Introduction

As the need to understand the long term evolution of climate has become a major scientific challenge, Mesozoic palaeoclimates have been revealed to be of great interest. With a distribution of continental masses different from the current one, during a globally warm climatic mode, and during a now well-documented and important shift in pCO₂, Mesozoic conditions are an interesting point of comparison for understanding the relative roles of major climate determinants. Several recent papers (e.g., Gröcke et al. 2003; Galli et al. 2005; Joral et al. 2011) focused on specific events, such as the Toarcian Oceanic Anoxic Event (OAE) (Mailliot et al. 2009; Gomez and Goy 2011), in order to decipher the determinism of major Mesozoic climate shifts. Most studies, if not all, have focused on the marine realm, as data from terrestrial environments are relatively scarce.

As trees are long-lived sessile organisms, they are of interest for documenting terrestrial environments, and in particular, climate changes at various scales. Using Quaternary trees, growth-rings are often used as a climate proxy. However, such an approach is difficult to transpose to the Mesozoic (Brison et al. 2001). In the fossil record, trees are represented by avatars, such as fossilised leaves, pollen, and wood. Fossilised wood is of interest for palaeoecological syntheses, as it is not very mobile and is more resistant than other plant megafossils. However, due to their rather chaotic systematics and nomenclature, fossilised wood samples have not frequently been used for palaeoecological purposes (Philippe 1993). Recently, nomenclatural (Philippe and Bamford 2008) and taxonomic (Philippe 1994; Philippe and Hayes 2010; Philippe et al. 2013) reappraisals of fossilised wood genera have been carried out. These reappraisals now allow for a reconsideration of the fossil record for wood, in search of global distribution patterns.

The distribution of fossil wood genera has been demonstrated to be an efficient proxy for Mesozoic terrestrial climates (Tsunada and Yamazaki 1984; Philippe et al. 2004; Oh et al. 2011). Specifically, the distribution of the wood of the extinct (Late Triassic to Late Cretaceous) conifer genus Xenoxylon Gothan, 1905 was confirmed as being a potential marker of a cool and/or wet climate in the boreal hemisphere (Philippe and Thévenard 1996; Philippe et al. 2009). Inferences from the Xenoxylon distribution data recently received support when they found agreement with the results obtained from analysing oxygen isotopes in vertebrate tooth enamel from the Cretaceous in Far-East Asia (Amiot et al. 2011).

In our current study, we investigated fossilised wood specimens associated with Xenoxylon, i.e., Xenoxylon-phytocoenoses. During preliminary research, we had noticed that Xenoxylon often co-occurs with two other wood genera, Protocedroxylon Gothan, 1910 and Taxodioxylon Hartig, 1848, in Far-East Asia. The frequent association of these genera in assemblages of fossilised wood was first noted by Gothan (1910). Since then, several researchers have noticed that Protocedroxylon and Xenoxylon often co-occurred (Shilkina 1967; Shilkina and Khudayberdyev 1971; Yamazaki and Tsunada 1981). Yamazaki and Tsunada (1982: 605) hypothesised that the corresponding trees may have inhabited similar ecologies. The co-occurrence of these genera was later hypothesised as being characteristic of cool and wet temperate climates (Philippe and Hayes 2010: 61). However, no study has yet investigated the association of Xenoxylon with other wood genera.

Therefore, this study addresses the synecology of Xenoxylon, a fossil wood genus that is restricted to the Mesozoic of Eurasia, and asks the following questions: Was Xenoxylon a part of diversified vegetation? When and where did this (these) phytocoenosi(e)s appear, spread, and disappear? Does the geographic distribution of this (these) phytocoenosi( e)s provide a climatic signal?

Institutional abbreviations.—BMNH, Natural History Museum in London; BIN, Komarov Botanical Institute of the Russian Academy of Sciences in Saint Petersburg; CBNU, Chonbuk National University, South Korea; HMB, Humboldt Museum in Berlin; NCU, Prof. Nishida's collection at Chuo University in Tokyo; NHMD, Natural History Museum of Denmark in Copenhagen; SMNH, Swedish Museum of Natural History in Stockholm; TUMS, Tohoku University Museum in Sendai.

Other abbreviations.—DI, diversity index; EMJ, Early to Middle Jurassic; FAD, First Appearance Date; GD, the number of genera; LJEK, Late Jurassic to Early Cretaceous; LK, Late Cretaceous; LT, Late Triassic; N, the number of occurence; OAE, Oceanic Anoxic Event.

Fig. 1.

A. In situ erect stump of Mesozoic Xenoxylon from Bayakodan, near Kuwajima in Shiraminemura, Japan. Used with the permission of Kazuo Terada (specimen no. 53901 in Suzuki and Terada 1992). B. Typical xenoxylean radial pitting and window-like cross-field pits in radial section of Xenoxylon (slide CBNU 73112).

Historical background

Xenoxylon is one among the approximately fifty Mesozoic fossil wood genera (Philippe and Bamford 2008). Its secondary xylem has distinctive anatomical features, which have no equivalents among extant wood, i.e., radial bordered pits that are exclusively, or at least partly, flattened and contiguous (xenoxylean radial pitting) and one or rarely two window-like simple cross-field pits in the early wood (Fig. 1).

There have been several arguments related to Xenoxylon's systematic affinity. For example, Podozamites Braun, 1843, Baiera Braun, 1843, Sciadopitys Siebold and Zucc, 1842, or some extant Podocarpaceae have been suggested as its allied taxa (Nathorst 1897; Jarmolenko 1933; Arnold 1953; Bailey 1953; Shilkina and Khudayberdyev 1971). More recently, based on geochemical analysis, it appeared that Xenoxylon was closer to the Podocarpaceae, Cupressaceae sensu lato, and to a lesser extent, to the Cheirolepidiaceae than to the Pinaceae and Araucariaceae (Marynowski et al. 2008). The latter authors also suggested that Xenoxylon could be related to the Miroviaceae, which is an extinct Mesozoic conifer family. However, because Xenoxylon has never been found connected with foliage or any reproductive structures, it is difficult to make a decision about its systematic position.

Although the systematic position of this morphogenus is unclear, it is considered to be a monophyletic taxon (Marynowski et al. 2008). In addition, it is a bioindicator for a cool and/or wet climate (Philippe and Thévenard 1996; Philippe et al. 2009).

Material and methods

Time intervals and geographic subdivisions .—The Xenoxylon time range covers the Late Triassic-Late Cretaceous interval. As significant palaeobiogeographic changes occurred in the Northern Hemisphere during this time interval, we subsequently subdivided the interval into four more palaeoecologically uniform time slices: the Late Triassic (LT); the Early to Middle Jurassic (EMJ); the Late Jurassic to Early Cretaceous (LJEK); and the Late Cretaceous (LK). None of these are perfectly homogeneous, but separating the data onto more time slices would affect the significance.

Asia is here defined as the continental area east of a line running from the Ural Mountains southward to the Caucasus and then westwards to the Bosporus. Greenland is here included in Europe, as it was a part of that continent during that time. We investigated the palaeogeographic distribution of fossilised wood assemblages containing Xenoxylon specimens. For this purpose, we merged data, on the basis of geographic origin, into twenty-two super-localities (Table 1).

Data sources .—For this study, we performed a literature survey for Mesozoic fossilised wood, limiting ourselves to the assemblages that included Xenoxylon. We considered only fossilised wood data that were based on isolated secondary xylem (tracheidoxyls). Some unpublished data were also used, either established by samples sent by colleagues (see acknowledgments section), or obtained while revising collections at HMB, NHMD, BMNH, BIN, TUMS, SMNH, and NCU.

The process of extracting the data set considered in this paper was not an easy task. In order to decrease taxonomic biases, which are, unfortunately, all too common in the palaeoxylological literature, we decided to perform our analysis on the generic level. We also decided not to consider any fossilised wood assemblages that included Protocedroxylon and/or Taxodioxylon (two genera commonly associated with Xenoxylon) but not Xenoxylon, or those fossilised wood floras that were limited to Xenoxylon alone (like e.g., in Terada et al. 2011). Not considering Xenoxylon-free assemblages is a problem, as it limits the validity of co-occurrence results to the Xenoxylon distribution area. However, considering Xenoxylon-free assemblages would force us to incorporate a huge number of wood floras (a number estimated at over 300) into the data set that were far from being taxonomically and nomenclaturally uniform. Moreover, incorporating the Xenoxylon-free assemblages would be the equivalent of using negative data, which is always a perilous task in palaeontology. Finally, it must be kept in mind that fossilised wood taxa are fossil taxa; thus, two specimens belonging to the same fossil taxon did not necessarily originate from the same biological taxon. This is a strong limitation, but taking into account only those assemblages including both Xenoxylon and other genera, i.e., using a palaeoecological restriction, narrows the bias induced by form taxonomy.

Another limitation results from the incompleteness of the fossil record. Indeed, some Xenoxylon data might have been published for locations that have not been extensively sampled, or for which all wood specimens have not been determined. Little can be done about this limitation, as it is impossible to decipher a priori which assemblages may have been under-sampled.

Table 1.

List of Xenoxylon super-localities.

To avoid redundancy, data in the database are understood to represent the occurrence of a fossil genus in a geological formation, not taking into account if this genus (or a synonym) had been reported several times from that formation, in one or several localities. Stratigraphically uncertain data were not considered for our study, specifically, the reworked fossilised wood assemblages from the Mekong River terraces (Vozenin-Serra and Privé-Gill 1991), or the loose wood assemblage from the Choshi area (Nishida et al. 1993). Several Mesozoic fossilised wood data points, particularly from China, are poorly age constrained. Geological formations were sometimes renamed after the original publication, or their age was reappraised. For Chinese data, we generally followed the recent review of Zheng et al. (2008).

A wood assemblage was reported from Korea by Kim et al. (2005) as being Late Triassic in age. However, the age of the corresponding Jogyeri Formation was subsequently revised to be Middle Jurassic by Egawa and Lee (2011).

Oh (2010), in his thesis, reported some fossilised wood assemblages, which included Xenoxylon from the Chengzihe and/or the Muling formations of the Jixi Group from the Early Cretaceous in Heilongjiang Province, China. Some of these fossilised wood specimens clearly originated from the Muling Formation, but the origin of some of the others was less clear; therefore, we grouped the two formations into one.

Parrish and Spicer (1988) reported two unidentified fossil wood taxa, taxon A and taxon B, with Xenoxylon latiporosum from the Nanushuk Group, Central North Slope, Alaska. The Nanushuk Group was then considered to be from the mid-Cretaceous (Albian to Cenomanian), based on Foraminifera (Sliter 1979) and palaeobotanical evidence (Spicer and Parrish 1986). Recently, the Nanushuk Group was renamed as the Nanushuk Formation, and its age was revised to be middle Albian to Cenomanian (Mull et al. 2003). According to Mull et al. (2003: 14), most of the formation is Albian; therefore, we assigned these fossilised wood assemblages to the Albian.

Nomenclature and taxonomy.—The nomenclature here follows Philippe (1993, 1995), Philippe and Bamford (2008), and Philippe and Hayes (2010).

In 2008, Philippe and Bamford suggested that the genera with Abietineentüpfelung (i.e., rounded pits occurring on the transverse wall of ray cells) Metacedroxylon Holden, 1913, Protocedroxylon and pro parte Cedroxylon Kraus in Schimper, 1870, might be synonyms of Araucariopitys Hollick and Jeffrey, 1909. However, examination by Philippe and Hayes (2010: 59–60) of a topotype for Araucariopitys revealed that this latter name should be used only for short shoots with preserved pith and araucarioid cross-fields, while Protocedroxylon should be used for tracheidoxyls with a clearly mixed type of radial pitting and few spaced taxodioid oculipores in the cross-fields. Some Araucariopitys species have been mentioned in our database, but all can be reassigned to Protocedroxylon. Consequently, we used Protocedroxylon here as the correct name for the data originally published for the study area and interval as Cedroxylon, Metacedroxylon, Protocedroxylon, and Araucariopitys.

In the case of araucaria-like fossilised woods (i.e., tracheidoxyls with araucarian radial pitting and araucarioid cross-field pitting, lacking Abietineentüpfelung and resin canals), many different names have been used until now (e.g., Araucarioxylon Kraus, 1870, Dadoxylon Endlicher, 1847, Agathoxylon Hartig, 1848, Dammaroxylon Schultze-Motel, 1966). Recently, the wood anatomist community was polled to clarify which name should be retained for araucaria-like fossilised wood (Rößler et al. 2014). In accordance with the conclusions from this poll, we used Agathoxylon here.

Despite these preliminary taxonomic reappraisals, the discrepancies between the systematic approaches of the various authors still obscured our database. Indeed, the same wood specimen can be assigned to different genera by different authors according to their taxonomic choices, e.g., a typical Podocarpoxylon Gothan, 1906, could be assigned to Taxodioxylon or to Circoporoxylon Kräusel, 1949 (see e.g., Müller-Stoll and Schultze-Motel 1989). It was considered to be too risky to revise all of the systematic attributions in the database only on the basis of the publications, without checking the types. Thus, to limit the taxonomical bias, during a second stage of data management, we chose to merge the data into seven informal taxonomical groups (Table 2), uniting similar-appearing genera. Group A (or the Protocedroxylon group) was for wood with Abietineentüpfelung, most likely related to the Pinaceae. Wood that resembled the extant Cupressaceae sensu lato was assigned here to group B (or the Taxodioxylon group). Group C (or the Phyllocladoxylon group) was for wood specimens with unbordered (or slightly bordered) oopores, which have unclear systematic affinities. Group D (the Podocarpoxylon group) contains wood specimens with taxodioid to podocarpoid oopores in the early wood cross-fields, and these mostly also have unclear systematic positions. All of the wood specimens having araucarian radial pitting and araucarioid cross-fields, group E, limited to Agathoxylon, appear more uniform, but certainly do not accommodate only the Araucariaceae and their relatives. Group F (Brachyoxylon group) is perhaps the most heteroclite, with wood specimens featuring a mixed type of radial pitting and araucarioid cross-fields. Group G is for all of the data that are not assigned to any of the previous groups. This partitioning does not mean that we consider that the various genera within each group are taxonomical synonyms or that they originate from the same botanical group, but only that the different genera in each group were often substituted one for the other in the literature, inconsistently, for the time interval we studied. The few wood specimen data that were nomenclaturally reassigned are listed in Table 3.

Table 2.

Informal taxonomical groups of Mesozoic woods discussed in this study.

Results

Our bibliographical review and unpublished observations allowed to assemble 148 data of co-occurrence (Supplementary Online Material, SOM available at  http://app.pan.pl/app60-Oh_etal_SOM.pdf). On this basis, we prepared a table and a graph of the fossilised wood genera associated with Xenoxylon (Table 4 and Fig. 2), summarising the raw information for the entire time range for Xenoxylon, i.e., from the Late Triassic to the Late Cretaceous. Because the different nomenclatural approaches had induced an artificial scattering of data, the graphs of Fig. 2, were then redrawn to Fig. 3, using the taxonomic grouping (Table 5). The circle charts were drawn on the corresponding palaeomaps, illustrating the shares of the seven recognised taxonomic groups (A–G) in the geographically merged data (Figs. 5A, B, 6A, B). All of them have the same diameters, although they do not represent the same quantities of data. They merely reflect the relative frequencies of each taxonomic wood group at a specific super-locality.

Table 3.

The data of nomenclatural and taxonomically modified taxa in this study.

Co-occurrence of Xenoxylon with other genera .—Fig. 2 shows that Protocedroxylon is the fossil genus most commonly associated with Xenoxylon. The co-occurrence of Xenoxylon with other genera is much more infrequent (at least 50% less common). Interestingly, Agathoxylon is more commonly associated with Xenoxylon than we expected based on previous results from Asia (Oh et al. 2011).

The association of the Protocedroxylon group with Xenoxylon in the fossil strata is reported for the entire time range of the latter, from the Late Triassic to the Late Cretaceous, shortly before Xenoxylon disappeared (Yamazaki et al. 1980; Yamazaki and Tsunada 1981; Nishida and Nishida 1986; Spicer and Parrish 1990). The Protocedroxylon group is usually dominant in the Xenoxylon wood assemblages, more markedly so during the Late Triassic and Late Cretaceous. The oldest Xenoxylon-Protocedroxylon co-occurrence is reported from the Late Triassic (Carnian to Norian) of the Nariwa and Mine groups in Japan (Yamazaki et al. 1980; Yamazaki and Tsunada 1981). In the Russian Arctic, Shilkina (1967) reported Xenoxylon with Araucariopitys (in this case, a taxonomic synonym of Protocedroxylon) in the Late Triassic of the Heiss Island formation in Franz Josef Land. The youngest Xenoxylon wood assemblage with the Protocedroxylon group is dated as Campanian to Maastrichtian, from the Central North Slope, Alaska (Spicer and Parrish 1990).

Fig. 2.

Fossil wood genera co-occurring with Xenoxylon from the Late Triassic to the Late Cretaceous in Northern Hemisphere.

Table 4.

Number of fossil wood genera co-occurring with Xenoxylon from the Late Triassic to the Late Cretaceous in Northern Hemisphere. Abbreviations: LT, Late Triassic; EMJ, Early to Middle Jurassic; LJEK, Late Jurassic to Early Cretaceous; LK, Late Cretaceous; A, Asia; E, Europe; NA, North America.

The first association of Xenoxylon with Taxodioxylon is mentioned from the Liassic (Lower Jurassic) of the Toyama Prefecture, Japan (Suzuki et al. 1982). This time frame is also the First Appearance Date (FAD) for the latter genus. Group B (Taxodioxylon group) is found associated with Xenoxylon during the entire Jurassic-Cretaceous interval. This group is one of the few groups that remained associated with Xenoxylon during the Late Cretaceous (Fig. 3).

Xenoxylon is also associated at several stages during the Late Triassic-Early Cretaceous interval with the Phyllocladoxylon group (group C) (Fig. 3). It must be emphasised, however, that some of the wood specimens included in the group C appear to be poorly preserved and/or taphonomically biased Xenoxylon specimens.

In contrast, Xenoxylon is rarely associated with the Agathoxylon or Brachyoxylon groups (groups E and F) (Fig. 3). Such associations are commonly reported only from the Middle Jurassic, mostly in Europe and sometimes in the Middle East and East Asia (e.g., Gothan 1906; Holden 1913; Kim et al. 2005; Poole and Ataabadi 2005; Marynowski et al. 2008). However, the Middle Jurassic fossilised wood record in East Asia, is rather limited (Oh et al. 2011). Neither of the two genus groups (E and F) co-occurred with Xenoxylon during the Late Triassic. After the Middle Jurassic, the genera Agathoxylon and Brachyoxylon became almost exclusively disassociated from Xenoxylon in Far-East Asia and in Europe as well (Fig. 3). During the Late Cretaceous, Agathoxylon was only found associated with Xenoxylon once, in Sakhalin (Nishida and Nishida 1986).

Several other genera of wood specimens are only occasionally associated with Xenoxylon. The small number of data, as well as the broad spectrum of possible biological affinities for these fossilised wood genera (specifically, Podocarpoxylon Gothan, 1906) make interpretations difficult and risky.

Evolution of the diversity of Xenoxylon-taphocoenoses.— For the Late Triassic, the fossilised wood assemblages containing Xenoxylon have a low diversity globally (3 genera) (Fig. 4). After the Triassic, the global generic diversity of Xenoxylon-taphocoenoses increased to 27 genera for the Early to Middle Jurassic interval, as shown in Fig. 4, with a well distributed share for the different genera, although the Protocedroxylon group (A), the Podocarpoxylon group (D), and Agathoxylon (E) are co-dominant (Fig. 3). The diversity remained high (28 genera) during the Late Jurassic to Early Cretaceous interval, but the contributions from groups D, E, and F decreased, in favour of those from groups B (Taxodioxylon group) and C (Phyllocladoxylon group) (Figs. 3, 4). For the Late Cretaceous, the number of associations reported in the database is low (Fig. 4), and their generic diversity is low as well (4 genera). The Protocedroxylon group remained dominant (Fig. 3). The evolution of the generic diversity in the Xenoxylon wood assemblages from the various geographic areas is relatively similar, while Asia hosts more diversity for every stage. North America only had a Xenoxylon wood flora in its northernmost regions (Canada) and only during a short period (late Early Cretaceous) (Gordon 1932; Arnold 1953; Parrish and Spicer 1988; Selmeier and Grosser 2011).

Table 5.

Number of data for the seven taxonomical groups and for four intervals. Abbreviations: LT, Late Triassic; EMJ, Early to Middle Jurassic; LJEK, Late Jurassic to Early Cretaceous; LK, Late Cretaceous (for the composition of each taxonomical group see Table 2).

Fig. 3.

Number of data for each taxonomical wood group (for the composition of each taxonomical group see Table 2); LT, Late Triassic; EMJ, Early to Middle Jurassic; LJEK, Late Jurassic to Early Cretaceous; LK, Late Cretaceous.

Fig. 4.

Generic diversity of Xenoxylon wood-assemblages during the Mesozoic times—global curve and curve for specific geographic areas. Note that the real diversity was probably much lower, because of taxonomical bias. Abbreviations: LT, Late Triassic; EMJ, Early to Middle Jurassic; LJEK, Late Jurassic to Early Cretaceous; LK, Late Cretaceous.

The palaeobiogeographic distribution of Xenoxylon-taphocoenoses .— From the Late Triassic, only four fossilised wood assemblages containing Xenoxylon are documented, all from eastern Asia and the Arctic region (Fig. 5A). The Protocedroxylon group and the Phyllocladoxylon group are associated with Xenoxylon only in the three northernmost super-localities. The southernmost one only has wood specimens from the Phyllocladoxylon group associated with Xenoxylon. According to our data, in the Early to Middle Jurassic, the global distribution of Xenoxylon-containing fossilised wood floras expanded widely, while their diversity also increased sharply (Fig. 5B). It is worth mentioning, however, that the southernmost Xenoxylon flora in Vietnam was isolated from the main range of the genus as this region belongs to the Indochina block, which most likely shifted significantly southwards during the Cretaceous (Carter and Bristow 2003).

During the Late Jurassic and Early Cretaceous, the global distribution of Xenoxylon-containing fossilised wood floras was somewhat reduced globally and shifted slightly northward (Fig. 6A). This pattern is particularly clear in Europe, but not so marked in Asia, while, curiously, Xenoxylon extended its range in North America. For the Late Cretaceous (Fig. 6B), there are, unfortunately, few assemblages containing Xenoxylon reported in the Arctic or in Asia, all at relatively high palaeolatitudes (Nishida and Nishida 1986; Spicer and Parrish 1990). The Late Triassic and Late Cretaceous figures are strikingly similar for both the distribution and the composition of fossilised wood assemblages containing Xenoxylon.

Discussion

Limits to the interpretation of the results.—In most cases the fossil wood taxa do not represent true biological entities, what introduces some degree of uncertainty in integrating the Mesozoic fossilised wood taxa in the plant systematics. Distribution data (Figs. 5A, B, 6A, B) show that Xenoxylon had a narrower latitudinal distribution than most coeval fossilised wood genera and therefore most likely had a more restricted ecology. Limiting the area being studied to the area of Xenoxylon distribution (i.e., taking into account, for the purpose of this discussion, only the floras containing Xenoxylon) limits the systematic bias somewhat by focusing on an ecologically more coherent unit. By the Late Triassic, all significant conifer clades were present (Renner 2009), although wood from the B, C, D, and F groups are only documented with certainty from the Early Jurassic onwards. All wood groups studied are represented from the Early Jurassic to the Late Cretaceous. Other limits encountered are the geological megabiases (Benson et al. 2009), in particular the fact that the various geological units do not outcrop on equivalent surfaces, and have not been subjected to the same investigation pressures. A global approach decreases this type of bias.

We evaluated the quality of the data set by calculating a diversity index (DI) for each time interval (LT, EMJ, LJEK, and LK). For this index, we calculated DI = GD / N, where GD is the number of genera found associated with Xenoxylon for a given interval, and N is the number of data for this interval (see Table 5). The results are given in Fig. 7. For the Late Triassic, the diversity index is low. This finding, however, is congruent with the fact that fossilised wood flora are poorly diversified globally at this time, including those that do not feature Xenoxylon. With the DI decreasing slightly from the LT to the EMJ, while the GD is nine times larger, it can legitimately be hypothesised that the sampling was good for this time period. Increasing the number of data would most likely not reveal new associated genera. The figure is similar for the LJEK. For the Late Cretaceous, the DI doubles. This result might be due to either a decrease in the number of data or to an increase in the GD. Because even in well-documented Late Cretaceous fossilised wood flora containing Xenoxylon the generic diversity is always low, we interpret the increase in DI as reflecting a strong reduction in Xenoxylon distribution and thus in the amount of data. Once this was considered, we assumed that the data set in Table 4 was not strongly biased.

Fig. 5.

Distribution and diversity of the fossil wood groups associated with Xenoxylon. A. Late Triassic (modified from the DINODATA;  www.dinodata.org; accessed Oct. 11. 2011) (1, Japan; 2, Korea; 3, Southern China-Guangdong; 4, Arctic. B. Early to Middle Jurassic (1, Japan; 2, Korea; 3, Northeast China-Beijing, Liaoning, Jilin, Heilongjiang, and Inner Mongolia; 4, Central China-Henan; 5, Vietnam; 6, Uzbekistan; 7, Iran; 8, Lithuania; 9, Poland; 10, Georgia; 11, Germany; 12, UK; 13, Greenland). Abbreviations: A, Protocedroxylon-group; B, Taxodioxylon-group; C, Phyllocladoxylon-group; D, Podocarpoxylon-group; E, Agathoxylon; F, Brachyoxylon-group; G, others (for the composition of each taxonomical group see Table 2).

Was Xenoxylon part of a diversified vegetation type?— From our data, Xenoxylon is typically associated with other genera in the outcrops. Our data originate from taphocoenoses, which represent the post-mortem gathering of fossilised wood pieces. There is a little evidence from in situ fossilised wood floras that include Xenoxylon (Shimakura 1936; Ogura et al. 1951; Wang et al. 2000), with Protopiceoxylon (a type of wood from the Protocedroxylon group) being reported in Wang et al. (2000). Nevertheless, the fact that, among fossilised wood assemblages, Protocedroxylon co-occurs with Xenoxylon more than eight times more than with any other fossilised wood genus from the Late Triassic to the Late Cretaceous demonstrates that Protocedroxylon was a component of the Xenoxylon-phytocoenoses. The results for the other genera are not as conclusive. From our data, it is impossible to assert that Xenoxylon was growing locally in exclusive stands. It seems, however, that when sufficient numbers of wood specimens are analysed from a locality, Xenoxylon is usually associated with other taxa.

As far as we know, Xenoxylon has never been reported as co-occurring with angiosperm wood. There have been reports of angiosperm wood specimens from Cretaceous deposits, mostly from the Upper Cretaceous, in North America, Japan, Europe, and Antarctica (e.g., Wheeler et al. 1987, 1994; Meijer 2000; Poole and Francis 2000; Takahashi and Suzuki 2003). These are, however, mostly from low- to mid-latitudes, while during this time span, the Xenoxylon range shifted towards high latitudes.

Fig. 6.

Distribution and diversity of the fossil wood groups associated with Xenoxylon. A. Late Jurassic to Early Cretaceous (1, Japan; 2, Northeast China-Beijing, Liaoning, Jilin, Heilongjiang, and Inner Mongolia; 3, Southeast Mongolia; 4, Western China-Xinjiang; 5, Primorye; 6, Arctic; 7, Greenland; 8, Canada-Alberta; 9, Canadian Arctic Archipelago; 10, Alaska-Central North Slope). B. Late Cretaceous (1, Sakhalin; 2, Alaska-Central North Slope). Abbreviations: A, Protocedroxylon-group; B, Taxodioxylon-group; C, Phyllocladoxylon-group; D, Podocarpoxylon-group; E, Agathoxylon; F, Brachyoxylon-group; G, others (for the composition of each taxonomical group see Table 2).

The global diversity of Xenoxylon floras, as judged from wood assemblages, was low during the Late Triassic but increased sharply at the Triassic-Jurassic boundary. This is in accordance with a global southward expansion of Xenoxylon during the Early Jurassic (Philippe and Thévenard 1996), as well as with a simultaneous strong diversification of wood types (Philippe and Harland 2007). During the Late Jurassic to Early Cretaceous interval, the diversity of the Xenoxylon flora is similar to that during the Early to Middle Jurassic, although it decreased slightly in the southern part of the range. By the Late Cretaceous, the diversity of Xenoxylon-floras decreased and their distribution was reduced.

Fig. 7.

Diversity index (DI = GD/N, where GD is the number of genera associated with Xenoxylon, and N the number of occurrences) for each time period. Abbreviations: LT, Late Triassic; EMJ, Early to Middle Jurassic; LJEK, Late Jurassic to Early Cretaceous; LK, Late Cretaceous.

Xenoxylon-phytocoenoses and palaeoclimates .—The Pangaean supercontinent became fragmented during the Triassic, while the climate remained warm and dry globally. It is noteworthy that Xenoxylon-phytocoenoses appeared worldwide during the Carnian, with Protocedroxylon as an associated element, during a time that might correspond to a cooling and pluvial event (Simms and Ruffell 1989). Continents became increasingly separated during the Early Jurassic, while concomitantly, the world climate became more equable and more humid than during the Triassic (Chandler et al. 1992; Golonka and Ford 2000; Ruckwied and Götz 2009). Such a globally more humid climatic environment was most likely the primary cause of the spread and the increase in the diversity of the Xenoxylon-phytocoenoses.

This diversity increased slightly during the Late Jurassic to Early Cretaceous interval, except in northern East Asia. The Cretaceous is regarded as having been globally warmer than the Jurassic (Spicer et al. 1993; Ziegler et al. 1993 ), although significant cooling events most likely occurred during the Early Cretaceous (Pucéat et al. 2003; Maurer et al. 2012). In Europe, specifically after the Middle Oxfordian, the climate became more arid (Dromart et al. 2003), while the range of Xenoxylon decreased northward (Philippe and Thévenard 1996). However, in Far-East Asia, this distribution restriction did not occur for Xenoxylon, and the associated phytocoenoses remained rich and diverse, especially in northeast China (Liaoning, Jilin, Heilongjiang, and Inner Mongolia provinces). Recent geochemical results (Amiot et al. 2011) evidenced that cool climates prevailed in northern Far-East Asia, at least during the Early Cretaceous (Barremian-early Albian). The prevalence of a cool environment during the Early Cretaceous most likely determined the persistence of diverse Xenoxylon-phytocoenoses in northern Far-East Asia.

The globally warm Late Cretaceous saw a concomitant severe decrease in the distribution of Xenoxylon and a sharp decrease in the diversity of Xenoxylon-phytocoenoses, confirming a relationship, for this genus, between its distribution and associated diversity. The most recent data for Xenoxylon-phytocoenoses, and for the genus, are from northern Far-East Asia and Alaska, supporting the idea that this area remained cool and/or humid throughout the Cretaceous, although Maastrichtian data from fossilised leaves indicate a significant local warming slightly to the south (Hermann and Spicer 1996).

Two genera (Brachyoxylon and Agathoxylon), found episodically associated with Xenoxylon, have special palaeoclimatic significance. Brachyoxylon was found in southwestern Europe, during the Middle to Late Jurassic interval, to be bound mostly to the tropophilous climates of tropical lowlands and perireefal systems (Garcia et al. 1998), while during the Early Jurassic, it had a more diverse palaeoecology (Philippe 1995). Brachyoxylon is only rarely associated with Xenoxylon after the Early Jurassic, except in Greenland, an exception that might be due to a taxonomic bias. Indeed, Brachyoxylon is related to diverse types of plants, and the Greenland associations might be caused by the local persistence of the more hygrophilous Early Jurassic Brachyoxylon plants. Agathoxylon, most likely a “bin” taxon for several botanical groups (pteridosperms, caytoniales, possibly also some cycads and Bennettitales) does not present a clear palaeoecological pattern. In Eurasia, it was, however, distributed mostly in the south throughout the Middle Jurassic-Late Cretaceous interval. After the Middle Jurassic, in Asia, Agathoxylon and Xenoxylon appear to be exclusive (Oh et al. 2011), whereas this is not the case in Europe. However, Agathoxylon might include taxa there that were not represented in Asia and that inhabit a different ecology.

The results and discussions confirm that Xenoxylon was bound to “northern climates”, i.e., cooler and/or humid ones, and that it was associated with a more diverse flora in the southernmost (i.e., warmest) parts of its distribution. This result suggests that the occurrence of Xenoxylon at lower latitudes is bound to more humid climatic episodes, which were not necessarily cooler.

Conclusion

From its appearance in the Late Triassic to its disappearance at the K–T boundary, Xenoxylon most likely lived in association with other genera. During the globally most arid/warmest intervals (Late Triassic and Late Cretaceous), Xenoxylon-phytocoenoses had a limited diversity, with Protocedroxylon being the primary associated genus, and a northern distribution. During the globally more humid/cooler times, and especially during the Early Jurassic, Xenoxylon-phytocoenoses are encountered in significantly more southern regions and are more diverse, often including Taxodioxylon, particularly during the Early Cretaceous. The distribution patterns for Xenoxylon-phytocoenoses in Europe, North America and Asia differ sharply. Our results suggest that within Xenoxylon's range, corresponding phytocoenoses were differentiated on a latitudinal gradient according to the climate change patterns during the Mesozoic, and the occurrence of diverse Xenoxylon-phytocoenoses at mid- to low palaeolatitudes characterise more humid episodes.