Coccoliths were classified at the species level following and the electronic guide to the biodiversity and taxonomy of coccolithophores Nannotax3 by . Specific taxonomical considerations regarding E. huxleyi specimens were taken into account, and morphotypes were differentiated as far as it was possible on single coccoliths directly during the counts.

In an additional count, we differentiated between five E. huxleyi morphotypes within morphogroups A and B. These are type A, type A overcalcified and type R (comprised in morphogroup A) as well as type B/C and type O (within morphogroup B). Our approach thus slightly differs from the morphotype identification of within morphogroup B. Due to the fact that we could not always distinguish with certainty between types B, B/C and C, we considered here type B/C as a mixed classification for coccoliths resembling characteristics of these three types, with a size ranging across the typical threshold at 3.5 µm . In our studied samples, size was the only coherent characteristic that differed between specimens within morphogroup B (excluding morphotype O) with a normal distribution maximum between approximately 3 and 4 µm and showing no indication for distinct morphotype distributions (see Table  and Fig. ). Hence, we classified all E. huxleyi coccoliths from morphogroup B into either type B/C or type O, depending on the central area . In total, we classified between 53 and 115 E. huxleyi morphotypes in most samples in the SAZ and four samples in the AZ, while we could only classify between 11 and 48 E. huxleyi morphotypes in eight samples from the AZ and SZ + CZ and in one sample in the SAZ (see Table ).

For a subset of 22 samples, the morphometry of E. huxleyi coccoliths was performed and analysed on 570 SEM micrographs of single flat-lying E. huxleyi coccoliths at a magnification of 10 000 × (347 of morphotype B/C, 156 of morphotype O, and 67 of morphotypes A and A overcalcified together). Images of E. huxleyi coccoliths were measured with the Coccobiom2 macro in the Fiji software programme . Measurements were done in micrometres (µm) based on the scale bar of the SEM images. They were scaled to 100 % with a Coccobiom2 SEM calibration of 1.09 and the chosen magnification.

From the morphometrical measurements on E. huxleyi the coccolith masses were estimated based on two different formulas, that of and that of . studied coccospheres and coccoliths of seven different Noëlaerhabdaceae species including three strains of E. huxleyi by means of X-ray nanotomography. They found that coccolith mass correlates with grid perimeter and with crystal number. They developed the following empirical formula (Eq. ) to calculate the coccolith mass m (in pg) from the number of segments n (with two constants, kn=4.73×10-5±0.28×10-5 and β=3.175±0.251). The number of segments (n) was calculated with p being the peripheral grid perimeter (central-area perimeter) and w being the average width of the calcite crystals (also known as T elements) that we set to 0.12 µm (see ). This is based on a calcite density of 2.71 pg µm-3. However, this formula does not account for calcification of the central area itself, as found in the E. huxleyi morphotype A overcalcified observed in this study.

4m=kn⋅nβwithn=pw

We additionally used the well-established approach to estimate coccolith masses by . Their approach (Eq. ) is based on the length l of a coccolith together with a species-specific and morphotype-specific shape factor ks and the calcite density C (for coherency with the formula of we used here of 2.71 pg µm-3).

5m=C⋅ks⋅l3

The different shape factors used were based on the identified morphotype following : ks=0.02 for morphotypes A and B/C and ks=0.04 for morphotype A overcalcified. The shape factor for morphotype O (ks=0.015) was introduced by in a plankton study along the Patagonian Shelf for a morphotype with a central area described as an “open or thin plate” which the authors called type B/C but that we identified as morphotype O based on the published images and description of . The approach of performing morphometric measurements on the coccoliths followed by the estimation of their coccolith mass assuming a systematic relation between length and thickness was chosen because all data were determined on the same material using SEM. compared coccolith carbonate estimates from a birefringence-based approach with the morphometrics-based approach from and showed, on average, slightly higher but largely comparable carbonate contents for E. huxleyi coccoliths using the latter approach.

We used biogeochemical parameters to test how much of the species assemblage composition could be explained by environmental factors (see also Sect. ). Those are annual salinity, temperature and phosphate at 10 m water depth which we extracted from the 1∘ GLODAPv2 mapped climatology Global Ocean Data Analysis Project; to stay consistent with the calculated carbon system parameters based on the same data product. Carbon system parameters for the location of each sample at its respective water depth were calculated using the CO2SYS macro for PC based on salinity, temperature, silicate, phosphate, alkalinity and total CO2. The data from GLODAPv2 and the derived carbonate system data have a seasonal bias towards Southern Hemisphere winter (December to March) because samples in this database were mostly taken during the austral summer . Coccolithophores in the southeastern Pacific sector of the SO bloom mostly during austral spring and summer months , and thus the coccolithophore assemblages in surface sediments are biased towards the same season as the environmental dataset.

Austral winter photosynthetically active radiation (PAR) at 10 m water depth was estimated using a model of light penetration , the diffuse attenuation coefficient for downwelling irradiance at 490 nm and Eq. (1) in . Mixed-layer depth (MLD) was extracted from monthly 1∘ Argo MLD climatology (; based on a density algorithm). Data from austral spring and summer months (September to March) were averaged and extracted from the respective sample positions.

Prior to any statistical analysis, we excluded three samples (PS97/077-1, PS97/079-1 and PS97/071-2) because of the very low number of coccoliths counted (<40 per sample). We identified suitable ordination methods by applying detrended correspondence analysis (DCA) on the species relative-abundance dataset. DCA resulted in a first axis length of 1 SD, suggesting a short gradient for which linear ordination methods are more appropriate. The relative-abundance data were standardized using a Hellinger transformation, which is well suited for species abundance data, to make variability of the species abundances comparable and give low weights to rare species , using R package adespatial 0.3-8 . We chose the Hellinger distance measure because it is metric and performs well in linear ordination . To find groups of samples that are most similar to each other, average-linkage (unweighted pair group method with arithmetic mean, UPGMA) hierarchical clustering was performed on the assemblage data with the R function hclust , because it takes into account the average pair-wise distance between all members of clusters. The best number of clusters was suggested by the majority of 30 indices calculated with the R package NbClust version 3.0 . The significance of each cluster was assessed by multiscale bootstrap resampling with 10 000 replications using the R package pvclust version 2.2-0 to assess the stability of the clusters.

We assessed relationships between environmental and biotic data using a transformation-based redundancy analysis (tb-RDA). We constrained our assemblage data (response of 16 species at 25 sites) to seven standardized environmental variables (explanatory): salinity, temperature, phosphate and PAR at 10 m water depth; CO3 as a representative variable for the carbonate system at surface sediment sample depth; the MLD; and the respective surface sediment sample depth itself. The adjusted R2 was calculated, and the significance of the tb-RDA was tested at 9999 permutations. Analysis was performed using the R package vegan version 2.5-6 (decostand, rda, RsquareAdj and anova.cca; ). We determined the similarity between the water column and fossil coccolithophore assemblages in this region by calculating the analogue distance of a subset of the sediment surface samples to the nearest plankton samples using the R package ggpalaeo version 0.0.0.9005 (; see Fig. ). All figures were made with the R package ggplot2 version 3.3.2 and Inkscape version 1.0.2 and Inkscape 1.1.1 .

All surface sediment coccolith assemblages in the study area consist of E. huxleyi, Gephyrocapsa muellerae and Calcidiscus leptoporus (intermediate morphotype; hereafter only referred to as C. leptoporus according to ), and in all, except in one sample, we identified Gephyrocapsa ericsonii as a main contributor. Those species make up on average 94 % of the coccolith assemblages in the SAZ, 93 % in the AZ and 87 % in the SZ + CZ. Additionally, C. leptoporus small, Florisphaera profunda, Gephyrocapsa oceanica, Helicosphaera carteri and Syracosphaera spp. are present in low relative numbers in several of the studied stations in the DP and the Chilean margin (see Fig. ).

The majority of the surface sediment assemblages are dominated by E. huxleyi coccoliths with average abundances of 72 % in the SAZ, 60 % in the AZ and 37 % in the SZ + CZ. Emiliania huxleyi gets replaced by gephyrocapsids (G. muellerae and G. ericsonii) and C. leptoporus at the deeper stations (i.e. below 3.1 km), from the SAZ (gephyrocapsids: 30 %, C. leptoporus: 40 %) and AZ (gephyrocapsids: 36 %, C. leptoporus: 19 %) to the SZ + CZ (gephyrocapsids: 47 %, C. leptoporus: 6 %). Gephyrocapsa muellerae and G. ericsonii are found in all stations, while G. oceanica is not observed in some samples from the SAZ along the Chilean margin and is completely absent in the open-ocean SAZ. Following those gephyrocapsids, the next most abundant coccoliths belong to the family Calcidiscaceae (mainly C. leptoporus), which makes up 13 % on average of the coccolith assemblage. Calcidiscus leptoporus is found in every sample, while C. leptoporus small is missing in some samples from the SAZ along the Chilean margin and in one sample from the SZ + CZ (Fig. ). Calcidiscus quadriperforatus is exclusively present in samples along the Chilean margin. Noteworthy minor species include F. profunda, H. carteri and Syracosphaera spp. with average relative abundances of a maximum of 2 % and with occurrences in all frontal zones (Fig. ). Other species scattered across all biogeographic zones are Reticulofenestra sessilis, Umbellosphaera spp., Coccolithus braarudii, Algirosphaera robusta and other unidentified taxa. Rare species include Syracosphaera pulchra, Umbilicosphaera foliosa, Umbilicosphaera sibogae, Calciosolenia brasiliensis, Coccolithus pelagicus, Oolithotus fragilis, Umbilicosphaera hulburtiana, Coronosphaera mediterranea, Pontosphaera spp. and further reworked specimens scattered across the DP in three stations (Fig. ). See the additional material referenced in the Data availability section for further details.

We performed a hierarchical cluster analysis on the relative species assemblages resulting in two groups (see Sect. ). Group A consists of samples that stem from water depths above 3.1 km from the SAZ and AZ. Group B consists of samples that are either from the SZ + CZ or from the AZ below or just above 3.1 km, with one sample from the SAZ far below 3.1 km.

Additionally to sample depth, we assessed possible further drivers of the assemblage distributions using RDA. Constraining the surface sediment species assemblages to sample depth, CO3 as a representative variable for the carbon system at sample depth and the MLD, PAR, temperature, phosphate and salinity at 10 m water depth result in a significant (p=0.0004) RDA and a significant first axis (pRDA1=0.0005) at 10 000 permutations. In line with the previous findings, the highest significance is found for the surface sediment sample water depth (p=0.0011). Further important drivers are PAR at 10 m (p=0.0222), salinity at 10 m (p=0.0269) and phosphate at 10 m (p=0.0314). Depending on the significance threshold, temperature at 10 m could also be considered relevant (p=0.0541). These environmental parameters are negatively correlated to the first RDA axis (RDA1) with scores of -0.76 (sample depth), -0.59 (PAR), -0.67 (salinity), -0.63 (phosphate) and 0.51 (temperature).

Emiliania huxleyi coccoliths belonging to morphotypes A, A overcalcified and R are mainly found in the SAZ (Fig. ). Morphotype A is found in very low numbers across all zones, making up on average 6 %, ranging from 1 % to 27 %, of the identified E. huxleyi morphotypes. In particular, morphotype A overcalcified is abundant in the northernmost samples along the Chilean margin (reaching up to ca. 16 %) with decreasing numbers towards the south (ca. 5 %). Few coccoliths of E. huxleyi morphotype A overcalcified were identified in some samples in the AZ, but they are absent from SZ + CZ locations. Morphotype R is restricted to samples in the SAZ, and it is present only in low relative abundances (1 %–7 %) with an average of 3 %. Morphogroup B is dominating the E. huxleyi coccolith assemblages in the studied surface sediment samples, with 13 % to 56 % of the coccoliths belonging to morphotype O (on average 31 %) and ca. 33 % to 83 % to morphotype B/C (on average 59 %). Morphotype O was found to be slightly more abundant in the SAZ, while morphotype B/C was more prominent in the AZ.

Emiliania huxleyi coccoliths belonging to morphogroup B predominate in the study area, showing a size variation of type B/C coccoliths from 2.22 to 4.78 µm and of type O coccoliths from 2.21 to 4.63 µm (Fig. ). Coccoliths belonging to morphotypes A and A overcalcified were only measured in the SAZ with lengths of 2.47 to 3.85 µm. Within morphotype B/C there is a decreasing size trend from north to south, with maximum lengths of 4.78 µm in the SAZ, 4.15 µm in the AZ and 4 µm in the SZ + CZ. Morphotype O reaches comparable lengths of up to 4.63 µm in the SAZ.

Overall, the number of coccoliths in morphogroup B (types B/C and O) longer than 4 µm decreases from 3 % to 50 % in the SAZ to up to 6 % in the AZ, while the number of coccoliths smaller than 3.5 µm increases from 16 % to 72 % in the SAZ to 60 % to 70 % in the AZ. Coccoliths in morphogroup A (types A and A overcalcified were measured, but morphotype R was not) are considerably smaller than the coccoliths in morphogroup B.

Mass estimates after and (see Sect. ) based on the morphometrical measurements of E. huxleyi coccoliths are comparable (Eq.  after , and Eq.  after ) with an overall mean of 2.12 pg (Eq. ) and 2.29 pg (Eq. ), respectively (see Table  and Fig. ). The mass estimates using the first formula were generally a little lower, although it also yielded maximum mass estimates of >8 pg for some coccoliths of morphotype B/C, for which Eq. () estimates maximum values of up to 6 pg. The lowest masses were found for morphotype A coccoliths (ca. 1–3 pg) with both formulas. At the same time, Eq. () largely underestimates the masses of morphotype A overcalcified coccoliths (to mostly under 2 pg), while Eq. () estimates masses of 2–6 pg for those coccoliths. The highest range in mass was found in the SAZ, with 8.42 and 5.7 pg, respectively.

The observed high numbers of coccoliths in the surface sediments offshore southern Chile are in good agreement to previously reported higher coccolith accumulation rates off central Chile , an area that is however, influenced by an active coastal upwelling system. The generally high numbers of coccoliths along the Chilean margin within the SAZ also suggest relatively high coccolithophore productivity conditions in the surface waters. For the present study area, elevated nutrient supply via freshwater runoff by precipitation and seasonal glacier melting (e.g. ) and relatively warm surface water temperatures in the CHC in comparison to the overall study area are plausible causes for an elevated coccolithophore production and the related increase in coccolith sedimentation off southern Chile. Slight differences in coccolith abundances between deeper mid-slope sediments (between 1.8 and 2.5 km depth, >2500×106 Coc/g sed.) and shallower sediments from the upper slope (0.6–1.2 km depth, < max 1000×106 Coc/g sed.) in the SAZ are probably due to dilution of the shallow samples by high sedimentation rates via freshwater runoff near the coast. However, steadily increased coccolithophore abundances and diversity from coastal to oceanic regions as described for areas further to the north may also account for variations in coccolith abundances in the studied slope sediments off southern Chile.

The coccolithophore species composition in surface sediments studied here, dominated by E. huxleyi and Gephyrocapsa species, agrees well with assemblages found in nearby surface sediment stations located further north offshore Chile as well as to those found in overlying plankton samples (e.g. ). Some species are identified in both surface sediment and plankton samples (i.e. E. huxleyi, G. muellerae, C. leptoporus and Syracosphaera spp.), while a few other species are only present in plankton samples. Emiliania huxleyi is the dominant species present in all the surface sediment samples from the study area, found even in samples in the SZ + CZ and AZ (see Fig. ). This species dominates the coccolithophore communities in the open-ocean surface waters of the study area (e.g. ), but it was also the dominant coccolith species in sediment traps of the SAZ and AZ south of Tasmania and of the AZ on the Kerguelen Plateau . Furthermore, it was also the only coccolithophore widely distributed along the fjords and inner channels of southern Patagonia . Previous studies in the SO have observed a distribution clearly limited up to the extent of the SACCF , which was interpreted as a minimum thermal tolerance of 1 ∘C for this taxon in the SO. This is supported by culture experiments which found out that temperature was the most important driver controlling both cellular particulate organic and inorganic carbon content in E. huxleyi . However, differences in the composition of the morphotypes within this species complex may indicate locally adapted populations with genotypes that have very different temperature sensitivities. The more common occurrence of heavily calcified E. huxleyi types A overcalcified and R offshore Chile (Fig. ) are well in accordance with observations with plankton studies from the study area (e.g. ) and may indicate an adaptation to the physicochemical conditions in the CHC. In addition, a similar shift in the morphotype composition of E. huxleyi assemblages towards morphogroup B in the AZ (Fig. ) along the Patagonian Shelf and in the Australian sector was interpreted as shift in dominance of different ecotypes. This highlights the important role of the variation in E. huxleyi morphotypes on the control of coccolith shape and stresses that any morphometrical information should not be interpreted isolated from morphotypic information.

Calcidiscus leptoporus, G. muellerae and G. ericsonii are common species in the surface sediments off southern Chile (Fig. ). While C. leptoporus was often present in relatively high abundances also in the plankton of the study area up to the PF , G. muellerae and G. ericsonii have been described mainly further north in the SAZ . These species were also secondary components of the coccolith sinking assemblages in sediment traps located in the SAZ south of Tasmania . Calcidiscus leptoporus is generally known as a species with a tendency to be found towards cool waters and moderately to highly nutrient-rich environments . Gephyrocapsa muellerae has also been observed in cold and nutrient-rich environments characterized by high productivity and temperate and mesotrophic regions . The species was even found at moderate abundances (up to 36 % of total numbers) south of the Falkland Islands and close to the SAF . Gephyrocapsa ericsonii was linked to cool and fresh surface waters especially in upwelling or frontal mixing zones, for example at the Chilean coast or at the Argentinean margin, where the Malvinas Current and the SAF transport Antarctic waters northward . Thus, the common abundances of these species in the study area could be explained by their preference for mixed and moderately to highly nutrient-rich environments, where high productivity occurs. The slight discrepancy between the continuous presence of C. leptoporus, G. muellerae and G. ericsonii in sediment and their episodic occurrence in plankton could be due to the episodic coccolithophore blooming, which might not have coincided with the timing of the plankton sampling. This finding is supported by the pronounced seasonality in the coccolith fluxes with marked intensifications during the peak blooming season in austral summer . In this regard, the coccolith content and diversity in surface sediments would represent a smoothed multi-annual signal, being potentially more representative of the conditions in the overlying surface ocean.

Coccolith species occurrence documented in the present SAZ surface sediments is consistent with previous reports on coccolithophore assemblage compositions in the surface waters , sediment traps and surface sediments and are more diverse than those found in the AZ sediment traps . Thus, the generally good agreement between (sub-)fossil sediment samples and living assemblages indicate that the regional oceanography plays an important role in shaping the structure of the coccolithophore community offshore Chile and broadly north of the PF.

South of the PF, coccolith abundances in surface sediments are strikingly lower than at the Chilean margin, but coccoliths are still relatively common in this area (up to 650×106 Coc/g sed.). It is also noticeable that a decrease in coccolith diversity is not observed in the present surface sediment samples (Fig. ). Thus, the situation south of the PF contrasts with the rapid decline in diversity and coccolithophore abundance observed in plankton samples in the DP and in other sectors of the SO (e.g. ). At the same time, no analogues were found between the species assemblages of a subset of surface samples and the nearest plankton samples from (see Fig. ). This discrepancy could be explained by the episodic nature of coccolithophore bloom events south of the PF , which might not have coincided with the timing of the plankton sampling in the DP, while the sediment record in surface samples averages hundreds or even thousands of years.

Emiliania huxleyi remains the dominant species in the DP and is – as off southern Chile – accompanied not only by G. muellerae and C. leptoporus but also by relatively dissolution-resistant taxa such as G. oceanica and F. profunda. While the latter species are selectively enriched here due to the dissolution of the smaller and more fragile taxa, some other species are not even observed in the overlying plankton samples offshore southern Chile and across the DP. These subordinate taxa found in DP sediments are more typical of lower latitudes where they usually live in warmer surface waters. For example, one of the taxon found in unusual abundances at these high latitudes is F. profunda, which is a typical subtropical–temperate species that dwells in the lower photic zone and is rarely present in high abundance at latitudes outside 30∘ N and S . Some of these low-latitude species were found in surface water samples retrieved via the ship's water pump system by during a cruise in austral autumn 1992 in the Weddell Sea, east of the DP. Species belonging to the genus Syracosphaera were mostly found offshore Chile and in the SAZ, in agreement with previous plankton studies in the DP , but they were also found in the AZ. Although different authors also found taxa of this genus in the SO (e.g. ), it was never observed so close to Antarctica, except by . However, Syracosphaera species have been described from a very similar setting off southeastern Greenland . There, a quite diverse upper photic zone assemblage dominated by E. huxleyi and Syracosphaera spp. once occurred, despite the harsh environmental conditions with sea surface water covered by ice most of the year. Therefore, the higher diversity in sediment samples may be explained by low-latitude coccolithophores species occasionally thriving south of the PF.

However, although the settling of biogenic material is directly related to surface production and reflects the seasonality of that production , the sinking of the particles can strongly be influenced by their drifting due to strong surface and deep currents in the DP. It should be noted that at least the largest part of the coccolith material sinks to the sea floor, incorporated into faecal pellets or in macro-aggregates (marine snow, e.g. ). Factors such as dilution and resuspension processes or drifting of the coccolith material due to strong surface and deep currents may have further influenced the surface sediment assemblage, as has been observed for other microplankton groups which are deposited in the seafloor (e.g. ). showed that bias in microfossil assemblages in surface sediments occur in most oceanic regions and are dependent on current strength and direction, sinking speed and sample depth. The strong ACC flow and frequent eddy formation in the area of the DP are likely to influence the sinking pathways of coccoliths. Thus, temperate taxa observed might not be in situ but could have been transported by currents, mostly the CHC, Peru–Chile Countercurrent and the Peru-Chile Undercurrent, which flow southward along the Chilean margin and which carry relatively warm water masses towards and into the SO and the DP through eddy circulation. These taxa may also originate around Patagonia, the southernmost region of South America, or the Patagonian Shelf in the southwesternmost South Atlantic. Patagonia is at least one of the most important sources for terrigenous fine-grained sediments, which are predominantly transported by bottom currents into the deep DP . The same transport mechanism can also be assumed for similar-sized coccoliths from the southern Patagonian Shelf, from where species such as G. muellerae, C. leptoporus and Gephyrocapsa small (G. ericsonii) were described in surface sediments . Thus, transportation via surface ocean and deep-ocean currents is another factor possibly influencing the surface sediment assemblages south of the PF.

One of the potential factors that could influence coccolith assemblages in surface sediment samples is the depth at which coccoliths are settled. In order to explore the geographical effect of this variable on our dataset, we performed a hierarchical cluster analysis considering the coccolith relative abundances (Fig. ). The cluster analysis identifies two groups of samples which belong to different oceanic regions and sample depths. Samples in cluster A are mostly found in the SAZ zone, with sample depths above 3.1 km, while samples in cluster B are mostly located south in the AZ, at depths below 3.1 km. Hence, the strongest assemblage dissimilarity between clusters is concomitant with sample depth shallower and deeper than 3.1 km. This pattern is exactly matching the distinction of samples by frontal zone and depth above or below the calcite saturation horizon (CSH, where the water becomes undersaturated with respect to calcite, i.e. Ωcalcite=1, ) shown in Fig. , with the exception of samples PS97/052-3 and PS97/074-1. These samples fall out of line, as they are located in the AZ above CSH (2.8 km), but appear grouped with samples in cluster B (located below the CSH). Thus, the clustering of the samples located along the DP transect indicates that not only oceanographic variables but also sample depth, which controls the calcium carbonate preservation, influence the composition of the coccolith assemblage, particularly south of the PF.

Based on the clustering findings and considering that potential factors other than only surface ocean productivity seem to affect the coccolith assemblages, we decided to further explore and test the potential influence of the sediment sample depth compared to other environmental variables driving the coccolithophore species composition, with a redundancy analysis (RDA; see Fig. ). The scores of the surface sediment samples located along RDA1 (which explains almost 68 % of the variance) are separated in two groups which correspond to the two clusters, A and B, identified with the hierarchical analysis. The assemblages located in a more open-ocean setting in the SAZ are comparable to those south of the PF (Fig. ). In both regions (southern SAZ and AZ) we found consistent differences in the assemblages in samples above and below 3.1 km (Fig. ). At depths deeper than 3.1 km, assemblages are characterized by having lower abundances of the relatively fragile E. huxleyi and higher abundances of more robust species such as G. muellerae, G. oceanica and C. leptoporus. This feature is even more striking at the deepest sample, PS97/094-1, located at 4 km water depth in the SAZ, in which the proportion of E. huxleyi compared to C. leptoporus and gephyrocapsids is even smaller (Fig. ). Furthermore, the lack of other species smaller and more delicate than Syracosphaera spp. (e.g. Ophiaster spp., Calciopappus caudatus, Papposphaera sp., Pappomonas sp., Wigwamma antarctica, etc. that were found in the plankton; ) can be attributed to selective dissolution through mechanical destruction when grazed by zooplankton or to dissolution when sinking through the water column .

Based on these observations as well as on the clustering and RDA results, we suggest that carbonate preservation influences the species composition at greater depth north and south of the PF. This is supported by the Calcidiscus leptoporus–Emiliania huxleyi Dissolution Index (CEX), an indicator for dissolution processes in coccolith assemblages . Typically, CEX values below 0.6 indicate selective dissolution of E. huxleyi compared to the more robust C. leptoporus . However, in our case, CEX < 0.75 (below around 3.1 km; see Fig. ) would already suggest preservational issues, coincident with the CSH threshold, which occurs in the DP at a water depth of around 3 to 3.25 km . Below this level, we argue for a selective species-specific dissolution of the delicate E. huxleyi in favour of G. muellerae dependent on the sample depth.

In the surface sediment samples from the DP, the distribution of E. huxleyi reaches unexpectedly high latitudes south of the PF. Except for morphotypes A overcalcified and R, the rest of E. huxleyi morphotypes are present in all bioregional zones. Type A overcalcified is absent in most samples from the SZ + CZ, and type R is absent from the AZ and SZ + CZ. While E. huxleyi morphotypes belonging to morphogroup A are much more abundant in the SAZ, morphotypes within morphogroup B dominate in each of the biogeographic zones. This pattern of dominance is relatively similar to studies of extant communities from the SO, in which E. huxleyi morphotype A is typically restricted to relatively warm waters, north of the PF (usually north of the SAF), and morphotype B/C dominates in cooler waters, in some cases even south of the PF , or it is even the only morphotype present .

In a plankton study along the Patagonian Shelf, suggested that the morphotypes A and B/C are different ecotypes because of their respective dominance in warmer, nutrient-poor water with a higher calcite saturation state versus cooler, nutrient-rich water with a lower calcite saturation state, respectively. Therefore, shifts in E. huxleyi morphotype assemblage composition seem to correlate to some extent with changes in carbonate system parameters. On the other hand, and occasionally observed in the eastern South Pacific morphotypes A and R overcalcified dominating in specific stations offshore Chile and Peru, at surface waters with a relatively low calcite saturation state. Even moderately calcified E. huxleyi type A has been observed in the southern Patagonian fjords and channels by , who highlight that the ecological response of this taxon is plastic and that it has high potential for adaptation to different niches.

Within morphotype B/C, we observed decreasing numbers of large coccoliths >4 µm towards the south, with nearly all occurrences of larger coccoliths in the SAZ (Fig. ). A comparable observation was made by in the open South Pacific Ocean, where the authors differentiated morphotypes B, B/C and C. The largest type B was found exclusively north of the PF, while the smaller types B/C and C dominated the overall E. huxleyi assemblage. In E. huxleyi morphotype B/C strains from the Arctic Ocean, observed decreasing coccolith sizes with rising temperatures and decreasing salinities, likely changes to happen in the SO due to global warming, which would further enhance this signal.

Morphotype O was identified in all the surface sediment samples, and we noted slightly higher relative abundances of this morphotype in the SAZ. Coccoliths with both a plated and open central area were identified in plankton samples in the DP . This morphotype is, however, also often ignored even in more recent studies or integrated into the B/C morphogroup (e.g. ). Up to now it has only been occasionally differentiated in plankton studies and not at all into coccolith assemblages in sediments . Such a differentiation is possible, although the presence of an open central area might also be an artefact of dissolution, especially in sedimented coccoliths. It is therefore possible that coccoliths classified as morphotype O would have been originally morphotype B/C. However, such a differentiation is of importance, since different E. huxleyi morphotypes exhibited different sensitivities in regards to seawater carbonate chemistry in cultures . Therefore, documenting the diversity of E. huxleyi morphotypes, in the SO in general and in the DP in particular, and showing their biogeographical distribution in relation to changing environmental conditions is of critical importance to assess their response to projected environmental change in the SO.

The coccolith carbonate mass estimations obtained are in close agreement, with an overall mean mass of 2.12 pg using Eq. () and of 2.29 pg using Eq. () (see Table ). The highest weights were estimated for coccoliths in the SAZ (up to 8.42 pg with Eq.  or 5.7 pg with Eq. ; Fig. ). The lowest weights are around and below 1 pg across frontal zones independently of the estimation method used. In general, there is a high mass range in the SAZ compared to the other frontal zones, most likely due to the actual E. huxleyi morphotype diversity recorded here. Total mass estimations vary considerably between different studies in the DP, although there is always a common decreasing trend towards the south. Even if absolute values differ due to the various techniques used, the high coccolith mass calculated and the relatively large mass range in the SAZ observed in our study are in agreement with mass estimates based on plankton studies from and (see Fig.  and Sect. ). obtained generally higher mass values using a technique based on circularly polarized light microscopy and a piece of image processing software, C-Calcita , while masses calculated based on length data from were lower (mostly under 2 pg) using the approach from .

In comparison to mass estimations based on light microscope images of surface sediment samples in other sectors of the SO, our results fall into a common range: there were slightly higher masses of E. huxleyi with an average of 3.6 ± 0.8 pg, but comparable ranges of 1.73 to 4.85 pg were reported from the Atlantic sector in the area south of South Africa by using a mass estimation method from . In the Indian sector south of Tasmania, showed comparable mass estimates and a similar north-to-south trend based on C-Calcita measurements with a total average of 2.65 ± 1.2 pg. Average E. huxleyi masses of approximately 2.5 to 2.8 pg were determined for the SAZ, which is slightly higher than the average of our findings. However, although the mass estimations were based on coccoliths whose morphotypes were not identified due to methodological reasons, linked the decreasing mass trend to morphotype distribution across the ACC. With our mass estimates on identified morphotype coccoliths, we can support this interpretation and thus generally attribute changes in the masses across the fronts to a change in the morphotype assemblages.

Combining the decreasing coccolith sizes and carbonate masses observed in this study with predictions of SO warming and freshening due to sea-ice melting and responses of cultured strains to those changes (e.g. ), it seems likely that produced coccoliths south of the PF will get smaller and lighter in the future.

Our study suggests that well preserved (sub-)fossil coccolith assemblages in the DP mirror the overlying extant coccolithophore communities which respond to environmental (i.e. physical, chemical and biological) variables, specifically north of the PF. Therefore these assemblages constitute a robust and valuable dataset for qualitative and quantitative calibrations and subsequent reconstructions of surface ocean conditions (i.e. transfer functions and oceanographic/climate models).

The phytoplankton dynamics in the SO is complex, and coccolithophores and diatoms (among other planktonic groups) coexist in this region (e.g. ). The coccolithophore diversity and abundance in surface sediment samples located south of the PF is high compared to plankton studies in the SO (e.g. ), DP and surface sediment samples in the Pacific sector of the SO . This demonstrates that calcareous phytoplankton can successfully thrive and be incorporated into the geological record at this latitude and that conditions for their preservation south of the PF are rather variable depending on the region. Although the similarity between living (plankton) coccolithophore communities and fossil surface sediment assemblages is high at the Chilean margin, there are some differences south of the PF. In particular, we observe a southward enrichment in dissolution-resistant species, such as C. leptoporus, at the expense of a decrease of more fragile species, such as E. huxleyi. A combination of statistical techniques (hierarchical clustering and RDA) and dissolution-sensitive indices (CEX) indicates that this shift in the composition is due to the preferential dissolution of calcium carbonate occurring mainly below 3.1 km, which coincides with the depth of the CSH in this region . In addition, a proportion of the fossil coccoliths (i.e. temperate taxa) found in sediments in the southernmost samples are most likely not in situ. Coccolithophores and detached coccoliths are subject to transport from lower latitudes over long distances while settling in the water column, as has been shown by other microplankton groups, such as planktonic foraminifera or dinoflagellates (e.g. ). Therefore, the hydrodynamic and post-depositional processes, which altered the original composition of the coccolith assemblages in some of the DP stations, provide insights into the deep-ocean biogeochemistry and hydrography of the study area. These processes distort the original ecological information and limit the potential of coccolith assemblages in surface sediments as surface ocean indicators. These processes need to be taken into account when interpreting downcore coccolith records at high latitudes.

Remote sensing studies over the last decades have detected coccolithophore blooms further north in the North Atlantic linked to anthropogenic-induced climate change (e.g. ). Furthermore, although plankton studies on coccolithophores are mostly limited to the Norwegian–Greenland seas and the Fram Strait south of the central Arctic Ocean (e.g. ), coccoliths have been found in sediments of the central Arctic Ocean and have been used intensively for stratigraphic purposes (e.g. ). Their occurrences in those sediments were interpreted as an indicator of partly ice-free conditions during at least some summers. Furthermore, by expanding poleward and doubling its areal extent in the northernmost Barents Sea, the occurrence of E. huxleyi attests the ongoing “Atlantification” of the Arctic Ocean . The primary driver of the dynamics of this species seem to be, in fact, stronger surface currents, which in turn intrinsically shape the temperature field and frontal structures. In the DP, the occurrence of coccoliths in surface sediments indicate that, despite the lack of observations in the plankton, coccolithophores seem to be continuously present south of the PF or are at least continuously drifted south via eddies during the Holocene. From the available data, however, it is difficult to deduce whether it is an increasing process or an increased shift towards the south. Plankton samples covering the last 5 decades, however, have shown a gradual poleward expansion of E. huxleyi in the SO , which currently seems a more permanent member of the summer phytoplankton community south of the PF. The clear dominance of E. huxleyi morphotype B/C (including B, B/C and C according to ) over morphotype O (with an opened central area and with lamella) in plankton samples but not in surface sediments in the DP indicates that this species is highly sensitive to lower calcite saturation state in depth south of the PF. The high numbers of E. huxleyi type O recorded in the surface sediment samples studied in the AZ and SZ + CZ (Fig. ) suggest that some of these specimens could have been originally B/C whose central area would have been affected by dissolution. Other SO living calcifiers, like the pteropod Limacina helicina antarctica, are already experiencing the effects of ocean warming and acidification. Since pteropod shells consist of aragonite, a more soluble carbonate mineral than calcite, the decrease in their shell growth, extensive shell dissolution and malformations are the first evidence of ocean acidification (e.g. ). This loss in carbonate fossils in the geological record could be a consequence of the increase in anthropogenic acidification, which is particularly significant in the SO, and it is dissolving the most recent calcium carbonate geological record . Distinguishing E. huxleyi morphotype O in future plankton, sediment trap and surface sediment studies will provide new insights into calcium carbonate dissolution processes due to increasing CO2 dissolved in the SO, which may affect coccolithophores communities either living in the water column or incorporated immediately to the fossil record.

The higher-than-expected occurrence of coccoliths in surface sediments south of the PF as well as the aforementioned preservational limitations open the possibility that sediment records in sub-polar ice distal regions could have also borne more coccoliths during older time intervals. Currently, these type of records in the SO are very scarce, and some of them only show coccoliths during deglaciations or interglacials, which is especially evident for instance during Termination IV or Marine Isotope Stage 11 (e.g. ).

It is imperative to assess the contribution of coccolithophores to changes in present and past processes with in situ SO observations and fossil datasets in order to develop high-resolution and well-constrained regional and global climate models. Model simulations projecting future coccolithophore growth and calcification in an acidified ocean have also proposed that coccolithophores will expand with increasing CO2 availability, but they will become more lightly calcified, with even “naked” coccolithophores (i.e. without coccoliths) dominating in polar areas (e.g. ). To test these future projections, further in situ field observations will be needed during the upcoming years in the already changing polar realm.