• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Macroecological Responses of Terrestrial Vegetation to Climatic and Atmospheric Change Across the Triassic/Jurassic Boundary in East Greenland

Posted on: Thursday, 8 November 2007, 06:00 CST

By McElwain, Jennifer C Popa, Mihai E; Hesselbo, Stephen P; Haworth, Matthew; Surlyk, Finn

Abstract.- The magnitude and pace of terrestrial plant extinction and macroecological change associated with the Triassic/Jurassic (Tr/ J) mass extinction boundary have not been quantified using paleoecological data. However, tracking the diversity and ecology of primary producers provides an ideal surrogate with which to explore patterns of ecosystem stability, collapse, and recovery and to explicitly test for gradual versus catastrophic causal mechanisms of extinction.

We present an analysis of the vegetation dynamics in the Jameson Land Basin, East Greenland, spanning the Tr/J extinction event, from a census collected paleoecological data set of 4303 fossil leaf specimens, in an attempt to better constrain our understanding of the causes and consequences of the fourth greatest extinction event in earth history. Our analyses reveal (1) regional turnover of ecological dominants between Triassic and Jurassic plant communities, (2) marked structural changes in the vegetation as reflected by potential loss of a mid-canopy habit, and (3) decline in generic-level richness and evenness and change in ecological composition prior to the Tr/J boundary; all of these findings argue against a single catastrophic causal mechanism, such as a meteorite impact for Tr/J extinctions. We identify various key ecological and biological traits that increased extinction risk at the Tr/J boundary and corroborate predictions of meta-population theory or plant ecophysiological models. These include ecological rarity, complex reproductive biology, and large leaf size.

Recovery in terms of generic-level richness was quite rapid following Tr/J extinctions; however, species-level turnover in earliest Jurassic plant communities remained an order of magnitude higher than observed for the Triassic. We hypothesize, on the basis of evidence for geographically extensive macrofossil and palynological turnover across the entire Jameson Land Basin, that the nature and magnitude of paleoecological changes recorded in this study reflect wider vegetation change across the whole region. How exactly these changes in dominance patterns of plant primary production affected the entire ecosystem remains an important avenue of future research.

Introduction

The nature, causes, and consequences of the Triassic/Jurassic (Tr/ J) mass extinction event (200 Myr ago) have received increasing attention over the past decade. Sepkoski's (1981) original global compendium of marine faunal extinction rates classified the Tr/J boundary as one of the "big five" extinction events in earth history. Estimates of the magnitude of diversity loss across the boundary vary from group to group and are dependent on the scale of study. Stage-level compilations for marine families and genera (Sepkoski 1981, 1993) indicate losses of 23% and 50% respectively. At the regional or single locality level, extinction magnitudes of species are extremely high: >80% of terrestrial plant species in Greenland and Sweden (Harris 1937; McElwain et al. 1999) and 42% of terrestrial vertebrates families in North America (Olsen et al. 1987), as well as widespread extinction of ammonites (Newell 1963; Tozer 1979), bivalves (McRoberts and Newton 1995), radiolarians (Tipper et al. 1994; Carter and Hori 2005), and coral reefs (Kiessling 2001, 2005). In contrast, other authors have questioned whether the boundary can be characterized unequivocally as a mass extinction. Hallam (2002) has argued that the tempo of extinction was gradual rather than catastrophically rapid, whereas others (Tanner et al. 2004) suggest that most of the apparent biodiversity losses across the Tr/J boundary are due to biases or artifacts of sampling or poor stratigraphie control. Reanalysis of Sepkoski's global marine faunal data has also demonstrated that low origination rates were more responsible than high extinction rates for Tr/J biodiversity loss (Bambach et al. 2004).

Uncertainties regarding the nature and tempo of the Tr/J boundary extinctions are further complicated by the fact that most biotic records spanning the Rhaetian-Hettangian interval are based on presence-absence data, which can artificially indicate a catastrophic event if taxonomic groups are "oversplit" or bias interpretation in favor of a gradual event owing to taphonomic control of last occurrences (Signer III and Lipps 1982). Presence- absence data sets also limit our understanding of the ecological and physiological selectivity of extinctions. For instance, is there evidence for gradual ecosystem decline or instability prior to the "extinction event" at the Tr/J boundary? Does the extinction event represent a biotic threshold response to long-term gradual forcing or was it truly catastrophic? Were the taxa that suffered extinction ecologically important dominants within ecosystems or rare? In the absence of paleoecological data these critical questions about the responses of terrestrial plant communities across the Tr/J boundary remain unanswered. Without paleoecological data it is also difficult to decipher potential forcing mechanisms of biotic turnover, many of which make explicit predictions about the nature and tempo of macroecological response. To address these uncertainties we have undertaken a detailed paleoecological study of terrestrial plant communities through Rhaetian and Hettangian strata of the Kap Stewart Group in Jameson Land, East Greenland. From a database of 4304 census-collected macrofossil plant specimens we have investigated the tempo of macroecological change across the Tr/J boundary and tracked stability, collapse, and recovery of plant communities, the primary production and therefore foundation of terrestrial ecosystems in this region. We define census-collected as follows: where every fossil specimen discovered within a specific fossil plant bed was collected within a standardized time frame for all plant beds.

We have conducted this paleoecological analysis in the context of a dramatically changing global environment throughout the Triassic- Jurassic interval, with peak environmental changes coinciding with the Tr/J boundary, as reflected by a negative carbon isotopic excursion in both organic and inorganic carbon (Palfy et al. 2001; Ward et al. 2001; Hesselbo et al. 2002; Guex et al. 2004; Galli et al. 2005). We have used relative abundance data and two measures of biodiversity (evenness and richness) collected from nine fossil plant beds to assess whether there was a gradual decline in ecosystem functioning prior to the Tr/J boundary, or whether an observed >80% species-level extinction at the boundary (Harris 1937; McElwain et al. 1999) represents a geologically instantaneous event. We have also investigated compositional changes (including the degree of heterogeneity) in terrestrial vegetation and the ecological and physiological selectivity of the Tr/J event in an attempt to understand better the causal mechanisms of Tr/J boundary floral turnover.

Abiotic Context for Tr/J Extinctions

Causal mechanisms for Tr/J extinctions remain unresolved. Suggested forcing factors for Tr/J biodiversity loss include climatic and environmental disturbance due to a catastrophic meteorite impact (CAMP) (Olsen et al. 2002), gradual climatic and environmental change associated with emplacement of the Central Atlantic Magmatic Province (Marzoli et al. 1999, 2004; McElwain et al. 1999; Hesselbo et al. 2002), catastrophic release of methane due to methane hydrate destabilization resulting in global warming (Palfy et al. 2001; Beerling and Berner 2002), and sea-level change involving a fall followed rapidly by a rise (Hallam 1997). Global cooling has also been invoked as a potential extinction mechanism (Hubbard and Boulter 2000) although supporting evidence for such a mechanism is open to serious challenge.

FIGURE 1. Stable carbon isotopic (delta^sup 13^C) record from fossil wood (Hesselbo et al. 2002) (A), cuticle (McElwain et al. 1999) (B), and a paleoatmospheric carbon dioxide record (McElwain et al. 1999) (C), plotted against height of the section at Astartekloft (cf. Fig. 3) across the Triassic (Rhaetian)/Jurassic (Hettangian) boundary. The carbon isotopic data in A are all derived from fossil wood collected at Astartekleft. The cuticle material in B is derived from Astartekloft and other sections throughout the Jameson Land Basin (see McElwain et al. 1999 for details).

Irrespective of the exact causal mechanism, stable carbon isotopic records have identified a pronounced double negative excursion of 2-3[per thousand] coincident with the boundary in East Greenland (McElwain et al. 1999; Hesselbo et al. 2002), Hungary (Palfy et al. 2001), England (Hesselbo et al. 2002), the United States (Guex et al. 2004; Ward et al. 2006), Italy (Galli et al. 2005, 2006), and Canada (Williford et al. 2006). These records clearly document a major perturbation of the carbon cycle, which was global in nature, and they suggest major environmental upheaval in the latest Rhaetian with maximum environmental change coinciding with the Tr/J boundary. Hesselbo et al. (2002) interpreted the single carbon isotope excursion apparent in the Jameson Land succession as an amalgamation of an "initial" and a "main" isotope excursion observed in expanded marine sections. In this paper we explore in detail the terrestrial plant communities in the region of Astartekloft, East Greenland, in the context of the stable carbon isotopic composition from fossil wood in the same section (McElwain et al. 1999; Hesselbo et al. 2002) (Fig. 1A,B). Although it is difficult to infer directly from isotopic records how the climate or atmospheric composition was changing we interpret the isotopic profile to provide an integrated record of environmental upheaval across the interval. We therefore interpret any significant global deviation in delta^sup 13^C (>2[per thousand] from average background levels) as a perturbation in the carbon cycle associated with a change in environmental conditions. We also investigate vegetation dynamics in the context of a likely fourfold increase in paleoatmospheric CO2 across the Tr/J boundary from levels that were approximately three times higher than present before the boundary, derived using the stomatal proxy approach (McElwain et al. 1999), and a 16[degrees]C regional climatic warming, inferred from a coupled ocean atmosphere general circulation model forced with this CO2 reconstruction (Huynh and Poulsen 2005). However, an estimated regional warming of 16[degrees]C is remarkably high and likely to be overestimated, as their model does not take into account the possible effect of higher SO2 levels due to CAMP volcanism (which would cool climates) on global or regional climate warming. Furthermore, stomatal density-based CO2 records are required from additional genera to Ginkgoales, following recent procedures for other time intervals (i.e., McElwain et al. 2005) to test the fidelity of the existing Ginkgo-based record. Paleo-CO2 records based on multiple independently calibrated genera with temporally overlapping stratigraphie ranges are more robust, as they minimize calibration uncertainty that can be introduced due to genetic variability in stomatal frequency within and between genera, such as that demonstrated by (Cantor et al. 2006).

Material

Kap Stewart Flora.-In a seminal series of monographic papers published between 1926 and 1937 Tom Harris documented a Rhaetian- Hettangian aged fossil flora of over 200 species from approximately 13 localities in Jameson Land, East Greenland (Fig. 2). This fossil flora is represented by a rich assemblage of bryophytes, pteridophytes and gymnosperms (Harris 1926, 1931, 1932a,b, 1935, 1937). Two distinct plant macrofossil biozones are recognized within the Kap Stewart Group: a Rhaetian assemblage zone characterized by the presence of Lepidopteris ottonis and a Hettangian-Sinemurian assemblage zone with Thaumatopteris brauniana (Harris 1937). A >80% species-level turnover of plant macrofossils occurs between the highest occurrences of L. ottonis zone taxa and the lowest occurrences of T. brauniana zone taxa (Harris 1937). This floral "transitional zone" was used by Harris to define the Tr/J boundary. Subsequent fieldwork (2000-2004) has shown that the first appearance of Thaumatopteris brauniana zone elements occur contemporaneously with the last appearances of Lepidopteris zone taxa in Bed 5 of our current study locality, Astartekloft. This level coincides with the most negative carbon isotopic values in the stable carbon isotopic record from the same section (Fig. 3) (Hesselbo et al. 2002). Palynological studies of the Kap Stewart Group have defined two microfloral zones that parallel the macrofloral biozones of Harris and correlate with established Rhaetian aged Rhaetipollis- Limbosporites zone and Hettangian aged Pinuspollenites- Trachysporites zones of Europe, Canada, and Svalbard (Pedersen and Lund 1980). For these reasons, and in the absence of a global boundary stratotype section and point (GSSP) for the Tr/J boundary, we interpret the top of Bed 5 as representing the Tr/J boundary at Astartekloft.

FIGURE 2. Map of the Jameson Land Basin showing locations of the fossil plant localities and position of paleolake.

Geological and Depositional Setting.-The Rhaetian-Hettangian aged Kap Stewart Group in Jameson Land was deposited in environments that ranged from fluvial to lacustrine (Dam and Surlyk 1993b) (Table 1). The group comprises three formations (Surlyk 2003): the predominantly conglomeratic and sandy Innakajik Formation at the base, deposited in coarse-grained alluvial plain environments; a mixed sandy and shaley Primulaelv Formation in the middle, deposited in a delta-plain setting; and a mixed sandy and shaley Rhaetelv Formation at the top, deposited in a lacustrine setting. The plant fossils collected for this study come exclusively from the Primulaelv Formation at Astartekloft (Figs. 2, 3). The succession at Astartekl0ft comprises mainly sandstone (~75-80%), of fine to coarse grade and commonly texturally and compositionally immature (Dam and Surlyk 1993b). Shaley units, which constitute ~20-25% of the succession, are made up of pale blocky mudstones ranging in thickness up to a maximum of ~4 m, and dark-gray leaf-rich shale in beds <~1 m. Poorly developed black coal and coaly mudstone occurs about the middle of the formation. Rootlets are present at the base of the coal, and at several other horizons in the mudstone facies. Thin beds of siderite occur in the more organic-rich shaley layers. Most of the mudstone units are traceable across the exposure at individual localities and evidently have lateral extents on the order of tens to hundreds of meters. Mudstone beds also occur within channelized sandstone bodies. Thin sand laminae and wave-ripple structures have not been observed in the mudstone facies at Astartekleft.

The thick sandstones were laid down from river channels that were commonly several meters deep (Sykes 1974; Dam and Surlyk 1993a,b). The textural and mineralogical immaturity of the sand indicates a local source terrain that was undergoing strong physical erosion; in the case of the Hurry Inlet (east Jameson Land) localities this was likely to have been the Caledonian basement that is currently exposed in Liverpool Land to the east (Fig. 2). Neither discrete erosional channel margins nor mud-plugged abandoned channels are commonly observed, indicating rapid and continuous lateral migration of channels during deposition. The thin coarsening-upward sandstone sequences that form "wings" to adjacent channel fills represent sheet splay or crevasse splay deposits formed during river flood episodes (Dam and Surlyk 1993b). These are the primary facies in which plant fossils occur at Astartekloft (sheet splay in Table 1, SS in Fig. 3). A secondary setting for the plant fossils is in sediments deposited at upward transitions from channel sand facies into floodplain facies (abandoned channel in Table 1, AC in Fig. 3). These are interpreted to have been shallow pools developed in semi- abandoned sand-filled channels.

The predominant gray color of the mudstone units, combined with moderate lateral extent and sporadically rooted nature, indicates deposition on a permanently or semi-permanently waterlogged floodplain or in an interdistributary lacustrine bay. The difference between floodplain and interdistributary lacustrine bay environments is one of degree of connection to open lake conditions. In our study, in the Hurry Inlet area, interpretation of these facies as floodplain is preferred because sedimentary structures indicative of open waters (e.g., wave ripples) are absent, palynofacies are exclusively of terrestrial origin (Koppelhus 1997), and there is no observed interdigitation of alluvial or delta plain with lacustrine facies. The true coal (coal swamp [CS]), a third category of plant bed in Jameson Land Basin, possibly only occurs at one stratigraphic level (equivalent with the level of Bed 6 at Astartekloft; Table 1, Fig. 3) but is not fully developed at Astartekloft. Only rarely did the floodplain areas dry out enough for oxidizing conditions to become established, as shown by the volumetrically subordinate purple-gray mudstone. The abundance of greenclay cementation of the tops of sandstone beds indicates the common development of anaerobic early diagenetic conditions beneath the waterlogged floodplain.

The floras preserved in these three plant-bed settings are likely to be of somewhat different origins. Floodplain deposits and coal swamps are likely to represent mainly autochthonous assemblages, containing mostly in situ representatives of the floodplain and swamp communities, respectively (Gastaldo 1989). Sheet or crevasse splays contain autochthonous assemblages from local wetland and floodplain communities, combined with allochthonous plant material derived from channel levees and possibly from plant communities further upstream (Gastaldo 1989; Spicer 1989). These types of deposits are thought to provide some of the most accurate representations of the overall regional source vegetation despite taphonomic biases that tend to overestimate upper delta or alluvial plain communities and underrepresent those of the local floodplain (Ferguson 1985; Gastaldo 1989; Spicer 1989). The abandoned channels, which according to sedimentological analysis were formed by avulsion, would be expected to preserve predominantly parautochthonous plant communities growing in close proximity to the channel, with only very rare occurrences of allochthonous components transported from upstream (Behrensmeyer et al. 2000). Fossil plant assemblages in these deposits would therefore most likely provide insights only into plant communities in the immediate region of Astartekloft.

FIGURE 3. Measured section from Astartekloft, East Greenland from Hesselbo et al. (2002), updated based on further fieldwork in 2002 and 2004. Biostratigraphy and lithostratigraphy from Harris (1937), Pedersen and Lund (1980), Dam and Surlyk (1993a,b) and Surlyk (2003). The fossil plant beds located in the present study are shown in boldface; Beds 0, 5.5, and 9 and the "moss bed" have not yet been investigated in detail. Harris's (1937) plant bed names are also shown, together with his barometer-based estimated distance below the Jamesoni Horizon (Harris scale). We re-located Harris's beds on the basis of position in the section and on the contained macrofloras. Interpreted depositional environments are indicated in italics. SS, sheet splay; CS, coal swamp; AC, abandoned channel. TABLE 1. Fossil plant collection statistics from 2002 field season.

Taphonomic Considerations

Taphonomic studies have shown that untransported and undecayed litter from temperate and relatively low diversity subtropical floodplain forests provides an accurate indication of both the richness and the dominance-diversity relationship of the source forest (Burnham 1989; Burnham et al. 1992). It is noteworthy that the majority of these live-dead studies have been carried out on angiosperm dominated plant communities, with only one detailed study to date on a forest where a gymnosperm was co-dominant (Taxodium- Acer swamp [Burnham et al. 1992]). Nonetheless, observations in different ecological settings and hemispheres have revealed that representation of live species in leaf litter from both angiosperm- and gymnosperm-dominated (Taxodium) forests is predominantly controlled by the same factors (e.g., relative abundance, canopy height, leaf area, distance from tree) (Burnham et al. 1992; Steart et al. 2005). In the absence of a live-dead study of broad-leaved conifer dominated vegetation, we have therefore assumed that the same primary factors controlled preservation of leaf litter in the Triassic and Jurassic. General biases that should be noted include slight overrepresentation of taxa with a woody vine (liana) habit (Burnham et al. 1989), and better representation of larger canopy trees than smaller sub-canopy taxa (Gastaldo 2001). Biases due to decay and transport of leaf litter can also be considerable (Ferguson 1985; Spicer 1989).

Despite these caveats, the macrofossil assemblages from Astartekloft are exquisitely preserved and abundant, primarily autochthonous or parautochthonous and to a lesser extent allochthonous, and show little evidence for decay or long-distance transport. Furthermore all fossil plant beds were census-collected to avoid collector bias, and lateral sampling of each bed was as wide as the nature of exposure would allow, so as to minimize biasing of the rank-abundance data by very localized input of leaves from individual plants. Every bed was therefore laterally sampled across a minimum of eight to a maximum of ~20 meters. Six out of a total of nine fossil plant beds are isotaphonomic, including all five of the Triassic beds and the Tr/J bed (Table 1, Fig. 3). Isotaphonomic assemblages afford the least-biased means of tracking biodiversity trends through time (Behrensmeyer et al. 2000; Gastaldo 2001). The large database of fossil leaf counts from macrofossil assemblages at Astartekloft therefore provides an ideal study system with which to investigate paleoecological changes in Triassic and Jurassic vegetation. In the case of the six isotaphonomic plant beds, we can explicitly examine paleoecological changes that occurred leading up to, and coincident with, the Tr/J extinction boundary bed, and test gradual versus catastrophic mechanisms of floral extinction. However, because the depositional setting of the Jurassic plant beds at Astartekloft (two abandoned channels and one swamp) are different from those of the Triassic (all sheet splay), we use preliminary fossil plant occurrence data from a Jurassic abandoned channel assemblage at a second Jameson Land field locality (South Tancrediacloft [McElwain, I. Glasspool, Popa, Hesselbo, D. Sunderlin, and Surlyk unpublished data]) to distinguish the potential effects of taphonomy from larger-scale paleoecological patterns across the Tr/J boundary.

Collection Methods

Macrofossil specimens were collected in 2002 using census style techniques from a total of nine fossil plant beds at Astartekloft, Jameson Land, East Greenland (Table 1, Fig. 3). Most these fossil plant beds were originally discovered by Harris and were relocated by measuring distance in meters below a pectenrich carbonate- cemented sandstone (Jamesoni Horizon) of the Neill Klinter Group (Fig. 3). Harris used this as his marker bed throughout Jameson Land localities to correlate macrofloras across the basin (Harris 1937: p. 71). Collection procedures were standardized for all nine fossil plant beds. Each bed was collected for a total of 48 hours by excavating four small quarries, spaced two to five meters apart laterally, as constrained by the nature of the exposure. Roughly the same amount of bulk sediment was excavated per bed with the exception of Beds 1.5 and 6, which were only census-collected for a total of eight hours each. All macrofossil specimens excavated during the collection interval were collected irrespective of preservation state and are currently stored in the paleobotanical collections at the Field Museum, Chicago. Every identifiable macrofossil specimen was identified to genus in the laboratory; specimens of each genus were counted and recorded in a fossil leaf occurrence matrix (Table 2). We use the term "macrofossil specimen" to refer to any identifiable vegetative or reproductive structure preserved on a rock slab/specimen. Individual slabs commonly preserved multiple fossil leaves and reproductive structures from one or many fossil plant taxa (Table 1). To estimate the relative abundances of Triassic and Jurassic genera at Astartekloft we used a sampling strategy that represents a hybrid method between the widely used "quadrat" method of counting (Pfefferkorn et al. 1975), which reports the occurrence of a taxon once per quadrat irrespective of how many leaves there were per quadrat, and those methods that count every individual leaf (Wing et al. 1993). Quadrat methods can inflate the relative abundances of ecologically rare taxa and underestimate the abundances of dominants (see Wing and DiMichele 1995 for a full discussion), whereas individual counting methods are extremely time consuming and it can be difficult to resolve when, and if, leaf fragments should be counted. We have counted every individual leaf or leaf fragment per rock slab as one occurrence in our abundance matrix (Table 2). However, in cases where there were more than five leaves or five leaf fragments per rock slab we recorded a count of five rather than counting every individual leaf. We recognize that this counting strategy may underestimate the relative abundances of the ecological dominants; however, we feel that this is preferable to the loss of potentially important ecological data caused by the inflation of abundances of rare taxa. Furthermore, a sensitivity analysis comparing our counting strategy for Bed 1 at Astartekloft with a standard quadrat style counting approach has revealed only subtle differences in the resultant relative and rank-order abundances of fossil leaf genera and insignificant differences in standard measures of biodiversity based on the relative occurrence data (Appendix 1). Census collection of the Kap Stewart Group at Astartekloft resulted in a data set of 4303 recorded occurrences of macrofossil leaf and reproductive specimens, derived from a total of 40 Rhaetian and Hettangian plant genera.

TABLE 2. Generic abundance (number of occurrences) of macrofossil leaf and reproductive specimens at Astartekloft, East Greenland.

Paleoecological Analysis

Paleoecological analyses have all been carried out at the genus level. An analysis of Harris's original Tr/J macrofossil data (Harris 1937) indicates that 95% of generic occurrences, from a total of 38 different beds, are monospecific. Those genera containing more than one species occur predominantly in Triassic (~9%) rather than Jurassic plant beds (~0.5%) and are typically represented by two (e.g., Ctenis, Equisetites) and rarely three to four species (e.g., Pterophyllum) in a single assemblage. From these data we infer that a generic-level analysis is a fairly robust proxy for analysis at the species level (which will be the subject of future work) but that generic-level richness may underestimate species richness in the Triassic more than in the Jurassic. We note that a generic-level paleoecological analysis may also mask interesting species dynamics between stratigraphie levels. However, these cannot be resolved until we have a firmer understanding of the real taxonomic value of leaf surface micromorphological features (e.g., stomatal density and distribution, trichome abundance, presence and absence) on which species determinations for the Kap Stewart Group flora were largely based (Harris, 1926, 1931, 1932a,b, 1937). It is now known than many of these micromorphological traits are highly plastic in response to changes in the climatic and atmospheric environment. A number of species as determined by Harris could therefore represent a continuum of ecophenotypes responding to environmental changes associated with CAMP volcanic activity, rather than biological species.

Composition.-Small changes in the ecological composition of plant communities can greatly alter whole ecosystem processes, biotic interactions, and feedbacks with the climatic system. Significant functional shifts in ecosystem processes can result from changes in the relative dominance of different taxa without any significant change in taxonomic composition. To track changes in generic-level floral composition and dominance patterns across the Tr/J boundary, we calculated relative generic abundances per bed for all 41 genera as a percentage of the total bed abundance using the occurrence matrix (Table 2).

Detrended correspondence analysis (DCA) was carried out on the resultant relative abundance matrix (Table 3) using PAST(c) Version 1.33 (Hammer et al. 2001). This eigenvector technique was developed for detecting floristic gradients in modern botanical data sets and displaying complex compositional differences between sites in two dimensions (Hill and Gauch 1980). Using this technique, "sites" or "plant beds" that are compositionally similar, plot close together in ordination space, whereas those which are compositionally distinct plot far apart. In the case of this study we display compositional change using the first eigenvector (Axis 1) only. To minimize potential biases introduced by disarticulated reproductive structures of unknown phylogenetic affinity, we restricted our investigations of compositional changes to vegetative macrofossils only. DCA was also repeated with the inclusion and exclusion of genera occurring in only one bed to test for any distortion effects that these may introduce in the analyses (Wilf and Johnson 2004). Biodiversity.-Biodiversity is variously defined by different authors. However, we define biodiversity here in terms of both richness (the number of taxa) and evenness (the equality of relative abundances among taxa) of the paleovegetation. The more diverse or rich an ecosystem the more likely that it will contain functional types that are highly productive (Loreau et al. 2001). Although there are many exceptions to the rule, richer ecosystems tend to be more productive as long as climatic and edaphic variables are favorable. Also the richer an ecosystem, the more likely it is to withstand major abiotic changes, because there is a larger species pool from which new dominant taxa or new productive functional types can be recruited.

Evenness is a measure of niche partitioning and facilitation within an ecosystem (Loreau et al. 2001). Ecosystems that have many codominant species all contributing to ecosystem function, such as in modern tropical lowland forests, have a high degree of evenness and a higher diversity of phenotypic traits. They are therefore more likely to be resilient to environmental changes or to exceeding critical thresholds (Loreau et al. 2001). These generalized ecological relationships between biodiversity and stability have been demonstrated for both short-term (Loreau et al. 2003) and geological time scales (Kiessling 2005). For the current study we use generic richness, evenness, and composition of plant communities as proxies for overall ecosystem "persistence stability," which we define as the ability to withstand abiotic perturbations as indicated by changing isotopic composition.

Richness.-Trends in generic richness across the Tr/J boundary were investigated by standardizing absolute generic richness for all nine fossil plant beds using rarefaction analysis. Analytical Rarefaction 1.3 by S. M. Holland was used to estimate rarefied leaf generic richness and 95% confidence intervals for seven of the nine fossil plant beds, all of which had more than 220 specimens (Table 1). In the case of Beds 1.5 and 6, which had less than 220 specimens (62 and 128, respectively), their respective richness values were forward-projected to 220 specimens by fitting the most statistically appropriate curve to their respective rarefaction curves: logarithmic in the case of Bed 1.5 (R^sup 2^ = 0.99) and power in the case of Bed 6 (R^sup 2^ = 0.99). Generic richness reported for these two plant beds should therefore be considered preliminary. Further collecting will be undertaken to test these preliminary richness estimates.

TABLE 3. Precentage representation of leaf genera at Astartekloft from generic abundance matrix (Table 2).

Taphonomic studies on modern leaf litter demonstrate that multiple samples (four) of temperate leaf litter capture on average 85% of the standing species richness of the source vegetation, whereas subtropical and tropical leaf litter typically underestimates standing richness by four and eight times, respectively, owing to high spatial heterogeneity (Burnham 1993). We have addressed this important potential bias by estimating within- bed heterogeneity following the methods of Burnham (1993), which are interpreted to reflect the degree of ecological heterogeneity in the source vegetation, using Sorensen's Coefficient of Similarity (Sorensen's Index, SI; 2C/[A + B]), where C is the number of species in common between two samples and A and B are the total number of species in each of the two samples (Burnham 1993). A mean Sorensen's Index was calculated for Beds 1, 2, 3, 4, 5, 7, and 8 from six pairwise comparisons between different collectors' individual quarries, and for Bed 1.5 from two pairwise comparisons. Richness estimates derived from forest leaf litter with Sorensen's Indices of less than 50% can underestimate that of source vegetation by as much as three times, whereas leaf litter samples with mean SIs greater than 50% underestimate source vegetation richness by only 1.4 times (Burnham 1993).

FIGURE 4. Standing species richness (A), number of species extinctions expressed as a proportion of standing species richness (B), species originations (C), range-through species (D), and number of singletons (species occurring in one fossil plant bed only) expressed as a proportion of A (E). A-E calculated from presence absence data of Harris (1937) for Astartekleft (Appendix 2). see text for more details on methods. Phases 1 to 4 represent distinct evolutionary/ecological phases in the vegetation of East Greenland which are described in detail in the text.

For comparison we have also estimated standing species richness (Fig. 4A), species origination/immigration (Fig. 4C), species extinction/emigration (Fig. 4B), and rangethrough species (Fig. 4D) from Harris's (1937) original presence/absence data set for Astartekloft following the methods of Foote (2000) and elaborated on for application to paleobotanical specimens by Wilf and Johnson (2004). This analysis excludes singletons (species occurring in one fossil plant bed only) and assumes the presence of taxa in a plant bed if it occurs in the beds both immediately above and below, i.e., with range-through occurrences (Appendix 2).

Evenness.-Changes in the patterns of dominance versus evenness of the vegetation spanning the Tr/J boundary were assessed from analyses of the relative abundance matrix obtained from the census- collected fossil plant beds of Astartekloft using Shannon's Diversity (H) divided by the logarithm of number of taxa, using PAST(c) Version 1.33. Specifically, we were interested in addressing whether there was any evidence for ecosystem instability, as indicated by a decrease in evenness, prior to the Tr/J boundary. The advantage of tracking changes in evenness in addition to richness is that evenness can be accurately estimated from the fossil record irrespective of sample size (Peters 2004).

Results

Biodiversity Richness.-There is an 85% decline in standing species richness of plant communities in the Astartekloft region throughout the Rhaetian with minimum levels coinciding with the Tr/ J boundary and the most negative carbon isotope values (Figs. 1, 4A). This marked decline in standing species richness is attributable to a combination of elevated levels of species extinction at and immediately prior to the Tr/J boundary (70% in Bed 4 and 80% in Bed 5) (Fig. 4B), depressed levels of species originations (Fig. 4C), and a sharp decline in the number of range- through taxa (Fig. 4D), all initiated prior to the deposition of Bed 3 and continuing through the Rhaetian. At the species level, therefore, species turnover at this locality appears to have increased sharply at the onset of the negative carbon isotope excursion and remained exceptionally high through the latest Rhaetian with peak proportional extinction coinciding with the Tr/J boundary.

FIGURE 5. Rarefaction curves of summed Triassic (Beds 1, 1.5, 2, 3, 4, 5) and Jurassic (Beds 6, 7, 8) leaf generic richness versus number of macrofossil genera. Phases 1 to 4 represent distinct evolutionary/ecological phases in the vegetation of East Greenland, which are described in detail in the text.

FIGURE 6. Rarefaction curves of leaf generic richness versus number of macrofossil specimens for each of the nine fossil plant beds at Astartekl0ft normalized for 220 specimens per bed with the exception of Beds 1.5 and 6 (see Fig. 4).

A comparison of lumped Rhaetian and Hettangian rarefied leaf generic richness indicates no significant differences (Fig. 5), suggesting that, at the generic level at least, recovery of Jurassic plant communities following Tr/J turnover must have occurred relatively rapidly (within a few million years). When the data are broken down into individual beds, a clearer pattern of changes in generic diversity emerges (Fig. 6). Beds 1, 1.5, and 2, the oldest Triassic fossil plant beds, contain the richest macrofossil assemblages at Astartekloft, with on average 12 to 13 genera each. A marked ~35% loss in generic richness occurs in Beds 3 and 4 (with 7.2 and 8.3 leaf genera, respectively). For Bed 3 in particular, the diversity loss coincides with the most positive stable carbon isotope values recorded for the whole succession. Bed 5, which marks the transition zone between the Lepidopteris zone of the Triassic and Thaumatopteris zone of the Jurassic, is richer than Beds 3 and 4, with rarefaction results indicating 10.9 leaf genera. However, Bed 5 groups with Beds 3 and 4 in demonstrating the lowest generic richness among all nine fossil plant beds analyzed. A conservative estimate suggests an average generic loss of ~ 17% between the oldest and youngest (including the boundary) Triassic fossil plant beds. Importantly, the loss occurs in both genus and species richness, and prior to onset of the negative carbon isotope excursion but coincident with an initial 250 to 500 ppmV rise in atmospheric CO2 concentration and presumed warming of global temperatures (Fig. 1). This suggests that profound changes in richness and stability of these plant communities were in place before maximum environmental change occurred at the Tr/J boundary. The lowest Jurassic assemblage (Bed 6), which represents the only coal, has an unexpectedly high rarefied generic richness of 14.5. It must be noted that this value was extrapolated from 120 to 220 individuals on the basis of the relationship between rarefied generic richness and 120 individuals. However, even if further collection of fossil plant specimens from this bed demonstrates that the true generic diversity has already been captured with 120 specimens, a raw diversity measure of 12 (for 120 individuals) is still higher than for the underlying three beds. These results indicate that plant generic richness may have rebounded rapidly in the Astartekloft region of Jameson Land (i.e., within a few million years rather than tens of millions of years) following Tr/J extinctions. As Bed 6 represents the only coal within the sections at Astartekloft, an alternative interpretation of these results would be that generic richness rebounded rapidly among peat-forming plant communities or that they were unaffected by events at the Tr/ J boundary. We cannot exclude any of these interpretations until a Triassic assemblage with a similar depositional environment has been identified, for comparison with Bed 6.

FIGURE 7. Stable carbon isotopic record from fossil wood (A) compared with changes in rarefied leaf generic richness (B), evenness (which is opposite to ecological dominance) (C), and ecological composition (D), calculated from relative abundance data derived from a total of 4303 macrofossil leaf specimens from Astartekloft.

Considering richness in isolation from other measures of biodiversity, these data suggest that ecosystem instability, as indicated by loss in plant primary productivity via emigration and regional extinction, was in place before the Tr/J boundary. Although generic diversity may have rebounded rapidly in the Jurassic (Figs. 7, 8B), the depleted levels of standing species diversity at Astartekloft (Fig. 4A) in both swamp and non-waterlogged plant communities suggest that species turnover remained high in the Jurassic. If the communities rebounded from Tr/J boundary environmental disturbances, their composition shows little temporal persistence. This pattern is supported by the observation that Jurassic plant communities are characterized by an extremely high relative proportion of singletons (on average >40%)-an order of magnitude greater than observed among those in the Rhaetian (on average <4%)-suggesting exceptionally high species flux in the post- boundary interval.

Biodiversity Evenness.-The pattern of change in generic richness from the Triassic to Jurassic was in part mirrored by observed shifts in evenness (Fig. 7C). Plant community composition was extremely even in the oldest Triassic plant beds (1, 1.5, and 2) with no single taxon dominating. Forest canopy dominants of the levees include Podozamites, Ginkgo, and Baiera, whereas trees of the floodplain swamps were represented by Elatocladus, Stachyotaxus, and to a much lesser extent Podozamites. A rich and even understory of bennettites (Pterophyllum and Anomozamites) occurred in the better- drained areas, whereas dipteridaceous ferns (Dictyophyllum, Hausmannia), cycads (Pseudoctenis and Nilssonia), and Anomozamites were prevalent in the floodplain settings. A dramatic decrease in evenness is evident in Beds 3 and 4. The levee and swampy floodplain environments became completely dominated by Podozamites (86% of relative abundance) and Pterophyllum (50% relative abundance) respectively. This pattern of increasing ecological dominance by a single taxon and decreasing evenness continues to the Tr/J boundary in Bed 5 where Stachyotaxus septentrionalis became one of the dominant taxa. Evenness did not rebound to pre-excursion levels until the youngest Hettangian plant bed (Bed 8, Fig. 7C).

FIGURE 8. A, Changes in heterogeneity of the source vegetation calculated using Sorensen's Index (SI) on different quarries from nine fossil plant beds at Astartekleft. An SI of 70% indicates that fossil leaf specimens collected from two different quarries from the same fossil plant bed (and assumed two different samples from the source vegetation) share 70% of the same taxa. B, Estimates of paleolatitude and inferred paleotemperature based on Rees's morphocat system (Rees et al. 2000). A lower mean DCA score indicates that the composition of the particular fossil plant bed in East Greenland provides a signal that was typically associated with lower paleolatitudes throughout the entire Jurassic and therefore from assumed higher paleotemperature (Rees et al. 2000).

Heterogeneity.-Moving from the oldest to the youngest fossil plant beds of the Triassic, a slight but insignificant increase in mean Sorensen's Index (SI) is observed, indicating a decrease in spatial heterogeneity of the vegetation through the Triassic (Fig. 8). The most unexpected shift in ecological heterogeneity occurs in Bed 5 at the Tr/J boundary, which has a lower SI (47 +- 3%) than all other Triassic and Jurassic plant beds analyzed. If our paleoecological data provide an accurate reflection of the ecological makeup of the standing vegetation, these results suggest that despite decreasing evenness the vegetation was slightly more patchy at the Tr/J boundary than any other time before or after the extinction event. The increased patchiness could reflect increased isolation of dry-ground communities to relatively higher and therefore drier topography as wet and swampy habitats favoring wet- loving plant communities expanded. Increased ecological importance of the wet-loving communities in Bed 5 is indicated by increased relative abundances of Stachyotaxus (from 2% to >40%) and the presence of Neocalamites and Ptilozamites (Appendix 3). An alternative explanation is that patchy extinction among Late Triassic plant communities and subsequent invasion by different Jurassic plant taxa resulted in increased spatial heterogeneity in the boundary bed. The shift in heterogeneity at the boundary is not, however, statistically significant and is unlikely to bias our estimates of generic richness any more than the other Triassic and Jurassic plant beds, as the SI is above 50% if standard errors are taken into account.

Composition.-Marked compositional differences are evident between Triassic and Jurassic plant communities in the region of Astartekloft. In the Triassic, four genera, Podozamites, Pterophyllum, Anomozamites, and Dictyophyllum make up on average over 75% of the relative abundance, but collectively these genera constitute less than 10% of the relative abundance in Jurassic communities. Similarly, four taxa (Czekanowskia, Sphenobaiera, Ginkgo and Cladophlebis) that collectively dominate Jurassic plant communities (>89% of relative abundance) were only very minor components of Triassic ecosystems, each contributing less than 5% of the relative abundance. Results from detrended correspondence analysis indicate that the age of the fossil plant assemblage exerts a stronger control on composition than does depositional environment (Fig. 9). These results suggest large-scale turnover of the ecological dominants after the Tr/J boundary, that cannot be explained by taphonomic differences between Triassic (all sheet splay) and Jurassic (swamp and abandoned channel) fossil plant assemblages. For instance, the floral composition of Bed 1 from South Tancrediakl0ft groups more closely with the same-aged (Triassic) plant beds from Astartekl0ft than with assemblages from the same depositional environment (Fig. 9).

FIGURE 9. Scatter plot results of detrended correspondence analysis (DCA) of fossil plant occurrence data from Astartekloft compared with preliminary occurrence data collected from South Tancrediakloft (Tan.). Triassic fossil plants beds are in italics and Jurassic fossil plant beds are underlined. Note that Bed 1 at Tancrediakloft groups more closely with similarly aged fossil plant beds from Astartekloft (despite differences in their depositional setting) than it does with Jurassic aged fossil plant beds at Astartekloft from the same depositional setting (abandoned channels). Axis 1, eigenvalue = 0.4462; Axis 2, eigenvalue = 0.3563.

The rate of compositional change at Astartekloft is reflected by a stepped change in the first eigenvector (axis 1 using EXIA) throughout the Rhaetian and a steady recovery after Tr/J extinctions (Fig. TD). Importantly, the exclusion of genera that occur in only one bed from our DCA does not change the observed pattern of compositional change (data not shown). DCA also reveals that the Tr/ J boundary bed (Bed 5) is the most compositionally distinct of all nine fossil plant beds analyzed (Fig. 7D). Relative abundance data from Bed 5 indicate that vegetation was co-dominated by Podozamites (~50%) and by Sachyotaxus (40%), a rare component of the vegetation in the oldest Rhaetian beds, with an average relative abundance of <2%. The floodplain ground cover and perhaps subcanopy, where present, were dominated almost exclusively by Pseudoctenis with rare occurrences of Neocalamites, Ptilozamites, and Taeniopteris. Other paleoecological studies have shown that such a floristic composition is usually associated with permanently or frequently flooded environments (Archangelsky et al. 1995; Howe and Cantrill 2001).

These ecological changes parallel those observed for both richness and evenness, indieating that marked changes in plant communities were initiated before the Tr/J boundary and delta^sup 13^C excursion. Together the results more strongly support a gradual or pulsed triggering mechanism for Tr/J boundary biodiversity loss rather than a single catastrophic event coinciding with peak negative isotopic values used to define the boundary globally. Long- term environmental change appears to have resulted in a major change in the dominance-diversity structure of plant communities prior to and coincident with the Tr/J boundary at Astartekloft. Ecological Selectivity of Tr/J Extinction

Comparison of the relative abundance distributions of summed Rhaetian and Hettangian macrofossil plant beds (Fig. 10) indicates that several key ecological and biological traits are associated with increased extinction risk at the Tr/J boundary. These include ecological rarity and complex reproductive biology, two traits that are predicted to increase extinction risk alone or synergistically according to meta-population theory (Gaston 1994; Lawton et al. 1994; McKinney 1997), and large leaf size, a trait predicted to be disadvantageous by ecophysiological modeling under a Tr/J global warming scenario (McElwain et al. 1999). At the generic level, extinctions recorded at Astartekl0ft and in the Jameson Land region occurred predominantly among (1) rare genera (defined here as genera with relative abundances in the last quartile of the rank abundances, following Gaston, 1994) including dipteridaceous ferns Clathropteris and Hausmannia, the bennettite "flower" Weltrichia, and foliage genera Anthrophyopsis (incertae sedis, ?Cycadales), Pachypteris (Corystospermales), and Macrotaeniopteris (incertae sedis, ?Cycadales); (2) genera with extremely large leaves (Anthrophyopsis, Clathropteris, Hausmannia, Macrotaeniopteris); and (3) those genera, judging from their reproductive anatomy, that likely would have required specialist insect vectors for pollination (Wielandiella and Weltrichia).

Physiognomic Selectivity of Tr/J Extinctions.-The selective regional extinction of rare large-leaved Triassic taxa such as Anthrophyopsis (~20 x 100 cm), Clathropteris (~30 x 40 cm), Hausmannia (>~20 cm in width), and Macrotaeniopteris (20 x >50 cm) is consistent with model expectations based on leaf physiological responses to increased atmospheric CO2 concentration and mean summer temperature. McElwain et al. (1999) modeled leaf temperature responses to a fourfold increase in atmospheric CO2 and a 3- 4[degrees]C global temperature rise across the Tr/J boundary. They predicted that large-leaved taxa would become more dissected and/or reduced in size in order to avoid exceeding lethal leaf temperature limits under a 5[degrees]C global warming. Recent paleoclimatic modeling shows that a fourfold increase in atmospheric CO2 may have raised local summer temperatures in the Jameson Land Basin by 16[degrees]C, resulting in mean summer temperatures of 36[degrees]C (Huynh and Poulsen 2005). Under such high summer temperatures, leaf temperatures with average undissected leaf widths of >3 cm would readily exceed 50-55[degrees]C, a temperature range considered lethal for tropical/subtropical woody taxa (Gauslaa 1984; Larcher 1994). The demonstration by McElwain et al. (1999) of the replacement of large canopy leaves with progressively smaller and more dissected leaves throughout the Late Triassic corroborates expectations based on ecophysiological modeling of leaf temperature. The results presented here suggest that lethal leaf temperatures may also have adversely affected large-leaved and presumed understory or ground cover taxa such as Clathropteris, Hausmannia, Anthrophyopsis, and Macrotaeniopteris.

FIGURE 10. Rank-order abundance plots for summed Triassic versus Jurassic fossil leaf and reproductive genera calculated from 4303 macrofossil specimens collected from Astartekloft, East Greenland. Dotted line indicates the four quartile (to the right) of the data, which are taxa considered ecologically rare using the definition of Gaston (1994).

Reproductive Selectivity of Tr/J Extinctions.-Analysis of Harris's (1937) presence/absence data set for the entire Jameson Land Basin reveals that 11 of 13 known bennettite taxa discovered in the Kap Stewart Group fossil plant beds (Wielandiella and Weltrichia) are known only from the Lepidopteris zone (Triassic). Our relative abundance data set (Table 3) indicates that Pterophyllum and Anomozamites, both bennettite foliage taxa, experienced the second and third most severe average losses in relative abundances of all taxa recorded from the Triassic to Jurassic, respectively: Pterophyllum declines from 20% to 1.3 %, and Anomozamites from 17% to 0.1% (Table 3). Together these data suggest that bennettites were severely reduced, in terms of both their taxonomic richness and ecological importance in Jurassic compared with Triassic plant communities across the entire Jameson Land Basin. The protection of ovules among interseminal scales and within flower-like reproductive structures (Delevoryas 1963; Harris 1974) and the presence of nectaries (Crane 1985; Crepet 1974; Delevoryas 1963) are all indicative of a specialized pollination syndrome, most likely involving insects, among bennettites.

For cycads, only Doratophyllum experienced regional extinction, and Ctenis, Pseudoctenis, and Nilssonia, all of which had low to rare abundances in the Triassic, are completely absent from Jurassic plant beds of Astartekloft. The majority of modern cycad taxa display mutualistic relationships with highly effective specialist insect vectors including beetles, weevils, and thrips (Oberprieler 1995a,b; Mound and Terry 2001; Terry et al. 2005). Volatiles released from both male and female cones attract insect vectors to visit, acquire, and transport pollen (up to 10,000 pollen grains per individual) between male and female cones, thus pollinating ovules (Donaldson 1997). In return, insect vectors receive a ready food source (pollen from the male and pollen droplet from the female), protection, warmth, and oviposition sites (Terry et al. 2005). If these highly mutualistic relationships between insects and cycads were in place by the Mesozoic, then extinction of one of the partners in the relationship would result in the relatively rapid demise of the other. Recent observations of Triassic insect coprolites loaded with pollen within a permineralized cycad cone suggest that such mutualistic relationships may have a very long evolutionary history (Klavins et al. 2003, 2005). We hypothesize therefore that the higher levels of extinction among entomophilous (including bennettites and cycads) than in anemophilous plants at the Tr/J boundary were caused by a contemporaneous extinction of insect taxa. Alternatively, extinction and ecological demise of these plants could have triggered extinction of their insect vectors. In the absence of insect body fossils, or a comprehensive record of in sect feeding damage from the Jameson Land Basin, this hypothesis remains to be tested.

The extinction of Lepidopteris at the Tr/J boundary may also be due to reproductive specialization; however, this hypothesis is not conclusive and requires further investigation. Peltaspermum, the female reproductive structures of Lepidopteris, consist of relatively complex radially symmetrical peltate ovuliferous disks from which hang the ovules (Harris 1932a). The micropyle of the ovule in Lepidopteris is extremely elongated and angled at about 30[degrees], an anatomical character that may indicate adaptation for insect rather than wind pollination. Lepidopteris pollen does not provide strong support for an entomophilous reproductive habit, as it is small, 23 to 40 [mu]m in length (Townrow 1960), and unsculptured with a single sulcus. However, cycads, which are entomophilous, also have pollen is this size range. Additional, but admittedly more indirect, evidence for relatively complex reproductive biology in Lepidopteris can be gleaned from its likely habit, which we interpret as vinelike or liana-like (see Appendix 3).

Vegetation Recovery

Twenty-two percent (five) of fossil plant genera observed in Jurassic aged plant beds were not recorded in Triassic assemblages at Astartekloft, suggesting that they likely immigrated from elsewhere. All of these "new" Jurassic immigrants can be classified ecologically as either rare or of low abundance in terms of their relative abundance distributions (Fig. 10). Phlebopteris and Matonia (matoniaceous fern), Sagenopteris (Caytoniales), Marattia (marattiaceous fern), and Pagiophyllum, all new genera to East Greenland in the Jurassic, have relative abundances of less than 2.5%. Recovery of the plant communities following the marked biodiversity loss before and coincident with the Tr/J boundary was therefore achieved by recruitment of new dominants from within existing plant communities from the same region, rather than through origination at the generic level or via significant levels of immigration of exotic taxa. For instance, there is evidence for ecological expansion of Cladophlebis,which increased in relative abundance from <4% in the Triassic to ~39% in the earliest Jurassic plant bed (Bed 6), and of Stachyotaxus, which increased in abundance from <2% to 40% at the Tr/J boundary (Bed 5) and to 27% in Bed 6.

Bed 6 at Astartekloft occurs 14 meters above the Tr/J boundary, but well within the negative isotopic excursion (Figs. 1, 3) that characterizes the boundary interval globally. It is noteworthy in containing the highest proportion of thermophilous elements, which likely represent immigrants from much lower paleolatitudes. These include Pagiophyllum, which is commonly associated with the thermophilous pollen taxa Classopollis (Vakhrameev 1991; Axsmith et al. 2004; McElwain et al. 2005) and Sagenopteris. We estimated an average Jurassic paleolatitude, as a qualitative proxy for relative global paleotemperature, for all nine fossil plant beds following the methods of Rees et al. (2000). This method averages the observed paleolatitudes that each taxon typically occupied in the whole Jurassic period, using an extensive biogeographical database of Mesozoic fossil plants (Rees et al. 2000). The analysis indicates that the taxonomic composition of Bed 6 is more typical of much lower paleolatitudes, thus supporting our suggestion that maximum regional temperatures for the entire Rhaetian-Hettangian interval likely occurred in the earliest Hettangian, coincident with Bed 6 (Fig. 8B). Post Tr/J Boundary Fern Spike?

The earliest Jurassic macrofossil plant assemblages (represented by Bed 6) are notable for their high relative proportion of fern taxa. Bed 6 contains the highest abundance of fern taxa compared with any of the other Triassic or Jurassic fossil plant beds: 49% compared with a maximum of 31% recorded in the Triassic. The high fern sum in Bed 6 is mainly due to a high relative abundance of Osmundaceae (Cladophlebis denticulata); however, three other fern families are also present (Dipteridaceae, Matoniaceae, Marattiaceae). Fieldwork in 2002-2004 shows that this coal, although thin, is laterally extensive across the whole of the Jameson Land Basin, suggesting that conditions favoring peat-forming vegetation, such as high precipitation and/or reduced clastic input, must have prevailed at this time (Fig. 3). Higher precipitation or an invigorated hydrological cycle is an expected consequence of higher global temperatures, because of increased vapor holding capacity of the troposphere (Hay and DeConto 1999) and increased latent heat transfer from low to high latitudes (Ufnar et al. 2004).

Peat-forming vegetation and a high relative proportion of fern (trilete) spores (89%) are recorded from tropical paleolatitudes in the Newark Basin of North America at the Tr/J boundary (Fowell and Olsen 1993; Olsen et al. 2002). These coaly beds and the contained "fern spike" are unique to the Tr/J boundary interval and have been interpreted as representing an expansion of opportunistic "disaster taxa" following the catastrophic environmental effects of a meteorite impact (Fowell and Olsen 1993; Olsen et al. 2002). It is not unusual for peat-forming vegetation to contain a high proportion of fern taxa. It is intriguing, however, that the only coal deposits in both basins occur within the negative isotopic anomaly that characterizes the Tr/J boundary globally. Further work is now required to determine whether these high fern compositions are merely a taphonomic artifact or whether they signify a truly unique ecological and/or climatic event across the Northern Hemisphere at the Tr/J boundary and in the earliest Jurassic.

Tr/J Vegetation Dynamics at Astartekloft

Our paleoecological results enable subdivision of the Tr/J vegetation dynamics at Astartekloft into four distinct phases. In phase 1, prior to the negative isotopic excursion and CO2-induced global warming (Fig. 1), the Triassic vegetation of the Astartekloft region was a rich and heterogeneous broad-leaved, gymnosperm- dominated forest. Agathis-dominated forests of New Zealand and Australia today, with Podocarpus and an understory of the cycad Lepidozamia hopei, are a good modern analogue (White 1994). Podozamites, Ginkgo, Elatocladus, and Baiera most likely co- dominated the canopy, whereas Pagiophyllum and Anomozamites, and to a lesser extent Nilssonia and Pseudoctenis, likely formed a sub- canopy. Dipteridaceous ferns, Equisetites, and other rare large- leaved taxa such as Anthrophyopsis and Macrotaeniopteris made up the ground cover (Appendix 3). No single taxon dominated the canopy, sub- canopy, or ground-cover habitats, and the flora comprised a high percentage of taxa with complex reproductive cycles requiring animal vectors for fertilization and dispersal (cycads and bennettites). If our interpretations are correct, vines were also present in low to rare abundances (Lepidopteris) (Appendix 3).

In phase 2, generic richness, evenness, and standing species richness began to decline and ecological composition changed markedly. These vegetation changes are indicative of instability in local plant communities and coincide with evidence for deterioration in climatic and atmospheric conditions (McElwain et al. 1999; Huynh and Poulsen 2005). The canopy was first dominated by Podozamites, and Stachyotaxus became an additionally important element at the boundary. We hypothesize that a mid-canopy habit in the Triassic was almost completely eradicated in the Jameson Land Basin, owing to high emigration and/or extinction of erect cycads and bennettites (Appendix 3). Marked changes in dominance patterns of ground-cover taxa also occurred during the Triassic to Jurassic transition as an abundance of dipteridaceous ferns gave way to osmundaceous ferns. Phase 3 is characterized by low evenness, depressed generic richness, peak species extinction, and the most compositionally distinctive vegetation of the entire Rhaetian-Hettangian interval, as indicated from detrended correspondence analysis. This phase coincides with the first peak of the main negative isotopic excursion and maximum estimated pCO^sub 2^ and global temperature.

The post Tr/J recovery interval, defined here as phase 4, is characterized by a rebound of generic richness at Astartekloft; however, standing species richness for the entire Jameson Land Basin remained low throughout this phase, owing to low species origination and low numbers of range-through taxa. As Bed 6 is the only coal swamp deposit at Astartekloft, it is not yet possible to evaluate whether this apparent rebound in generic richness in Bed 6 is real or a taphonomic artifact. This will be possible only when a pre- boundary bed from the same depositional environment has been identified for direct comparison. Most of the raw species richness in Hettangian plant beds at Astartekloft is made up of singletons, suggesting extremely high species turnover during the post-boundary interval (Fig. 4E). The macroecological characteristics of the plant communities such as evenness and composition did not fully rebound to pre-excursion levels until the end of phase 4 as reflected in Beds 7 and 8, which occur >46 m above-and likely on the order of 3 to 4 million years after-the Tr/J boundary, indicating a long recovery period. It is noteworthy that the nature and tempo of vegetation dynamics spanning the Tr/J boundary parallel those observed in Carboniferous swamp forest spanning the Westphalian/ Stephanian boundary (DiMichele et al. 1996), hinting that there may be a common macroecological response of plants to global climate change.

Source: Paleobiology

More News in this Category