Isotopic filtering reveals high sensitivity of planktic calcifiers to Paleocene–Eocene thermal maximum warming and acidification

Significance

Human-induced carbon emissions are causing global temperatures to rise and oceans to acidify. To understand how these rapid perturbations affect marine calcifying communities, we investigate a similar event in Earth’s geologic past, the Paleocene–Eocene thermal maximum (PETM). We introduce a method, isotopic filtering, to mitigate the time-averaging effects of sediment mixing on deep-sea microfossil records. Contrary to previous studies, we find that tropical planktic foraminifers in the central Pacific ocean were adversely affected by PETM conditions, as evidenced by a decrease in local diversity, extratropical migration, and impaired calcification. While these species survived the PETM through migration to cooler waters, it is unclear whether marine calcifiers can withstand the rapid changes our oceans are experiencing today.

Abstract

Ocean warming and acidification driven by anthropogenic carbon emissions pose an existential threat to marine calcifying communities. A similar perturbation to global carbon cycling and ocean chemistry occurred ∼56 Ma during the Paleocene–Eocene thermal maximum (PETM), but microfossil records of the marine biotic response are distorted by sediment mixing. Here, we use the carbon isotope excursion marking the PETM to distinguish planktic foraminifer shells calcified during the PETM from those calcified prior to the event and then isotopically filter anachronous specimens from the PETM microfossil assemblages. We find that nearly one-half of foraminifer shells in a deep-sea PETM record from the central Pacific (Ocean Drilling Program Site 865) are reworked contaminants. Contrary to previous interpretations, corrected assemblages reveal a transient but significant decrease in tropical planktic foraminifer diversity at this open-ocean site during the PETM. The decrease in local diversity was caused by extirpation of shallow- and deep-dwelling taxa as they underwent extratropical migrations in response to heat stress, with one prominent lineage showing signs of impaired calcification possibly due to ocean acidification. An absence of subbotinids in the corrected assemblages suggests that ocean deoxygenation may have rendered thermocline depths uninhabitable for some deeper-dwelling taxa. Latitudinal range shifts provided a rapid-response survival mechanism for tropical planktic foraminifers during the PETM, but the rapidity of ocean warming and acidification projected for the coming centuries will likely strain the adaptability of these resilient calcifiers.

Unabated carbon emissions are projected to raise atmospheric carbon dioxide (CO₂) concentrations to levels (850 parts per million volume [ppm], shared socio-economic pathway [SSP]3-7.0) that have not existed on Earth for nearly 50 million y and increase global temperatures by ∼4 °C over the coming centuries (1, 2). Much of the carbon emitted over this timeframe will be absorbed by the oceans, lowering seawater pH and increasing the solubility of carbonate minerals such as calcite (CaCO₃) that many marine organisms use to build their shells and exoskeletons (3–5). Thus, the combined effects of ocean warming and acidification pose an existential threat to marine calcifying organisms (6), including various calcareous plankton that are foundational to the marine food web (7–9).

To better understand the biotic consequences of human-induced climate change, many researchers have turned their attention to geological records of past greenhouse climate states. One such global warming event, the Paleocene–Eocene thermal maximum (PETM), occurred 56 million y ago and is considered a natural analog for ocean warming and acidification driven by anthropogenic CO₂ emissions (10, 11). Geochemical records show that sea surface temperatures (SSTs) warmed by ∼4 to 5 °C globally during the PETM (12, 13) and that this transient (∼170 ka) warming was coupled to a sharp rise in atmospheric CO₂ levels and a sustained period of ocean acidification (14–16). A global hallmark of the PETM is a decrease in the carbon isotope (δ¹³C) compositions of inorganic and organic materials in both marine and terrestrial records (17, 18). This negative carbon isotope excursion (CIE) signals a geologically rapid (≤5 kyr) release of massive quantities of ¹³C-depleted carbon into the ocean–atmosphere system (11, 19–21), making it a reliable chemostratigraphic marker for correlating marine and continental PETM records from various sites around the world. However, close examination of the CIE signature in deep-sea sedimentary records has demonstrated that pelagic PETM records are often distorted by such preservation biases as carbonate dissolution, diagenesis, and sediment mixing (11, 14, 22, 23).

Of the aforementioned taphonomic processes, sediment mixing is arguably the most underappreciated impediment to achieving a clear understanding of the impacts that PETM warming and ocean acidification had on marine calcifiers. The marine microfossil record provides one of our best insights into past oceanic and biotic change, yet ocean currents and the burrowing activities of benthic organisms (bioturbation) blend anachronous microfossils into aggregate assemblages before they are incorporated into the deep-sea sedimentary archive (24). Vertical sediment mixing typically homogenizes surficial sediments on time scales of 1,000s to 10,000s of years in pelagic settings (25, 26), which can potentially mask transitory ecological responses to rapid ocean-climate change. Indeed, high-resolution δ¹³C records show that PETM assemblages of planktic foraminifers are time-averaged mixtures of ¹³C-enriched shells calcified prior to the CIE and ¹³C-depleted shells calcified during the CIE (e.g., refs. 27–29). And as we shall show, failure to account for the commingling of pre-CIE and CIE microfossils has perpetuated a longstanding misconception that tropical planktic foraminifers diversified in open-ocean settings during the PETM (27, 30).

Here, we reconstruct the response of tropical planktic foraminifers to PETM conditions by deconvolving the effects of sediment mixing through isotopic filtering, where the δ¹³C signatures of 548 individual planktic foraminifer shells are used to filter out reworked, pre-CIE specimens from a series of samples taken from within the CIE interval of the PETM record from Ocean Drilling Program (ODP) Site 865. Paleolatitude projections (31) place Site 865 within a few degrees of the equator during the late Paleocene (Fig. 1A), and foraminifer δ¹³C records have established that two parallel PETM records were recovered at Site 865 (32), one from drill Hole B and another from nearby drill Hole C. A wealth of single-shell δ¹³C data (n = 235) was compiled for the 865C section by previous studies [Kelly et al., 1996 (27) and 1998 (30)], which we expanded by conducting supplementary single-shell analyses (n = 313 this study) on additional species and specimens from the CIE interval of the same section. Analytical limitations necessitated the use of relatively large (>250 μm) specimens for single-shell δ¹³C measurements. Still, we are able to constrain the proportions of reworked pre-CIE and in situ CIE specimens for the nine most abundant taxa (SI Appendix, Figs. S1 and S2). The relatively high δ¹³C signatures (∼4‰) of pre-CIE morozovellids and acarininids indicate that these taxa inhabited the ¹³C-rich waters of the euphotic zone, while the lower δ¹³C values (∼2‰) of pre-CIE subbotinids indicate biocalcification within deeper, ¹³C-depleted waters of the thermocline (33, 34) (Fig. 1B).

Fig. 1.

Site 865 location and CIE marking the PETM. (A) Paleogeographic map showing tropical location of open-ocean Site 865 (red star) during the late Paleocene (source: Ocean Drilling Stratigraphic Network, odsn.de). (B) Planktic foraminifer carbon isotope (δ¹³C) records for the 865C section constructed with pooled, multishell samples of M. velascoensis (orange squares), Acarinina spp. (green diamonds), and Subbotina spp. (blue triangles) (32). The red arrow denotes stratigraphic position (meters below sea floor = mbsf) of the benthic foraminifer extinction event associated with CIE onset (32). (C and D) Bulk-carbonate δ¹³C records of CIE in the 865C and 865B PETM sections, respectively. The break along vertical axis in panel D signifies a coring gap. The dark gray shading delimits the correlative lower CIE interval in the two PETM records. The light gray shading delimits the portion of 865B record sampled for planktic foraminifer assemblage counts. (E and F) Relative abundance curves showing acme of excursion marker taxa (M. allisonensis, A. sibaiyaensis, and A. africana) based on unfiltered census counts (>125 μm) for 865C and 865B PETM planktic foraminifer assemblages, respectively. Assemblage count data for the 865C PETM section taken from published literature (27). For comparative purposes, data shown are plotted relative (±cm) to stratigraphic level of CIE onset in bulk-carbonate δ¹³C records, which was prescribed a reference depth of 0 cm.

Removal of foraminifer shells for stable isotope analyses has biased the taxonomic composition of the 865C assemblages. As a result, point-counted census data (>250 μm size fraction) were compiled for a fresh set of correlative samples from the CIE interval of the nearby 865B section and corrected for sediment mixing using the per-taxon estimates of reworked shells in the 865C section (Materials and Methods). Several stratigraphic features permit close cross-correlation of samples between the 865B and 865C PETM sections. Parallel bulk-carbonate δ¹³C records show that the early stages of the PETM, extending from the CIE onset to just above the CIE minimum where bulk-carbonate δ¹³C values start to return to higher values, were recovered at both sites (Fig. 1 C and D). Furthermore, initial studies of the Site 865 PETM planktic foraminifer assemblages identified a suite of short-lived “excursion taxa” (Morozovella allisonensis, Acarinina sibaiyaensis, Acarinina africana) whose biostratigraphic ranges are confined to the CIE interval (27, 30). Parallel assemblage counts (>125 μm size fraction) show that the peak abundance of these marker taxa is captured by both PETM records (Fig. 1 E and F). Astronomical tuning of deep-sea PETM records indicates that this portion of the CIE spans the first 70 kyr of the event (26), thus the stratigraphies of the two Site 865 PETM records correlate well at millennial timescales (Fig. 1 C–F).

Results

Isotopic filtering reveals that nearly one-half (49.5%) of the 548 individual foraminifer shells analyzed are pre-CIE contaminants within the 865C PETM record (Fig. 2). The distributions of single-shell δ¹³C values for the shallow-dwelling morozovellids and acarininids are distinctly bimodal (Fig. 2), indicating that these genera are distinguishable mixtures of reworked pre-CIE and in situ CIE shells. Conversely, the subbotinid δ¹³C distribution is unimodal and centered on a pre-CIE value of ∼2‰ (Figs. 1B and 2). This observation, in combination with the precipitous decline in subbotinid abundances in unfiltered census records over the CIE interval at Site 865 and other subtropical sites in the Pacific (35), indicates that all subbotinid shells are pre-CIE contaminants, as previously surmised (27, 30). Only 0.5% (n = 3) specimens register intermediate δ¹³C values, which are believed to reflect either infilling of shells by diagenetic calcite (36) or the occasional downward mixing of shells from the overlying recovery stages of the CIE where δ¹³C values begin to transition back toward higher post-PETM values (Fig. 1C).

Fig. 2.

Frequency distributions of single-shell δ¹³C values for planktic foraminifers from the lower CIE of the Site 865C PETM record (n = 548). The bimodal distributions for shallow-dwelling genera Morozovella (n = 274) and Acarinina (n = 220) indicate that these taxa are mixtures of reworked, pre-CIE specimens with high δ¹³C values and in situ CIE specimens with relatively low δ¹³C values. The unimodal distribution of thermocline-dwelling genus Subbotina (n = 54) is centered on a pre-CIE value for this particular taxon, indicating that all subbotinids are reworked specimens.

Examination of single-shell δ¹³C records for individual taxa reveals that the deep-dwelling subbotinids are not the only group represented by solely reworked, pre-CIE specimens (Fig. 3A). For instance, 100% of Morozovella aequa/subbotinae shells register pre-CIE values (Fig. 3B), as do the overwhelming majority of Morozovella acuta, Morozovella velascoensis, and Acarinina soldadoensis shells within the three lowermost samples (Fig. 3 C–E). We collectively refer to these five taxa as “Lazarus” taxa (37) because their single-shell δ¹³C records demonstrate that they temporarily disappear over the CIE interval only to reappear higher in the Site 865 stratigraphy (Fig. 4C). By contrast, virtually none of the shells belonging to the shallow-dwelling species Acarinina esnaensis and Acarinina esnehensis register pre-CIE values within the CIE interval (Fig. 3 F and G). Thus, these two acarininid species are present before, during, and after the PETM, and so are called “holdover” taxa due to their uninterrupted occurrence. Finally, the “excursion” taxa (M. allisonensis and A. sibaiyaensis) have observed stratigraphic ranges restricted to the CIE interval at Site 865 (27) and their shells register almost exclusively CIE values (Fig. 3 H and I).

Fig. 3.

Proportions of reworked specimens within the lower CIE of the 865C PETM section used to reconstruct population dynamics for individual planktic foraminifer taxa during the early stages of the CIE. (A–E) “Lazarus” taxa disappear over part or all of the early CIE only to subsequently reappear in the fossil record. (F and G) “Holdover” taxa occur continuously before, during, and after the PETM. (H and I) “Excursion” taxa occur almost exclusively during the early CIE. The red error bars delimit the 95% CI as determined by nonparametric bootstrapping (i.e., resampling) of proportions (1,000 iterations). The data are plotted relative (+cm) to stratigraphic level of CIE onset in Site 865C bulk-carbonate δ¹³C record.

Fig. 4.

Comparison of isotopically filtered and unfiltered planktic foraminifer assemblages from the 865B PETM section. Diversity metrics and relative abundances for the nine taxa sampled for single-shell stable isotope analyses (Materials and Methods). Comparisons include (A) diversity (Shannon-H) and (B) combined relative abundance of excursion taxa (M. allisonensis and A. sibaiyaensis). (C) Stacked-area plot showing changes in taxonomic composition and taxon relative abundances for isotopically filtered assemblages over the lower CIE in the 865B PETM section. The break along the vertical axis denotes coring gap in the 865B section (mbsf = meters below sea floor).

Correcting for sediment mixing leads to a reversal of previously reported trends in local diversity (27, 30) and establishes that the transient excursion taxa were a major component of planktic foraminifer communities during peak PETM conditions (Fig. 4). Specifically, the uncorrected record yields an apparent increase in diversity, whereas the isotopically filtered record shows a sharp decrease in local diversity over the CIE interval (Fig. 4A). The relative abundances of the excursion taxa, A. sibaiyaensis and M. allisonensis, are greater than originally reported, accounting for nearly one-half (∼45%) of the early CIE assemblages (Fig. 4B). Isotopic filtering also reaffirms that both the deep-dwelling Subbotina and shallow-dwelling M. aequa/subbotinae lineages disappear entirely over the main body of the CIE (Fig. 4C). Our data thus show that planktic foraminifer communities at Site 865 suffered an abrupt decrease in both taxonomic richness and evenness during the PETM, leaving a depauperate (oligotaxic) assemblage composed predominantly of acarininids (A. esnaensis, A. esnehensis, and A. sibaiyaensis) and the excursion morozovellid, M. allisonensis (Fig. 4C).

Discussion

PETM Extirpations: Tropical Exodus or Taphonomic Artifact?

All taxa that disappear eventually reappear higher in the stratigraphy (Fig. 4C), indicating that no true extinctions are observed among planktic foraminifers at Site 865. Ocean acidification increased carbonate dissolution during the initial stages of the PETM (14, 38), and the dearth of intermediate values in our single-shell δ¹³C dataset (Fig. 2) suggests that the transition from pre-CIE to CIE conditions was either extremely rapid or lost from the Site 865 record due to dissolution (36). Thus, the temporary disappearances of such Lazarus taxa (i.e., Subbotina spp., M. aequa/subbotinae, M. velascoensis, M. acuta, and A. soldadoensis) could be construed as a taphonomic artifact stemming from incomplete preservation (35). However, the presence of delicate A. sibaiyaensis and M. allisonensis shells contradicts a strong dissolution effect. Furthermore, the shells of some Lazarus taxa (A. soldadoensis, M. aequa/subbotinae, M. velascoensis, and M. acuta) are relatively resistant to dissolution (35, 39), which is inconsistent with a strong taphonomic overprint.

A more likely explanation is that the Lazarus taxa emigrated out of the study area toward cooler waters in response to extreme tropical warmth during the PETM. The collapse and subsequent reestablishment of M. aequa/subbotinae populations, which accounted for ∼40 to 50% of the assemblages prior to the CIE, typifies such an extratropical migration (Fig. 4C). The hypothesis for extratropical migration by M. aequa/subbotinae is further supported by marked increases in their abundances throughout the circum-Antarctic region during the PETM (40)—increases that cannot be attributed to sediment mixing because these taxa were absent or present only in trace amounts in the southern high latitudes prior to the CIE. Our isotopically filtered census data thus suggest that some tropical planktic foraminifer taxa experienced both leading- and trailing-edge dynamics to their biogeographic ranges, where poleward range shifts were accompanied by local extirpations in the tropics. Similar range shifts are underway today (41).

Comparable spatiotemporal shifts in the biogeographic ranges of calcareous marine algae (coccolithophores) (42) and temperate continental vegetation (43) have been attributed to PETM warming, suggesting transient range shifts driven by transient climate change. In addition, transient disappearances of planktic foraminifers somewhat similar to those uncovered by this study have been observed in hemipelagic sections from the eastern (Tanzania) (44) and western (Nigeria) (45) margins of Africa, where complementary geochemical records indicate that tropical SSTs (>36 °C) likely exceeded the physiological temperature tolerances of many marine calcifiers. Likewise, foraminifer δ¹⁸O temperature records indicate that exceptionally warm SSTs (∼33 °C) prevailed at Site 865 during the PETM (46). These observations suggest that the warming tropics became uninhabitable for some calcifying plankton, resulting in a tropical exodus of taxa as they emigrated out of the study area to escape extreme PETM warmth.

Site 865 shows that tropical pelagic ecosystems were profoundly perturbed during the PETM, yet planktic foraminifers did not suffer a major extinction event like that experienced by their benthic foraminifer counterparts, in which 40 to 60% of species were lost during the PETM (10, 47). Latitudinal range shifts among planktic foraminifers appear to have been temporary, with local communities returning to pre-event levels of richness and biodiversity following the PETM (Fig. 4A). The reestablishment of planktic foraminifer communities at Site 865 indicates that extratropical migration served as a rapid response for the survival of some taxa, which in turn allowed the expatriated taxa to quickly repopulate the tropics as PETM conditions waned. Hence, these results suggest a differential sensitivity of planktic and benthic foraminifers to ocean warming, perhaps because of a higher thermal heterogeneity in the surface ocean and/or potential for plankton to migrate more readily.

Synergistic Environmental Stressors.

The tropical exodus of many planktic calcifiers indicates that heat stress was the primary driver for the temporary loss of diversity at low latitudes during the PETM (42, 44, 45), yet it is likely that attendant ocean acidification and deoxygenation also contributed to the planktic foraminifer response at Site 865. Stratigraphic unmixing by isotopic filtering shows that the taxa M. acuta and M. velascoensis were temporarily replaced by the excursion taxon, M. allisonensis, during the early stages of the CIE (Figs. 3 C, D, and H and 4C). The highly gradational morphologies of these taxa (30), in combination with the sequential order of their appearance and disappearance, suggest a plexus of forms in which the biconvex axially compressed shells of M. allisonensis represent deformed variants of the M. acuta/velascoensis lineage. We therefore view the brief appearance of the excursion taxon, M. allisonensis, as an ecophenotypic response to ocean acidification. Comparable morphological variation occurred among other calcareous plankton during the PETM, most notably the nannofossil genus Discoaster (48–50). As in the M. velascoensis lineage, increased morphological variability during the PETM gave rise to a continuum of Discoaster morphotypes, some of which are thought to be malformed ecophenotypes resulting from decreased calcite saturation (49, 51). Both culturing experiments (52) and field studies (9, 53–55) indicate that calcification in extant planktic foraminifers is impeded by lower pH and carbonate ion concentrations. Thus, the succession of phenotypes seen at Site 865, where M. allisonensis initially replaces M. velascoensis only to revert back to the antecedent M. acuta/velascoensis phenotype midway through the CIE interval, is consonant with the view that biocalcification and inorganic carbon production were inhibited during the PETM (49, 56), producing strong ecophenotypic responses.

Other researchers have suggested that thermal stress caused a breakdown in the symbiotic association between planktic foraminifers and photosynthetic microalgae (i.e., symbiont bleaching) during the PETM (57, 58). The photosynthetic activity of algal symbionts results in a positive relationship between δ¹³C and shell size in photosymbiotic planktic foraminifers (59), and this δ¹³C-size relationship prevails in both acarininids and morozovellids (34, 60). However, size-segregated analyses of excursion taxa M. allisonensis and A. sibaiyaensis from Site 865 (30) indicate that these taxa retained the characteristic photosymbiotic δ¹³C-size relationship, and the δ¹³C distributions for these two excursion taxa exhibit values similar to those registered by in situ CIE shells of M. velascoensis, M. acuta, and A. soldadoensis. Moreover, if symbiont bleaching occurred at Site 865, δ¹³C signatures of large shells of symbiotic taxa would be expected to decrease by ∼1‰ (59), while the statistical populations of reworked pre-CIE and in situ CIE specimens are separated by ∼3‰ (Fig. 2). Hence, these observations do not fully rule out symbiont bleaching at the PETM but do suggest that its effect on our single-shell foraminifer δ¹³C records is negligible.

Finally, ocean deoxygenation—driven by some combination of temperature solubility effects, changing ocean ventilation, increased nutrient input from continents (eutrophication), and/or methane hydrate destabilization—is thought to have contributed to the turnover in benthic foraminifer communities during the PETM (47, 61–64). There is now a large body of evidence to indicate that the oxygen content of seawater decreased globally during the PETM (65–68), including molecular biomarker (69) and barite sulfur isotope (70) records signaling the development of euxinic conditions within the water column. A warming-induced decrease in oxygen solubility and concomitant expansion of oxygen minimum zones (71) would have had dire consequences for planktic foraminifers with deeper depth ecologies such as the thermocline-dwelling subbotinids. Though paleoredox records indicate that oceanic bottom waters remained relatively well ventilated at Site 865 during the PETM (72), it does not preclude the possibility that an expanded oxygen minimum zone within the overlying water column contributed to the demise of local subbotinid populations by impinging upon their thermocline habitat.

Isotopic Filtering and Implications for the Future Health of Pelagic Calcifiers.

This case study demonstrates how isotopic filtering can be used to unmix microfossil assemblages that have been blended by sediment mixing processes, thereby enhancing the fidelity of fossil records chronicling transitory biotic responses to past episodes of rapid biogeochemical and climate change. Here, isotopic filtering reveals a previously unrecognizable biotic crisis among tropical planktic foraminifer communities at Site 865, as signaled by an abrupt decrease in local diversity during the PETM. Though no true extinctions are observed (73), extreme PETM warmth triggered a tropical exodus that profoundly altered the complexion of pelagic calcifying communities across the low latitudes (42, 44, 45). Our results thus highlight the sensitivity of tropical planktic foraminifer communities to the synergistic environmental stressors of extreme temperatures, ocean acidification, and deoxygenation at low latitudes during the PETM. Moreover, isotopic filtering could be readily extended to other abrupt events marked by rapid geochemical excursions recorded by marine biocalcifiers (e.g., other Eocene hyperthermal events, abrupt deglaciations) or terrestrial microfossils (e.g., pollen assemblages).

Planktic foraminifer communities have endured for nearly 170 million y and proven resilient in the face of past carbon cycle perturbations and episodes of ocean warming (74). Yet the rapidity of anthropogenic carbon emissions, which is ∼10 times faster than carbon emission rates during the PETM (75), will strain their adaptability to future environmental change. Continuing on the current emissions pathway without major policy interventions (SSP3-7.0) will result in a global temperature increase of ∼4 °C by the end of the century (1), while a similar magnitude of temperature change occurred over several millennia during the PETM (11–13, 45). Moreover, foraminifer boron isotope records indicate that surface-ocean pH dropped by ∼0.3 units during the PETM (15), whereas surface-ocean pH has already dropped by 0.1 units since preindustrial times and global ocean pH is projected to drop by an additional 0.3 to 0.4 pH units by the end of this century (76). Similarly, oxygen concentrations in the global ocean are projected to decrease by an additional 1 to 7% percent by the year 2100, beyond the 2% decrease observed since 1960 (77, 78). In short, the Anthropocene may pose the gravest threat to these prominent marine calcifiers since their near extermination by a cataclysmic asteroid impact some 66 million y ago (79).

Materials and Methods

ODP Site 865 is located at intermediate water depths (∼1,510 m) atop Allison Guyot (18°26’N, 179°33’W) in the Mid-Pacific Mountains (31). Benthic foraminifer assemblages suggest a midbathyal (∼1,300 m) setting during the late Paleocene (32), and δ¹³C records show that PETM records were recovered from holes 865B and 865C (32). Both PETM sections are part of a pelagic cap composed of pure calcareous foraminifer–nannofossil ooze deposited atop Allison Guyot (31). For correlative purposes, bulk-carbonate isotopic analyses (δ¹³C_bulk) were performed on a series of samples taken across both PETM records (Datasets S1 and S2). The two δ¹³C_bulk records feature comparable CIE magnitudes (∼1.5 to 2.0‰) and similar degrees of stratigraphic separation (∼10 to 12 cm) between the CIE onset and minimum (Fig. 1 C and D). Accordingly, the CIE onset and minimum in the δ¹³C_bulk records delimit the main body of the isotopic excursion and served as chronostratigraphic tie points, with intervening samples in the two sections being cross-correlated via linear interpolation assuming a constant sedimentation rate (SI Appendix, Table S1). The upper part of the 865B record falls within a coring gap, thus the gradual return to higher values (CIE recovery) seen in the 865C δ¹³C_bulk record is missing from the 865B record. Biostratigraphic correlation is based on the combined relative abundance of the excursion marker taxa (A. sibaiyaensis, A. africana, M. allisonensis), as determined from assemblage counts (>300 specimens per sample) carried out on the 865B planktic foraminifer assemblages using the same size fraction (>125 μm) employed by earlier studies (27, 30) to delineate taxonomic change in the 865C PETM record (Dataset S4). The two biostratigraphic records (>125 μm) are in good agreement, showing correlative abundance acmes for the excursion taxa (Fig. 1 E and F).

Foraminifer shells were gleaned from the calcareous ooze by soaking each bulk-sediment sample in a pH-buffered sodium hexametaphosphate hydrogen peroxide (30%v) solution and wet sieving the disaggregated sediment. Resulting coarse fraction (>63 μm) residues were rinsed with distilled water and oven dried (30 °C) overnight. Specimens used for single-shell stable isotope analyses were handpicked from the >250-μm sieve size fraction of the 865C samples (Dataset S3). Single-shell δ¹³C analyses (n = 548) were limited to the nine most abundant taxa to ensure statistically robust estimates of reworking in each taxon. Taxonomic coverage of previously published single-shell δ¹³C datasets (27, 30) was updated and expanded by including four additional taxa (A. esnaensis, A. esnehensis, M. aequa/subbotinae, and M. acuta), generating a new single-shell δ¹³C dataset for A. soldadoensis, and analyzing additional Subbotina spp. specimens. Isotope analyses conducted for this study (n = 313), in combination with published single-shell δ¹³C data (n = 235) from the same samples (27, 30), provided an average of ∼12 values per taxon per stratigraphic sample. All stable isotope measurements (Datasets S1–S3) were performed at the University of California, Santa Cruz Stable Isotope Laboratory using a ThermoScientific Kiel IV carbonate device interfaced with a ThermoScientific MAT 253 dual-inlet gas-source isotope ratio mass spectrometer. External analytical precision for δ¹³C on this instrument, as determined by replicate analyses of the calcite standard Carrera Marble, was ≤0.1‰ (±2 SD).

Standard point-counting methods (>300 specimens per sample) were used to compile census data on the nine study taxa (>250 μm) in the 865B PETM section (Dataset S5). Major taxa include A. sibaiyaensis, A. esnehensis, A. esnaensis, A. soldadoensis, M. allisonensis, M. acuta, M. aequa/subbotinae, M. velascoensis, and Subbotina spp. (SI Appendix, Fig. S2). The new abundance tallies for each taxon were isotopically filtered (corrected) by subtracting the proportion of reworked specimens indicated by their corresponding single-shell δ¹³C values from the 865C section. CIs (95%) on the proportions of reworked specimens were determined via nonparametric bootstrapping (Fig. 3) using the R package mosaic (version 1.8.3). The δ¹³C values of a taxon in a given sample were randomly resampled (1,000 iterations) and the SEs of the bootstrapped distributions computed (SI Appendix, Table S2). R code used for calculating proportions of reworked specimens and bootstrapped error estimates is provided in the supporting information. The few specimens (n = 3 of 548) showing intermediate δ¹³C values (between +2.0 and +3‰) were not used for isotopic filtering (i.e., they were not included in the per-species, per-sample estimates of the proportion of reworked and in situ taxa). All geochemical and micropaleontological data generated by this study (Datasets S1–S5) will be archived and made available via the National Oceanic and Atmospheric Administration National Center for Environmental Information Paleoclimatology database upon acceptance of the manuscript.

Data Availability

All study data are included in the article and/or supporting information.