Dynamic responses of picophytoplankton to physicochemical variation in the eastern Indian Ocean

Abstract Picophytoplankton were investigated during spring 2015 and 2016 extending from near‐shore coastal waters to oligotrophic open waters in the eastern Indian Ocean (EIO). They were typically composed of Prochlorococcus (Pro), Synechococcus (Syn), and picoeukaryotes (PEuks). Pro dominated most regions of the entire EIO and were approximately 1–2 orders of magnitude more abundant than Syn and PEuks. Under the influence of physicochemical conditions induced by annual variations of circulations and water masses, no coherent abundance and horizontal distributions of picophytoplankton were observed between spring 2015 and 2016. Although previous studies reported the limited effects of nutrients and heavy metals around coastal waters or upwelling zones could constrain Pro growth, Pro abundance showed strong positive correlation with nutrients, indicating the increase in nutrient availability particularly in the oligotrophic EIO could appreciably elevate their abundance. The exceptional appearance of picophytoplankton with high abundance along the equator appeared to be associated with the advection processes supported by the Wyrtki jets. For vertical patterns of picophytoplankton, a simple conceptual model was built based upon physicochemical parameters. However, Pro and PEuks simultaneously formed a subsurface maximum, while Syn generally restricted to the upper waters, significantly correlating with the combined effects of temperature, light, and nutrient availability. The average chlorophyll a concentrations (Chl a) of picophytoplankton accounted for above 49.6% and 44.9% of the total Chl a during both years, respectively, suggesting that picophytoplankton contributed a significant proportion of the phytoplankton community in the whole EIO.

. In particular, Syn and PEuks all coexist in various environments, even from pole to pole (Chen et al., 2011). Observations in the oligotrophic Pacific Ocean and Atlantic Ocean showed that picophytoplankton contributed to approximately 60%-80% of the total marine primary productivity (Agustí & Llabrés, 2007;Campbell, Liu, Nolla, & Vaulot, 1997), suggesting that picophytoplankton has crucial roles in primary productivity in (sub)tropical oligotrophic regions (Matsumoto, Abe, Fujiki, Sukigara, & Mino, 2016). Collectively, owing to their high abundance, wide distribution, and large contribution to primary production, picophytoplankton have been known to have large impacts on marine ecosystem and biogeochemical cycles. Thus, presenting their biogeographic patterns is critical to understand different regional contributions to carbon cycle of these special taxa.
To date, many biostatistical models (such as quantitative niche models, parametric regression models, and neural network models) combining the variable characteristics of picophytoplankton and oceanic environments have been successfully constructed to simulate their biogeographic distributions (Flombaum et al., 2013). As one of the largest oligotrophic areas, however, the Indian Ocean (IO) particularly with respect to picophytoplankton has received far less attention than other oceans. Meanwhile, Clokie, Millard, Mehta, and Mann (2006) presented that picophytoplankton are the most abundant primary producers in the IO. Based on our size-fractionated chlorophyll a analysis, indeed, picophytoplankton are responsible for a large fraction of phytoplankton community in eastern Indian Ocean (EIO), further confirming they are potentially important for the IO ecosystem and primary productivity. To enhance our appreciation of the importance of various forms of picophytoplankton, similar exercises should be conducted in regional scale with higher spatial resolution in the IO. The tropical IO forms the major part of the largest warm pool on the earth, and its interaction with the monsoon plays an important role in shaping complex circulation systems on both regional and global scales (Raven, 1998). In other words, surface circulations and water masses in the IO are considerably complex and highly variable because of its response to the annually reversing monsoon winds.
Therefore, we speculated that the spatio-temporal variability in picophytoplankton might be closely related to the annual variations of circulations or water masses in the EIO. We thereafter compared the picophytoplankton communities and associated environmental variables between spring 2015 and 2016 to address the following ques- In the present study, the dynamic responses of picophytoplankton communities to physicochemical variations associated with circulations or water masses were investigated by flow cytometry, to address the lacking data of picophytoplankton, and to understand the annual effects of environmental variables on picophytoplankton in the EIO.

| Sampling strategy
Two cruises were conducted by the R/V Shiyan-1 in the EIO: March 15, 2015to May 18, 2015and March 20, 2016to May 12, 2016 representing spring 2015 and 2016, respectively. Our study area extending from near-shore coastal waters to oligotrophic open waters covered the entire EIO and its adjacent shelf, and 56 stations were investigated (Figure 1). At each station, seawater samples were collected from 7 depths within the upper 200 m water column using 12 L Niskin bottles equipped with a SeaBird CTD (Conductivity, Temperature, and Depth; SBE 19 Plus). Temperature, salinity, and depth were recorded in situ at the same time. Photosynthetically active radiation (PAR, 400-700 nm, µmol quanta m −2 s −1 ) was measured by an RBR sensor (XRX-620).
F I G U R E 1 Study area and sampling stations during spring 2015 (red dots) and 2016 (blue dots) Seawater samples for picophytoplankton were fixed with paraformaldehyde (1% final concentration) on board. To avoid loss of resolution and changes in cell counting due to fixation or freezing, FCM samples were kept in the dark without treatment at room temperature for 10-15 min, and then quickly freeze-trapped in liquid nitrogen until analysis in the laboratory (Marie, Simon, Guillou, Partensky, & Vaulot, 2000). Seawater samples for nutrient analysis were filtered through 0.45 µm cellulose acetate membrane filters and then immediately refrigerated at −20°C for further analysis.
Analyses for the determination of nutrient concentrations including ammonium, phosphate, nitrate, nitrite, and silicate were performed by a Technicon AA3 Auto-Analyzer (Bran + Luebbe) according to the classical colorimetric methods. Dissolved inorganic nitrogen (DIN) defined as ammonium + nitrite + nitrate was analyzed using the copper-cadmium column reduction method (Guo et al., 2014). Dissolved inorganic phosphorus (DIP) and silicate (DSi) were measured using spectrophotometry with standard molybdic acids and Murphy Riley molybdenum blue reagents according to Brzezinski and Nelson (1986) and Karl and Tien (1992), respectively.
Subsamples for size-fractionated chlorophyll a (Chl a) analysis were filtered serially through 20 µm × 20 mm silk net, 2 µm × 20 mm nylon membrane, and 0.2 µm × 20 mm polycarbonate filters under a filtration vacuum of less than 100 mm Hg, then immediately refrigerated at −20°C. After returning to the laboratory, these filters were placed into 20 ml glass tubes, the pigments were then extracted by 5 ml 90% acetone, and quickly stored in the dark at 4°C for 24 hr.
Finally, the Chl a contents were determined using a CE Turner Designs Fluorometer (Liu et al., 2016;Welschmeyer, 1994).

| Flow cytometry analysis
Abundances of three picophytoplankton groups were enumerated using a flow cytometer (FCM, Becton-Dickinson Accuri C6) equipped with a laser emitting at 488 nm. Data collection of all parameters was triggered by the 488 nm scatter signal. As a total volume of only 198 µl (flow rate at 66 µl/min running for 3 min) was analyzed, and the upper size limit for picoeukaryotes was usually 5 µm, above which the cells were very rare and could not be accurately quantified. Two µm fluorescent beads (Polysciences) were added as the instrument internal standard (Olson, Zettler, & DuRand, 1993).
Different picophytoplankton populations were manually classified according to their amplitude, shape, and position of optical signals in the scatterplots of relative red fluorescence (FL3, >670 nm) versus relative orange fluorescence (FL2, 585 ± 42 nm) and FL3 versus side scatter (SSC). Three dominating populations including Syn, Pro, and PEuks were identified by FCM ( Figure 2). Otherwise, PEuks in some special stations were also differentiated three subclusters (I-III) by their distinct red fluorescence signals. However, it was difficult to identify and separate these subclusters from all stations, so that these several subclusters of PEuks would be combined when describing their abundance and distribution.
To avoid loss of dispersion and resolution of target cells in the scatterplot due to the overlapping signals with noise, the trigger threshold (~600) was set well below the lowest scatter signals from surface Pro cells. Spurious scatter events and fluorescence cross-talk were minimized by setting the polarization of excitation laser perpendicular to the axis of flow. After these settings, 2 ml DI water was preferentially run at a steady flow rate of 66 µl/min to collect and gate the scatter and fluorescence signals of noise. If higher background counts were detected, the FCM was thoroughly cleaned with 5% bleach. The fluorescence and scatter signals were captured with user-built detector assemblies with an extended range. However, these detectors cover a signal range of more than six decades by combining the signals from two photomultipliers that operate at different gains ( Van et al., 2017). The relative gain settings were calibrated by a regression analysis of the events that fall within the linear window of both detectors. This approach can accurately detect all instantaneous picophytoplankton sizes ranging from dim surface dwelling Pro to bright PEuks.  (Wyrtki, 1973). The high levels of surface temperature and salinity were particularly observed between 80°E and 90°E around the equator. This was probably because the WJ was strongest between 60°E and 90°E and profoundly changed its water layer structure by removing the relatively high salinity and warm surface seawater from west and accumulating it in the east. The northeast part of the studied area represented a cold tongue with a low temperature of approximately 29°C, where was primarily influenced by the surface freshwater from the Bay of Bengal (BBR) (Sengupta, Bharath Raj, & Shenoi, 2006). Nevertheless, there were contrasting differences in physical background nearby the Sumatra between The vertical profiles of temperature and salinity along the Sumatra are shown in Figure 4. Water columns were apparently stratified during both years due to high surface temperature in spring. Shetye et al., (1993)

| Comparisons of picophytoplankton abundance and chlorophyll a between spring 2015 and 2016
During spring 2015 and 2016, Syn were most abundant in the surface layers and only very few cells were counted below 100 m.
Notwithstanding PChl a concentrations showed considerably low values throughout the EIO, they are responsible for a large fraction of the total Chl a concentrations, accounting for average percentage 49.6% and 44.9% of PChl a/Total Chl a in the whole EIO, respectively. Very similar temporal variations were found between PChl a concentrations and Syn and Pro abundances.

| Horizontal distributions of picophytoplankton abundance and chlorophyll a
Since Pro abundance was seldom observed at the surface layer, the water column integrated abundance involving a series of abundance variations with depth was better suited for spatial distributions of picophytoplankton than the averaging through the sampled layer. The

| Vertical distributions of picophytoplankton abundance and chlorophyll a
To clearly understand and compare the vertical pattern, all the data points of picophytoplankton abundance and PChl a concentration against depth were analyzed to fit the nonlinear curves ( Figure 8).
This nonlinear analysis provided the real shape of the response curve of abundance to depth and highlighted the variance (R 2 ) in regression models. There were striking similarities between the two years in vertical distribution of three picophytoplankton groups.
For Syn, the maximum abundance distinctly occurred in the surface However, the depths between the maximum abundance of Pro and PEuks and the DPCM were fairly similar.

| Relationship between picophytoplankton abundance and biological and environmental factors
In both spring, abundances of Syn, Pro, and PEuks were positively correlated with each other (Pearson rank correlation coefficient r > 0.46, p < 0.001). Furthermore, Pro and PEuks abundances were also positively correlated with PChl a concentration (r > 0.19, p < 0.001). Differently, Syn abundance was positively correlated with PChl a concentration in spring 2016 (r = 0.64, p < 0.001), while no significant correlation was found in spring 2015 (Table 2).
Canonical correspondence analysis (CCA) is a popular multivariate extension analysis of weighted averaging ordination, particularly developed to relate biological assemblages of species to known variation in the environment (Braak, 1986;Braak & Verdonschot, 1995).
Questions in biological ecology that have typically been studied by indirect gradient analysis can be answered more directly by the CCA.
In addition, the CCA is an efficient ordination analysis when species have regression response curves or surface (vertical) with respect to environmental gradients, and is therefore more appropriate for analyzing data on species and environmental variables than other analysis (Braak, 1986;Hill, 1991). In summary, picophytoplankton variation can be directly related to environmental variation by the CCA analysis.
The CCA leads to an ordination diagram in which point represents species, and vector represents environmental factor (Figure 9). The

| Characteristics of picophytoplankton abundance
Remarkable differences in Syn and Pro abundances were observed between spring 2015 and 2016, which showed these picocyanobacteria were approximately 2 ~ times higher abundances in spring 2016 (Table 1)

| Characteristics of picophytoplankton horizontal distribution
It is well known that the variation in temperature and salinity (T-S) is a useful aid in discriminating the various water bodies and studying their sources and mixing (Tomczak, 1999). Combined with our spring data of water temperature and salinity, the possible vital water masses were accurately discriminated, such as the Wyrtki jets, coastal current from Bay of Bengal and upwelling along the Sumatra (Figures 3 and 4). However, only our T-S dataset is not yet fine enough to fully explain the changes in circulations throughout the EIO. Therefore, we supplemented the circulation system based on the references and summarized a intuitive schematic of circulations or water masses in the EIO (Figure 10). Sri Lanka island is a TA B L E 1 Range and median of picophytoplankton abundance (×10 3 cells ml −1 ), PChl a concentration (µg L −1 ), and average percentage of PChl significant place encircled by the Indian Ocean, and is considered as a center of import and export commercial harbors. Most of coastal waters around the Sri Lanka island are influenced by the increases of pollution and eutrophication (Silva, 2007). Simultaneously, the East India Coastal Current (EICC) along the western boundary of the Bay of Bengal flows equatorward and bifurcates east of the Sri Lanka island, but one bifurcation of its source waters characterized by nutrient enrichment continues along the coast of Sri Lanka island (Vinayachandran et al., 2005). This studied coastal area is relatively abundant in nutrients and is therefore suitable for picophytoplankton development ( Figure 5). This increase in nutrients near the coast can induce the regional increase in abundance and PChl a concentration, however, which is not yet fine enough to fully represent the total PChl a concentration throughout the EIO. The total PChl a concentration in the EIO remained lower than other oceans. Syn can be subdivided into open ocean and coastal phylogenetic clusters which have not salt requirements for growth (Dufresne et al., 2008;Sohm et al., 2015). Our CCA analysis revealed that Syn abundance was negatively correlated with nutrients and salinity (r < −0.26, limiting factor for Syn growth in the EIO. In situ as well as experimental observations showed that nutrients and its efficient uptake significantly influence the growth of Syn (Bemal & Anil, 2016). Syn also exhibit a wide range of elemental stoichiometry, including carbon-to-nitrogen ratios and increased their carbon-to-phosphorus ratios in response to low dissolved phosphorus availability (Baer et al., 2017). The higher growth rate of coastal strains than the open ocean strains is mainly attributed to differences in nutrient concentrations between these two regions (Liu et al., 1998;Sohm et al., 2015). As such, Syn were generally the predominant groups in coastal waters of the Sri Lanka island attributing to a combined result of different taxonomic composition and superior ability to adapt a wide variety of nutrient concentrations. Previous studies revealed that PEuks are most abundant in nutrient-rich than in oligotrophic oceans (Wang, Huang, Liu, & Chen, 2014;Worden & Not, 2008). In general, Pro dominate in the oligotrophic environments of subtropical and tropical oceans, but their distributions are limited at high latitudes by low temperature (Johnson et al., 2006) and are also limited in coastal waters, upwelling areas and temperate oceans by environmental factors such as nutrients availability, heavy metal toxicity, and competition among groups (Lee, Choi, Youn, & Roh, 2014).
No significant correlations were found between Pro abundance and temperature in both spring (Figure 9), thus implying temperature is not the most important factor controlling the horizontal distribution of Pro in the tropical EIO. However, low temperature has significant limiting effect on the vertical distribution of Pro (r > 0.23, p < 0.001).
Interestingly, Pro exhibited a remarkably high density in the coastal zones nearby the Sumatra and Sri Lanka island, while this finding was also observed in the Pearl River Estuary where an intrusion of oligotrophic seawater enriched Pro abundance (Zhang et al., 2013).
Within this study, Pro's persistent or sudden appearance nearby the nutrient-rich coastal waters or upwelling zones could not be simply associated with the intermittent intrusion phenomenon. Based on our CCA analysis, the close associations between nutrients and Pro abundance suggested that nutrient availability was the crucial vari- However, their abnormal appearance along the equator appears to be associated with the annual changes of the circulations or water masses. Surface water structure around the equator was profoundly modulated by the WJ through removing the relatively high salinity (above 34) and warm (>30°C) surface seawater from west and accumulating it in the east (Figures 3 and 4).

| Characteristics of picophytoplankton vertical distribution
A simple conceptual model for vertical patterns of picophytoplankton based upon environmental parameters is proposed ( Figure 11).
Three picophytoplankton groups showed two different characteristics of vertical distribution in both spring. One group, including Syn, was frequently abundant in the surface and then gradually decreased with depth. Another group, including Pro and PEuks, was characterized by the subsurface maximum ( Figure 8). Jiao, Yang, Koshikawaz, and Watanabez (2002) stated that the vertical distribution of Syn can be understood in terms of nutrient and light availability. In the present study, however, temperature and light irradiation tended to be the factors mostly affecting Syn abundance and distribution. This finding well agreed with previous temperature analysis on diverse strains of Syn reported by Mackey et al., (2013) and Sohm et al., (2015), who presented Syn have evolved a suite of temperature acclimation strategies to underlie the larger geographic range of this group. In addition to the influence of temperature, Grébert et al., (2018) Olson, and Chisholm (2001) reported the depth of Pro maximum abundance is significantly correlated with the nutricline depth.
Indeed, the nutricline in the EIO was near the subsurface 50-75 m during both spring ( Figure 6). Meanwhile, Moore et al., (1995) proposed the growth of most Pro strains is inhibited at temperatures higher than 25°C. The temperature of surface seawater in the EIO was fairly high with an average value of approximately 30°C ( Figure 3). Therefore, the high temperature at the surface potentially limited the Pro growth. Although not statistically significant in the CCA analyses, too low temperature similarly has limiting effect on the vertical distribution of Pro as discussed above (r > 0.23, p < 0.001). Malmstrom et al., (2010) similarly observed Pro ecotypes each respond differently to the variation in temperature and light.
The temperature may be another reason for the vertical variation of Pro abundance in the EIO. Consequently, Pro with a maximum abundance near the subsurface were probably formed by the combined effects of temperature and nutrient availability in the EIO.
PEuks were less light dependent and could thrive over a very wide light gradient resulting from their ecophysiological heterogeneity (Worden & Not, 2008). Previous studies implied that PEuks are ubiquitous in the marine environment with population maximum occurring frequently in low irradiance, but high nutrient environments (Zhang et al., 2013). In particular, PEuks are able to tolerate lower temperature. There were thus high abundance of PEuks evidently at 50-75 m depth (which did not extend to the surface).

ACK N OWLED G M ENTS
The authors are grateful to Professor Dongxiao Wang for supplying the temperature and salinity data for the cruise. We also express thanks to the crews of R/V Shiyan I and the Open Cruise of National Natural Science Foundation of China for the Indian Ocean. This study was supported by the National Natural Science Foundation of China (41676112)