Cool tropics in the Middle Eocene: Evidence from the Changchang Flora, Hainan Island, China

Cool tropics in the Middle Eocene: Evidence from the Changchang Flora, Hainan Island, China

Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16 Contents lists available at ScienceDirect Palaeogeography, Palaeoclimatology, Pala...

13MB Sizes 0 Downloads 8 Views

Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

Contents lists available at ScienceDirect

Palaeogeography, Palaeoclimatology, Palaeoecology journal homepage: www.elsevier.com/locate/palaeo

Cool tropics in the Middle Eocene: Evidence from the Changchang Flora, Hainan Island, China Robert A. Spicer a,c,d, Alexei B. Herman b,c, Wenbo Liao c, Teresa E.V. Spicer d, Tatiana M. Kodrul b,c, Jian Yang d, Jianhua Jin c,⁎ a

Environment, Earth and Ecosystems, Centre for Earth, Planetary, Space and Astronomical Research, The Open University, Milton Keynes MK7 6AA, UK Laboratory of Paleofloristics, Geological Institute, Russian Academy of Sciences, 119017 Moscow, Russia State Key Laboratory of Biocontrol and Guangdong Key Laboratory of Plant Resources, School of Life Sciences, Sun Yat-sen University, Guangzhou 510275, China d Key State Laboratory for Systematic and Evolutionary Botany, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China b c

a r t i c l e

i n f o

Article history: Received 25 March 2014 Received in revised form 9 July 2014 Accepted 10 July 2014 Available online 17 July 2014 Keywords: Eocene Cool tropics Climate Leaf Analysis Multivariate Program Hainan Island Plant fossils

a b s t r a c t The middle Eocene (Lutetian–Bartonian, 48.6–37.2 Ma) near-equatorial megafossil flora from swamp and lacustrine facies of the lower Changchang Formation, Hainan Island, South China (19.631463°N, 110.445049°E) is highly diverse (N 200 taxa) dominated by an unusual mixture of angiosperms typical of modern temperate, subtropical and tropical evergreen and deciduous forms. It is also rich in palms. Multivariate analysis of the architecture of minimally transported woody dicot leaves reveals a mean annual air temperature (MAT) of ~22 ± 4.7 °C with a marked thermal seasonality range of ~21 °C. The year-round humid climate lacked any monsoonal signature. The overall climate signal is compatible with the growth characteristics exhibited by fossil wood, but is warmer than the climate signal derived from pollen and spores using Co-existence Analysis. Corrections for possible palaeoelevation of the basin bring the megafossil-derived MAT estimate in line with 54–52 Ma sea surface and soil temperatures obtained from the Gulf Coast, USA, (palaeolatitude ~30°N) using multiple geochemical proxies and supports the claim that the low latitude Eocene climate was not uniformly warm. This challenges previous conclusions based on ∂18O analysis of unaltered calcareous microfossils. Our air temperature data also adds to the challenge of understanding heat transport away from the equator to higher latitudes during ‘hothouse’ climate regimes. © 2014 Elsevier B.V. All rights reserved.

1. Introduction 1.1. Paleogene climate questions The Eocene global mean annual surface temperature is inferred to have been markedly warmer than now (e.g. Greenwood and Wing, 1995), and based on both marine and terrestrial records witnessed one of the warmest prolonged intervals in the last 65 million years (Zachos et al., 2001; Eberle and Greenwood, 2012). A general characteristic of such ancient ‘hothouse’ climates is that numerous proxies indicate a markedly lower equator-to-pole temperature gradient than during cooler periods. However, the failure of climate models to reproduce such shallow gradients persists despite decades of attempts to warm the poles without overheating the equator (Valdes, 2000). That the Eocene high latitudes were warm is beyond doubt as evidenced by highly productive ecosystems populated by cold intolerant

⁎ Corresponding author. Tel.: +86 20 84113348; fax: +86 20 84110436. E-mail address: [email protected] (J. Jin).

http://dx.doi.org/10.1016/j.palaeo.2014.07.011 0031-0182/© 2014 Elsevier B.V. All rights reserved.

organisms (Eberle and Greenwood, 2012). More controversial has been the characterisation of low latitude climates, particularly when isotopic systems are so prone to diagenetic effects (Norris and Wilson, 1998; Schrag, 1999; Wilson et al., 2002). Diagenetic alteration initially resulted in surprisingly cool equatorial temperatures during the Cretaceous and Paleogene (Shackleton and Boersma, 1981; Bralower et al., 1995; D'Hondt and Arthur, 1996; Dutton et al., 2005) but analysis of unaltered material subsequently revised these temperatures upwards (Pearson et al., 2001; Pearson et al., 2007) such that maximum sea surface temperatures (SSTs) were estimated to have been mostly N30 °C, and some as high as 34 °C, implying even warmer terrestrial daily peak air temperatures. Such temperatures compromise primary productivity in land plants by enhancing photorespiration and destroying lipid membranes (Muraki et al., 2000; Sharkey, 2000). Organic geochemical proxies, while less prone to diagenetic effects, yield even higher tropical SSTs of between 35 and 40 °C during the warmest Eocene interval and if these temperatures reflect past reality it is likely that tropical vegetation suffered mass mortality (Huber, 2008). There is some evidence for this, but only during extremely warm but shortlived (~ 102 ka) episodes such as that at the PETM (Harrington and Jaramillo, 2007) when temperatures peaked well above the longer-

2

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

term background temperatures. To characterise better near-tropical Eocene terrestrial background climates we here report on a diverse Lutetian–Bartonian (48.6–37.2 Ma) palaeoflora from a palaeolatitude of ~ 22°N using minimally transported leaf megafossils preserved in lacustrine and swamp environments. 1.2. Geological setting Numerous plant fossils representing the Changchang middle Eocene flora occur throughout the Changchang Formation, Changchang Basin, on Hainan Island, South China (19.631463°N, 110.445049°E; Fig. 1). The Paleogene deposits of the Changchang Basin are subdivided into three formations (Zhou and Chen, 1988; Lei et al., 1992), the lowermost being the Changtou Formation (Paleocene), overlain by the Changchang Formation (Eocene), above which occurs the Wayao Formation (Eocene). The lower coaliferous part of the plant-bearing Changchang Formation is about 52–54 m thick and consists of clastic terrigenous and coal-bearing deposits probably formed in lacustrine and paludal environments. This is overlain by part, 37–40 m of predominantly lacustrine and fluvial mudstones, siltstones and sandstones. The deposits of the lower part consist of dark grey mudstone, greyish black coaly shale, brownish grey oil-bearing shale, yellowish brown, greyish yellow, greyish white muddy siltstone and sandstone, and coal (Fig. 2). A simplified sedimentary log of the section yielding the material used in the study reported here is given in Fig. 3. Well-preserved plant megafossils were collected mainly from the coaly shales, grey mudstones and siltstones of the lower part of the Changchang Formation. As well as leaves, fruits, seeds and fossil wood these deposits contain diverse spore and pollen assemblages (Zhang, 1980; Lei et al., 1992; Yao et al., 2009; Ch.Ch. Hofmann, pers. comm. 2009), gastropod and bivalve mollusc shells, and fish bones, teeth and scales. 1.3. The age and palaeolatitude of the Changchang Flora Following the discovery of abundant plant fossils in the Changchang Formation by the 4th geological survey team (Bureau of Geology and Mineral of Hainan Province) Guo S.X. identified 30 species including

Sabalites sp., Cinnamomum spp., Nelumbo nucifera, and Osmunda lignitum, which exhibit similarities with the Miocene flora of the Shangchun Formation in the Maoming Basin. In 1979, Guo reinvestigated these plant fossils and published identifications for 10 species (see section 1.4.1). Based on floral composition, Guo argued for a late Palaeocene to early, middle Eocene age for the Changchang Formation. Based on palynological data from the Changtou, Changchang and Wayao formations, Lei et al. (1992) dated the upper (coaliferous) member of the Changchang Formation as late early Eocene to early late Eocene, although in an earlier work by Zhang (1980) it was assigned to the early to middle Oligocene. Yao et al. (2009) regard their palynological assemblages from the Changchang Formation as being of “early middle Eocene age”. Sychevskaya (pers. comm., 2008), who made a preliminary study of fossil fish remains from the Changchang Formation, suggested an early middle to late Eocene age due to the co-occurrence of fishes belonging to families Catostomidae and Cyprinidae. Two taxonomically diverse Eocene floras have been recently reported from the Youganwo and Huangniuling formations of the Maoming Basin (southwestern Guangdong Province, South China) (Kodrul et al., 2012a; Kodrul et al., 2012b). The Youganwo Flora yields horsetails (Equisetum sp.), ferns (Osmundaceae, Polypodiaceae, Salviniaceae), conifers (Pinaceae, Podocarpaceae) and numerous angiosperms (Nelumbonaceae, Lauraceae, Fagaceae, Fabaceae, Platanaceae, Altingiaceae, Anacardiaceae, Celastraceae, Ulmaceae, Euphorbiaceae, Myrtaceae, etc.), whereas the overlying Huangniuling Flora is composed of ferns (Lygodiaceae), conifers (Pinus, Nageia, Sciadopitys) and angiosperms (Lauraceae, Fagaceae, Fabaceae, Hamamelidaceae, Altingiaceae, Myricaceae, Juglandaceae, Myrtaceae, Rhamnaceae, Ulmaceae, Celastraceae, Dipterocarpaceae, Apocynaceae, Annonaceae, Apiaceae) (Feng and Jin , 2012; Kodrul et al., 2012a, 2012b; Oskolski et al., 2012). Preliminary palynological study of three sections within the nearby Chinese mainland Maoming Basin, and a comparison of the data obtained with other palynocomplexes of the South China Sporopollen Region (Ye et al., 1996), show that the Youganwo Formation is most probably middle Eocene (Lutetian–Bartonian, 48.6 Ma to 37.2 Ma, (dates from Gradstein et al., 2005) and

Fig. 1. Map showing the position of the Changchang Basin and fossil flora, Hainan Island, China.

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

3

Fig. 2. Panoramic photograph of the Changchang leaf fossil site showing different lithostratigraphic units within the exposed Changchang Formation.

Huangniuling Formation–late Eocene (Priabonian, 37.2–33.0 Ma) in age (Aleksandrova et al., 2012)). Finds of fossil fishes, turtles, crocodiles and mammals from oil shales of the Youganwo Formation evidence a late Eocene age of the fossiliferous strata (Wang et al., 2007; Jin, 2008; Jin and Kodrul, 2008; Tong et al., 2010). According to a magnetostratigraphic data (Wang et al., 1994), the Youganwo and Huangniuling formations were deposited during Chrons 18 to 16 and therefore are middle to late Eocene in age. The Changchang Flora of Hainan Island exhibits a significant taxonomic similarity to the Youganwo Flora of the Maoming Basin in that they have a number of taxa in common (e.g. O. lignitum (Giebel) Stur,

Nageia, Nelumbo, Laurophyllum, Quercus, Celastraceae, Altingiaceae, Podocarpium). Due to this similarity, these two floras appear to be of the same age (Kodrul et al., 2012a, 2012b); a conclusion that is corroborated by a palynological data (Lei et al., 1992; Yao et al., 2009; Aleksandrova et al., 2012). Based on the observations listed above the Changchang flora is middle Eocene (Lutetian–Bartonian) in age (~48–~38 Ma). The latitudinal position of Hainan Island has remained remarkably stable since the Eocene. Based on palaeogeographic reconstructions the Changchang plant remains were deposited at ~ 22°N (Fig. 4) (http://www.odsn.de; Scotese, 2001).

Fig. 3. Lithostratigraphic column of the Changchang Formation, Hainan Island derived from the section shown in Fig. 2.

4

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

Fig. 4. Palaeogeographic map showing the location of Hainan Island at 40 Ma. Plate I. Woody dicot leaf morphotypes of the Changchang Flora, Hainan Island, South China; collection no. Scale bar 1 cm. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40.

OTU2, sample no CC8-2; OTU3, sample no CC772a-1; OTU4, sample no CC692; OTU5, sample no CC691a; OTU7, sample no CC967-2; OTU8, sample no CC11; OTU10, sample no CC3; OTU11, sample no CC1; OTU17, sample no CC789; OTU17, sample no CC959; OTU18, sample no CC94; OTU18, sample no CC97; OTU21, sample No CC758; OTU22, sample no CC14-1; OTU23, sample no LC1b-2; OTU25, sample no CC130-1; OTU26, sample no CC925; OTU27, sample no CC490a-2; OTU29, sample no CC270; OTU30, sample no CC271-2; OTU31, sample no CC769-1; OTU32, sample no CC302a-1; OTU34, sample no CC415; OTU36, sample no CC303; OTU37, sample no CC950-2; OTU38, sample no CC664; OTU39, sample no CC308a; OTU40, sample no CC306; OTU41, sample no CC334a; OTU43, sample no CC338; OTU44, sample no CC479; OTU45, sample no CC1012a; OTU46, sample no CC1124a; OTU48, sample no CC322; OTU49, sample no CC364; OTU50, sample no CC294; OTU51, sample no CC1010a; OTU52, sample no CC884; OTU54, sample no CC1156; OTU57, sample no CC821.

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

Plate I.

5

6

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

1.4. Taxonomic composition of the Changchang Flora 1.4.1. Megafossils Impressions and compressions of leaves, fruits, rhizomes and roots, as well as petrified wood remains, are most abundant in a grey mudstone underlying and overlying two main coal seams. In a preliminary report Guo (1979) identified 10 species belonging to 9 genera (O. lignitum, Salvinia sp., Cyclocarya scutellata, Neulumbo protospeciosa, Cinnamomum larteti, Ocotea sinensis, Citrus niger, Sabalites szei, S. changchangensis, Nordenskioeldia borealis). Subsequently Jin (2009) published a detailed study of two fossil fruits belonging to Paleocarya sp. (Juglandaceae) and Acer cf. A. miofranchetii Hu et Chaney (Aceraceae). Following extensive collecting the Changchang Flora is now known to contain possible horsetails, ferns, conifers and angiosperms belonging to over 200 species (morphotaxa). With respect to diversity, the Changchang Flora is comparable to the richest Eocene floras known. Possible horsetails are represented by numerous rhizome remains, probably belonging to Equisetum (Equisetaceae). Although overall the Changchang Flora is rich in species, ferns are not at all diverse. The most abundant fern is O. lignitum, which probably inhabited wetlands and swamps. Aquatic ferns are represented by abundant Salvinia (Wang et al., 2014). Other ferns are represented by only a few Dryopteridaceae-like remains and one segment of a fern probably belonging to the Polypodiaceae. Conifers are extremely rare in the Changchang Flora. In the collection to date (N 5000 specimens) there are only a few compressions of podocarpaceous leaves (e.g. Nageia) (Jin et al., 2010).

Angiosperms, both dicots and monocots, dominate in the flora in terms of diversity and, presumably contributed the greatest biomass in the ancient vegetation. Woody dicotyledons are assignable to the families Lauraceae, Fagaceae, Altingiaceae, Myricaceae, Fabaceae, Malvaceae, Juglandaceae and Ulmaceae. Lauraceae are an almost ubiquitous component of the Changchang megaflora but because the pollen of this family does not preserve well Lauraceous pollen is absent from the microflora. Based on leaf morphology and epidermal–cuticular characters, several genera of this family can be recognised (Alseodaphne, Beilschmiedia, Cinnamomum, Cryptocarya, Litsea, Machilus, Neolitsea, and Laurophyllum). In several localities, the most abundant plant remains are those of Nelumbo (Nelumbonaceae) leaves, rhizomes, roots, tubers, receptacle and fruits. Newly collected material has allowed some of us to reinterpret C. scutellata, described by Guo (1979), as a Nelumbo receptacle (Jin and Kodrul, 2008; Jin et al., 2009b; He et al., 2010). The Fagaceae are represented by several species belonging to two or three genera (possibly Castanopsis, Lithocarpus and Quercus), while the family Altingiaceae is documented by three-lobed Liquidambar leaves similar to those of L. miosinica Hu et Chaney. A single species of Myrica represents the Myricaeae. Representatives of the Fabaceae make up a notable component of the Changchang megaflora. Leaflets with an asymmetrical base and a short petiolule are assigned in a preliminary way to the morphogenus Leguminosites. Other leaflets possessing a lanceolate lamina, rounded mucronate apex, an asymmetric base and prominent basal veins

Plate II. Woody dicot leaf morphotypes of the Changchang Flora, Hainan Island, South China; collection no. scale bar 1 cm. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.

OTU59, sample no CC752b; OTU61, sample no CC997; OTU62, sample no CC1106a; OTU62, sample no CC1118a-1; OTU63, sample no CC960a; OTU66, sample no CC1112; OTU67, sample no CC437; OTU68, sample no CC1126; OTU70, sample no CC438; OTU71, sample no CC1201; OTU72, sample no CC1104; OTU73, sample no CC1198; OTU74, sample no CC288; OTU75, sample no CC1193; OTU76, sample no CC915a; OTU77, sample no CC1195b-1; OTU78, sample no CC1196; OTU79, sample no CC769-4; OTU80, sample no CC806a; OTU81, sample no CC297; OTU83, sample no CC864; OTU84, sample no CC1069a; OTU85, sample no CC6-1; OTU86, sample no CC1080a; OTU89, sample no CC1031a; OTU92, sample no CC738; OTU93, sample no CC737; OTU94, sample no CC451a; OTU95, sample no CC390; OTU96, sample no CC377a; OTU97, sample no CC944a-1; OTU102, sample no CC134; OTU103, sample no CC978; OTU104, sample no CC788a; OTU105, sample no CC699; OTU106, sample no CC855; OTU107, sample no CC976; OTU108, sample no CC784; OTU108, sample no CC785; OTU109, sample no CC854-2; OTU110, sample no CC354a; OTU116, sample no CC397a; OTU116, sample no CC1022; OTU118, sample no CC801; OTU121, sample no CC351; OTU122, sample no CC714b.

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

Plate II.

7

8

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

extending along leaflet lamina margins, possibly belong to the genus Podocarpium. Single seeded fruits associated with these leaflets are also assigned to Podocarpium. This genus is known in the early Oligocene to Pliocene floras of Eurasia (Wang et al., 2007). New finds of Podocarpium from the Changchang Formation support the assumption that an ancestral population of this genus may have originated in the early Paleogene of eastern Asia and then spread to most of middle latitude Eurasia during the Miocene (Wang et al., 2007). The Malvaceae is represented by the genus Craigia (Jin et al., 2009a). This is the first fossil record of Craigia from within its modern distribution centre and provides new data on the origin and phytogeographic history of this genus. A juglandaceous winged fruit was assigned to Palaeocarya sp. (Jin, 2009). The family Ulmaceae is represented by leaves belonging to Celtis and possibly Ulmus. Monocotyledons are the characteristic component of the Changchang Flora, with palm leaves being the most numerous. These leaves were assigned by Guo (1979) to two species of Sabalites (Arecaceae). Recently, Zhou et al. (2013) described seven species belonging to three genera (Coryphoideae of Arecaceae): Sabalites asymmetricus, S. robustus, S. tenufolius, S. szei, S. changchagnensis, Livistona sp., and Amesoneuron sp. Fragmentary elongate leaves with parallel venation possibly belong to the Musaceae (Musophyllum) and the Araceae. Petrified wood is also common in the Changchang megaflora. At least two species of morphogenus Paraphyllantoxylon similar in

anatomy to the modern Elaeocarpaceae or Euphorbiaceae have been recognised (Feng et al., 2010). A new species of Altingioxylon has recently been described (Oskolski et al., 2012). 1.4.2. Microfossils Zhang (1980) considered that the flora comprised a mixture of evergreen and deciduous temperate, tropical and sub-tropical plants but in a more detailed study Lei et al. (1992) recognised three palynological assemblages. Collected from the lowest layer of coal-bearing series, Assemblage I was characterised by Procolpopollenites miikensis– Polypodiaceaes porites haardti and was dominated by ferns (52.5%), with the proportion of angiosperms and gymnosperms being 34.85% and 12.65%, respectively. The main elements reported from the assemblage were Poprocolpopollenites miikensis, Gothanipollis bassensis, Momipites coryloides, Caryapollenites triangulus, Monocolpopollenites tranquillus, Abietineae pollenites microaladus f. minor, Pinuspollenites banksianaeformis, Polypodiaceaesporites haardti, Osmundacidites crassiprimarus, Cyclophorusisporites. After comparison with similar assemblages in Japan, France and Hungary, Lei et al. (1992) considered the age of this group to be early Eocene. Assemblage II was referred to as the “Abietineae pollenites–Momipites triletipollenites–Operculumpollis operculatus” association, in which the ratio of angiosperms, gymnosperms, and ferns was 59.72%, 23.19%, and 18.89% respectively. The main taxa reported were Momipites triletipollenites, Ulmipollenites undulosus, Liquidambarpollenites

Plate III. Woody dicot leaf morphotypes of the Changchang Flora, Hainan Island, South China; collection no. scale bar 1 cm. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45.

OTU123, sample no CC710; OTU124, sample no CC417a; OTU126, sample no CC701; OTU127, sample no CC695; OTU128, sample no CC676; OTU130, sample no CC831a; OTU130, sample no CC967-1; OTU134, sample no CC356a; OTU136, sample no CC1035; OTU137, sample no CC1051-2; OTU138, sample no CC812; OTU139, sample no CC940; OTU140, sample no CC796; OTU141, sample no CC1024; OTU142, sample no CC1132a; OTU143, sample no CC1123; OTU144, sample no CC1142; OTU145, sample no CC885; OTU146, sample no CC879; OTU147, sample no CC1181; OTU148, sample no CC1205; OTU149, sample no CC638; OTU150, sample no CC1143; OTU151, sample no CC429-1; OTU152, sample no CC934b; OTU152, sample no CC980a; OTU153, sample no CCF017-2; OTU154, sample no CC974; OTU155, sample no CC1014; OTU156, sample no CC854-1; OTU159, sample no CC281-1; OTU164, sample no CC347-1; OTU164, sample no CC875; OTU166, sample no CC348a; OTU167, sample no CC349a; OTU168, sample no CC359; OTU169, sample no CC365; OTU171, sample no CC398b-1; OTU173, sample no CC413; OTU174, sample no CC434; OTU176, sample no CC439; OTU177, sample no CC917; OTU178, sample no CC498; OTU180, sample no CC755-1; OUT183, sample no CC833.

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

Plate III.

9

10

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

mangelsdorfianus, Operculumpollis operculatus, Cupliferoipollenites cingulum, Cupliferoidaepollenites quisqualis, Abientineaepollenites microalatus f. minor, Pinuspollenites bankisianeformis, Polypodiaceaesporites haardti, P. megahaardti. The age of this composition was regarded as middle Eocene. Assemblage III was recovered from the upper part of the coalbearing series and lower part of the Wayao Formation. This association was composed of Alnipollenites–Quercoidites–Liquidambarpollenites, and the percentage of angiosperms, gymnosperms, and ferns were 73.99%, 15.97% and 10.04% respectively. The main taxa are Quercoidites microhenrici, Q. minor, Q.asper, Alnipollenites verus, Liquidambar pollenites orientaliformis, L. mangelsdorfianus, L. stigmosus, Salixipollenites hians, Ulmipollenites undulosus, Cupliferoidaepollenites quisqualis, Tricolpites, Retitricolpites delicates. The age of this assemblage was considered to be late Eocene. More recent work by Yao et al. (2009) similarly reported three palynological zones from the same section but they were compositionally distinct from those of Lei et al. (1992). Zone 3 of Yao et al. (2009) represents the lower most 19.5 m of their section and is approximately equivalent to Assemblage I of Lei et al. (1992). Yao et al. (2009) report that their Zone 3 is characterised by Faguspollenites, Momipites, Betulaepollenites and Juglanspollenites Raatz that represent deciduous broad-leaved tree taxa. A few herbaceous taxa (Cucurbitaceaepollenites, Corsinipollenites Nakoman, Cruciferaeipites Zheng and Chenopodipollis Krutzsch) are also present at low frequency (0.1–0.2%). The pollen types of tropical and subtropical taxa Myrtaceidites, Palmaepollenites, Proteacidites, Gothanipollis Krutzsch, Symplocoipollenites Potonié, Margocolporites Ramanujam ex Srivastava and Hamamelidaceae are also present in this zone. Gymnosperm taxa (3.0%) are represented by Pinuspollenites, Ephedripites and Taxodiaceaepollenites. Pteridophyte spores account for 4.6% of the assemblage composition with Polypodiaceaesporites, Pterisisporites, Triletes, Lygodiumsporites and Hymenophyllumsporites being present. Zone 2 of Yao et al. (2009) spanned 16 m of their section and was comparatively depauperate in palynomorph diversity with only 18 forms being recovered. Of these 14 represented angiosperms, two gymnosperms and two pteridophyte taxa. As in zone 1, angiosperm pollen dominated the palynoassemblage of this zone (97.9%) and pollen of the Fagaceae predominated. A few tropical and subtropical taxa of Monocolpopollenites Pflug et Thomson and Hamamelidaceae were recovered together with some temperate deciduous broad-leaved tree

taxa such as Betulaepollenites Potonié, Alnipollenites Potonié, Momipites Wodehouse and Ulmipollenites Wolff. Gymnosperm taxa accounted for only 2.0% of the assemblage represented by Pinuspollenites and Taxodiaceaepollenites. Ferns were represented by just two forms, Polypodiaceaesporites (0–0.2%) and Pterisisporites (0–0.1%), and even these only occurred sporadically. Zone 1 of Yao et al. (2009) came from the upper 63.8 m of their section but the topmost 46 m was barren of any fossil pollen or spores. The remaining 17.8 m of section yielded 30 pollen and spore types made up of 22 angiosperm, four gymnosperm and four pteridophyte taxa. Angiosperm pollen (96.9%) represented mostly deciduous broadleaved trees of temperate and subtropical affinities. Fagaceae pollen was particularly abundant in this zone (94.36%) represented by Faguspollenites, Cupuliferoipollenites, Quercoidites and Castanopsis. Myrtaceidites, Proteacidites and Hamamelidaceae indicate tropical and subtropical taxa present in the assemblage. Gymnosperm taxa (2.6%) were represented by Pinuspollenites, Tsugaepollenites, Ephedripites and Taxodiaceaepollenites. Pteridophytes were uncommon (0.5%), being represented by Polypodiaceaesporites, Pterisisporites, Triletes and Cyathidites. Taken together the 46 pollen and spore taxa present in all three of the zones described by Yao et al. (2009) were overwhelmingly those representing angiosperms (36 taxa), while 4 presented gymnosperms and 6 ferns. This reflects a similar composition to that seen in the megafossils even though some families that are abundantly represented by megafossils such as the Lauraceae are absent from the microflora.

2. Material and methods Our palaeoclimatic analysis of the Changchang Flora was performed using the multivariate foliar physiognomic proxy known as Climate Leaf Analysis Multivariate Program (CLAMP) (Wolfe, 1993; Kovach and Spicer, 1995; Spicer, 2008; Yang et al., 2011). Underlying the technique is the use of Canonical Correspondence Analysis (CCA) (ter Braak, 1986) to position modern and fossil leaf physiognomy data relative to one another in multidimensional physiognomic space calibrated using highresolution gridded climate data. The technique has been widely applied to leaf fossil assemblages from mid Cretaceous to Pliocene times (Spicer and Herman, 2010; Yang et al., 2011) and full details of the methodology are given on the CLAMP website (http://clamp.ibcas.ac.cn).

Plate IV. Woody dicot leaf morphotypes of the Changchang Flora, Hainan Island, South China; collection no. scale bar 1 cm. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

OTU187, sample no CC897; OTU188, sample no CC891; OTU189, sample no CC887; OTU192, sample no CC939-1; OTU196, sample no CC1019; OTU198, sample no CC1032; OTU199, sample no CC1043; OTU200, sample no CC1057; OTU201, sample no CC1067; OTU202, sample no CC1180.

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

Based on comparative architecture, including venation as well as overall morphology, 202 woody dicot leaf morphotypes, approximating as far as possible to fossil species, were recovered from the section shown in Figs. 2 and 3. Of these all but 135 morphotypes were fragmentary and the perceived diversity may well be inflated due to specimen incompleteness. Even so the assemblage is extremely diverse. All 202 morphotypes were then scored for a maximum of 31 different characters states following the CLAMP protocols (http://clamp.ibcas.ac.cn) subject to the characters being present in the specimens. The inclusion of fragmentary material yielded a low completeness factor (Herman and Spicer, 1997) of 0.45, which is regarded as being too low to yield reliable palaeoclimate results so the 135 most complete morphotypes were selected to form an edited fossil dataset that yielded a completeness statistic of 0.76. These two fossil datasets are designated Changchang1.csv and Changchang2.csv respectively. Illustrations of representative specimens of the 135 morphotypes (OTUs) are shown in Plates 1–4. CLAMP was calibrated using a high-resolution gridded climate calibration dataset based on New et al. (2002), interpolated and adjusted for altitude following the method of Spicer et al. (2009). The data making up this climate calibration, designated HiResGRIDMetAsia2.csv, included monsoon and non-monsoonal, tropical, subtropical, and temperate climates from 177 modern sites distributed across North America, Japan, China, India, Thailand, the Caribbean and Pacific islands. A leaf physiognomy calibration dataset was then constructed from leaves recovered from natural or naturalised stands of vegetation at the same 177 locations yielding the climate data. At each of these sites leaves from a minimum of 20 woody dicot taxa including trees, vines and shrubs were scored for the same 31 characters states used in the scoring of the fossil laves. A description of the scoring methodology and collecting protocols is given on the CLAMP website. The resulting training dataset was designated PhysgAsia2.csv and is given in the supplementary information

Fig. 5. CLAMP CCA Axis 1 v Axis 2 biplot showing the position of the two Changchang Formation floral assemblage samples (solid red circles) relative to modern vegetation sites from different geographic regions in the PhysgAsia2 dataset. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

11

Fig. 6. CLAMP CCA Axis 1 v Axis 3 biplot showing the position of the two Changchang Formation floral assemblage samples (solid red circles) relative to modern vegetation sites from different geographic regions in the PhysgAsia2 dataset. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

together with the accompanying high-resolution gridded climate data (HiResGRIDMetAsia2.csv). The same datasets were used in Khan et al. (2014).

Fig. 7. CLAMP CCA Axis 2 v Axis 3 biplot showing the position of the two Changchang Formation floral assemblage samples (solid red circles) relative to modern vegetation sites from different geographic regions in the PhysgAsia2 dataset. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

12

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

The analysis was carried out following the procedure given in Yang et al. (2011).

3. Results Fig. 5 shows the CLAMP CCA Axis 1 v Axis 2 biplot showing the position of the Changchang1 and 2 fossil samples (red filled circles) relative to modern calibration sites from different geographic regions. Both the samples plot close to one another despite the low completeness statistic for Chanchang1. They both plot in a region of physiognomic space populated by modern vegetation sites from eastern Asia, southern Asia (India) and southeastern Asia (Thailand), Polynesia and Melanesia. In the Axis 1 v Axis 3 plot (Fig. 6) there is an apparent proximity to sites from central America but this is an artefact of viewing the plot along the direction of Axis 2, which separates the non-monsoonal central American sites (negative Axis 2 scores) from those experiencing a strong Asian monsoon (positive Axis 2 scores) (Fig. 7). Above an axis 2 score of + 1 the strength of the monsoon becomes more pronounced and increases with increasing axis scores. Note however that the Changchang sites have scores slightly less than one and plot close to those sites having a weak monsoon or none at all. Figs. 8–12 show the 2nd order polynomial regression model used to generate the predictions for mean annual temperature (MAT) (Fig. 8), mean temperatures for the warmest and coldest months (WMMT and CMMT) (Figs. 9 and 10 respectively), and precipitation during the three consecutive wettest and driest months (3-WET and 3-DRY) (Figs. 11 and 12 respectively). Uncertainties (± 1 s.d.) in predictions are shown by the vertical bars from the yellow-filled circles representing the Changchang samples. Note that the fossil sites plot at the peak of the regression curve for the WMMT (Fig. 9) where there is a large scatter of calibration sites. At this part of the plot the prediction carries larger uncertainties than indicated by the standard deviation value. The predictions for the MAT and CMMT are more precise. If the

Fig. 8. Mean annual temperature (MAT) vector scores of vegetation sites in the PhysgAsia2 physiognomic dataset plotted against the observed (MAT) as derived from a highresolution (0.16° by 0.16°) gridded meteorological dataset collected between 1961 and 1990. The positions of the Changchang fossil data are shown by yellow-filled circles with 1 s.d. error bars. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 9. Warm month mean temperature (WMMT) vector scores of vegetation sites in the PhysgAsia2 physiognomic dataset plotted against the observed WMMT as derived from a high-resolution (0.16° by 0.16°) gridded meteorological dataset collected between 1961 and 1990. The positions of the Changchang fossil data are shown by yellow-filled circles with 1 s.d. error bars. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

difference between the MAT and the CMMT (10.5 °C) is added to the MAT this gives a WMMT of 32 °C which may be more reliable than that suggested by the regression in Fig. 9.

Fig. 10. Cold month mean temperature (CMMT) vector scores of vegetation sites in the PhysgAsia2 physiognomic dataset plotted against the observed CMMT as derived from a high-resolution (0.16° by 0.16°) gridded meteorological dataset collected between 1961 and 1990. The positions of the Changchang fossil data are shown by yellow-filled circles with 1 s.d. error bars. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

13

dry season in the time of the Changchang Flora lacked pronounced drought (Fig. 10). The predicted values for all the CLAMP-returned climate variables are given in Table 1. The Changchang samples plot at the wet end of the growing season precipitation (GSP) regression and because the length of the growing season is close to 12 months the GSP is effectively the mean annual precipitation. Relative humidity, averaged over the year was close to 70%. 4. Discussion 4.1. Mean annual temperatures

Fig. 11. Precipitation during the three consecutive wettest months (3-WET) vector scores of vegetation sites in the PhysgAsia2 physiognomic dataset plotted against the observed 3WET as derived from a high-resolution (0.16° by 0.16°) gridded meteorological dataset collected between 1961 and 1990. The positions of the Changchang fossil data are shown by yellow-filled circles with 1 s.d. error bars. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

In respect of the precipitation regime, Figs. 9 and 10 indicate an overall humid regime. The Changchang samples plot close to modern day sites in India that have amongst the driest wet seasons, while the dry season is much wetter than any site in modern India. Thus even the

Both fossil sample datasets (Changchang 1 and 2) plot in almost identical positions suggesting that the inclusion of large numbers of fragmentary leaf forms in dataset Changchang 1 does not significantly degrade the obtainable climate signal. Nevertheless, for simplicity, results for the Changchang 2 sample, with its higher proportion of complete leaves, will be used in this discussion. The MAT derived from the CLAMP analysis for the Changchang 2 sample is 21.6 ± 4.7 °C (Table 1) and cooler than might be expected at the margins of the Eocene palaeotropics given the present day thermal regime and the known low equator-to-pole temperature gradient throughout the Eocene. It is markedly cooler than the SSTs obtained from minimally altered isotopic signatures and organic geochemical proxies such as TEX86 (Pearson et al., 2001; Pearson et al., 2007), and ~ 3–4 °C cooler than the MAT at sea level on Hainan Island today (see supplementary data). However, the CLAMP values are higher than MAT estimates obtained by Yao et al. (2009) from the Changchang palynoflora using the Co-existence Approach (CoA) (Mosbrugger and Utescher, 1997). CoA yielded MATs of 14.2–19.8, 13.3–22.6, and 14.2– 19.8 °C for their zones 3–1 respectively, and an overall MAT for the whole section of 14.2–19.8 °C (Yao et al., 2009). These cool temperatures immediately raise the issue of the altitude of the Changchang Basin in middle Eocene time. The CLAMP-derived enthalpy estimate of 340 kJ/kg is close to that obtained for other Eocene Asian floras (unpublished data) known to have been deposited at or close to sea level and suggests a palaeoelevation no higher than 1300 m. Assuming this was the altitude of the Changchang Basin in middle Eocene times, and applying a lapse rate correction, this could explain no more than ~6 °C lower MAT than for coeval sea level locations. This only raises the MAT to 27 °C which is still 3–7 °C cooler than the equatorial SSTs obtained by Pearson et al. (2001 and 2007). However, recent studies of early Eocene (54–52 Ma) bivalves from the Gulf Coastal Plain of the USA (palaeolatitude ~30°N, Müller et al., 2011) indicate an MAT of ~27 °C (Keating-Bitonti et al., 2011). This study combined stable oxygen isotope, clumped isotopes, and strontium isotope analyses of shell carbonate with tetraether lipid proxies (TEX86, BIT [branched and isoprenoid tetraether], MBT/CBT [methylation of branched tetraethers/ cyclization of branched tetraethers]) from organic matter in sediment enclosed by articulated shells. This allowed correction for salinity effects that influence stable isotope-derived temperature estimates. While there are dangers in extrapolating across time and space, both our floral, and the coastal water derived temperatures obtained by Keating-Bitonti et al. (2011), suggest that the Eocene low latitudes were perhaps not as uniformly warm as previously thought. More specific to the terrestrial realm is that the MBT/CBT proxy specifically targets terrestrially derived organic matter, and indicates that soil palaeotemperatures on the Gulf Coastal Plain were between 25 and 28 °C. 4.2. Seasonality

Fig. 12. Precipitation during the three consecutive wettest months (3-DRY) vector scores of vegetation sites in the PhysgAsia2 physiognomic dataset plotted against the observed 3DRY as derived from a high-resolution (0.16° by 0.16°) gridded meteorological dataset collected between 1961 and 1990. The positions of the Changchang fossil data are shown by yellow-filled circles with 1 s.d. error bars. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Both CLAMP and CoA indicate that some degree of seasonality was experienced by the Changchang palaeovegetation. The CLAMPpredicted CMMT is ~11 °C ± 7.2 °C (Table 1) while CoA yielded a combined estimate of 1.7–11.9 °C for the palynoflora as a whole. Only zone 2

14

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

Table 1 CLAMP-derived estimates for mean annual temperature (MAT), warm month mean temperature (WMMT), cold month mean temperature (CMMT), length of the growing season (LGS), precipitation during the growing season (GSP), mean monthly growing season precipitation (MMGSP), precipitation during the three consecutive wettest months (3-WET), three consecutive driest months (3-DRY), relative humidity (RH), specific humidity (SH) and enthalpy.

Changchang1 Changchang2 Regression R2 Uncertainties (2 s.d.)

MAT (°C)

WMMT (°C)

CMMT (°C)

LGS (Months)

GSP (cm)

MMGSP (cm)

3_WET (cm)

3_DRY (cm)

RH (%)

SH (g/kg)

ENTHAL (kJ/kg)

21.6 21.3 0.87 4.7

28.4 28.4 0.63 5.6

11.2 10.8 0.84 7.2

11.7 11.5 0.87 2.1

200.7 202.0 0.52 122

19.5 19.7 0.44 12.2

87.7 88.0 0.34 71.6

21.2 22.6 0.38 18.9

68.4 69.0 0.55 16.7

11.4 11.3 0.78 3.7

340 339 0.83 17

departs from this estimate in that it suggested the range could be extended to give a CMMT as high as 15.3 °C (Yao et al., 2009). It seems that both the pollen and the megaflora suggest similar results in that the CLAMP mean value falls within the CoA range, but towards its upper end. At ~11 °C the cold month mean is 10.6 °C cooler than the annual mean temperature. For comparison purposes the sea level CMMT (air) on Hainan Island today is 18–19 °C (Supplementary data). Mean temperatures for the warm month are likely to be underestimated in our analysis because the Changchang megafloral samples plot at the peak of the PhysgAsia2 regression curve (Fig. 9) where the correlation between leaf form and the WMMT breaks down. An alternative and potentially more reliable estimate of the WMMT can be gained from adding the difference between the CMMT and MAT to the MAT. Thus the WMMT rises to ~ 32 ± 7.2 °C. Today the sea level WMMT (air) on Hainan Island is 28–29 °C (Supplementary data). Tree species diversity, canopy and leaf architecture all affect leaf temperature and mean canopy temperatures of temperate forest trees deviate substantially from air temperature, but in a highly species-specific manner (Leuzinger and Körner, 2007). When closed canopy forests, whether temperate or tropical, growing in moist soils are exposed to dry air masses significant evapotranspirational cooling can take place, lowering the sub-canopy temperatures by as much as 7 °C (Fritts, 1961; Spicer et al., 2011). This requires adjustment to the CLAMP thermal estimates (Srivastava et al., 2012). However, with both 3-WET, 3-Dry, GSP and RH all indicating a moist regime (Table 1) such cooling is likely to be b 4 °C. Even in the summer the monthly average and daily peak temperatures are unlikely to have been lethal to plant life and only occasionally would have compromised photosynthetic efficiency. The mean annual range of temperature of approximately 22 °C is surprisingly large for such a low latitude flora. However, the CLAMP analysis suggests that neither temperature nor water availability would have been limiting to growth and with a temperature defined duration of the growth season (11.5 ± 2.1 months) it is unlikely that seasonality would have been expressed as strongly developed tree rings. Fossil wood from the Changchang Formation confirms this with only one taxon (Paraphyllanthoxylon) exhibiting a single weakly developed ring (Oskolski, pers. comm., 2013). In five other samples of Paraphyllanthoxylon there are no rings and rings are also absent in samples of Altingioxylon. At most there appears to be some rhymicity in changes of vessel size but these are subtle. In the coeval Maoming Basin sediments wood samples display similar rhythmicity, but again these variations are indistinct. Overall the lack of ring development indicates a climate with insufficient seasonal variation in temperature or water availability to affect growth, and there is no evidence of a cold and/or dry season. The position of the Changchang Flora in CLAMP physiognomic space, particularly along CCA axis 2 (Fig. 7), also suggests that a strong monsoon was not a feature of the Hainan region climate in middle Eocene times. The 3-WET : 3-DRY ratio is b6:1 which defines it as nonmonsoonal (Lau and Yang, 1997; Zhang and Wang, 2008). On Hainan Island today the ratio is ~ 11:1 (Supplementary data) and the GSP is ~128–177 cm depending on aspect and altitude; somewhat drier than the 200 cm predicted by CLAMP for the Eocene. Note, however, that because the Eocene regime is overall wet, leaf physiognomy is poorly

constrained by soil water availability and the prediction uncertainties are correspondingly high. By contrast palynology-based CoA suggests a much drier regime with mean annual rainfall (roughly equivalent to GSP because of the year round growing season) predicted as being ~ 78 cm (Yao et al., 2009). This difference is not surprising. The drier regime and lower temperatures predicted by CoA could be a function of the larger catchment area sourcing the pollen and spores. Cooler, better-drained upland sites are more likely to be presenting in the palynomorph spectrum than the leaf flora because pollen and spores can be transported further. It is worth noting that the composition of the palynofloras reported by Lei et al. (1992) and Yao et al. (2009) exhibit some notable differences in the proportion of angiosperms to gymnosperms and ferns, as well as species composition within the angiosperms. This may be due to differing sampling and preparation techniques. It is unlikely, however, that the disparity is due to within-section variation because the CoA suggests a remarkably stable environment throughout the period of Changchang Formation deposition. This stability justifies our approach of analysing a megaflora collected from throughout the section. By sampling several sedimentary facies we minimise facies-specific taphonomic biases. The occurrence of thermophyllic animals (crocodiles, turtles, etc.) and plants is compatible with the temperature ranges indicated by the megaflora using CLAMP, but less so with the low end of the CMMT range suggested by palynofloral estimates using CoA (~1 to 2 °C). Because CoA is based on the assumption that environmental tolerance is conserved over time it is usually regarded as being most reliable for Neogene and younger material. Its use in the Eocene is likely to be associated with higher uncertainty. Nevertheless the possibility that the palynoflora may contain a significant proportion of material derived from plants growing at high (cooler) elevations could also explain why the technique yields cooler estimates overall. Some elements of the palynoflora show clear affinities with some modern seasonally deciduous temperate as well as tropical taxa (Lei et al., 1992; Yao et al., 2009). The CLAMP-derived temperatures, corrected for palaeoelevation, are similar to SSTs recently obtained from Eocene low latitude coastal settings using multiproxy isotopic and organic geochemical techniques (Keating-Bitonti et al., 2011), but distinctly cooler than those derived from oxygen isotopes alone. The δ18O data of Keating-Bitonti et al. (2011) also showed a degree of thermal seasonality (summer maximum temperatures ranging from 31 to 43 °C, compared to an MAT of 26.6 ± 1.1 °C using the same proxy and assuming normal marine salinity), but summer and winter temperatures were indistinguishable using clumped isotopes and organic geochemical proxies. While thermally buffered shallow marine settings are likely to display less of a seasonality signal than our estimated range in air temperature, that our seasonal range is so high is intriguing. A moderate palaeoelevation could explain some, but not all, of our perceived seasonality. 5. Conclusions The middle Eocene Changchang Flora of Hainan Island provides an important insight into both the vegetation and climate in near equatorial latitudes during a previous ‘hothouse’ climate. Taxa whose present day relatives exist in both tropical and temperate climates were

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

admixed in an angiosperm-dominated highly diverse near-equatorial forest. It appears that both micro and megafloral records in nonmarine sediments reveal a climate that was humid year round with a wet season perhaps four times as wet as the dry season but no extreme rainfall seasonality so typical of the modern Southeast Asian monsoon system. The middle Eocene vegetation of Hainan Island lived at a time before the onset of the current Hainan monsoon-driven seasonality of marked wet and dry seasons. There was, however distinct seasonality in temperature even though the growth characteristics of wood and leaf physiognomy suggest year round productivity. Multivariate analysis of the leaf flora indicates a mean annual temperature of 22 ± 4.7 °C, with a mean annual temperature range of 21 °C. The pollen and spore record may capture more seasonal conditions in vegetation at higher elevations nearby, but the general lack of ring development in fossil woods suggest that temperatures were never low enough to limit tree growth within the catchment basin. By comparison with ∂18O records our temperatures are cool but, allowing for a possible higher elevation of the Changchange depositional basin during the middle Eocene, are similar to those obtained from a shallow marine setting in the Gulf Coast of North America at a palaeolatitude of ~ 30°N. These cool terrestrial and marine temperatures suggest significant heat transport took place from the equator to the poles during the Eocene hothouse world by mechanisms yet to be fully understood and modelled. Acknowledgements This study was supported by the National Natural Science Foundation of China (nos. 41210001, 31070200), the National Basic Research Program of China (973 Program) (no. 2012CB822003), the joint Project of the National Natural Science Foundation of China and the Russian Foundation for Basic Research (nos. 413111040, 14-05-91163), the Fundamental Research Funds for the Central Universities (no. 12lgjc04), and the Key Project of the Sun Yat-sen University for inviting foreign teachers. The work of RAS was in part supported by a Visiting Professorship for Senior International Scientists awarded to R.A.S. by the Chinese Academy of Sciences (2009S1-20), by an International S & T Cooperation Project of China no. 2009DFA32210. The authors thank the graduate students majoring in Botany at Sun Yat-sen University for their field work and fossil collecting. We also offer sincere gratitude to Dr. A. A. Oskolski, Komarov Botanical Institute (RAS), Russia and Ch. -Ch Hofmann, Department of Paleontology, Geocentre, Austria for their suggestions. References Aleksandrova, G.N.,Kodrul, T.M.,Liu, X.,Song, Y.,Jin, J.H., 2012. Palynological characteristics of the upper part of the Youganwo Formation and lower part of the Huangniuling Formation, Maoming Basin, South China. In: Jin, J.H., Tang, B. (Eds.), Proceedings of the 2nd Sino-Russian Seminar on Evolution and Development of Eastern Asian Flora. Sun Yat-Sen University, pp. 3–15. Bralower, T.J.,Zachos, J.,Thomas, E.,Parrow, M.,Paull, C.K.,Kelley, D.C.,Primoli Silva, I.,Sliter, W.V.,Lohmann, K.C., 1995. Late Paleocene to Eocene paleoceanography of the equatorial Pacific Ocean: stable isotopes recorded at Ocean Drilling Program Site 865: Allison Guyot. Paleoceanography 10, 841–865. D'Hondt, S.,Arthur, M., 1996. Late Cretaceous oceans and the cool tropics paradox. Science 271, 1838–1841. Dutton, A.,Lohmann, K.C.,Leckie, R.M., 2005. Data report: stable isotope and Mg/Ca of Paleocene and Eocene foraminifers, ODP Site 1209, Shatsky Rise. In: Primoli Silva, I., Matione, M.J. (Eds.), Proceedings of the Ocean Drilling Program, Scientific results, College Station, Texas, pp. 1–19. Eberle, J.J., Greenwood, D.R., 2012. Life at the top of the greenhouse Eocene world — A review of the Eocene flora and vertebrate fauna from Canada's High Arctic. Geol. Soc. Am. Bull. 124, 3–23. Feng, X.X., Yi, T.M., Jin, J.H., 2010. First record of Paraphyllanthoxylon from China. IAWA J. 31, 89–94. Feng, X.X., Jin, J.H., 2012. First record of extinct fruit genus Chaneya in low-latitude tropic of South China. Sci. China Ser. D Earth Sci. 55, 728–732. Fritts, H.C., 1961. An analysis of maximum summer temperatures inside and outside a forest. Ecology 42, 436–440. Gradstein, F.M., Ogg, J.G., Smith, A.G., 2005. A Geologic Time Scale 2004. Cambridge University Press, Cambridge, p. 610. Greenwood, D.R., Wing, S.L., 1995. Eocene continental climates and latitudinal temperature gradients. Geology 23, 1044–1048.

15

Guo, S.X., 1979. Late Cretaceous and Early Tertiary Floras from the Southern Guangdong and Guangxi with their Stratigraphic Significance, Mesozoic and Cenozoic Red Beds of South China. Science Press, Beijing. Harrington, G.J.,Jaramillo, C.A., 2007. Paratropical floral extinction in the Late Palaeocene– Early Eocene. J. Geol. Soc. Lond. 164, 323–332. He, X.Y., Shen, R.J., Jin, J.H., 2010. A new species of Nelumbo from South China and its palaeoecological implications. Rev. Palaeobot. Palynol. 162, 159–167. Herman, A.B., Spicer, R.A., 1997. New quantitative palaeoclimate data for the Late Cretaceous Arctic: evidence for a warm polar ocean. Palaeogeogr. Palaeoclimatol. Palaeoecol. 128, 227–251. Huber, M., 2008. A hotter greenhouse? Science 321, 353–354. Jin, J.H., 2008. On the age of the Youganwo Formation in the Maoming Basin, Guandong Province. J. Stratigr. 32, 47–50. Jin, J.H., 2009. Two Eocene fossil fruits from the Changchang Basin of Hainan Island, China. Rev. Palaeobot. Palynol. 153, 150–152. Jin, J.H.,Kodrul, T.M., 2008. Review of “Cyclocarya scutellata” Guo. 75th Anniversary Annual Meeting of Botanical Society of Chinapp. 49–50 Abstracts. Jin, J.H.,Kodrul, T.M.,Liao, W.B., 2009a. Review of “Cyclocarya scutellata” Guo. Acta Sci. Nat. Univ. Sunyatseni 48, 149–150. Jin, J.H., Kodrul, T.M., Liao, W.B., 2009b. Discovery of Craigia (Malvaceae, Tilioideae) from the Eocene Changchang Formation of the Hainan Island, South China. Climatic and Biotic Events of the Paleogene (CBEP 2009), Wellington, New Zealand, p. 109a. Jin, J.H., Qiu, J., Zhu, Y., Kodrul, T.M., 2010. First fossil record of the genus Nageia (Podocarpaceae) in south China and its phytogeographic implications. Plant Syst. Evol. 285, 159–163. Keating-Bitonti, C.R., Ivany, L.C., Affek, H.P., Douglas, P., Samson, S.D., 2011. Warm, not super-hot, temperatures in the early Eocene subtropics. Geology 39, 771–774. Khan, M.A.,Spicer, R.A.,Bera, S.,Ghosh, R.,Yang, J.,Spicer, T.E.V.,Guo, S.-X.,Su, T.,Jacques, F., Gote, P.J., 2014. Miocene to Pleistocene floras and climate of the Eastern Himalayan Siwaliks, and new palaeoelevation estimates for the Namling–Oiyug Basin, Tibet. Glob. Planet. Chang. 113, 1–10. Kodrul, T.M., Jin, J.H., Aleksandrova, G.N., Herman, A.B., Feng, X.X., Song, Y., Liu, X., 2012a. Phytogeographical relationships of the low-latitude Eocene floras of South China (Changchang Basin, Hainan Province and Maoming Basin, Guangdong Province). Proceedings of the 2nd Sino-Russian Seminar on Evolution and Development of Eastern Asian Flora. Sun Yat-Sen University, Gunagzhou, China, pp. 37–46. Kodrul, T.M., Jin, J.H., Aleksandrova, G.N., Zhou, W., Li, W., Song, Y., Liu, X., 2012b. Preliminary data on the Cenozoic floras of the Maoming Basin (Guangdong Province, China). Internat. Conf. Dedicated to the 100th Jubilee of V.A. Vakhrameev (1912–1986), Moscow, Russia, p. 28 ((in Russian), Abstracts). Kovach, W.L., Spicer, R.A., 1995. Canonical correspondence analysis of leaf physiognomy: a contribution to the development of a new palaeoclimatological tool. Palaeoclimates 1, 125–173. Lau, K.M., Yang, S., 1997. Climatology and interannual variability of the southeast Asian summer monsoon. Adv. Atmos. Sci. 14, 141–162. Lei, Y.Z.,Zhang, Q.R., He, W.,Cao, X.P., 1992. Tertiary, Geology of Hainan Island. Geological Publishing House, Beijing, pp. 218–266. Leuzinger, S.,Körner, C., 2007. Tree species diversity affects canopy leaf temperatures in a mature temperate forest. Agric. For. Meteorol. 146, 29–37. Mosbrugger, V.,Utescher, T., 1997. The coexistence approach — a method for quantitative reconstructions of Tertiary terrestrial palaeoclimate data using plant fossils. Palaeogeogr. Palaeoclimatol. Palaeoecol. 134, 61–86. Müller, D., Gurnis, M., Torsvik, T., 2011. GPlates version 1.0.1. Muraki, Y., Tsuyama, M., Kobayashi, Y., Kodama, H., Iba, K., 2000. Trienoic fatty acids and plant tolerance of high temperature. Science 287, 476–479. New, M.,Lister, D.,Hulme, M.,Makin, I., 2002. A high-resolution data set of surface climate over global land areas. Clim. Res. 21, 1–15. Norris, R.D., Wilson, P.A., 1998. Low-latitude sea-surface temperatures for the midCretaceous and the evolution of planktic foraminifera. Geology 26, 823–828. Oskolski, A.A., Kodrul, T.M., Jin, J.H., 2012. Altingioxylon hainanensis sp. nov.: earliest fossil wood record of the family Altingiaceae in Eastern Asia and its implications for historical biogeography. Plant Syst. Evol. 298, 661–669. Pearson, P.N.,Ditchfield, P.W.,Singano, J.M.,Harcourt-Brown, K.G.,Nicholas, C.J.,Olsson, R.K. ,Shackleton, N.J.,Hall, M.A., 2001. Warm tropical sea surface temperatures in the Late Cretaceous and Eocene epochs. Nature 413, 481–487. Pearson, P.N.,van Dongen, B.E.,Nicholas, C.J.,Pancost, R.D.,Schouten, S.,Singano, J.M.,Wade, B.S., 2007. Stable warm tropical climate through the Eocene epoch. Geology 35, 211–214. Schrag, D.P., 1999. Effects of diagenesis on the isotopic record of late Paleogene tropical sea surface temperatures. Chem. Geol. 161, 215–244. Scotese, C.R., 2001. Earth System History Geographic Information System v. 02b. PALEOMAP Project, Arlington, Texas. Shackleton, N.J.,Boersma, A., 1981. The climate of the Eocene ocean. J. Geol. Soc. Lond. 138, 153–157. Sharkey, 2000. Some like it hot. Science 287, 435–437. Spicer, R.A., 2008. CLAMP. In: Gornitz, V. (Ed.), Encyclopedia of Paleoclimatology and Ancient Environments. Springer, Dordrecht, pp. 156–158. Spicer, R.A.,Herman, A.B., 2010. The Late Cretaceous environment of the Arctic: a quantitative reassessment using plant fossils. Palaeogeogr. Palaeoclimatol. Palaeoecol. 295, 423–442. Spicer, R.A., Valdes, P.J., Spicer, T.E.V., Craggs, H.J., Srivastava, G., Mehrotra, R.C., Yang, J., 2009. New developments in CLAMP: calibration using global gridded meteorological data. Palaeogeogr. Palaeoclimatol. Palaeoecol. 283, 91–98. Spicer, R.A., Bera, S., De Bera, S., Spicer, T.E.V., Srivastava, G., Mehrotra, R.C., Mehrotra, N.C., Yang, J., 2011. Why do foliar physiognomic climate estimates sometimes differ from those observed? Insights from taphonomic information loss and a CLAMP case study from the Ganges Delta. Palaeogeogr. Palaeoclimatol. Palaeoecol. 299, 39–48.

16

R.A. Spicer et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 412 (2014) 1–16

Srivastava, G., Spicer, R.A., Spicer, T.E.V., Yang, J., Kumar, M., Mehrotra, R.C., Mehrotra, N.C., 2012. Megaflora and palaeoclimate of a Late Oligocene tropical delta, Makum Coalfield, Assam: evidence for the early development of the South Asia Monsoon. Palaeogeogr. Palaeoclimatol. Palaeoecol. 342–343, 130–142. ter Braak, C.J.F., 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67, 1167–1179. Tong, H.,Zhang, J.-Y.,Li, J.-J., 2010. Anosteira maomingensis (Testudines: Carettochelyidae) from the Late Eocene of Maoming, Guangdong, southern China: new material and redescription. N. Jb. Geol. Paläont. (Abh.) 256, 279–290. Valdes, P.J., 2000. Warm Climate Forcing Mechanisms. Cambridge University Press, Cambridge. Wang, J., Li, H., Zhu, Z., Seguin, M.K., Yang, J., Zhang, G., 1994. Magnetostratigraphy of Tertiary rocks from Maoming Basin, Guangdong Province, China. Chin. J. Geochem. 13, 165–175. Wang, Y.-Y., Zhang, Z.-H., Jin, J.-H., 2007. Discovery of Eocene fossil mammal from Maoming Basin, Guandong. Acta Sci. Nat. Univ. Sunyatseni 46, 131–133. Wang, L.,Xu, Q.Q.,Jin, J.H., 2014. A reconstruction of the Fossil Salvinia from the Eocene of Hainan Island, South China. Rev. Palaeobot. Palynol. 203, 12–21. Wilson, P.A.,Norris, R.D.,Cooper, M.J., 2002. Testing the Cretaceous green-house hypothesis using glassy foraminiferal calcite from the core of the Turonian tropics on Demerara Rise. Geology 30, 607–610. Wolfe, J.A., 1993. A method of obtaining climatic parameters from leaf assemblages: United States Geological Survey Bulletin. U. S. Geol. Surv. Bull. 2040, 1–73.

Yang, J., Spicer, R.A., Spicer, T.E.V., Li, C.-S., 2011. ‘CLAMP Online’: a new web-based palaeoclimate tool and its application to the terrestrial Paleogene and Neogene of North America. Palaeobiodivers. Palaeoenviron. 91, 163–183. Yao, Y.F.,Bera, S., Ferguson, D.K.,Mosbrugger, V., Paudayal, K.N., Jin, J.H., 2009. Reconstruction of paleovegetation and paleoclimate in the Early and Middle Eocene, Hainan Island, China. Clim. Chang. 92, 169–189. Ye, D.Q.,Zhong, X.C.,Yao, Y.M.,Yang, F.,Zhang, S.B.,Jiang, Z.X.,Wang, Y.C.,Sun, Z.C.,Yang, S.Z., Zhao, X.L., Shen, H., Liang, H.D., Tang, W.S., Guan, X.T., Zhao, C.B., 1996. Tertiary in Petroliferous Regions of China. Petroleum Industry Press, Beijing. Zachos, J., Pagani, M., Sloan, L., Thomas, E., Billups, K., 2001. Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292, 686–693. Zhang, Q.R., 1980. Stratigraphy and Palaeontology. Yichang Institute of Geology and Mineral Resources, Chinese Academy of Geological Sciences, Yichang, pp. 106–117. Zhang, S.P., Wang, B., 2008. Global monsoon summer rainy seasons. Int. J. Climatol. 28, 1563–1578. Zhou, G.Q., Chen, P.Q., 1988. Tertiary, Regional Geology of Guangdong Province, People's Republic of China. Geological Publishing House, Beijing, pp. 237–263. Zhou, W.J., Liu, X.Y., Xu, Q.Q., Huang, K.Y., Jin, J.H., 2013. New coryphoid fossil palm leaves (Arecaceae: Coryphoideae) from the Eocene Changchang Basin of Hainan Island, South China. Sci. China. Earth Sci. 56 (9), 1493–1501.