Milankovitch-scale palynological turnover across the Triassic–Jurassic transition at St. Audrie's Bay, SW UK

Abstract: A high-resolution palynological study of the Triassic–Jurassic boundary in the St. Audrie's Bay section revealed a palynofloral transition interval with four pronounced spore peaks in the Lilstock Formation. Regular cyclic increases in palynomorph concentrations can be linked with periods of increased runoff, and correspond to the orbital eccentricity cycle. Spore peaks can be related to precession-induced variations in monsoon strength. An implication is that the initial carbon isotope excursion lasted for at least 20 ka. Emergence during deposition of the Cotham Member had an influence on one of the peaks, which is dominated by spore-producing pioneer plants (e.g. horsetails and liverworts). There is no compelling evidence of a global end-Triassic spore spike that, by analogy with the K–T boundary fern spike, could be related to a catastrophic mass extinction event. Climate change is a more plausible mechanism to explain the increased amount of spores. Supplementary material: An alphabetical list of palynomorphs identified in the St. Audrie's Bay section is available at http://www.geolsoc.org.uk/SUP18406.

The Triassic-Jurassic (T-J) boundary interval, spanning one of the 'big five' episodes of mass extinction, is characterized by large-scale volcanism (Marzoli et al. 2004;Schaltegger et al. 2008), major carbon cycle perturbations (e.g. Hesselbo et al. 2002;Ruhl et al. 2009), climate change (e.g. McElwain et al. 1999;Korte et al. 2009) and pronounced vegetation changes (e.g. McElwain et al. 2007;Bonis et al. 2009). Explanations for the end-Triassic biotic crisis have included both gradual (e.g. sea-level change) and catastrophic mechanisms such as volcanism or a bolide impact (Olsen et al. 2002a,b;Tanner et al. 2004;Hesselbo et al. 2007). The climate was significantly different from today: there was no ice present at high palaeolatitudes (Frakes et al. 1992;Satterley 1996;Hallam & Wignall 1999) and the thermal contrast between the large low-latitude continental interior and the sea drove a strongly developed monsoonal circulation (Parrish 1993;Buratti & Cirilli 2007;Sellwood & Valdes 2007). The monsoonal activity influenced precipitation patterns, and consequently the floral distribution and vegetation development.
In palynological records, major biotic crises are sometimes characterized by a highly increased spore abundance. The best known example is at the Cretaceous-Tertiary (K-T) boundary, 65 Ma ago, where a spore spike was described at the boundary clay level (Nichols & Johnson 2008). For the T-J boundary, the evidence for a global spore spike is equivocal. A spore peak in the T-J transition interval was recognized in the Newark Basin (USA) and subsequently linked to an impact-induced mass extinction (e.g. Fowell & Olsen 1993;Olsen et al. 2002a,b). In contrast, gymnosperm forests on land adjacent to the Eiberg Basin (Austria) were gradually replaced by ferns and fernassociated vegetation (Kürschner et al. 2007;Bonis et al. 2009). Van de Schootbrugge et al. (2009) interpreted T-J fern proliferation as a pioneer assemblage after the disturbance of the terrestrial ecosystems by the release of pollutants during flood basalt volcanism. However, that interpretation is rather ambig-uous because it does not account for possible changes in the pollen/spore ratio caused by sedimentological facies changes and related taphonomical processes (Neves effect) or environmental changes that may have resulted in an increased dominance of spore-producing plants (e.g. Traverse 2007).
In St. Audrie's Bay (Somerset, UK), a key T-J boundary section, an interval with increased spore abundance is present, 8-6 m below the first occurrence of Jurassic ammonites (Hounslow et al. 2004;Warrington et al. 2008). In this paper, we present high-resolution and quantitative palynological data from St. Audrie's Bay to shed new light on the continuing discussion of the dynamics of end-Triassic vegetation changes. We focus on the spore record, and consider whether the spore increase is caused by a sudden proliferation of pioneer vegetation in the aftermath of the end-Triassic crisis, or if other factors such as changes in climate and sea level are more important. We compare the spore record in St. Audrie's Bay with records from other regions to assess any similarities, and whether they can be used for long-range correlation.

Study area and lithology
St. Audrie's Bay is a classic T-J marine boundary section located on the west Somerset coast in southwestern England (Fig. 1). The lithology and depositional environment have been reported in detail by Hesselbo et al. (2004), Hounslow et al. (2004) and Warrington et al. (2008) (and references therein). A short summary is given below.
The earliest Rhaetian Williton Member (upper part of the Blue Anchor Formation) was deposited in a shallow marine environment (Warrington et al. 2008). The distribution of facies in the succeeding Westbury Formation was controlled by fluctuations in relative sea level (Hesselbo et al. 2004). Three sedimentary cycles, representing alternating deposition in transgressive, littoral, high-energy environments and lower energy, stagnant or weakly oxygenated water bodies, may be present (Warrington et al. 2008). The transition from the Westbury Formation to the lower Cotham Member (Lilstock Formation) represents a shallowing of the depositional environment from shelf to peritidal water depths (Wignall & Bond 2008). A 0.5 m thick unit of deformed beds in the middle of the member (e.g. Mayall 1983) is followed by an erosional surface penetrated by deep cracks that are considered to reflect temporary emergence (Hesselbo et al. 2004). The exposure was extremely brief as suggested by the unstratified, single-generation fill of these desiccation cracks (Hesselbo et al. 2004). The deformed beds are interpreted as a seismite (Mayall 1983;Simms 2007). An alternative explanation could be a more prolonged period of seismic activity associated with the onset of the Central Atlantic Magmatic Province (Wignall & Bond 2008). The upper Cotham Member represents a coastal, lagoonal environment (Mayall 1983;Hesselbo et al. 2004;Mander et al. 2008) and fully marine beds alternate with beds containing freshwater bivalve taxa (Radley et al. 2008). The Cotham Member contains the initial negative carbon isotope excursion (Hesselbo et al. 2002). The Cotham Member-Langport Member junction was interpreted by Hesselbo et al. (2004) to represent a flooding surface. The Langport Member (Lilstock Formation) was deposited in a warm and shallow saline lagoonal environment (Warrington et al. 2008), or in a broad shallow seaway, or during sea-level rise on a carbonate ramp (Hesselbo et al. 2004). The depositional environment of the top of Langport Member is also disputed, as it may have formed either during relative sea-level fall, with regression culminating in sea-floor erosion and emergence at the top of the Langport Member (Wignall & Bond 2008), or during sea-level rise during the final drowning of the carbonate ramp (Hesselbo et al. 2004). The Blue Lias Formation was deposited during a phase of rapid flooding, indicated by development of a laminated, organic-rich shale (Hallam 1995(Hallam , 1997Warrington et al. 2008). The first occurrence (FO) of ammonites of the genus Psiloceras was proposed as a boundary marker for the base of the Jurassic (Cope et al. 1980;Warrington et al. 1980Warrington et al. , 1994Warrington et al. , 2008. The FO of Psiloceras planorbis is at the base of bed 13 in the Blue Lias Formation (Warrington et al. 2008).

Methods
Sixty-three rock samples from the St. Audrie's Bay section were selected for palynological analysis. The samples range from the upper part of the Williton Member to the Blue Lias Formation. Average sample spacing was c. 1 m, but 10 cm or less throughout the Lilstock Formation (Fig. 2). All sample levels cited are relative to the base of the Williton Member (0 m). Between 5 and 20 g of sediment was crushed into small fragments and dried for 24 h at 60 8C. A Lycopodium spore tablet was added to each sample. Subsequently, the samples were treated twice alternately with cold HCl (30%) and cold HF (40%) to remove the carbonates and silicates. The residues were sieved using a 250 ìm and a 15 ìm mesh. ZnCl 2 was applied to separate the lighter organic material from the heavier mineral particles. The lighter fraction was transferred from the test-tube and sieved once more using a 15 ìm mesh. The remaining organic material was mounted on two slides per sample with glycerine jelly. The slides are stored in the collection of the Section Palaeoecology, Laboratory of Palaeobotany and Palynology, Utrecht University, Netherlands. Pollen and spore identification was mainly based on Schulz (1967), Morbey (1975), Schuurman (1976Schuurman ( , 1977Schuurman ( , 1979 and Lund (1977).
The identified morphotaxa of spores, pollen and aquatic palynomorphs are listed in the Supplementary Material. About 300 terrestrial palynomorphs were counted per sample (quantitative analysis; see palynomorph sums in Figs 3 and 4). Lycopodium spores were counted concomitantly, but excluded from the terrestrial palynomorph sum. The palynomorph concentrations (absolute number of palynomorphs per gram in samples) were calculated based on the fossil palynomorphs counted, the Lycopodium spores counted, the dry weight of the samples, and the total number of Lycopodium spores added to the sample. Relative abundances were calculated and plotted using the Tilia and TgView programs (Grimm 1991(Grimm -2001. Terrestrial palynomorph assemblages were established by constrained cluster analysis using CONISS (Grimm 1987) within Tilia. A subsequent qualitative analysis, scanning two complete slides per sample, was carried out to check if rare palynomorph taxa were present that could be of biostratigraphic value. The complete presence/ absence dataset is available on request.
A linear ordination method, principal components analysis (PCA), was carried out on the relative pollen and spore abundance data. By relating the palynomorph taxa with their botanical affinity we interpreted the axes and revealed temperature and humidity gradients in the data.
Frequency analysis was performed on the terrestrial palynomorph concentration record (0.3-28.5 m) and the relative spore abundance record (12-14.5 m) using AnalySeries 1.1.1 (Paillard et al. 1996). Data were linearly detrended before the Blackman-Tuckey method was applied (compromise predefined level, Barlett window). Power spectra for each proxy record are reported in cycles cm À1 with a 90% confidence interval, and a Gaussian band-pass filter from the main peaks of each proxy record is reported (see below).

Terrestrial v. aquatic palynomorphs
In the lower part of the Westbury Formation (below 1010 cm) the palynomorph assemblages are generally dominated by terrestrial palynomorphs. Higher aquatic palynomorph abundance has been recorded in the Williton Member (up to 75%) and in the lower Westbury Formation (up to 68%) (Fig. 2). Between 1010 and 1580 cm most samples are dominated by aquatic palynomorphs. Between 1580 cm and the top of the section sampled the assemblages consist mainly of terrestrial palynomorphs (55-97%). A cyclic pattern is visible in the palynomorph concentration. The Lilstock Formation is characterized by a very low total palynomorph concentration and a higher aquatic than terrestrial palynomorph content (Fig. 2). Most samples from the Williton Member and Westbury Formation have a higher terrestrial than aquatic palynomorph content. Exceptions are the samples from 130, 400, 1010 and 1120 cm. The terrestrial palynomorph content is also higher in the Blue Lias Formation, except in some of the lowest samples (1460-1580 cm). The trends in terrestrial and aquatic palynomorph concentrations are simultaneous (Fig. 2).

Terrestrial palynomorphs
Significant changes occur in the pollen:spore ratio across the T-J boundary interval (Fig. 2). In the Westbury Formation the samples are characterized by a high amount of pollen. Only in the sample at 610 cm does spore abundance exceed 50% (spore peak 1). Samples from the Lilstock Formation have varying pollen and spore peak abundances. The total palynomorph concentration is very low in this interval (Fig. 2). Two spore peaks (2: 53%; 3: 68%) are present in the Cotham Member, and two even higher spore peaks (4: 96%; 5: 76%) in the Langport Member. Pollen represents the most abundant palynomorph category throughout the Blue Lias Formation although there are some small increases in spore abundances. Within the terrestrial palynomorph fraction, four assemblages (SAB1-SAB4) are distinguished, based on cluster analysis (Fig. 3), as follows.
Assemblage SAB1. This assemblage occurs between the base of the section studied and the top of the Westbury Formation. It is characterized by high amounts of pollen, mainly Classopollis meyeriana, C. classoides, Ovalipollis pseudoalatus and Rhaetipollis germanicus. In the lower part of the assemblage, C. classoides and O. pseudoalatus are more abundant than in the upper part. A prominent spore type is Ricciisporites tuberculatus, with a major peak (71%) at 610 cm and minor peaks at 290 and 880 cm. Granuloperculatipollis rudis has its last common occurrence at the top of this assemblage.
Assemblage SAB3. This assemblage occurs between 1275 cm in the Cotham Member and the top of the Langport Member (1435 cm). A main feature is the absence of V. bjuvensis and T. pseudomassulae. Samples within this assemblage show a large variation in species composition. Classopollis meyeriana is the most abundant pollen in the lower samples and an acme of C. classoides (74%) occurs at 1390 cm. The most abundant spores are H. reissingeri, Deltoidospora spp., Acanthotriletes varius, Baculatisporites spp., Conbaculatisporites spp., Todisporites spp., Concavisporites spp. and Trachysporites fuscus. Two spore peaks are present (Fig. 4). The first peak (4: up to 96% spores) is based on four samples of which the lower two (1340 and 1350 cm: peak 4a) differ in species composition from the upper two (1360 and 1370 cm: peak 4b). The two lower samples consist mainly of Deltoidospora spp., A. varius, H. reissingeri, Concavisporites spp., Conbaculatisporites spp. and Trachysporites fuscus. The two upper samples have a high abundance of H. reissingeri, accompanied by Deltoidospora spp. The major components of the second peak (5: 75% spores) at 1430 cm, are Deltoidospora spp., A. varius and Conbaculatisporites spp.
Assemblage SAB4. This assemblage occurs between 1435 cm and the top of the studied section. This assemblage is characterized by the dominance of C. meyeriana (75-100%) and common Pinuspollenites minimus. The amount of spores increases slightly at some levels, with H. reissingeri as a major component. The largest spore 'peak' is only 22%. Of biostratigraphic importance are the first occurrences of Cerebropollenites thiergartii and Ischyosporites variegatus at 1850 cm.

Aquatic palynomorphs
The three most prominent aquatic palynomorphs in the record are dinoflagellate cysts (Heibergella sp. A, Rhaetogonyaulax rhaetica, Dapcodinium priscum), Micrhystridium spp., and cf. Leiosphaeridia (Fig. 5). Notably, Heibergella sp. A is present in only one sample in the Williton Member, where it has an abundance of 88%. At the transition from the Cotham Member to the Langport Member there is a shift from an assemblage dominated by dinoflagellate cysts (mainly R. rhaetica) and prasinophytes to one dominated by prasinophytes (cf. Leiosphaeridia) and acritarchs (Micrhystridium spp.). Botryococcus disappears in the Langport Member at 1360 cm. Leiofusa jurassica is an acritarch of biostratigraphic importance, which has its first occurrence at 1850 cm and a minor peak abundance (5%) in the Blue Lias Formation at 1890 cm.

Principal components analysis
The two main ordination axes are the dimensions through the dataset with the largest variance in species composition (Fig. 6). These axes are explained in terms of the environmental or climatic gradient that controls the dataset. The first axis represents a gradient from relatively warm (e.g. C. meyeriana) to 'cold' (e.g. C. classoides) palynomorph taxa (Fig. 6). The second axis represents a gradient from relatively dry (e.g. C. meyeriana and other conifers) to wetter taxa (spore-producing plants). The sample scores on the second axis are used to derive a trend in relative humidity (Fig. 7). In general, climate was relatively dry throughout the time represented by the section studied. The Lilstock Formation is characterized by an interval of wetter phases, corresponding with the spore peaks.

Frequency analysis
Time-series analyses of terrestrial palynomorph proxy data show strong oscillations throughout the T-J transition (Fig. 7c). Six periodic increases in terrestrial palynomorph concentrations, of up to 200 3 10 3 palynomorphs g À1 sediment, coincide with low relative spore abundances. A power-spectrum (Fig. 7a), reflecting the main periodic oscillation in the terrestrial palynomorph concentration record, is marked by a .90% significant peak with a c. 470 cm periodicity. Large fluctuations in relative spore abundance are confined to the Lilstock Formation. At least four periodic fluctuations with a c. 100 cm periodicity (Fig. 7b) coincide with one c. 470 cm periodic cycle in the terrestrial palynomorph concentration record. The fundamental precession frequencies of the Earth's orbital parameters decreased to c. 20 ka in the early Jurassic (Berger et al. 1992). The duration of eccentricity cycles, however, remained constant. The distinct wavelengths of the oscillations in our palynological records can be tentatively linked to the orbitally controlled precession and c. 100 ka eccentricity frequencies that are corrected for the late Triassic.

Palynology
The St. Audrie's Bay palynological record is characterized by a marked palynofloral change within the Cotham Member at 1275 cm. Conifer-dominated hardwood vegetation was replaced by a monotonous Cheirolepidiaceous forest (Fig. 3). Orbell (1973) proposed older (Rhaetipollis) and younger (Heliosporites) palynomorph zones from the British T-J transition and placed the boundary between these at a rapid decline in the abundances of Ovalipollis ovalis, Rhaetipollis germanicus and Ricciisporites tuberculatus. This boundary corresponds to the boundary between SAB2 and SAB3 (Fig. 3) but its exact position is questionable as the decline in the numbers of these taxa is not usually synchronous (Warrington 2005). Our results are in good agreement with the previous palynomorph studies of St. Audrie's Bay (e.g. Hounslow et al. 2004;Warrington et al. 2008). However, the present higher resolution reveals new palynological findings. The first finding is that there are two transitional zones within each of the palynofloras recognized by Orbell (1973). The upper part of the Rhaetipollis zone, SAB2, shows an acme of Vitreisporites bjuvensis and Tsugaepollenites pseudomassulae and an increase in relative spore abundance (e.g. Porcellispora longdonensis, Polypodiisporites polymicroforatus). The lower part of the Heliosporites zone, SAB3, is characterized by the absence of O. pseudoalatus and Rhaetipollis germanicus. This zone also has a high abundance of spores, dominated by Heliosporites reissingeri, Deltoidospora spp. and Acanthotriletes varius. The second finding is that the FO of Cerebropollenites Fig. 6. Principal components analysis (PCA) biplot of the pollen and spore percentage data. The plot shows a gradient from relatively warm to cooler taxa along axis 1, whereas axis 2 represents a drier to wetter vegetation gradient. Fig. 7. Power spectra of (a) the terrestrial palynomorph concentration with a main periodicity of c. 470 cm and (b) the relative spore abundance with main periodicity of c. 100 cm. Gaussian band-pass filters reflect periodic changes in both proxy records (c) and are tentatively assigned to the astronomical c. 100 ka eccentricity (continuous grey lines) and c. 20 ka precession cycles (dashed grey lines), based on the ratio between the thicknesses of both cycles. Periodic alternations in spore abundance occur on a similar regular basis as periodic precession-induced black shale occurrences higher in the section (Ruhl 2010). The sample levels are relative to the base of the Williton Member (0 m). thiergartii appears to be a useful biostratigraphic marker allowing the correlation between terrestrial and marine realms (Kürschner et al. 2007;Bonis et al. 2009). In the St. Audrie's Bay section the FO of this pollen taxon is close to that of Psiloceras planorbis (Fig. 3). Ischyosporites variegatus and the acritarch Leiofusa jurassica have their FO in the Blue Lias Formation at the same level as C. thiergartii (Figs 3 and 5). Also, in an earlier study of St. Audrie's Bay, Leiofusa jurassica was found in the T-J transition interval (Warrington 1981;Hounslow et al. 2004;Warrington et al. 2008). A Jurassic acme of this species was recorded from Danish sections and from the NW German Basin (Lund 1977;Dybkjaer 1991). According to Lund (1977), the basal part of the Pinuspollenites-Trachysporites zone is characterized by this acme. This is in concordance with results from the Eiberg Basin (Austria, western Tethys), where the base of the Pinuspollenites-Trachysporites zone correlates with the base of the Jurassic TH zone, containing the FO of C. thiergartii (Bonis et al. 2009). Van de Schootbrugge et al. (2007) reported an FO of Leiofusa jurassica slightly lower in the St. Audrie's Bay section (c. 1820 cm). The third finding is that the increased spore abundance in the Lilstock Formation, as described by Hounslow et al. (2004) and Warrington et al. (2008), now appears to comprise multiple spore peaks (Figs 2-4), the main constituents of which are summarized in Table 1; the nature and cause of these peaks are discussed below.

Climate change
Orbitally induced variations in solar radiation (Milankovitch cycles) have exerted a strong influence on the Earth's climate throughout geological time (Berger et al. 1992;Olsen & Kent 1996;Van der Zwan 2002;Popescu et al. 2006;Ruddiman 2006). The amount of incoming solar radiation in present-day low-latitude systems depends mainly on precession. Maximum precession leads to maximum insolation and corresponds to times of maximum monsoon intensity (Vollmer et al. 2008). Changes in monsoonal activity have immediate consequences for atmospheric circulation (Crowley et al. 1992), the magnitude of precipitation rates (Vollmer et al. 2008), runoff, and weathering patterns, which potentially translates terrestrial changes to the marine realm (Crowley et al. 1992). The large Pangaean landmass may have intensified the monsoon system because the larger land area could retain more heat (e.g. Crowley et al. 1992). A modelling study by Kutzbach (1994) suggested that rainfall and runoff would undergo cyclic changes with periods of 23 ka over a substantial part of the Pangaean (sub)tropics. The low palaeolatitude (c. 308) of the St. Audrie's Bay section (Kent & Tauxe 2005) suggests an influence of monsoonal activity. Cyclic fluctuations in palynomorph concentrations (with a c. 470 cm periodicity) are present in the St. Audrie's Bay record (Figs 2 and 7). Additionally, four distinct peaks in relative spore abundance (with c. 100 cm periodicity) occur within one of these longer cycles (Fig. 7). Periods of increased spore abundance probably reflect wet phases, which may be related to intensified monsoon activity on the precession scale. The different periodic oscillations in terrestrial palynomorph records may be tentatively assigned to astronomical climate forcing with a c. 100 ka eccentricity and c. 20 ka precession wavelength (Fig. 7).
A qualitative and quantitative analysis of palynofacies revealed orbital cycles in the Early Jurassic in the southern UK  Fowell & Olsen (1993) and Fowell et al. (1994); data for the NW German Basin are from Van de Schootbrugge et al. (2009). The percentage in parentheses following the location name is the maximum total relative spore abundance. *Granulatisporites infirmus and Converrucosisporites cameronii dominate the fern-spike assemblages (Fowell & Olsen 1995). (Waterhouse 1999a,b). The precession cycle acted mainly on the terrestrial environment, probably via climate-controlled variations in runoff that affected terrestrial organic debris (Waterhouse 1999a), and the 100 ka eccentricity cycle controlled relative sea level (Waterhouse 1999b). However, in the absence of ice-sheets during the T-J period (e.g. Frakes et al. 1992;Satterley 1996;Hallam & Wignall 1999), a 100 ka sea-level cycle is unlikely. Late Triassic (Norian) playa cycles in the Mid-German Basin were associated with varying monsoon activity (Vollmer et al. 2008). The palaeoclimate model for the Mid-German Basin implies highest rainfall when summer solstice passed through perihelion (Earth closest to the Sun) in the northern hemisphere (Vollmer et al. 2008, p. 12, fig. 9). Insolation was highest, Tethyan seawaters evaporated, and moisture was transported to the north, where it precipitated in the playa system (monsoonal maximum). Insolation was lowest when the northern hemisphere summer occurred during passage of the solstice through aphelion. Decreasing evaporation of Tethyan seawater supplied less moisture and a drier climate resulted (monsoonal minimum). The wet and dry periods in the Mid-German Basin occur within a single precession cycle. A similar palaeoclimate model may be applied to the St. Audrie's Bay record, with the spore peaks representing monsoonal maxima. Also, Late Triassic palynological data from northern Spain revealed peak abundance of hygrophytic plants that may reflect the strong monsoon precipitation regime (Gómez et al. 2007). Olsen & Kent (1996) considered that precession-related cycles in precipitation (including the powerful effect of the eccentricity cycles on the expression of the precession) were a consistent feature of tropical climate during most times in Earth history. A modelling study by Crowley et al. (1992) suggested that climate responses to 100 ka eccentricity forcing can occur over low-latitude land areas involved in monsoon fluctuations. Kemp & Coe (2007) recognized 100 ka eccentricity cycles in the Late Triassic at St. Audrie's Bay on the basis of rock colour. The Newark Basin succession shows that lake levels are controlled by Milankovitch modulation of the monsoon systems of Pangaea (Olsen & Kent 1996; unfortunately, pollen and spores were too sporadically preserved to produce time-series data for that succession (Olsen & Kent 1996). We suggest that the high terrestrial palynomorph concentrations in St. Audrie's Bay are linked to an abrupt increase in seasonality in a semi-arid region (enhanced monsoonal activity) and more runoff during eccentricity maxima (Fig. 7c). Coincidence of occurrences of high total organic carbon (TOC) values and black shales with the high terrestrial palynomorph concentrations support this interpretation (Ruhl 2010). An intensified monsoon system may induce extension of the climate belts. Because moisture penetrated further into the hinterland, vegetation such as Cheirolepidiaceae and other gymnosperms could cover a larger area, and enhanced seasonal runoff would transport a relatively large amount of pollen, as is reflected in the low spore abundance during eccentricity maxima (Fig. 7c). A similar pattern, with enhanced seasonal runoff causing high terrestrial palynomorph concentrations, has been suggested for the Eiberg Basin in Austria (Bonis et al. 2010).

Sea-level change
A marine extinction scenario driven by sea-level fall and the loss of shallow-marine habitat space has been put forward (Hallam & Wignall 1999), although according to Hesselbo et al. (2004) 'it is unlikely that sea-level fall played a significant role in the T-J boundary extinctions in either a local or a global context'. A main question is to what extent sea-level changes influenced the observed palynomorph distribution patterns, as the marginal marine facies of the St. Audrie's Bay section may have been highly sensitive to such changes (Hesselbo et al. 2004). A transgression represented by the upper part of the Blue Anchor Formation (Hesselbo et al. 2004;Warrington et al. 2008) is confirmed by a peak abundance of the dinoflagellate cyst Heibergella sp. A, which was reported by Palliani & Buratti (2006), but any ecological preferences of this cyst are unknown. The transgressive systems tract represented by the Westbury Formation (Hesselbo et al. 2004) is reflected in the high abundance of dinoflagellate cysts in the aquatic palynomorph association; the sample from 1010 cm in particular has a very high concentration (and relative abundance) of the dinoflagellate cyst Rhaetogonyaulax rhaetica (Figs 2 and 5), a feature described by Orbell (1973) as a Rhaetogonyaulax population bulge possibly related to a decrease in the salinity. Dinoflagellate cysts (Dapcodinium, Beaumontella and Rhaetogonyaulax) have a very low abundance in the basal Jurassic deposits from the UK. Prasinophytes as well as acritarchs dominate the Blue Lias Formation, which is in agreement with the findings of Van de Schootbrugge et al. (2007), who suggested that seawater was warmer and had a lower salinity, and that conditions prone to stratification and the development of anoxia were created.
Sea-level changes could influence the relative amount of spores in the record. Spores are relatively heavy, and more difficult to transport than pollen (e.g. Dybkjaer 1991). Therefore, spores should predominate in more proximal marine facies, whereas the more floatable pollen (e.g. bisaccates) would predominate in more distal facies. At St. Audrie's Bay the spore spike interval that occurs in the Lilstock Formation is characterized by a sea-level lowstand ( Fig. 2; Hesselbo et al. 2004). The Cotham Member was deposited in an environment that was subject to subaerial exposure (Hesselbo et al. 2004) and spore peak 3 coincides with a level with desiccation cracks (Fig. 4). One of the most abundant spores in this peak is Porcellispora longdonensis, a large, heavy spore, which may imply that sealevel change influenced the presence of this peak. Furthermore, P. longdonensis and Calamospora tener are both produced by pioneer plants (bryophytes and horsetails, respectively), which could have invaded the newly emergent coastal areas. Another feature of the St. Audrie's Bay record is spore peak 4b, which consists almost completely of Heliosporites reissingeri. In the NW German Basin, the relatively common occurrence of Heliosporites in the brackish Early Rhaetian and marine Hettangian suggests that the parent plant inhabited a coastal environment, possibly in a marsh or mangrove-swamp (Lund 2003). Heliosporites was produced by lycophytes, possibly Selaginellaceae (Schulz 1967), and may represent a pioneer plant (J. H. A. Van Konijnenburg-Van Cittert, pers. comm.). Heliosporites is also reported from a Rhaeto-Liassic flora from a lacustrine environment in Airel, northern France (Muir & Van Konijnenburg-Van Cittert 1970). Almost the entire palynomorph assemblage from this locality consists of Classopollis harrissii sp. nov. (.99%). This assemblage is comparable with the pattern in St. Audrie's Bay. One could argue that this H. reissingeri peak was caused by a sea-level fall resulting in an extension of the coastal area to be invaded by these lycophytes. However, this peak occurs within a transgressive systems tract, suggesting that the increased H. reissingeri abundance reflects a vegetation change induced by climate and not by sea level. Although the lowest total palynomorph concentrations are present within the Lilstock Formation, most samples are characterized by a higher concentration of aquatic palynomorphs, which strengthens the idea that the Lilstock Formation was at least partially marine most of the time (Figs 2 and 4; Hounslow et al. 2004;Warrington et al. 2008). Furthermore, the spore peaks do not correspond to a lower abundance (both per cent and concentration) of aquatic palynomorphs. In all samples with spore peaks dinoflagellates are also present. Peak 2 even coincides with a very high abundance of the dinoflagellate cyst Rhaetogonyaulax rhaetica, which is thought to have been more adapted to open marine conditions (Courtinat & Piriou 2002;Kürschner et al. 2007). Apart from spore peak 3, there is no direct evidence that the increased amount of spores coincided with a lower sea level.
Terrestrial palynomorph associations from the Blue Lias Formation are dominated by C. meyeriana (75-100%). End-Triassic-earliest Jurassic Classopollis meyeriana-dominated palynofloras are documented from the continental Newark Basin, USA (Fowell et al. 1994;Olsen et al. 2002a), the Argana Basin, Morocco (Whiteside et al. 2007), and shallow marine or coastal successions in northern and eastern Spain (Barrón et al. 2006;Gómez et al. 2007). These occurrences imply a change to a warmer and/or more arid climate. However, a monotonous Classopollis assemblage is absent from the earliest Jurassic units of the Danish Sub-basin (Dybkjaer 1991), which suggests that the relatively drier climate was restricted to the interior of Pangaea. Cheirolepidiaceae are probably wind pollinators (Ziaja 2006) that produce a large amount of pollen per plant. Such small pollen are easily transported, so that sea-level change (transgression) could well have affected Classopollis dominance. Although part of the Cheirlepidiaceous group may have had a coastal habitat (Batten 1974;Watson 1988;Abbink 1998;Abbink et al. 2004), the interval of increased Classopollis meyeriana abundance in the Blue Lias Formation of St. Audrie's Bay is unlikely to be related to an extension of the coastal plain area as this part of the succession represents a transgression.
The sea-level curve constructed by Hesselbo et al. (2004) does not follow changes in the spore:pollen or terrestrial:aquatic palynomorph ratios (Fig. 2). Therefore, it is more likely that climate, rather than sea level, was the main influence on the palynomorph record. Another reason for regarding the spore peaks as climate-related is that it is difficult to explain a sealevel change every 20 ka, especially because the Late Triassic was a non-glacial interval and glacio-eustasy could not be expected (Satterley 1996). The larger cycles in palynomorph concentrations are also hard to explain by a sea-level change occurring every 100 ka. The changes in terrestrial and aquatic palynomorph concentrations are simultaneous (Fig. 2), which may imply that marine productivity was controlled by enhanced nutrient supply via river runoff.

The end-Triassic spore spike
It has been suggested that the palynological records from the Newark Basin and St. Audrie's Bay, with an upward increase in spore diversity and abundance, followed by low-diversity assemblages dominated by Classopollis, represent comparable microfloral turnovers (e.g. Hesselbo et al. 2002;Hounslow et al. 2004;Whiteside et al. 2007). However, correlation of these records is equivocal. The species composition of the spore spike (or 'fern spike') in the Newark Basin is different from that from St. Audrie's Bay (Table 1), and it is the only one documented in a continental succession (Olsen et al. 2002b), which could indicate local climate changes. The Newark Basin spore peak coincides with a 60% regional extinction of palynoflora and occurs in a coal-smectite clay (Fowell & Olsen 1993;Olsen et al. 2002b) deposited in a swamp environment where higher spore abun-dance would be expected. The fern spike in the Newark Basin has been linked to an impact, based on a small iridium anomaly and the occurrence of this spike prior to (c. 20 ka before) the extrusion of the Orange Mountain basalt (Olsen et al. 2002a). Whiteside et al. (2007) suggested that there is no relationship between the T-J extinction and the onset of Central Atlantic Magmatic Province volcanism. However, there is continuing discussion about the exact position of the T-J boundary in the eastern North America basins (Hounslow et al. 2004;Kozur & Weems 2005;Lucas & Tanner 2007). There are no agediagnostic palynomorphs or other fossils to prove that the extinction and replacement of the diverse P. densus microflora, of taxa with Norian to Rhaetian, or even longer, ranges (such as Enzonalasporites vigens, Carnisporites spiniger, Patinasporite densus, Vallasporites ignacii, Granuloperculatipollis rudis, Classopollis meyeriana and Classopollis classoides) by the lowdiversity C. meyeriana palynoflora occurred at the T-J boundary (Kozur & Weems 2005). Palynological assemblages from sedimentary rocks just above the North Mountain basalt in the Fundy Basin are dominated by bisaccates such as Lunatisporites rhaeticus and Alisporites parvus and appear to be Triassic in age, indicating that Central Atlantic Magmatic Province volcanism may have triggered the T-J environmental crisis (Cirilli et al. 2009). Although spore peaks from the Newark Basin and St. Audrie's Bay may be contemporaneous, they might not have a causal relationship. Fowell et al. (1994) suggested that the T-J boundary spore spike is analogous to that at the Cretaceous-Tertiary (K-T) boundary. However, in contrast to the T-J boundary, the spore spike just above the level of the extinction of Cretaceous pollen (65 Ma) has a global occurrence (Nichols & Johnson 2008). It occurs at 40 localities in four basins in the USA, two in Canada, one in Japan and two in New Zealand, and in most places consists of a single taxon such as Cyathidites or Laevigatosporites (Nichols & Johnson 2008, and references therein). The K-T boundary spore spike was linked to an extraterrestrial impact scenario. An impact site has been identified, and anomalous iridium concentrations and impactsourced shock-metamorphosed mineral grains occur in numerous K-T boundary sections. After the impact, ferns took temporary advantage of the absence of seed plants and dominated the landscape as pioneer communities (e.g. Tschudy et al. 1984). Fleming & Nichols (1990) formally defined the spike as a palynological assemblage composed of 70-100% fern spores of a single species occurring within an interval 0-15 cm above the K-T boundary. The reported end-Triassic spore spikes from the Newark Basin (Fowell & Olsen 1993;Fowell et al. 1994), St. Audrie's Bay, and the NW German Basin ( Van de Schootbrugge et al. 2009) all consist of various species (Table 1). The transitional palynomorph assemblages in St. Audrie's Bay (SAB2 and SAB3) even include four distinct peaks (Fig. 4) but with very low spore concentrations. These characteristics are inconsistent with a scenario of pioneer species invading bare lands after a mass extinction. Therefore, the end-Triassic spore spike is not comparable with that at the K-T boundary and was probably not caused by an impact. This is further indicated by the lack of definite evidence of an impact scenario, such as shocked quartz or an impact crater (Lucas & Tanner 2007;Simms 2007).
Gymnosperm forests in NW Europe were transiently replaced by ferns and fern-associated vegetation (Van de Schootbrugge et al. 2009). This has been interpreted as a pioneer assemblage commonly found in disturbed ecosystems, in this case suggested to be caused by global warming and the release of pollutants during Central Atlantic Magmatic Province flood basalt volcanism. It is important to note that these assemblages in Germany are derived from the so-called Triletes Beds, named after the regular occurrence of lycophytic megaspores that have not been transported over long distances. Hence the increase of fern spores could also be facies-induced. All the locations mentioned by Van de Schootbrugge et al. (2009) that are marked by fern proliferation were positioned at about the same palaeolatitude, suggesting that they may have been within the same humid climate belt. For example, a spore increase in the Tatra Mountains, Slovakia, has been interpreted as reflecting a sudden increase in humidity (Ruckwied & Götz 2009). Volcanic activity associated with changes in oceanic and atmospheric circulation patterns could regionally result in increasing precipitation and/or humidity. Ruckwied et al. (2008) also related an increase of trilete spores within the T-J boundary interval in a terrestrial coal-bearing series in Hungary to increasing humidity, rather than catastrophic events. A spore spike (mainly Concavisporites and Deltoidospora) in the marine Csővár section (Hungary) has been linked to Central Atlantic Magmatic Province volcanism (Götz et al. 2009) but the maximum spore abundance is only c. 35%. Van de Schootbrugge et al. (2009) also mentioned a volcanism-induced fern proliferation in northern Spain (Gómez et al. 2007), but this interpretation is debatable as the very high percentages of Classopollis were represented separately from the other palynomorphs by Bárron et al. (2006), who interpreted the fern increase as due to a short humid event in an arid desert-like palaeoenvironment at the beginning of the Jurassic. After this humid event, plant communities were reduced to a Cheirolepidiaceous vegetation with undergrowth containing scarce lycophytes and ferns (Bárron et al. 2006). It was suggested that peaks of relative abundance of hygrophytic plants might reflect the strong monsoon precipitation regime that dominated Pangaea during the Late Triassic (Gómez et al. 2007). According to Van de Schootbrugge et al. (2009), Polypodiisporites polymicroforatus dominates T-J boundary assemblages in Austria. However, although it occurs in the palynoflora of the end-Triassic Schattwald beds it forms only c. 10-20% of the terrestrial palynomorph assemblage, and occurs with a wide variety of other taxa (Bonis et al. 2009, pp. 13-16, fig. 4): a gradual proliferation of ferns in the Austrian sections is here interpreted as reflecting a change to a more humid climate. A fern spike has, according to Van de Schootbrugge et al. (2009), also been documented in continental successions in Greenland. However, the high relative proportion of fern taxa there comprises macrofossils (McElwain et al. 2007) and at present there is no published high-resolution palynological study from Greenland. A taphonomic artefact cannot be excluded in this instance, as this material was deposited in a coal swamp and 'it is not unusual for peat-forming vegetation to contain a high proportion of fern taxa' (McElwain et al. 2007). Limited quantitative information from North China suggests that a change from gymnosperm dominance to fern dominance can be recognized as far east as the Junggar Basin. There Late Triassic assemblages containing a wide variety of gymnosperm pollen, but typically without Classopollis, are succeeded by one with nearly 60% fern spores (Lu & Deng 2005).
We consider that linking increased spore abundance to volcanism and pollutants is premature and that the reality of a supraregional end-Triassic 'spore spike' has to await confirmation by further high-resolution palynological records. Possibly, local depositional environment and climate changes influenced the fern abundances to a greater degree than suggested by Van de Schootbrugge et al. (2009).

Concluding remarks
High-resolution palynological study of the St. Audrie's Bay section revealed that there are two transitional zones within each of the palynofloras recognized by Orbell (1973) and that these contain four pronounced end-Triassic spore peaks. The present study shows that there is no single unambiguous global end-Triassic spore spike, but that there is a more complex pattern of spore distribution. Therefore, one must be careful when using end-Triassic spore spikes as a correlation tool. Cyclic patterns are observed in the palynomorph records. It is unlikely that the spore peaks from the St. Audrie's Bay section would show a cyclic pattern if they were linked to fern proliferation after major mass extinction caused by an impact or volcanic activity. We suggest that the spore peaks in the St. Audrie's Bay section are related to precession-induced changes in monsoon strength and precipitation and/or humidity. This implies that the total duration of the spore peak interval was about 80-100 ka and that the initial carbon isotope excursion lasted for at least 20 ka.
We acknowledge funding from the 'High Potential' stimulation program of Utrecht University. We are grateful to S. Hesselbo for providing the samples. J. van Tongeren and N. Welters are acknowledged for their assistance in the laboratory. M. Hounslow is thanked for his guidance in the field. We thank M. Deenen, W. Krijgsman and H. van Konijnenburgvan Cittert for the useful discussions. We thank the editor and two anonymous reviewers for their constructive comments on the manuscript. We gratefully acknowledge the thoughtful comments by H. Visscher on an earlier version of the manuscript. This is publication no. 20100302 of teh Netherlands Research School of Sedimentary Geology.