Phytoplankton growth and microzooplankton grazing dynamics across vertical environmental gradients determined by transplant in situ dilution experiments

Abstract

The Costa Rica Dome (CRD) represents a classic case of the bloom-forming capacity of small phytoplankton. Unlike other upwelling systems, autotrophic biomass in the CRD is dominated by picocyanobacteria and small eukaryotes that outcompete larger diatoms and reach extremely high biomass levels. We investigated responses of the subsurface phytoplankton community of the CRD to changes associated with vertical displacement of water masses, coupling in situ transplanted dilution experiments with flow cytometry and epifluorescence microscopy to assess group-specific dynamics. Growth rates of Synechococcus (SYN) and photosynthetic picoeukaryotes (PEUK) were positively correlated with light (R_{pearson_SYN} = 0.602 and R_{pearson_PEUK} = 0.588, P < 0.001). Growth rates of Prochlorococcus (PRO), likely affected by photoinhibition, were not light correlated (R_{pearson_PRO} = 0.101, P = 0.601). Overall, grazing and growth rates were closely coupled in all picophytoplankton groups (R_{spearman_PRO} = 0.572, R_{spearman_SYN} = 0.588, R_{spearman_PEUK} = 0.624), and net growth rates remained close to zero. Conversely, the abundance and biomass of larger phytoplankton, mainly diatoms, increased more than 10-fold in shallower transplant incubations indicating that, in addition to trace-metal chemistry, light also plays a significant role in controlling microphytoplankton populations in the CRD.

INTRODUCTION

Picophytoplankton (0.2–2 µm) are commonly viewed as a relatively stable component of the photosynthetic community, while larger phytoplankton cells respond more dynamically to physical perturbations that affect nutrient and light supply (Chisholm, 1992; Marañón et al., 2012). This is the prevailing paradigm for large regions of tropical and subtropical oceans where autotrophic biomass is dominated by picophytoplankton (Li et al., 1983; Letelier et al., 1993), which exhibit little temporal variability despite high turnover rates (Laws et al., 1984; Banse and English, 1994; Gasol et al., 1997). The stability of picophytoplankton standing stocks is achieved to a large extent through protistan grazing activity, which keeps phytoplankton populations in check while efficiently recycling the nutrients that sustain growth of their prey (Landry et al., 1997; Landry, 2002; Banse, 2013). Conversely, larger phytoplankton cells, by virtue of their lower susceptibility to grazing (Kiørboe, 1993), nutrient storage (Geider et al., 1986) and higher carbon-specific photosynthetic rates in the case of diatoms (Cermeño et al., 2005a, b) dominate in productive systems (Chisholm, 1992; Cullen et al., 2002) and respond to physical perturbations such as mesoscale eddies that inject nutrients into the nutrient-impoverished surface waters of tropical and subtropical oceans (Bibby et al., 2008; Brown et al., 2008).

The “rising tide hypothesis” provides field evidence from the equatorial Pacific that cellular growth rate and population abundance of picophytoplankton respond to the same physical processes that stimulate larger phytoplankton growth (Barber and Hiscock, 2006). Picophytoplankton-dominated blooms reported in oceanic waters of the tropical and subtropical northeast Atlantic (Partensky et al., 1996; Gutiérrez-Rodríguez et al., 2011) and the Arabian Sea (Latasa and Bidigare, 1998), although relatively modest (0.5–1 μg Chl a L⁻¹) in comparison to nano- and microphytoplankton dominated blooms (1–20 μg Chl a L⁻¹) (Marañón et al., 2012), further support the ability of small phytoplankton to, at least transiently, avoid grazing control and increase their abundance. The Costa Rica Dome (CRD) upwelling, an offshore thermal dome that develops seasonally in the eastern tropical Pacific (Fiedler and Talley, 2006), represents a classic case of the bloom-forming capacity of small phytoplankton that makes it unusual in terms of community structure and primary production levels (Taylor et al., 2016; Selph et al., 2016). Unlike other upwelling systems typically dominated by larger eukaryotic phytoplankton (Cullen et al., 2002), photosynthetic biomass and primary production in the CRD are dominated by picophytoplankton, with Synechococcus (SYN) reaching extremely high concentrations (∼3 × 10⁶ cells mL⁻¹) (Li et al., 1983; Saito et al., 2005). Trace-metal availability has been identified as one of the primary factors allowing SYN and Prochlorococcus (PRO) to outcompete large eukaryotic plankton in the CRD upwelling system (Franck et al., 2003; Saito et al., 2005; Ahlgren et al., 2014; Chappell et al., 2016). However, in addition to the trace-metal conditions that select for picocyanobacterial growth, the high abundances of SYN, PRO and photosynthetic picoeukaryotes (PEUK) during the CRD summer upwelling suggest unusual mechanisms that allow small phytoplankton to escape the tight grazing control typically exerted by microzooplankton.

Given the significant contribution of picophytoplankton to primary production globally (Li, 1994; Agawin et al., 2000) and the high proportion of this production channeled through protistan grazers (Sherr and Sherr, 2002; Chen et al., 2009), growth and grazing imbalances between these trophic levels can have profound implications for the structure and functioning of pelagic ecosystems, particularly in those like the CRD dominated by small phytoplankton (Li et al., 1983; Ahlgren et al., 2014; Taylor et al., 2016). However, we know very little about the mechanisms that regulate the coupling between phytoplankton growth and microzooplankton grazing and how these mechanisms interplay with physical and chemical processes to ultimately enhance (or suppress) the accumulation of picophytoplankton biomass (Irigoien, 2005). Light can play an important structuring role in the community due to the size dependence of optical light absorption and the packaging effect (Finkel, 2001; Mei et al., 2009). The increasing contribution of small phytoplankton to community biomass during reduced light and nutrient-sufficient winter conditions of temperate seas, for instance, has been related to the lower susceptibility of smaller cells to the package effect (Raven, 1998; Finkel et al., 2004; Marañón, 2015). There is also evidence that phytoplankton vulnerability to protist-mediated grazing mortality (e.g. Dolan and Simek, 1999; Strom, 2001; Gutiérrez-Rodríguez et al., 2009) and resulting net phytoplankton growth (Landry et al., 2011) can be affected by light. However, these studies are scarce, and little overall is known about how light influences predator–prey interactions, and the net outcomes of these trophic interactions at population and community levels. In this context, the main motivation for our work was to gain insights on how light affects phytoplankton dynamics directly and indirectly through growth and grazing, thereby structuring the community. The strong vertical stratification and picophytoplankton-dominated community that develop during the summer upwelling season of the CRD offer a good opportunity to investigate the simultaneous effects of light upon growth and grazing rates of specific phytoplankton groups. Ultimately, we aim to better understand the potential roles of light in establishing the distinctive phytoplankton community structure of the CRD and, more generally, the conditions that favor the accumulation of dominating concentrations of picophytoplankton in the pelagic ecosystem.

Toward these ends, we conducted 24-h in situ dilution experiments with subsurface phytoplankton communities that we transplanted to shallower and deeper depths. With this approach, we aimed to simulate light changes associated with the shoaling (deepening) of the thermocline on subsurface phytoplankton populations brought to higher (lower) isolumes during the activation (relaxation) of the CRD summer upwelling. We prepared a series of replicated dilution experiments with subsurface water collected from a single depth (i.e. the same initial community composition and physiological conditions), which was subsequently incubated across eight different depths in situ (one replicate per depth) spanning the euphotic zone. We analyzed the variability of SYN, PRO and PEUK growth and grazing rates, and daily changes of microphytoplankton biomass when incubated at different irradiance levels to test the effects of light on phytoplankton growth and grazing mortality, and community structure resulting from these population dynamics. We hypothesized the following: (i) that the growth of phytoplankton community exposed to higher irradiance would be enhanced relative to grazing, leading to positive phytoplankton net growth rates, while the opposite would be expected when the community was switched to lower irradiance conditions at deeper layers; (ii) that SYN would maintain higher growth rates than PRO when transplanted to high illuminated surface layers, while the latter would maintain higher net growth rates when transplanted to deeper layers; (iii) that diatoms would be favored over picophytoplankton when transplanted to shallower waters by virtue of their fast-growth capacity and lower susceptibility to protistan grazing.

METHOD

Study area, hydrography and sample collection

We sampled the CRD upwelling region (Fig. 1) on R/V Melville cruise MV1008 (22 June–25 July 2010) as part of the CRD FLUx and Zinc Experiments cruise, which had the broad goals of characterizing plankton community structure, biogeochemical fluxes and mechanisms of phytoplankton control in this ecosystem (Landry et al., 2016a). Sampling and experiments were conducted in 3–4 day periods of activity, called “cycles”, during which the water parcel tracked by drift array was followed and sampled daily (Landry et al., 2009). Hydrographic data (temperature, salinity, fluorescence and dissolved oxygen) and discrete water samples were acquired with a CTD-rosette system with 24 10-L Niskin bottles with Teflon-coated springs. Seawater samples were collected from eight depths in the upper 80–120 m at each station, extending from the surface to the depth of penetration of ∼0.1% surface irradiance (E₀). Each depth was sampled for macronutrients, chlorophyll a (Chl a), flow cytometry (FCM) and analysis of microplankton by digitally enhanced epifluorescence microscopy (Taylor et al., 2011, 2016). Nutrient samples were filtered directly from the Niskin bottle through an acid-washed, seawater-rinsed 0.1-µm Suporcap filter capsule to 45-mL plastic tubes and stored frozen at −18°C until analysis. The tubes were rinsed twice, filled, immediately frozen (−20°C) and later analyzed for major nutrient concentrations (NO₃⁻ + NO₂⁻, NO₂⁻, PO₄³⁻, NH₄⁺ and SiOH₃) by flow injection by the Analytical Laboratory at the Marine Science Institute, University of California, Santa Barbara using standard wet-chemistry methods (Gordon et al., 1992). Samples (250 mL) for Chl a were immediately filtered onto 25-mm GF/F filters, and the pigment was extracted with 90% acetone in a dark freezer (−20°C) for 24 h and quantified on a calibrated Turner Designs model 10 fluorometer (Landry et al., 2009). For FCM analysis, 2-mL seawater samples were preserved with 0.5% paraformaldehyde (final concentration) (Sigma Aldrich), flash-frozen in liquid nitrogen and transferred to a −80°C freezer until analysis. For microscopical assessment of nano- and microplankton, aliquots of 250 mL were collected, preserved and then stained with proflavin (0.33 w/v) allowing them to fix for at least 1 h before filtering onto 8.0 µm and staining with 4′,6-Diamidino-2-Phenylindole (DAPI) (Taylor et al., 2011).

Fig. 1.

Map of the Costa Rica Dome area showing the stations occupied daily during each of the experimental cycles. Stations where transplant experiments were carried out are indicated on the map as C2, C3, C4 and C5.

Picophytoplankton analysis by flow cytometry

Cell abundances of SYN, PRO and PEUK were measured using a Becton-Dickinson FACSort flow cytometer equipped with a 488-nm argon laser following the procedure described in Collier and Palenik (Collier and Palenik, 2003). Prior to analysis, fixed samples were thawed and kept in the dark at room temperature for less than an hour, and a known concentration of standard beads (0.97 µm, Green Fluorescent, Duke Scientific) was added to each sample for fluorescence normalization. SYN, PRO and PEUK populations were distinguished based on their Chl a (red fluorescence, 680 nm), phycoerythrin (orange fluorescence, 564–606 nm), forward-angle (FALS) and 90° light side-scatter (SSC) signatures. Cell counts were converted into cell concentrations (cells mL⁻¹) by calculating the volume injected for each sample from the weight of the sample tube measured before and after each run. PRO and SYN cell abundances were converted into carbon using 32 and 101 fg C cell⁻¹, respectively (Garrison et al., 2000; Brown et al., 2008), adjusted for biovolume changes with depth using bead-normalized FALS^0.55 as scaling factor (Binder et al., 1996; Landry et al., 2003). To convert PEUK cell abundances to carbon biomass, we used the total PEUK biomass and cell abundances estimated by Taylor et al. (Taylor et al., 2016) to calculate the mean biomass for PEUK cells in each cycle with such data available (Cycle 2 = 156 fg C cell⁻¹, Cycle 3 = 149 fg C cell⁻¹ and Cycle 4 = 267 fg C cell⁻¹), and the mean cell biomass calculated from all other cycles in Cycle 5 (Cycle 5 = 185 fg C cell⁻¹).

In situ incubation experimental design

We conducted a series of transplant dilution experiments at four different locations in the CRD region corresponding to the last day of experimental Cycles 2–5. At each station, seawater collected from the subsurface chlorophyll maximum (∼20–30 m, 10% E₀) was used to prepare eight replicates of two-treatment dilution assays that contained the same initial microbial community (Landry et al., 1984, 2008). For each dilution assay, one 0.26-L polystyrene tissue culture container was filled with whole seawater (WSW), and a second bottle was filled with 0.2 L of 0.1-µm filtered seawater before filling it to the top with WSW. Each unreplicated dilution assay containing two bottles was then incubated for 24 h at one different depth attached to a drifting array with a surface float and a mixed-layer drogue at 15 m (Landry et al., 2009). This setup allowed us to expose the same microbial community sampled from a single depth to a gradient of in situ irradiance and temperature conditions across the euphotic zone and to assess the variability of growth and grazing rates as a function of natural light fields difficult to simulate ex situ. FCM samples were taken at Time 0 from the collection depth and after 24 h from each experimental incubation and depth. These abundance estimates were used to calculate the net rates of change (k) of SYN, PRO and PEUK. Instantaneous growth (µ, day⁻¹) and grazing (m, day⁻¹) rates for each group were estimated assuming that grazing mortality declines linearly with dilution, as confirmed by full dilution experiments (Selph et al., 2016). Accordingly, the net rate of change (k) of cell abundance is k = µ − m in the undiluted bottles and k_d = µ − x_m in diluted bottles, where “x” is the fraction of natural grazer density in the dilute treatment (0.24 in these experiments). The two equations, $m={\displaystyle \frac{k_{\mathrm{d}}-k}{1-0.24}}$ and µ = k + m, were then solved for the two unknowns, µ and m.

Microscopical assessment of nano- and microplankton

Seawater samples (250 mL) for nano- and microplankton abundance and biomass assessment were processed and analyzed using digital epifluorescence microscopy as described in Taylor et al. (Taylor et al., 2016). Briefly, slides were imaged and digitized with a Zeiss AxioVert 200 M inverted epifluorescence microscope equipped with a fully motorized stage and controlled by Zeiss AxioVision software. Slides were viewed at ×200, and at least 20 random fields per slide were imaged. Each field image consisted of three to four different fluorescent channels: Chl a, DAPI, fluorescein isothiocyanate and phycoerythrin. The separate channel images for each field were composited into 24-bit RGB images for analysis. Counting and sizing of eukaryotes >8-µm cell lengths was semi-automated with ImagePro software. Cells were identified and grouped manually into three functional groups (diatoms, autotrophic flagellates and autotrophic dinoflagellates). All cells were binned into 5–10, 10–20, 20–40 and >40-µm based on measurements of the longest cells axis. Length (L) and width (W) measurements were converted to biovolumes (BV; µm³) by applying the geometric formula of a prolate sphere (BV = 0.524 LWH), assuming the unmeasured cell height (H) equals cell width (H = W). Carbon biomass (pg C cell⁻¹) was computed: C = 0.215 × BV^0.939 for non-diatoms, and C = 0.288 × BV^0.811 for diatoms (Menden-Deuer and Lessard, 2000).

Statistical analysis

Statistical analyses were conducted using GraphPad 5.0 software.

RESULTS

Hydrography, nutrients and chlorophyll a

The surface mixed layer was markedly shallow for Cycles 2 and 4 (10–15 m) located inside the dome and extended deeper for Cycles 3 and 5 (25–30 m) located outside of the dome (Fig. 2A). Surface temperature was similar for all experiments [27.6 ± 0.53°C, mean ± standard deviation (SD)] and decreased dramatically down to ∼16°C by ∼40 m, which gave a sharp temperature gradient in the euphotic zone (Fig. 2A). Surface salinity was virtually identical in all profiles (33.50 ± 0.01), but a steep halocline reached asymptotic values at shallower depths than the thermocline (20–30 m), particularly in Cycles 2 and 4 (Fig. 2A). The resulting strong pycnocline separated two water masses with distinctive physical, chemical and biological characteristics within the euphotic zone. Chl a concentration in surface waters ranged from 0.25 to 0.30 mg m⁻³, increased to subsurface maxima at 20–30 m depth (∼0.4 mg m⁻³) and decreased rapidly (<0.1 mg m⁻³) below 40 m. However, the chlorophyll maximum in Cycle 5 was less conspicuous, with relatively higher values (∼0.2 mg m⁻³) down to 70 m (Fig. 2A).

Fig. 2.

Vertical profiles of physico-chemical (A) and biological (B) properties at the beginning of transplant experiments conducted during each cycle. Fluorescence (FL), chlorophyll a (Chl a), Prochlorococcus (PRO), Synechococcus (SYN), picoeukaryotes (PEUK).

Figure 2B shows the vertical distributions of picophytoplankton populations in the water samples used to prepare the transplant experiments (Fig. 2B). More detailed descriptions of the structure and dynamics of picophytoplankton populations across the CRD are presented elsewhere (Selph et al., 2016; Taylor et al., 2016). Maximum concentrations of SYN were high in the surface mixed layer and decreased abruptly below the pycnocline, particularly in Cycles 3 and 5 (Fig. 2B; Taylor et al., 2016). In addition to lower surface concentrations, SYN was virtually absent below the mixed layer at these stations (Fig. 2B). Conversely, SYN concentrations below the mixed layer remained substantial in Cycles 2 and 4, and showed a secondary peak (∼10⁵ cells mL⁻¹) (Fig. 2B), which in some cases reached higher abundances than at the surface (Taylor et al., 2016). PRO vertical distributions also differed between Cycles 2 and 4 and Cycles 3 and 5 located inside and outside of the dome, respectively (Fig. 2B). In Cycles 2 and 4, PRO abundances peaked in subsurface waters (∼40 m, 1–3 × 10⁵ cells mL⁻¹), while they were highest in the surface mixed layer in Cycle 3 and 5 (0.2–0.8 × 10⁵ cells mL⁻¹) and decayed rapidly below the pycnocline (Fig. 2B). PEUK concentrations in the surface mixed layer were similar in Cycles 2, 3 and 4 (2 to 2.5 × 10⁴ cells mL⁻¹), each showing a subsurface peak that reached concentrations up to 4.0 × 10⁴ cells mL⁻¹ just below the pycnocline (Fig. 2B). In Cycle 5, PEUK abundance was lower (∼1.0 × 10⁴ cells mL⁻¹) but represented a larger proportion of the total picophytoplankton community (Fig. 2B). Light conditions assessed as daily incident irradiance showed substantial variability among cruise days (Fig. 3). The vertical extinction coefficients, however, were similar during the different experiments providing a relatively stable depth-dependent light field for transplant experiments (Table I, Fig. 3).

Fig. 3.

Daily integrated incident irradiance (bars) measured at the sea surface with a ship-mounted radiation sensor. Black and white bars represent days within experimental cycles and navigation days, respectively. Values above bars are mean coefficients of light extinction for the euphotic zone estimated from noon profiles of photosynthetically active radiance measured with a PAR sensor mounted on a fast repetition rate fluorometer system (Goes et al., 2016). Asterisks indicate days when transplant experiments were conducted.

Table I:

Picophytoplankton growth (μ), grazing (m) and net growth (μ_net) rate estimates for the transplant experiments conducted at different depths in the Costa Rica Dome, July 2010

Experimental cycle	Depth (m)	Optical depth (%E0)	Daily integrated light (mol m−2 day−1)	T (°C)	SYN			PRO			PEUK
					μ (day−1)	m (day−1)	μnet (day−1)	μ (day−1)	m (day−1)	μnet (day−1)	μ (day−1)	m (day−1)	μnet (day−1)
Cycle 2	2	85.9	12.4	27.8	nd	nd	nd	nd	nd	nd	nd	nd	nd
Cycle 2	12	40.1	5.8	27.6	0.14	0.45	−0.31	−0.02	0.40	−0.43	0.19	0.65	−0.46
Cycle 2	20	21.8	3.2	20.5	0.19	0.20	−0.01	−0.17	−0.08	−0.09	0.08	0.20	−0.12
Cycle 2	30	10.2	1.5	17.2	0.83	0.97	−0.14	0.68	0.84	−0.16	1.28	1.52	−0.24
Cycle 2	40	4.8	0.7	16.0	0.10	0.41	−0.31	0.17	0.55	−0.38	0.11	0.60	−0.49
Cycle 2	50	2.2	0.3	14.8	−0.12	0.12	−0.24	−0.12	0.14	−0.26	−0.20	0.18	−0.38
Cycle 2	60	1.0	0.1	14.2	−0.18	0.11	−0.29	−0.14	0.15	−0.29	−0.28	−0.04	−0.25
Cycle 2	80	0.2	0.0	13.5	−0.02	0.34	−0.36	0.04	0.42	−0.38	−0.07	0.32	−0.39
Cycle 3	2	89.4	33.6	28.3	1.27	2.31	−1.05	0.53	1.69	−1.16	1.17	1.30	−0.13
Cycle 3	12	50.9	19.1	27.9	0.24	−0.50	0.74	0.70	0.75	−0.05	1.02	1.32	−0.29
Cycle 3	20	32.5	12.2	26.0	0.40	0.59	−0.19	0.43	0.44	−0.01	0.70	0.60	0.10
Cycle 3	30	18.5	6.9	19.1	0.04	0.21	−0.17	0.19	0.29	−0.11	0.18	0.22	−0.04
Cycle 3	45	8.0	3.0	15.5	0.08	0.13	−0.06	0.16	0.24	−0.09	0.15	0.13	0.02
Cycle 3	60	3.4	1.3	14.6	0.27	0.31	−0.05	0.33	0.45	−0.12	0.36	0.17	0.19
Cycle 3	80	1.1	0.4	13.9	−0.11	−0.16	0.05	−0.06	−0.10	0.03	0.04	−0.12	0.15
Cycle 3	100	0.4	0.1	13.3	0.05	−0.04	0.09	0.02	−0.10	0.12	0.44	0.16	0.28
Cycle 4	2	86.3	47.1	27.2	−2.44	0.93	−3.37	−2.34	0.79	−3.14	−0.38	0.40	−0.78
Cycle 4	12	41.3	22.5	26.5	0.28	0.48	−0.20	−0.28	0.46	−0.74	0.09	0.44	−0.35
Cycle 4	20	22.9	12.5	19.5	0.28	0.35	−0.07	0.13	0.26	−0.13	0.72	0.65	0.07
Cycle 4	30	11.0	6.0	16.3	0.38	0.47	−0.09	0.43	0.58	−0.15	0.47	0.40	0.07
Cycle 4	40	5.3	2.9	14.7	0.20	0.20	0.00	nd	nd	nd	nd	nd	nd
Cycle 4	50	2.5	1.4	14.4	0.21	0.04	0.17	0.18	0.05	0.12	0.31	0.18	0.13
Cycle 4	60	1.2	0.7	13.9	0.28	0.27	0.02	0.34	0.33	0.01	0.32	0.36	−0.05
Cycle 4	80	0.3	0.1	13.5	0.13	0.23	−0.10	0.11	0.22	−0.11	0.00	0.19	−0.19
Cycle 5	2	88.5	57.5	27.7	0.67	−0.08	0.75	−0.68	−0.24	−0.44	0.86	0.31	0.55
Cycle 5	12	48.0	31.2	27.7	0.87	0.47	0.39	−0.02	0.57	−0.59	0.52	0.02	0.50
Cycle 5	20	29.4	19.1	27.1	0.83	0.52	0.31	0.25	0.56	−0.31	0.84	0.78	0.07
Cycle 5	30	16.0	10.4	19.8	0.68	0.42	0.26	0.80	0.45	0.35	1.14	0.96	0.18
Cycle 5	40	8.7	5.6	17.5	0.26	0.44	−0.18	0.38	0.28	0.10	0.36	0.65	−0.29
Cycle 5	50	4.7	3.1	16.2	0.11	0.01	0.09	0.26	0.13	0.13	0.23	0.70	−0.48
Cycle 5	70	1.4	0.9	14.6	0.17	0.10	0.08	−0.10	−0.27	0.17	−0.20	−0.22	0.02
Cycle 5	100	0.2	0.14	14.1	0.00	−0.06	0.05	0.07	−0.12	0.19	0.08	−0.03	0.11

Optical depth (%E₀) and temperature (T) for each incubation are also given. Categories are: Synechococcus (SYN), Prochlorococcus (PRO) and picoeukaryotes (PEUK). Units are percentage of incident irradiance for optical depth, °C for temperature and day⁻¹ for growth, grazing and net growth rates. Data from the sample collection depth are in bold.

Picophytoplankton growth, grazing and net growth rates

Growth and grazing rates estimated for subsurface populations SYN, PRO and PEUK incubated across the water column are summarized in Table I and graphically represented in Fig. 4. Intrinsic growth rates estimated from incubations conducted at the same depth of sample collection were higher in Cycles 2 and 5 than in Cycles 3 and 4 for all groups (Table I, Figs 4 and 5). Overall, PEUK showed higher growth rates (µ = 0.90 ± 0.38 day⁻¹, mean ± SD, n = 4) than SYN (µ = 0.57 ± 0.22 day⁻¹) and PRO (µ = 0.58 ± 0.19 day⁻¹) (F = 8.45, P = 0.018, repeated-measures analysis of variance (ANOVA)). Grazing rates were similar or higher than growth rates in all cycles, except for Cycle 5, where growth exceeded grazing (Table I, Fig. 4). Mean group-specific grazing rates mirrored those for growth, with PEUK showing higher values (m = 0.87 ± 0.49 day⁻¹, mean ± SD, n = 4) than SYN (m = 0.61 ± 0.25 day⁻¹) and PRO (m = 0.58 ± 0.19 day⁻¹), although the differences were not significant in this case (F = 2.24, P = 0.18, repeated-measures ANOVA). Net growth rates at the depth of sample collection were slightly negative for SYN and PRO while the opposite was true for PEUK (Table I, Fig. 4).

Fig. 4.

Rates of growth, grazing and net growth of PRO, SYN and PEUK populations in transplant incubations conducted at different irradiance levels expressed as a function of percent surface irradiance in the water column. Dashed horizontal line indicates the depth where the sample water used for the preparation of the eight replicated dilution experiments was collected. Solid vertical line is located at zero rate values for reference.

Fig. 5.

Bivariable plot of grazing mortality and growth rates for each of the transplant experiment for Prochlorococcus (PRO), Synechococcus (SYN) and picoeukaryotes (PEUK). Spearman correlation coefficient (R) estimated for each group is indicated within each plot. Dashed lines represent the 1 : 1 relationships.

Growth rates of picophytoplankton groups decreased gradually when transplanted to deeper waters with lower irradiance levels (Fig. 4). This trend was consistent among all groups and experiments, although the decay was more pronounced in Cycles 2 and 5 (Fig. 4). In these cycles, growth rates measured at ∼5% E₀ were ∼5-fold lower than those at the sample collection depth (∼10% E₀) for all picophytoplankton groups, with SYN showing the strongest decrease (Table I, Fig. 4). Grazing rates for all groups decreased with decreasing irradiance levels as well, although the minimum values in the lower euphotic zone differed among cycles. While grazing rates at irradiance levels ≤1% E₀ were virtually zero in Cycles 3 and 5, they remained substantial (∼0.25 day⁻¹) for Cycles 2 and 4 (Table I and Fig. 4), which may indicate differences in the protistan grazing impact on primary production and nutrient recycling at the lowest light levels of the euphotic zone. When transplanted to shallower depths, the picophytoplankton community response was more variable among cycles and populations (Fig. 3). For Cycles 2 and 4, all populations had lower growth rates when transplanted to the surface mixed layer while the opposite response was observed in Cycle 3 (Fig. 4). In Cycle 5 experiments, however, cell growth of SYN and PEUK remained constant or increased slightly in the vicinity of the collection depth and sharply decreased at the shallowest incubation depth, whereas PRO showed decreasing growth rates in all incubations above the sample collection depth (Fig. 4). Mean growth rates estimated across transplanted depths were significantly higher for PEUK (0.29 ± 0.38 day⁻¹, mean ± SD) and SYN (0.24 ± 0.34 day⁻¹) compared with PRO (0.09 ± 0.27 day⁻¹) (Table I, F = 4.65, P = 0.014, repeated-measures ANOVA), although the 75th percentile of PEUK growth rate (µ_75th__percentil = 0.48 day⁻¹, n = 25) was about 2-fold higher than that for SYN and PRO (µ_75th = 0.28, and 0.25, respectively).

Grazing rates displayed similar vertical trends to phytoplankton growth, and both were strongly correlated either within each experiment (Fig. 5) or combining the data from all experiments in all groups (R_{spearman_PRO} = 0.572, R_{spearman_SYN} = 0.588, R_{spearman_PEUK} = 0.624). The slopes and Y-intercepts of grazing and growth model II regressions were not significantly different from 1 and 0, respectively (Fig. 6), suggesting an overall balance between both rates across all incubations. As a consequence of this coupling, net growth rates remained close to zero, although the occasional imbalance between growth and grazing led to substantial net growth rates, mainly for the shallowest incubations (Fig. 4 and 6). In Cycle 5, for instance, the enhanced growth rates of SYN and PEUK at higher irradiance levels were not matched by grazing and yielded positive net growth rates, while for Cycle 4 we observed the inverse situation in which the opposing trends of growth and grazing rates yielded net consumption (Fig. 4). In contrast to growth, mean grazing rate for PRO (0.28 ± 0.08 day⁻¹) was not significantly different than for PEUK (0.36 ± 0.08 day⁻¹) and SYN (0.27 ± 0.09 day⁻¹) (F = 1.04, P = 0.35, repeated-measures ANOVA), which further reflects the growth and grazing imbalance of PRO in the surface transplant experiments (Fig. 4).

Fig. 6.

Model II linear regressions between grazing mortality and growth rates estimated for Synechococcus (SYN), Prochlorococcus (PRO) and picoeukaryotes (PEUK) across all transplant experiments (dashed line). Regression parameters are estimated excluding shallowest incubations (solid line, cross-filled symbol). Error estimates are standard errors of the mean.

Picophytoplankton growth and grazing relationships with light and temperature

The effects of light and temperature on picophytoplankton growth, grazing and net growth rates were explored by simple linear regression analysis (Fig. 7, Table II). Growth rates of SYN and PEUK were positively correlated with irradiance levels (log %E₀) (R_{pearson_SYN} = 0.602 and R_{pearson_PEUK} = 0.588, P < 0.001) while significant correlation was absent for PRO (R_{pearson_PRO} = 0.101, P = 0.601; Fig. 7, Table II). This was also true for the ambient community (Selph et al., 2016). Growth and grazing rates of SYN and PEUK were also positively correlated with the temperature change, expressed as the difference between ambient temperature at incubation and at collection depth (T′ – T₀) (R_{pearson_SYN} = 0.602 and R_{pearson_PEUK} = 0.588, P < 0.001), but this was not the case for PRO (R_{pearson_PRO} = 0.04, P = 0.84). Conversely, grazing mortality for PRO was positively correlated with irradiance (R_{pearson_PRO} = 0.528, P < 0.01), underlying the decreasing net growth observed with increasing irradiance (R_{pearson_PRO} = −0.503, P < 0.01) (Fig. 7). Grazing rates of SYN and PEUK were positively correlated with irradiance levels and yielded regression parameters (Table II) not significantly different from the growth-irradiance relationships (analysis of covariance, P > 0.2). On the other hand, net growth rates of SYN and PEUK varied closely around zero despite dramatic changes in ambient irradiance and temperature during the transplanted incubations, yielding non-significant relationships with irradiance and temperature (Table II, Fig. 7).

Fig. 7.

Simple linear regressions of growth (solid circles), grazing (empty circles) and net growth (triangles) rates against incubation irradiance levels (log %E₀) estimated for Synechococcus, Prochlorococcus and picoeukaryotes. The slope (±standard error of the mean) of the regression and the correlation coefficient are shown.

Table II:

Functional relationships between picophytoplankton growth and grazing mortality rates and light

Phytoplankton group	Dependent variable	Independent variable	Slope ± SEM	Y-Intercept ± SEM	r2	P-value
Synechococcus	μ	%E0	0.27 ± 0.06	0.070 ± 0.069	0.4073	0.0001
	m		0.22 ± 0.10	0.13 ± 0.12	0.1366	0.0444
	μnet		0.05 ± 0.08	−0.069 ± 0.09	0.0143	0.5366
Prochlorococcus	μ	%E0	0.048 ± 0.07	0.12 ± 0.08	0.0147	0.5304
	m		0.23 ± 0.08	0.14 ± 0.09	0.218	0.0107
	μnet		−0.18 ± 0.07	−0.02 ± 0.07	0.207	0.0131
Picoeukaryotes	μ	%E0	0.33 ± 0.08	0.11 ± 0.09	0.3671	0.0005
	m		0.29 ± 0.09	0.20 ± 0.10	0.2833	0.003
	μnet		0.038 ± 0.07	−0.090 ± 0.074	0.01233	0.5663

Results of simple linear regression analysis on the growth (μ), grazing (m) and net growth (μ_net) rate estimates for Synechococcus, Prochlorococcus, and picoeukaryotes using the percentage of incident irradiance (%E₀) as independent variable. Irradiance data were log₁₀ transformed prior to analysis. Error represents the standard error of the mean (SEM, n = 30).

Nano- and microphytoplankton net growth dynamics

The biomass of larger phytoplankton (>8 µm) increased in shallower incubations, although responses to the incubation conditions differed among functional groups (Fig. 8A). Autotrophic dinoflagellates for instance, seemed to do best when incubated at the depth of collection, while transplants to different light and temperature decreased their growth (Fig. 8A). Conversely, net accumulations of autotrophic flagellates and especially diatoms were favored by conditions prevailing in shallower water incubations (Fig. 8A). This trend was consistent among all cycles, and shifted the community structure of surface incubations towards higher contributions of larger phytoplankton (Fig. 9). Maximum biomass levels in transplant incubations were observed in experiments conducted outside the dome (Cycles 3 and 5, Fig. 8A), particularly in Cycle 5 where diatoms took over the community at the shallowest incubation depth. In experiments conducted within the dome (Cycles 2 and 4), the accumulation of diatoms was less marked, although the modest increase of larger phytoplankton observed in Cycle 4 (Fig. 8A) was clearly reflected in community structure after 24 h of incubation due to strong photoinhibition of picophytoplankton in shallow waters.

Fig. 8.

Phytoplankton carbon biomass (µg C L⁻¹) and cell abundance at the end of the incubations in the non-diluted treatment (A–C) and relative change in carbon biomass between sample incubation and collection depths (D). Diatoms, autotrophic flagellates (AF), autotrophic dinoflagellates (ADinos).

Fig 9.

Relative contribution of major phytoplankton groups to total autotrophic community carbon biomass in the non-diluted treatment at the end of the incubations. Synechococcus (SYN), Prochlorococcus (PRO), picoeukaryotes (PEUK), diatoms, autotrophic flagellates (AF), autotrophic dinoflagellates (ADinos). Arrows indicate the original depth of sample collection.

A closer look into the size structure of diatoms in three size fractions (10–20, 20–40 and >40-µm) showed that the intermediate fraction had the highest abundance in surface incubations (Fig. 8B). However, the >40-µm diatom fraction contributed most to biomass accumulation (Fig. 8C). Normalization of the diatom biomass in transplanted incubations relative to collection depth incubations showed that the stimulation of net growth rate was strongest, in relative terms, for Cycles 2 and 5 (9-fold). Biomass of the 20- to 40-µm size fraction often showed a disproportionate response of up to a 25-fold increase (Fig. 8D), a somehow counterintuitive result given the dominant contribution of the larger fraction to total diatom biomass.

DISCUSSION

Experimental approach and potential artifacts

Because light is essential for phytoplankton photosynthesis, light availability is recognized as a key factor affecting the dynamics and structure of planktonic communities (Huisman et al., 1999a, b). Besides being a resource for primary producers, there is evidence that light can also affect protistan grazing activity (e.g. Strom, 2001) and, therefore, impact population dynamics and community structure. The objective of this study was to analyze the variability of phytoplankton community growth and grazing dynamics in response to irradiance changes associated with the CRD upwelling. In an attempt to simulate physico-chemical changes experienced by the phytoplankton and microzooplankton community during the seasonal shoaling and mixing of thermocline waters, we transplanted the microbial community from the subsurface chlorophyll maximum to shallower and deeper layers, and estimated the daily rates of growth, grazing and net accumulation of major phytoplankton groups from in situ dilution assays. This study circumvents at least some of the problems associated with on-deck incubation experiments by using a pseudo-Lagrangian drift array (Landry et al., 2009) that allowed us to incubate simultaneously eight dilution assays containing the same initial microbial community across a gradient of irradiance conditions spanning the euphotic zone (Table I). Although we did not evaluate the experimental variability of the two-treatment dilution assay in the current study, previous experience has shown the relatively low standard deviation for growth (0.07 day⁻¹) and grazing (0.10 day⁻¹) rates estimated from independent triplicate experiments (Landry et al., 2008). The coherence of estimated process rates within each cycle experiment (Fig. 4) and the consistency with growth and grazing rates estimated independently at the collection depths (Landry et al., 2016b; Selph et al., 2016) also support the validity of this experimental approach.

The highly stratified waters and vertically structured phytoplankton community of the CRD upwelling region (Gutiérrez-Rodríguez et al., 2014, Fig. 2) provide a relatively stable gradient of depth-dependent light exposures, which minimizes photoacclimation issues associated with fixed-depth incubations (Table I). In this sense, the similar vertical extinction coefficients measured in the different experiments (Fig. 3) facilitate comparison among transplant experiments. To minimize complications inherent to pigment-based rate estimates, we assessed population dynamics from changes in cell abundance and biomass measured by FCM and epifluorescence microscopy. Vertical profiles of fluorescence per cell volume quantified by FCM (FL3/SSC) (Fig. 10) indicate that picophytoplankton cells adjusted their pigment contents to changes in light exposure during the 24-h transplant incubations. Cell pigment contents decreased in samples transplanted to shallow depths, remained relatively constant in samples incubated at the depth of collection and remained relatively constant or decreased in deeper water incubations (Fig. 10). These changes are largely consistent with the physiological responses expected for acclimation of light-limited phytoplankton transferred to increasing photon flux density dose, and support the validity of the experimental design to investigate phytoplankton population dynamics under a realistic range of physiological conditions.

Fig. 10.

Vertical profiles of the ratios between red fluorescence per side scatter (FL3/SSC) at the sample collection and incubation depths at the end of the experiment shown as an index of strength of photoacclimation occurring during the 24-h transplant incubations.

Picophytoplankton growth and grazing response to environmental gradients

We hypothesized that growth rates of phytoplankton transplanted to shallower waters would be enhanced by favorable light conditions that allowed the community to fully exploit the higher nutrient concentrations carried upward from subsurface waters. The significant positive linear relationship between growth rate and irradiance observed for SYN and PEUK (Fig. 7, Table II) is consistent with this hypothesis. The absence of such relationship for PRO is mainly due to high-light depression of growth rate observed at shallow depths (Fig. 4). This trend has been previously observed from both high pressure liquid chromatography and FCM-based growth rate estimates in the equatorial Pacific (Selph et al., 2011) and is also consistent with the higher cell death observed for PRO compared with SYN and PEUK at high irradiance levels (Llabrés and Agustí, 2006). The persistence of the growth decaying trend across the mixed layer (ΔT < 0.125°C) (e.g. Cycle 4) and the higher sensitivity of PRO to increasing irradiance compared with SYN and PEUK (Figs 3 and 9) (Partensky et al., 1999; Llabrés and Agustí, 2006; Scanlan et al., 2009) point toward excess light as the likely cause for the lack of systematic growth stimulation in shallower incubations (Fig. 5). The growth photoinhibition that we observed for subsurface PRO populations transplanted to shallower waters is consistent with what we know about HL- and LL-adapted PRO ecotypes (Moore et al., 1998; Rocap et al., 2003; Biller et al., 2015) and would suggest the genotypic and phenotypic distinction of PRO populations in the CRD (Gutiérrez-Rodríguez et al., 2014).

In addition to enhancing phytoplankton growth through photosynthesis, light can also stimulate digestion, grazing and growth of phagotrophic protists feeding on pigmented prey (Strom, 2001). In our dataset, grazing rates on all picophytoplankton groups were significantly and positively related to irradiance (Fig. 7), supporting the hypothesized effect of light on community grazing. Although the initial and 24-h final samplings of our experiments preclude evaluation of day–night rhythms in grazing, the positive relationship between grazing and light levels is in agreement with the higher protistan grazing activity during the day compared with night reported in experiments conducted with natural (Dolan and Simek, 1999; Kuipers and Witte, 2000) and cultured grazer populations (Jakobsen and Strom, 2004). On the other hand, the decreasing grazing mortality observed in deeper incubations seems consistent with the systematic lower grazing mortality estimated from dilution experiments incubated at lower light conditions (Gutiérrez-Rodríguez et al., 2009). In contrast with these observations, a recent mesocosm experiment conducted in eastern Mediterranean oligotrophic waters failed to detect systematically enhanced microzooplankton grazing in high-light (60% E₀) compared with low-light conditions (6% E₀) (Calbet et al., 2012), suggesting that other factors such as trophic structure and interactions (e.g. mixotrophy) could affect the light dependence of grazing.

Close coupling between growth and grazing

In our dataset, growth and grazing rates are strongly correlated both within experiments (Fig. 5) and picophytoplankton group, indicating that they remained tightly coupled along the gradient of conditions explored (Table I) despite large variability in process rates (Fig. 5). Growth and grazing rates estimated from independent dilution experiments are often positively correlated, with phytoplankton groups showing higher growth rates being usually consumed at higher rates as well (Burkill et al., 1987; Strom and Welschmeyer, 1991; Latasa et al., 2005). The lack of independence between growth and grazing in the ecological model assumed by the dilution method can contribute to this correlation (Gutiérrez-Rodríguez et al., 2009). In our dataset, however, the growth and grazing correlations associated with this methodological issue (R_{spearman_methodological} = 0.12 ± 0.19, 0.18 ± 0.18 and 0.10 ± 0.19 for PRO, SYN and PEUK, respectively) were significantly lower than the correlations observed (R_{spearman_PRO} = 0.572, R_{spearman_SYN} = 0.588, R_{spearman_PEUK} = 0.624). The strong ecological link between growth and grazing rates suggested by these results is consistent with the close coupling observed between picophytoplankton growth and mortality rates (Chen et al., 2009; Cáceres et al., 2013) and further highlights the highly dynamic nature of this coupling (Strom, 2002). In this sense, despite the significant relationship between grazing and irradiance (r² ∼ 0.3, Fig. 7), it is worth noting that growth rate itself remains the best predictor of grazing rates for picophytoplankton (r² ∼ 0.5, Figs 5 and 6). During the same cruise, Freibott et al. (Freibott et al., 2016) found remarkably constant depth-integrated biomass of heterotrophic protists in the region, and although this biomass was significantly related to the carbon consumed by microzooplankton grazing, a large fraction of this variability remained unexplained by the biomass of grazers alone.

The high correlation between growth and grazing rates makes it difficult to distinguish between the direct (e.g. “light-aided” digestion) and indirect effects that light may exert on grazing rates through internal regulation (i.e. higher grazing on faster growing prey). In experiments where the processes are clearly decoupled due to photoinhibition of phytoplankton growth, grazing tends to increase with increasing irradiance, particularly for PRO (e.g. Cycle 4, Fig. 5), supporting a direct link between light and community grazing. Laboratory experiments performed with Oxyrrhis marina and Dunaliella tertiolecta growing on a two-stage chemostat system have shown significantly higher prey content and slower vacuole evacuation rates in cultures maintained in the dark compared with light (Décima and Gutiérrez-Rodríguez, unpublished data), suggesting a direct effect of light upon digestion. Taken together, these laboratory and field results support the role of light not only as a resource for phytoplankton growth but also as an important factor modulating grazing mortality dynamics. Regardless of the mechanism, the rapid response of protistan grazers to sudden environmental perturbations affecting picophytoplankton growth highlights the resilience of this trophic coupling and its role in stabilizing planktonic ecosystems.

Nano- and microphytoplankton net growth response and community structure

The scarcity of larger cells, particularly in the diluted treatment, precludes robust determination of microphytoplankton growth rates from standard microscopy counts. Yet, comparisons of cell abundance and biomass in the non-diluted incubations across the different incubation depths allows us to assess net responses of the microphytoplankton community to irradiance changes associated with the transplant incubations. These results show an accumulation of larger phytoplankton, mainly diatoms, at shallower incubation depths (Figs 8 and 9), in agreement with the capacity of diatoms to respond quickly to sudden increase of resource supply (Sarthou et al., 2005). Given that nutrient concentrations were the same across all replicated incubations (i.e. the same water sample), the rapid accumulation of diatoms in high-light (HL) transplanted incubations suggest suboptimal light conditions for diatom growth in the deeper waters. Only after exposure to higher irradiance levels could diatoms fully exploit their physiological and maximum growth capacities (Furnas, 1991; Cermeño et al., 2005b), producing a major shift in community structure in shallow incubations after only 24 h (Fig. 9). This observation agrees with results from grow-out experiments that show enhanced response of phytoplankton populations to nutrient enrichment (silicate, Fe and Zn) under high-light conditions (Goes et al., 2016). In both cases, the response of diatoms to increasing light seem to reflect the disadvantage of larger cells under reduced light conditions due to the size dependence of the package effect (Finkel, 2001; Mei et al., 2009) and the subsequent alleviation of their light absorption constraints under higher irradiance. According to the theoretical framework developed by Huisman et al. (Huisman et al., 1999a), critical light intensity ($I_{\mathrm{out}}^{\ast }$) is a key feature of light competition that allows species with lower $I_{\mathrm{out}}^{\ast }$ to outcompete those with higher values in well-mixed environments. At low mixing rates, however, this rule does not hold because the relative position in the light gradient can favor the growth of a species at a physiological disadvantage by their ability to shade and outcompete the species below (Huisman et al., 1999b). Under low turbulent mixing conditions such as those prevailing in the CRD, the competition model predicts sharp transitions between the dominance of different species, which is consistent with the dramatic increase of diatoms and resulting changes in community structure observed in shallow transplant incubations (Figs 8 and 9). A growing body of evidence suggests that the unusual community structure of the CRD upwelling results from the unique trace-metal conditions of the region (Franck et al., 2003; Saito et al., 2005), including higher cobalt concentrations and metal complexation (Ahlgren et al., 2014) and Zn depletion (Chappell et al., 2016). Nonetheless, the proliferation of diatoms in well-lit shallow transplant incubations compared with those kept at low-light (LL) in deeper waters, having both the same initial nutrient availability (Fig. 8), suggests that competition for light, in addition to trace-metal chemistry, plays a significant role in modulating phytoplankton growth and community structure in the CRD.

Culture experiments have shown an increase in cellular iron requirement for phytoplankton under low-light conditions, allowing for simultaneous limitation of phytoplankton growth by iron and light (Sunda and Huntsman, 1997, 2011). In natural systems, iron and light co-limitation has been described for surface phytoplankton populations in the Southern Ocean and high-latitude Pacific Ocean with mixed layers extending beyond the euphotic zone (Mitchell et al., 1991; Boyd et al., 1999; Smith et al., 2000), as well as for subsurface populations forming the deep chlorophyll maximum over a wide region of the Pacific (Hopkinson and Barbeau, 2008; Johnson et al., 2010). Using microcosm manipulation experiments, Hopkinson and Barbeau (Hopkinson and Barbeau, 2008) observed that the response of diatoms to iron enrichment was strongest under high-light conditions, while iron addition at ambient low-light levels was often not sufficient to trigger significant changes in phytoplankton biomass and composition. Similarly, Johnson et al. (Johnson et al., 2010) found a positive and comparable response of diatoms to both “+light” and “+light + iron” treatments in grow-out experiments conducted with natural phytoplankton assemblages sampled from the DCM at several locations across the equatorial and subtropical Pacific. The accumulation of diatoms in high-light transplant incubations is consistent with these experimental observations and further supports the central role of light in determining the dynamics and composition of subsurface populations in well-stratified systems.

In addition to the nutrient fertilization of shallower waters, the doming of the isopycnals that defines the CRD upwelling can potentially represent a light-enriching mechanism (sensu Diehl et al., 2002). Yet, the diatom accumulation generally observed in our light-enriched incubations is at odds with the picophytoplankton-dominated community that develops in the CRD where diatoms remain rare members despite the upwelling nature of the system (Li et al., 1983; Saito et al., 2005; Taylor et al., 2016). Results of extensive studies on phytoplankton control mechanisms in the eastern equatorial Pacific (EEP) may help to reconcile these differences between experimental and in situ observations. The EEP is a high-nutrient low-chlorophyll area where low levels of iron prevent the growth of larger diatoms (Coale et al., 1996b; Landry et al., 1997). A recent comprehensive study of phytoplankton structure and dynamics across the iron-limited EEP showed that iron supply associated to the westward shoaling of the pycnocline stimulated PRO rather than diatom biomass (Selph et al., 2011). This result contrasts with the diatom-dominated community response in bottle and in situ mesoscale experiments in this (Chavez et al., 1991; Martin et al., 1991; Coale et al., 1996a, b) and other putative iron-limited oceanic regions such as the subartic Pacific (Welschmeyer et al., 1991) and the Southern Ocean (Boyd et al., 2007; Smetacek et al., 2012). Selph et al. (Selph et al., 2011) argued that the low-light conditions prevented diatoms from utilizing new nutrients entering the lower euphotic zone, which were instead consumed by small phytoplankton before they reached the well-lit shallower waters. Similarly, our transplant experimental design implies a different resource supply scenario for phytoplankton in which relatively nutrient-rich subsurface waters are suddenly exposed to high-light conditions where diatoms can fully exploit their high-nutrient uptake and fast-growth capacity before small picophytoplankton exhaust the limiting nutrient. These results support the inability of diatoms to compete with picophytoplankton for the limiting nutrient (i.e. cobalt or zinc) under low-light conditions, and stress the importance that resource supply mechanics associated with different fertilization mechanisms may have on the community response.

Mesoscale and sub-mesoscale physical features such as fronts, eddies and internal waves are known to affect the growth environment for phytoplankton as well as system trophic status (Benitez-Nelson and McGillicuddy, 2008). Our results show that protistan grazing may also respond to light changes caused by vertical displacements of isotherms associated with such physical events, influencing the fate of the primary production stimulated by the same process. These experiments provide further evidence of the robustness and highly dynamic nature of the close coupling between picophytoplankton growth and grazing dynamics, which remained tight despite perturbation along strong environmental gradients. In addition, accumulation of larger phytoplankton, and particularly diatoms, was systematically enhanced in well-lit shallow incubations illustrating the light-limited conditions of subsurface populations and the capacity of larger cells to bloom when that limitation is lifted suddenly. Nonetheless, the surface phytoplankton community that develops naturally in response to summer upwelling conditions of the CRD is dominated by small phytoplankton. The different community responses to manipulative experiments compared with natural fertilization processes suggest that the “sudden” versus “gradual” pairing of resource availability (i.e., light and nutrient) may be a key factor influencing changes in the size and taxonomic structure of the phytoplankton community that can develop in response to transient physical features.

FUNDING

This study was supported by U.S. National Science Foundation grant OCE-0826626 to M.R.L. and by a Ramón Areces Foundation Fellowship to A.G.R.