For SEM imaging of sedimented and plankton net samples, individuals were mounted on sticky carbon disks placed on steel stubs and were typically gold or gold/palladium coated (Table 2). Type material of G. exumbilicata Herman 1974, G. egelida Cifelli and Smith 1970, and Globigerina atlantisae Cifelli and Smith 1970, all archived in the USNM Cushman Collection of foraminifera, was imaged for the first time as part of this study. New images of an uncoated type specimen of T. quinqueloba (G. quinqueloba Natland, 1938) were also obtained using an FEI Apreo high-resolution field emission SEM, providing improved characterization of its wall texture.

SEM images of other relevant type specimens were downloaded from the USNM collections (https://collections.nmnh.si.edu/search/paleo/, last access: 2 February 2018). Whole-specimen and wall texture images were taken to document test morphology and ultrastructure, including evidence of spines.

Morphometric analyses were made on three populations: one population of CAO Morphotype-1; (Sample AO16-8GC-2, 2.5–3.5 cm, 68.5–69.5 cmbsf, Alpha Ridge); one population of Morphotype-2 (Sample P1-92-AR-P40, 374–375 cmbsf, Northwind Ridge (Cronin et al., 2019)); and as an extra-Arctic out-group, one population of “classic” T. quinqueloba from MIS 5d (107 ka) (Sample ODP 907A-1H-2, 37–39 cm, Iceland Sea) (Table 1). In each case, measurements were made on ca. 300 individuals from the > 63 µm fraction. Samples were first split into manageable aliquots using an aluminum micro-splitter. All planktonic foraminifera were extracted from the splits until the target of ca. 300 was reached. All specimens were imaged in umbilical view using a Hitachi TM 3000 desktop SEM (Table 2). Measurements were made using the manual measuring tool in the open-source software Image-J (Schneider et al., 2012). Morphometric traits were recorded as continuous/discontinuous measurements or presence/absence conditions (coded as 1 = present, 0 = absent) (Table 4, Appendix Fig. A2). The proloculus was not always identifiable in SEM due to test thickening that obscures early chambers, especially in “classic” T. quinqueloba. Proloculus diameter, therefore, was measured on light microscope images, using a Leica M205C microscope and camera system, and water to emphasize chamber contrasts (Appendix Fig. A1). Proloculus diameter measurements are at the limits of accuracy due to this approach; they represent maximum estimates and likely carry greater uncertainty than measurements obtained using, e.g., micro-CT methods.

To assess morphological differentiation among populations, a simple one-way multivariate analysis of variance (MANOVA) and linear discriminant analysis (LDA) were conducted on the continuous morphometric parameters (maximum test diameter, proloculus diameter, maximum final chamber diameter, minimum final chamber diameter, and apertural lip thickness) using base R and the MASS package (R version 4.4.2). MANOVA tests whether groups differ in their multivariate means and is well suited to datasets like ours that reasonably meet assumptions of normality and linear relationships among variables (see Results section). Morphometric variables were first log-transformed to stabilize variance and approach multivariate normality, followed by z-score standardization to ensure equal weighting and mitigate scale-related biases. The MANOVA was used to test the null hypothesis that the multivariate centroids (i.e., the mean morphology in multivariate space) of the three populations are equal. Multivariate test statistics (Pillai's Trace) were obtained from the base R stats implementation of MANOVA. The LDA was performed to further explore group separation and assess the diagnostic potential of the measured traits. The classification accuracy of the LDA model was evaluated using leave-one-out cross-validation (LOOCV) to provide a conservative estimate of the model's predictive performance. This was also performed using base R.

Nannofossil analyses were undertaken on 5 cores from the Northwind Ridge, Mendeleev Ridge, and southern Lomonosov Ridge (Fig. 2; Table 1) in an attempt to identify independent age and paleoenvironmental constraints within documented subpolar planktonic foraminifera horizons from across the Arctic Ocean (Appendix Table 1) (Hanslik, 2011; Hanslik et al., 2013; Cronin et al., 2014). These were combined with existing nannofossil data from multiple sites in the CAO (Razmjooei et al., 2023) and foraminifera assemblages from core AO16-8GC on the Alpha Ridge (Vermassen et al., 2021). We specifically targeted records where we could investigate nannofossil occurrences associated with Morphotype-2 horizons.

Small amounts of unprocessed sediment were taken throughout the core, including horizons containing Morphotype-2. Nanno smear slides were prepared using the technique of Bown and Young (1998). The slides were examined using a ZEISS Axio Scope A1 polarizing light microscope at ×1000 magnification and imaged. Species identification followed the taxonomy of Perch-Nielsen (1985) and Young (1998), as well as the most up-to-date descriptions and illustrations in the online database Nannotax3 (Young et al., 2022). In order to examine the abundance of different taxa, all nannofossils were counted in 100 fields of view (FOVs). Additionally, two to three more traverses of each slide were scanned to detect rare and very rare species.

Our Holocene core top samples contain “classic”-looking T. quinqueloba, which, under the optical light microscope, appear either reflective and shiny or white and non-reflective when a thickened outer crust is present (Plate 1, panels 1 and 2). It is these encrusted forms that are here referred to as “classic” T. quinqueloba. In the eastern Nordic Seas samples (Fig. 2; Plate 1, panel 2; Plate 3), specimens often exhibit the classic features diagnostic of the taxon, including an elongate/ampullate (tear-drop-shaped) final chamber that extends tightly over the umbilicus and a finely pustulose test surface. Tests commonly have 4.5 to 5 chambers in the final whorl or occasionally 6 (Plate 3, fig. 22), a petaloid peripheral outline, and a relatively flattened coil (see amended diagnosis by Pearson and Kucera, 2018). The terminal chamber is usually bordered by a distinctive lip or flap. The holotype of T. quinqueloba, which is described from recent sediments from the Californian margin (Natland 1938), exemplifies these features (Plate 3, fig. 26; Plate 6, figs. 1–6). We do, however, observe variability in the morphology of the terminal chamber in the Nordic Seas examples, with respect to the degree of final chamber elongation and overhang into the umbilicus (Plate 1, panel 2; Plate 3). The spinose, normal-perforate wall is visible in close-up SEM views of “classic” T. quinqueloba (Plate 3, figs. 5, 10, 25). The spine bases, which take the form of small pustules or mounds, are distributed between the pores and are mostly separated from each other. Gametogenic calcite tends to thicken the spine bases, which can conceal the spine holes. Extreme gametogenic test thickening can create highly rugose test surfaces and an N. pachyderma-like appearance (or N. incompta depending on coiling direction), emphasizing the value of SEM wall texture analysis in some cases (Plate 3, figs. 14, 17). Both right and left coilers are imaged here.

SEM images of water column T. quinqueloba from south of Iceland document tests with and without their life covering of fine spines (Plate 4, figs. 1–16), which are easily wiped away during sampling. The imaged specimens are all left coiling. However, we believe this is an artifact of the original sampling procedure. Populations of this species have previously been observed to be approximately 50:50 right : left coiling (Brummer and Kroon, 1988; Darling et al., 2006). All the plankton-sample specimens in the pump samples were relatively small, and the classic overhanging final chamber characteristic of mature, sedimented T. quinqueloba was not well developed. We believe that the left-coiling bias can be explained by the fact that, at the time of sampling, efforts were focused on targeting left-coiling N. pachyderma. Mature specimens of any species were rare under the prevailing bloom conditions, and juvenile, thin-walled N. pachyderma were difficult to distinguish from T. quinqueloba. Consequently, even the T. quinqueloba dataset may have inadvertently become biased toward left-coiling individuals. Some test surfaces in our images are observed to be covered with fine particulate material, which is most likely adhering organic debris, having been picked from a concentrated plankton sample.

Comparing the wall texture of the netted forms with and without spines confirms that the evenly distributed bumps are spine bases. All specimens lack test thickening seen in sedimented T. quinqueloba from the Nordic Seas (Plate 1). Importantly, among the water column specimens, we see differences in the shape of the final chamber, which is typically rounded and rarely ampullate or overhanging the umbilicus. This emphasizes how one of the defining features of T. quinqueloba is limited to adult and terminal ontogenetic stages. Comparison of a right-coiling neogloboquadrinid from the plankton pump, which could be a right-coiling N. pachyderma or an N. incompta (Plate 4, figs. 17, 18), emphasizes the difference in wall texture between Turborotalita and Neogloboquadrina – Neogloboquadrina has pores and “bumps”: these are not spine bases as in T. quinqueloba (there are no spines or spine holes) but solid pustules that become more abundant with increasing test size, joining and coalescing to produce the characteristic thickened reticular texture typical of Neogloboquadrina.

The investigated central Arctic material includes core top (Holocene) and subsurface samples. To provide stratigraphic context, we refer to the distribution of planktonic foraminifera in the three exemplary CAO cores, summarized together with other lithological data (Fig. 1). Our preliminary observations suggest that the Morphotype-2 zone marks a faunal transition from dominantly agglutinated foraminifers to calcareous-rich assemblages that predates the dominance of N. pachyderma in CAO sediments, although this interpretation requires further testing and refinement as additional records become available. The selected cores include published records from the “Drifting Ice Station T-3”, where the Arctic subpolar invasions were first documented on the Mendeleev Ridge (Herman, 1970, 1974), Alpha Ridge core AO16-8GC (Vermassen et al., 2021), and LOMROG07-4PC on the Lomonosov Ridge north of Greenland (Hanslik, 2011) (Fig. 2, Table 1). These cores span the Amerasian and Eurasian basins. As is typical for Pleistocene-age CAO strata, the cored lithologies involve alternations of brown (manganese-rich) bioturbated and microfossil-rich intervals, and lighter-colored intervals devoid of microfossils, interpreted to be interglacial and glacial periods, respectively (O'Regan et al., 2008; Löwemark et al., 2014).

Due to the challenges of dating Arctic sediments, including the lack of a continuous oxygen stable isotope record, debates persist over the sequence of MIS stages captured in recurrent sedimentological variations beyond the range of radiocarbon dating (Razmjooei et al., 2023; Knies et al., 2025; O'Regan et al., 2026). Nonetheless, at least three discrete correlative subpolar plankton acmes are widely reported from across the CAO (Fig. 2). A first-order stratigraphic interpretation is based on occurrences above or below the proposed last occurrence (LO) of the nannofossil P. lacunosa or its correlative horizon in cores where it has not been directly identified. This species, only recently recognized in CAO sediments (O'Regan et al., 2020; Razmjooei et al., 2023), became globally extinct during MIS 12 (424–478 ka) (Raffi et al., 2006). In the absence of further constraints, we infer that the multiple horizons of subpolar species in the younger interval (post-MIS 12) are dominated by populations of Morphotype-1. By contrast, the older interval (pre-MIS 12) is notable for the presence of Morphotype-2, although some specimens attributable to Morphotype-1 may also occur. Neogloboquadrina pachyderma is present throughout the studied horizons to varying degrees; however it is notably less common and even absent in some cores during Morphotype-2 acmes. Our preliminary observations suggest that the Morphotype-2 zone marks a faunal transition from dominantly agglutinated foraminifers to calcareous-rich assemblages that predates the dominance of N. pachyderma in CAO sediments. This agglutinated to calcareous transition is recognized across the CAO and linked to a critical phase in Cenozoic Arctic Ocean climate evolution (Cronin et al., 2008) (Fig. 1). These intervals may even predate the dominance of N. pachyderma in the CAO, which is of great significance to Arctic marine geology and paleoceanography.

Our observations confirm that “modern” Holocene core top sediment samples from the CAO and the early Holocene, like the modern CAO water column beneath perennial sea ice, are devoid of subpolar species, including T. quinqueloba, with the exception of the occasional specimens considered expatriates from lower latitudes (O'Regan et al., 2019; Vermassen et al., 2025) (Plate 2, panel 1). The stratigraphically highest subpolar horizon, whose youngest age could be in MIS 5 (O'Regan et al., 2019; Vermassen et al., 2021; Razmjooei et al., 2023; Vermassen et al., 2023), yields a small spinose morphotype (63–125 µm fraction) with an ampullate final chamber, which corresponds to Morphotype-1. This morphotype has recently been reported as T. quinqueloba (Vermassen et al., 2023) (Plate 1, panel 4). The morphometric data exhibit approximately normal distributions, as shown in the histograms (Fig. 3).

Our Morphotype-1 population (AO16-8GC-2, 2.5–3.5 cm, 68.5–69.5 cmbsf – Alpha Ridge), considered representative of this morphotype in the CAO, is readily identified in the 63–125 µm size fraction by its petaloid shape, distinctive ampullate final chamber and relatively smooth reflective test surface, which appears shiny in light microscope view compared to rough-surfaced, co-occurring N. pachyderma (Plate 1, panel 4), with sparse spine bases revealed in SEM images (Plate 5, figs. 1–13). The mean maximum test diameter is 113.8 µm (range: 82.3–146.9 µm) (Table 4, Fig. 4). This morphotype does not occur in the larger size fractions where N. pachyderma dominates (Plate 1, panel 3). SEM images provide detail of the ampullate final chamber and wall ultrastructure: well-preserved specimens reveal evenly distributed spine bases, often with spine holes (Plate 5, fig. 13). The juvenile whorls and proloculus are clearly visible in spiral views (Plate 5, figs. 9, 11; Appendix Fig. A1). In the CAO, Morphotype-1 attains between 30 %–60 % of the total planktonic foraminiferal assemblage in these stratigraphically higher acmes, the remainder being N. pachyderma (Herman, 1974; Vermassen et al., 2023). In contrast, T. quinqueloba in MIS 5 sediments from the Iceland Plateau (ODP 907A; 1H-2, 37–39 cm, 1.87 mbsf) (Clotten et al., 2018; West et al., 2023) exhibit the “classic” T. quinqueloba morphology, differing dramatically from CAO Morphotype-1 in size. The morphometric analysis comparing MIS-5-age turborotalitids from the Nordic Seas (the ODP 907 “classic” T. quinqueloba) and CAO Morphotype-1 (< MIS 12 – MIS age range), shows a clear difference in overall test size between Morphotype-1 and the other two morphotypes, with Morphotype-1 having smaller mean test diameters and a narrower range (mean 113.8 µm, range 82.3–146.9 µm) compared to “classic” T. quinqueloba (mean 189.6 µm, range 118.6–278.2 µm) and Morphotype-2 (mean 176.3 µm, range 81.5–323.4 µm) (Table 4). “Classic” T. quinqueloba also occurs in the > 125 µm fraction, while CAO Morphotype-1, being considerably smaller, is largely restricted to the 63–125 µm fraction. However, the range and mean of the proloculus size of the two morphotypes is almost identical (Fig. 3, Table 4).

Planktonic foraminifera assemblages are dramatically different in the deeper bio-horizon (Fig. 1), both in Herman's (1974) T3 67-3 and T3 67-9 Mendeleev Ridge material and in our Lomonosov Ridge and Alpha Ridge stations (LRG07-PC04, AO16-9PC, and P1-92-AR-P40). Here they are dominated by Morphotype-2 (Plate 2, panels 2, 4–6). Notably, while N. pachyderma dominates in the corresponding core top samples (Plate 2, panels 1 and 3), N. pachyderma is significantly less abundant or absent where Morphotype-2 dominates. Specimens comparable to Morphotype-1, with a reduced final chamber encroaching the umbilicus, co-occur with Morphotype-2 (Plate 2; Plate 5, figs. 36, 37, 41); however, these are relatively few and are mostly restricted to the 63–125 µm fraction. The dominant Morphotype-2 exhibits shiny/reflective and slightly hispid tests, seen in light microscope view (Plate 2, panels 2, 4–6), a distinctly lobate periphery, evolute coiling, 4–6 chambers in the final whorl, and a globular final chamber with a pronounced apertural lip. The final chamber is often reduced in size compared to the previous chambers and typically does not overhang the umbilicus (Plate 5, figs. 14–34). The umbilicus is often open and maybe surrounded by the remnants of apertural lips from previous chambers. Morphotype-2 attains larger test sizes than Morphotype-1 (AO16-8GC), having a mean size of 176.3 µm in the morphometric sample (range: 81.5–323.5 µm) (Fig. 3, Table 4), occurring in both the 125–250 and the 63–125 µm fractions. There is considerable overlap between the mean and maximum test diameter of Morphotype-2 (P1-92-AR-P40) and “classic” T. quinqueloba from the Norwegian–Greenland Sea (ODP 907). The proloculus of Morphotype-2 is also often strikingly large (Plate 5, figs. 28, 33; Fig. 4). The total number of chambers varies from 6 to 12 in Morphotype-1, 5 to 13 in Morphotype-2, and 7 to 15 in “classic” T. quinqueloba (Table 4). The histogram of the total number of chambers per test shows three distinct distributions and peaks. Morphotype-2 exhibits the fewest chambers (mean = 7.82), followed by Morphotype-1 (mean = 9.45) and “classic” T. quinqueloba, which has the highest count (mean = 11.01) (Fig. 3, Table 4).

In terms of wall ornamentation morphology, Morphotype-1 is spinose, as demonstrated by the presence of spine bases, spine holes and/or the remains of broken spines (Plate 5, figs. 23, 27). Spine bases are also visible in Morphotype-2 but are broader and flatter. Spine bases of both Morphotype-1 and Morphotype-2 are sparsely distributed compared to “classic” T. quinqueloba, although this has not been quantified in this study. In spiral side view, Morphotype-2 often shows rows of small pustules that align to form linear ribs that radiate towards the periphery, most commonly on the spiral side (Plate 5, figs. 28, 29, 32–34, Plate 6, figs. 34, 35). These features are reminiscent of costellae, narrow, and ridge- or rib-like ornamentations, seen on test surfaces in some Cretaceous planktonic foraminifera species (Falzoni et al., 2014). To our knowledge, they are not seen in Morphotype-1 or “classic” T. quinqueloba. Coiling ratios in the three populations (% sinistral), are ca. 50 %, 58 %, and 56 % for Morphotype-1, Morphotype-2, and “classic” T. quinqueloba, respectively (Table 4).

The MANOVA and LDA multivariate statistical approaches further explore morphological differentiation among populations. MANOVA evaluates whether the overall multivariate shell morphology differed among the three predefined groups. Accordingly, it tests the null hypothesis of no significant difference in the centroids (multivariate means) of the three populations. The analysis, based on five log-transformed and standardized traits (maximum test diameter, proloculus diameter, maximum and minimum final chamber diameter, and apertural lip thickness), implies that the centroids of the three populations occupy significantly different positions in multivariate morphospace, demonstrated by the high value of the generally robust Pillai's Trace statistic (1.1763), with a very high approximate F(10 1776) = 253.61 and significance (p<0.001).

LDA identifies the linear combinations of variables that best separate predefined groups and evaluates how reliably individual specimens can be assigned to them. The analysis revealed visual separation among all three samples, aligning with the statistically significant separation of centroids shown by MANOVA (Fig. 4a, Table 5). Visually, distinction is particularly strong for Morphotype-1 (AO16-8-GC1). Prior probabilities were nearly identical among groups (AO16-8-GC1 Morphotype-1 = 0.328, ODP 151/907A Morphotype-1 = 0.336, and P1-92-AR-40 Morphotype-2 = 0.337), reflecting the balanced sample sizes and ensuring an unbiased classification model (Table 6).

The first linear discriminant (LD1) explains 70.54 % of the between-group variance, while LD2 explains the remaining 29.46 %. LD1 was primarily driven by the minimum final chamber diameter (LD1 coefficient = 1.04), maximum test diameter (0.78), and proloculus diameter (0.51), indicating that overall shell size and final chamber geometry represent the principal axes of morphological separation. LD2 was strongly influenced by proloculus diameter (1.24), with additional contributions from maximum test diameter (-0.62) and apertural lip thickness (-0.43), highlighting proloculus size as an important secondary discriminator among the three populations (Table 5). Group means on the standardized scale indicate that Morphotype-1 from AO16-8-GC1 consistently occupies the “smallest end” of morphospace across all measured traits, with strongly negative mean values for the maximum test diameter (-1.11), minimum final chamber diameter (-1.12), and lip thickness (-0.73). In contrast, “classic” T. quinqueloba exhibits the largest average dimensions, particularly for maximum test diameter (0.70) and lip thickness (0.69). Morphotype-2 displays intermediate overall test size but a notably large proloculus (1.00), separating it clearly from the other two groups in the second discriminant dimension (LD2).

The predicted classification of individual specimens into the three groups (Fig. 4b) showed high discriminatory success with minimal misclassification. Cross-validated classification accuracy reached 88.93 %, demonstrating that the selected morphometric parameters are sufficient for robust specimen-level discrimination (Table 6). Morphotype-1 was particularly distinct, with 98.3 % of specimens correctly classified, forming a compact and isolated cluster with minimal overlap. In contrast, moderate overlap was observed between “classic” T. quinqueloba and Morphotype-2, indicating closer morphometric affinity between these two groups based on the compared traits.

The inter-group morphological differences, including size, are also clearly visible in the individual histograms (Fig. 3). “Classic” T. quinqueloba and Morphotype-2 reach similar overall test sizes and show comparable size variance, although “classic” T. quinqueloba exhibits the highest mean test size (189.63 µm; Table 4). Umbilical area shows distinct distributions for each sample, with Morphotype-2 being the most variable and showing a distinct right-skewed distribution with a long upper tail, a pattern similarly observed in the “Test projection area”. “Final chamber globulation” also displays three distinct peaks, consistent with separation into distinct populations. Of particular importance is the distribution of proloculus diameter, which emerged as one of the strongest discriminating variables in the LDA, especially LD2 (Table 5). Morphotype-1 and “classic” T. quinqueloba, both dated to MIS 5, show nearly identical proloculus sizes (means of 13.72 and 13.95 µm, respectively) despite very different overall test sizes (means of 113.8 and 189.6 µm, respectively) (Fig. 5). In contrast, Morphotype-2 from the older (> MIS 12) CAO horizon has a substantially larger mean proloculus diameter (21.90 µm) and greater variance. Further, “classic” T. quinqueloba and Morphotype-2 show greater visual separation along the LD2 axis (Fig. 4a), where separation is predominantly driven by proloculus diameter (Table 5). Hence, using histogram observations and the results of LDA, we observe that proloculus size is a critical feature that unites Morphotype-1 and “classic” T. quinqueloba and distinguishes them from Morphotype-2.

To summarize, across all analyses, the morphometric data reveal consistent and statistically robust differentiation among the three microfossil populations. Univariate histograms show distinct and often non-overlapping distributions in key traits (e.g., test size, chamber number, and proloculus diameter), while multivariate analyses (MANOVA and LDA) confirm that the populations occupy significantly different regions of morphospace, with high classification accuracy and clear separation primarily driven by overall size and proloculus–chamber geometry.

A fourth morphotype, which under the light microscope can resemble a small Turborotalita, must be considered for accurate identification of subpolar taxa in the central Arctic realm. This form corresponds to a morphotype of N. pachyderma, referred to as “Nps-5” (Eynaud et al., 2009; El Bani Altuna et al., 2018; Vermassen et al. 2025; Weitkamp et al., 2025a). It occurs in all samples where N. pachyderma is common, especially in the > 125 µm sieve fractions. These specimens display a petaloid outline similar to Turborotalita-like morphotypes, with five slightly globular chambers in the final whorl and an apertural lip. They also appear shiny and reflective under reflected light (Plate 1, panels 5 and 6). They are usually smaller than Morphotype-2 (Plate 7, figs. columns A–C) but can also appear in the 125–250 µm fraction, in the water column, at the core top, and in older samples where typical N. pachyderma are common (Plates 1 and 2). Coiling direction is both sinistral and dextral (Plate 1, panel 6). Under SEM, the test wall exhibits irregularly distributed, slightly angular bumps. Importantly, there are no signs of spine holes or broken spine bases, indicating a non-spinose wall and clearly distinguishing this morphotype from Turborotalia or other spinose taxa. Examination of multiple specimens reveals a gradation in wall texture, from isolated bumps to coalescing structures and eventually to coarse reticulating ridges characteristic of N. pachyderma. This non-spinose wall character is clear under SEM, but it can be difficult to resolve under the optical microscope. These have not been included in our morphometric analyses. Despite their striking gross test homeomorphy, the contrasting wall textures confirm their phylogenetic difference.

Consistent with previous analyses of CAO cores, nannofossil occurrences are extremely scarce (Appendix Table A1). They occur primarily within dark brown finer-grained intervals typically associated with bioturbation and the presence of other calcareous microfossils (benthic and planktonic foraminifera) (Hanslik et al., 2013; Razmjooei et al., 2023). Preservation was moderate to poor. Sampling focused on intervals with high abundances of T. quinqueloba (Morphotype-1) and Morphotype-2, in the hope of identifying key nannofossil taxa such as Gephyrocapsa huxleyi and P. lacunosa. These have proved valuable for anchoring recent biostratigraphic chronologies in other CAO cores from the central Lomonosov Ridge and Alpha Ridge (Razmjooei et al., 2023).

LOMROG07-PC04 (Lomonosov Ridge) contained the highest number of nannofossil counts, but no G. huxleyi and P. lacunosa were recorded (Appendix Table A1). The most common species were “small” indiscriminate Gephyrocapsa species and G. caribbeanica, both members of the family Noelaerhabdaceae and indicative generally of a Quaternary age. Gephyrocapsa muellerae and Coccolithus pelagicus are extremely rare, each observed in only two samples. Gephyrocapsa caribbeanica occurs in all samples containing T. quinqueloba but is not present in samples below 185 cm where Morphotype-2 occurs.

On the Mendeleev and Northwind ridges, samples were mainly taken from cores where the Morphotype-2 horizon had been identified (Cronin et al., 2014). In core P1-94-AR-P9 (Mendeleev Ridge), this horizon lies between the 413–427 cm subsurface. No G. huxleyi were found in this core, but a single P. lacunosa was identified at 435 cm. G. caribbeanica occurs in two samples above the Morphotype-2 horizon, and a single nannofossil (Reticulofenestra asanoi) was identified in one of the three samples from the Morphotype-2 horizon.

All but one sample from the other Northwind cores were barren of nannofossils (Appendix Table A1). These cores spanned water depths of 700–1470 m. In P1-92-AR-P40, 17 samples were analyzed, with 13 being from the Morphotype-2 interval, identified between 340–410 cm by Cronin et al. (2014). In P1-92-AR-30, 10 samples were analyzed from the Morphotype-2 interval, identified between 585–625 cm core depth. In P1-93-AR-P21, 2 of the 5 samples were from the Morphotype-2 interval between 200–210 cm. A single specimen of G. caribbeanica was recorded in the lowermost of these samples (210 cm).

In CAO core material, the stratigraphically upper subpolar planktonics (Morphotype-1), found in horizons dated to interglacials between MIS 5 and 11 (Vermassen et al., 2021; Razmjooei et al., 2023; Vermassen et al., 2023), have an enrolled, laterally compressed test and a reduced or tear-drop-shaped final chamber covering the umbilicus. They are clearly spinose, as shown by the regularly distributed spine bases from which spines would have emerged in life. The tear-drop-shaped final chamber is a “classic” T. quinqueloba trait. However, Morphotype-1 differs from classic Nordic Seas/North Atlantic T. quinqueloba, from both core tops (Holocene) and an < MIS 12–MIS 5 horizon, in several ways: the distinctly smaller test size (Fig. 3, mean = 113.8 µm, compared to 189.6 µm in “classic” T. quinqueloba), fewer chambers in the whole test (9.45 compared to 11.01, respectively), more sparsely distributed pores, and a lack of late-stage surface thickening, seen in our sedimented sub-Arctic (Nordic Seas) examples (Plates 1 and 3, figs. 4–5).

Although Morphotype-1 (AO16-8GC-2, 2.5–3.5 cm, Alpha Ridge) and “classic” T. quinqueloba from the MIS 5 samples in the Iceland Sea (ODP 907A-1H-2, 37–39 cm) are statistically distinguishable and fall into separate groups, this alone does not justify their designation as distinct morphotypes or species, since test size is known to vary substantially within planktonic foraminifera in response to environmental and climatic conditions and stress (Bauch, 1994; Schmidt et al., 2003; Todd et al., 2020; Martinelli et al., 2026). Instead, we argue that key taxonomic traits, such as tendency to have an ampullate final chamber, enrolled coiling, and a compressed test, define T. quinqueloba (Natland 1938; Brummer and Kučera 2022) and unite both “classic” T. quinqueloba and Morphotype-1 as a single species. Notably, these diagnostic traits do not appear to be optimally represented or detected in our choice of traits in the quantitative morphometric analysis. In the case of the ampullate final chamber, this may be due to limitations in how we parameterized the variable “final chamber globularity” or to a high degree of intra-population variability that obscures this feature at the population level.

A further striking similarity between Morphotype-1 and “classic” T. quinqueloba is their near-identical mean proloculus size and variance, which contrasts strongly with the consistently larger mean size and broader range in Morphotype-2 (Fig. 3). Proloculus size should be treated cautiously as a taxonomic character because it can exhibit substantial intraspecific variability, including due to reproductive dimorphism (Keating-Bitonti and Payne, 2018; Darling et al., 2023; Meilland et al., 2023). Indeed, wide ranges in proloculus size have been documented in several planktonic foraminifera species, including N. pachyderma from the Greenland Sea and Baffin Bay (ca. 7–40 µm; Meilland et al., 2023) and the CAO (ca. 6–34 µm; Weitkamp et al., 2025a), as well as Globigerina aequilateralis (ca. 10–30 µm; Brummer et al., 1987). Notably, laboratory cultures of both species show that offspring produced in the lab exhibit similar proloculus size ranges to field collections, confirming a broad and stable intraspecific range in these species (Brummer et al., 1987; Meilland et al., 2023; Weitkamp et al., 2025a). However, other taxa exhibit consistently narrow proloculus size ranges; G. ruber (pink and white forms) both have proloculus diameters in the ca. 8–16 µm range compared to G. sacculifer, showing a range of ca. 12–19 µm (Brummer et al., 1987). This indicates that, while variability exists, some species nevertheless express relatively characteristic proloculus size ranges, suggesting a biologically constrained component of variation that may be detectable at the population level. We therefore interpret the close match in proloculus size range between Morphotype-1 and “classic” T. quinqueloba from different regions as supporting their conspecificity. In this context, differences in adult size and other test features are interpreted as environmentally driven, likely reflecting differing or suboptimal growth conditions in the central Arctic Ocean compared to the Iceland Sea. In contrast, we argue that the consistently larger mean proloculus sizes and broader size range observed in Morphotype-2 are unlikely to be explained by ontogenetic stage alone. We therefore interpret proloculus size range as a meaningful biological signal in this context. Together with other morphological traits, this pattern supports interpreting Morphotype-2 as a distinct morphospecies and likely a separate biological taxon, rather than an ontogenetic or ecophenotypic variant of T. quinqueloba.

Based on this evidence, we remain confident in assigning Morphotype-1 to T. quinqueloba, recognizing its morphological and developmental continuity with the “classic” form. Similar morphological variability in Arctic-realm T. quinqueloba has previously been linked to polar environmental gradients. For example, under polar conditions and ice cover of the southern Nansen Basin (north of Svalbard, > 85° N), which is the modern northern limit of the species, relatively small test sizes in water column T. quinqueloba were observed (modal class size ca. 90–120 µm) (Carstens and Wefer, 1992; Jensen, 1998; Stangeew 2001; Meilland et al., 2020; Greco et al., 2022) compared to relatively larger tests ranging from 110 to 140 µm in the subpolar zone (Norwegian Sea) (Stangeew, 2001). Considered “pre-adult” stages, the Nansen Basin specimens were interpreted to be transported individuals, expatriated from their production habitats further south by north-flowing Atlantic Water (West Spitsbergen Current) (Fig. 2) (Carstens and Wefer 1992). Significant expatriation via Atlantic currents was also invoked to explain increasing T. quinqueloba abundance in the Fram Strait over recent decades (Greco et al., 2022). Images of the collected specimens were not included in either of these studies, preventing morphologic comparisons with our CAO Morphotype-1 assemblages.

Variability in T. quinqueloba size is also reported in down-core assemblages in the Nordic Seas (Bauch, 1994). Microscope imagery is again lacking, but the simple sketch included in this study implies a morphology comparable to our CAO Morphotype-1, involving a dwarf form that retains the diagnostic ampullate final chamber covering the umbilicus. Importantly, this study found that Holocene (core top) T. quinqueloba in the vicinity of incoming Atlantic waters (eastern Nordic Seas, “Atlantic domain”) had larger mean test sizes than those from the western part of the Nordic Seas (Greenland Sea, “Arctic domain”) (210 µm versus 140 µm median test sizes, respectively), consistent with a stronger polar influence in the west from the East Greenland Current (Bauch, 1994). It is unclear whether these small polar specimens also had fewer chambers in the final test as seen in our CAO populations. Turborotalita quinqueloba was largely absent from the Nordic Seas during Pleistocene glacials, while N. pachyderma persisted. The former reappeared during the glacial-termination but tests were initially small (100–125 µm) in both the eastern Atlantic and western polar Nordic Seas domains, enlarging, especially on the eastern side, only when full interglacial conditions were established (Bauch, 1997). A similar pattern occurred during the previous termination and the transition into the MIS 5 interglacial (Bauch, 1997). Thus, reduced test size in T. quinqueloba has been broadly linked with cooler waters (Bauch, 1994), though in reality, size (and numbers of chambers in the test) likely reflects multiple interacting factors including temperature, stratification, sea ice, salinity and food availability. At its modern northern limit, a stable ice margin, along which high primary production occurs, may also play a role (Johannessen et al., 1994; Carstens et al., 1997).

Importantly, T. quinqueloba does not currently occur in the water column north of 83° N (Carstens and Wefer, 1992; Volkmann 2000; Snoeijs-Leijonmalm and Party, 2022; Vermassen et al., 2025; Weitkamp et al., 2025a). The similarity in test sizes between our CAO Morphotype-1 T. quinqueloba and glacial-termination or Arctic-domain populations from the Nordic Seas (Plates 1 and 3) suggests parallels in the environmental conditions that shaped their morphology.

Morphotype-2 attains a similar size range to “classic” T. quinqueloba, and the multivariate plots imply some overlap in morphospace. However, there are also clearly different morphological traits, both in the morphometric data and SEM observations of the test ultrastructures. Our microscope images illustrate how Morphotype-2 deviates from Morphotype-1 (T. quinqueloba) in the systematically larger test area, more evolute coiling, a more open umbilicus with distinct apertural lips, and a more rounded final chamber that typically does not extend into the umbilical region. This is supported by the morphometric data, which highlight both the overall larger test size range and the significantly larger proloculus in Morphotype-2 (Fig. 3), compared to CAO T. quinqueloba, leading to a distinct cluster in the LDA (Fig. 4). The low average number of chambers in the whole test of Morphotype-2 (median 8) compared to “classic” T. quinqueloba (11) is likely a function of the typically larger proloculus, as shown in other species (Brummer et al., 1987). The Morphotype-2 state likely reflects a different life strategy or ecological adaptation compared to T. quinqueloba, including more resources allocated to initial growth, possibly linked to different reproductive modes or an overwintering strategy. Fewer chambers with a larger initial chamber might suggest faster early growth or a life cycle adapted to short periods of favorable conditions.

Although our metric for “final chamber globularity” does not appear to fully distinguish the distinctive ampullate/tear-drop shape of the T. quinqueloba final chamber concept, Morphotype-2 exhibits a markedly narrower variance in this parameter (Fig. 3). This suggests a more consistent, rounded chamber shape in Morphotype-2, particularly when compared to the “classic” T. quinqueloba, which shows more variability. In contrast, the umbilical area in Morphotype-2, intended to capture the loose coiling and large umbilicus, shows no clear separation from “classic” T. quinqueloba, although Morphotype-2 does show a right-skewed distribution and a long tail toward larger umbilical areas, consistent with observations. This indicates that while most specimens have moderate-sized umbilical areas, a subset exhibits much larger ones, reflecting substantial coiling variability within the morphospecies and a capacity to have very large open umbilici. This could reflect real biological variability, perhaps related to developmental plasticity or opportunistic ecology/growth. That our measurements do not clearly differentiate the umbilical characteristics of Morphotype-2 may be a consequence of the “umbilical area” metric used: delineation of the umbilical boundary can be subjective and can blur the differences between morphotypes. Instead, measuring the “umbilicus maximum diameter” may better capture these differences and could be explored in future work. Importantly, the use of area does not affect the distinction between morphotypes or the study's conclusions.

Based on the combination of Morphotype-2's distinct and consistent final chamber morphology, proloculus, and growth patterns; its unusually high variability in umbilical area; and its restricted occurrence in deeper stratigraphic levels, together with additional surface texture features not captured by our morphometric analysis, such as the broad spine bases and costellae-like linear alignments of pustules common in spiral view (Plate 5, figs. 28, 29, 32–34; Plate 6, figs. 34, 35), we conclude that Morphotype-2 deserves taxonomic differentiation. This is consistent with previous work that utilizes the occurrence of the morphotype as a stratigraphic marker with paleoceanographic value (Cronin et al., 2013, 2014, 2019; Xiao et al., 2020; Jang et al., 2026). Lumping these forms with T. quinqueloba would result in a loss of important biostratigraphic and paleoceanographic information. In this study, our morphometric populations are limited to single locations. Whether the same conclusions would apply if we repeated the experiments with populations from other stations in the CAO is unknown, although the SEM images of specimens from these different regions imply that the morphologies are similar across the region.

The following question then arises: what name should be applied to Morphotype-2? Although the designation T. egelida has been widely used in recent years, lending its name to the “T. egelida zone” (Cronin et al., 2019), our investigations reveal that this assignment rests on a rather insecure taxonomic foundation. This requires a comprehensive reassessment. In the following sections, we detail the ambiguities and reconsider the taxonomic framework for subpolar planktonic foraminiferal invasions, beginning with a critical review of the Turborotalita in northern polar and subpolar regions.

Since early plankton surveys, T. quinqueloba, originally described as Globigerina quinqueloba Natland (Natland, 1938), has been recognized as a key component of high-latitude North Atlantic and sub-Arctic planktonic foraminifera assemblages (Bé and Hamlin, 1967; Cifelli and Smith, 1970), where it is now known to occur as genetic Types IIa and IIb (Darling et al., 2007, 2017; Morard et al., 2024). High T. quinqueloba abundance is linked to northward-flowing Atlantic surface waters in the eastern Nordic Seas, Barents Sea, and Laptev Sea, where it thrives under “open water” and “sea-ice edge” conditions (Carstens and Wefer, 1992; Bauch 1994; Johannessen et al., 1994; Carstens et al., 1997; Volkmann, 2000; Simstich et al., 2003; Jonkers et al., 2010; Werner et al., 2011; Husum and Hald, 2012; Zamelczyk et al., 2018; O'Regan et al., 2019; Anglada-Ortiz et al., 2025). Turborotalita quinqueloba has long been regarded as a symbiotic surface dweller due to its spinose habit (Hemleben et al., 1989), although its small size has hindered culturing efforts, leaving this interpretation uncertain. Recent plankton net observations, however, report no evidence of algal symbiosis in T. quinqueloba from the mid-latitude western North Pacific (Takagi et al., 2019) or the high-latitude North Atlantic (Armitage et al., 2026), although it remains possible that some subtropical or tropical genotypes host facultative symbionts. The occurrence of significant proportions (ca. > 10 %) of T. quinqueloba, and/or diagnostic N. pachyderma–T. quinqueloba δ18O offsets, is commonly used as paleoceanographic tracers, including mapping of sub-Arctic frontal systems; the sea-ice edge; the strength of seasonality; and, more recently, evidence of increasing “Atlantification” of sub-Arctic and Arctic regions (Johannessen et al., 1994; Alonso-Garcia et al., 2011; Bauch et al., 2012; Husum and Hald 2012; Moffa-Sánchez et al., 2014; Mokeddem et al., 2014; Zhuravleva et al., 2017; Tesi et al., 2021; Greco et al., 2022; Vermassen et al., 2023). The Atlantic Water connection seems robust: today, north of Svalbard, T. quinqueloba quickly disappears as Atlantic Water subsides beneath the strong Arctic halocline, and it is entirely absent from the CAO water column (Carstens and Wefer 1992; Volkmann 2000; Nørgaard-Pedersen et al., 2007; Ding et al., 2014; Vermassen et al., 2025), making the observed fossil T. quinqueloba occurrences in the CAO all the more significant.

The discovery of subpolar plankton invasions in the CAO dates back to the surveys of the 1950s–1970s, notably “Fletcher's Ice Island (T-3)” and the “Arlis and Alpha Ice Island” expeditions, which marked the earliest Arctic marine geology investigations (Schindler, 1968). Yvonne Herman, who led many of the pioneering sedimentological and micropaleontological studies of the time, identified distinct horizons of “subpolar planktonic foraminifera” in multiple sediment cores, including T3 67-3, T3 67-9, and DA2-II from the Mendeleev and Alpha ridges, respectively (Herman, 1974) (Fig. 1). Originally referred to as the “G. quinqueloba complex”, the multiple horizons variably comprised G. quinqueloba and a second morphotype at deeper horizons, which Herman initially referred to as “Globigerina sp., cf. G. quinqueloba” (Herman, 1970) and later named Globigerina exumbilicata (Herman, 1974). Globigerina exumbilicata was described as differing from G. quinqueloba in the spherical shape of the final chamber, which does not overhang the umbilicus, and by having a more distinct apertural lip and a deep and open umbilicus, often surrounded by umbilical teeth. This description is consistent with the holotype (Plate 6, figs. 28–30) and paratypes (Plate 6, figs. 31–36) of the species, illustrated here for the first time, as well as our Morphotype-2 assemblages from cores in the Eurasian and Amerasian basins. We find multiple examples of “Morphotype-2” that bear a striking resemblance to the holotype of G. exumbilicata (Plate 5, figs. 14, 19, 20).

As the first central Arctic studies were unfolding during the 1970s, Herman was contacted by Richard Cifelli regarding the taxonomy of G. exumbilicata. He sent Herman comparative specimens of his species G. egelida, described from his North Atlantic plankton net studies (37–47° N; Fig. 2) (Cifelli and Smith, 1970), and suggested that her Arctic morphotype was conspecific with his egelida. Herman accepted Cifelli's interpretation and, since G. egelida had priority, she recognized exumbilicata as its junior synonym and referred to the Arctic material as G. egelida. This classification was later adopted by Cronin et al. (2013), in the combination Turborotalita egelida. As part of our assessment of subpolar species in the CAO, we revisited this synonymization to address two questions: (1) what was the original concept of G. egelida and (2) is the synonymy of G. exumbilicata with it justified?

Critical to Cifelli and Smith's (1970) work is how they classified quinqueloba. From their descriptions and illustrations, it is apparent that they recognized quinqueloba-like forms in their North Atlantic water column assemblages. However, they proposed that these Atlantic occurrences differed from Natland's species described from California margin sediments and suggested that the Atlantic morphotype should instead be referred to as a new subspecies, which they named G. quinqueloba egelida (Cifelli and Smith). The argument was that the Pacific forms have a thicker test, more compact form, less depressed sutures, and a less lobulate periphery compared to the Atlantic occurrences (Plate 3). The holotypes and paratypes of egelida have been imaged for the first time as part of this study (Plate 6, figs. 7–18). Unfortunately, the holotype of egelida was lost during imaging, and only the spiral view was captured. Nevertheless, we consider the paratype images, together with the original illustration of the umbilical side of the holotype, sufficient to illustrate the range of morphological variability attributable to the taxon. The implications of this are significant: in Cifelli and Smith's classification, the “true” Globigerina quinqueloba (i.e., Natland's nominal species) was implicitly restricted to the Pacific, while the Atlantic and Arctic forms were treated as a different taxon, i.e. G. quinqueloba egelida. Cifelli and Smith's Atlantic vs. Pacific quinqueloba classification was never adopted. Kennett and Srinivasan (1983) considered egelida to be a “phenotypic variant of quinqueloba”. Recently, Brummer and Kučera (2022) also placed egelida in synonymy with T. quinqueloba, concluding that since it was recovered from the water column, it was likely a growth stage of this species.

When comparing the type material, we recognize the broad similarities between Cifelli and Smith's (1970) egelida and Herman's (1974) exumbilicata, particularly the rounded final chambers and open umbilicus, which contrast with the classic ampullate final chamber of T. quinqueloba, in both the “classic” and the dwarf CAO populations recovered from sediments (Plates 3 and 5). However, we point out that these egelida specimens are from the water column, not sedimented. Therefore, we argue that Cifelli and Smith's egelida represents a sub-adult form and are thus not an appropriate reference for comparing the mostly mature, sedimented shells described by Herman (1974) and referred to as the taxon Globigerina exumbilicata. SEM images of live T. quinqueloba from our northern North Atlantic plankton pump samples, confirmed to be a known genotype of T. quinqueloba (Type IIb, Kucera and Darling 2002), support this: the identified specimens, pumped from 5 m subsurface, are concluded to largely represent juvenile specimens that lack a well-developed ampullate final chamber (Plate 4). Since the ampullate overhanging final chamber generally appears only in sedimented populations, this must be a final-stage adult feature occurring in individuals that have reached full maturity. Our plankton material in fact closely resembles the G. egelida-type specimens (Plate 4; Plate 6, figs. 7–18). Ambiguous identification of immature planktonic foraminifera in the water is a general challenge for many species, as the final-stage chamber morphologies and test ornamentations useful for taxonomy are often added only in the terminal stages (Hemleben et al., 1989). From these arguments, we assert that central Arctic exumbilicata Herman, as sedimented fossils, should not be referred to as Cifelli and Smith's taxon “G. egelida”; the former represents a mature morphotype, while the latter is an immature form (Kennett and Srinivasan, 1983; Brummer and Kučera, 2022). Since our morphometric and SEM analyses find clear differences between Morphotype-1 (T. quinqueloba) and Morphotype-2 in the CAO, they require taxonomic differentiation. We therefore reinstate exumbilicata Herman as the most appropriate name for Morphotype-2.

Of relevance here is another species described from the same study, Globigerina atlantisae Cifelli and Smith, also SEM illustrated here for the first time (Plate 6, figs. 19–24). Although outwardly similar to T. quinqueloba in its tight coiling and ampullate final chamber, Cifelli and Smith distinguished G. atlantisae by its more curved sutures, elongate chambers (umbilical view), and consistently umbilicus-covering final chamber. There are also fewer total numbers of chambers (10–11 vs. 13–14) in G. quinqueloba egelida, with generally only 4 chambers in the final whorl of G. atlantisae. Our new SEM images show that G. atlantisae lacks pores and bears randomly distributed small pustules rather than spine bases, indicating a non-spinose, microperforate wall (Plate 6, fig. 22). This confirms its affinity with Tenuitellita iota (Parker, 1962), as proposed by Brummer and Kučera (2022). Notably, Cifelli and Smith's taxonomic checklist (fig. 4, p. 7, Cifelli and Smith, 1970) lists atlantisae as “Globigerinita atlantisae”, reflecting uncertainty about its generic assignment. We therefore exclude atlantisae from further discussion of Arctic Turborotalita, although we note that T. iota may be mistaken for small T. quinqueloba.

After Herman's work during the 1980s, no further reports of egelida were published until the 2010s when new findings of subpolar planktonic foraminifera invasions at relatively deep subseafloor depths (several meters) were reported and referred to as the “egelida zone” (Cronin et al., 2014). Cronin et al. (2013) were the first to use the binomial Turborotalita egelida, later illustrated with SEM images in Cronin et al. (2014). We interpret Cronin et al.'s (2013) assertion that “T. egelida-dominated … assemblages have no known modern analogue” as correct. In contrast, Herman implied that egelida is extant and can be differentiated from quinqueloba (Herman and Hopkins, 1980; Herman, 1983), an interpretation that is inconsistent with Cifelli and Smith's concept and the synonymization of egelida under quinqueloba.

The complex history of the application of egelida, quinqueloba, and exumbilicata in the Arctic and sub-Arctic Atlantic, as outlined here, reveals significant confusion and misapplication of these taxa. This underscores the need for taxonomic clarification to advance stratigraphic and paleoenvironmental interpretations of subpolar planktonic foraminifera occurrences in the central Arctic. With our accompanying image catalogue – featuring whole-specimen and wall texture SEMs that underpin higher taxonomy – alongside new morphometric data, we propose a path forward toward improved classification and taxonomic resolution. To summarize, the key issues that we have unearthed are the following.

In the CAO, at least two stratigraphic horizons can be recognized involving acmes of species other than N. pachyderma: the stratigraphically upper (younger) horizons (Morphotype-1) involve acmes of dwarf T. quinqueloba, whereas the deeper (older) horizons involve a different morphotype (Morphotype-2), which was originally named G. exumbilicata Herman 1974 but was subsequently synonymized under a poorly known species, G. egelida Cifelli and Smith 1970, to which it bears some resemblance.

Cifelli and Smith (1970) had established egelida as a subspecies of G. quinqueloba from living plankton samples in the North Atlantic. It was differentiated from Natland's (1938) species G. quinqueloba described from the Pacific Ocean, on the grounds that the Atlantic forms lack the elongate final chamber typical of G. quinqueloba. The intention was that “egelida” should be a general replacement for occurrences of G. quinqueloba in the North Atlantic. This is not consistent with the subsequent usage of egelida in the context of its application in Arctic Pleistocene sediments (Herman, 1974; Aksu, 1985; Cronin et al., 2019; Vermassen et al., 2021).

Our new SEM images of “live” T. quinqueloba from the water column south of Iceland closely match the morphology of Cifelli and Smith's egelida-type material in lacking the elongated final chamber typical of the species. As acknowledged previously, the presence of an elongate, embracing final chamber thus represents a mature stage seen in sediment specimens (Brummer and Kučera, 2022). DNA analysis of the netted immature T. quinqueloba imaged herein also supports the morphotype being T. quinqueloba (Type IIb) (Kucera and Darling, 2002; Darling and Wade, 2008). We therefore conclude that the concept of egelida is (a) not appropriate for defining a new species that will largely be identified in sediments and (b) should not replace G. exumbilicata Herman 1974 as the valid prior synonym.

The consistent morphological differences observed between Morphotype-2, in particular the larger proloculus, more open umbilicus, and differing wall texture and ornamentation, provide strong support for biological and taxonomic differentiation from Morphotype-1 (T. quinqueloba). Combined qualitative, quantitative, and stratigraphic evidence confirms that they represent separate morphospecies.

There is a potential to confuse both CAO T. quinqueloba and Morphotype-2 with a smooth-walled open-coiling morphotype of N. pachyderma (Nps-5), leading to the potential for misinterpretation in paleoceanographic reconstructions. Comparative wall texture SEM analysis can easily resolve this.

Given the stratigraphic and morphologic arguments for taxonomically differentiating Morphotype-1 and Morphotype-2 and the unavailability of egelida (being a growth stage of T. quinqueloba), we reinstate Herman's (1974) species name exumbilicata for Morphotype-2, replacing the current usage of T. egelida. Although no image of the holotype was provided in the original description, a type specimen and repository were designated (Herman, 1974, p. 300), satisfying the requirements of the International Code of Zoological Nomenclature (Article 10.1); the taxon is therefore available.

The generic placement of exumbilicata is less straightforward. Both the original and the emended definitions of the genus Turborotalita focus on the presence of an ampullate final chamber (originally described as a bulla) that extends into the umbilicus, variously described as “encroaching”, or “adumbilically displaced”, and the potential to be covered by a thick calcite crust (Blow and Banner, 1962; Pearson and Kucera, 2018; Brummer and Kučera, 2022). Systematic absence of these features in the populations of exumbilicata investigated in this study suggests that the morphospecies does not belong in Turborotalita but instead may be better placed in Globigerina. The species “exumbilicata” broadly satisfies the criteria for Globigerina in having a “bulloides-type” wall, although modified and more similar to that seen in e.g. Oligocene Globigerina officinalis (Spezzaferri et al., 2018) (plate 6.3), and with broader spine bases. The costellae-like pustule ridges may also be significant. This feature has, to our knowledge, not been observed in any species of Turborotalita. However, similar features have been illustrated in some specimens of Globigerina falconensis (e.g., Schiebel and Hemleben, 2017, pl. 2.4, fig. 4) and G. neofalconensis (Fabbrini et al., 2023, pl. 4, fig. f). Morphologically, exumbilicata also shares a lobulate outline and apertural lip development with members of the G. falconensis group, although its wall is generally less reticulate. On the other hand, the umbilical–extraumbilical aperture of “exumbilicata” is inconsistent with the current genus level definition of Globigerina, which has an umbilical aperture (Fabbrini et al., 2023).

Moreover, as discussed above, the distinction between Turborotalita and Globigerina is complicated by ontogenetic variability. In T. quinqueloba, final chamber elongation, umbilical overhang, possession of a terminal flap, and encrustation develop late in ontogeny. As a result, “simplified” T. quinqueloba specimens can resemble exumbilicata, highlighting a broader issue in planktonic foraminiferal taxonomy: key diagnostic taxonomic characters often emerge during ontogeny and are variably represented, making their recognition dependent on sampling. This challenge is widespread, with immature specimens of many spinose genera, including Globigerina, Globigerinoides (and Trilobatus), and Orbulina being difficult to distinguish (Brummer et al., 1987). In addition, as reported here, statistical analyses of morphological traits show that different taxa commonly occupy partially overlapping regions of morphospace (Tabachnick and Bookstein, 1990; Malmgren et al., 1983; Pearson, 1993; Ezard et al., 2010; Pearson and Ezard, 2014). Consequently, even genus-level boundaries, such as those between Turborotalita and Globigerina, become less distinct when juvenile and ecophenotypic variability are taken into account and are most robust when based on adult (sedimented) population morphologies.

We thus conclude that within this framework, exumbilicata forms a consistent and distinct population-level morphology in CAO sediment assemblages, characterized by a globular final chamber, wide umbilicus, umbilical–extraumbilical aperture, non-encrusted wall, pustulose ridges, and a commonly large proloculus and variable proloculus. This repeated association across sites indicates a coherent morphotype that is not fully explained by incidental overlap with immature T. quinqueloba. Moreover, it lacks many of the diagnostic traits that would support assignment to genus Turborotalita. We therefore adopt a conservative approach and refer to this taxon as “Globigerina” exumbilicata, recognizing that a definitive generic assignment will require broader comparative analyses of globigerinid and turborotalitid morphology and test ultrastructure, extending beyond the central Arctic Ocean. In this respect, the co-occurrence of Morphotype-2 and T. quinqueloba observed in some of our assemblages also warrants further investigation. This provisional nomenclature is presented in the “Systematic taxonomy” section below.

Thus far, “G.” exumbilicata has not previously been reported outside the CAO. A morphotype bearing a striking resemblance to it was recorded as abundant by Poore (1979) in samples biostratigraphically dated as “late Pliocene to mid-Quaternary age” (nannofossil zone range N16/N19) (Bukry, 1979; Martini, 1979; Poore, 1979; Steinmetz, 1979), from DSDP Site 409, south of Iceland. Referred to as “Globigerina sp. A”, Poore's SEM images (pl. 19, figs. 5–12 of Poore, 1979) reveal various features common to “G.” exumbilicata, including open coiling, multiple relict apertural lips, a large proloculus, and linear pustulose ridges, suggesting the morphotypes are conspecific. Likewise, Globigerina antarctica Keany and Kennett 1972, a little-known taxon that briefly appeared as an acme species in the southern Indian Ocean Antarctic polar frontal system during the mid-Pleistocene (Matuyama magnetic reversal chron), represents a potential southern high-latitude analogue that warrants further investigation. This species has long been regarded as a junior synonym of Globigerina falconensis (Kennett and Srinivasen, 1983) although this was rejected by Fabbrini et al. (2023).

With our revised taxonomy providing a new framework for differentiating subpolar species in the CAO, we return briefly to the question of the timing of the enigmatic “G.” exumbilicata acmes. Herman (1974) had originally assigned it to the Matuyama magnetic chron based on the paleomagnetic reversal stratigraphy of Clark (1970). This implied an age in the range of 2.595 Ma–773 ka (Ogg, 2020), which spans MIS 104 to MIS 19 of the global benthic δ18O chemostratigraphy (Lisiecki and Raymo, 2005). These initial paleomagnetic-based age assignments in Arctic sediments have been questioned, and subsequent age models, which utilized biostratigraphy and cyclostratigraphy, assigned the exumbilicata horizon to MIS 11 (Jakobsson et al., 2001; Cronin et al., 2013; Cronin et al., 2014; Xiao et al., 2020).

In 2020, the calcareous nannofossil Pseudoemiliana huxleyi was first identified in marine sediment cores from the CAO (O'Regan et al., 2020). The relative of P. huxleyi, P. lacunosa went extinct globally during MIS 12 (424–478 ka) and now provides an important middle Pleistocene biostratigraphic age constraint for CAO sediments. Shortly after its discovery in the CAO, Vermassen et al. (2021) showed, through stratigraphic correlation between the LOMROG12-3PC record containing P. lacunosa and the AO16-8GC record containing “G.” exumbilicata, that the “G.” exumbilicata horizon occurred below the highest occurrence (HO) of P. lacunosa, indicating that it was not an MIS 11 (424–374 ka) marker but must, therefore, be older than MIS 12 (434–432 ka). At the time they suggested an age of MIS 15/MIS 17 (> 580 ka).

Subsequently, Razmjooei et al. (2023) provided a detailed nannofossil biostratigraphy for AO16-8GC and, in doing so, revised the stratigraphic placement for the HO of P. lacunosa, placing it higher in the lithostratigraphic column than previously suggested. In AO16-8GC, the HO of P. lacunosa is at 138 cm, and the highest level containing “G.” exumbilicata is 187–198 cm (Vermassen et al., 2021) (Fig. 1), indicating that the “G.” exumbilicata interval is “considerably” older than MIS 12. Surprisingly, AO16-8GC is the only reported record from the CAO containing both “G.” exumbilicata and P. lacunosa. None of our new nannofossil studies on the Northwind Ridge, Mendeleev Ridge, or the Lomonosov Ridge provide further constraints on the age of this zone. In fact, our new nannofossil observations presented here indicate that, despite the distinct nature and remarkably good carbonate preservation of the “G.” exumbilicata interval, the nannofossil assemblage is unremarkable, being characterized by extremely low abundance and diversity or complete absence of nannofossils.

Tentatively, a “younger-than age” is suggested by the dinoflagellate cyst biostratigraphy from IODP Expedition 302 (Arctic Coring Expedition), where the correlative horizon for “G.” exumbilicata occurs above the acme of Filisphaera filifera and the highest persistent occurrence of Habibacysta tectata. These events, when correlated to records from the Nordic Seas and Barents Sea margin, are argued to be > 1.8 and 2.0 Ma, respectively (Matthiessen et al., 2018). This suggests that the “G.” exumbilicata interval is older than MIS 12 (ca. 0.44 Ma) and younger than 1.8 Ma. This age framework is broadly consistent with recent stratigraphic interpretations (Jang et al., 2026), although resolving the age of the “G.” exumbilicata event remains a major question in Arctic paleoceanography. Addressing this uncertainty will require multidisciplinary efforts integrating stratigraphy, geochronology, and paleoenvironmental analysis to better understand Pleistocene climatic and biotic transitions both within and outside of the CAO, where more diverse dating tools, including robust paleomagnetic reversal and oxygen isotope stratigraphy, are available. This is a priority for future research.