The integrated Ocean Drilling Program Expedition 321 Site U1338 (Hole 1338A) (2∘30.469′ N, 17∘58.162′ W), situated in the EEP, was drilled to 410 m below seafloor (mbsf) through Holocene – early Miocene pelagic sediments (Pälike et al., 2010). At ∼ 3 Ma, the site was at ∼ 2∘ N paleolatitude (Drury et al., 2014) in a deep-water pelagic environment of similar water depth to the modern (∼ 4 km). The primary lithologies represented are calcareous, diatomaceous and radiolarian nannofossil oozes and chalks. Despite the deep-water settings, and primarily calcareous nature of the sediments, excellent microfossil preservation has been recorded in planktonic foraminiferal specimens through intervals of this core (Fox and Wade, 2013; Woodhouse et al., 2021). A preliminary assessment of core U1338A was carried out to determine the approximate position of the extinction of several species of Dentoglobigerina (∼ 3 Ma) based on tropical biostratigraphy (Wade et al., 2011), and shipboard paleomagnetic data (Pälike et al., 2010).

Sediment volumes of 20–40 cm3 were collected and washed with de-ionized water over a 63 µm sieve; the residues were dried in an oven at 40 ∘C and split. All samples were examined using a Zeiss Stemi 305 Compact Stereo Microscope. Planktonic foraminifers were identified following the taxonomy of Kennett and Srinivasan (1983), Schiebel and Hemleben (2017), and Wade et al. (2018). We performed assemblage counts on >63 µm splits yielding >300 individuals (Table S1 in the Supplement). Assessments of taxonomy and test preservation of foraminifera were performed via analysis with the Tescan VEGA3 XM Scanning Electron Microscope (SEM) at the University of Leeds, UK.

Species were grouped by their ecology to assess the relative abundances of taxa for paleoceanographic information, where taxa were assigned as either shallow- and warm-water taxa (the ecogroups “symbiotic” and “asymbiotic” of Aze et al., 2011), or deep- and cold-water taxa (the ecogroups “thermocline”, “subthermocline” of Aze et al., 2011, and Globigerinita glutinata, Lutz, 2010).

Specimens of the genus Dentoglobigerina were measured to compare species size with stable isotope ratios to investigate species ecology and ontogeny. Complete specimens of Dentoglobigerina were picked and mounted in umbilical position on card slides pierced with a fine needle to accommodate the variably spired nature of species in the genus (Wade et al., 2018). Specimens were imaged umbilically using a Zeiss Axio Zoom V16 microscope with attached Canon EOS 100D camera at ×19.4 magnification. All specimens were then rotated 90∘ laterally and imaged whilst propped onto their penultimate chamber. Images were processed using the image analysis software Image Pro Premier, and the maximum test diameter, previously deemed a statistically repeatable measurement amongst Dentoglobigerina (Brombacher et al., 2017b, 2018; Woodhouse et al., 2021), was captured from both orientations, and the lower of the two values assigned as test size.

Following morphometric analysis, well-preserved specimens of Dentoglobigerina altispira (>200 µm), Dentoglobigerina baroemoenensis (>200 µm) and Dentoglobigerina globosa (>200 µm) were picked, ultrasonicated in deionized water for 10–15 s, and dried for stable isotope analysis. This process was repeated for extant taxa representing known discrete ecological habitats through the water column to determine the ecological niche habits of the extinct dentoglobigerinids: Globigerinoides ruber (212–350 µm, surface mixed-layer (SML)), Neogloboquadrina incompta (212–350 µm, subsurface), Globorotalia tumida (>300 µm, thermocline and/or photic zone base, corrected for a 1.0 ‰ δ13C enrichment due to this species occupying the shallow oxygen minimum zone and the consequential effects of reduced ambient pH, Lohmann, 1995; Bijma et al., 1999; Uchikawa and Zeebe, 2010; Birch et al., 2013), Globorotalia scitula (212–300 µm, subthermocline), and Cibicidoides wuellerstorfi (>212 µm, bottom-water) (see Cramer et al., 2009, 2011; Rasmussen and Thomsen, 2010; Aze et al., 2011; Woodhouse et al., 2021). Single specimens of dentoglobigerinids and multiple specimens of other foraminifer species were analyzed using an Elementar IsoPrime Dual-Inlet Isotope Ratio Mass Spectrometer in the School of Earth and Environment at the University of Leeds, and data were reported to the Vienna Peedee belemnite (VPDB) scale using a Carrara marble standard (Elemental Microanalysis B2214), where analytical precision was better than 0.07 ‰ and 0.13 ‰ for δ13C and δ18O, respectively.

Specimens of Cibicidoides wuellerstorfi were also used to create a benthic foraminiferal δ18O record for this study to supplement the shipboard paleomagnetic data (Pälike et al., 2010). This record was constructed and tuned to the Ocean Drilling Program Site 849/IODP Site 1338 stack constructed by Lyle et al. (2019) using QAnalySeries software (Kotov and Pälike 2018) to better constrain the timing of events (see Woodhouse et al., 2021), where the study section is calculated to represent the Pliocene interval from ∼ 3.47–2.98 Ma.

To investigate how planktonic foraminiferal macroevolution and paleoecology has developed since the Pliocene, the Triton dataset (Fenton and Woodhouse et al., 2021) was downloaded, and all macroperforate planktonic foraminiferal records occurring from 5.3–0 Ma (early Pliocene-Recent) were binned into 53-time bins with equal length (100 kyrs). Species were assigned the speciation and extinction datums in accordance with Aze et al. (2011) and Fenton and Woodhouse et al. (2021), and all species occurrences located outside of these assigned stratigraphic ranges were removed. This range trimming was applied to eliminate much of the occurrence data likely attributable to misidentification and/or reworking that may create artificial “tails” within speciation and extinction data (Liow et al., 2010; Lazarus et al., 2012; Flannery-Sutherland et al., 2022). The trimming of taxa resulted in a dataset of 239 317 planktonic foraminiferal occurrences. Furthermore, all species were assigned to their respective “ecogroups”, which represent broad ecological categories based on paleoecological and phylogenetic data (Aze et al., 2011). These ecogroups are defined as: ecogroup 1 = surface mixed layer dweller with photosymbionts, ecogroup 2 = surface mixed layer dweller without photosymbionts, ecogroup 3 = thermocline dweller, and ecogroup 4 = subthermocline dweller, ecogroup 5 = high-latitude.

All sediments contain a highly abundant well-preserved (Fig. 1) open-ocean planktonic foraminifer assemblage comprising ∼ 70 morphospecies. The dominant genera through the section included Neogloboquadrina, Globigerinoides, Pulleniatina, and Globigerinita (see Supplement). The isochronous extinction of the species D. altispira, D. baroemoenensis, and D. globosa in U1338A occurs ∼ 35.50 m b.s.f. (∼ 3.037 Ma). There is also a notable influx of the species Menardella cf. exilis and Menardella cf. pertenuis occurring from ∼ 42.58–40.56 m b.s.f. (∼ 3.45–3.36 Ma), after which they are absent within the study section.

Comparing the assemblage composition of warm and surface dwellers against cold and deep dwellers, we document generally consistent species abundances until ∼ 36.26 m b.s.f. (∼ 3.08 Ma), where the assemblage exhibits consistently greater abundances of cold and deep dwellers for the rest of the record (Fig. 2).

Test preservation is excellent throughout the sampled interval, where specimen walls appear optically translucent, and SEM images (Fig. 1) indicate no observable diagenetic alteration, clean pore spaces, and spines preserved within tests (Fig. 1a). Following artificial test breaking, inspection of the wall ultrastructure (Fig. 1b) showed that no wall recrystallization had taken place, suggesting stable isotope signals remain unaltered (Sexton et al., 2006; Edgar et al., 2013b). Single-specimen δ13C values for D. altispira, D. baroemoenensis, and D. globosa vary from -0.54 ‰ to +3.53 ‰, +0.30 ‰ to +3.50 ‰, and +0.51 ‰ to +2.84 ‰ (Fig. 3, Tables S3 and S4), respectively, whilst δ18O values vary from -3.56 ‰ to -0.03 ‰, -2.99 ‰ to -0.17 ‰, and -1.92 ‰ to -1.05 ‰, respectively. In all three species, a positive correlation is calculated between test size and δ13C; however, this relationship is only significant (p<0.0001) in D. altispira and D. baroemoenensis, potentially due to the greater number of specimens analyzed for these species (Fig. 3). Regression slopes are similarly steep; however the slopes for D. altispira and D. globosa are more alike. Dentoglobigerina baroemoenensis and D. globosa show no significant correlation between δ18O and test size; however, a significant (p = 0.0006) negative relationship is recorded in D. altispira. Once again, regression slopes for D. altispira and D. globosa are similar, whereas for D. baroemoenensis, this slope is almost flat (Fig. 3).

Amongst the extant species picked from strict size fractions throughout the study section, G. ruber δ13C values show the highest δ13C variability (+0.7 ‰ to +2.8 ‰) and δ18O values varying from -2.3 ‰ to -0.8 ‰ (Fig. 4, Table S2). Corrected G. tumida values show the lowest δ13C variability (0 ‰ to +0.8 ‰), and the highest variability for δ18O (-1.3 ‰ to +1.1 ‰). G. scitula δ13C values vary from -0.6 ‰ to +0.4 ‰, whilst δ18O varies from +0.8 ‰ to +2.9 ‰, whereas bottom-water C. wuellerstorfi δ13C values vary from -0.8 ‰ to +0.5 ‰, and δ18O varies +2.2 ‰ to +3.1 ‰, showing the lowest variability in δ18O (Fig. 4). The stable isotope records of extant species generally tend to show low inter-sample variability, however there is notable geochemical stochasticity within the G. tumida record ∼ 36.17–35.86 m b.s.f. (Fig. 2).

All three species of Dentoglobigerina analyzed in this study record mean stable isotopic ratios similar to the extant species G. ruber, where all species indicated δ18O ratios less negative than G. ruber, and D. baroemoenensis and D. globosa show δ13C ratios more positive than this species (Figs. 3 and 4).

The global proportions of planktonic foraminiferal ecogroups within the Triton dataset (Fenton and Woodhouse et al., 2021) show that throughout the time interval studied (5.3–0 Ma), surface mixed layer dwellers without photosymbionts (orange) proportions remain relatively consistent, making up ∼ 10 % of the total global fauna (Fig. 5). From 5.3–3 Ma, global ecogroup proportions are relatively consistent, where the dominant forms are thermocline dwellers (cyan) and surface mixed layer dwellers with photosymbionts (red) (Fig. 5), making up ∼ 40 % and ∼ 35 % of the global fauna, respectively. During this interval, the proportion of subthermocline dwellers (blue) and high-latitude forms (yellow) also remain relatively consistent with ∼ 10 % and <5 % of total proportions, respectively. After ∼ 3 Ma, both thermocline dwellers and surface mixed layer dwellers with photosymbionts show a steady decline in global representation approaching the Recent, making up ∼ 20 % each in the modern ocean. At ∼ 3 Ma, both the subthermocline dwellers and high- latitude ecogroup start to show increases in their proportions, though this increase approaching the modern is much greater in the latter group. During the last 100 kyrs, subthermocline and high- latitude species have constituted ∼ 15 % and ∼ 25 %, respectively (Fig. 5).

The extinction of D. altispira is a useful marker in the mid-Pliocene, previously recorded within the East Equatorial Pacific Ocean (3.46 Ma; Shackleton et al., 1995; Wade et al., 2011); however, this study and that of Woodhouse et al. (2021) provide recalibration for this event, which also includes the co-extinction of D. baroemoenensis and D. globosa. The co-extinction of the dentoglobigerinids is significant as Woodhouse et al. (2021) demonstrated that D. altispira and D. baroemoenensis show unique phenotypic responses leading up to their termination, despite their shared phylogenetic and ecological affinity. We, therefore, suggest that all three species share an ecological habit that ultimately proved inefficient to mitigate the changing abiotic conditions associated with this critical period of bipolar cryosphere development (Kleiven et al., 2002; Brierley and Fedorov, 2010; Cramer et al., 2009, 2011; Willeit et al., 2015; Hayashi et al., 2020; Westerhold et al., 2020).

Studies of extant species suggest a strong positive correlation between planktonic foraminifera δ13C and test size, paired with a lack of a strong negative significant relationship in δ18O and test size, is indicative of species bearing photosymbiotic algae (e.g., Berger et al., 1978; Bouvier-Soumagnac and Duplessey, 1985; Spero and Williams, 1988, 1989; Spero et al., 1991; Spero, 1992; Spero and Lea, 1993; Ravelo and Fairbanks, 1995; Norris, 1996; Birch et al., 2013). Our data suggests that all three species of Dentoglobigerina studied here are symbiont-bearing, displaying mean stable isotope data similarly to the symbiont-bearing surface mixed layer dwelling G. ruber (Figs. 2–4). Moreover, the regression lines for the ancestor-descendent pair D. globosa – D. altispira are remarkably similar, suggesting that the species share similar stable isotope-test size relationships, despite the speciation of D. altispira occurring ∼ 20 Myrs prior (Aze et al., 2011; Wade et al., 2011, 2018). However, the substantially fewer data for D. globosa currently limits the significance of this observation for this particular species (Fig. 3).

The results presented here contribute to the variety of interpretations from previous studies on the paleoecology of D. altispira. Spanning different ocean basins throughout the species' stratigraphic range they suggest: a shallow-dwelling (e.g., Keller and Savin, 1985; Vincent et al., 1985; Prentice and Matthews, 1988; Corfield and Cartlidge, 1991; Norris et al., 1993; Hodell and Vayavananda, 1994; Sosdian and Lear, 2020), deep-dwelling (Opdyke and Pearson, 1995; Pearson and Shackleton, 1995), or lower mixed layer/upper thermocline (Zou et al., 2022) ecological niche habit.

Woodhouse et al. (2021) suggested that prior to extinction, the D. altispira geochemical signal may signify that this species underwent adaptive ecological niche migration across multiple depth habitats. Upon further investigation, by isolating all D. altispira specimens present after the migration horizon at ∼ 3.061 Ma (Woodhouse et al., 2021), we show that these specimens do, in fact, display a steeper negative relationship between δ18O and test size (Fig. S1), consistent with asymbiosis; however, this relationship lacks statistical significance. Nevertheless, this observation lends support to the conclusions of Woodhouse et al. (2021) that this species underwent niche adaptation prior to extinction.

Another plausible scenario may be the existence of indeterminate cryptic diversity within the D. altispira morphospecies complex where multiple ecological strategies or phylogenetic expressions may be present (e.g., Huber et al., 1997; Bijma et al., 1998; de Vargas et al., 1999, 2002; Weiner et al., 2012; Schiebel and Hemleben, 2017; Nirmal et al., 2021; Pearson and Penny, 2021), and the stepwise changes observed by Woodhouse et al. (2021) may indicate the systematic loss of “cryptic genotypes” within this morphospecies complex. Indeed, Pearson and Penny (2021), hypothesized that dramatic abundance switches in the Indo-Pacific Warm Pool of ecologically distinct, alternately coiled populations of Pulleniatina morphospecies may signify replacement by distinct cryptic genotypes, and such coiling switches are noted throughout the planktonic foraminiferal fossil record (Ericson et al., 1955; Saito et al., 1975; Bossio et al., 1976; Hallock and Larsen, 1979; Hornibrook, 1982; Scott et al., 1990; Norris and Nishi, 2001; Winter and Pearson, 2001; Crundwell and Nelson, 2007; Wade et al., 2011; Pearson and Ezard, 2014; Crundwell, 2015a, b; Levin et al., 2016; Wallace et al., 2019; Crundwell and Woodhouse, 2022a, b). Therefore, the range of interpreted paleoecologies in D. altispira may, in fact, be due to the occurrence of distinct cryptic populations from across the geological record. Irrespectively, the abiotic conditions at this time appear to have become fatally detrimental to the entire ecological habit of all three morphospecies of Dentoglobigerina analyzed in the EEP, and subsequently across the entire globe (Shackleton et al., 1995; Chaisson and Pearson, 1997; Wade et al., 2011; Raffi et al., 2020; Groeneveld et al., 2021; Woodhouse et al., 2021).

The transient influx and co-occurrence of M. cf. exilis and M. cf. pertenuis from ∼ 3.45–3.36 Ma may indicate a shared ecological affinity between these species (Fig. 2; Kennett and Srinivasan, 1983; Knappertsbusch, 2016). Furthermore, this association may suggest that these two similar forms exist within the same genetic species complex, though further study on internal shell ontogeny and external morphology would be required to confirm this hypothesis. Notably, however, Kaneps (1970) and Chaisson (2003) suggested that Indo-Pacific occurrences of these two species are, in fact, “aberrant” forms of Menardella limbata and Menardella menardii, with the M. exilis-pertenuis plexus being solely endemic to the Atlantic basin (Sexton and Norris, 2011).

Scanning election photomicrographs and pore density analysis of Menardella morphospecies within this study (Fig. 6; Table 1) suggests that the M. cf. exilis and M. cf. pertenuis morphotypes ascribed to “aberrant” forms may, in fact, be phylogenetically distinct from both M. menardii and M. limbata, as well as from the endemic Atlantic M. exilis-pertenuis plexus. Where the Atlantic type specimens exhibit finely perforate tests (Kennett and Srinivasan, 1983), the specimens within this study appear to show test perforation intermediate between “normally perforate” menardellids such as M. menardii and M. limbata (Fig. 6), and the “finely perforate” M. exilis-pertenuis plexus (Kennett and Srinivasan, 1983; Chaisson, 2003). Whether all Indo-Pacific occurrences of these morphotypes (e.g., Brönniman and Resig, 1971; Jenkins and Orr, 1972; Thunell, 1981; Keigwin, 1982; Thompson, 1982; Chaisson and Leckie, 1993) exhibit similar test perforation remains to be determined. However, these forms could represent either: (1) convergent evolution of a geographically isolated endemic population restricted to the Indo-Pacific triggered by vicariance, or (2) specimens within the M. exilis-pertenuis plexus exhibiting a differential phenotypic expression of pore density, potentially due to regional differences in paleoceanography between the Atlantic and Indo-Pacific basins at this time (Haug et al., 2001). Irrespective of the biological implications, this acme event may represent a regionally valuable biostratigraphic marker horizon.

Comparing the assemblage composition of warm and surface dwellers against cold and deep dwellers, we document a gradual turnover from an assemblage exhibiting relatively even abundances of these two groups to one of increasing occupation of cold and deep dwellers from ∼ 3.08 Ma onwards (36.26 m b.s.f.; Fig. 2). This may reflect regional thermocline shoaling similar to patterns observed at DSDP Site 84 in the proximal Panama Basin (Lutz, 2010). Other geological intervals associated with significant global cooling such as the Eocene-Oligocene transition and late Miocene cooling exhibit similar patterns of gradual, successive turnover of warm-water dwellers by cold-water taxa potentially associated with contractions and expansions in their respective ecospaces (Keller, 1983; Boersma and Premoli Silva, 1991; Keller et al., 1992; Molina et al., 1993; Wade and Pearson, 2008; Ezard et al., 2011; Lowery and Fraass, 2019; Lowery et al., 2020; Boscolo-Galazzo and Crichton et al., 2021; Boscolo-Galazzo et al., 2022; Woodhouse and Swain et al., 2023).

The establishment of the Isthmus of Panama played a key role in Atlantic Meridional Overturning Circulation intensification by barring Caribbean inflow from the Atlantic (O'Dea et al., 2016; Hayashi et al., 2020), leading to a new paleoceanographic state defined by an equatorial Pacific thermocline exhibiting a high-angle east–west tilt (Fiedler and Talley, 2006; Yang and Wang, 2009; Zhang et al., 2012; Ford et al., 2015). Changes in the EEP water column structure are compounded within biotic and abiotic records, where global changes in ice volume, as shown by benthic δ18O (Fig. 5), and the formation of the isthmus contributed to the gradual cooling and shoaling of the thermocline (Ford et al., 2015). Isochronous records of ice-rafted debris in the North Atlantic and Pacific infer a substantial change to high-latitude global climate ∼ 3 Ma (Shackleton et al., 1984; Maslin et al., 1996; Kleiven et al., 2002; Lawrence et al., 2006). Consequently, the already restricted CAS was likely subjected to substantial eustatic sea level fluctuations (Chaisson, 2003; Bartoli et al., 2005; O'Dea et al., 2016), causing cool water within or below the thermocline to become shallow enough to be mixed by surface winds (Philander and Fedorov, 2003; Fedorov et al., 2004, 2006; Ford et al., 2012, 2015). In turn, this contributed to a dramatic increase in regional upwelling associated with further development of the EEP cold tongue (Herbert et al., 2010). We suggest that the increasing dominance of cold and deep dwellers in the EEP (Figs. 2 and 5) are a direct manifestation of the water column structure changes associated with closure of the CAS and cryosphere intensification, marking the initiation of transition from the early Pliocene “El Padre” mean state to one more similar to the modern ocean (Seki et al., 2012; Ford et al., 2012, 2015).

The geochemical fluctuations noted within G. tumida from 3.071–3.061 Ma (36.17–35.86 m b.s.f.; Fig. 2) are coeval with the interval assigned to the “phase transition” by Woodhouse et al. (2021), in which D. altispira and D. baroemoenensis exhibit dramatic fluctuations in size and shape preceding rapid ecological changes. Here, the δ18O signal of G. tumida exhibits substantial and rapid changes up to ∼ 2.1 ‰, switching from typical thermocline values to those indicative of the subsurface, and back again, whereas all other water column δ18O value changes are nominal (Fig. 2).

If the fluctuations within the δ18O signal of G. tumida were related to fluctuations in its depth habitat, we would expect the inverse pattern to occur in its δ13C signature, as δ13C values decrease with depth (Cannariato and Ravelo, 1997; Ford et al., 2012; Birch et al., 2013); however, this is not the case (Fig. 2, Table S2). The maximum δ18O excursion value (∼ 2.1 ‰) is equivalent to a temperature change of ∼ 10 ∘C, based on a mid-Pliocene SMOW value of -0.3 ‰ (Williams et al., 2005; Medina-Elizalde, Lea and Fantle, 2008; Tindall and Haywood, 2015) and the paleotemperature equations of Kim and O'Neil (1997). Previous studies from EEP sites show Mg / Ca-derived thermocline temperature fluctuations of ∼ 3 ∘C (Site 1241; Steph et al., 2006) to ∼ 5 ∘C (Sites 848, 849, and 853; Ford et al., 2012) across the same interval, and modern ocean seasonal deviations within EEP surface waters are minor at ± 1 ∘C (Fiedler, 1992). However, these can be intensified by El Niño conditions by up to +3.8 ∘C (Pérez-Angel and Molnar, 2017).

Modern regional evaporation–precipitation balance within the EEP also shows very little variation, wherein δ18Osw values average 0.26 ‰ (Fairbanks et al., 1992), ranging from ∼ 0.2 ‰–0.5 ‰ for the entire tropical Pacific (Ravelo and Hillaire-Marcel, 2007). Factors other than temperature and salinity (e.g., chlorophyll a, the lunar cycle) that trigger species-specific habit changes (Rebotim et al., 2017) could be at least partially responsible for these dramatic changes in the G. tumida δ18O records; however, this species is known to calcify at the base of the photic zone (Ravelo and Shackleton, 1995), irrespective of thermocline depth (Ravelo and Fairbanks, 1992; Rincon-Martinez et al., 2011).

Despite the present low range in Pacific equatorial δ18Osw values, modern Atlantic-Caribbean surface waters record a salinity value ∼ 3 ‰ less than the modern EEP (Haug et al., 2001; Garcia et al., 2006; Schmidt et al., 2016; Öğretmen et al., 2020), equating to ∼ 1.5 ‰ lower in mean δ18Osw values (Ravelo and Hillaire-Marcel, 2007). This salinity contrast was fully established by ∼ 4.2 Ma (Haug et al., 2001), and consistent breaching of Atlantic-Caribbean waters over the still-submerged isthmus may have occurred as late as 1.9 Ma (Coates and Obando, 1996; Keller et al., 1989; Schmidt et al., 2016). Therefore, a breaching event is the most probable cause for the substantial variations in the δ18O record of G. tumida from 3.071–3.061 Ma, where this apparently thermocline-restricted (Fig. 2) disturbance may have contributed to disruption of the Dentoglobigerina ecological niche habit at this time (Woodhouse et al., 2021).

It appears that the faunal turnover documented in the EEP at ∼ 3.08 Ma (Fig. 5) may represent a critical point in the global development of the bipolar cryosphere, which typifies the Pleistocene and Holocene (Kleiven et al., 2002; Mudelsee and Raymo, 2005; Brierly and Fedorov, 2010; Cramer et al., 2009, 2011; Willeit et al., 2015; Hayashi et al., 2020; Westerhold et al., 2020). Major Antarctic ice sheet expansion (Shevenell et al., 2004; Holbourn et al., 2015; Frigola et al., 2018; Westerhold et al., 2020) following the Miocene Climatic Optimum (∼ 17–15 Ma; Methner et al., 2020) initiated global cooling, intensifying meridional gradients (Gaskell et al., 2022). This triggered unprecedented mean size increases in low-latitude planktonic foraminifera (Schmidt et al., 2004a, b) and higher efficiency of the biological carbon pump, promoting greater endemism and exploitation of new deep-water niches in calcifying plankton (Olsson, 1982; Scott, 1982; Keller, 1985; Malmgren and Berggren, 1987; Scott et al., 1990; Norris et al., 1993, 1994, 1996; Norris, 1999, 2000; Rögl, 1999; Chaisson, 2003; Kucera and Schönfeld, 2007; Ezard et al., 2011; Crundwell, 2018; Rosenthal et al., 2018; Spezzaferri et al., 2018; Lam and Leckie, 2020; Boscolo-Galazzo and Crichton et al., 2021; Boscolo-Galazzo et al., 2022; Woodhouse and Swain et al., 2023).

From ∼ 6.5 Ma, the equatorial Pacific shows meridionally distinct planktonic foraminiferal faunal provinces that display a turnover in dominance from Miocene species to more Recent taxa ∼ 4 Ma (Chaisson, 1995; Chaisson and Ravelo, 2000) linked with the closure of the CAS (Haug et al., 2001). Global temperatures record a gradual, stable decline towards ∼ 3 Ma (Fig. 5; Westerhold et al., 2020), after which planktonic foraminiferal morphospecies diversity shows a notable decline (Aze et al., 2011; Ezard et al., 2011; Fraass et al., 2015; Lowery et al., 2020), whilst ecogroups (Fig. 5) exhibit the expansion of cold-water forms coincident with the development and intensification of Northern Hemisphere ice sheets (Kleiven et al., 2002; Brierly and Fedorov, 2010; Cramer et al., 2009, 2011; Willeit et al., 2015; Hayashi et al., 2020). The greater proportion of cold and deep dwellers in the EEP at ∼ 3.08 Ma (Fig. 2) coincides with global ecogroup patterns (Fig. 5), signifying the initiation of increasing global dominance of planktonic foraminiferal species with cold-water affinity, alongside many other phylogenetic groups (Slater et al., 2017; Steinthorsdottir et al., 2020), as the world descended into a bipolar icehouse state.

Despite the reduction in morphospecies diversity from ∼ 3 Ma, the late Cenozoic closure of the Tethyan and Central American Seaways (Crame and Rosen, 2002; Brierly and Fedorov, 2010; Hamon et al., 2013; Matthews et al., 2016) may have contributed to the notable rise in diversity through the Neogene (Aze et al., 2011; Ezard et al., 2011; Peters et al., 2013; Fraass et al., 2015; Lowery et al., 2020), due to a significantly more heterogenous ocean structure via longitudinal obstruction of tropical-subtropical waters by continental reconfiguration, and latitudinal partitioning caused by the steepening of global meridional temperature gradients (Haug et al., 2001; Schmidt et al., 2004a, b; Knappertsbusch, 2016; Ford et al., 2022; Friesenhagen, 2022; Gaskell et al., 2022). Moreover, this intensified, heterogenous icehouse climate may have played a significant role in shaping the incredible diversity observed within modern planktonic foraminiferal cryptic genotypes (Darling and Wade, 2008; Aurahs et al., 2009; Morard et al., 2009, 2013, 2019; Ujiié et al., 2010; Norris and Hull, 2012; Weiner et al., 2012, 2014; André et al., 2014; Ujiié and Ishitani, 2016), though further work is required on the quantification of planktonic foraminiferal cryptic diversity within deep time (André et al., 2013).

It should be noted however, that there is a prominent increase in sampling of Quaternary high latitudes that is yet to be replicated in deep-time marine records (Lazarus, 1994; Diepenbroek et al., 2002; Sellén et al., 2010; O'Regan, 2011; Siccha and Kučera 2017; Waelbroeck et al., 2019; Renaudie et al., 2020; Fenton and Woodhouse et al., 2021). This is due to the difficulties encountered in deep-sea scientific drilling of higher latitudes through the combined impacts of unpredictable and detrimental oceanographic conditions, and ephemeral sea ice (Barker et al., 1977; Backman et al., 2006; Lamy et al., 2019; McKay et al., 2019). Consequently, the substantial increase in cold-water forms observed from ∼ 3 Ma to the Recent (Fig. 5) may be at least partly driven by sampling efforts.