The Paleocene–Eocene Thermal Maximum (PETM, 56 Ma) was a phase of rapid
global warming associated with massive carbon input into the ocean–atmosphere
system from a
Long-term gradual warming during the late Paleocene and early Eocene
(
The integrated changes in climate and the carbon cycle lead to global
average surface warming during the PETM of the order of
The magnitude of warming during the PETM and environmental and ecological
effects have been extensively documented at midlatitude and high-latitude sections
(see review in Sluijs et al.,
2014). Although micropaleontological studies indicate assemblage shifts and
environmental perturbation in several tropical open ocean regions (e.g.,
Ocean Drilling Program (ODP) Site 865; Kelly et al.,
1996, ODP Site 1001; Bralower et al., 1997),
the underlying mechanisms remain unclear because causal relations with
physicochemical parameters are difficult to establish. Evidence from
Tanzania, Nigeria, and ODP Site 865 indicates that tropical surface oceans
warmed by
Map of eastern tropical Atlantic and location of ODP Site 959.
Recently, a massive decrease in abundance and diversity of dinocysts was
found during the PETM in Nigeria, which was attributed to heat stress
(Frieling et al., 2017). The impact and geographical
extent of this heat-stress-driven biodiversity drop, however, remain unknown,
although similar heat-stress effects on marine biota may have been
widespread in tropical regions
(Aze et al., 2014; Yamaguchi and
Norris, 2015). Although other stressors like stratification, salinity
fluctuations, and acidification may play a role, it should be noted that
dinoflagellates typically thrive under such conditions, including high
CO
The ODP recovered a relatively complete Cretaceous and Cenozoic sediment
sequence from Hole 959D, located on the CIGTM, in 1995 (3
The PETM as recorded by dinoflagellate cyst assemblages at Site
959.
Frieling et al. (2018a) recorded a
All discrete samples taken from the working halves were freeze-dried and measured for bulk magnetic susceptibility before splitting them into fractions for palynology and inorganic and organic geochemical analysis.
Freeze-dried samples were weighed and measured for bulk magnetic susceptibility on an MFK1-FA at the paleomagnetism laboratory, Fort Hoofddijk, Utrecht University. Reproducibility was determined with replicate measurements and was always better than 1 %.
Compilation of tropical PETM sites and sites that may show evidence for heat stress. Abbreviations are as follows: sea surface temperature (SST), Ocean Drilling Program (ODP), Tanzania drilling program (TDP).
References:
We processed and counted a total of 155 samples, using standard protocols
used at the Laboratory of Palaeobotany and Palynology at Utrecht University
(Sluijs et al., 2003). In brief, a spike of exotic spores
XRF Core Scanner data were collected every 1 cm down-core over a 1 cm
Bulk sediment chemistry was determined on 28 samples in the interval that
could not be used for XRF core scanning and 22 samples from intervals with
XRF scanning data to assess consistency between the two methods.
Approximately 125 mg of powdered freeze-dried sediment was dissolved in
2.5 mL of HF (40 %) and 2.5 mL of a HClO
The PETM as recorded with organic and inorganic geochemical proxies
at Site 959.
The carbonate concentration of Site 959 samples was determined at the
University of Southampton using a UIC CM5015 coulometer operated with a UIC
CM5011 emulator and coupled to an AutoMate FX autosampler and carbonate
digestion system. The samples were oven-dried at 50
Dinocysts typically dominate the palynological residues. Pollen and spores
derived from terrestrial higher plants are present but in very low
abundances (average
Apart from sediments close to the PETM, the studied interval typically
comprises 10–30 weight % (wt %) CaCO
We use elemental ratios to reconstruct hinterland hydrology and redox-sensitive trace elements to reconstruct bottom-water oxygenation for the
site location. The Ti
Magnetic susceptibility is anticorrelated with carbonate percentages in the upper Paleocene. At 804.2 m b.s.f., before the onset of the CIE, this relation disappears. The porcellanite layer is marked by low magnetic susceptibility and the body of the CIE is marked by stable high values. A maximum is reached during the recovery phase, at 802.97 m b.s.f., followed by a second maximum at 802.81 m b.s.f. Bulk magnetic susceptibility measurements closely resemble Fe concentrations derived from both ICP-OES and XRF (Fig. 3j) and correlate to TOC wt % in the upper Paleocene and lower part of the CIE.
The latest Paleocene at the core location is characterized by cyclic
variations in proxy records, including TOC,
dinocysts per gram of sediment, magnetic susceptibility, and CaCO
Dinocyst assemblages in the Paleocene are characterized by abundant
Occasionally, Goniodomideae are present in great abundance. In the modern ocean (Zonneveld et al., 2013) and in the Paleogene (Sluijs and Brinkhuis, 2009), high abundances of Goniodomideae occur in very shallow marine settings such as lagoons, and are typically associated with warm, stratified waters and very high and/or seasonally fluctuating salinity. In open ocean settings the group can be indicative of strong stratification (Reichart et al., 2004). Considering the offshore location of Site 959, and relatively low abundances of low-salinity-tolerant taxa, we interpret high abundances of Goniodomideae to indicate seasonally strong stratification, either by temperature or salinity.
From 804.4 m b.s.f., we find an increase in abundance of dinocysts belonging to,
or closely related to, the genus
The PETM is associated with a drop in
Following inferences from previously published records
(Sluijs et al., 2008b), higher
sea level should hence result in a decrease in the relative abundance of
However, Goniodomideae, typically associated with stratification
(Sluijs et al., 2005; Zonneveld et al., 2013), are abundant lower in the analyzed
section and increase in relative abundance together with
The quasi-global acme of the tropical dinocyst genus
In addition to the acme of
The abundance of
At Site 959,
At Site 959, we find the highest abundance (95 %) of
Many sites globally show decreased bottom-water oxygen content during the
PETM and at some shelf sections bottom waters
(Dickson et al.,
2012, 2014) and even the photic zone became euxinic
(Sluijs
et al., 2006, 2014; Frieling et al., 2014, 2017; Schoon et al., 2015).
Oxygen minimum zones were expanded (Zhou et al.,
2016) and deep ocean waters were affected by deoxygenation, although anoxia
did not develop in the deep sea
(Chun et al., 2010;
Pälike et al., 2014). At Site 959, we find strong indications of
decreased oxygen concentrations in bottom waters in the form of increased
organic matter burial fluxes, as assessed through reconstructed accumulation
rates and TOC wt %. Moreover, increasing C
The dinocyst assemblages, but also other microfossil groups, in
midlatitudes have been successfully used for detailed environmental
reconstructions (e.g.,
Gibbs
et al., 2006; Hollis, 2007; Sluijs and Brinkhuis, 2009; Stassen et al.,
2012). Here, we find extremely low numbers of dinocysts (100–200 g
A few dinocyst genera, however, are present. The CIE is notably marked by
higher relative abundances of Protoperidinioid cysts, as well as
At 803.0 m b.s.f., a small (
Only the variability of dinocyst assemblages during the body of the CIE at the New Jersey Shelf sites Bass River and Wilson Lake resemble the results at Site 959 (Sluijs and Brinkhuis, 2009). However, Bass River and Wilson Lake are shallower and closer to the paleoshore and therefore more susceptible to environmental swings than the offshore Site 959. The strong variation at Site 959, representing a fully open ocean setting is therefore much more surprising.
Potentially, the recorded variability at Site 959 is related to sub-Milankovitch climate variability, which has been observed in a range of tropical sections from the Proterozoic onwards (Wu et al., 2012; Wilson et al., 2014) and also within the PETM (Abdul Aziz et al., 2008) and in dinocyst records (Reichart and Brinkhuis, 2003). However, we observe no cyclic recurrence of these dinocyst assemblage variations, nor do we find the far more obvious cyclic (precession scale) variation in the upper Paleocene in the dinocyst assemblages. Therefore, this explanation appears unlikely. Unfortunately, none of our proxy records show similar variability in this period; thus, we cannot resolve the underlying cause of such extreme variations in dinocyst assemblages.
Early Eocene dinocyst assemblages and thus presumably environmental
circumstances are similar to those found in the Paleocene. Only
The recorded warming across the onset of the PETM (3.9
Sediments typically contain less than 30 % CaCO
After carbonate accumulation ceased, without apparent change in other
proxies or large changes in sediment accumulation rates, dinocyst
concentrations decreased from 5000 to 100–200 g
In addition to these overall similarities, the dinocyst assemblages at Site
959 are very similar to those found at the PETM in Nigeria. Both sections
are marked by relatively high percentages of Protoperidinioid dinocysts
within the body of the PETM, which is an exclusively heterotrophic group
(Jacobson and Anderson, 1986). Oxidation experiments show that
Protoperidinioid cysts are usually less resistant to oxidation than
The combined information suggests that eukaryote activity in the mixed layer was suppressed not only in Nigeria (Frieling et al., 2017) but also at the more offshore Site 959. Similar to Nigeria, there is hence no evidence that the low numbers of dinocysts resulted from severe stratification and anoxia, strong variations in salinity, or biases resulting from preservation. Although the initial absence of carbonate is perhaps related to a secondary factor, we attribute the drop in eukaryote production in this region to heat stress, as this is among the few physicochemical factors that can generate the same effect in both the open ocean (Site 959) and on the shelf (Nigeria).
It is important to note that primary productivity did not collapse due to the lack of eukaryotes. Organic linings of benthic foraminifera are abundant within the PETM interval at Site 959 and in Nigeria, which suggests that food supply was likely similar to before the PETM. Both sections are characterized by enhanced organic carbon content and burial. This apparently prokaryote-dominated ecosystem, with enhanced organic carbon burial, hence bears some similarities to tropical ecosystems during the Cretaceous ocean anoxic events (OAEs; Kuypers et al., 2001).
While heat stress seems the most likely option for the demise of eukaryotes
in the equatorial Atlantic, other open ocean equatorial and tropical (20
It is far from certain if heat stress affected planktic calcifiers not only in the eastern equatorial Atlantic and the western Indian Ocean but also in larger areas in the tropics. However, if it did, it might have affected sedimentary sequences in the deep sea to the extent that it had implications for global carbon cycling during the PETM and thus calculations of the mass of carbon that was injected into the ocean–atmosphere system during the PETM (e.g., Panchuk et al., 2008; Zeebe et al., 2009; Luo et al., 2016).
First, it would imply that the reduction in carbonate accumulation rates at tropical sites were potentially a combined result of acidification and heat stress. If so, the reduction in carbonate accumulation across the PETM should have been relatively high at tropical sites relative to higher latitudes (see Sect. 5.4). Interestingly, the reduction in carbonate accumulation was most severe at Walvis Ridge (Zachos et al., 2005) and the Caribbean (Bralower et al., 1997), both located towards the borders of the tropical band, which is in sharp contrast to sustained carbonate deposition in the Atlantic Southern Ocean sites 689 and 690 (Kelly et al., 2005, 2010). Although carbonate wt % at Pacific tropical sites remains high (Colosimo et al., 2006; Leon-Rodriguez and Dickens, 2010), this is largely due to very low accumulation rates of the sole other sedimentary component, clay, and bioturbation, complicating robust estimates of carbonate accumulation through time.
Second, because calcification consumes alkalinity and produces CO
Collectively, it requires thorough reinterpretation of numerous published records to evaluate to what extent tropical heat stress might have contributed to reduced carbonate accumulation in the global ocean.
The Paleocene–Eocene transition at Site 959 is marked by a warming of 3.9
Importantly, based on our multi-proxy records, we conclude that these
extreme temperatures caused a remarkable drop in dinocyst abundances, and
most likely also in the production of biogenic carbonate and opal during the
PETM. Crucially, only heterotrophic dinoflagellate species, which likely
resided somewhat deeper in the water column, persisted. Together with
relatively abundant organic benthic foraminifer linings, this indicates
sustained primary production, likely dominated by prokaryotes. If the drop
in calcification was global, it might have contributed to the recorded
decline in deep sea carbonate accumulation, and, as such, it is an important
factor to constrain because it potentially affects calculations of global
ocean acidification and CO
Combined evidence from dinocysts and inorganic chemistry points to an altered hydrological cycle at the PETM, resulting in fluctuations in stratification. Furthermore, progressively less oxygenated bottom waters characterize the body phase of the PETM. In sharp contrast to the general absence of eukaryotes during the body phase of the PETM, the recovery phase of the PETM is highly dynamic, with several groups of dinocysts dominating assemblages on timescales shorter than 10 kyr.
Data used in this publication will be made available
through the online database Pangaea (
JF and AS designed the study. JF, SMB, JJM, UR, TW, and GJR performed inorganic geochemical analyses. JF generated palynological data. All authors analyzed and discussed the data. JF and AS wrote the paper, with input from all authors.
The authors declare that they have no conflict of interest.
We thank Natasja Welters, Arnold van Dijk, Ton Zalm, and Dominika Kasjaniuk (Utrecht University) for analytical support and Alex Wülbers and Walter Hale (IODP Bremen Core Repository) and Willem Schinkel (Erasmus University Rotterdam) for sampling assistance. The European Research Council (ERC) under the European Union Seventh Framework Program provided funding for this work with ERC starting grant 259627 to Appy Sluijs. The Netherlands Organization for Scientific Research (NWO) supported this work through grant no. 834.11.006 to Gert-Jan Reichart. The Deutsche Forschungsgemeinschaft supported Ursula Röhl and Thomas Westerhold. This work was carried out under the program of the Netherlands Earth System Science Centre (NESSC), financially supported by the Ministry of Education, Culture and Science (OCW). The International Ocean Discovery Program (IODP) is acknowledged for access to materials and data. Edited by: Luc Beaufort Reviewed by: two anonymous referees