Large diatom bloom off the Antarctic Peninsula during cool conditions associated with the 2015/2016 El Niño

Due to contrasts in oceanographic properties along the NAP²⁴, the sampling grid was split in two subregions: north and south (Fig. 1; see “Methods”). The north and south subregions showed from the satellite data a much higher spring/summer (November–February) mean chlorophyll-a (Chl-a) in 2015/2016 than the decadal average time series (2010–2019; Table 1). In agreement with the El Niño effects^10,16, the sea surface temperature (SST) and the air temperature showed substantially lower mean values during the spring/summer of 2015/2016 along the subregions (Table 1). However, there was an evident spatial/temporal variability in sea ice concentration/duration between the subregions, with a northward (southward) lower (higher) mean value during 2015/2016 in relation to the decadal average (Table 1). Along the south subregion during the spring/summer of 2015/2016, the increased Chl-a during January followed the decline in the sea ice concentration over the spring and early summer, concurrent with increased SST, which was markedly colder throughout the seasonal phytoplankton succession (Fig. 2a). These results to the south subregion are consistent with previous studies along the WAP, in which years characterized by longer sea ice cover in winter have led to higher phytoplankton biomass in the following summer associated with a more stable water column^11,16,26. To the north subregion, however, although there was a similar pattern between Chl-a and SST, the increased Chl-a during January was not related with the sea ice retreat (Fig. 2b). Moreover, there was a clear difference between the Chl-a peaks (the highest Chl-a value reached) along the subregions from the satellite data. The Chl-a peak in the south subregion occurred in early January (10 January 2016, reaching 1.73 mg m^–3), whereas in the north subregion the Chl-a peak was observed in late January (29 January 2016, reaching 2.23 mg m^–3).

It has been estimated that drifters entrained in the Gerlache Strait Current and the Bransfield Strait Current exit the Bransfield Strait in 10–20 days¹⁷, which is consistent with the interval of 19 days between both Chl-a peaks when considering the extreme distance between the subregions (see Fig. 1). These authors also estimated that drifters deployed in the Gerlache Strait Current were quickly advected out of the Gerlache Strait in less than 1 week (i.e., low residence time)¹⁷, which supports the similar diatom species assemblages identified in our microscopic analysis between stations of the Gerlache Strait and southwestern Bransfield Strait²⁴. Therefore, it is plausible that phytoplankton growth in the north of the Gerlache Strait may be laterally advected northward into the Bransfield Strait, explaining the observed concomitant increase of satellite Chl-a data in both subregions from spring, associated with sea ice retreat southward (Fig. 2). In addition, as phytoplankton biomass tends to accumulate northward^17,27,28, the advection processes could also explain the temporal and intensity differences of the Chl-a peaks along the subregions (see Fig. 2). This suggests that there was a link between the sea ice dynamics, phytoplankton biomass (Chl-a) and advection processes along the NAP during the spring/summer of 2015/2016, in which the sea ice melting first triggered an increase in phytoplankton biomass through water column stratification along the south subregion, and the advection processes led to a subsequent increase northward.

The satellite Chl-a data require extensive validation with in situ data, especially in polar regions, where cloud cover is ubiquitous and performance is typically poor, due to not properly accurate Chl-a algorithms^12,29. For that, although the mean Chl-a in 2015/2016 from the satellite data was approximately twice as large as the decadal average, there was a severe discrepancy in the mean Chl-a values observed between the in situ and remote sensing data (see Table 1 and Supplementary Table 1). This highlights the importance of the in situ dataset reported here, especially evident during February 2016, when the signal of an intense diatom bloom (> 40 mg m^–3 Chl-a)²⁴ was not captured in the satellite data (Supplementary Fig. 1), supporting that phytoplankton biomass accumulation during this summer was much higher than recorded by remote sensing observations (see Table 1). In general, the in situ Chl-a achieved its maximum (40 mg m^–3) and higher mean value (17.4 mg m^–3) during February comparing to November and January (Supplementary Table 1).

Phytoplankton community structure during the spring/summer of 2015/2016 was assessed through Chemical taxonomy (CHEMTAX) software, using accessory pigments versus in situ Chl-a concentrations measured via high-performance liquid chromatography (HPLC; see “Methods”). The main phytoplankton group over the season were diatoms, followed by haptophytes (Phaeocystis antarctica), cryptophytes, and dinoflagellates, according to the succession stage (Fig. 3a). Diatoms dominated the phytoplankton community composition in relation to the other groups along the whole in situ sampling period, although their relative biomass (to the total in situ Chl-a) was lower in some stations compared to others in different moments during spring/summer (Fig. 3a). To assess the degree to which the water column structure was a primary driver for development and intensity of diatom growth^3,24, the mixed layer depth (MLD) and water column stability were calculated as a function of seawater potential density (see “Methods”). There was an inverse polynomial relationship between MLD and mean upper ocean stability (averaged over 5−150 m depth; hereafter referred to as upper ocean stability) (Fig. 3b). The significant positive exponential relationship between the upper ocean stability and diatom absolute concentrations (in situ Chl-a) demonstrates that stability, associated with MLD, was an important driver of diatom dynamics (Fig. 3b). This elucidates the increase in biological production during summer months of 2016, when upper ocean physical structures (MLD and stability) were sufficiently shallow and stable to produce the high phytoplankton biomass (in situ Chl-a) registered here. However, as MLD and stability showed similar values between summer months (Supplementary Table 1), only the upper ocean physical structures cannot be accounted for the high differences of in situ Chl-a values observed between diatom blooms in January (maximum of 12 mg m^–3) and February (maximum of 40 mg m^–3). Likewise, also not explaining these differences of in situ Chl-a values between summer months, macronutrients were highly abundant throughout the seasonal phytoplankton succession (Supplementary Table 1). Furthermore, although no measurements of dissolved iron, which can be considered as a limiting factor to primary productivity³⁰, were carried out here, the Antarctic Peninsula continental shelves have been depicted as a substantial source of this micronutrient to the upper ocean, not limiting phytoplankton growth even during intense blooms^31,32.

The stability was also seen to influence the dominant phytoplankton size class. While micro-phytoplankton dominated over the whole range of the upper ocean stability, smaller size classes showed a more restricted distribution (Fig. 3b). These differences on biomass and dominant phytoplankton size class could be due to distinct moments of the succession stages linked to interspecific differences. In general, the nano- and pico-phytoplankton were more abundant when: (i) stability was lower in the late spring/early summer (mainly small diatoms and small flagellates) (Fig. 3b and Supplementary Fig. 2), and when (ii) the large micro-phytoplankton cells left the system in a fast-sinking process, leading to a shift to small dinoflagellates (mainly Gymnodiniales < 20 μm) in the late summer (Fig. 3b)²⁴.

To further examine these phytoplankton composition and biomass differences along the seasonal succession, the physiological condition and grazing pressure were also assessed through indices, using accessory pigments and Chl-a degradation products via HPLC analysis (see “Methods”). Predominantly, the highest proportions of photosynthetic carotenoids were observed in the mid/late summer, indicating high photosynthetic activity from the phytoplankton cells associated with diatom blooms. During the late spring, however, phytoplankton communities accumulated more photoprotective carotenoid pigments (see Supplementary Fig. 3a, b), likely to mitigate the cells photooxidative damage caused by enhanced irradiation, which favors species adapted to light stress³³. This indicates that the large environmental fluctuations characteristic of the spring, due to the physiological stress, typically favor a more heterogeneous community dominated by nanoflagellates and other smaller phytoplankton cells (Supplementary Fig. 2 and Fig. 3a, b)^26,34. The grazing and senescence indices, in turn, indicate that top-down control and phytoplankton cell deterioration, respectively, were more intense during January, compared to February, which may explain part of the monthly in situ Chl-a differences (Supplementary Fig. 3c, d).

The top-down control performing on phytoplankton biomass may be exerted by both microheterotrophs (e.g., bacteria, heterotrophic flagellates, and ciliates) and metazoan plankton, such as krill and salps. The fecal pellets of metazoans tend to sink quickly, and hence are improbable to be sampled³⁵. However, it is unlikely that in situ Chl-a could have reached the high values recorded here in January and February 2016 under a grazing pressure of metazoans^30,36. Therefore, it is expected that a grazing pressure associated with microheterotrophs was the major top-down control in this situation. Microheterotrophs differ from metazoans in that they are included together with phytoplankton during sample filtration for HPLC/CHEMTAX analysis and, consequently, pigments associated with their grazing, ingested and/or digested, are extracted and analyzed³⁵. Moreover, their small fecal pellets sink slowly and are thus likely to be captured with the sample³⁵. To support this case, converging with our grazing index data (Supplementary Fig. 3c), diatom blooms generally characterized by large centric cells were clearly different between both summer periods. While in February the diatom bloom was mainly composed by the large centric Odontella weissflogii (mostly > 70 µm in length; ref. ²⁴), during January a large number (> 2.5 × 10⁶ cells L^–1) of small (< 20 µm in length) pennate Fragilariopsis spp. contributed substantially to the total of diatom bloom biomass. Given that large cell size of phytoplankton can reduce, or prevent, efficiency of zooplankton grazing^37,38, this intra-group difference between the diatom blooms may be indicating that a selective activity of microheterotrophic grazing occurred on smaller pennate diatoms during January 2016.

Furthermore, the effects of a strong positive SAM trend on sea ice dynamics over the previous years may have caused a severe impact on the survival of larval krill², which was reflected in low krill recruit density during 2016 along the NAP³⁹. This further supports a low top-down control from the krill population during this summer, which is also consistent with the grazing index values that were much lower than expected (Supplementary Fig. 3c, d)²⁴. It has been recorded along the Antarctic regions that these grazing index values are greater than 10% when associated with diatom blooms^40,41, thus indicating a prevalence of active grazing pressure. Therefore, it is likely that the low top-down control on produced biomass further contributed to the high anomalous values of in situ Chl-a registered over the seasons; even though a grazing pressure associated with microheterotrophic zooplankton assemblages may have driven differences in Chl-a concentrations between the mid/late summer (Supplementary Fig. 3c, d).

Despite the clear dominance of diatoms throughout the in situ sampling periods in the 2015/2016 spring/summer, there was a notable separation between November and January/February in respect to the biomass contribution (as a proportion to the total Chl-a) of the main phytoplankton functional groups (Supplementary Fig. 4). Using a canonical correspondence analysis (see “Methods”), the diatoms and dinoflagellates were positively associated with temperature, stability and Chl-a, and negatively associated with salinity, MLD and macronutrients to the mid/late summer months. The opposite was observed with the other phytoplankton groups (mostly small nanoflagellates), mainly associated with the stations sampled during November (Supplementary Fig. 4), as discussed above. This indicates that, during summer months, dinoflagellates (mainly Gymnodiniales < 20 μm) were the second most representative taxonomic group, after diatoms.

To determine the statistical relationship between diatoms and dinoflagellates along the NAP seasonal succession, all in situ data collected during February 2016 (ref. ²⁴), along with the dataset presented here, were further assessed. In general, there was a positive relationship between absolute concentrations of both groups, suggesting their coexistence (Supplementary Fig. 5a). It is possible to observe that there was a decline in diatom biomass under higher values of upper ocean stability, mainly associated with waters influenced by ice melt coming from the Bellingshausen Sea (Supplementary Fig. 5a)²⁴. This was linked to the sedimentation process of larger diatom cells out of the upper layers, due to water column structure dynamics²⁴. As a result, following the diatom sinking process, the significant positive relationship between upper ocean stability and dinoflagellates further indicates that these small flagellates were a major contributor for the phytoplankton community composition in the late summer 2016 (Supplementary Figs. 5b and 6). In recent years, it has been observed an increasingly contribution of small dinoflagellates (especially Gymnodiniales < 20 μm) to phytoplankton community composition along the NAP waters during the mid/late summer months^42,43,44, drawing attention to potential poleward shifts of Antarctic phytoplankton species¹². These small flagellates have been reported apparently occupying sites/conditions less favorable to diatoms and cryptophytes^24,42. Furthermore, going into autumn (from March onwards), the decline in SST may still suggest a period of increased wind speed, and thus vertical mixing, which indicates a potential deepening of MLD along the region. This event, together with progressive reduced daylight length⁴⁵, could further explain the decline in Chl-a values from the satellite data to the end of the seasonal phytoplankton succession (Fig. 2), potentially favouring a community dominated by small mixed flagellates and small diatoms (Supplementary Fig. 6)⁴⁵.

Kruskal−Wallis tests were performed (see “Methods”), comparing our February time series data (2008–2017) with the atypical situation of February 2016 (our focus here), to examine differences on water physical parameters and on indices of physiological state and grazing pressure associated with high (> 50%) and low (< 30%) diatom proportions (to the total in situ Chl-a). In general, high diatom proportion (representing values in the time series and in 2016 higher than 50%) showed a strong association with deeper MLD (> 20 m), higher surface salinity (majority > 34), and lower upper ocean stability values (< 4 × 10^–6 m^–1) relative to low diatom proportion (representing values in the time series lower than 30%; see Fig. 4). These results demonstrate a preferential fundamental niche to diatoms along the NAP (Fig. 4), although there was a marked difference in biomass accumulation during February 2016 (see Supplementary Fig. 7). In addition, there were also strong differences on indices of physiological state and grazing pressure between high and low diatom proportions. The low proportion of diatoms was associated with higher (lower) ratios of photoprotective (photosynthetic) carotenoids and lower (higher) grazing (senescence) index values, which suggest a poorly productive phytoplankton community relative to high diatom proportion conditions (Supplementary Fig. 8).

The only physical parameter showing a difference between both high diatom proportions was surface temperature, which was colder in 2016. However, this parameter solely cannot explain the much higher diatom biomass observed in the late summer 2016 (mean Chl-a value of 12.4 mg m^–3) relative to the other years (mean Chl-a values lower than 1 mg m^–3; see Supplementary Fig. 7). The ratios of photoprotective carotenoids and senescence index showed higher values in the other years (2008−2015, 2017) compared with 2016, pointing for a less favorable condition during these years, even though no statistically significant difference between the ratios of photosynthetic carotenoids and grazing index was found (see Supplementary Fig. 8). As high diatom proportions showed similar behavior regarding water physical parameters (Fig. 4), differences on both photoprotective and senescence indices may be indicating that less favorable conditions^35,40,41,42 were preventing diatom biomass accumulation in the other years compared with 2016. However, only through the indices is not possible to disentangle the potential critical factor associated with these unfavorable conditions. Analyses of diatom life cycle, considering interspecific differences, prevailing irradiance, and nutrient availability (mainly those in trace amounts) will be needed to evaluate, and understand, the critical factor (factors) which is (are) hampering diatom bloom formation, even under favorable upper ocean physical structures. The difference in space and time should be considered, assuming the likelihood of sedimentation process (or consumption by metazoan plankton) of large diatom cells before the sampling, that is, low diatom biomass but favorable upper ocean physical structures. Furthermore, although there was no statistical difference between grazing index values associated with high diatom proportions; activity of grazing in the other years cannot be disregarded as a potential driver on diatom biomass, as the grazing index in these years in respect to 2016 showed a wide range of values, which in part exceeded 10% (see Supplementary Fig. 8c).

Over a longer spatial/temporal scale, the effects of climate warming are progressing southward beyond the NAP^1,3. The progression of these effects is causing impacts on ocean physics and biological compartments. Accordingly, a decrease in diatom proportion has been pointed northward along the mid/southern WAP, concurrent with an increase in cryptophyte proportion³. This shift has also been reported along the NAP as a result of glacier meltwater processes during summer^12,13,14. Due to meltwater inputs, the cryptophyte fundamental niche has been associated along the NAP with mean values of salinity lower than 34 (refs. ^13,14), which is consistent with other regions in the mid/southern WAP^3,11,46. Moreover, It has also been observed that the peak of cryptophyte abundance is related to warmer air temperature (> 2.3 °C), contrasting with diatom peak that was found at lower mean air temperature (~1.2 °C)¹¹. Such evidence links the reported trends of an increase in cryptophyte proportion and warmer ocean and atmospheric temperatures, which play a pivotal role in glacier retreat rates, driving a substantial glacial melting over the past warming decades along the Antarctic Peninsula^8,9,16.

During the spring/summer of 2015/2016 reported here, diatom blooms were associated with lower mean air temperatures and high mean surface salinity values (< 0.2 °C and > 34, respectively; see Fig. 4, Table 1, and Supplementary Table 1), which could be indicating a lower incoming of meltwater¹⁶. Likewise, during the moderately strong 2009/2010 El Niño event, large diatom blooms along the NAP coastal region were associated with low air temperature, which caused a delay in the melting of surrounding glaciers, leading to average salinity values greater than 34 (ref. ⁴⁷). These findings are consistent with our results, which reinforce that the effects of the extreme 2015/2016 El Niño on atmospheric/ocean physics were substantial, triggering anomalous diatom blooms during the seasonal phytoplankton succession (Fig. 4). However, even with large diatom blooms during recent El Niño events, the cryptophyte growth has still been reported along the NAP^13,45. The emergence of this nanoflagellate was associated with localized inputs of glacial meltwater, decoupled from the diatom blooms, supporting the segregation and/or competition between these two phytoplankton functional groups^11,12. This differentiation in upper ocean physical structures caused by the influence of glacial meltwater input, which leads to shallower, stabilized, and well-lit upper layers, based on results found here, apparently prevents diatom bloom formation along the NAP due to the fast-sinking process, decreasing biomass accumulation (Fig. 4d and Supplementary Fig. 5a). It is important to note that, in a first moment, diatom contribution increased concomitantly with the upper ocean stability (Fig. 3), decreasing its biomass on contrary under higher stability values (> 4 × 10^–6 m^–1; Supplementary Fig. 5a). These results converge with the distribution of low diatom proportion associated with higher (lower) upper ocean stability (surface salinity) and shallower MLD (Fig. 4b−d). Additionally, other factors such as daily, seasonal and latitudinal fluctuations in solar radiation^41,48, top-down control^30,49, interspecific differences³³, and micronutrient limitations¹⁶ within the upper ocean cannot be disregarded for further understanding diatom dynamics along the NAP, and in other Antarctic coastal regions.

The substantial long-term physical changes along the NAP have been linked to strengthening winds, driven by a positive SAM trend and its co-variability with La Niña^1,10,16. However, this is opposed to the results reported here, in which the effects of the extreme 2015/2016 El Niño notably superimposed on the persistent positive SAM trend. In this study, we demonstrated a close link between large diatom blooms, upper ocean physical structures and sea ice cover during the spring/summer of 2015/2016 along the NAP (Figs. 2–4). Moreover, there was also evidence of low top-down control on the diatom biomass (Supplementary Figs. 3c and 8c), which further supports that bottom-up control is a key mechanism of the marine ecosystems along the WAP⁵⁰. As it is predicted by models that positive SAM frequencies will increase for the next 50 years⁵¹, coupled with strong winds and warmer temperatures, further glacial retreat and sea ice decline⁶ should be expected. Consequently, this would provide a more favorable fundamental niche for cryptophytes^3,12. This hypothetical future scenario would make large diatom blooms typically episodic events, intensifying and/or establishing what it has been observed over the past warming decades across the NAP^{1,12,13,14,47}. Therefore, despite a seasonal phytoplankton succession with massive concentrations of diatoms during the austral spring/summer of 2015/2016; due to continued warming trends, diatom blooms may become more episodic and restricted at spatial/temporal scales with increasingly meltwater inputs along the NAP.

Source: Ecology - nature.com