The palynological records show that terrestrial vegetation of the extensive coastal areas and lowlands of the Danish Basin was remarkably stable during the middle Rhaetian (Figure 2). Variations in climate and sea level likely influenced vegetation changes on land as well as changes in depositional environment (Nielsen, 2003), contributing to abundance variations in the palynological records of the Gassum Formation (Figure 2). The pollen records show that cheirolepidiacean and taxodiacean/cupressacean conifers dominated the vegetation (Figure 2), occupying the upper canopy tier in drier and wetter environments, respectively (Tables 1 and 2; Figure 3A). Apart from these and Ricciisporites tuberculatus, most other typical Rhaetian tree pollen taxa were less common, but their parent plants were still important contributors to the upper canopy. The Rhaetian mire forests were most likely multi-storeyed and lush, with mid-canopy elements including very rare Erdtmannithecales, rare Bennettitales and Ginkgoales/Cycadales, and common Caytoniales (Tables 1 and 2; Figure 3A), similar in composition to macroplant assemblages described from Greenland, southern Sweden and Germany (McElwain et al., 2007; Pott and McLoughlin, 2011; Van Konijnenburg-Van Cittert et al., 2018). Understorey plants, including ferns, lycophytes, bryophytes and sphenophytes, were generally rare to common and many of these plants were probably adapted to living in the shade of the trees or even as epiphytes on the trees (Figure 3A).

Two phases of mass rarity are evident in the Stenlille pollen records, here referred to as MR1 and MR2, respectively (Figure 2). MR1 occurs within the lowermost part of the Fjerritslev Formation (Figure 2) and encompasses the upper part of a black shale and the lowermost part of the succeeding “grey siltstone” interval of the Fjerritslev Formation (Lindström et al., 2017b) (Figure 2). At MR1, upper canopy plants appear to have been particularly hard hit. Both the coastal, salinity tolerant and xerophytic cheirolepidiacean conifers, as well as the taxodiacean/cupressacean conifers that preferred wet habitats like mires and estuaries (Mussard et al., 1997), were severely reduced in abundances within a c. 2 m thick interval from the middle of the black shale in the Stenlille-1 record (Figure 2A; Table 1). Four different cheirolepidiacean pollen species exhibit rarity at this level commencing with Granuloperculatipollis rudis, followed in ascending order by Classopollis meyerianus, Geopollis zwolinskai and C. classoides (Figure 2). Rarity is also registered in Rhaetipollis germanicus, Monosulcites minimus, Ovalipollis ovalis and Perinopollenites elatoides, showing that other tree taxa were also affected (Figure 2A). The affinity of R. germanicus is unknown, but O. ovalis may have been produced by voltzialean conifers (Scheuring, 1970), and M. minimus by a bennettitalean parent plant (Pott et al., 2016). The Cheirolepidiacean, taxodiacean/cupressacean and voltzialean conifers had scale- or needle-like leaves that would make them more fire prone than e.g., bennettitaleans that had elongated entire-margined or segmented leaves (Belcher et al., 2010) (Table 1). Wildfires under high temperatures indicative of crown fires in fire-prone vegetation are known to have occurred in the mid-Rhaetian mires (Petersen and Lindström, 2012). Palynofacies analyses of the Stenlille-4 samples show some intervals in the Gassum Formation where increased amounts of black phytoclasts and wood remains indicate that more intense wildfires may have occurred at times (Figure 2B). Rhaetian bennettitaleans seem to have had an ecological preference for highly disturbed and deltaic habitats (Pott, 2014), and M. minimus was found to be common to abundant in the fire struck Rhaetian mires of the Danish Basin (Petersen and Lindström, 2012). Except for the Bennettitales that may have been insect pollinated (Pott, 2014), the tree taxa affected at MR1 are all considered to have been wind pollinated (Table 1). In the Stenlille-4 record MR1 appears to be extended from the uppermost part of the Gassum Formation to lower part of the “grey siltstone” interval (Figure 2B). Discrepancies between the upper part of MR1 in this record and Stenlille-1 likely relates to the fact that the base of the “grey siltstone” in Stenlille-4 contains clay clasts probably derived from the black shale (Figure 2). Apart from the probable reworking of black shale material into the lowermost “grey siltstone”, the overall pattern is similar in the two wells. Although neither the parent plants of Classopollis classoides, C. meyerianus, Perinopollenites elatoides nor Monosulcites minimus were extirpated in the end-Triassic extinction, the cheirolepidiacean conifers that produced G. zwolinkai and G. rudis, as well as the voltzialean conifer that shed O. ovalis and the unknown parent plant of R. germanicus, were all victims of the crisis.

The ecological disruption during MR1 resulted in a major reduction of the upper and mid-canopy tree elements, suggesting major deforestation in the area and a marked shift towards a more open landscape (Tables 1 and 2; Figure 3B). Marked and consistently increased levels of black phytoclasts and black wood remains, and significantly decreased amounts of spores and pollen compared to prior to MR1 may indicate that increased wildfire (Figure 2B). Both the vegetation density and volume were likely severely decreased during this floral crisis interval as indicated by low TOC values (Lindström et al., 2019), and by significantly lower amounts of spores and pollen during this interval (Figure 2B). The deforestation allowed some previously less prominent taxa to flourish during the deposition of the “grey siltstone” (Figures 1, 2). In particular the understorey spore-producing schizaeacaean (Polypodiisporites polymicroforatus spores) and dipterid and/or matoniaceous ferns (Van Konijnenburg-Van Cittert et al., 2020) (Deltoidospora spores; not shown on Figure 2) seem to have benefitted from the open landscape (Table 2; Figure 3B). These opportunistic ferns had a significantly faster life cycle than trees and could perhaps reproduce both sexually and through asexual spreading via rhizomes. These ferns flourished together with an unknown, possibly ruderal, gymnosperm (Ricciisporites tuberculatus pollen) (Kürschner et al., 2014) that also increased in abundance compared to in the Gassum Formation. A similar vegetation response has been documented across Europe and interpreted as reflecting widespread distribution of pioneering plant communities after deforestation (van de Schootbrugge et al., 2009; Lindström, 2016; Lindström et al., 2017b; Gravendyck et al., 2020). Interestingly, Monosulcites minimus is significantly decreased during the deforestation interval, i.e. the “grey siltstone” beds, despite the parent plants possible preference to disturbed habitats (Pott, 2014) (Figure 2). Some typical Rhaetian spore-pollen taxa, e.g., the lycophyte spores Limbosporites lundbladiae and Cingulizonates rhaeticus, and the probable conifer or seed fern pollen Lunatisporites rhaeticus, show none or only marginal changes in abundance during this interval (Figure 2). Another typical Rhaetian lycophyte spore, Semiretisporis gothae, even increased in abundance during the deposition of the “grey siltstone” beds.

In the uppermost part of the “grey siltstone” beds, a second mass rarity event, MR2, was instigated with the virtual disappearance of the low-abundance species S. gothae, L. rhaeticus, C. rhaeticus, and L. lundbladiae (Figures 2, 3C). Two of the opportunistic palynofloral elements in the “grey siltstone” beds, P. polymicroforatus and R. tuberculatus, also show marked abundance losses, reflecting that the collapse of the pioneering ecosystem occurred during MR2 (Figure 2). MR2 coincided with the onset and first negative peak of the Spelae CIE (Figure 2), which coincides with a sudden relative increase in the equisetalean spore Calamospora tener and in the freshwater algae Botryococcus braunii (CCM zone in (Lindström et al., 2017b). As equisetaleans preferred wet environments like riverbanks or lake shores, this signals an increase in freshwater input at that time. After this there was a return of, in particular, the taxodiacean/cupressacean (P. elatoides pollen), but also to a lesser extent cheirolepidiacean (C. meyerianus and C. classoides pollen) and bennettitalean (M. minimus pollen) trees (Figure 2). Although cupressacean conifers appear to have returned in large numbers signalling re-establishment of coastal mires, the drier coastal habitats that supported the cheirolepidiacean conifers appear to have been greatly diminished (Tables 1 and 2; Figure 3C). In general, the post-crisis vegetation appears less dense in the upper canopy while the mid-canopy tier seems almost eradicated, similar to macroplant fossil observations in East Greenland (McElwain et al., 2007). The post-crisis vegetation contained several rare or ghost taxa, the latter including e.g., G. rudis, O. ovalis, L. rhaeticus, P. polymicroforatus and R. tuberculatus, that were victims of the crisis (Figure 3C). Later, Pinuspollenites minimus, a previously rare pinacean conifer pollen increased in abundance during the post-crisis and recovery interval, probably inhabiting hinterland or drier areas (Lindström, 2016) (Figure 3C).

Diversity indices provide additional clues to the timing and severity of the environmental changes that led to the mass rarity phases. In accordance with the intermediate disturbance hypothesis (IDH) (Svensson et al., 2012), two intervals where species richness increased and peaked indicate that the ecosystem was subjected to disturbances of intermediate magnitude and/or frequency; namely intermediate disturbance 1 (ID1) and 2 (ID2) (Figure 4). At ID1 species richness increased and peaked just prior to and during the lower part of MR1 (Figure 4), indicating an intermediate level of disturbance that hindered competitive exclusion by supressing species that were previously dominant and allowing new species to colonize free space (Svensson et al., 2012). This is reflected by the initial increase in richness coinciding with a minimum in dominance, <0.1, and an increase in evenness to >0.8 (Figure 4). Just prior to MR1, while species richness was still high, evenness dropped continuously to the top of MR1 (Figure 4), indicating high level disturbances promoting dominance of a few disturbance specialists, namely P. polymicroforatus, R. tuberculatus and Deltoidospora spp. (Figure 2). The decline in evenness continued to the top of MR1 as species richness also declines, reflecting a subsequent loss in biodiversity (Figure 4).

At ID2 in the upper part of the “grey siltstone” beds, richness again reached maximum levels indicating intermediate disturbance (Figure 4). At the top of the beds, richness fell successively and reached a minimum during MR2 (Figure 4). This coincided with a fall in dominance and a rise in evenness, suggesting that the taxa that were ecologically dominant during the “grey siltstone” beds were now suffering, whereas other species were recovering or establishing. The diversity indices suggest that conditions improved after the Spelae CIE and the palynology suggests that some of the previously lost coastal habitats were restored, as primarily taxodiacean/cupressacean, but also to lesser extent cheirolepidiacean, conifers increase in abundance (Figures 1–3).

The magmatic activity in the Central Atlantic Magmatic Province may have affected the terrestrial ecosystem in multiple ways, following the flow chart in Figure 6. The successive abundance losses at MR1 were unprecedented during the Rhaetian of NW Europe. The preceding increase in richness, ID1, which signals intermediate levels of ecosystem stress, coincided with and immediately succeeded the Marshi CIE, believed to reflect added excess of light carbon to the carbon cycle as greenhouse gases (Hesselbo et al., 2002; Ruhl and Kürschner, 2011; Lindström et al., 2021), which may have included methane from magma-sediment interaction (Ruhl et al., 2011; Capriolo et al., 2021) (Figure 4). The local/regional effects of the subsequent global warming would likely have varied depending on the prevailing palaeoclimatic conditions at each site. The maximum values in species richness also coincided with increased Hg-loading and moderately increased occurrences of mutated fern spores (Lindström et al., 2019) interpreted to reflect volcanic emissions from CAMP volcanism (Thibodeau et al., 2016; Percival et al., 2017; Lindström et al., 2019; Lindström et al., 2021). Increased occurrences of mutated Classopollis pollen at a corresponding stratigraphical level in Germany (Gravendyck et al., 2020) further suggests that mutagenic stress also affected the cheirolepidiacean conifers. The mutagenesis could have been caused by Hg-toxicity as suggested for the ferns (Lindström et al., 2019), however, other heavy metal toxicities or increased UV-B radiation, as suggested for plant mutagenesis at the end-Permian event (Benca et al., 2018; Chu et al., 2021), cannot be excluded at this time (Figure 6).

The coastal and near-coastal lowland terrestrial ecosystems along the margins of the European continental sea and the northern Tethys margin were vulnerable to sea-level changes, but the vegetation appears to have remained relatively stable during the middle Rhaetian. The black shale at the base of the Fjerritslev Formation was deposited during the late Rhaetian transgression (marked by a maximum flooding surface, MFS7; Figure 2) (Nielsen, 2003; Lindström and Erlström, 2006) which in contrast to previous transgressive events was associated with massive abundance losses in trees growing in coastal habitats or near-coastal lowlands during MR1 (Figure 2). The fact that this mass rarity occurred just prior to, during and after the culmination of this late Rhaetian transgression indicates that the rate of sea level rise outpaced the organic and inorganic sediment supply in the coastal region, leading to habitat fragmentation of the terrestrial coastal and lowland ecosystems to such an extent that it instigated the mass rarity (Figure 6). The fast relative sea-level rise in combination with global warming (Ruhl et al., 2011), increased wildfire frequency (Petersen and Lindström, 2012) and heavy metal pollution (Lindström et al., 2019) likely exerted profound stress on the terrestrial ecosystem at that time (Figure 6).

The late Rhaetian transgression was succeeded by a prominent regression during which the “grey siltstone beds” were deposited (Figure 2). This major drop in sea-level in the late Rhaetian was part of a regression-transgression couplet that can be recognized over large parts of Europe (Hallam, 1997; Hallam and Wignall, 1999), but the mechanism behind this sea-level drop remains to be explained. Hallam (1997) favoured a tectono-eustatic mechanism other than sea-floor spreading, comparing it in pace and magnitude to glacio-eustacy. Hallam and Wignall (1999) suggested that the regression-transgression couplet was linked to the formation of the CAMP, and widespread occurrences of seismites in co-eval strata across Europe (Simms, 2003; Lindström et al., 2015) seem to support this. In the Danish Basin, this regression culminated with coastal deposition far into the basin, and in some marginal areas fluvial erosion and incision took place (Nielsen, 2003). As a consequence of this regression, a marked lowering of the base-level groundwater table would have had devastating effects on the remaining fragmented coastal mires and lowland ecosystems (Figure 6). Normally, sea-level changes affect the relative abundances of spores and pollen in such a way that wind-transported pollen increase in relative abundance compared to water-transported spores with increasing distance from the shore, i.e. the “Neves effect” (Chaloner and Muir, 1968). The fact that the opportunistic palynoflora dominated by P. polymicroforatus, Deltoidospora spp. and Ricciisporites tuberculatus, has been recovered from various terrestrial to marine settings (Lindström and Erlström, 2006; Lindström et al., 2017b) without major changes in the relative abundances, suggests that the floral shift was not due to the “Neves effect” (Chaloner and Muir, 1968), but that the deforestation on land was extensive (van de Schootbrugge et al., 2009).

On land, in northwest Scania in southern Sweden, the spore-pollen records indicate decreasing tree density from the middle to late Rhaetian, coincident with increased wildfire frequency but decreasing burning temperatures, where the latter may also reflect decreased tree density (Petersen and Lindström, 2012). In fluvial terrestrial strata, stratigraphically equivalent to the “grey siltstone” beds, perfectly zoned sphaerosiderite suggest strong seasonal climate variations with major fluctuations between high groundwater levels and drought (Weibel et al., 2016) (Figure 6). Fluctuations in groundwater levels could have been the result of extreme weather, with alternations between drought and excessive rain fall and flash floods during storms. It is also possible that changes in groundwater level was directly related to the on-going seismic activity (Weibel et al., 2016) (Figure 6). Both increased wildfire activity and deforestation would exacerbate sediment loads due to lack of vegetation stabilizing channel-margins and could potentially have hindered colonization of newly emerged surfaces by other than opportunistic plants. An abundance of reworked palynomorphs within the grey siltstone beds and equivalent strata in NW Europe, further suggest rapid ongoing soil erosion (van de Schootbrugge et al., 2020) (Figure 6).

Towards the top of this interval, increased species richness during ID2 (Figure 4) again indicate intermediate disturbance to the ecosystem. However, ID2 neither coincided with or immediately succeeded a negative δ¹³C_org excursion or Hg-loading (Lindström et al., 2019) (Figure 4), hence disturbance stressing the opportunistic ecosystem at this level must have been something else. Multiple records provide evidence of further intensification of storminess and wildfire frequency at this level, as evidenced by charcoal records (Belcher et al., 2010; Petersen and Lindström, 2012), and pyrolytic polycyclic aromatic hydrocarbons (PAHs) (van de Schootbrugge et al., 2009; Lindström et al., 2021). In addition, PAH formed by incomplete combustion of organic matter suggest that intrusive coking of organic-rich sediments in the CAMP area also occurred (van de Schootbrugge et al., 2009). At the same level (Lindström et al., 2021), cuticles from bennettitaleans and ginkgoaleans in Greenland exhibit lesions and distortions typical of sulfuric acid deposition and this was further intensified across the TJB (Steinthorsdottir et al., 2018), suggesting ongoing sulphur-rich volcanic emissions from the CAMP, in line with that some CAMP lavas contained especially high levels sulphur compared to other LIPs (Callegaro et al., 2014). Acid rain would have exacerbated the physiological responses that TJB plants had to ongoing heat and water stress, possibly further increasing the hydrological cycle (Steinthorsdottir et al., 2012), which could explain why the opportunistic plants adapted to drought began to decline in abundance. MR2 preceded the prominent Spelae CIE (Figures 2, 4). This negative excursion has been interpreted to reflect atmospheric increase in light carbon either through degassing of volcanic CO₂ or methane (Hesselbo et al., 2002; Ruhl et al., 2011). However, this was recently questioned by Fox et al. (2020) who instead suggested that the negative CIEs at the Spelae level were regional phenomena caused by fractionation by microbial mats under brackish water conditions. An increase in freshwater algae in the Stenlille record also seems to indicate enhanced runoff (Lindström et al., 2012), but because negative CIEs at the Spelae level are also present outside the European epicontinental sea, e.g. in Argentina (Ruhl et al., 2020), further studies are needed to resolve this issue. Regardless of which, the Spelae CIE did not instigate the MR2 mass rarity in land plants. Instead, living conditions for some of the plants that had suffered during MR1 (e.g. Classopollis, Perinopollenites elatoides) appear to have improved during the Spelae CIE, perhaps due to stabilizing climatic conditions. A second relative sea-level rise may have helped to restore some of the coastal and near-coastal lowland habitats that were lost during and after MR1 (Figures 2–5). Thus, at least in the northern hemisphere both the destruction and resurrection of coastal and near-coastal terrestrial ecosystems were tightly linked to relative sea-level changes.

Today, areas with long term stable climatic conditions have been found to contain higher numbers of rare plants, because of the reduced extinction risks (Enquist et al., 2019). However, because rare plants species are more susceptible to reductions in populations size, they are more likely to go extinct by orders of magnitude than more abundant species during times of rapid climate change (Enquist et al., 2019). The geographical and numerical reduction of plant species during the two end-Triassic mass rarity phases attest to the severe impact of this crisis on the vegetation in NW Europe. The causality behind the mass rarity amongst the plants was clearly complex, including both direct effects of the volcanic activity in the CAMP, as well as subsequent feedback mechanisms. The volcanic activity of the CAMP emitted multiple stressors, including greenhouse gases, SO₂, Hg-pollution, PAHs, and likely also halocarbons (Figure 6). The results emphasize that rapid sea-level changes, probably driven by tectono-eustatic activity in the CAMP area, severely affected the coastal and near-coastal lowland vegetation, to the extent that many species already stressed by climate change and volcanic pollution, may have had difficulties coping with fragmentation of habitats and changes groundwater levels. The mass rarity was profound as it affected both already rare as well as previously abundant species and must have had a cascading negative impact on the terrestrial ecosystem as a whole, altering the conditions for a range of organisms from the lowest to the highest trophic levels. The mass rarity in plants was supraregional across NW Europe (Table 2), although small populations of typical Rhaetian plant taxa lingered without recovery into the earliest Jurassic (early Hettangian) in some areas. Most of these ghost taxa are only known by their spores or pollen and have not been identified as macroplant fossils, thus they were rare already prior to the crisis. Their extirpation in the aftermath of the end-Triassic crisis attest to the vulnerability of plant species to reductions in abundance and geographic range, especially in coastal regions. This should resonate when considering present day anthropogenic ecosystem disturbances and ongoing and future climate change scenarios (IPCC, 2021) that, similar to the end-Triassic crisis include rapid sea-level rise (De Conto et al., 2021) in combination with increased frequency and intensity of climate extremes such as prolonged heatwaves and drought, wildfires, major storms and heavy precipitation.

109 sedimentary rock samples from Stenlille-1 and 87 ones from Stenlille-4 were processed using standard palynological methods at the Palynology Laboratory at GEUS. Approximately 20 g of each sample was crushed and treated with hydrochloric and hydrofluoric acids to remove carbonate and silicate minerals, respectively. Heavy minerals were removed from the residues using heavy liquid separation. The organic residues were mildly oxidized using nitric acid and filtered on 11 μm mesh filters. Strew slides were prepared after each step in the preparation after the hydrofluoric acid treatment, using glycerine gel as mounting medium. The palynomorph content of the samples were assessed by counting 300 specimens in one representative strew slide from each sample. Spore and pollen taxa abundances were normalized against the total spore-pollen count in each sample. It is important to remember that spore/pollen production in a plant is not necessarily related to the actual abundance of that plant in the ecosystem, and the parent plant of continuously rare spore/pollen taxa may still have played an important role in the ecosystem. In addition, plants that rely on insect or animal assistance with dispersal and pollination may have a more restrained pollen production compared to those that rely on wind for dispersal. In the late Triassic, prior to the diversification of the angiosperms, possible insect pollination has only been suggested for the Bennettitales (Labandeira et al., 2007).

Seventy-five samples from Stenlille-4 were analysed for palynofacies. Two-hundred particles or more were counted from each sample on the slide prepared after heavy liquid separation but prior to swirling. The palynodebris and palynomorphs were divided into ten different categories: Black opaque phytoclasts which include opaque usually rounded particles; black wood encompasses all black wood remains, both blade shaped, needle shaped and equidimensional ones; plant tissue includes all non-woody and non-cuticular plant remains; cuticles include translucent to semi-translucent tissue with cellular structures with or without stomata; Amorphous organic matter (AOM) encompasses degraded material either as structureless amorphous or structured amorphous matter; marine phytoplankton includes acritarchs, dinoflagellate cysts and prasinophytes, however, for graphic purposes unidentifiable palynomorphs were also included in this group but are listed separately in Supplementary Table S3; freshwater microalgae primarily include Botryococcus braunii and zygnematacean cysts; Spores also include rare fragments of clitellate cocoon fragments; and finally pollen. The counts are listed in Supplementary Table S3.

In palynology, loss of species abundance has often been used as an indicator of biotic crises without specifically referring this to mass rarity (Twitchett et al., 2001; Lindström and McLoughlin, 2007; van de Schootbrugge et al., 2009). The mass rarity of a taxon was determined as either: 1) a marked shift from consistently common to abundant abundance values to consistently or inconsistently rare occurrences, i.e. the last common occurrence (LCO), or 2) as a marked shift from consistently present in low abundances to inconsistently present in low or lower abundances, i.e. the last consistent occurrence (LCon). In fact, mass rarity is routinely used in palynostratigraphy as the last common occurrence (LCO) of a taxon, or for taxa of low abundance the last consistent occurrence (LCon) may also reflect mass rarity in the parent plant.

Diversity trends in the successions were analyzed using species richness, dominance (1-Simpson index) and Pielou’s evenness calculated using the software PAST4.05. Because many of the spore- and pollen-genera at the Triassic–Jurassic boundary are monospecific, species richness is only marginally higher than genus richness. Reworked taxa were excluded from the diversity analyses. The diversity patterns were assessed primarily according to the intermediate disturbance hypothesis (Connell, 1978; Svensson et al., 2012).