presumably are the sulci for the tendons of musculus flexor perforatus digit II, and musculus flexor hallucis longus respectively. The widths of these three sulci are roughly the same. In Rhynchaeites messelensis, as in Sanshuiornis zhangi, the medial crista is the most developed, but the remaining three cristae are more weakly developed compared to Sanshuiornis zhangi, so that the middle and lateral sulci are not as prominent (compare Fig. 3 with Mayr and Bertelli 2011: fig. 4). The distal end of crista 1 converges with the basis of the remaining three cristae into a single ridge, and extends proximo-distally, which is not the case in Rhynchaeites messelensis. This ridge is very wide, measuring nearly one third of the width of the shaft. The medial parahypotarsal fossa is much deeper than the lateral one, and the two fossae extend distally to one fourth of the length of the shaft, where the single ridge begins to be confluent with the shaft. There is a marked tubercle on the mid-section of the proximal rim of the medial parahypotarsal fossa in a newly reported Rhynchaeites sp. (Mayr and Bertelli 2011), but such a tubercle is absent in Sanshuiornis zhangi and extant Threskiornithidae.
The dorsal surface of the tarsometatarsus is deeply excavated by the dorsal infracotylar fossa on the proximal end. The extensor sulcus occupies about one third the width of the shaft through most of its length. Its lateral section appears deeper than its medial section, but this may be an artefact of preservation. The sulcus begins to angle gently from the medial to the lateral side of the shaft, at three-fourths of the length of the shaft, and extends into the distal vascular foramen. The fossa infracotylaris dorsalis area is crushed and difficult to discern. The distal vascular foramen is situated just upon the upper margin of the lateral intertrochlear incision, and is situated more distally than in Eociconia sangequanensis. The distal vascular foramen is elliptical, with the width being half its length.
In their proportions, the trochleae resemble those of extant Threskiornithidae.
The distal end of the trochlea for the second toe is expanded and ball-shaped in both dorsal and distal view, lacking an articular furrow on its dorsal surface. There is, however, a groove on the plantar surface, and the medial section of the trochlea is plantarly deflected. The collateral ligamental fovea on its medial side is exposed. The distal articular facet is medially deflected, forming an angle of 33° with the longitudinal axis of the tarsometatarsus. The medial intertrochlear incision is not as deep as the lateral one, but mediolaterally wider. The trochlea metatarsi III projects more dorsally and distally than the other trochleae, which reach almost the same dorsal and distal level. It is grooved on its dorsal surface, and even more strongly so on the plantar surface. The width of the groove decreases slowly in the disto-dorsal direction. On the proximal end of this trochlea there is a fossa, which is wider than the distal vascular foramen; its proximal boundary is at the same distal level with the distal end of the distal vascular foramen. The distal articular facet of the trochlea for the fourth toe projects laterally, forming an angle of 20° with the longitudinal axis of the tarsometatarsus. Its dorsal surface is grooved, but the plantar groove is only very shallow. The dorsal groove is wider than that of the third trochlea, but less deep. The distal end of the trochlea bears a curving groove, and the groove becomes shallower distally. The proximal end of the trochlea metatarsi IV is not expanded significantly, meeting with the shaft of the tarsometatarsus with little demarcation. The trochlea metatarsi IV projects a little more dorsally than the trochlea metatarsi II, but less so than the trochlea metatarsi III. There is a distal interosseus canal opening into the lateral intertrochlear incision, which is connected with distal vascular foramen in plantar view, but cannot be seen in dorsal view.
Metatarsal I measures about two-thirds of the length of the first phalanx of digit I and attaches to the tarsometatarsus tightly throughout most of its length. It is “J” shaped in dorsal view, and also exposes its lateral side, where a collateral ligamental fovea can be seen. The plantar supratrochlear fossa is shallow, and demarcated medially by a high ridge, which separates the base of metatarsal I and the trochlea metatarsi II.
Pedal phalanges.—The pedal digits are almost completely preserved in articulation, except that digits II, III are displaced. Like in most birds, the phalangeal formula is 2–3–4–5.
Digit I is well developed and reaches about half the length of digit III. Assuming that the unguals are roughly similar in length for all the digits, which is confirmed by the unguals of digit I, III and IV, digit II is shorter than digit III. The proximal phalanx of digit I, as well as the proximal phalanx of digit II, are the longest phalanges; both are proportionally shorter than the corresponding phalanges of Rhynchaeites messelensis. The proximal phalanges of digit I and IV are slimmer than the proximal phalanges of the other toes.
The proximal phalanx of digit II measures 133% of the length of its penultimate one, and both are compressed in the mid section, with expanded distal and proximal ends.
Digit III is the longest and measures more than half of the length of the tarsometatarsus, and 195% and 122% of the length of digits I and IV respectively. Compared with Rhynchaeites messelensis, it is proportional shorter. The estimated length of digit III is nearly equal to the length of the tarsometatarsus in Rhynchaeites messelensis, but because of the longer tarsometatarsus the same ratio for IVPP V18116 is 70%. All phalanges of the third toe are exposed in medial view, and the flexor tubercles and collateral ligamental foveae can be seen. The length of the phalanges decreases proximodistally, with the proximal phalanx being 150% as long as the penultimate one, and the second one being the most robust.
The lengths of the phalanges of digit IV also decrease distally, and the proximal phalanx is twice as long as the penultimate one. The proximal phalanx of digit IV is the shortest, and the last three phalanges including the ungual are equal in length.
The ungual phalanges have a similar morphology, including the developments of the flexor tubercles and vascular sulcus. The unguals of digit I and II are equal in size, and smaller than that of digit IV. The ungual of the third toe is the biggest.
Stratigraphieal and geographical range.—Huayong Formation, Middle Eocene, Foshan, Guangdong Province.
List of the 20 scored taxa held in IVPP.
Measurement of Sanshuiornis zhangi gen. et sp. nov, from the middle Eocene Huayong Formation of Guangdong Province, south China, IVPP V18116.
Result of phylogenetic analysis
The new species was added to the data matrix of Mayr and Clarke (2003), and fourteen characters were scored for the new fossil. 405 most parsimonious trees were obtained (length = 728, CI = 0.23, RI = 0.48, HI = 0.77), the strict consensus tree of which is shown in Fig. 6. In the strict consensus tree, the relationships of a large number of taxa, including Sanshuiornis zhangi, is unresolved. In 66 of the 405 most parsimonious trees, S. zhangi is sister taxon of a clade including Threskiornithidae, Ciconiidae, and Phoenicopteridae.
The morphological features displayed by Sanshuiornis zhangi rule out some taxa as close relatives. For example, the hypotarsus lacks bony canals, which are present in Galliformes, all Pelecaniformes except Phaethontidae (Mayr and Bertelli 2011), and most psittaciform birds (Mayr 2008). Unlike Accipitridae, Falconidae, Horusornithidae and Sagittariidae (Mayr 2006), the central phalanges of the fourth toe and the proximal phalanx of the second toe are not strongly reduced. The trochleae metatarsorum II and IV extend to the same level distally, which is different from the condition seen in Eogruidae, Gruidae, Rallidae, Procellariidae and Phoenicopteridae (Wetmore 1934; Mayr and Clarke 2003; Clarke et al. 2005).
In overall morphology of the bones, Sanshuiornis zhangi most closely resembles some fossil “ciconiiform” birds, especially the poorly known and much larger Eociconia sangequanensis and the stem-group threskiornithid Rhynchaeites. The holotype of E. sangequansis preserves only the distal end of the left tarsometatarsus (Fig. 5), and the trochleae metatarsorum are largely broken. S. zhangi differs from E. sangequanensis in: (i) being much smaller, (ii) width of the medial intertrochlear incision smaller, (iii) trochlea metatarsi III positioned in the center of the tarsometatarsus, not as medially displaced as in E. sangequanensis, and (iv) the extensor groove proximal of the distal vascular foramen being deeper.
Sanshuiornis zhangi resembles Rhynchaeites messelensis in many respects, especially that (i) the hypotarsus consists of four cristae without bony canals, with the medial crista being the most prominent (compare Fig. 3 with Mayr and Bertelli 2011: fig. 4), and (ii) the length of the phalanges of the three anterior toes decreases distally. The key traits in which Rhynchaeites differs from Sanshuiornis are the presence of the tubercle on the proximal rim of the medial parahypotarsal fossa (see Mayr and Bertelli 2011), and the difference in the proportional length of some elements, especially pedal digits (proportionally shorter in Rhynchaeites) and tarsometatarsus (proportionally longer in Sanshuiornis).
Unfortunately, the affinities of the fossil are not resolved in our phylogenetic analysis. Although 66 of the 405 most parsimonious trees recover a sister-group relationship between Sanshuiornis zhangi and the clade (Threskiornithidae + Ciconiidae + Phoenicopteridae), the evidence for this grouping is weak. Five characters were optimized as apomorphies of this clade: (i) distal rim of medial condyle of tibiotarsus notched (character 102: 1, CI = 0.33, HI = 0.60), (ii) hypotarsus without bony canals for musculus flexor digitorum longus (character 105: 0, CI = 0.14, HI = 0.40) and musculus flexor hallucis longus (character 106: 0, CI = 0.11, HI = 0.38), (iii) presence of distal interosseus canal (character 107: 0, CI = 0.50, RI = 0), (iv) tarsometatarsal trochleae II and IV extending to same dorsal and distal level (character 108:1, CI = 0.11, HI = 0.42), and (v) hallux not reduced (character 110: 0, CI = 1, RI = 1). At least the latter three of these represent, however, the plesiomorphic condition in neornithine birds, and the first two have a wide distribution and are thus of little significance. Moreover, whereas Threskiornithidae are traditionally considered to be closely related to the Ciconiidae (Mayr and Clarke 2003; Clarke et al. 2005; Livezey 2007), a clade including (Threskiornithidae + Ciconiidae + Phoenicopteridae) is not supported by current phylogenetic analyses, which show Phoenicopteridae to be the sister taxon of Podicipedidae (e.g., Ericson et al. 2006; Hackett et al. 2008).
Although we note morphological similarities with some “ciconiiform birds”, we thus conclude that a definitive phylogenetic placement of the fossil is not possible with the material at hand.
We thank Xianqiu Zhang and Xiaojun Zhang (both SINOPEC Star Petroleum Corporation Limited, Guangdong, China), Zhao Wang and Qiuyuan Wang for help in the field, Yutong Li for fossil preparation, and Wei Gao for photos (all Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, China). The research was supported by National Natural Science Foundation of China and the Chinese Academy of Sciences. We thank Gareth Dyke (School of Biology and Environmental Science, University College Dublin, Belfield Dublin 4, Ireland) and Nathan Smith (Department of Geology, The Field Museum of Natural History, Chicago, USA) for reviewing the manuscript and providing helpful comments.
The modified characters in the data matrix (Mayr and Clarke 2003) according to Mayr (2011) are as follows: Gruidae (character 6: 1), Hesperornithidae (character 22: 1), Phoenicopteridae (character 31: 1), Apterygidae (character 32: 1), Phalacrocoracidae (character 57: 0), Rheidae (character 65: 1), Anatidae (character 85: 0), Opisthocomidae (character 85: 1), Apterygidae (character 100: 0), Coliidae (character 105: 0; character 106: 1)
Sanshuiornis zhangi scoring in Mayr and Clarke (2003) data matrix: 102: 1, 103: 1, 104: 0, 105: 0, 106: 0, 107: 0, 108: 1, 109: 0, 110: 0, 111: 0, 112: 0, 113: 0, 114: 0, 149: 0
Character 149 scored for all the taxa: Apsaravis: ?, Hesperornis: ?, Ichthyornis: ?, Rheidae: 1, Apterygidae: 0, Tinamidae: 0, Galliformes: 0, Anhimidae: 0, Anatidae: 0, Opisthocomidae: 0, Podicipedidae: 0, Phoenicopteridae: 0, Threskiornithidae: 0, Cariamidae: 0, Strigiformes: 2, Recurvirostridae: 0, Burhinidae: 0, Accipitridae: 1, Falconidae: 2, Sagittariidae: 0, Cuculidae: 0, Musophagidae: 0, Cathartidae: 0, Gaviidae: 0, Spheniscidae: 0, Phaethontidae: 0, Fregatidae: 0, Phalacrocoracidae: 0, Balaenicipitidae: 0, Eurypygidae: 0, Ardeidae: 0, Ciconiidae: 0, Procellariidae: 0, Otididae: 0, Pteroclidae: 0, Columbidae: 0, Rallidae: 0, Psophiidae: 0