Leaves and fruits of Bauhinia (Leguminosae, Caesalpinioideae, Cercideae) from the Oligocene Ningming Formation of Guangxi, South China and their biogeographic implications

Background The pantropical genus Bauhinia, along with the northern temperate Cercis and several tropical genera, bear bilobate, bifoliolate, or sometimes unifoliolate leaves, which constitute the tribe Cercideae as sister to the rest of the family Leguminosae based on molecular phylogenetics. Hence, the fossil record of Cercideae is pivotal to understand the early evolution and biogeographic history of legumes. Results Three fossil species of Bauhinia were described from the Oligocene Ningming Formation of Guangxi, South China. Bauhinia ningmingensis sp. nov. is characterized by its bifoliolate, pulvinate leaves bearing basal acrodromous primary veins and brochidodromous secondary veins. B. cheniae sp. nov. bears moderately or deeply bilobate, pulvinate leaves, with basal actinodromous primary veins and eucamptodromous secondary veins. B. larsenii D.X. Zhang et Y.F. Chen emend. possesses shallowly or moderately bilobate, pulvinate leaves bearing basal actinodromous primary veins and brochidodromous secondary veins, as well as elliptic, stipitate, non-winged, and oligo-seeded fruits. Meanwhile, previously reported Bauhinia fossils were reviewed, and those pre-Oligocene foliage across the world are either questionable or have been rejected due to lacking of reliable evidence for their pulvini or/and basal actinodromous or acrodromous venations. Besides Oligocene leaves and fruits presented here, foliage and/or wood of Bauhinia have been documented from the Miocene–Pliocene of Thailand, India, Nepal, Uganda, and Ecuador. Conclusions Bauhinia has exhibited a certain diversity with bifoliolate- and bilobate-leafed species in a low-latitude locality–Ningming since at least the Oligocene, implying that the tropical zone of South China may represent one of the centres for early diversification of the genus. The reliable macrofossils of Bauhinia and Cercis have made their debut in the Eocene–Oligocene floras from mid-low latitudes and appeared to lack in the coeval floras at high latitudes, implying a possible Tethys Seaway origin and spread of legumes. However, detailed scenarios for the historical biogeography of Bauhinia and its relatives still need more robust dataset from palaeobotany and molecular phylogeny in future research.

The goals of this paper are to (1) investigate and evaluate the fossil record of Bauhinia, with special reference to that of Cercis, by comparing both extinct and extant angiosperms with the similar lobed leaf forms, (2) describe the foliage and fruit fossils from the Oligocene Ningming Formation of Guangxi, South China, and discuss their biogeographic implications.

Macrofossils
The fossil foliage and fruits studied in this paper were collected from the Ningming Formation at 22°07.690'N, 107°02.434'E in the western region of Ningming County, Guangxi Zhuang Autonomous Region, South China ( Figure 2). The Ningming Formation is primarily shallow lacustrine deposits consisting of gray to dark gray mudstone, light yellow shaly siltstone, and finely grained sandstone. No volcanic rocks and mammals are hitherto found in the Ningming Formation [28], and an absolute age for this formation is therefore unavailable. The Ningming Basin is among the late Palaeogene basins (e.g., mammal-bearing Figure 1 A simplified phylogenetic tree of the Leguminosae, with special reference to the phylogeny of the tribe Cercideae (after [10,17]) and iconic leaf forms enhanced. The purple taxa belong to Bauhinia sensu lato.
In China, all the land belongs to our country. Our fossil-collecting fieldwork was done in non-National Nature Reserves (NNR) and non-private areas and has been allowed by the local government. We did not violate the Chinese fossil collection and mining laws and management regulations.
The macrofossils are preserved as compressions/impressions only with a little organic material remaining in mudstone. Cuticle preparations were unsuccessful because organic material has been greatly weathered. All the marcofossil specimens used herein are deposited at Natural History Museum of Guangxi (NHMG), Nanning, P.R. China (see Additional file 2).

Herbaria
The exsiccatae examined in this study are kept at the following herbaria: Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu (CDBI), Guangxi Institute of Botany, Chinese Academy of Sciences, Guilin (IBK), South China Botanical Garden, Chinese Academy of Sciences, Guangzhou (IBSC), the Herbarium of Northeast China, Shenyang (IFP), Kunming Institute of Botany, Chinese Academy of Sciences, Kunming (KUN), and the Chinese National Herbarium, Beijing (PE) (see Additional file 2).
(2) Chinese Virtual Herbarium (CVH) [41]. The images of herbarium specimens were browsed. (3) ILDIS (International Legume Database & Information Service) [42]. The geographic distribution of living species in Bauhinia is compiled by ILDIS, with a few newly published records ( Figure 4; see Additional file 1). (4) Hunt Institute for Botanical Documentation [43]. Standardized abbreviations of plant-family names and periodical titles in this paper were consulted and applied (see References; Additional file 3).

Comparative morphology
Both fossil and extant taxa bearing similar unifolioate, bilobate, and bifoliolate leaves in Leguminosae and other families ( Figure 3; see Additional file 3) were compared to evaluate the fossil record and biogeographic history of Bauhinia. Based upon an extensive review on the literature and specimens of previously reported Bauhinia and other similar foliage from the Cretaceous and Cenozoic, we summarized the reliable fossil record of Bauhinia (see Additional file 4).

Figures
Photographs of specimens were taken with digital cameras (Panasonic DMC-FZ30 and Nikon D90). A simplified phylogenetic tree (Figure 1) of the Cercideae within Leguminosae was partially adapted from the literature [10,17], with the iconic leaf forms enhanced. A map for the fossil locality ( Figure 2) was partially adapted from the literature [36]. Line drawings of leaf specimens for fossil and living taxa as well as of distributional map of Bauhinia were drawn (Figures 3,4,5,6,7 and 8) and arranged using CorelDRAW 10.0 (Corel, Ottawa, Ontario, Canada) and Adobe Photoshop 6.0 (San Jose, California, USA) programmes.

Abbreviations
The standardized abbreviations for the family name of plants, the author citation of plant names in this paper and journal titles in References follow Brummitt and Powell [62] as well as Botanico-Periodicum-Huntianum (BPH), its supplement (BPH/S), and BPH-2 [63], respectively. The herbarium codes refer to Index Herbariorum [64].

Similar leaf architectural comparisons
A leaf lamina partially or fully divided into two lobes is not very common but remarkable among angiosperms. Such leaf forms can be traced back to the Late Cretaceous, for example extinct Liriodendron-like angiosperms Liriodendrites bradacii K.R. Johnson [44] and Liriophyllum kansense Dilcher et P.R. Crane [45] (Figure 3d The taxonomy of Bauhinia L.
Morphological characters are the features that ultimately support the distinctiveness of real biological entities, so integral studies mutually illuminating between morphology and molecular systematics will be key in the discrimination of elusive relationships within Bauhinia sensu lato. However, considerable convergence, parallelism or evolutionary conservativeness in the organs (especially leaves) of Bauhinia sensu stricto and its relatives often place palaeobotanists in a predicament. Without reproductive organs (especially the calyces, fertile stamens, and petals), bilobate leaves of some species in Bauhinia sensu stricto, Phanera, and Schnella would not be distinguished from each other even by neobotanists. Hence, bilobate or bifoliolate leaf fossils in Cercideae were often assigned to Bauhinia sensu lato (see Additional file 4).

The fossil record of Bauhinia and other bilobate leafed taxa
Overall, Bauhinia bears mostly bilobate, bifoliolate, or sometimes unifoliolate leaves having characteristic upper and lower pulvini, basal actinodromous or acrodromous primary veins (3-13 in number per leaf), brochidodromous, eucamptodromous or craspedodromous secondary veins, and alternate or opposite percurrent tertiary veins, character combinations of which are noticeably different from the lobed foliage of genera in the tribe Detarieae sensu lato Polhill of Leguminosae, as well as in Apocynaceae, Convolvulaceae, Liriodendraceae, Oxalidaceae, Passifloraceae, Proteaceae Juss., and Zygophyllaceae (see Figure 3;

Fossil species
Three fossil species of Bauhinia are described as follows.
All the voucher specimens were collected from the same locality and stratigraphy, and they are deposited at the same museum.

Etymology
The specific epithet is derived from Ningming County, where the fossils were collected. diverging at ca. 60°-90°from the primary veins on the exmedial side. Tertiary veins alternate and opposite percurrent or ramified, straight, convex or sinuous (Figure 5a-c). Quaternary veins forming irregular polygons. Aerolation well developed. Freely ending veinlets mostly branching once. Marginal ultimate veins looped and fimbriate (Figure 5b-d).  (Figures 6a-g, 7a-e).

Etymology
The specific epithet is dedicated to Prof. Chen Dezhao (Chen Te-chao) (South China Botanical Garden, CAS) for her important contribution to the taxonomy of Cercideae.

Description
Bilobate leaves, wide ovate or suborbicular, ca. 2.0-6.0 cm long and 2.2-6.5 cm wide, often folded along the midvein (Figures 6a-c, 7a-e). The petiole glabrescent or covered with dense, spreading hairs (Figure 6d,e).The petiole stout, ca. 1.6-2.0 cm long, bearing thickened, upper and lower pulvini (Figures 6d,e 7c). The upper pulvinus connecting the laminar base via a tiny, semicircular laminar joint. Bifid to ca. 2/3-4/5 of laminar length or almost to the laminar base, forming a narrow or flaring sinus. Two lobes symmetrical or slightly asymmetrical. Lobe apex slightly acuminate, obtuse or rounded (Figures 6a-d,f and 7a-e). Laminar base symmetrical, shallowly to deeply cordate (Figures 6a-d,f,g, 7a-e). Margin entire. Texture apparently chartaceous. Primary veins basal actinodromous, 7-9 in number, the outmost pair and midvein being weaker than the inner pairs. Midvein terminated in a short or long spine within the sinus (Figures 6g, 7c). Lateral primary veins straight or curved, branched or unbranched, and the innermost pair reaching the lobe apex and outer pairs approaching to the margin. Secondary veins eucamptodromous, diverging at ca. 30°-80°mainly from the innermost and outmost lateral primary veins and arching upward along the margin (Figures 6a-c,f,g, 7a,b). A pair of secondary veins usually emitting from the midvein near the sinus and approaching to the inner margin of lobes (Figure 6b-d, f). Intersecondary veins sometimes present, parallel to subjacent secondary veins. Tertiary veins alternate and opposite percurrent or ramified, mostly convex, sinuous or rarely straight. Quaternary veins alternate and opposite percurrent, forming irregular polygons (Figures 6b,c,g, 7a,b). Aerolation well developed. Freely ending veinlets mostly unbranched. Marginal ultimate veins absent.

Emended description
Leaves suborbicular or slightly ovate to wide ovate, ca. 2.1-4.5 cm long and 1.8-4.8 cm wide, usually folded along the midvein (Figure 8a-d, f-i). Bifid to ca.1/2-3/5 of laminar length, forming a narrow sinus. Two lobes symmetrical or slightly asymmetrical. Lobe apex rounded to obtuse (Figure 8a,c,d,f). Laminar base symmetrical, rounded or shallowly cordate (Figure 8a,c,d,g). Margin entire. Texture apparently chartaceous to coriaceous. Primary veins basal actinodromous, 5-9 in number, the outmost pair being weaker than the midvein and inner pairs. Midvein terminated in a short spine within the sinus. Lateral primary veins branched or unbranched, and the innermost pair reaching the lobe apex. Major secondary veins brochidodromous, diverging at ca. 45°-60°from the lateral primary veins mainly on the exmedial side and sporadically on the admedial side (Figure 8d)

Comparisons
This fossil species B. larsenii was first described by Chen and Zhang [27] on the basis of four specimens. Here, we emended this species, especially regarding the leaf architecture and fruit morphology, based upon the type specimens and newly collected materials. It is very similar to living B. viridescens Desv. and B. brachycarpa Wall. ex Benth. [3,5,41] in bearing shallowly or moderately bilobate leaves, brochidodromous secondary vein, and elliptic fruits (see Additional file 3). However, no adequate characters guarantee the fossils to belong to any living species. In addition, B. larsenii is different from B. cheniae sp. nov. and other fossil species [33,[50][51][52][53][54][55][56][57] in the leaf architectural detail (see Additional files 3 and 4). In particular, B. larsenii represents the first recognition of Bauhinia fruit and foliage organically connected in the fossil record.

Discussion
The Leguminosae is the third largest angiosperm family only after Orchidaceae Juss. and Asteraceae Bercht. et J. Presl, varying in habit from herbs to shrubs, vines, lianas, and trees, with an extremely high diversity of ca. 751 living genera and ca. 19,500 species [10,17] across different habitats of the world. Meanwhile, this family has an abundant and diverse fossil record, and its characteristic fruits, flowers, pollen, foliage, and wood have been well recognized from numerous Cenozoic localities around the world [65,66]. However, an outstanding incongruence between the palaeobotanical finds and molecular systematics of legumes is that the earliest fossil record of the tribe Cercideae as sister to all other lineages in the molecular phylogenetic trees of Leguminosae [9][10][11][12][13][14][20][21][22][23][24] has so far occurred later than that of some derived tribes bearing compound leaves such as Sophoreae Spreng. ex DC. in the subfamily Papilionoideae L. ex DC. [67]. Such an incongruence implies that the extant Cercideae bearing the simple, entire or bilobate to bifoliolate foliage is unlikely to be the most primitive in Leguminosae, but the derived as some authors formerly suggested from an extinct legume ancestor possibly with palmately compound leaves [68] or pinnately compound leaves [27,69]. The fossil record of the Cercideae lacking or being fewer than those derived tribes in the early Palaeogene of middle latitudes is because either the early distribution of the Cercideae might be restricted to low latitudes, or palaeobotanical studies on the coeval legumes from low latitudes are relatively inadequate [70]. Hence, the Cercideae fossils, especially from low latitudes, can provide an historical perspective for their early evolution, adaptive radiation, and biogeographic history.
Leaves of Cercis have been reported from Late Cretaceous and Cenozoic sediments, but the overwhelming majority of these reports have been rejected, questioned, revised [59,71], or in need of confirmation by reinvestigation of the original materials and discovery of better preserved materials [69] (see Additional file 4). The oldest reliable fossils of Cercis are represented by the foliage and/or fruits (i.e., C. parvifolia Lesq., C. herbmeyeri H. Jia et Manchester) from the Late Eocene Florissant Formation, Colorado and John Day Formation, Oregon, western USA [71,72]. In contrast, the foliage fossils of Bauhinia have been previously reported from the Late Cretaceous and Palaeogene of North America and Eurasia [73][74][75][76][77][78][79][80][81][82][83][84][85][86][87][88], but these identifications are erroneous or unreliable [70] (see Additional file 4). Although Bauhinia or Bauhinia-like bilobate foliage have been recently reported from the middle Eocene of Tanzania [89], the late Eocene of Vietnam [30], the late Eocene-early Miocene of Brazil [90], and possibly the latest Oligocene-mid-late Miocene of Australia [91], the preservation of these pre-Miocene fossils, which are observed from the originally published figures, appears too poor to reliably assign these leaf fossils to either Bauhinia or even Cercideae, because neither the pulvinus nor basal actinodromous or acrodromous venations can be confirmed (see Additional file 4). Instead, the oldest reliable evidence of Bauhinia and Bauhinia-like foliage are provided from the Oligocene Ningming Formation, Guangxi, South China (i.e., Bauhinia ningmingensis sp. nov., B. cheniae sp. nov., and B. larsenii D. X. Zhang et Y. F. Chen [33], this paper) and Coatzingo Formation, Puebla, Mexico (i.e., Bauhcis moranii Calvillo-Canadell et Cevallos-Ferriz [92]). By the Miocene-Pliocene, various species of Bauhinia have existed in Thailand, India, Nepal, Uganda, and Ecuador ( Figure 4) while those of Cercis have become widespread in mid-latitudes of the northern hemisphere [69,71].
In addition, some other unifoliolate or bilobate foliage or fruit fossils from the Oligocene-Miocene of Jinggu (Yunnan), Zhangpu (Fujian), and Ningming (Guangxi) in South China have been reliably described [93][94][95] or preliminarily identified as Cercis and Bauhinia (Unpublished observation by Qi Wang, Institute of Botany, Beijing, October 18 th , 2013). Also, the bilobate foliage extremely similar to Bauhinia has been discovered from the Eocene -Oligocene coals of West Sumatra, western Indonesia (vide the image of this leaf fossil and information provided by Drs. Yahdi Zaim, Institute of Technology, Bandung, Indonesia, Peter Wilf, Pennsylvania State University, and Gregg F. Gunnell, Duke University Lemur Center, March 17 th 2014). Hence, more Cercideae fossil will be studied and reported from low-latitude tropical zone [96] of East and Southeast Asia. Recently, a strictly east-towest vicariance for the historical biogeography of Cercis has been postulated by molecular data [97]. The Cercideae macrofossils occurring in the Eocene to Oligocene of mid-low latitudes and apparently lacking in the coeval sediments at high-latitudes appear to partially support a tropical Tethys Seaway origin and spread [11,22,23] or an "Out-of-Tropical Asia" dispersal [26] of the Cercideae and the Leguminosae as formerly hypothesized by some authors. However, detailed historical biogeography of Cercideae still need more palaeobotanical and molecular dataset.

Conclusions
Bauhinia has exhibited a certain diversity with three species (i.e., B. ningmingensis, B. cheniae, and B. larsenii) bearing bifoliolate or bilobate leaves in a low-latitude locality-Ningming since at least the Oligocene, implying the tropical zone of South China may represent one of the centres for early diversification of the genus. The reliable macrofossils of Bauhinia and Cercis have made their debut in the Eocene-Oligocene floras from mid-low latitudes and appeared to lack in the coeval floras at high latitudes. By the Miocene-Pliocene, various species of Bauhinia have existed in Thailand, India, Nepal, Uganda, and Ecuador while those of Cercis have become widespread in mid-latitudes of the northern hemisphere. Such a biogeographic pattern implies a possible Tethys Seaway origin and spread for legumes. However, detailed scenarios for the historical biogeography of Bauhinia and its relatives still need more robust dataset from palaeobotany and molecular phylogeny in future research.