Chemotaxonomic characterization of the key genera of diatoms in the Northern Antarctic Peninsula

: Diatoms are successful in occupying a wide range of ecological niches and biomes along the global ocean. Although there is a recognized importance of diatoms for the Southern Ocean ecosystems and biogeochemical cycles, the current knowledge on their ecology and distribution along the impacted Antarctic coastal regions remains generalized at best. HPLC–CHEMTAX approaches have been extensively used to this purpose, providing valuable information about the whole phytoplankton community, even for those small-size species which are normally diffi cult to identify by light microscopy. Despite that, the chemotaxonomic method has reserved minimal focus on great diversity of types associated with diatom genera or species. Here, we show a coupling between the key genera and the corresponding chemotaxonomic subgroup type-A or type-B of diatoms via HPLC–CHEMTAX and microscopic analysis, using chlorophyll– c 1 and chlorophyll– c 3 as biomarker pigments, respectively. The results demonstrated strong correlations for nine of the fi fteen most abundant diatom genera observed along the Northern Antarctic Peninsula, from which fi ve (four) were statistically associated with chlorophyll– c 1 (chlorophyll– c 3 ). Our study highlights the importance to observe diatoms in greater detail, beyond being only one functional group, for a better understanding on their responses under a climate change scenario.


INTRODUCTION
Most of the excess energy stored in the climate system due to anthropogenic greenhouse gas emissions has been absorbed by the oceans (Stocker 2015, Zanna et al. 2018. Covering 30% of the global ocean's surface, the Southern Ocean has been pointed out to account for up to 75% of the heat uptake by the oceans over the past century (Frölicher et al. 2015). In addition, it is also responsible by nearly half of the oceanic CO 2 uptake (Takahashi et al. 2012), mainly associated with its high seasonal phytoplankton production, indicated as a critical factor in driving oceanic CO 2 uptake during the austral summer (Brown et al. 2019, Costa et al. 2020, Monteiro et al. 2020. Due to progressive southward warming, however, shifts in the distribution of polar species have been observed in Antarctic coastal regions experiencing strong physical alterations (Ducklow et al. 2007, Schofi eld et al. 2010, especially at the base of the food web in marine ecosystems (Montes-Hugo et al. 2009), which may be impacting the fundamental role of the Southern Ocean on the Earth's energy balance and global biogeochemical cycles.
The Northern Antarctic Peninsula (NAP), which encompasses the Gerlache and Bransfi eld Straits, the southernmost sector of the Drake Passage, and the northwestern Weddell Sea, is a good example of how environmental changes can alter and transform the polar ecosystem through time (Kerr et al. 2018). The NAP marine ecosystem has experienced substantial physical changes due to long-term increases in atmospheric and oceanic temperatures (Henley et al. 2019, Ferreira et al. 2020. Marine glaciers are retreating (Cook et al. 2005(Cook et al. , 2016, and the sea ice season has been shortened (Stammerjohn et al. 2008). These long-term physical changes have led to shifts in phytoplankton community composition, potentially threatening trophic structure and biological carbon pump in the local ecosystems (Costa et al. 2020, Ferreira et al. 2020. Diatoms, considered the core base of the Southern Ocean food web (Armbrust 2009, Rousseaux & Gregg 2014, have been replaced by small flagellate cryptophytes (Montes-Hugo et al. 2009, Mendes et al. 2018a. Nonetheless, the critical factors, which are hampering the diatom blooms development during the austral summer along the NAP, as well as in other impacted Antarctic coastal waters, remain an open question (Henley et al. 2019, Costa et al. 2020, Pan et al. 2020. The study of phytoplankton along the NAP has started more than forty years ago (Ferreira et al. 2020). Since then, different methodological approaches have been adopted to measure the biomass and community composition of these microalgae (e.g., Holm-Hansen & Mitchell 1991, Rodriguez et al. 2002, Mendes et al. 2012, Costa et al. 2020. Among them, the chemotaxonomic (HPLC-CHEMTAX) approach has been extensively used in the NAP phytoplankton studies (Rodriguez et al. 2002, Mendes et al. 2012, 2018a, 2018b, Costa et al. 2020, Pan et al. 2020). This approach based on the chemical taxonomy (CHEMTAX) software uses a factor analysis and steepest descent algorithm to best fit the pigment data onto an initial matrix of pigment ratios, providing valuable information about the whole phytoplankton community, even for those small-size species which are normally difficult to identify by light microscopy. As a result, the CHEMTAX initial matrix ratios based on class-specific accessory pigments and chlorophyll-a (Chl-a) have been modified and improved though time (Rodriguez et al. 2002, Kozlowski et al. 2011, Mendes et al. 2012, 2018b, Costa et al. 2020. Recently applied for the NAP, a separation for the functional group of diatoms has been proposed due to their great diversity of types, which has resulted in two chemotaxonomic subgroups: diatoms type-A and diatoms type-B, identified through chlorophyll-c 1 and chlorophyll-c 3 , respectively (Costa et al. 2020). However, despite this improvement, results from the chemotaxonomic subgroups have not been simultaneously correlated with measurements of diatom genera or species through microscope identification.
Diatoms are successful in occupying a wide range of ecological niches and biomes along the global ocean, mostly due to their great diversity in morphology and physiology (Malviya et al. 2016, Tréguer et al. 2017. Indeed, a great number of diatom genera and species has been recorded along the Antarctic Peninsula coastal waters (Annett et al. 2010), distributed over a large range of salinity and temperature (Schofield et al. 2017, Brown et al. 2019. Although there is a recognized importance of diatoms for the Southern Ocean ecosystems and biogeochemical cycles, the current knowledge on their ecology and distribution along the impacted Antarctic coastal regions remains generalized at best (Schofield et al. 2017, Henley et al. 2019. Most previous studies have been limited to a taxonomic (microscopic identification) or chemotaxonomic approach and/or with minimal focus on types of diatoms based on CHEMTAX. Here, we show a coupling between the key genera and the corresponding chemotaxonomic subgroup of diatoms. Furthermore, we assess the distribution patterns of the two types of diatoms using both chemotaxonomic and taxonomic dataset, which encompass seasonal succession periods (November-March) collected during eight cruises (between 2013-2016) along the NAP. Our study provides a baseline to the use of an initial matrix of pigment ratios to be applied to further phytoplankton ecology research, particularly diatom chemotaxonomic studies across the NAP, as well as in other Antarctic regions. A thorough analysis of major diatom genera/species with their pigment suite may help to disentangle the large range of hydrography features associated with diatom distribution.

Study area and sample collection
Data for this study were collected during eight cruises conducted on board the RV Almirante Maximiano of the Brazilian Navy during the late spring and summer months (November, January, February, and March, respectively) between 2013 and 2016 along the NAP. The study area covered the Gerlache Strait, which separates the Anvers and Brabant Islands from the Antarctic Peninsula; Bransfield Strait, located between the Southern Shetland Islands and the Peninsula; and regions located near the Elephant Island ( Fig. 1). February 2015 was sampled across the The discrete seawater samples were collected using a combined Sea-Bird CTD (conductivity-temperature-depth instrument)/ Carrousel 911+system ® equipped with 24 fivelitre Niskin bottles. Surface seawater samples (5 m depth) were taken in all stations for analyses of phytoplankton pigments and microscopy ( Fig.  1).

Microscopic analysis
The samples collected to the microscopic analysis were preserved in amber glass flasks (± 250 mL) with 2% alkaline Lugol's iodine solution in order to evaluate the species composition and cell concentrations of the phytoplankton assemblage. Settling chambers (2, 10 or 50-mL settling volume) were inspected on an Axiovert 135 ZEISS inverted microscopic (Utermöhl 1958, Sournia 1978 at 100x, 200x, and 400x magnification, following previous literature (e.g., Lund et al. 1958). Species-specific cell biovolumes were estimated by measuring cell dimensions (such as length, width and height) and applying respective similar geometric shapes (Hillebrand et al. 1999). Carbon biomass (µgC.L -1 ) was obtained from the biovolume values through a carbon to volume ratio used for diatoms (Menden-Deuer & Lessard 2000). The most abundant diatom genera were selected for cell carbon biomass calculation; that is, those genera that were identified in at least 10% of the samples counting via microscopic analysis (Fig. 2), as a significant statistical number of data was needed for making reasonable correlations with the results of HPLC-CHEMTAX analysis. Additionally, diatom family Naviculaceae was also considered in the analysis.

HPLC analysis
For phytoplankton pigment analysis, seawater samples (0.5-2.5 L) were filtered under low vacuum through GF/F filters, which were subsequently frozen in liquid nitrogen for later HPLC pigment analysis. In the laboratory, the filters were placed in a screw-cap centrifuge tube with 3 mL of 95% cold-buffered methanol (2% ammonium acetate) containing 0.05 mg L -1 trans-β-apo-8'-carotenal (Fluka) as internal standard. Samples were sonicated for 5 min in an ice-water bath, placed at -20°C for 1h, and then centrifuged at 1100 g for 5 min at 3°C. The supernatants were filtered through Fluoropore PTFE membrane filters (0.2 μm pore size) to separate the extract from remains of filter and cell debris. Immediately prior to injection, 1000 µL of sample was mixed with 400 µL of Milli-Q water in 2.0 mL amber glass sample vials, which then were placed in the HPLC cooling rack (4°C). The pigment extracts were analysed using a Shimadzu HPLC constituted by a solvent distributor module (LC-20AD) with a control system (CBM-20A), a photodiode detector (SPDM20A), and a fluorescence detector (RF-10AXL). The chromatographic separation of the pigments was performed using a monomeric C8 column (SunFire; 15 cm long; 4.6 mm in diameter; 3.5 μm particle size) at a constant temperature of 25 °C. The mobile phase (solvent) and respective gradient followed the method developed by Zapata et al. (2000), which was discussed and optimized by Mendes et al. (2007), with a flow rate of 1 mL min −1 , injection volume of 100 μL, and 40 min runs. All the studied pigments were identified from both absorbance spectra and retention times, and the concentrations were calculated from the signals in the photodiode array detector in comparison with commercial standards obtained from DHI (Institute for Water and Environment, Denmark). Peaks were integrated using LC-Solution software, and all of the peak integrations were checked manually and corrected when necessary. A quality assurance threshold procedure, through application of quantification limit (LOQ) and detection limit (LOD), was applied to the pigment data (Hooker et al. 2005) to reduce the uncertainty of pigments found in low concentrations. The LOQ and LOD procedures were performed according to Mendes et al. (2007). In order to correct for losses and volume changes, the concentrations of the pigments were normalized to the internal standard.

CHEMTAX ANALYSIS
The absolute and relative contribution of phytoplankton groups to the overall biomass were calculated from the class-specific accessory pigments and total Chl-a using the chemical taxonomy (CHEMTAX) software v1.95 (Mackey et al. 1996). The basis for calculations and procedures are fully described by Mendes et al. (2018a). Although Chl-a concentration is not an absolute measure of algal biomass, such as carbon, it can be used as a proxy for biomass (Jeffrey et al. 1997, Huot et al. 2007, as this photosynthetic pigment is common to all autotrophic phytoplankton species. Therefore, here we use the term Chl-a referring to either absolute biomass or relative biomass attributed to the corresponding chemotaxonomic groups (Mendes et al. 2012, 2018a, 2018b, Costa et al. 2020. In this study, we applied two different initial matrices of pigment ratios to CHEMTAX (Supplementary Material -Tables SI and  SII), which gave rise to two independent methodological approaches: "Method A" and "Method B". In the Method A, seven algal groups were chosen: diatoms type-A, diatoms type-B, dinoflagellates type-A, dinoflagellates type-B, haptophytes type 8 (Phaeocystis antarctica), cryptophytes, and green flagellates. In this method, the initial matrix of pigment ratios was based on three biomarker pigments (fucoxanthin, chlorophyll-c 1 and chlorophyll-c 3 ) to estimate total diatom biomass (sum of biomass of diatoms type-A and diatoms type-B). Fucoxanthin and chlorophyll-c 1 (Chl-c 1 ) were used as biomarkers of diatoms type-A, and fucoxanthin and chlorophyll-c 3 (Chl-c 3 ) as biomarkers of diatoms type-B (Table SI). On the other hand, in the Method B, the initial matrix of pigment ratios was based on only one biomarker pigment (the fucoxanthin) to estimate total diatom biomass (Table SII). We applied both methods to investigate potential differences in the determination of diatom biomass, that is, over-or underestimation, assessing how much concentration of Chl-c 1 and Chl-c 3 influences in estimation of diatom chemotaxonomic subgroups. The carbon biomass of the identified diatom genera via microscopic analysis was used as a baseline to assess a potential difference between both Methods.
For optimization of the input matrix, series of 60 pigment ratio matrices were generated by multiplying each ratio from the initial matrix by a random function as described in Wright et al. (2009). The averages of the best six (10%) output matrices, with the lowest residual or mean square root error, were taken as the optimized results. The optimization of the input matrix was performed separately to each one of the eight performed oceanographic cruises, allowing to assess variations in pigment: Chl-a ratios, due to potential differences in light availability between years (Schlüter el al. 2000, Higgins et al. 2011). In addition, these procedures were performed separately to the two distinct initial matrices of pigment ratios, i.e., to the Methods A and B. The CHEMTAX derived results of diatoms were quality-controlled by microscopic analysis.

Statistical analysis
Given that there is a relationship between cell size and chlorophyll concentrations in diatoms (Stauber & Jeffrey 1988, Montagnes et al. 1994, we correlated the carbon biomass values of the most abundant diatom genera ( Fig. 2) with Chl-c 1 and Chl-c 3 of our dataset to determine their chemotaxonomic subgroups type-A or type-B (Figs. 3 and 4), which were estimated via HPLC-CHEMTAX analysis. Data were excluded from analysis whenever the concentration of either Chl-c 1 or Chl-c 3 was zero or missing, and/or carbon biomass value was zero or missing. This qualitycontrol step was performed to each diatom genus. It was considered significant relationship between carbon biomass and Chl-c 1 or Chl-c 3 , those genera that the slope significance level (that is, the Chl-c 1 or Chl-c 3 to carbon ratio) was p < 0.0001 (Figs. 3 and 4). This p-value threshold minimizes an artefact possibility on correlations of the parameters. To support the relationship between diatom genera and their respective chemotaxonomic subgroups, we summed all carbon biomass of diatom genera according to the identified chemotaxonomic subgroup (that is, type-A or type-B), and compared it with respective diatom-type biomass, estimated via HPLC-CHEMTAX analysis (mg m -3 of Chl-a; Supplementary Material - Figure S1). Furthermore, we also assessed differences between Methods A and B, comparing them with respect to the total sum of carbon biomass associated with the identified diatom genera (Fig. S2).

RESULTS
The main identified and counted diatom genera via inverted microscope analysis are shown in Fig.  2. Standing out among diatom genera, much due to its large cell size, the genus Odontella showed higher carbon biomass values when compared with the others (Fig. 2a). In the stretched carbon biomass scale of Fig. 2b, excluding the genus Odontella, was possible to observe a wide distribution of diatom genera contributing with relatively high values of carbon biomass (Fig.  2b).
There was a large variability in mean surface values of Chl-a between the sampling periods. The highest mean Chl-a value was reported during January 2016, whereas the lowest value was found in November 2013 (Table I). Fucoxanthin was the second most abundant pigment. Its highest mean value was also reported during January 2016, whereas the lowest occurred in November 2014 (Table I). Chlorophyll-c 1 (Chl-c 1 ), chlorophyll-c 2 (Chl-c 2 ), and chlorophyll-c 3 (Chl-c 3 ) showed similar patterns, although the highest mean Chl-c 3 value was found in March 2015, which contrasted with the highest values of   (Table I). In general, the chlorophylls-c (Chls-c) and the sum of other pigments showed the lowest values in November months (Table I).
The applied Methods showed a strong statistical relationship with each other through their estimated diatom absolute concentrations (R 2 : 0.99, p < 0.0001). Further, when compared with carbon biomass measurements, both methods showed strong correlations (see Fig.  S2). The chemotaxonomic subgroups of diatoms showed a large interannual variability of relative biomass along the sampled months (Fig. 5). The diatoms type-B proportion to the total Chl-a was almost always higher than diatoms type-A proportion, with exception of February 2014 (Fig.  5). An interannual variability was also observed Table I. Average concentrations (mg.m -3 ) of chlorophyll-a (Chl-a), chlorophyll-c 1 (Chl-c 1 ), chlorophyll-c 2 (Chl-c 2 ), chlorophyll-c 3 (Chl-c 3 ), fucoxanthin (Fuco), and the sum of all other pigments during the late spring and summer months (November, January, February, and March) between 2013 and 2016 along the NAP. Standard deviation is showed in parentheses. Other pigments are represented as a sum of 19 '-hexanoyloxyfucoxanthin, gyroxanthin diester, prasinoxanthin, peridinin, alloxanthin, chlorophyll b, 19'-butanoyloxyfucoxanthin, 19'-hexanoyloxy-4ketofucoxanthin, and (Table SIV). In general, there was no marked seasonal pattern on these diatom genera (Tables SIII and SIV).

DISCUSSION
It is estimated that diatoms are responsible for about 40% of marine primary productivity, and particulate carbon exported to depth as part of the biological pump (Tréguer et al. 2017). For this reason, considering that global ocean covers about 70% of the Earth's surface, diatoms can be considered as a key group among the autotrophic organisms, playing crucial functions in trophic structure and biogeochemical cycles (Assmy et al. 2013, Costa et al. 2020. The ecological success of diatoms is strongly associated with their pigment profile, composed by photosynthetic and photoprotector pigments, which in general include Chl-a, Chls-c, fucoxanthin and some other carotenoids (Stauber & Jeffrey 1988). Chlorophylls and fucoxanthin are photosynthetic pigments responsible by photosynthesis activity, whereas β-carotene, diadinoxanthin, and diatoxanthin are associated with photoprotection mechanisms to cope with light intensity fluctuations in marine ecosystems (van Leeuwe et al. 2005, Araujo et al. 2017. In diatoms, as in all phototrophic phytoplankton species, the Chl-a plays a central role in photochemical energy conversion, while Chls-c and fucoxanthin are accessory pigments responsible to efficiently rise photon accumulation through different absorption maxima (Kuczynska et al. 2015). The Chls-c cover a family of chlorophylls mainly comprising Chl-c 1 , Chl-c 2 , and Chl-c 3 , all of which serve exclusively as an antenna pigment (Kuczynska et al. 2015). Stauber & Jeffrey (1988), through analyses of the photosynthetic pigments of 51 species (28 genera) of tropical and sub-tropical diatoms, depicted that 88% of the diatoms tested have shown Chl-c 2 together with Chl-c 1 . The authors have still shown that when Chl-c 1 was absent or in trace amounts (8 species, 3 genera) it was replaced by Chl-c 3 , having only one exception of a species containing approximately equal amounts of the three Chls-c. These results, which were further investigated (e.g., Zapata et al. 2011), have provided a base to diatom type separation, applied latter to HPLC/ CHEMTAX analysis (Higgins et al. 2011, Rozema et al. 2017, Costa et al. 2020). Therefore, from among different forms of Chls-c described to diatoms, the most abundant are thought to be Chl-c 1 and Chl-c 2 (Kuczynska et al. 2015), which contrasts with our results. Here, Chl-c 3 was predominantly higher than Chl-c 1 (Table I). This could be explained by the fact that Chl-c 3 is an ambiguous biomarker pigment also present in haptophytes and dinoflagellates type-B, unlike Chl-c 1 (see Table SI).
We found strong correlations for nine of the fifteen diatom genera, from which five (four) were statistically associated with Chl-c 1 (Chl-c 3 ). The genera of diatoms Chaetoceros, Eucampia, Thalassiosira, Haslea, and Fragilariopsis were classified as diatoms type-A, whereas Naviculaceae, Odontella, Proboscia, and Rhizosolenia were classified as diatoms type-B (see Figs. 3 and 4). Surprisingly, the genus Corethron was correlated neither with Chl-c 1 nor with Chl-c 3 , even showing high carbon biomass values. Pseudo-nitzschia and Nitzschia also did not show statistical relationships with either of these two biomarker pigments (Fig. S3). This could be due to the concomitant presence of Chl-c 1 and Chl-c 3 in these genera, such as it has already been observed in the literature (Stauber & Jeffrey 1988, Zapata et al. 2011. Another critical factor preventing a potential correlation to small pennate or small centric diatom genera, for which a chemotaxonomic subgroup was not ascribed here, would be interference of Chl-c 1 or Chl-c 3 content of large diatoms of other genera. This latter case becomes evident when we observe that a high correlation with either Chl-c 1 or Chl-c 3 for most of the identified diatom genera is strongly dependent on their high carbon biomass values (see Figs. 2-4).
Furthermore, the diatom genera associated with Chl-c 1 showed higher R 2 than those associated with Chl-c 3 . As already mentioned above, this could also be due to the fact that Chl-c 3 is an ambiguous biomarker pigment, which still explains the higher (~0.2) additive constant (referred to as the y-intercept) associated with diatoms assigned as type-B than those pointed out as type-A (~0) (Figs. 3 and 4).
The strong statistical relationship for both chemotaxonomic subgroups with respect to their respective total carbon biomass values supports that those identified diatom genera are reliably represented (see Fig. S1). Moreover, the significant relationship (R 2 : 0.99; p < 0.0001) between the Method A and Method B through their estimated diatom absolute concentrations (data not shown), and the also significant relationship that both methods showed in respect to the carbon biomass values (as the total sum of carbon biomass of the identified diatom genera; see Fig. S2), indicate that regardless the applied method the chemotaxonomic estimation of diatom biomass is reliable according to the carbon biomass measurements. These results also support the diatom type separation performed by the Method A.
I t h a s b e e n re p o r t e d t h a t t h e chemotaxonomic subgroups of diatoms showed distinct biomass distribution, indicating a possible seasonal pattern between them during spring in the coastal Arctic Ocean (Alou-Font et al. 2013). The authors have found that the reported bloom was dominated by diatoms type-B associated with low snow cover sites and high irradiance, while diatoms type-A were only observed under high snow cover and low irradiance conditions (Alou-Font et al. 2013). Additionally, although no statistical relationships have been performed, this previous study still associated the chemotaxonomic subgroup type-A to the centric diatom genera Chaetoceros and Thalassiosira, converging with our analysis (Fig. 3), and type-B with the pennate diatom genera Navicula and Nitzschia. Although we did not get conclusive results on the genus Nitzschia, the genus Navicula, belonging to the family Naviculaceae, also converges with our analysis (Fig. 4). However, despite some similarity with these previous results, in our study there was no marked seasonal pattern associated with chemotaxonomic subgroups of diatoms during four sampled austral summers along the NAP (Fig. 5). Furthermore, our data did not either demonstrate any link between subgroup type-A and centric diatoms, or subgroup type-B and pennate diatoms. Two of five diatom genera assigned as type-A were pennate in our analysis (Haslea and Fragilariopsis; Fig. 3).
In our study, there was a great interannual variability in phytoplankton biomass during the austral summers along the NAP (Table  I). Although fucoxanthin is an ambiguous biomarker pigment, its concentration in other groups is in minor amounts compared with diatoms (Higgins et al. 2011), indicating that high fucoxanthin values recorded in our dataset were likely due to the contributions of diatoms (Table I), associated with their chemotaxonomic subgroups. In general, no patterns were found with chemotaxonomic subgroups of diatoms (Fig. 5), as well as with their correlated genera (Table SIII and SIV). On the contrary, the majority of diatom genera showed their highest carbon biomass contributions during January 2016 (Table SIII and Table SIII and SIV).
Likewise, a considerable interannual variability between diatom species or species groups has also been reported in the phytoplankton community over three austral summers in the Ryder Bay, a small coastal embayment of the West Antarctic Peninsula (Annett et al. 2010). These previous results reinforce that our dataset is not sufficiently large to depict with certainty any pattern associated with diatom genera, and their respective chemotaxonomic subgroups. Despite that, here we identified the key diatom genera and potentially assigned their corresponding chemotaxonomic subgroups, using both taxonomic and HPLC-CHEMTAX approaches. Although no definitive seasonal patterns were found with diatom genera and their corresponding chemotaxonomic subgroups, as Chls-c have differences in the absorption spectrum (Kuczynska et al. 2015), it is likely that the pigment profile difference between diatoms type-A and diatoms type-B may have an ecological implication. For this reason, our study highlights the importance to observe diatoms in greater detail, beyond being only one functional group, and indicates the need for further studies about diatom taxa (including small-size species) along the NAP region. These approaches will allow a better understanding on diatom responses under a climate change scenario.  (FURG). CAPES also provided free access to many relevant journals though the portal "Periódicos CAPES". We are thankful for the constructive criticism of the anonymous reviewer, who helped to improve the manuscript. All authors declare they have no conflict of interests.  Table SI. Pigment to chlorophyll a ratios used for CHEMTAX analysis in the Method A. Initial ratios before analysis (input matrix; higher panel), and one example of ratios after analysis (output matrix; lower panel). Chl-c3 = chlorophyll-c3 ; Chl-c1 = chlorophyll-c1; Perid = peridinin; But-Fuco = 19'-butanoyloxyfucoxanthin; Fuco = fucoxanthin; Hex-Fuco = 19'-hexanoyloxyfucoxanthin; Hex-kfuco = 19'-hexanoyloxy-4-ketofucoxanthin; MGDG-Chl c2 = Chl c2-monogalactosyldiacylglycerol ester; Gyro-e = gyroxanthin diester; Allo = alloxanthin; Chl-b = chlorophyll-b; Chl-a = chlorophyll-a. Table SII. Pigment to chlorophyll a ratios used for CHEMTAX analysis in the Method B. Initial ratios before analysis (input matrix; higher panel), and one example of ratios after analysis (output matrix; lower panel). Chl-c3 = chlorophyll-c3; Perid = peridinin; But-Fuco = 19'-butanoyloxyfucoxanthin; Fuco = fucoxanthin; Hex-Fuco = 19'-hexanoyloxyfucoxanthin; Allo = alloxanthin; Chl-b = chlorophyll-b; Chl-a = chlorophyll-a Table SIII. Average carbon biomass (µgC.L -1 ) of diatom genera assigned as chemotaxonomic subgroup type-A during the late spring and summer months (November, January, February, and March) between 2013 and 2016 along the NAP. Standard deviation is showed in parentheses. The asterisk (*) demonstrates carbon biomass values that are lower than 0.01. Table SIV. Average carbon biomass (µgC.L -1 ) of diatom genera assigned as chemotaxonomic subgroup type-B during the late spring and summer months (November, January, February, and March) between 2013 and 2016 along the NAP. Standard deviation is showed in parentheses. The asterisk (*) demonstrates carbon biomass values that are lower than 0.01. Figure S1. Relationships between chemotaxonomic subgroups type-A (on the left panel) and type-B (on the right panel) of diatoms (mg.m -3 Chl-a) and their respective carbon biomass values (µgC.L -1 ), as the sum of the corresponding identified diatom genera. Figure S2. Relationships of the diatom absolute concentrations (mg.m -3 Chl-a) from the Method A (on the left panel) and from the Method B (on the right panel) with carbon biomass values (µgC.L -1 ), as the total sum of the identified diatom genera. Figure S3. The lack of statistical relationship between carbon biomass values (µgC.L -1 ) of the remaining of the most abundant diatom genera and chlorophyll-c3 (Chl-c3; mg.m -3 ), represented by dark gray circles at background, and chlorophyll-c1 (Chl-c1; mg.m -3 ), which is represented by black open circles.