IMPACT OF THE LATE TRIASSIC MASS EXTINCTION ON FUNCTIONAL DIVERSITY AND COMPOSITION OF MARINE ECOSYSTEMS
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
[Palaeontology, Vol. 61, Part 1, 2018, pp. 133–148] IMPACT OF THE LATE TRIASSIC MASS EXTINCTION ON FUNCTIONAL DIVERSITY AND COMPOSITION OF MARINE ECOSYSTEMS by ALEXANDER M. DUNHILL 1 , WILLIAM J. FOSTER 2 , JAMES SCIBERRAS 3 and RICHARD J. TWITCHETT 4 1 School of Earth & Environment, University of Leeds, Leeds, LS2 9JT, UK; a.dunhill@leeds.ac.uk 2 Jackson School of Geosciences, University of Texas, Austin, TX 78712, USA 3 Department of Biology & Biochemistry, University of Bath, Claverton Down, BA2 7AY, UK 4 Department of Earth Sciences, Natural History Museum, London, SW7 5BD, UK Typescript received 23 June 2017; accepted in revised form 15 September 2017 Abstract: Mass extinctions have profoundly influenced the the extinction event was, however, highly selective against history of life, not only through the death of species but also some modes of life, in particular sessile suspension feeders. through changes in ecosystem function and structure. Impor- Although taxa with heavily calcified skeletons suffered higher tantly, these events allow us the opportunity to study ecologi- extinction than other taxa, lightly calcified taxa also appear to cal dynamics under levels of environmental stress for which have been selected against. The extinction appears to have there are no recent analogues. Here, we examine the impact invigorated the already ongoing faunal turnover associated and selectivity of the Late Triassic mass extinction event on with the Mesozoic Marine Revolution. The ecological effects the functional diversity and functional composition of the glo- of the Late Triassic mass extinction were preferentially felt in bal marine ecosystem, and test whether post-extinction com- the tropical latitudes, especially amongst reefs, and it took munities in the Early Jurassic represent a regime shift away until the Middle Jurassic for reef ecosystems to fully recover to from pre-extinction communities in the Late Triassic. Our pre-extinction levels. analyses show that, despite severe taxonomic losses, there is no unequivocal loss of global functional diversity associated with Key words: mass extinction, Triassic, Jurassic, functional the extinction. Even though no functional groups were lost, diversity, niche, regime shift. T H E Late Triassic mass extinction event (LTE), which to understanding mass extinction events (Barnosky et al. occurred ~201.6 million years ago (Blackburn et al. 2013), 2012; Aberhan & Kiessling 2015). The LTE has been under- is the second biggest biodiversity loss (Alroy 2010) and the studied in comparison with other major Phanerozoic biotic third biggest ecological crisis (McGhee et al. 2004) since crises (Twitchett 2006) and, despite the importance of the Cambrian. The proposed mechanism for the crisis was functional diversity changes or ecological regime shifts on CO2-induced environmental changes, including global ecosystem function, a palaeoecological perspective on long- warming (McElwain et al. 1999; Wignall 2001), sea-level term trends through deep-time, including the effects of changes (Hallam 1981), ocean anoxia (Hallam & Wignall mass extinctions, is still lacking (Villeger et al. 2011). Con- 2000; Jaraula et al. 2013) and ocean acidification (Haut- sequently, little is known about the ecological impact of the mann 2004; Hautmann et al. 2008a) as a result of the erup- LTE, aside from the preferential extinction of tropical taxa tion of the Central Atlantic Magmatic Large Igneous inhabiting shallow marine environments, particularly Province (CAMP) (Whiteside et al. 2010; Blackburn et al. hypercalcifiers, and the subsequent reef crisis in the Early 2013). This makes the LTE an attractive candidate for Jurassic (Kiessling & Aberhan 2007; Kiessling et al. 2007; drawing ecological comparisons with the current anthro- Kiessling & Simpson 2011). pogenically-driven biodiversity decline. Recognizing when Understanding the nature and patterns of extinction ecosystems reach a threshold in response to environmental selectivity during this event may help in better under- pressures, resulting in a change in ecosystem structure standing the cause(s) of the event, and/or may help pin- often characterized by a shift in dominance among organ- point sampling biases or deficiencies. Selective loss of reef isms with different modes of life (i.e. a regime shift) is key ecosystems is a common outcome of past biotic crises ©The Authors. doi: 10.1111/pala.12332 133 Palaeontology published by John Wiley & Sons Ltd on behalf of The Palaeontological Association. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
134 PALAEONTOLOGY, VOLUME 61
driven by CO2-induced environmental change (Fl€ ugel Database (PaleoDB) on 13 April 2016 (Clapham et al.
2002; Veron 2008; Kiessling & Simpson 2011; Foster & 2016; Dunhill et al. 2017). This database consists of
Twitchett 2014) and may be interpreted as a consequence 55 608 occurrences of 2615 genera compiled at stage
of their sensitivity to environmental change. Apart from level. Filtering protocols were used to exclude ichnotaxa,
selection against specific regions, ecosystems or habitats form taxa and any uncertain generic assignments (e.g.
(Foster & Twitchett, 2014), other studies have suggested aff., cf., ex gr., sensu lato, ‘[quotation marks]’, and ‘infor-
that traits such as skeletal composition, motility and feed- mal’). Although there is concern that some of the data
ing are also critical in explaining patterns of selectivity. within the PaleoDB is inaccurate in terms of taxonomic
Past extinction events associated with elevated CO2 have and stratigraphic assignment, a number of studies have
been shown to be selective against epifaunal, sessile, sus- shown that taxonomic errors in these datasets are ran-
pension feeders (Bush & Bambach 2011; Foster & Twitch- domly distributed and have only minimal effects on long-
ett 2014; Clapham 2017), and selective against heavily term diversification and extinction patterns (Miller 2000;
calcified organisms (Hautmann 2004; Knoll et al. 2007; Peters 2007; Wagner et al. 2007; Miller & Foote 2009).
Hautmann et al. 2008a; Clapham & Payne 2011; Kiessling Erroneous taxonomic and/or stratigraphic assignments
& Simpson 2011; Bush & Pruss 2013; Payne et al. 2016a) were identified and corrected as far as was possible. This
but as epifaunal, sessile suspension feeders tend to be process involved drawing on the expertise of the authors
heavily calcified, it is unclear which of those traits is most and consultation with experts in specific taxonomic
important. Also, apparent selectivity may simply be an groups where necessary (all cited in Acknowledgements
artefact of preservation or sampling biases that may exist below). As well as systematic information, from phylum
between different modes of life, at least regionally (Man- to generic level, information relating to palaeolongitude
der & Twitchett 2008). and palaeolatitude, depositional environment, and both
The main aim of this study is to understand the patterns chrono- and lithostratigraphy were downloaded. Strati-
of diversity change and extinction selectivity in marine graphic assignments were made to the stage level, and
ecosystems during the Triassic–Jurassic interval and, in par- Triassic and Jurassic occurrences falling outside of a Ladi-
ticular, in relation to the LTE. In order to investigate the nian to Aalenian range were omitted from the data set.
functional diversity of ecosystems and to study key traits Occurrences without stage-level designations were omit-
such as feeding, tiering and motility, the ecospace model of ted. However, occurrence data do not capture Lazarus
Bambach et al. (2007) was used to provide a quantitative taxa (i.e. an organism that disappears from the fossil
autecological analysis of all known Middle Triassic to Middle record, often for millions of years, before reappearing
Jurassic marine animal genera. This approach has been pre- unchanged; Flessa & Jablonski 1983) which can lead to
viously successfully applied to analysis of the Cambrian radi- the overestimation of extinction rates, particularly across
ation (Bambach et al. 2007), comparisons of Palaeozoic and major extinction events (Twitchett 2001). A second data-
modern ecosystems (Bush et al. 2007; Villeger et al. 2011; base was therefore constructed by ranging-through 2844
Knope et al. 2015) and studies of both the late Permian genera between first and last occurrences at substage level
(Dineen et al. 2014; Foster & Twitchett 2014) and end- derived from the PaleoDB and from Sepkoski et al.
Cretaceous (Aberhan & Kiessling 2015) mass extinction (2002). To avoid edge effects in the range database, pre-
events. The following individual hypotheses were tested: (1) Middle Triassic and post-Middle Jurassic occurrences
in common with the Late Permian extinction (Foster & were used for determining the stratigraphic ranges of gen-
Twitchett 2014), despite significant taxonomic losses, the era. This approach was used by Foster & Twitchett (2014)
LTE did not result in a reduction in global functional diver- and compensates for the out-of-date nature of much of
sity; (2) sessile suspension feeders were selected against, Sepkoski et al. (2002) such as truncation of generic ranges
compared to other modes of life across the LTE; (3) heavily in the Norian which are now known to extend into the
calcified organisms were selected against and suffered higher Rhaetian, whilst filling in the gaps of the incomplete
extinction rates than lightly calcified taxa; and (4) taxonomic PaleoDB. The occurrence-based PaleoDB also allows for
and functional diversity loss was greater in the tropics than the accounting of variation in sampling intensity which is
at higher latitudes. not possible with a range-through data set. The PaleoDB
bins data at the stage level whilst the ranged-through data
(Sepkoski et al. 2002) is binned at the substage level.
MATERIAL AND METHOD Therefore, taxonomic and functional richness data are
calculated from ranged-through data at the substage level
Marine animal database and occurrence data (raw and subsampled) at the stage
level. As it is only possible to use the occurrence data in
A database of all Middle Triassic to Middle Jurassic mar- the regional subset analyses, all of these are at stage-level
ine animal genera was downloaded from the Paleobiology only.DUNHILL ET AL.: IMPACT OF THE LATE TRIASSIC MASS EXTINCTION 135
Functional characterization of marine animal genera classified as photosymbiotic or predatory; this would
only change the mode of life code from 261 to 265 or
Modes of life for all marine genera within the database 266.
were inferred using data from previous publications, Some previous studies have also considered skeletal cal-
functional morphology, online taxonomic databases and cification to be an important trait that may confer selec-
data from extant relatives. Each genus was assigned to a tive advantage during extinction events (P€ ortner et al.
functional bin in the ecospace model of Bambach et al. 2005; Knoll et al. 2007; Helmuth 2009; Payne et al.
(2007) based on its tiering, motility and feeding habits 2016a). In order to test this, the fossil genera were also
(Tables 1, 2). Scleractinian coral genera were all assigned to one of three bins based on their calcification
recorded as having suspension-feeding modes of life, (i.e. light, moderate or heavy), following the classification
despite evidence suggesting that many may have been scheme of Knoll et al. (2007). We do not go as far as to
photosymbiotic (Stanley & Swart 1995; Frankowiak et al.
2016; Stanley & Swart 2016). Modern scleractinian corals
may employ a variety of feeding strategies, but most, TABLE 2. Mode of life assignments (Bambach et al. 2007) pre-
sent in Middle Triassic to Middle Jurassic marine environments.
including zooplankton microcarnivory, are consistent
with suspension-feeding (Bambach et al. 2007; Bush Numeric Mode of life defintion
et al. 2007). However, as the dominant organisms within (tiering; motility; feeding mechanism)
this mode of life (which suffer the greatest reduction in
115 Pelagic; freely, fast; predatory
generic richness and occurrences at the LTE) it would
141 Pelagic; facultative, attached; suspension
not alter the results of this analysis if corals were to be
241 Erect; facultative, attached; suspension
261 Erect; non-motile, attached; suspension
312 Surficial; freely, fast; surface deposit
TABLE 1. Mode of life categories for tiering, motility, and
315 Surficial; freely, fast; predatory
feeding following Bambach et al. (2007).
321 Surficial; freely, slow; suspension
Ecologic Description 322 Surficial; freely, slow; surface deposit
category 324 Surficial; freely, slow; grazing
325 Surficial; freely, slow; predatory
Tiering 331 Surficial; facultative, unattached; suspension
1 Pelagic: in the water column 341 Surficial; facultative, attached; suspension
2 Erect: benthic, extending into the water mass 351 Surficial; non-motile, unattached; suspension
3 Surficial: benthic, not extending significantly into 361 Surficial; non-motile, attached; suspension
the water column 415 Semi-infaunal; freely, fast; predatory
4 Semi-infaunal: partly infaunal, partly exposed 421 Semi-infaunal; freely, slow; suspension
5 Shallow infaunal: living in the top ~5 cm of the 422 Semi-infaunal; freely, slow; surface deposit
sediment 425 Semi-infaunal; freely, slow; predatory
6 Deep infaunal: living more than ~5 cm deep in the 441 Semi-infaunal; facultative, attached; suspension
sediment 442 Semi-infaunal; facultative, attached; surface deposit
Motility 451 Semi-infaunal; non-motile, unattached; suspension
1 Freely, fast: regularly moving, unencumbered 456 Semi-infaunal; non-motile, unattached; other
2 Freely, slow: as above, but strong bond to substrate 461 Semi-infaunal; non-motile, attached; suspension
3 Facultative, unattached: moving only when necessary, 521 Shallow infaunal; freely, slow; suspension
free-lying 522 Shallow infaunal; freely, slow; surface deposit
4 Facultative, attached: moving only when necessary, 523 Shallow infaunal; freely, slow; mining
attached 525 Shallow infaunal; freely, slow; predatory
5 Non-motile, unattached: not capable of moving, 526 Shallow infaunal; freely, slow; other
free-lying 531 Shallow infaunal; facultative, unattached; suspension
6 Non-motile, attached: not capable of moving, 532 Shallow infaunal; facultative, unattached; surface
attached deposit
Feeding 533 Shallow infaunal; facultative, unattached; mining
1 Suspension: capturing food particles from the water 535 Shallow infaunal; facultative, unattached; predatory
2 Surface deposit: capturing loose particles from a 541 Shallow infaunal; facultative, attached; suspension
substrate 546 Shallow infaunal; facultative, attached; other
3 Mining: recovering buried food 621 Deep infaunal; freely, slow; suspension
4 Grazing: scraping or nibbling food from a substrate 631 Deep infaunal; facultative, unattached; suspension
5 Predatory: capturing prey capable of resistance 641 Deep infaunal; facultative, attached; suspension
6 Other: e.g. photo- or chemosymbiosis, parasites 646 Deep infaunal; facultative, attached; other136 PALAEONTOLOGY, VOLUME 61
consider skeletal mineralogy, however, previous results Statistical analyses
from Kiessling et al. (2007) found no overall selectivity
associated with skeletal mineralogy. To account for biases brought about by uneven sampling
Where sufficient geological information was available, across space and through time, we applied a subsampling
fossil occurrences were assigned to one of three deposi- regime to standardize both taxonomic and functional
tional settings: inshore, offshore or reef (Table 3). Occur- diversity on the basis of the number of fossil occurrences.
rences with insufficient environmental data were not used Subsampling quotas were set at n = 2000 occurrences per
in the depositional settings analyses. All occurrences were bin for global analyses, and at n = 350 for the smaller-
assigned to 30° palaeolatitudinal bands, irrespective of scale analyses pertaining to palaeolatitude, depositional
hemispheric orientation: polar (> 60°N or S), temperate setting and oceanic basin. Any time bins that did not
(between 30 and 60°N or S) or tropical (< 30°N or S). reach the subsampling quota were omitted from the anal-
All occurrences were, using palaeolatitude and palaeolon- yses. In each case, subsampling was carried out across
gitude, plotted on to Triassic–Jurassic plate-tectonic 1000 iterations. We do not use shareholder quorum sub-
basemaps (Scotese 2010) to identify which palaeo-oceano- sampling (SQS) (Alroy 2010), which has been proven to
graphic basin they were derived from (i.e. Tethys, Pantha- avoid the problems rarefaction has with dampening true
lassa, Boreal) (Dunhill et al. 2017, fig. S1). differences in diversity, as our data show no trend in
alpha diversity (i.e. number of taxa per collection)
through time or across the Triassic–Jurassic boundary.
TABLE 3. Categorization of depositional environments Dissimilarity in ecological community structure between
recorded in the PaleoDB following Kiessling et al. (2007). time bins was tested for using Bray–Curtis dissimilarity
(Bray & Curtis 1957) tests on subsampled occurrence data
PaleoDB classification Category from this study to detect changes in relative abundances of functional
Coastal indet. Inshore groups and significance testing was achieved by randomly
Delta plain Inshore assigning fossil occurrences to pre- and post-extinction
Deltaic indet. Inshore time bins across 1000 iterations and performing a Manly
Estuary/bay Inshore test (Manly 1997) to detect whether the observed dissimi-
Foreshore Inshore larity is significantly different to the distribution of ran-
Lagoonal Inshore dom occurrences. The simulations were achieved by
Lagoonal/restricted shallow subtidal Inshore randomly assigning Rhaetian and Hettangian occurrences
Marginal marine indet. Inshore to pre- and post-extinction assemblages and calculating
Open shallow subtidal Inshore
Bray–Curtis dissimilarity across the boundary.
Paralic indet. Inshore
Peritidal Inshore
Prodelta Inshore
Sand shoal Inshore RESULTS
Shallow subtidal indet. Inshore
Shoreface Inshore Global changes in functional diversity
Transition zone/lower shoreface Inshore
Basinal (carbonate) Offshore Global genus-level extinction across the LTE is estimated
Basinal (siliciclastic) Offshore at between 46% (using ranged-through data) and 72–73%
Deep subtidal (indet.) Offshore (raw and subsampled occurrence) (Fig. 1A). There are
Deep subtidal ramp Offshore considerable differences between the ranged-through and
Deep subtidal shelf Offshore
occurrence data throughout the results, with the former
Deep-water indet. Offshore
providing lower extinction estimates. This is unsurprising
Offshore Offshore
Offshore indet. Offshore as the incomplete nature of the fossil record leads to
Offshore ramp Offshore inflated levels of extinction in occurrence data due to dis-
Offshore shelf Offshore appearances brought about by Lazarus taxa. The same
Slope Offshore difference between the datasets is evident in functional
Submarine fan Offshore diversity loss too: seven modes of life apparently disap-
Basin reef Reef peared across the LTE according to the occurrence data,
Intrashelf/intraplatform reef Reef whereas five of those modes of life are recorded as being
Perireef or subreef Reef present by the range-through data (Fig. 1B). A maximum
Platform/shelf-margin reef Reef of two modes of life, therefore, disappeared across the
Reef, buildup or bioherm Reef
Triassic–Jurassic (T–J) boundary: pelagic, facultatively
Slope/ramp reef Reef
mobile, suspension feeders and semi-infaunal, unattachedDUNHILL ET AL.: IMPACT OF THE LATE TRIASSIC MASS EXTINCTION 137
non-motile, ‘other’ feeders, associated with the extinction between the Carnian–Norian (Tuvalian–Lacian) and across
of pelagic crinoids and megalodontid bivalves, respec- the Pliensbachian–Toarcian boundary (Fig. 1B). Neither of
tively. these reductions in functional diversity are reflected in the
In the immediate aftermath of a mass extinction, it is generic richness curve, but approximately correspond to
hypothesized that there would be a radiation in func- known lesser extinction events: the Carnian Pluvial Event
tional diversity as taxa evolve to fill vacant ecospace left (CPE) in the mid-Carnian (Dal Corso et al. 2015) and the
by the high levels of extinction (Bambach et al. 2007). early Toarcian Ocean Anoxic Event (OAE) (Little & Benton
Despite this, only one new mode of life originated in the 1995). The early Toarcian OAE is only detected in the
aftermath of the LTE (Fig. 1B): semi-infaunal, slow-mov- occurrence (raw and subsampled) data and not in the ran-
ing, deposit-feeding gastropods of the family Nerinellidae. ged-through data, suggesting that extinctions may have
However, a greater increase in ecospace occupation been regional with functional groups rapidly recolonizing
apparently preceded the LTE, in the Rhaetian, with the affected environments in the Middle Jurassic.
occupation or reoccupation of three modes of life: semi-
infaunal, slow-moving, predators; shallow infaunal, slow-
moving, ‘other’ feeders; shallow infaunal, facultatively Regional changes in functional diversity
mobile attached, ‘other’ feeders (Fig. 1B).
Similar losses of global functional diversity that are Based on subsampled data, the LTE records the biggest
recorded in the aftermath of the LTE are also recorded declines in both generic and functional richness in
FIG. 1. Global diversity from the A
Middle Triassic to the Middle Juras- 1200
sic. A, ranged, raw sampledand sub-
sampled generic richness. B, ranged, 1000
raw sampled and subsampled func-
Generic richness
tional diversity. Alternate white and 800
grey bars represent chronostrati-
graphic stages (from left to right):
600 ●
L, Ladinian; C, Carnian; N, Norian; ●
R, Rhaetian; H, Hettangian; S, Sine- ●
murian; P, Pliensbachian; T, Toar- 400 ● ●
● ● ●
cian; A, Aalenian. ●
200
0
B
40
Functional richness
30 ● ●
● ● ●
● ●
●
20
10 raw occurrence
● n = 2000
ranged-through
0
L C N RH S P T A
Mid Late Early Mid
Triassic Jurassic
240 230 220 210 200 190 180 170
Age (Ma)138 PALAEONTOLOGY, VOLUME 61
tropical latitudes (Fig. 2A; Dunhill et al. 2017, fig. S2). Hettangian. After subsampling, functional richness shows
Erect, facultatively mobile, suspension feeders disappeared a much more pronounced drop from the Pliensbachian
from the tropics and did not return until the Middle to the Toarcian, suggesting that although the LTE pro-
Jurassic. A number of mostly suspension-feeding, surficial foundly affected tropical and polar latitudes, the Toarcian
and infaunal modes of life also vacated the tropics, but OAE extinction was a more important event in the mid-
returned in the Sinemurian–Pliensbachian. Tropical func- latitudes. The drop in functional richness across the
tional diversity returned to pre-extinction levels by the Pliensbachian–Toarcian boundary is associated with the
Sinemurian–Pliensbachian, but generic richness remained temporary loss of ten infaunal modes of life, of which
much lower than it was prior to the extinction, in the seven returned in the Aalenian.
Carnian–Norian. Similar patterns are recorded in polar The effects of the LTE are much more apparent in Pan-
latitudes, but both generic and functional richness took thalassa than in the Tethys or Boreal oceans (Fig. 2B).
longer to recover and remained low until the Aalenian Pelagic suspension feeders apparently disappeared earlier
(Fig. 2A; Dunhill et al. 2017, fig. S2). in Panthalassa, at the end of the Carnian. Facultatively
In contrast, the mid-latitudes record a markedly differ- mobile, erect suspension feeders, all of which are crinoids,
ent pattern, with only a slight decrease in generic and are absent from Panthalassa during the Hettangian. A
functional richness across the T–J boundary (Fig. 2A; number of epifaunal and infaunal suspension feeding
Dunhill et al. 2017, fig. S2). Although pelagic suspension- guilds are also absent until the Sinemurian or Pliens-
feeders and a single guild of deep-infaunal suspension- bachian. Panthalassan generic and functional richness
feeders disappeared, most other guilds are present in the recovered to pre-extinction levels by the Sinemurian,
A 20 B C
● 25
15 ●
● ●
Functional richness
●
● ●
15 20 ●
● ●
10 ●
15
10
10
5
5
5 raw occurrence
Polar (>60ºNS) Boreal Ocean Inshore ● n = 350
0 0 0
25
25
30 ●
Functional richness
20 ●
● ●
● ●
20 ●
● ● ● ●
● ● ● ● ●
● ● ●
20 ● ● 15 ● ●
● ●
15
●
10 10
10
5 5
0 Mid (30-60ºNS) 0 Panthalassa 0 Offshore
30 30 20
Functional richness
●
● ● ●
20 ●
● ● 15 ● ●
● 20 ● ● ●
● ● ● ●
● ●
10
●
10 10
5
0 Tropical (< 30ºNS) 0 Tethys Reefs
0
Mid Late Early Mid Mid Late Early Mid Mid Late Early Mid
Triassic Jurassic Triassic Jurassic Triassic Jurassic
240 230 220 210 200 190 180 170 240 230 220 210 200 190 180 170 240 230 220 210 200 190 180 170
Age (Ma) Age (Ma) Age (Ma)
FIG. 2. Regional functional diversity from the Middle Triassic to the Middle Jurassic. A, subdivided by latitude (polar, mid-, tropi-
cal). B, subdivided by palaeo-oceanic basin (Boreal Ocean, Panthalassa, Tethys). C, subdivided by depositional environment (inshore,
offshore, reef). Gaps in the subsampled time series represent time bins where the subsampling quotas were not met. Chronostrati-
graphic scale is the same as Figure 1.DUNHILL ET AL.: IMPACT OF THE LATE TRIASSIC MASS EXTINCTION 139 although functional richness decreased once more fig. S2). Facultatively motile, erect suspension feeders (i.e. through the Pliensbachian and Toarcian before recovering crinoids) are present in offshore environments during the again in the Aalenian (Fig. 2B; Dunhill et al. 2017, fig. Hettangian, suggesting that those settings remained hos- S2). The effects of the LTE appear less pronounced in the pitable to this mode of life whilst they were excluded Tethys Ocean: the slight drop in raw generic and func- from shallower ecosystems. There is a drop in functional tional richness vanishes entirely after subsampling richness, although not in generic richness, between the (Fig. 2B; Dunhill et al. 2017, fig. S2). Although generic Pliensbachian and Toarcian in offshore environments richness was higher in the Late Triassic than the Early (Fig. 2C; Dunhill et al. 2017, fig. S2), with the loss of Jurassic, the decrease appears in the Norian–Rhaetian many shallow and deep-infaunal guilds. transition (Dunhill et al. 2017, fig. S2), rather than across the T–J boundary, and functional richness appears to be stable across the entire study interval. However, pelagic Changes in ecological structure suspension feeders disappear at the LTE along with the temporary absence of some semi-infaunal and deep infau- There is a significant reduction in within-mode-of-life nal guilds until the Sinemurian and Pliensbachian. The richness from the Rhaetian to the lower Hettangian (Wil- Toarcian OAE is characterized by the temporary loss of coxon paired test: V = 293, p < 0.001) (Fig. 3A), with 19 some infaunal guilds, which all returned in the Aalenian. modes of life, including the most taxonomically diverse We see no evidence for a pronounced LTE or early Toar- groups, suffering a significant reduction in richness. cian event in the Boreal Ocean (Fig. 2B), although the Attached, erect, suspension feeders suffer a 50% reduction data are too sparse to carry out meaningful subsampling. in their generic richness, whilst pelagic, fast-moving, Reefs suffered greater losses than other depositional set- predators and surficial, attached suspension feeders suffer tings during the LTE (Fig. 2C). The generic and func- a 30–40% reduction in generic richness. No change in tional richness of reef environments plummeted across generic richness was recorded for 11 modes of life, and the T–J boundary, with the loss of over 90% of genera increases in richness were recorded in 4 modes of life. (Fig. 2C; Dunhill et al. 2017, fig. S2). With the exception There is no significant reduction in the occurrence (i.e. of benthic suspension feeders, represented by one bivalve, number of occurrences within PaleoDB collections) of one echinoid and seven coral genera, all functional guilds each mode of life across the T–J boundary (Wilcoxon are absent from Hettangian reef ecosystems. All known paired test: V = 256, p = 0.89) (Fig. 3B). However, there Hettangian reefs are from mid-palaeolatitudes. The com- is a reduction in the occurrence of attached, surficial sus- plete absence of tropical reefs is not due to sampling fail- pension feeders of over 25% and attached, erect, suspen- ure as there is a large sample size of Hettangian tropical sion feeders suffer a decline of around 75%. There are occurrences in general. Reef ecosystems also record low reductions in the occurrence of many other suspension functional diversity in the Toarcian (Fig. 2C). This feeding groups whilst the number of occurrences of decline in the Toarcian is similar to the LTE, with the motile, surficial grazers, deposit feeders, and predators temporary exclusion of most functional guilds apart from increase. There is also a dramatic increase in the number certain surficial suspension-feeding coral and bivalve gen- of occurrences of fast-moving, pelagic carnivores, which era. However, in contrast to the Hettangian, all Toarcian is unexpected given the reduction in generic richness reefs are from the tropics, with none recorded from the recorded within this group. mid-latitudes despite a large sample size of mid-latitude Despite the significant losses to generic richness across Toarcian occurrences in general. all modes of life, the relative richness amongst modes of Inshore environments record a drop in raw generic and life does not deviate a great deal across the T–J boundary functional richness across the T–J boundary, although this (Fig. 4A). Both erect and surficial, stationary, attached is not retained after subsampling (Fig. 2C; Dunhill et al. suspension feeders suffered a reduction in proportional 2017, fig. S2). The LTE appears to driven the loss of some richness across the LTE and remained at lower relative epifaunal and infaunal groups, which returned in the richness throughout the subsequent Early Jurassic. In con- Sinemurian, but pelagic and facultatively mobile suspen- trast, infaunal groups and motile surficial modes of life sion feeding crinoids disappeared earlier. Neither of these show increases in proportional richness whilst pelagic functional guilds returned to inshore environments until predators suffer a reduction in proportional richness the Middle Jurassic, suggesting that Early Jurassic shallow across the extinction but recover to pre-extinction levels marine conditions remained inhospitable to some groups by the late Hettangian (Fig. 4A). Long-term proportional of motile, suspension feeding crinoids. occurrence patterns were, as expected, much more vari- Offshore environments record modest drops in generic able than the smoother trends of proportional richness, and functional richness across the T–J boundary, with the with the LTE appearing more pronounced (Fig. 4B). Sur- loss of a few infaunal guilds (Fig. 2C; Dunhill et al. 2017, ficial and erect, non-motile, suspension feeders suffered a
140 PALAEONTOLOGY, VOLUME 61
A
200 pre-extinction
115 415
post-extinction
36
Generic richness
1
150 526
261
533
100
50
0
331
341
351
361
415
421
422
425
441
442
451
456
461
521
522
523
525
526
531
532
533
535
541
546
621
631
641
646
115
141
241
261
312
315
321
322
324
325
Mode of life
B
531 115
600
541
36
1
Occurrences
261 324
400
200
0
331
341
351
361
415
421
422
425
441
442
451
456
461
521
522
523
525
526
531
532
533
535
541
546
621
631
641
646
115
141
241
261
312
315
321
322
324
325
Mode of life
FIG. 3. A, pre-extinction (Rhaetian) and post-extinction (lower Hettangain) within-mode-of-life generic richness based on ranged-
through richness data. B, pre-extinction (Rhaetian) and post-extinction (Hettangian) within-mode-of-life occurrence data based on
subsampling to n = 2000. Silhouettes represent examples of modes of life that increased/decreased in richness/occurrence across the
extinction event. See Tables 1 and 2 for mode of life definitions.
reduction in proportional richness whereas surficial, slow- Global ecological structure of the Rhaetian is more dis-
moving grazers, deposit feeders and predators showed an similar to that of the Hettangian than would be expected
increase. A large increase is also recorded in pelagic, fast- by chance (Bray–Curtis dissimilarity value of 0.3,
moving predators. Surficial, non-motile suspension feed- p < 0.001), and the change recorded across the LTE is
ers recovered quickly, reaching their pre-extinction level greater than any other time period, bar the Toarcian–Aale-
of proportional occurrence by the Sinemurian, but erect, nian (Fig. 5A and see Table 4 for Bray–Curtis scores).
non-motile suspension feeders (a guild dominated by However, these dissimilarity values are only slightly higher
scleractinian corals and crinoids) remained depressed in than those recorded between other time bins (Fig. 5A).
terms of number of occurrences with only a slight hint of There is a clear dip in dissimilarity between the Hettangian
recovery into the Aalenian. Another change in propor- and Sinemurian, before an increase in dissimilarity across
tional occurrence occurs between the Pliensbachian and the Pliensbachian–Toarcian boundary, and then a further
the Toarcian, which is coincident with the biotic crisis at increase in dissimilarity into the Middle Jurassic, possibly
that time. signifying the recovery of reef ecosystems (Fig. 5A). TheDUNHILL ET AL.: IMPACT OF THE LATE TRIASSIC MASS EXTINCTION 141
FIG. 4. Proportional richness and A
occurrence of modes of life from 1.0 115
141
the Middle Triassic to the Middle
241
Jurassic. A, proportional generic
261
richness of modes of life, based on 0.8 311
ranged-through data. B, propor- 312
Proportional richness
tional occurrence of modes of life 315
based on occurrence data subsam- 321
pled to n = 2000 occurrences. See 0.6 322
Tables 1 and 2 for mode of life defi- 324
nitions. 325
331
0.4 336
341
351
361
0.2 415
421
422
424
● 425
0.0
431
B 1.0 441
442
451
456
0.8 461
Proportional occurrences
513
521
522
0.6 523
525
526
531
0.4 532
533
535
541
0.2 546
561
621
631
0.0 636
Mid Late Early Mid 641
Triassic Jurassic 646
230 220 210 200 190 180
Age (Ma)
Rhaetian is more ecologically similar to the Sinemurian calcified genera (Fig. 6A). This result is also reflected in
than it is to the Hettangian, suggesting the onset of recov- extinction magnitude, where heavily calcified genera suffer
ery through the Hettangian (Fig. 5A). However, the Pliens- the highest level of extinction, closely followed by light
bachian and Toarcian are less similar to the Rhaetian than calcifiers, with much lower extinction among moderately
to the Hettangian or the Sinemurian. The functional com- calcified genera (Fig. 6C). Analyses were also carried out
position of the Aalenian is more similar to the Rhaetian on just the benthic taxa, to test whether the erratic
than any of the Early Jurassic stages (Fig. 5A). extinction rates of pelagic organisms, such as ammonoids,
Skeletal calcification appears to have had an influence may have influenced these results, but no difference was
on richness and proportional extinction across the T–J recorded. Benthic taxa record very similar results, in
boundary (Fig. 6). Moderately calcified genera record a terms of both generic richness (Fig. 6B) and extinction
small net increase in relative richness, whereas the greatest magnitude (Fig. 6D). However, in the benthic-only analy-
loss of richness is recorded by the lightly and heavily ses, there is a slightly larger difference between the142 PALAEONTOLOGY, VOLUME 61
A B
0.5
OAE
LTE
● sequential BC 150
Bray–Curtis dissimilarity
BC compared to
0.4 Rhaetian
100
Frequency
0.3 ● ●
●
● ●
●
0.2 ●
● 50
0.1 reef recovery
level bottom recovery
0.0 pelagic recovery 0
Mid Late Early Mid
Triassic Jurassic 0.0 0.1 0.2 0.3
240 230 220 210 200 190 180 170 Bray–Curtis dissimilarity
Age (Ma)
FIG. 5. Bray–Curtis dissimilarity analyses of global faunal composition from the Middle Triassic to the Middle Jurassic. A, Bray–Cur-
tis dissimilarity values (BC) amongst sequential time bins (black); each point represents the dissimilarity between the time bin and the
previous time bin and Bray–Curtis dissimilarity values between the Rhaetian and Jurassic time bins; all Bray–Curtis analyses were per-
formed on occurrence data subsampled to n = 2000. B, Manly test for significance of Bray–Curtis dissimilarity between observed Rhae-
tian and Hettangian occurrences (dotted line) (p 0.001) as compared to randomized occurrences across 1000 iterations
(histogram).
TABLE 4. Bray–Curtis dissimilarity analyses. very little evidence for significant reductions in global
Stage Sequential Dissimilarity to Rhaetian
functional diversity. Global functional stability appears to
be a common factor among mass extinction events, hav-
Raw n = 2000 Raw n = 2000 ing been previously recorded for the Late Triassic (Kies-
sling et al. 2007), late Permian (Foster & Twitchett 2014)
Ladinian
and the end-Cretaceous (Aberhan & Kiessling 2015). The
Carnian 0.37 0.25
Norian 0.25 0.25 LTE fits the ‘skeleton crew hypothesis’ of Foster &
Rhaetian 0.28 0.27 Twitchett (2014), in which each ecological guild survives,
Hettangian 0.38 0.3 0.38 0.3 but is ‘manned’ by only a few individual taxa during the
Sinemurian 0.29 0.2 0.27 0.27 recovery period.
Pliensbachian 0.54 0.15 0.57 0.31 The two modes of life that are apparently lost across
Toarcian 0.22 0.22 0.55 0.44 the extinction interval are both equivocal because of
Aalenian 0.53 0.3 0.25 0.25 uncertain taxonomy and autecology. The pelagic, faculta-
tively mobile, suspension feeders are all represented by
Sequential analyses represent mean value of dissimilarity across
1000 iterations from previous time bin to labelled time bin. Dis-
one order of crinoids, the Roveacrinida, which originated
similarity to Rhaetian analyses represent dissimilarity from Rhae- in the Ladinian (Hess et al. 2016) and range-through to
tian to labelled time bin. the Rhaetian when they apparently became extinct at the
LTE. However, different taxa assigned to the Roveacrinida
extinction magnitude of heavy and light calcifying genera, appear again in the Bajocian (Middle Jurassic) and range-
suggesting a slightly greater selection against heavy calci- through to the Miocene (Gorzelak et al. 2011). As this
fiers (Fig. 6D). The only time interval that appears to be mode of life is re-occupied by the same order of organ-
exclusively selective against heavily calcifying organisms is isms in the Middle Jurassic, we consider it unlikely that
across the Pliensbachian–Toarcian boundary (Fig. 6D). this mode of life was completely vacated at the LTE, and
instead infer that the Roveacrinida persisted in abun-
dances that were too low to be recorded in the fossil
DISCUSSION record, or in areas that have yet to be sampled for Early
Jurassic fossils, as observed for Cenozoic occurrences of
Although the LTE caused significant reductions in generic this group (Gorzelak et al. 2011). Alternatively, if the
richness within almost all classes of animal life, we find Roveacrinida is in fact paraphyletic, as has been suggestedDUNHILL ET AL.: IMPACT OF THE LATE TRIASSIC MASS EXTINCTION 143
A B
1.0 1.0
0.8 0.8
Proportional richness
Proportional richness
0.6 0.6
0.4 0.4
0.2 light 0.2 light
moderate moderate
heavy heavy
0.0 0.0
Mid Late Early Mid Mid Late Early Mid
Triassic Jurassic Triassic Jurassic
240 220 200 180 240 220 200 180
Age (Ma) Age (Ma)
C D
0.6 light 0.6 light
● moderate ● moderate
heavy heavy
Extinction magnitude
Extinction magnitude
0.4 0.4
●
●
●
●
0.2 ● ●
●
●
●
0.2
●
● ● ●
● ●
● ●
● ● ● ●
●
● ● ● ● ● ●
0.0 0.0
Mid Late Early Mid Mid Late Early Mid
Triassic Jurassic Triassic Jurassic
240 220 200 180 240 220 200 180
Age (Ma) Age (Ma)
FIG. 6. Proportional generic richness and extinction magnitude of groups of calcifying organisms, from the Middle Triassic to the
Middle Jurassic, distinguishing between light, moderate and heavy calcification. A–B, proportional generic richness; A, all groups; B,
benthic groups. C–D, extinction magnitude; C, all groups; D, benthic groups.
(Simms 1990), then this may represent a real functional palaeoecological interpretation, where either: (1) no
extinction and later re-occupation of the same niche by a megalodontids were photosymbiotic; or, (2) some photo-
different lineage of hemi-pelagic crinoids. symbiotic megalondontids also survived the extinction.
The surficial, unattached, ‘other’ feeders are represented The single new mode of life that appears in the after-
by certain genera of megalondontid bivalves that have math of the extinction and the three new modes of life
been interpreted as possessing photosynthetic symbionts that appear in the Rhaetian, immediately preceding the
(Yancey & Stanley 1999). A number of megalondontids extinction, are all reoccurrences of niches that are found
that are not regarded as photosymbiotic are, however, earlier in the Permian or Triassic, but not previously in
known to have survived the extinction (Ros-Franch et al. the Late Triassic (i.e. Carnian and Norian). The reappear-
2014). Given that the identification and interpretation of ance of semi-infaunal, slow-moving, deposit feeders and
photosymbiosis in alatoform bivalves is regarded as prob- predators, represented by nuculoid bivalves, gastropods,
lematic (Yancey & Stanley 1999; Vermeij 2013) the appar- asteroids and ophuiroids, which were apparently absent
ent loss of this mode of life may be an artefact of through the Carnian and Norian, is probably an artefact.144 PALAEONTOLOGY, VOLUME 61 These taxa are rare throughout the entire sequence and early Hettangian and maintain this throughout the Early their absence from the Carnian and Norian is probably Jurassic and into the Middle Jurassic. This is a common due to sampling failure. A further two modes of life in pattern that is also recorded in the aftermath of the late the Rhaetian are reoccupied following gaps in the Mid- Permian (Twitchett 2006; Foster & Twitchett 2014; Foster dle–Late Triassic and are represented by shallow infaunal, et al. 2016, 2017) and end-Cretaceous (Aberhan & Kies- slow moving or facultatively mobile, chemosymbiotic sling 2015) extinction events, when suspension feeders bivalves of the family Lucinidae known also from the suffered a crash primarily attributed to epicontinental Early Triassic (Hautmann & N€ utzel 2005). Therefore, as shelf turbidity (Foster & Twitchett 2014) or a productivity has been recorded for the late Permian mass extinction crisis (Aberhan & Kiessling 2015) respectively. Given the (Erwin et al. 1987; Foster & Twitchett 2014), even though similarities between the causal mechanisms of the late taxonomic richness was low in the aftermath of the Late Permian and LTE events, it seems more likely that high Triassic extinction, global ecosystem functioning of key- sediment fluxes and eutrophication-driven dysoxia caused stone species was maintained with each functional group by increased runoff from the terrestrial realm may have occupied by only a few individual genera. spelled the downfall of benthic suspension feeding groups As happened in the late Permian mass extinction, the at the LTE (Algeo & Twitchett 2010; Jaraula et al. 2013; LTE was felt most intensely in the tropics and dispropor- Foster et al. 2015; Schobben et al. 2015, 2016). Such a tionally affected reef-dwelling organisms with suspension scenario may also explain the earlier exclusion of suspen- and/or photosymbiotic feeding habits, whilst deposit-feed- sion feeding guilds after the Carnian as a result of the ing modes of life appear to benefit (Kiessling & Aberhan CPE (Dal Corso et al. 2015; Ruffell et al. 2016). Increased 2007; Kiessling et al. 2007; Martindale et al. 2012). The runoff and subsequent eutrophication of shallow shelf sea strong extinction signal in the tropical latitudes and the settings would have simultaneously choked suspension preferential extinction of reef taxa supports the hypothesis feeders, pushing them offshore, whilst benefiting deposit- of a CAMP-driven, climate-warming kill mechanism driv- feeding organisms (Algeo & Twitchett 2010). It is also ing high sea surface temperatures and anoxia for the LTE possible that reductions in suspension feeder richness (Hallam & Wignall 2000; Kiessling et al. 2007; McRoberts may be, in part, because many are heavily calcified (i.e. et al. 2012; Jaraula et al. 2013; Bond & Wignall 2014; van crinoids, brachiopods etc.), and heavily calcified organ- de Schootbrugge & Wignall 2016). isms, particularly in reef settings, were selected against. Previous studies have suggested that global warming dri- Although the LTE appears to have been concentrated ven by rapid CO2 release should also preferentially impact in the tropical latitudes, the Toarcian OAE appears to heavily calcified marine organisms, due to the effects of have been felt most severely in the mid-latitudes, with the acidification (Hautmann 2004; Knoll et al. 2007). Although temporary exclusion of infaunal taxa being consistent heavily calcified organisms do record the highest extinction with widespread dysoxia and/or anoxia being the main rates across the T–J boundary, it is not clear that acidifica- driving force of the early Toarcian extinction (Little & tion was driving extinction because lightly calcified organ- Benton 1995). This is further supported by the complete isms also record high rates of extinction. This result absence of reef environments in the tropics during the contrasts with those of some previous studies where there Hettangian and the mid-latitudes during the Toarcian. appear to be clear differences in extinction rates between High extinction levels in Panthalassa, as compared to physiologically buffered and unbuffered organisms (Haut- Tethys, particularly amongst erect suspension feeders and mann et al. 2008a; Kiessling & Simpson 2011). In other infaunal taxa may be a real signal, or may be a conse- warming-related extinctions in the study interval, such as quence of the high levels of extinction recorded in tropi- the early Toarcian, selection against heavily calcified organ- cal latitudes and reef ecosystems. Further investigation of isms is much clearer (Fig. 6). The extinction pattern across the determinants of extinction is required to answer this. the LTE is consistent with that of previous studies, which Although we see a general trend of reduced generic show that although heavy calcifiers experienced higher richness within modes of life across the LTE, a number of extinction rates, there are not huge differences between modes of life actually show an increase in richness. These light, moderate and heavy calcifiers (Bush & Pruss 2013) functional groups, although not particularly diverse, all and that selectivity against heavy calcifiers only becomes display feeding strategies that are expected to fare well in apparent after the removal of other confounding factors a stressed, post-extinction world (i.e. deposit-feeding, (Payne et al. 2016b). Therefore, it appears that even if acid- mining and chemosymbiosis) (Foster & Twitchett 2014; ification contributed to the LTE and reef collapse, other kill Clapham 2017). Some functional groups, such as pelagic, mechanisms (i.e. temperature change) were also important. fast-moving carnivores, suffered a significant reduction in The reductions in suspension-feeder richness contrast relative richness at the LTE, but then quickly recovered to with the slow moving, surficial, deposit feeders, grazers pre-extinction levels by the late Hettangian, reinforcing and predators, which increase in relative richness in the the idea that both vertebrates (Thorne et al. 2011)
DUNHILL ET AL.: IMPACT OF THE LATE TRIASSIC MASS EXTINCTION 145 (excluding conodonts which become extinct) and cepha- somewhat at odds with previous studies of a single clade lopods (Longridge & Smith 2015) passed through an evo- (i.e. bivalves) which showed selectivity against infaunal lutionary bottleneck at the T–J boundary where rapid taxa (McRoberts & Newton 1995; McRoberts et al. 1995). turnover rates aided rapid recovery. Despite this reduc- Apparent high rates of extinction seen in infaunal bivalves tion in relative richness across the LTE, we see an increase are most likely to be the result of preservational biases in relative occurrences of pelagic, fast-moving carnivores. (Mander & Twitchett 2008). Also, similar increases in An explanation for this discrepancy is that the ranged- deposit-feeders, grazers and predators, as well as in preda- through richness data is compiled at the substage-level, tion-resistant modes of life, have also been recorded in and thus distinguishes the lower and upper Hettangian, the aftermath of the late Permian (Foster & Twitchett whereas the occurrence data from the PaleoDB is com- 2014) and end-Cretaceous mass extinctions (Aberhan & piled at the stage-level, and thus bins together the entire Kiessling 2015). Although initial recovery from the LTE Hettangian. Several studies have demonstrated that recov- was rapid in level-bottom and pelagic communities, full ery was rapid and complete by the upper Hettangian benthic recovery, i.e. the reestablishment of reef ecosys- (Barras & Twitchett 2007; Hautmann et al. 2008b; Man- tems, appears to have been delayed by the subsequent cri- der et al. 2008) and it is likely that some of the extinction sis in the Early Toarcian. Overall, the similarities in effects are being masked due to the resolution of this functional, environmental and biogeographical extinction study. Nevertheless, some of the pre–post-extinction selectivity and recovery between the late Permian and the changes are still evident in the occurrence data, most Late Triassic events are striking, although not surprising notably the reduction in non-motile suspension feeders given the similarities in the causal mechanisms of the two and the increase in motile grazers, deposit feeders and events (van de Schootbrugge & Wignall 2016). predators. Whilst dissimilarity analyses show that the global eco- logical shift from Rhaetian to Hettangian communities CONCLUSIONS was greater than would be expected by chance, it is not notably larger than the ecological shifts witnessed at many Although the LTE does not appear to have resulted in other stage-level transitions and is similar to the ecologi- the extinction of any ecological guilds or a permanent cal shift witnessed between the Toarcian and Aalenian global ecological regime shift, it had very profound (the Early to Middle Jurassic transition). Comparing the regional and environmental effects, with the temporary dissimilarity between the Rhaetian and each Jurassic time complete collapse of tropical reef ecosystems (McGhee bin in turn shows that dissimilarity is higher between the et al. 2004; Kiessling & Simpson 2011; Martindale et al. Rhaetian and Hettangian than it is between the Rhaetian 2012) that did not fully recover until the Middle Juras- and Sinemurian, indicating that the global ecosystem was sic. The effects of the LTE were largely confined to the returning to its pre-extinction state by the Sinemurian. tropical latitudes and there was significant selection This is followed by a slight increase in dissimilarity against some modes of life, particularly stationary sus- between the Rhaetian and Pliensbachian, followed by a pension feeders. We did not find any conclusive evi- large increase in dissimilarity between the Rhaetian and dence that heavily calcifying organisms were selected Toarcian, as a result of the early Toarcian OAE. However, against during the LTE, as compared to light calcifiers, the Aalenian has a more similar ecological composition to but this may well have been the case during the early the Rhaetian than the latter does to any of the Early Toarcian OAE. The lack of evidence of permanent global Jurassic time intervals, thus signifying a recovery to some- regime shifts across major mass extinction events (Foster thing more similar to pre-extinction levels at the onset of & Twitchett 2014), coupled with reduced estimates of the Middle Jurassic, probably driven by the beginning of taxonomic extinction (Stanley 2016), suggest that mass the recovery of reef ecosystems (Hautmann 2012, 2016). extinctions may not have been as ecologically catas- Benthic marine deposit-feeders, predators and grazers, trophic as once suspected. However, profound regional dominated by gastropods, echinoderms and malacostra- ecological effects and global taxonomic losses of 30–80% cans, flourished during the earliest Jurassic in the after- of species (Stanley 2016) still represent global biotic math of the crisis. Non-motile epifauna declined across catastrophes of unimaginable magnitudes when thought the LTE, but infauna, motile epifauna and pelagic preda- of in terms of modern biodiversity. It therefore remains tors all increased in taxonomic richness and number of that the mass extinctions of the distant past, particularly occurrences. This trend is associated with the Mesozoic those linked to rapid greenhouse warming, ocean anoxia, Marine Revolution and was ongoing throughout the Late and ocean acidification (e.g. late Permian and Late Tri- Triassic (Tackett & Bottjer 2016), before being invigorated assic), offer a stark reminder of the possible future con- by the preferential extinction of non-motile, suspension- sequences of anthropogenic influences on the biosphere feeding benthic guilds at the LTE. Our findings are (Harnik et al. 2012; Hull 2015).
146 PALAEONTOLOGY, VOLUME 61
Acknowledgements. The authors thank the numerous authors of B L A C K B U R N , T. J., O L S E N , P. E., B O W R I N G , S. A.,
the original studies that provide the source data on which this M C L E A N , N. M., K E N T , D. V., P U F F E R , J., M C H O N E ,
study is based, and the many data-enterers of the Paleobiology G., R A S B U R Y , E. T. and E T - T O U H A M I , M. 2013. Zir-
Database for the provision of the fossil occurrence data, particu- con U-Pb geochronology links the end-Triassic extinction with
larly: Matthew Clapham, Wolfgang Kiessling, Franz F€ ursich, the Central Atlantic Magmatic Province. Science, 340, 941–945.
Martin Aberhan, Andy Rees, J ozsef Palfy, Matthew Carrano, B O N D , D. P. G. and W I G N A L L , P. B. 2014. Large igneous
David Bottjer, Alistair McGowan, Arnold Miller, Luc Villier, provinces and mass extinctions: an update. Geological Society
Roger Benson, John Alroy, and Richard Butler. Thanks also to of America Special Papers, 505, 29–55.
Martin Aberhan, Timothy Astrop, Kenneth De Baets, Mariel Fer- B R A Y , J. R. and C U R T I S , J. T. 1957. An ordination of the
rari, Andy Gale, Stefanie Klug, and Crispin Little for discussion upland forest communities of Southern Wisconsin. Ecological
of fossil modes of life, taxonomic assignment and stratigraphic Monographs, 27, 325–349.
range data. Thanks to Michael Hautmann, Andrew Bush, and B U S H , A. M. and B A M B A C H , R. K. 2011. Paleoecologic
two anonymous reviewers for constructive comments that megatrends in marine Metazoa. Annual Review of Earth &
greatly improved this manuscript. Thanks to Sally Thomas for Planetary Sciences, 39, 241–269.
editorial comments. Thanks to Autumn Pugh, Crispin Little, B U S H , A. and P R U S S , S. 2013. Theoretical ecospace for
and Paul Wignall for discussion of the manuscript. AMD is ecosystem paleobiology: energy, nutrients, biominerals, and
funded by a Leverhulme Early Career Fellowship (ECF-2015- macroevolution. The Paleontological Society Papers, 19, 1–20.
044) and a NERC research grant (NE/P013724/1). WJF is funded B U S H , A. M., B A M B A C H , R. K. and D A L E Y , G. M. 2007.
by a Distinguished Postdoctoral Fellowship from the University Changes in theoretical ecospace utilization in marine fossil
of Texas at Austin. This is Paleobiology Database publication assemblages between the mid-Paleozoic and late Cenozoic.
293. Paleobiology, 33, 76–97.
C L A P H A M , M. E. 2017. Organism activity levels predict mar-
ine invertebrate survival during ancient global change extinc-
DATA ARCHIVING STATEMENT tions. Global Change Biology, 23, 1477–1485.
-and P A Y N E , J. L. 2011. Acidification, anoxia, and extinc-
Data for this study are available in the Dryad Digital Repository: tion: a multiple logistic regression analysis of extinction selec-
https://doi.org/10.5061/dryad.bg30k tivity during the Middle and Late Permian. Geology, 39, 1059–
1062.
Editor. Michael Hautmann - K I E S S L I N G , W., F U € R S I C H , F., A B E R H A N , M.,
R E E S , A., P A L F Y , J., C A R R A N O , M. T., B O T T J E R , D.
B., M C G O W A N , A. J., M I L L E R , A. I., V I L L I E R , L.,
REFERENCES B E N S O N , R. B. J., A L R O Y , J. and B U T L E R , R. J. 2016.
Taxonomic occurrences of Triassic to Jurassic marine animals.
A B E R H A N , M. and K I E S S L I N G , W. 2015. Persistent eco- Paleobiology Database, accessed 13 April 2016. http://paleob
logical shifts in marine molluscan assemblages across the end- iodb.org
Cretaceous mass extinction. Proceedings of the National Acad- D A L C O R S O , J., G I A N O L L A , P., N E W T O N , R. J.,
emy of Sciences, 112, 7207–7212. F R A N C E S C H I , M., R O G H I , G., C A G G I A T I , M.,
A L G E O , T. J. and T W I T C H E T T , R. J. 2010. Anomalous R A U C S I K , B., B U D A I , T., H A A S , J. and P R E T O , N.
Early Triassic sediment fluxes due to elevated weathering rates 2015. Carbon isotope records reveal synchronicity between
and their biological consequences. Geology, 38, 1023–1026. carbon cycle perturbation and the “Carnian Pluvial Event” in
A L R O Y , J. 2010. The shifting balance of diversity among major the Tethys realm (Late Triassic). Global & Planetary Change,
marine animal groups. Science, 329, 1191–1194. 127, 79–90.
B A M B A C H , R. K., B U S H , A. M. and E R W I N , D. H. 2007. D I N E E N , A. A., F R A I S E R , M. L. and S H E E H A N , P. M.
Autecology and the filling of ecospace: key metazoan radia- 2014. Quantifying functional diversity in pre- and post-extinc-
tions. Palaeontology, 50, 1–22. tion paleocommunities: a test of ecological restructuring after
B A R N O S K Y , A. D., H A D L Y , E. A., B A S C O M P T E , J., the end-Permian mass extinction. Earth-Science Reviews, 136,
B E R L O W , E. L., B R O W N , J. H., F O R T E L I U S , M., 339–349.
G E T Z , W. M., H A R T E , J., H A S T I N G S , A., M A R - D U N H I L L , A. M., S C I B E R R A S , J., F O S T E R , W. J. and
Q U E T , P. A., M A R T I N E Z , N. D., M O O E R S , A., T W I T C H E T T , R. J. 2017. Data from: Functional diversity
R O O P N A R I N E , P., V E R M E I J , G., W I L L I A M S , J. W., of marine ecosystems across the Late Triassic mass extinction.
G I L L E S P I E , R., K I T Z E S , J., M A R S H A L L , C., Dryad Digital Repository. https://doi.org/10.5061/dryad.bg30k
M A T Z K E , N., M I N D E L L , D. P., R E V I L L A , E. and E R W I N , D. H., V A L E N T I N E , J. W. and S E P K O S K I , J. J. J.
S M I T H , A. B. 2012. Approaching a state shift in Earth’s bio- 1987. A comparative study of diversification events: the early
sphere. Nature, 486, 52–58. Palaeozoic versus the Mesozoic. Evolution, 41, 1177–1186.
B A R R A S , C. G. and T W I T C H E T T , R. J. 2007. Response of F L E S S A , K. W. and J A B L O N S K I , D. 1983. Extinction is
the marine infauna to Triassic-Jurassic environmental change: here to stay. Paleobiology, 9, 315–321.
ichnological data from southern England. Palaeogeography, FLU € G E L , E. 2002. Triassic reef patterns. SEPM Special Publica-
Palaeoclimatology, Palaeoecology, 244, 223–241. tion, 72, 391–463.You can also read
