A late Pliocene (1.8–1.9 Ma) wetland fossil forest community that was dominated by Metasequoia and Glyptostrobus was reconstructed based on the species composition of the stumps and other plant macrofossil assemblages. The plant fossils were recovered from a fossil forest preserved in deposits of the Kobiwako Group that are exposed in the Echi River, Shiga Prefecture, central Japan. Fossil wood of Metasequoia and Glyptostrobus was distinguished based on anatomical characteristics. Apportionment of the wood from different horizons in the fossil forest indicates Metasequoia grew over a long period of time in a stable environment, while Glyptostrobus and Alnus grew in unstable environments characterized by shortinterval floods. The fossil forest as a whole represents a fluvial back-marsh environment.
The upland forest was composed of mixed evergreen conifers and deciduous broad-leaved trees including Chamaecyparis pisifera, Tsuga, Magnolia, and Acer. Aquatic and wetland herbs such as Cyperus, Carex, Scirpus, Polygonum, and Menyanthes grew in and around the wetland forest. The assemblage of fossil plants recovered from the fossil forest consists of plants that are currently distributed in the cool temperate climate zone, such as Picea, Thuja, Betula maximowicziana, Pterocarya rhoifolia, and Menyanthes trifoliata. Glyptostrobus is distributed only in subtropical areas at present, but it also grew under a cool temperate climate zone in Japan during the latest Pliocene.
Introduction
Metasequoia and Glyptostrobus are monotypic genera of the Taxodiaceae (Cupressaceae sensu lato) and fossils of these genera are common in Tertiary-age floras of the Northern Hemisphere (Florin, 1963; Yu, 1995). Their fossil records show they were distributed widely in both the Eurasian and North American continents during the Paleogene, but went extinct in North America and Europe during the Miocene and Pliocene and are now restricted to small populations in eastern Asia (Matsumoto et al., 1997; LePage et al., 2005; LePage, 2007).
Metasequoia glyptostroboides Hu et Cheng is naturally distributed in humid and slightly waterlogged places in mountain valleys in the middle reaches of the Yangze River (Fu and Jin, 1992; LePage et al., 2005). On the other hand, Glyptostrobus pensilis (D. Don) K. Koch is distributed in wetlands and waterlogged places in the lowlands of northern Vietnam and southern China (Li and Xia, 2004; Farjon, 2005). Although the two taxa are now distributed in different habitats and in separate geographical regions, remains of both taxa have been found together in fossil assemblages from the Tertiary of the Canadian Arctic (e.g., McIver and Basinger, 1993), northeastern Russia (e.g., Budantsev, 1994), and Japan (e.g., Miki, 1948). Such fossil assemblages sometimes accompany fossil stumps that have been described as taxodiaceous wood (e.g., Christie and McMillan, 1991; Williams et al., 2003a, 2003b).
From the late Pliocene sediments (ca. 1.8 Ma) of the Kobiwako Group, a fossil forest dominated by taxodiaceous wood was discovered with cones and shoots of Metasequoia and Glyptostrobus in the bed of the Echi River, Shiga Prefecture, central Japan (Echigawa Petrified Forest Research Group, 1993; Figure 1). This fossil forest is the youngest in the world in which macrofossils of both genera occur together with taxodiaceous wood.
Itoh (1993) identified the fossil wood as that of “Taxodiaceae”, but did not recognize characters that would allow the wood to be identified further as either that of Metasequoia or Glyptostrobus.
Differences in anatomical features between Metasequoia and Glyptostrobus were clarified by Visscher and Jagels (2003) based on their observation of wood from extant trees. Kumagai et al. (1995) had already separated fossil wood from the Eocene fossil forests of the Canadian Arctic into Metasequoia and “cf. Glyptostrobus” based on wood anatomy. Given that these studies indicated that Metasequoia and Glyptostrobus should also be distinguishable on the basis of wood anatomy, this approach was used to provide a detailed reconstruction of the spatial and temporal distribution of Metasequoia and Glyptostrobus in the late Pliocene Echigawa fossil forest. Although these previous studies did not identify taphonomic differences between the layers containing Glyptostrobus and those containing Metasequoia, there should have been recognizable or at least subtle differences in the depositional environments given the different ecological requirements the extant species need to grow and reproduce (Fu and Jin, 1992; Vann, 2005).
In this study, we reexamined the fossil wood described by Itoh (1993) from the Echigawa fossil forest to distinguish Metasequoia and Glyptostrobus wood based on the criteria of Visscher and Jagels (2003) so that we could assess the temporal and spatial distribution of these species within this fossil forest. In addition, based on the species composition of the plant macrofossil and pollen assemblages preserved within the fossil forest deposits, we reconstructed the paleovegetation in and around this late Pliocene Metasequoia-and Glyptostrobusdominated fossil forest.
Geology and occurrence of the fossil forest
A late Pliocene fossil forest was found in the riverbed of the Echi River in Higashiomi City, Shiga Prefecture (35°5′N, 136°17′E, altitude 178 m), central Japan (Figures 1, 2). The sediment sequence containing the fossil wood is about 13.4 m thick and consists of sand, mud, and organic silt that is assigned to the Kobiwako Group. The uppermost part of the unit is 2.5 m thick (from HA to HC in Figure 3), and crops out widely in the riverbed. Because the dip of these sediments is almost horizontal, abundant in situ fossil stumps and roots are also exposed on the bedding plane (Figure 1). The part of the unit containing the stumps and roots is composed of silt, organic silt, mud, and alternating layers of silt and sand (Figure 3). Sand and gravel beds attest to fluvial channel-fills and intercalate with the silt and mud (Amemori et al., 1993).
Figure 3.
Columnar sections showing the horizons of fossil stumps and samples locations at the Echigawa site based on Amemori et al. (1993). EP1-7: horizons of plant macrofossil samples. HA, HA-HB, HB, HB-HC, HC: horizons of fossil stumps. ENP-11, 16, 17, 18: horizons of pollen analysis by Mizutani (1993).
Two volcanic beds crop out in and around the study site. The Naka volcanic ash layer indicating the Olduvai normal subchron (C2n) (Nakayama, 1993), lies about 10 m below the uppermost horizon (HC) of fossil stumps (Amemori et al., 1993), while the Kitawaki volcanic ash layer occurs in an outcrop near the study site, and is stratigraphically located about 20 m above the HC horizon (Yoshikawa, 1993). The Kitawaki volcanic ash layer is correlated with the Fukuda volcanic ash layer of the Osaka Group that indicates normal magnetic polarity above the Olduvai normal subchron (C2n) and has been fission-track-dated to 1.75 Ma (Yoshikawa, 1984; Yoshikawa et al., 1996). Therefore, the age of this fossil forest is estimated to be 1.8–1.9 Ma and is late Pliocene in age.
This fossil forest was composed of 129 in situ stumps distributed over an area of about 6,510 m2 (155 m long, 42 m wide) (Echigawa Petrified Forest Research Group, 1993). All of the stumps are mummified and show compression to some extent, as shown by the deformed annual rings. From this assemblage of fossils, three groups of fossils are recognized; stumps with roots, roots with stems that are eroded, and stems (trunks). The Echigawa Petrified Forest Research Group (1993) collected and described in detail 53 well preserved stumps from the 129 exposed specimens. These specimens comprised 19 stumps with trunks and roots, 22 with trunks only, and 12 with root only (Table 1).
Table 1.
List of fossil standing stumps in the bed of the Echi River, based on Echigawa Petrified Forest Research Group (1993).
In this study, the stratigraphy of the fossil forest beds by Amemori et al. (1993) was revised and the five fossil forest beds labeled A to E in descending order from the top of the section were renamed as follows: beds A, C, and E are renamed as HC, HB, and HA, and beds B and D are now called HA-HB and HB-HC, respectively (Figure 3). In beds HC, HB, and HA, the trunk bases and flattened root systems were distributed in the uppermost part of discrete organic-rich silt beds that were about 10 cm thick each. In contrast, the fossil stumps in HB-HC and HA-HB were scattered in sediments composed of poorly sorted silt, mud, and alternating silt and sand layers.
Materials and methods
To identify the fossil wood, we reexamined the specimens that were studied by Itoh (1993) that are now curated at the Research Institute for Sustainable Humanosphere, Kyoto University, Japan (registration numbers: KYO_ID_1_EW–KYO_ID_53_EW). Specimens of the 53 fossil samples collected in 1991 were mounted as cross, radial, and tangential sections on glass slides (Table, 1). Identification of the Metasequoia and Glyptostrobus woods was based on the methodology provided in Visscher and Jagels (2003), and that of the other taxa was based on the IAWA handbook (IAWA, 1989, 2004).
In 1991 plant macrofossils were collected from seven horizons (EP-1 to EP-7; Figure 3) and an inventory of the taxa present was compiled by Yamakawa (1993). However, counts of the number of specimens of each taxon were not compiled. In this part of the study, we reexamined the plant macrofossils previously identified and counted the number of macrofossils in each of the samples. Of the seven horizons, EP-1, 4, and 7 were located in the organic silt situated below the horizons in which the fossil stumps were rooted (HA, HB, and HC, respectively). Sediment blocks of about 2000 cm3 were taken from each horizon and soaked in a 2% aqueous solution of NaOH for two or three days. The solution was then neutralized with 2% acetic acid, and the residues were washed on a series of 1.0 mm, 0.5 mm, and 0.25 mm mesh sieves. The plant macrofossils remaining in the sieves were identified under a stereomicroscope, then placed in vials containing 70% ethanol. All of the plant macrofossil specimens collected for this study were deposited in the Lake Biwa Museum, Shiga Prefecture, Japan (LBM0112000273–0112000545, LBM 0112001671–0112001870).
Results
Identification of fossil wood
Itoh (1993) identified 53 fossil wood samples and grouped them as follows: 26 were Taxodiaceae, six were conifers, 15 were Alnus, two were Rosaceae, one was Fraxinus, two were diffuse-porous wood, and one was ring-porous wood. Registration numbers were not assigned to six samples and the results of their identification were not indicated in that paper. We reexamined the six samples and recognized KYO_ID_1_EW (stump no. 1) as diffuse-porous wood, KYO_ID_45, 48, and 53_EW (stump nos. 45, 48, and 53) as Taxodiaceae, and KYO_ID_49 and 51_EW (stump nos. 49 and 51) as Alnus. Taxodiaceous wood was characterized by possessing an abrupt transition from early wood to late wood and scattered parenchyma cells parallel to annual ring boundaries in cross section, and taxodioid-type cross-field pits.
KYO_ID_5_EW (stump no. 5) reported as a “conifer” by Itoh (1993), was reidentified as Alnus based on the aggregated rays visible in cross and tangential sections. KYO_ID_41_EW (stump no. 41) identified as “Rosaceae” by Itoh (1993), had characteristics of Sorbus. In cross section the ray parenchyma is arranged in two rows with medium-sized vessels scattered in the secondary xylem (Figure 4a), and in radial section many small pits are arranged as reticulate perforations (Figure 4b).
Figure 4.
Fossil wood of Sorbus (KYO_ID_41_EW, stump no. 41, horizon HB-HC): a, cross section, ×100, showing ray parenchyma arranged in two rows, with medium-sized vessels scattered in the secondary xylem; b, radial section, ×400, showing reticulate perforations.
Among the 26 samples identified as Taxodiaceae by Itoh (1993), we recognized four samples as Metase quoia (KYO_ID_2, 3, 28, and 40_EW) and seven as Glyptostrobus (KYO_ID_16, 26, 29, 33, 34, 35, and 44_EW). The most distinctive differences between the Metasequoia and Glyptostrobus woods were the presence of gaps between the ray cells and the arrangement and maximum height of the ray cell stacks (Table 2). Metasequoia lacked gaps between the ray cells in radial and tangential section (Figure 5-2b). The cross-field pits of the taxodioid type were arranged in one or two rows oppositely (Figure 5-2a). The ray cells formed stacks of less than 10 cells for the most part, but maxima of 33–47 cells were observed (Figure 5-2b). In contrast, Glyptostrobus had visible gaps between the ray cells in tangential as well as radial section (Figures 6-2a, 6-2c). The cross-field pits of the taxodioid type were randomly arranged in two or three rows (Figure 6-2a). Most of the stacks of the ray cells were less than 10 cells, as in Metasequoia, but the maximum cell stacks did not exceed 28 cells in height, which contrasts slightly with what is seen in Metasequoia (Figure 6-2b).
Table 2.
Anatomical features of fossil and extant Metasequoia and Glyptostrobus wood samples
Figure 5.
Extant and fossil wood of Metasequoia. 1. Extant Metasequoia glyptostroboides wood obtained from Nagai Park in Osaka City: a, radial section, ×400; b, tangential section, ×100. 2. Fossil Metasequoia wood (KYO_ID_3_EW, stump no. 3 in horizon HA): a, radial section, ×400, showing no gaps between ray cells, two or three cross-field pits in the radial walls, and uniseriate or opposite biseriate rows of rays in the early wood; b, tangential section, ×100, showing ray parenchyma cells stacked from three to more than 30 cells high.
Figure 6.
Extant and fossil wood of Glyptostrobus. 1. Extant Glyptostrobus pensilis wood from a tree of ca. 20 cm diameter at breast height (DBH) at the Lake Biwa Museum: a, radial section, ×400; b, tangential section, ×100; c, stack of ray parenchyma cells, ×400. 2. Fossil Glyptostrobus wood (KYO_ID_35_EW, stump no. 35 in horizon HB-HC): a, radial section, ×400, with arrow showing a gap between ray cells (also discernable in the tangential section), two cross-field pits in the radial walls, and random or biseriate rows of ray cells in the early wood; b, tangential section, ×100, showing straight uniseriate and rare partly biseriate ray cells, with maximal ray parenchyma cell stacking of usually less than 28 cells; c, ray parenchyma in tangential section, ×400, with arrow showing a gap between ray cells.
The remaining taxodiaceous wood samples were compared to extant samples of Sequoia, Cunninghamia, Cryptomeria, Taiwania, and Taxodium. The fossil samples possessed narrower biseriate rays compared to that of Sequoia and Cunninghamia, lacked the admixed cupressoid-type cross-field pits observed in Taxodium and Taiwania, and the maximum number of cell stacks was more than 11 and differed from Cryptomeria. Although it was not possible to identify the remaining taxodiaceous wood samples to the generic level, we are assuming that these specimens are probably either Metasequoia or Glyptostrobus.
Composition and distribution of fossil stumps
Of the 53 wood samples that we reexamined, 16 were identified as Taxodiaceae, seven were identified as Glyptostrobus, four were Metasequoia, five were conifers, 15 were Alnus, there was one each of Fraxinus, Sorbus, and Rosaceae, and two were of diffuse-porous wood (Table 1). Assuming that the Taxodiaceae samples are Metasequoia and Glyptostrobus, Metasequoia and Glyptostrobus comprised 51% of the samples that were reexamined, and Alnus accounted for 28% of the 53 fossil wood specimens (Table 3).
Table 3.
Composition of fossil wood samples identified from the Echi River
Twelve of the stumps that were more than 75 cm in trunk diameter were identified as conifers: four Metasequoia, one Glyptostrobus, five Taxodiaceae, and two conifers. All four Metasequoia samples were from the largest size class, with a trunk diameter of greater than 100 cm, with the largest having a major axis diameter of 190 cm. On the other hand, the Glyptostrobus samples were from smaller stumps with major axis diameters of between 5 and 29 cm (Table 1). The major axis diameter of the Alnus trunks varied between 14 and 45 cm (22.8 cm in mean) and the two Sorbus and Rosaceae stumps were both less than 11 cm in diameter (Table 1).
In horizons HA, HB, and HC, large Metasequoia and Taxodiaceae stumps were common and Glyptostrobus and Alnus were rare. In contrast, horizons HA-HB and HB-HC contained medium-size stumps of mainly Alnus and Glyptostrobus, with some Taxodiaceae and broadleaf trees such as Sorbus and Rosacae. The largest Glyptostrobus stump was 85 cm in diameter and one large conifer stump of 150 cm in trunk diameter was present in horizons HB-HC and HA-HB, respectively.
Composition of plant macrofossil assemblage
The seven plant macrofossil assemblages (EP-1 to EP-7 in Figure 2) included seed cones, fruits, seeds, leaves, and shoots of representatives from 32 families, 46 genera, and 60 species (Table 4; Figure 7). Tree taxa comprised 16 families, 23 genera, and 27 species. Three of the tree genera are extinct from Japan: Metasequoia, Glyptostrobus, and Hemiptelea. The fossil seeds of Sapium sebiferum described by Yamakawa (1993) was reidentified as Euphorbia sp., and a branch of “Sequoia sp.?” was reidentified as a shoot of Glyptostrobus pensilis. In addition, fruits of Pterocarya stenoptera were reidentified as P. rhoifolia, and endocarps of Cornus microcarpa as C. controversa.
Table 4.
List of plant macrofossils obtained from the Echigawa fossil forest.
Figure 7.
Plant macrofossils from the bed of the Echi River. (Scale bars 1 mm, but absent for specimens shown at actual size). 1. Metasequoia glyptostroboides: a, ovulate cone, (LBM0112000581, EP-1); b, seed (LBM0112000466-04, EP-1). 2. Glyptostrobus pensilis: a, cone (LBM0112000467-01, EP-1); b, shoot (LBM0112000467-03, EP-1). 3. Abies sp., leaf (LBM0112000495, EP-3). 4. Tsuga sp., leaf (LBM0112000468, EP-1). 5. Chamaecyparis pisifera: a, leaf (LBM0112000470, EP-1); b, seed (LBM0112001693, EP-1). 6. Pterocarya rhoifolia, fruit (LBM0112000515, EP-5). 7. Betula grossa, bract (LBM0112001764, EP-5). 8. Betula maximowicziana, fruit (LBM0112001843, EP-5). 9. Alnus japonica, infructescence (LBM0112000582, EP-1). 10. Hemiptelea mikii, fruit (LBM0112000498, EP-3). 11. Hamamelis sp., seed (LBM0112000517, EP-5). 12. Rubus sp., endocarp (LBM0112001794, EP-7). 13. Stephanandra incisa, seed (LBM0112001691, EP-1). 14. Acer sp., fruit (LBM0112000475, EP-1). 15. Cornus controversa, endocarp (LBM0112000516, EP-5). 16. Sparganium sp., fruit (LBM0112000520, EP-5). 17. Sagittaria sp., fruit (LBM0112000542, EP-6). 18. Carex sp. A, fruit (LBM011200 0478, EP-1). 19. Scirpus sp. A, fruit (LBM0112000535-1, EP-7). 20. Polygonum sp., fruit (LBM0112000505, EP-4). 21. Menyanthes trifollata, seed (LBM0112000531, EP-7). 22. Lycopus sp., fruit (LBM0112001745, EP-4). 23. Potentilla sp. B, fruit (LBM0112001820, EP-7). 24. Euphorbia sp., seed (LBM0112000530, EP-1). 25. Physalis sp., seed (LBM0112001809, EP-7). 26. Asteraceae trib. Cardueae gen. et sp. indet., fruit (LBM0112001716, EP-2). 27. Bidens sp., fruit (LBM0112000543, EP-6).
The composition of the assemblages from EP-1 to EP-7 was similar from one to the other. Among the tree taxa, Metasequoia, Glyptostrobus, and Alnus were included in all of the assemblages. Metasequoia was represented by abundant leaves, short shoots, seed cones, and seeds in assemblages except for EP-6 and EP-7 (Table 4). Fruits and infructescences of Alnus were common in most of the assemblages. Leaves, shoots and seed cones of Glyptostrobus occurred rarely, except in EP-1. Seeds and leaves of Chamaecyparis pisifera occurred commonly except in EP-7, whereas Tsuga, Abies, and Picea were represented only by leaves in EP-1, EP-2, EP-3, and EP-5. These evergreen conifers were rare or absent in assemblages EP-4, EP-6, and EP-7. Leaves of broadleaf trees were preserved, but it was only possible to identify those of the Leguminosae. Among broadleaf trees, seeds of Magnolia, fruits of Acer, and endocarps of Rubus and Cornus were relatively common compared to all other fruits in the assemblages. The extinct species Hemiptelea mikii occurred in EP-2 and EP-3. Broadleaf trees that are mainly distributed in the cool-temperate zone such as Pterocarya rhoifolia and Betula maximowicziana were found in EP-5 and EP-7.
The herbaceous taxa comprised 16 families, 23 genera, and 37 species, and consisted mainly of aquatic and wetland plants including Sparganium, Alismataceae, Cyperaceae, Polygonum, Lycopus, and Menyanthes. Among the seven assemblages, EP-7 was rich in herbaceous taxa and included 31 species. Fruits of Scirpus, Cyperus, Carex, and Polygonum were abundant in all of the assemblages; each genus was represented by two or more taxa based on external features of the fruits. Fruits of Ranunculus were relatively common in some of the assemblages. The representative cool-temperate taxon, Menyanthes trifoliata, occurred in EP-4, but some were also identified in EP-7 (Table 4).
The fruit fossil of Pterocarya rhoifolia is the oldest record in Japan. Momohara et al. (1997) reviewed the stratigraphic occurrence of P. rhoifolia and reported that it was found only in Pleistocene sediments. Although fossil seeds of Menyanthes trifoliata occur in the late Pliocene sediments of the Yamato Group in the Aizu Basin, northern Japan (Manabe and Suzuki, 1988), the occurrence of M. trifoliata in our deposits provide the oldest report of this taxon from central and southern Japan.
Discussion
Composition and ecological characteristics of the fossil forest
The fluvial sedimentary facies indicates that the Echigawa fossil forest was part of a wetland forest mosaic. The sediments containing the fossil stumps and plant macrofossil assemblages are composed of alternating silt and sand units that were deposited in a fluvial back marsh with occasional overflow of silt and sand from the channel (Amemori et al., 1993). The plant macrofossil assemblages, mainly composed of aquatic and wetland plants such as Cyperaceae, Sparganium, and Polygonum, also indicate the dominance of wetland vegetation in and around the fossil forest (Table 4). Amemori et al. (1993) examined the sedimentary facies of the fossil stumps and showed the large stumps stood in organic silt beds. The distribution of the fossil stumps among the five horizons (HA, HA-HB, HB, HB-HC, and HC: Figure 3) indicates successive stages of forest regeneration and growth following periods of major disturbance and burial of the forest by sediment.
The organic silt beds, in which are the fossil forests of horizons HA, HB, and HC, are the paleosols in which the forest trees were rooted (Figure 3). In horizons HA, HB, and HC, Taxodiaceae and Metasequoia stumps of more than 90 cm in diameter were present. Metasequoia was represented by various and abundant plant parts (e.g., leaves, cones, and seeds) from the organic silts of EP-1 located in horizon HA and EP-4 from horizon HB (Table 4). Shoots and seed cones of Glyptostrobus are rare compared to those of Metasequoia in those horizons. Based on the relative abundance of the various isolated fossil remains of Metasequoia, the stumps of Taxodiaceae and conifers in horizons HC and HB are most probably those of Metasequoia. If this assumption is true, then Metasequoia was clearly a dominant constituent in the wetland forest in horizons HA, HB, and HC.
The proboscidean footprints found in horizons HA and HB (Okamura, 1993) show that the upper boundary of such organic silt deposits was the actual ground surface prior to burial of the fossil stumps. The largediameter trunks in those horizons indicate that the influx of sediment into these depressional areas was minimal and had little effect on tree growth given the size of the preserved stumps and stems. The largest number of annual rings counted was 329, in stump no. 28 (Kobayakawa, 1993), which further supports the idea that these forests were stable and not subject to major disturbance for long periods of time.
In contrast, the fossil stumps in the sediments of horizon HB-HC comprise many specimens of Alnus and Glyptostrobus. The trunk diameters are smaller (< 22 cm) compared to those of trees found in horizons HA, HB, and HC, except for one stump of Glyptostrobus, which was 85 cm in diameter. Woody remains of Metasequoia were not found in this horizon. The horizons of the bases of the stumps and the roots are scattered throughout the sediment in contrast to horizons HA and HB. The occurrence of alternating silt and sand sediments in HB-HC indicates that the trees were growing in a dynamic environment and often buried by sediments in recurring floods. The small diameter of the trunks indicates that the wetland forest composed of Alnus and Glyptostrobus was probably disturbed by flooding occurring over short time intervals.
Because large stumps of Metasequoia and smaller stumps of Alnus and Glyptostrobus do not occur in the same horizons (Figure 8), two types of wetland forest are recognized. They may represent different growth or successional stages of the forest and/or different habitats within a larger forested floodplain mosaic. The forest composed of Alnus and Glyptostrobus may represent a pioneer stage following disturbance by flooding. These taxa occupy unstable, flood-prone environments, while Metasequoia could only have grown to a great size and age under more stable environments.
Figure 8.
Temporal and spatial distribution of the in situ fossil in the Echi River, based on Amemori et al. data (1993). Contour interval lines show the topography of the riverbed.
The habitats requirements of extant Metasequoia, Glyptostrobus, and Alnus are currently different from one another (Lang, 1999). Glyptostrobus is distributed in waterlogged places in the flood plains of the lower reaches of large rivers in southern China and northern Vietnam (Li and Xia, 2004; Farjon, 2005). Alnus also occupies back marshes of fluvial plains in the temperate zone of the Northern Hemisphere. Both taxa tolerate anaerobic stagnant water and Glyptostrobus develops pneumatophores similar to those of Taxodium (Fu and Jin, 1992). Natural stands of Metasequoia are distributed at present in wet places in small valleys along the upper reaches of the Yangtze River (LePage et al., 2005). Some of the roots of the trees along the river are waterlogged, but the water is not stagnant (Momohara et al., 1993). The habitat of extant Alnus and Glyptostrobus in the flood plain is susceptible to floods, but that of Metasequoia does not suffer minor flood disturbances. Considering the ecological characteristics of extant species, the composition and occurrence of the fossil forest indicate that the wetland forests dominated by Glyptostrobus and Alnus and those dominated by Metasequoia existed separately during the late Pliocene. This suggests that the differences in forest composition and habitats observed in these late Pliocene ecosystems were being controlled by the prevailing hydrologic and flood disturbance regimes. The results of this study indicate that the ecological preferences of Metasequoia and Glyptostrobus were distinct during the late Pliocene. While it is difficult to determine with certainty whether such differences in habitat partitioning were prevalent in Late Mesozoic and Cenozoic fossil forests where Metasequoia and Glyptostrobus coexisted, our study demonstrates the importance of performing high-resolution stratigraphic studies where these taxa cooccur.
Paleovegetation in and around the fossil forest and its paleoclimatic implications
Based on the composition of the fossil stumps that we examined, the wetland forest was dominated by Metasequoia, Glyptostrobus, and Alnus, and included Fraxinus, Sorbus, and Rosaceae as minor elements (Table 3). The other arboreal taxa included in the plant macrofossil assemblages EP-1, EP-4, and EP-7 (i.e., the same horizons as the fossil forest beds), seem to have been more adapted to riparian habitats than to the wetland forest itself. This suggests that there may have been limited transport of the plant remains from nearby extrabasinal sources into the wetlands. Examples that support this interpretation are the evergreen conifer Chamaecyparis pisifera, Abies, Picea, and Tsuga along with deciduous broadleaf trees like Pterocarya rhoifolia, Betula maximowicziana, Magnolia, and Acer.
Palynological analysis of the sediments in and below horizons HA (ENP11 and 18) and HC (ENP16 and 17) by Mizutani (1993) indicated that Alnus is the most dominant taxon (ca. 30–40%), while Tsuga, Cryptomeria- Sequoia and Metasequoia-Glyptostrobus occupy about 10–15% of the total arboreal pollen. Pollen of broadleaf trees other than Alnus, such as Juglans-Pterocarya, Carpinus, Fagus, Quercus subgen. Lepidobalanus, and Ulmus-Zelkova, occurs at less than 5% each. Most of the pollen of Alnus, Metasequoia, and Glyptostrobus would have been derived from the wetland forest, and thus the hinterland of the fluvial back marsh evidently featured a mixed evergreen conifer and deciduous broadleaf forest.
The palynological study of Mizutani (1993) also indicated that the aquatic and wetland herbs including Cyperaceae and Polygonum dominated the vegetation in and around the fossil forest. Pollen of Persicaria and Echinocauron was abundant (ca. 3–15% of the total arboreal pollen), with Gramineae (ca. 2–5%) dominant in the herbaceous pollen assemblages. Fossil fern spores amounted to 15–20% of the total pollen and spore count. The relatively high amount of fern spores suggests the occurrence of wetland fern vegetation in and around the fossil forest, but no fossil fern fronds were found in the plant macrofossil assemblages (Table 4).
The composition of the plant macrofossil assemblages accompanying the fossil forest shows characteristics of the Pleistocene cold stages (Momohara, 1994) except for the occurrence of Glyptostrobus. Plants distributed mainly in the cool-temperate zone, such as Picea, Thuja, Chamaecyparis pisifera, Betula maximowicziana, Pterocarya rhoifolia, and Menyanthes trifoliata, were included in all of the plant macrofossil assemblages associated with the fossil forest. The plants that are currently distributed in the warm-temperate zone are not included in the fossil forest assemblage except for Glyptostrobus. This suggests that the climate during deposition of the fossil forest was under a cool-temperate regime and the change of composition of wetland forest from one dominated by Glyptostrobus and Alnus to a Metasequoia -dominated forest was not caused by climatic changes. Glyptostrobus is now distributed in the subtropics at present, but in the latest Pliocene, it grew in a cool-temperate zone climate.
Acknowledgments
We thank T. Itoh (Professor Emeritus of Kyoto University) for readily providing the fossil wood samples, M. Yamamoto (Lake Biwa Museum) for technical support to preserve the plant macrofossils, and M. J. Grygier (Lake Biwa Museum) for greatly improving the manuscript through his careful reading. We very much appreciate the help of B. A. LePage (URS Corporation), T. Saito (Meijo University), and K. Uemura (National Science Museum) for their valuable comments and advice. We also wish to thank K. Takahashi, F. Satoguchi, and S. Miyamoto (Lake Biwa Museum) for helpful advice. Finally, we are grateful for the valuable data provided by the Echigawa Petrified Forest Research Group.
