We investigate early Eocene hyperthermals by complementing foraminiferal and bulk carbonate isotopes with benthic foraminiferal assemblages from three marine coreholes located along a paleoshelf transect on the New Jersey coastal plain (ODP 174AX Bass River, Double Trouble, and Ancora). Distinct negative δ13C and δ18O excursions likely correspond to the globally documented ETM-2, H2, I1, I2, and J events. Foraminiferal stable isotope data at Bass River reveal greater warming in benthic and thermocline communities compared to the surface dwellers during these excursion events. During the largest excursion event (ETM-2), thermocline-dwelling Subbotina not only experienced greater overall warming, but also recorded lower δ18O values than Morozovella (–5.1‰ vs. –4.3‰). This suggests either greater warming in the thermocline, habitat depth restructuring, or possibly a change in calcification season. We also demonstrate a potential biotic threshold, providing the first comprehensive evaluation of the sensitivity of shallow-marine taxa in response to these transient warming events.

The overall warming trend of the late Paleocene through early Eocene was punctuated by prominent, relatively short-lived warming events known as hyperthermals, the most extreme of which is known as the Paleocene-Eocene Thermal Maximum (PETM; ∼56 Ma; e.g., Kennett & Stott, 1991; Katz et al., 1999; Röhl et al., 2007; Zachos et al., 2007; Charles et al., 2011; Dickens et al., 2011). The PETM occurred amid both a 7-Myr warming trend (indicated by a decrease in δ18O) and a ∼4-Myr δ13C decline (Miller et al., 1987; Zachos et al., 2001; Cramer et al., 2009). During the PETM, global temperatures rose by ∼5–8°C (Kennett & Stott, 1991; Zachos et al., 2003; McInerney & Wing, 2011). The event was accompanied by the rapid adaptive radiation of terrestrial plants and mammals, hydrologic cycle acceleration, ocean acidification, and extinction of 40–60% of deep-sea benthic foraminifera (see Thomas, 2007, for summary). The abrupt PETM warming is linked to a massive injection of isotopically light carbon into Earth’s mobile carbon reservoirs, creating a negative ∼3–8‰ excursion in the stable carbon isotope (δ13C) record in both marine and terrestrial settings (Kennett & Stott, 1991; Koch et al., 1992; Bralower et al., 1997; Zachos et al., 2004, 2005; Schouten et al., 2007; Bowen et al., 2015). Although the exact source and magnitude of the 13C-depleted carbon released into the ocean-atmosphere system is still debated, hypotheses for the onset include methane dissociation (e.g., Dickens et al., 1995; Katz et al., 2001), organic carbon (e.g., Kurtz et al., 2003), extraterrestrial impact (Kent et al., 2003; Schaller et al., 2016), and volcanism (e.g., Storey et al., 2007; Gutjahr et al., 2017).

Following the PETM, a series of smaller negative carbon and oxygen isotope excursions (CIEs, OIEs) are found in numerous deep-sea sites, identified as H1/ETM-2 (or “Elmo”), H2, I1, I2, J, and K (or ETM-3 or X-event; Thomas & Zachos, 2000; Cramer et al., 2003; Lourens et al., 2005; Nicolo et al., 2007; Stap et al., 2009; Galeotti et al., 2010; Leon-Rodriguez & Dickens, 2010; Stap et al., 2010a, b; Westerhold et al., 2018). These distinct warming events primarily have been studied geochemically in deep-sea sections (e.g., using stable isotopes and bulk sediment carbonate content). This approach has provided information on the timing, duration, and magnitude of these events, yet our current understanding of the biotic impact of these post-PETM hyperthermals remains elusive (Turner & Ridgwell, 2013) because little work has been done on microfossil assemblage shifts associated with these δ18O and δ13C excursions (D’haenens et al., 2012), especially in shallow-water settings.

Early Eocene hyperthermal events share similar characteristics, such as increased temperatures associated with negative carbon isotope excursions, clay-rich layers often attributed to potential carbonate dissolution in bottom waters, and increased continental erosion due to perturbations in the hydrologic cycle (Stap et al., 2010a; Arreguín-Rodríguez & Alegret, 2016). Orbitally-tuned age models suggest that many of these hyperthermals share a similar origin (although not the PETM; Cramer et al., 2003) and were paced by variations in Earth’s orbit during periods of high eccentricity (both 100- and 405-kyr cycles) influencing the amount of insolation entering Earth’s climate system (Lourens et al., 2005; Zachos et al., 2010; Kirtland Turner et al., 2014; Littler et al., 2014; Westerhold et al., 2020). Phase lags between excursion events and maxima in eccentricity are documented at Site 1262 in the South Atlantic and are not constant, although they suggest a progressive weakening of the connection between climate and the carbon cycle from the late Paleocene into the early Eocene (Littler et al., 2014). Westerhold et al. (2018) suggests that not all hyperthermal events are necessarily fueled by the same carbon source (e.g., methane hydrate, permafrost, dissolved organic carbon) and that this is possibly related to their relationship with these orbital cycles.

The extent of the biotic response to these brief early Eocene hyperthermal events varies by site location (e.g., paleodepth, paleolatitude) and severity of environmental perturbation (i.e., specific hyperthermal event). Low benthic foraminiferal numbers, an increase in oligotrophic taxa, and low diversity are described at DSDP Site 401 (Bay of Biscay, NE Atlantic) at ETM-2 (D’haenens et al., 2012). Benthic and sedimentological records from Southeast Atlantic Sites 1263 and 1262 show a greater decrease in diversity and accumulation rates at the shallower Site 1263 across ETM-2, suggesting environmental changes were more pronounced at intermediate depths (Jennions et al., 2015). Planktonic foraminiferal studies across ETM-2, H2, and I1 from NE Italy indicate an increase in warm-indicator acarininds, with the most significant perturbations coinciding with the large ETM-2 event (D’Onofrio et al., 2016). During the PETM, significant changes in thermocline structure and calcification season have been proposed (Makarova et al., 2017) as well as depth migrations of surface-dwelling planktonics in response to extreme temperatures (>32°C; Si & Aubry, 2018). Similarly, shifts from oligotrophic to opportunistic, low-oxygen tolerant planktonic foraminiferal assemblages are documented during the Middle Eocene Climatic Optimum (MECO; ∼40 Ma) warming event (Luciani et al., 2010). Nannofossil reports from Pacific deep-sea Site 1209 show intense changes across the larger excursions (PETM, ETM2, I1) but do not display above background level changes associated with the minor events, implying that biotic responses to environmental perturbations are reached only at a critical threshold (Gibbs et al., 2012). Deep-sea foraminiferal studies indicate that the more intense hyperthermal events result in a greater impact on the biotic communities (e.g., Arreguín-Rodríguez & Alegret, 2016; Arreguín-Rodríguez et al., 2016, 2022; Thomas et al., 2018; Alegret et al., 2021), but more work is needed to evaluate the correlation between the excursions in δ18O and δ13C, critical temperature threshold, and degree of biotic perturbation, especially in shallow water settings.

The PETM and subsequent hyperthermal events have proven to be of great interest to the general public and scientific community as atmospheric, oceanic, and terrestrial changes associated with these events are seen as possible analogs to today’s state of global climate change. In this paper, we present high-resolution stable isotope records coupled with faunal studies across hyperthermal events at three paleoshelf sites from the U.S. mid-Atlantic coastal plain, providing us with the opportunity to reconstruct the response of foraminifera to environmental perturbations associated with transient warming events in the early Eocene. Comparison of planktonic stable isotope results with benthic foraminiferal assemblages provide a more complete picture of the response of biota to the excursion events. This provides the first comprehensive evaluation of the sensitivity of shelf marine faunas to these smaller hyperthermals.

ODP 174AX

Samples were obtained from Ocean Drilling Project (ODP) sites Bass River (Miller et al., 1998), Double Trouble (Browning et al., 2011), and Ancora (Miller et al., 1999), drilled as part of the New Jersey Coastal Plain Drilling Project (NJCPDP), Leg 174AX (Fig. 1). Our studied interval is primarily from Sequence E2 (∼53–54 Ma, GTS2012) from the Lower Eocene Manasquan Formation, which is dominated by calcareous silty clays (“marls”; Miller et al., 1998, 1999; Browning et al., 2011). The lower Eocene section at these sites consists mainly of silty clays deposited in middle to outer neritic (30–200 m) paleodepths, with sequence E2 deposited in the deepest waters of the Eocene (Miller et al., 1998, 1999; Browning et al., 2011). Bathymetric zonations are divided into inner neritic (0–30 m), middle neritic (30–100 m), and outer neritic (100–200 m; Van Morkhoven et al., 1986) zones. Sequence E2 at Ancora (the most updip site) displays greater lithologic variation than coeval downdip sites and contains highstand deposits of fine quartz and reworked glauconite sand (identified as reworked by covariance with quartz sand) throughout the Manasquan Formation (Miller et al., 1999). Within E2, Bass River shallows from 185 to 155 m paleodepth (Van Sickel et al., 2004), Double Trouble shallows from 155 to 125 m, and Ancora shallows from 145 to 115 m, based on a 1/750 m paleoslope model following Esmeray-Senlet et al. (2015). Bass River, Double Trouble, and Ancora were chosen because they contain early Eocene hyperthermal events recorded in Sequence E2 and allow us to compare sites along a paleodepth gradient.

Stable Isotopes

Bulk Sediment Calcium Carbonate

Bass River samples were taken every 0.25 ft (∼7.5 cm) for bulk isotopes (δ13C and δ18O) for a total of 122 samples and every ∼0.5 ft (15 cm) at Double Trouble and Ancora (69 and 86 samples, respectively). Samples were taken from the inner part of the core to avoid the edges, which are in contact with the core liner and potentially are affected by mud rind contamination. Samples were analyzed at the Stable Isotope Laboratory in the Department of Earth and Planetary Sciences at Rutgers University using a Micromass Optima Dual-Inlet mass spectrometer. Crushed samples were reacted with phosphoric acid at 90°C for 15 minutes. Stable isotope values are reported versus V-PDB by analyzing NBS-19 and an internal laboratory reference material during each automated run. Reference material is continuously calibrated against NBS-19, with a reference-NBS-19 offset of ±0.04‰ and ±0.10‰ for δ18O and δ13C, respectively. Results are reported relative to the VPDB standard. The laboratory standard error (1σ) is ±0.08‰ for δ18O, and ±0.05‰ for δ13C.

Benthic and Planktonic Foraminifera

Obvious changes in benthic foraminiferal assemblages across hyperthermals led us to evaluate the water column during these warming events by analyzing benthic and planktonic foraminiferal stable isotopes. The Bass River site was chosen because it is the deepest of the three sites, providing a sample of the entire water column and complete integration of the mixed layer and thermocline that may be otherwise missing at the more updip sites.

Stable isotope analyses of planktonic foraminifera were conducted on specimens of surface-dwelling Morozovella (M. aequa, M. subbotinae, M. gracilis, M. marginodentata, M. lensiformis, M. formosa, M. crater) and thermocline-dwelling Subbotina (S. patagonica, S. roesnaesensis) from 49 sample depths within Sequence E2 at Bass River. Planktonic species are identified using taxonomy of Pearson et al. (2006). A monogeneric record (i.e., Morozovella spp. and Subbotina spp.) is presented due to the uneven distribution of single species throughout the studied interval. Although this approach has been used in previous studies (Babila et al., 2016; Makarova et al., 2017), we acknowledge the possibility for inter-species offsets in the δ18O and δ13C of morozovellid and subbotinid species. Benthic foraminiferal stable isotopes were conducted on specimens of Cibicidoides pippeni to constrain the isotopic signals of shelf bottom water and to complement planktonic stable isotopes analyses. Analyses comprised approximately 10 (planktonics) and 5 (benthics) individual specimens solely in the 250–425-μm size-fraction to minimize size offsets (e.g., Birch et al., 2012). Only well-preserved transparent to translucent specimens were used and were briefly sonicated in distilled water to remove clays. Samples were analyzed at the Stable Isotope Laboratory in the Department of Earth and Environmental Sciences at Rensselaer Polytechnic Institute using the Isoprime-100 dual-inlet stable isotope ratio mass spectrometer. Stable isotope values are reported versus V-PDB by analyzing NBS-19 and an internal laboratory standard (Carrara Marble) during each automated run. Analytical precision (1σ) is 0.02‰ for δ18O and 0.01‰ for δ13C on these runs, and the long-term laboratory precision (1σ) is ±0.04‰ for δ18O, and ±0.02‰ for δ13C on NBS19.

Paleotemperatures were estimated using δ18Oforaminifera with Cibicidoides spp. yielding bottom water temperatures (BWT), Subbotina yielding intermediate thermocline sea surface temperature (SST), and Morozovella yielding the mixed layer SST reconstructions. We use δ18Oforaminifera as a relative indicator of ocean temperature. In order to compare our study to Makarova et al. (2017), we use δ18Oforaminifera only and assume 0.22‰ corresponds with a 1°C temperature change. Although we are most interested in establishing temperature changes, this approximation does not account for transient shifts in δ18Oseawater (which can influence δ18Ocarbonate), and we acknowledge that this is a potential limitation to our temperature estimates. Due to the location of our sites, an increase in continental runoff may cause changes in δ18Oseawater. This is a possible scenario, especially when an enhanced hydrological cycle accompanied these hyperthermal events.

Faunal Studies

Samples at Bass River, Double Trouble, and Ancora were taken every 1 ft (30 cm) for microfossils (27, 27, and 30 samples, respectively). In total, 84 samples among the three sites were taken for quantitative benthic foraminiferal faunal analysis. Samples 20 cm3 were soaked overnight in a sodium-metaphosphate solution made with deionized water (5.5 g/l), washed with tap water through a 63-µm sieve, and then oven-dried at ∼50°C overnight. Samples were weighed before and after washing to determine the weight percent coarse fractions (CF; >63 µm) to aid in paleo-depth interpretations. An increase in mud (<63 μm) generally indicates greater water depths. A microsplitter was used to obtain splits of ∼200 benthic foraminiferal specimens for quantitative analysis and sieved to acquire the >150-µm fraction, consistent with studies on the margin used for comparison (Browning et al., 1997; Charletta, 1980; Miller & Katz, 1987; Streeter & Lavery, 1982). Specimens were picked from the >150-µm size-fraction to ensure only adult tests were obtained. This approach was employed with the intent to limit the degree of uncertainty due to morphologic variability of juvenile populations. We acknowledge that some benthic foraminiferal taxa, such as the buliminids and bolivinids, can be smaller than the 150-µm size-fraction even as adults. There is the potential for some paleoenvironmental information loss due to the small sizes of these infaunal taxa. All benthic foraminifera in each sample split were mounted on microslides and identified to the species level (except genera such as Astacolus spp. and Lenticulina spp.). A subgrouping titled “unknown benthics” is included for specimens that are not identifiable due to preservation state. The faunal data were used to determine dominant species and quantitative, and statistical analyses were conducted to establish biofacies relationships and trends. Taxonomy from Howe (1939), Bandy (1949), Enright (1969), Jones (1983), Tjalsma & Lohmann (1983), Boersma (1984), Van Morkhoven et al. (1986), and Stassen et al. (2015) was used to identify the benthic foraminiferal species in each sample. Species were also compared to type slides and assemblage slides from Browning et al. (1997) and Charletta (1980). All planktonic foraminifera in each sample split were counted to determine planktonic foraminiferal abundance (relative to total foraminiferal abundance).

We determined the benthic foraminiferal abundances, planktonic foraminifera vs. total foraminifera ratio (%P), benthic foraminifera per gram of bulk sediment, planktonic foraminifera per gram of bulk sediment, % agglutinates vs. total benthics, proportion of epifaunal/infaunal morphotypes, diversity indices [Shannon-Wiener index: H(s), Dominance: D, Fisher alpha: F(α)], and multivariate analyses [e.g., Principal Component Analysis (PCA) and Detrended Correspondence Analysis (DCA); Appendix 1]. Agglutinates and calcareous taxa are described in Appendix 2, and infaunal and epifaunal classifications are summarized in Appendix 3. Both PCA and DCA were calculated on relative abundances of the most common taxa (>2.5% in at least one sample). The DCA was used to help identify the environmental variables that may have controlled or contributed to the distribution pattern of benthic foraminifera. Diversity indices were calculated using initial data sets, including all counted benthic specimens. Diversity indices, cluster analysis, PCA, and DCA were all performed using the PAST 3.13 software (Hammer, 2001).

Foraminifera are widely used as paleoenvironmental indicators, and benthic foraminifera that inhabit the shelf reflect environmental conditions that vary proportionally with depth (e.g., Natland, 1933; Douglas, 1979; Olsson & Wise, 1987; Speijer et al., 1996; Pekar et al., 1997; Sen Gupta, 1999; Leckie & Olson, 2003; Katz et al., 2003, 2013; Stassen et al., 2012a, b, 2015). The availability of oxygen and food are major factors controlling the distribution and abundance of benthic biofacies, although the major factor in shallow water settings is oxygen because food is abundant (Jorissen et al., 1995, 2007). The proportion of epifaunal (those that live at the sediment-water interface) versus infaunal (within the sediment) morphotypes can be used as a proxy for oxygenation and food supply (Rathburn & Corliss, 1994; Jorissen et al., 1999) and are based on test shape (e.g., Jones & Charnock, 1985; Corliss & Chen, 1988). Infaunal taxa become more prominent as the organic flux to the seafloor increases, and an increase in infaunal populations is observed under low-oxygen bottom conditions (Jorissen et al., 2007). The ratio of foraminifera with agglutinated versus calcareous tests can be used to measure the degree of carbonate dissolution, as they are dissolution-resistant taxa (Arreguín-Rodríguez et al., 2016; Nguyen et al., 2009). Percent planktonics (%P) is also used as a proxy for assessing dissolution as well as water depth (Grimsdale & Van Morkhoven, 1955; Nguyen et al., 2009).

Age Models and Correlation Between Sites

We use an updated age-depth diagram using biostratigraphic data to develop the age model for Sequence E2 at Bass River, placing E2 at 53.4 to 53.9 Ma (Fung et al., 2019. We apply this age model to an astronomical record using Analyseries software (Laskar et al., 2004). Age control for Double Trouble and Ancora are based on Browning et al. (2011) and Van Sickel et al. (2004), placing Sequence E2 ∼54–53 Ma. We use biostratigraphic and sedimentologic data to correlate Double Trouble and Ancora to Bass River. Specifically, the Trifarina wilcoxensis record (discussed below) exhibits a unique abundance record which helps tie the three sites together.

Bulk Stable Isotopes, %P, % Agglutinates, Morphotypes, and Diversity Indices

Hyperthermal events are recognized by a coincident rapid decrease in δ13Cbulk and δ18Obulk isotopes (Fig. 2). Individual excursion events are highlighted (gray shaded bars), and are based on the onset of the CIEs, similar to previous reports (e.g., Cramer et al., 2003; Zachos et al., 2010). Depths of individual hyperthermal events can be found in Appendix 4, and a figure showing δ13Cbulk and δ18Obulk records for all three sites is shown in Appendix 5. Distinct excursion events have been identified and the inferred hyperthermal events (ETM-2/H1/H2/I1/I2/J) described in Cramer et al. (2003), Lourens et al. (2005), Röhl et al. (2005), Agnini et al. (2009), and Galeotti et al. (2010) are labeled stratigraphically based on location within calcareous nannofossil Zones NP11 and NP12. A lack of magnetostratigraphic framework limits unequivocal correlations at this time. Similar to D’haenens et al. (2012), we assign α and β to observed excursions below ETM-2 at Bass River. Due to poor core recovery from 508–515 ft (154.8–156.9 m) at Ancora, hyperthermal event H2 is likely missing.

The carbon and oxygen stable isotope records at all three sites are comparable to records of hyperthermal events previously published from deep-sea sites (Table 1). At Bass River, values vary between –1.5 and 0.13‰ for δ13Cbulk and –3.1 and –1.9‰ for δ18Obulk. At Double Trouble, values vary between –2.1 and 0.21‰ for δ13Cbulk and –3.4 and –1.7‰ for δ18Obulk. At Ancora, values vary between –1.5 and 0.19‰ for δ13Cbulk and –3.6 and –1.7‰ for δ18Obulk. During the largest excursion event (ETM-2), bulk carbonate isotopes at Bass River, Double Trouble, and Ancora exhibit maximum average amplitude decreases of 0.63‰, 1.62‰, and 0.47‰ in δ13Cbulk and 0.44‰, 0.25‰, and 0.82‰ in δ18Obulk, respectively. Therefore, the two most updip sites, Double Trouble and Ancora, exhibit the greatest δ13C and δ18O changes. For comparison, the average amplitude of bulk carbonate δ13C change before and after the PETM CIE at Bass River is 3.46‰ (John et al., 2008). Similarly, bulk carbonate isotopes across the PETM from New Jersey coastal plain sites (Bass River, Ancora, Wilson Lake, and Clayton) vary strongly among sites (Fig. 2), with the largest excursions occurring at the most updip sites (Stassen et al., 2012b; Wright & Schaller, 2013). Shallow-water sites have the potential for increased riverine inputs bringing in low δ13C dissolved inorganic carbon, which could influence stable isotope records, increasing the amplitude particularly in the sites closest to shore.

Percent planktonics, % agglutinates, % epifaunal, diversity indices (Shannon-Wiener index, Dominance, Fisher alpha), benthics/gram, and planktonics/gram show changes associated with these excursions (Fig. 2). Overall, %P decreases with excursion events and is most consistent at Bass River and Ancora. The %P at Bass River remains high (average = 89%; standard deviation (S.D.) = 6.8; range = 69–98%), with an obvious decrease within H2. The %P is high at Double Trouble (average = 88%; S.D. = 6.8; range = 74–97%), with decreases at ETM-2, I1, and I2, though values increase in H2. The %P at Ancora varies (average = 38%; S.D. = 30; range = 0–82%) with falls in %P across three of the four events. At Ancora, %P drops to from 81% to <10% within ETM-2, decreases from 56% to <20% within event I1, and remains at 0% throughout event I2. The total number of benthics per gram of sediment (benthics/g) and total number of planktonics per gram of sediment (planktonics/g) show fluctuations across excursion events, although these changes are not consistent across all events. At Bass River, both benthics/g and planktonics/g reach some of the lowest values recorded during the ETM-2 event. At Double Trouble, minimum benthics/g and planktonics/g values are reached during the I1 event, and values fall across the I2 event. At Ancora, we observe a fall in both benthics/g and planktonics/g, although a clear increase in both metrics is noticed across the J event.

Calcareous taxa strongly dominate throughout the studied intervals at all three sites (Fig. 2; see Appendix 2 for classification). The % agglutinates remain relatively unchanged except for at key excursion events; agglutinate taxa clearly increase in abundance at ETM-2 (all three sites) and at I2 at Ancora. At Bass River, this increase in agglutinates occurs within the upper half of ETM-2 and continues rising until the following H2 event. Agglutinated taxa then decrease within the H2 excursion event at Bass River. Ancora displays the largest variation in agglutinated taxa with obvious increases at ETM-2 and I2; at excursion events ETM-2 and I2, the agglutinates reach 70% and 59% respectively.

Epifaunal morphotypes are slightly more abundant at all three sites (Fig. 2; average = 60%; see Appendix 3 for classification). Changes in the % epifaunal taxa are most consistent at the Bass River and Ancora sites. At Bass River, epifaunal taxa visibly decrease within events ETM-2 and H2. At Double Trouble, epifaunal taxa decrease across the H2 event, yet increase in I1 and I2 events. At Ancora, epifaunal taxa decrease within excursion events ETM-2 and I2. Turrilina robertsi (average = 2.0%), Trifarina wilcoxensis (average = 5.7%), Pseudoclavulina spp. (average = 6.0%), and Lenticulina spp. (average = 7.4%) are the most abundant infaunal species across the three sites. Cibicidoides micrus (average = 17%), C. cf. pseudoungerianus (average = 11%), C. cocoaensis (average = 3.5%), C. eocaenus (average = 3.9%), Hanzawaia ammophila (average = 3.0%), and Osangularia expansa (average = 3.8%) are the most abundant epifaunal species (Fig. 3).

Trends in diversity indices across excursion events are most consistent at Double Trouble and Ancora (Fig. 2). The heterogeneity [Shannon-Wiener index: H(s)] ranges from 1.1 to 3.0 at Bass River and Dominance (D) ranges between 0.06 and 0.63. At Bass River, H(s) remains relatively constant at events α and β, decreases followed by an increase at ETM-2, and decreases within H2. Dominance increases at ETM-2 (although slightly), and H2 events at Bass River and F(α) values share similar trends to H(s). At Double Trouble, H(s) ranges from 1.8 to 2.9 and D ranges from 0.07 to 0.28; Dominance increases while H(s) and F(α) decrease across excursion events. At Ancora, H(s) ranges between 1.3 and 2.8, while D values range between 0.08 and 0.49. At Ancora, we observe increases in D and decreases in both H(s) and F(α) across excursion events.

Diversity indices can be used to measure environmental stability with faunal communities considered stable when the Shannon H(s) index falls between 2.5–3.5, in transition when between 1.5–2.5, and stressed when below 1.5 (Magurran, 1988; Patterson & Kumar, 2000; Roe & Patterson, 2014). At Bass River, 81% of samples remain above stable levels [i.e., Shannon H(s) >2.5], with transition level values occurring at the ETM-2 event and transition-stressed conditions in the H2 event. At Double Trouble, 59% of samples are considered stable, with transition levels occurring at H2 and I1. At Ancora, only 23% of samples are considered environmentally stable. All excursion events are either in transition or stressed, with stressed levels occurring at the two largest CIEs (ETM-2 and I2) recorded at Ancora.

Faunal Studies: PCA and DCA

A total of 75, 58, and 53 species are identified from Bass River, Double Trouble, and Ancora, respectively. From all three sites, a total of 41 species are considered significant (>2.5% in at least one sample) and are used for faunal studies (see Appendix 6 for a taxonomic list of species). At Bass River, Double Trouble, and Ancora, specimens placed into the “unknown benthics” category average <1%, 1%, and 3%, respectively. Four components from PCA explain 81% of faunal variation from a total of 84 samples (Fig. 4; Appendix 7). Each principal component (PC) is characterized by taxa loadings, that is how much each original variable (= taxa) contributed, either positively or negatively, to the component. Curves for each PC express how each sample scored along that corresponding PC axis moving up through the stratigraphy of each studied section (Fig. 4). Benthic foraminiferal abundances of most common taxa present in PC1–4 at each site show changes associated with excursion events (Fig. 4).

The first component, PCA-1, explains 43% of total variance and is characterized by C. micrus (loading: 160) and C. cf. pseudoungerianus (loading: 76). Cibicidoides micrus increases in abundance across many excursion events at Bass River, Double Trouble, and Ancora, although not across I2 at Ancora, ETM-2 at Double Trouble, or H2 at Bass River. Cibicidoides cf. pseudoungerianus is not consistent across all three sites and increases drastically in abundance at ETM-2 at Bass River and I2 event at Double Trouble.

The second component, PCA-2, explains 16% of total variance and is dominated by a high loading of Trifarina wilcoxensis (loading: 123). We observe an increase in abundance of this species at the base of ETM-2 at Ancora, the base of H2 at Double Trouble, and near the top of H2 at Bass River. This increase (up to ∼80% at Bass River) occurs following the largest excursion events at all three sites.

The third component, PCA-3, explains 13% of total variance and is characterized by Pseudoclavulina spp. (loading: 92) and C. cocoaensis (loading: 40). Pseudoclavulina spp. increases in abundance at many excursion events across all three sites, although not across H2 at Bass River, I1 at Double Trouble, or J and I1 at Ancora. At Ancora, C. cocoaensis clearly increases across excursion events I1 and I2, yet is less consistent at Bass River and Double Trouble.

The fourth component, PCA-4, explains 8.3% and is characterized by C. cf. pseudoungerianus (loading: 45), Pseudoclavulina spp. (loading: 43), and C. cookei (loading: 17). At Bass River, C. cookei decreases in abundance across α and ETM-2 events. At Double Trouble, C. cookei decreases in abundance across H2 and I1 events, and at Ancora, C. cookei decreases in abundance across all events except for J.

R-mode DCA

R-mode DCA with interpreted environmental conditions are visualized in a bivariate plot of sample scores on the first and second axes (Fig. 5). Known environmental states of select benthic foraminifera (Appendix 8) help with reconstructions at our sites. Based on these characteristics, we infer axis 1 (x-axis) to represent bottom-water oxygenation (Fig. 5). Oxic species (e.g., Cibicidoides howeii, T. robertsi, Neoeponides lotus) can be differentiated by R-mode DCA from suboxic-tolerant taxa (e.g., T. wilcoxensis and Anomalinoides acuta). Organic flux levels are represented by axis-2 (y-axis), with food availability increasing upwards. Mesotrophic taxa are placed near the top (e.g., A. aragonensis), whereas oligotrophic taxa fall towards the bottom half (e.g., Oridorsalis plummerae). Based on PETM work by Stassen et al. (2015) and the inferred ecological preferences (Appendix 8), we place benthic foraminiferal taxa into three main groups that represent identifiable biofacies. Although not all taxa have distinct paleoecologic affinities, these groupings allow us to interpret overall paleoecologic conditions at our sites (Fig. 5).

Foraminiferal Stable Isotopes

Planktonic and benthic foraminiferal stable isotopes are measured on samples within sequence E2 and are focused across the ETM-2 event at Bass River 1122–1116 ft (342.0–340.2 m), where we observe the greatest negative excursions in bulk isotopes (Figs. 6a, 7; Tables 23). In general, Subbotina exhibits higher δ18O values than Morozovella and Morozovella δ13C values are consistently higher than Subbotina values, as expected for cooler and lower oxygen intermediate waters. Cibicidoides pippeni exhibits the highest δ18O values reflecting lower seafloor temperatures on the shelf and δ13C values are comparable to Subbotina. Although we observe these offsets among the three taxa, Morozovella, Subbotina, and C. pippeni follow comparable trends throughout the studied interval. During the excursions (α, β, ETM-2, H2), planktonic foraminifera results show larger negative δ13C excursions recorded in surface-dwelling Morozovella versus thermocline-dwelling Subbotina. In contrast, Subbotina exhibits larger negative δ18O excursions than Morozovella. Furthermore, during excursion events, Subbotina δ18O values attain comparable values to Morozovella. For example, leading into the α event, Morozovella and Subbotina δ18O values are offset by 1.2‰. At the peak excursion event, Morozovella and Subbotina δ18O values both reach –3.9‰. Comparable δ18O relationships are observed at β, ETM-2, and H2 events. On the other hand, the δ13C offset leading into the α events is 2.0‰ and is 1.4‰ within the excursion event. The δ13C offset between the two genera remains relatively consistent throughout the studied interval and indicates no consistent change in offset during the excursion events.

Higher-resolution planktonic and benthic foraminiferal stable isotope results across the ETM-2 event reveal potential changes in water column dynamics during this larger hyperthermal event (Figs. 6a, 7). A negative δ18O excursion in Morozovella coincides with the decrease in δ18Obulk at the base of ETM-2, and then is stable in the upper half of ETM-2. Slight decreases in Subbotina and C. pippeni δ18O values occur at the base of ETM-2, and then values decrease notably starting at 1118.8 ft (341.0 m), near the peak Morozovella and bulk sediment δ18O excursion values. Within ETM-2, Morozovella, Subbotina, and C. pippeni exhibit peak δ18O excursion values of –4.3‰ (at 1118.3 ft), –5.1‰ (at 1118.2 ft), and –4.0‰ (at 1118.2 ft), respectively. This is followed by simultaneous increases in Morozovella, Subbotina, and C. pippeni δ18O, although Subbotina values remain low, and the same taxa offsets recorded prior to ETM-2 are not observed. Synchronous CIEs occur in the Subbotina and C. pippeni records with an onset concurrent with the Subbotina and C. pippeni δ18O excursions. Morozovella appears to increase slightly at the base of ETM-2, with the negative CIE lagging Subbotina and C. pippeni. Within ETM-2, Morozovella, Subbotina, and C. pippeni reach peak δ13C excursion values of –1.1‰ (at 1118.3 ft), –3.0‰ (at 1118.0 ft), and –2.3‰ (at 1118.0 ft), respectively. Concurrent returns to near pre-excursion δ13C and δ18O values are recorded in Morozovella and C. pippeni. In contrast, Subbotina δ13C values remain low (with only a gradual increase) and do not return to pre-excursion levels until the top of E2. The average carbon isotopic gradient between surface dwellers and benthic foraminifera below the ETM-2 CIE onset is ∼1.7‰. Within the core of ETM-2, the average δ13C surface to bottom water difference increases slightly to ∼2.1‰. In contrast, we observe a reduction in the δ18O vertical gradient, from ∼1.2‰ below the ETM-2 excursion to ∼0.8‰ within the core of the event.

We focus on establishing temperature anomalies instead of absolute temperature values due to δ18Oseawater changes through time. Sea surface temperature (SST) reconstructions from Morozovella and Subbotina and bottom water temperature (BWT) reconstructions from C. pippeni across the ETM-2 event at Bass River show an average increase of 1°C for the surface mixed layer, 2°C for the thermocline, and 3°C for bottom water conditions (Table 2). Therefore, average temperature changes from pre-excursion to the excursion event are higher for C. pippeni (3°C) compared to Subbotina and Morozovella (2 and 1°C, respectively). Interestingly, Subbotina exhibits the lowest δ18O values (i.e., warmest temperatures) recorded within the peak of ETM-2, followed by Morozovella and C. pippeni.

Bulk carbonate δ18O largely covaries with δ13C, yet during ETM-2 the Morozovella δ18O excursion onset slightly precedes the carbon isotope excursion and then recovers more quickly. This discrepancy in timing may reflect particle size-dependent sediment mixing where fine-fraction carbonate (e.g., nannofossils) are more intensely mixed than larger foraminifera shells (see Hupp et al., 2019). This could result in the δ18Obulk and δ13Cbulk records to be more intensely smoothed compared to the δ18Oforaminifera and δ13Cforaminifera records, leading to the observed divergence. A similar scenario is observed across the largest hyperthermal event ETM-2 at deep-sea Site 401 (Bay of Biscay, North Atlantic; D’haenens et al., 2012).

Orbital-Scale Forcing

The Bass River age model provides sufficient detail to observe potential orbital-scale forcing mechanisms by comparing the bulk δ13C results with eccentricity (Fig. 6b). The largest CIE at 1120 ft (∼341.4 m) is in line with 405- and 100-kyr eccentricity highs and is in agreement with the ETM-2 event observed globally at this time (Cramer et al., 2003; Lourens et al., 2005). It appears that the bulk δ13C excursions of ETM-2 and H2 events align with eccentricity maxima, whereas excursions below ETM-2 (i.e., α and β events) more closely correspond with eccentricity minima.

Faunal Changes – Benthic Foraminifera

During hyperthermal events (as noted by negative excursions in δ13C and δ18O), we observe fluctuations in % planktonics, % agglutinates, % epifaunal taxa, heterogeneity [H(s)], dominance (D), diversity F(α), benthics/gram, planktonics/gram, and grain size at Bass River, Double Trouble, and Ancora (Fig. 2), yet the faunal response across sites and within a site to different events is not uniform. This is visualized in Fig. 4 where no one principal component describes all identified hyperthermal events at the three sites. The varied biotic response to these hyperthermal events suggests the presence of some threshold for environmental perturbation. We discuss two possible factors influencing this threshold idea: 1) magnitude of event and 2) distance of site from shore.

Although we do observe a faunal response to these excursion events, several of the smaller excursions (e.g., α and β) do not show significant faunal variability, indicating that the magnitude of the hyperthermal, including duration and rate, influences biotic responses. For example, the % agglutinates value remains relatively unchanged throughout our record, except for unmistakable abundance increases at the largest excursion event ETM-2 at Ancora and Double Trouble and a decrease followed by an increase at Bass River (Fig. 2). In addition, there is little to no change in diversity indices [H(s), D, F(α)] across the smallest excursions at Bass River and Double Trouble indicating the influence of different magnitude events. Furthermore, a peak in Aragonia aragonensis solely occurs following the largest excursion event at Bass River (ETM-2) (Fig. 3). Interestingly, at Double Trouble the increase in A. aragonensis does not appear to occur within an excursion event, and the taxon is not present at the shallowest site, Ancora. Aragonia aragonensis is found in neritic to bathyal deposits after the PETM and is considered a hyperthermal marker taxon (Thomas, 1998, 2003). It is possible that A. aragonensis only serves as an indicator for warming events in deeper settings and/or during maximum excursion events. These results suggest that only the largest events cause observable and/or significant biotic turnover and that reduced magnitude events show less consistent faunal fallouts. The subdued faunal response to the smaller excursion events could be the result of an inherent biotic threshold and/or recovery from preceding larger warming events.

The three paleoshelf sites sit along a depth transect, allowing us to evaluate the extent to which water depth influences the impact of an environmental perturbation on foraminiferal communities. Faunal changes across hyperthermal events are most apparent at the more updip sites, Ancora and Double Trouble, indicating greater environmental perturbations in shallower settings (Fig. 2). Trends in diversity indices across all hyperthermal events are most consistent at shallowest sites Double Trouble and Ancora. At the shallowest site (Ancora), all excursion events exhibit environmental stability values that represent in-transition or stressed levels, whereas diversity indices are less obvious at the deepest site, Bass River. Faunal communities are considered stressed when the Shannon H(s) index falls below 1.5 (Magurran, 1988; Patterson & Kumar, 2000; Roe & Patterson, 2014). The only event where H(s) values dip below 1.5 is during the largest event (ETM-2) and at the shallowest site, Ancora. Gibbs’ et al. (2012) nannofossil record at deep-sea Site 1209 suggests that the input of carbon and associated climate change must exceed a threshold in order for a biotic response to occur during these early Eocene hyperthermals. A biotic sensitivity threshold shows calcareous nannoplankton assemblage changes associated with CIEs of >0.6‰ or 2°C, although the sensitivity for different planktic groups varies (e.g., dinoflagellates, planktonic foraminifera; Gibbs et al., 2012). An important takeaway from our study is that not all hyperthermal events, specifically smaller magnitude events farther from shore, will exhibit foraminiferal changes and are therefore not always identifiable by faunal studies alone.

The distribution of taxa delineated by DCA (Fig. 5) and our corresponding paleoenvironmental interpretations help identify potentially important hyperthermal marker taxa for the paleoshelf. The overall low diversity and decrease in diversity and heterogeneity across specific events indicates intervals of periodic stress for shallow-water biota associated with the environmental perturbations, which is also noted across the PETM at Bass River (Stassen et al., 2012a, b). Specifically, we find peaks in stress-tolerant C. micrus (Stassen et al., 2015) at some excursion events, including at the largest excursion event (ETM-2) at Bass River (Fig. 3). The increase in dominance at excursions indicates that opportunistic taxa are associated with these transient warming events in the early Eocene. Opportunistic taxa include Anomalinoides acuta (Gibson et al., 1993; Gibson & Bybell, 1994; Stassen et al., 2012a, b) and Aragonia aragonensis (Thomas, 1998, 2003). Aragonia aragonensis peaks from 0 to 10% at ETM-2 at Bass River but not elsewhere; our results support the contention that it responds to transient food-rich environments (see Figs. 3, 5) and is a potential hyperthermal marker taxon, at least in some settings (e.g., Steineck & Thomas, 1996; Alegret et al., 2009; Nguyen et al., 2009; Ortiz et al., 2011). Evidence of increased surface-water eutrophication during periods of intense warmth, due to enhanced weathering and supply of nutrients, is also documented through planktonic foraminiferal hyperthermal studies (D’Onofrio et al., 2016). Following excursion events, we observe recovery-phase biofacies with an increase in the well-oxygenated indicator taxon N. lotus (Gooday & Rathburn, 1999; Jorissen, 1999; Duchemin et al., 2007). Cibicidoides cookei increases following excursion events, indicating that it also belongs in the recovery group. Diversity indices increase (and dominance decreases) following excursion events, consistent with a period of biotic recovery.

The increases in abundances of agglutinates and infaunal taxa during some excursion events could result from an increase in food delivery and low bottom-water oxygenation. This is further supported by an increase in dysoxic taxon T. wilcoxensis (Stassen et al., 2012a, b) at the H2 excursion event at Bass River and Double Trouble (Fig. 3). The increase in T. wilcoxensis across excursion events has been noted across the PETM at Bass River and Wilson Lake (although called Pseudouvigerina wilcoxensis; Stassen et al., 2015).

Benthic and Planktonic Foraminiferal Isotopes

When compared to PETM paleotemperature reconstructions at Bass River (John et al., 2008) and Millville (Makarova et al., 2017), our average temperatures and magnitude of temperature change are lower than the PETM, as expected (Table 2). Specifically, we observe an average 1–3°C increase across ETM-2 compared to an increase of 4–7°C at Bass River and 5–9°C at Millville across the PETM (John et al., 2008; Makarova et al., 2017). Although ETM-2 represents the second largest Eocene hyperthermal event, far less is known in terms of climatic change compared to the PETM, with fewer studies to make comparisons and temperature reconstructions limited to open ocean sites. For example, subtropical sea surface reconstructions from planktonic records across the ETM-2 indicate an increase of 2–4°C (Harper et al., 2017). Although our study suggests somewhat larger maximum temperature changes (3–7°C), PETM studies show that at shallow continental sections, such as Bass River and Millville, the CIE tends to be a few per mil larger than what is recorded in marine sections (John et al., 2008) and that warming on the shelf is higher than observed for open ocean low-to-middle latitude sites (Makarova et al., 2017), verifying that our temperature estimates are indeed reasonable. This further demonstrates the importance of evaluating these hyperthermal events on the shelf, where records are frequently expanded by high sedimentation rates, and are more complete (although deposition is variable). Open-ocean, pelagic δ18O and δ13C records are potentially muted by low sedimentation rates, carbonate dissolution, and seafloor diagenesis during these warming events, which act as a low-pass filter for high-amplitude events.

We attribute the obvious Morozovella and Subbotina offset in δ18O followed by comparable values during excursion events to three plausible scenarios (Figs. 6a, 7). Planktonic foraminiferal isotope results indicate that either: 1) the thermocline genus (Subbotina) experienced greater warming, possibly through a deepening of the mixed layer; 2) the surface-dwelling genus (Morozovella) migrated to the thermocline and cohabitated similar water depths with Subbotina; or 3) the taxa built their test during different seasons during the perturbation. In the first scenario, heat from warm surface waters propagates down, creating a more homogenous warm mixed layer. In this case, the two genera continue to live at different water depths but grow in comparable ocean temperatures. In the second situation, surface-dwelling taxa experienced too much warming and were unable to survive in the surface waters. As a result, they migrated deeper into the water column to seek refuge and shared environments with thermocline-dwelling genera or resituated calcification to a cool season. In this habitat restructuring, Morozovella and Subbotina precipitated and grew their tests in similar water temperatures. It is also possible that the surface and intermediate dwellers changed their seasons of calcification, with high summer surface waters precluding calcification.

The fact that comparable δ18O values are attained at multiple event levels at Bass River (α, β, ETM-2, H2) suggests that this is not a singular phenomenon. Furthermore, comprehensive work by Makarova et al. (2017) on the New Jersey paleoshelf reveals similar results, with thermocline taxa showing greater δ18O excursions than surface-dwelling genera across the PETM. They similarly attribute these changes to either a modification to the water column structure (i.e., a more gradual thermocline and thicker warmer mixed layer) or a change in habitat or dominant calcification season for surface-dwellers due to excessive stress in the surface (e.g., acidification, warmth, eutrophication; Makarova et al., 2017). A change in bloom season resulting from exceptionally warm SSTs could have restricted planktonic foraminiferal calcification to cooler months. This suggests that the smaller overall change in δ18O recorded by Morozovella compared to Subbotina reflects the inability of surface-dwelling foraminifera to calcify in summer during peak hyperthermals (Makarova et al., 2017). Work by Frieling et al. (2017) attributes plankton restructuring to loss of surface habitat because of extreme temperatures (>36°C) in the tropics during the PETM. Such warm temperatures are not recorded in our δ18O measurements, but it is possible that the warmest season temperatures were not recorded. Furthermore, Luciani et al. (2017) reports a reduction in the relative abundance of subbotinids during the Early Eocene Climatic Optimum (EECO) at Site 1263 (Walvis Ridge) and attributes this decline to destratification of the upper water-column with intermediate waters warming relatively more than surface waters. Similarly, Stap et al. (2010b) observes an absence of Subbotina during the ETM-2 horizon at Site 1265 (Walvis Ridge).

The higher-resolution results across ETM-2 at Bass River provide further insights into the benthic and planktonic foraminiferal response to environmental perturbations (Figs. 6a, 7). The greater temperature change in C. pippeni and Subbotina compared to Morozovella (2–3°C versus 1°C) supports the idea that bottom water and thermocline taxa experience greater warming or surface-dwelling Morozovella does not record the full seasonal variability due to changes in calcification seasons. Even more interesting is the fact that during peak warming at ETM-2, Subbotina records lower δ18O values (i.e., warmer temperatures) than Morozovella (–5.1‰ vs. –4.3‰). This is not consistent with PETM paleoshelf results, where both surface-dwelling genera Morozovella and Acaranina continue to record lower δ18O values than Subbotina through the CIE (Makarova et al., 2017). In our study, Subbotina does not return to pre-excursion values until after the subsequent (H2) excursion event, indicating continued warmth in the thermocline until well after the ETM-2 event. The warmer values recorded in Subbotina during ETM-2 may be the result of a seasonal change in calcification, such that Morozovella does not calcify during extreme surface temperatures, but rather calcifies during cooler months. In this scenario, the warming in surface waters may not be recorded fully in our Morozovella record at ETM-2, and is potentially greater than 1°C and could even be higher than 3°C. Work by Si & Aubry (2018) across the PETM suggests a calcification threshold of 32°C with a lack of growth above this temperature. If during ETM-2 surface waters experience comparable warming to the thermocline, then temperatures could have reached this threshold boundary, potentially limiting Morozovella growth. Planktonic foraminifera demonstrate a seasonal variation due to temperature and “optimal growth” conditions, and show stress-induced reduced calcification when those conditions are not met (e.g., de Villiers, 2004; Weinkauf et al., 2016). Furthermore, modern planktonic foraminifera show changes in oxygen isotopic calcification due to seasonal variability (Williams et al., 1979; Curry et al., 1983). There is also the potential that the apparent reversal in Subbotina and Morozovella δ18O values at ETM-2 could be an artifact of differential preservation during the warming event, though preservation of our samples shows no evidence for this.

Although a change in water column dynamics, change in calcification seasonality, and habitat restructuring in response to warming events seem possible, the consistently higher δ13C values for Morozovella compared to Subbotina make it difficult to reconcile similar life environments for the two genera. Specifically, we would expect Morozovella δ13C values to become lower with a descent in the water column, reflecting typical vertical nutrient profiles (e.g., Boersma et al., 1987). Therefore, we suggest that thermocline-dwelling Subbotina experiences greater warming due to a warming of the shelf thermocline waters, creating a more homogenous warm mixed layer in response to warming events. The isotopically lower Subbotina δ18O values reflect a change in Morozovella seasonal calcification due to extreme temperatures in the surface waters observed at ETM-2.

We examine early Eocene hyperthermals at shallow-water sites Bass River, Ancora, and Double Trouble by combining bulk carbonate and foraminiferal stable isotopes with faunal studies. During some hyperthermal events (as noted by negative excursions in δ13C and δ18O), we observe an increase in the ratio of agglutinated to calcareous foraminifera, an increase in infaunal taxa, a decrease in diversity [F(α)] and heterogeneity [H(s)], and an increase in dominance (D), indicating an overall decrease in benthic diversity in response to warming events. Prominent shifts to opportunistic, stress-tolerant taxa suggest increased nutrients and ocean productivity in response to environmental perturbations associated with the hyperthermal events. Faunal turnovers across hyperthermal events are most apparent at the more updip sites Ancora and Double Trouble, suggesting greater environmental perturbations occurred in shallower settings. Greatest faunal changes occur across the largest excursions, suggesting that biotic responses to environmental perturbations are reached only at a critical threshold. We present new stable isotope data from planktonic and benthic foraminifera focused on the largest event in our record (ETM-2) at Bass River. Comparable planktonic thermocline and surface-dwelling δ18O values during excursion events suggest greater warming in the thermocline (through deepening of the mixed layer), changes in calcification season, or depth habitat restructuring. All scenarios are plausible, although the consistently higher δ13C values for Morozovella compared to Subbotina argue against migration of surface-dwelling taxa into the thermocline. Comparing geochemical and biotic data from three shallow-water sites along a depth transect allows us to provide a consistent and comprehensive study of the benthic foraminiferal response to the transient warming of hyperthermals in the early Eocene.

This paper benefitted from the comments of two anonymous reviewers. All data employed in this paper are available in the supporting information (Appendices 1–8) and may also be requested from Megan Fung (mkfung@callutheran.edu). Appendix 1 can be found linked to the online version of this article.

APPENDICES

Appendix 1. Stable isotope and foraminifera data from Figures 2, 3.

Appendix 2. Agglutinate and calcareous taxa present at Bass River, Double Trouble, and Ancora. Classifications compared to Thomas (2003) and Arreguín-Rodríguez et al. (2016).

Appendix 3. Classification of inferred life position (epifaunal or infaunal) of all benthic foraminifera identified at Bass River, Double Trouble, and Ancora. Morphotype grouping is based on test-shape (e.g., Jones and Charnock, 1985; Corliss and Chen, 1988) and is compared to Arreguín-Rodríguez et al. (2016).

Appendix 4. Depths (ft.) of individual hyperthermal events at each site as illustrated by gray shaded bars in Figures 24, 67.

Appendix 5. Records for δ18Obulk and δ13Cbulk at Bass River, Double Trouble, and Ancora.

Appendix 6. Taxonomic list of taxa discussed throughout the manuscript.

Appendix 7. PCA results from Bass River, Double Trouble, and Ancora using the most significant taxa (>2.5% in at least one sample).

Appendix 8. Summary of known ecological preferences for specific benthic foraminiferal taxa.

Supplementary data