Introduction
Fragilariopsis kerguelensis (O’Meara) Hustedt is the most abundant diatom species in the surface sediments and in the down-core records of the open Southern Ocean (SO) (Zielinski & Gersonde Reference Zielinski and Gersonde1997, Crosta et al. Reference Crosta, Romero, Armand and Pichon2005a) and is believed to be a major silica carrier to the sea floor (Cortese & Gersonde Reference Cortese and Gersonde2008, Abelmann et al. Reference Abelmann, Gersonde, Knorr, Zhang, Chapligin, Maier, Esper, Friedrichsen, Lohmann, Meyer and Tiedemann2015). Fragilariopsis kerguelensis size variability has recently been proved useful to infer recent (Cortese & Gersonde Reference Cortese and Gersonde2007) and past oceanic conditions at the glacial–interglacial timescales (Cortese et al. Reference Cortese, Gersonde, Maschner and Medley2012, Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013, Nair et al. Reference Nair, Mohan, Manoj and Thamban2015). A study conducted on sediment traps and surface sediments from the sea-ice zone to the subtropical zone of the Atlantic SO suggested a strong correlation of F. kerguelensis size with iron availability (Cortese & Gersonde Reference Cortese and Gersonde2007), consistent with the findings of culture experiments (Assmy et al. Reference Assmy, Henjes, Smetacek and Montresor2006, Timmermans & van der Wagt Reference Timmermans and van der Wagt2010) and in situ iron fertilization experiments (Assmy et al. Reference Assmy, Henjes, Klaas and Smetacek2007). Similarly, the down-core records from the Atlantic and Indian Antarctic Polar Front (APF), and Atlantic and Indian Sub-Antarctic Front (SAF) show larger F. kerguelensis mean valve area (FkergArea) throughout the last glacial period during higher aeolian dust flux to the SO. FkergArea generally declined during the deglacial and the Holocene periods along with a reduction in the dust flux (Cortese & Gersonde Reference Cortese and Gersonde2007, Cortese et al. Reference Cortese, Gersonde, Maschner and Medley2012, Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013). This pattern was reproducible until termination V,~430 cal kyrs bp (Cortese et al. Reference Cortese, Gersonde, Maschner and Medley2012). However, other sources of iron, such as iceberg surges, may have mediated larger FkergArea in the Atlantic SO during the last deglaciation, which explains the observed differences in temporal variations of the species mean size between the different SO basins (Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013).
In addition to iron availability, sea surface temperature (SST) has been shown to play an important role on F. kerguelensis size, probably via the regulation of silica uptake, whereby diatoms seemed to grow larger initial cells at their optimal ecological range (Crosta Reference Crosta2009, Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013). Indeed, SST sets the upper bound on SO diatom growth (Smith Reference Smith1990). Furthermore, although higher SST increase the growth rate of Pseudo-nitzschia multiseries (Hasle) Hasle, a sub-Antarctic diatom related to F. kerguelensis, it also decreases cell size in this species (Boyd et al. Reference Boyd, Dillingham, McGraw, Armstrong, Cornwall, Feng, Hurd, Gault-Ringold, Roleda, Timmins-Schiffman and Nunn2016 and references therein).
Therefore, diatom size variation in the SO is controlled by both physical and biogeochemical components, and their non-linear interactions. Here, we present a 42 cal kyrs record of FkergArea from the Indian subtropical region of the SO, where the biogeochemical environment did not change much over this period (Crosta et al. Reference Crosta, Shemesh, Etourneau, Yam, Billy and Pichon2005b, Beucher et al. Reference Beucher, Brzezinski and Crosta2007), to better understand the impact of SST on F. kerguelensis size variation and associated silica export.
Materials and methods
The biometric measurements were performed on F. kerguelensis valves from sediment core MD97-2101 (43°30'S, 79°50'E, at the modern southern Subtropical Front (SSTF)) from the Indian sector of the SO (Fig. 1), retrieved during the Images III–IPHIS cruise on board RV Marion Dufresne II. The apical and transapical lengths were measured for 100 complete F. kerguelensis valves from each sample, for which the standard deviation was ~0.2 μm. Details on the measurements and calculations used in this study can be found in Shukla et al. (Reference Shukla, Crosta, Cortese and Nayak2013). To provide a direct comparison between all records, we recalculated published FkergArea records (Cortese & Gersonde Reference Cortese and Gersonde2007, Cortese et al. Reference Cortese, Gersonde, Maschner and Medley2012) with the formula presented in Shukla et al. (Reference Shukla, Crosta, Cortese and Nayak2013). The age model of core MD97-2101, along with details on slide preparation, diatom counts and absolute abundance calculations are detailed in Crosta et al. (Reference Crosta, Shemesh, Etourneau, Yam, Billy and Pichon2005b).
Map showing the locations of the sediment and ice cores used in the study. Red star represents the MD97-2101 core from the Indian southern Subtropical Front for F. kerguelensis mean valve area (FkergArea) (present analysis) and the black squares represent previously published FkergArea records. White circles represent environmental records: core MD94-102 for the opal flux (Dézileau et al. Reference Dézileau, Reyss and Lemoine2003), and the Vostok and EPICA Dome C (EDC) ice cores for dust fluxes. The colour gradient represents the yearly average sea surface temperature (World Ocean Atlas 2009, https://www.nodc.noaa.gov, Locarnini et al. Reference Locarnini, Mishonov, Antonov, Boyer, Garcia, Baranova, Zweng and Johnson2010). AFP=Antarctic Polar Front, SAF=sub-Antarctic Front, STF=Subtropical Front.
To assess the role of F. kerguelensis in biogenic silica export to the sea floor (FkergBSi) in the Indian SSTF over the last 42 cal kyrs bp, and to provide a direct comparison with opal flux, we multiplied FkergArea by F. kerguelensis absolute abundance (FkergAbun), as has been performed for Antarctic and sub-Antarctic records (Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013).
The down-core results and interpretations are described for three climatic periods namely the Last Glacial, the deglaciation and the Holocene which have been defined after Crosta et al. (Reference Crosta, Shemesh, Etourneau, Yam, Billy and Pichon2005b). The statistical analyses of FkergArea data with other data were performed using linear regression analysis and one-way ANOVA tests as detailed in Shukla et al. (Reference Shukla, Crosta, Cortese and Nayak2013). The down-core data for FkergArea, FkergAbun and FkergBSi are provided in Supplemental File A found at http://dx.doi.org/10.1017/S095410201600050X.
Results
The mean valve area of F. kerguelensis was largest during the Last Glacial period (~324 µm2, n=24), decreasing during the deglacial period (~296 µm2, n=11) and into the Holocene period (~275 µm2, n=10) (Fig. 2a green line). The results of the one-way ANOVA tests for FkergArea across the climatic periods are summarized in Supplemental File B found at http://dx.doi.org/10.1017/S095410201600050X. The FkergArea was statistically different between the Holocene and glacial periods, the Holocene and deglacial periods, and Last Glacial and deglacial periods (P>0.05).
Down-core records of a. F. kerguelensis mean valve area (green line) and absolute abundances (brown shading) in core MD97-2101 vs calendar age compared with previous biometric studies from the Southern Ocean: b. core PS-2498 and c. core PS-2499 from the Atlantic sub-Antarctic Front (SAF; Cortese et al. Reference Cortese, Gersonde, Maschner and Medley2012), d. core MD88-769 from the Indian SAF (Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013), e. core TNO57-13-PC4 from the Atlantic Antarctic Polar Front (APF; Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013), f. core PS1654-2 from the Atlantic APF (Cortese & Gersonde Reference Cortese and Gersonde2007), and g. core SO136-111 from the Indian APF (Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013). h. Dust flux record from EPICA Dome C (EDC) ice core (Lambert et al. Reference Lambert, Delmonte, Petit, Bigler, Kaufmann, Hutterli, Stocker, Ruth, Steffensen and Maggi2008) (pink line) and dust concentration record from Vostok ice core (Petit et al. Reference Petit, Jouzel, Raynaud, Barkov, Barnola, Basile, Bender, Chappellaz, Davis, Delaygue, Delmotte, Kotlyakov, Legrand, Lipenkov, Lorius, Pepin, Ritz, Saltzman and Stievenard1999) (black line), plotted on independent time scales; shown for direct comparison with F. kerguelensis mean valve area. The Last Glacial, deglacial and Holocene periods are highlighted with cyan, dark grey and light grey, respectively.
The absolute abundances of F. kerguelensis was highest during the glacial period (2.58×106 valves g-1), especially during 41–35 calkabp (2.61×106 valves g-1) and 20–17 calkabp (3.34×106 valves g-1), and the Holocene (2.78×106 valves g-1). It was lowest during the deglacial period (1.99×106 valves g-1) (Fig. 2a brown shading). The results of one-way ANOVA tests indicate that FkergAbun was statistically different between the Holocene and deglacial periods, and Last Glacial and deglacial periods (P>0.05), but statistically similar for the Holocene and Last Glacial periods (P<0.05) (for a summary of the one-way ANOVA tests see Supplemental File C found at http://dx.doi.org/10.1017/S095410201600050X).
Finally, FkergArea data were compared with FkergAbun to infer the relationship between valve size and productivity. A positive linear relationship was found for the Last Glacial (R=0.31, n=24, P=0.02) and deglacial periods (R=0.056, n=11, P>0.05). While an inverse relationship was found for the Holocene period (R=-0.58, n=10, P>0.05) (for a detailed data set see Supplementary File D found at http://dx.doi.org/10.1017/S095410201600050X).
The reconstructed biogenic silica associated to F. kerguelensis burial was high during the Last Glacial period, especially during 41–35 calkabp and 20–17 calkabp, and during the late Holocene, and low during the deglacial and early Holocene periods (Fig. 3c).
Discussion
Previous studies have suggested that there are three main factors that control F. kerguelensis size variability, which are superimposed on productivity-dependent size reduction during asexual cell division. First, alleviation of iron deficiency in the open SO was shown to generate larger valves (Cortese & Gersonde Reference Cortese and Gersonde2007, Cortese et al. Reference Cortese, Gersonde, Maschner and Medley2012, Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013, Nair et al. Reference Nair, Mohan, Manoj and Thamban2015). Second, lengthening of the growing season induced by higher SST and early sea ice break-up may promote the production of longer initial cells in the seasonal sea-ice zone (Crosta Reference Crosta2009). Third, increasing SST may have a subordinate role to dust-bearing iron flux drop on F. kerguelensis size reduction over glacial terminations (Cortese et al. Reference Cortese, Gersonde, Maschner and Medley2012). These findings demonstrate that the controls on F. kerguelensis size variability may change through space and time depending on the limiting factors and their interactions. Our new data on FkergArea from the Indian SSTF of the SO provide a northernmost end-member to published records and allow us to further explore the previous hypotheses.
Down-core records of F. kerguelensis size in the Southern Ocean
Our FkergArea record from the Indian SSTF (Fig. 2a green line) follow the general decreasing trend observed in previously published records (Fig. 2b–g). However, during the Last Glacial, FkergArea at the Indian SSTF site was larger than in the sub-Antarctic and Antarctic zones by 4–26 μm2 and 18–45 μm2, respectively (Fig. 4a). Likewise, during the deglacial period, FkergArea was larger at the Indian SSTF site than in the sub-Antarctic and the Antarctic zones by 25–64 μm2 and 10–44 μm2, respectively (Fig. 4b). During the Holocene, FkergArea at the Indian SSTF site was larger than in the sub-Antarctic Indian and Atlantic zones by ~9 µm2 and 85–103 µm2, respectively (Fig. 3c). The picture is more complicated for the Antarctic zone, where Indian SSTF FkergArea was larger than in the Atlantic sector core (by 24–36 μm2; Fig. 2f) but smaller relative to the Indian APF sector core (by ~32 μm2; Fig. 2g) where the largest FkergArea were found (Fig. 4c; Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013).
Down-core records of a. F. kerguelensis mean valve area in core MD97-2101 vs calendar age (present study) compared with b. February sea surface temperature records of the same core (Crosta et al. Reference Crosta, Shemesh, Etourneau, Yam, Billy and Pichon2005b). The reconstructed F. kerguelensis biogenic silica burial in c. is compared with the Opal flux record from the twin core MD94-102 in d. (Dézileau et al. Reference Dézileau, Reyss and Lemoine2003). The Last Glacial, deglacial and Holocene periods are highlighted with cyan, dark grey and light grey, respectively.
Box and whisker plots showing variations in F. kerguelensis valve area for the a. Last Glacial, b. deglacial and c. Holocene periods for different cores in the Southern Ocean: core MD97-2101 (present study), cores PS-2498, PS-2499 and PS1654-2 (Cortese et al. Reference Cortese, Gersonde, Maschner and Medley2012), and cores MD88-769, SO136-111 and TNO57-13-PC4 (Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013).
Iron stress usually decreases diatom growth rate, silica maximal uptake rate and cell size (Leynaert et al. Reference Leynaert, Bucciarelli, Claquin, Dugdale, Martin-Jézéquel, Pondaven and Ragueneau2004, Timmermans et al. Reference Timmermans, van der Wagt and de Baar2004, Timmermans & van der Wagt Reference Timmermans and van der Wagt2010). Conversely, it increases diatom silicification as a result of lengthening of the cell wall synthesis phase (Martin-Jézéquel et al. Reference Martin-Jézéquel, Hildebrand and Brzezinski2000). Therefore, the overall larger F. kerguelensis valves encountered in the warm, low-nutrient SSTF environment over the past 42 cal kyr bp probably resulted from low growth rates with fewer cell divisions, indicated by the very low absolute abundance in core MD97-2101 (Fig. 2a brown shading), which limited size reduction of the species communities. Indeed, F. kerguelensis averaged ~2.5×106 valves g-1 in core MD97-2101 and between 20×106 and 180×106 valves g-1 in the other cores. The large FkergArea recorded during the Last Glacial period (Fig. 2b–g) may be a result of increased iron availability (Fig. 2h) (Cortese & Gersonde Reference Cortese and Gersonde2007, Cortese et al. Reference Cortese, Gersonde, Maschner and Medley2012, Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013). However, both data and models suggest that iron stocks in SSTF surface waters via dust, upwelling and sedimentary sources have been permanently low over the last 40 cal kyrs bp (Andersen et al. Reference Andersen, Armengaud and Genthon1998, Lefèvre & Watson Reference Lefèvre and Watson1999, Mahowald et al. Reference Mahowald, Baker, Bergametti, Brooks, Duce, Jickells, Kubilay, Prospero and Tegen2005, Lamy et al. Reference Lamy, Gersonde, Winckler, Esper, Jaeschke, Kuhn, Ullermann, Martinez-Garcia, Lambert and Kilian2014). Indeed, equivalence of isotopic composition of diatom-bound intrinsic matter δ15Ndiat and silicon stable isotopes δ30Si data from the MD97-2101 core for the Last Glacial period suggested little impact of the iron supply on silicic acid and nitrate dynamics in the subtropical region of the SO (Beucher et al. Reference Beucher, Brzezinski and Crosta2007) (Fig. 5a & b). Therefore, we propose that other factors are responsible for the temporal variations in FkergArea recorded in core MD97-2101.
Down-core records of a. δ15Ndiat data from Crosta et al. (Reference Crosta, Shemesh, Etourneau, Yam, Billy and Pichon2005b) and b. δ30Si data from Beucher et al. (Reference Beucher, Brzezinski and Crosta2007) for core MD97-2101.
Fragilariopsis kerguelensis mean valve area and sea surface temperature
The SST record at the Indian SSTF site presents a glacial–interglacial pattern (Crosta et al. Reference Crosta, Shemesh, Etourneau, Yam, Billy and Pichon2005b) consistent with the global deglaciation. The SSTF FkergArea record appears inversely correlated to the SST record (Fig. 3a & b) materialized by a correlation coefficient of R=-0.63, n=31, P<0.01 (Fig. 6). Larger FkergArea occurred when SST ranged between 6–8°C during the Last Glacial period, while FkergArea started to decline during the deglacial period and continued into the Holocene period when SST increased by >8°C (Fig. 3a & b). FkergArea was smallest during the mid-Holocene period when SST was highest.
Linear correlation between F. kerguelensis mean valve area and sea surface temperature in core MD97-2101.
Maximum occurrences of F. kerguelensis in the phytoplankton occur between the winter sea ice limit and the SAF (Hasle Reference Hasle1969, Fenner et al. Reference Fenner, Schrader and Wienigk1976, Froneman et al. Reference Froneman, Perissinotto, McQuaid and Laubscher1995) and are similarly registered in underlying surface sediments (DeFelice & Wise Reference DeFelice and Wise1981, Zielinski & Gersonde Reference Zielinski and Gersonde1997). Preferential ecological conditions of F. kerguelensis generally fall within the 1–8°C SST range (Crosta et al. Reference Crosta2005a) where largest F. kerguelensis have been also observed (Cortese & Gersonde Reference Cortese and Gersonde2007). Conversely, smaller F. kerguelensis have been measured in surface sediments of the sub-Antarctic zone and the sea-ice zone where F. kerguelensis are less abundant. We propose that the occurrence of larger and more abundant F. kerguelensis at the Indian SSTF during the Last Glacial period (Fig. 3a) is a response to more favourable SST (6–8°C) (Fig. 3b). Conversely, a reduction in FkergArea and FkergAbun during the deglacial period may have resulted from increasing SST>8–9°C. The positive relationship between FkergArea and FkergAbun for 42–10 calkyrsbp further supports that diatoms thriving at the adequate SST range grow larger initial cells (Fig. 7), supporting more numerous divisions, leading to larger communities. However, we found that the smallest F. kerguelensis co-occurred with highest abundances in the MD97-2101 core during the Holocene, at odds with the modern model (Cortese & Gersonde Reference Cortese and Gersonde2007). Although we do not have a definitive explanation for such a relationship change during the Holocene, we note that a recent study conducted on P. multiseries shows a twofold increase in growth rate, leading to decreasing cell sizes, when SST increases by 3°C (Boyd et al. Reference Boyd, Dillingham, McGraw, Armstrong, Cornwall, Feng, Hurd, Gault-Ringold, Roleda, Timmins-Schiffman and Nunn2016). Furthermore, small cells are believed to have a competitive advantage, through higher surface-to-volume ratios, under limiting nutrient conditions (Kiørboe Reference Kiørboe1993). Small cells also present higher growth rates, though slower rates of size reduction per generation, than larger cells (Amato et al. Reference Amato, Orsini, D’Alelio and Montresor2005, D’Alelio et al. Reference D’Alelio, Amato, Luedeking and Montresor2009). We postulate that the co-occurrence of small but numerous F. kerguelensis during the Holocene may have resulted from the production of smaller initial cells (Fig. 7) for which rapid divisions could take place. The production of smaller F. kerguelensis may represent a specific adaptation to survive in the warm, low-nutrient, unfavourable conditions prevailing during this period (Fig. 2a).
Fragilariopsis kerguelensis maximum mean valve length variations for the Last Glacial, deglacial and the Holocene periods are shown for the different zones of the Southern Ocean: Antarctic zone (cores TNO57-13-PC4 and SO136-111; Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013), sub-Antarctic zone (core MD88-769; Shukla et al. Reference Shukla, Crosta, Cortese and Nayak2013) and subtropical zone (core MD97-2101; present study).
Our observations suggest that SST changes dictated both F. kerguelensis productivity and size variations in the Indian subtropical zone over the last 42 cal kyrs probably via the mediation of silica uptake and the size of the initial cells. Indeed, studies on silicon metabolism suggest that efficient silicic acid uptake by diatoms can only be possible within the preferential environmental conditions of the species (Blank et al. Reference Blank, Robinson and Sullivan1986, Hildebrand Reference Hildebrand2000, Martin-Jézéquel et al. Reference Martin-Jézéquel, Hildebrand and Brzezinski2000). Our observations also extend, at the species level, previous findings demonstrating an inverse relationship between SST and phytoplankton community size (Hilligsøe et al. Reference Hilligsøe, Richardson, Bendtsen, Sørensen, Nielsen and Lyngsgaard2011).
Fragilariopsis kerguelensis represents the most abundant diatom in core MD97-2101, with relative abundances between 20–40% (data not shown). The good concordance between the estimated FkergBSi (Fig. 3c) and the opal flux (Fig. 3d) recorded in a twin core (Dézileau et al. Reference Dézileau, Reyss and Lemoine2003) suggests that this species is one of the main opal carriers to the sea floor in the subtropical Indian sector of the SO. However, despite a lower contribution to the diatom assemblages (10–20%), Thalassiosira lentiginosa (Janisch) Fryxell, a large heavily silicified centric diatom, was shown to export more biogenic silica at the Indian SSTF core site (MD97-2101) over the last 42 calkyrs (Shukla et al. Reference Shukla, Crespin and Crosta2016). Similarly, T. lentiginosa was shown to be the main biogenic silica burial species to the sea floor in the open ocean sub-Antarctic and Antarctic zones (Shukla et al. Reference Shukla, Crespin and Crosta2016), although this species was never previously considered as such (Grigorov et al. Reference Grigorov, Pearce and Kemp2002, Abelmann et al. Reference Abelmann, Gersonde, Cortese, Kuhn and Smetacek2006).
Conclusions
We investigated F. kerguelensis size variation over a 42 cal kyr period in core MD97-2101 from the Indian SSTF of the SO, where micro- and macronutrients are believed to have been consistently low. Our data provide a northernmost end-member to published records of diatom size variation and allow us to further explore the previous hypotheses. FkergArea was inversely correlated with SST over the last 42 000 years, with larger FkergArea during the Last Glacial period when SST were ~6–8°C declining during the last deglaciation into the Holocene period when SST increased by >9°C. In contrast to other SO oceanographic realms where iron and nutrient concentrations strongly varied through time, our observations suggest that SST was the main factor controlling both the species’ productivity and mean size in the subtropical zone, probably via the modulation of growth rates and silicic acid uptake. Further investigations on time series and sediment cores from the Indian and the other SO basins are required to refine the hypothesis proposed here.
Acknowledgements
The research presented here was funded by the Scientific Committee on Antarctic Research (SCAR Fellowship 2010–11 awarded to SKS) and carried out at EPOC, Université de Bordeaux. SKS is thankful to Prof Sunil Bajpai for infrastructural facilities and encouragement to publish this work (BSIP/RDCC/Publication No. 24/2016-17). XC was funded by CNRS-INSU. Two anonymous reviewers are thanked for providing helpful suggestions which improved the final manuscript. The Editor, Prof Walker Smith, is thanked for a timely review of the paper and for making it press worthy.
Author contributions
SKS and XC designed the project work. SKS generated the down-core biometry data. Both SKS and XC discussed the data. SKS wrote the manuscript and XC substantially contributed to manuscript writing.
Supplementary Material
Supplemental files will be found at http://dx.doi.org/10.1017/S095410201600050X.
