Seasonal mixed layer depth shapes phytoplankton physiology, viral production, and accumulation in the North Atlantic

Mixed layer depth and phytoplankton accumulation dynamics in the North Atlantic

The NAAMES expeditions intensively measured biological, chemical, and physical properties from 4 to 7 locations, or stations, in each bloom phase during November (Winter Transition), March−April (Accumulation), May (Climax; same as Climax Transition²²), and September (Decline)²². Stations spanned a broad range in latitude (~37 °N to ~55 °N, Fig. 1a), sub-regional classifications (Gulf Stream and Sargasso Sea, Subtropical, Temperate and Subpolar)²⁴, and MLDs (tens to hundreds of meters) (Fig. 1b and Supplementary Fig. 1). MLDs were calculated using a density difference threshold of 0.03 kg m⁻³ from the top 10 m²⁵. Field data and associated analyses are derived from phytoplankton 1–20 µm in diameter and their associated communities sampled within the photic zone (40, 20, 1% surface irradiance) and within the mixed layer, unless otherwise noted.

Predominantly positive rates of phytoplankton accumulation were observed during three of the four phases of the annual bloom cycle (Fig. 1c, d and Supplementary Fig. 2b), including Winter Transition, consistent with prior observations^6,26. The Decline phase had the lowest accumulation rates and the highest proportion of net phytoplankton loss rates (Fig. 1c, d). Accumulation rates were statistically indistinguishable (p > 0.05, Kruskal−Wallis) between populations collected from 5 m in-line sampling throughout the day (in situ) and contemporaneous incubations of the same phytoplankton populations under simulated in situ irradiance and temperature (incubations; see ‘Methods’) (Fig. 1c, d). Accumulation rates using incubations calculated via cell concentration or via biovolume were not statistically different (Supplementary Fig. 2b).

Phytoplankton cell concentration and biovolume generally increased with water column stability (stratification), during the Winter Transition, Accumulation, and Climax phases (Fig. 1e and Supplementary Fig. 2c). Stratification was quantified by the buoyancy frequency averaged over the upper 300 m of the water column (see ‘Methods’). Higher values of buoyancy frequency indicate a more stratified water column where exchange with nutrient-rich water below the surface is reduced. Strongly stratified water columns (buoyancy frequencies above 2 × 10⁻⁵ s⁻¹) during the Decline phase were associated with lower cell concentrations (Fig. 1e), consistent with enhanced phytoplankton loss or reduced accumulation. Phytoplankton biovolume and cell size distribution within 1–20 µm-sized phytoplankton cells increased during the Decline phase (Supplementary Fig. 2c–e). These higher biovolumes could have been a result of changes in community composition. They could have also been attributed to aggregation caused by virus infection^20,21,28, as virus concentrations were highest during this season (discussed below), or by light stress²⁷, as mixed layer populations were more consistently exposed to daily higher irradiance levels characteristic of shallow mixed layers (Fig. 1e).

In situ phytoplankton cell concentrations increased from Winter Transition until the Climax phase, from ~1 × 10⁶ to 2.5 × 10⁷ cells L⁻¹ (Fig. 2a, c, gray boxes). On-deck incubations showed similar trends but had higher overall cell concentrations (Fig. 2a, c, white boxes). The Decline phase was characterized by a 4-fold reduction in median phytoplankton cell concentrations from the peak abundances observed during Climax phase (Fig. 2a, c). The stress markers utilized in this study provided a unique view into the physiological status of communities across these annual bloom phases (Supplementary Table 1). Our ROS and compromised cell membranes biomarkers specifically targeted eukaryotic phytoplankton, given the conditions used for flow cytometry analysis (see ‘Methods’). PCD-related proteases and lipids were extracted from biomass collected onto 1.2 and 0.2 µm diameter membrane filters, respectively. Consequently, these biomarkers could also include eukaryotic heterotrophs and bacteria in the system. Induction of caspase and metacaspase activities have been found in diverse phytoplankton, such as coccolithophores, diatoms, chlorophytes, nitrogen-fixing cyanobacteria, and dinoflagellates cells undergoing stress, senescence, and death²⁹. They have also been reported in stressed or dying grazers³⁰, although no marine species has been explicitly studied. TAGs are found mainly in marine eukaryotic phytoplankton^31,33,33 and grazers³⁴. The highly unsaturated fatty acids in the PC and OxPCs detected in our measurements are also indicative of eukaryotic organisms, and not marine cyanobacteria³² or heterotrophic bacteria³⁵.

Collectively, these markers provide a broad interpretive context of the physiological state of the microbial system in response to changes in mixed layer depths throughout the North Atlantic bloom. We point out that some of our metrics (e.g., ROS, lipid biomarkers, PCD proteolytic enzymes) are not individually diagnostic of a particular stress or death process, such as virus infection, nutrient limitation, or photo-damage. Instead, they represent core cell stress and death processes that are shared among biotic- and abiotic-stressors and are explicitly tied to cell fate, making them useful diagnostic indicators of physiological changes associated with bloom communities in relation to MLD and stratification.

Populations in the Accumulation phase had low ROS and few compromised cell membranes, indicative of healthy cells (Fig. 2b, d). Conversely, populations in the Climax phase were characterized by much higher ROS levels, diagnostic of oxidative stress (Fig. 2b). ROS production in phytoplankton has been linked to light stress^36,37, nutrient limitation^38,39, and virus infection¹³. The elevated ROS signatures during Climax phase were likely not attributable to macronutrient limitation (Supplementary Fig. 3). Nutrient levels throughout NAAMES differed based on sub-regional water type, with Subpolar stations having the highest concentrations in the Climax phase (NO₃+NO₂ > 5 µM; PO₄ > 0.4 µM). Notably, nutrient concentrations during the Climax phase were similar or higher than those observed for Accumulation phase samples, which had lower ROS signatures (Fig. 2b).

Phytoplankton cells in the Decline and Winter Transition phases had a higher percentage of compromised cell membranes, reaching levels as high as 80% (Fig. 2c, d). Both late stage viral infection and PCD have been linked to high levels of compromised membranes^13,29. The percentage of phytoplankton cells with compromised membranes was used to calculate concentrations of live and dead cells within the mixed layer across the bloom phases. Living phytoplankton cell concentrations generally increased from the Winter Transition through the Climax phase (Fig. 2e). The variability of dead cells was highest in the Decline phase, which also had the largest variation in total, living, and dead cell concentrations (Fig. 2c, e, f).

Targeted analysis of OxPC, and TAGs in resident phytoplankton communities provided further context of changes in physiological states due to their relevance in cellular stress and loss processes. The seasonal bloom phases were characterized by distinct levels of these lipids (Fig. 3 and Supplementary Fig. 4). OxPC levels were highest in the Climax phase (Fig. 3a), where mixed layers had recently shallowed (Fig. 1b) and were concomitant with high intracellular ROS levels (Fig. 2b). Subcellular environments lacking in adequate antioxidant capacity are expected to accumulate OxPC⁴⁰ particularly when a shallow mixed-layer enhances UV exposure¹⁵. Chlorophyll-normalized TAG was highest in the Decline phase (Fig. 3b), which also had the lowest accumulation rates (Fig. 1c, d). High cellular TAG levels have been observed in senescent^41,42 or nutrient limited⁹ diatoms, and virus infected haptophytes⁴³.

Caspase and metacaspase catalytic activity in community protein extracts provided additional subcellular diagnostics of stress and PCD activation²⁹. Unlike metacaspase activity, which was detected at all depths and stations sampled, the incidence of caspase activity was lowest in the Accumulation phase (20%), increased in prevalence through the Climax (58%) and Decline (86%) phases, and then remained relatively high into the Winter Transition (59%) (Fig. 3c, d; pie graphs). Caspase-specific activities showed a similar pattern, being lowest in the Accumulation phase and highest in the Decline and Winter Transition phases (Fig. 3c). Caspase activity has been extensively linked to viral infection and autocatalytic PCD pathways in diverse eukaryotic phytoplankton^29,44,45. Metacaspase-specific activities were highest in the Accumulation and Decline phases (Fig. 3d), which supports reported associations with normal cell function⁴⁶, nutrient stress, and viral infection^46,47 in unicellular eukaryotic phytoplankton.

Stress, death, and extracellular signatures of loss and removal

Given that TEP accumulation has an established mechanistic link with ROS and PCD activation in diverse marine phytoplankton²⁹, we analyzed concentrations of TEP in each phase of the annual biomass cycle. Both TEP L⁻¹ and TEP cell⁻¹ were highest during the Climax phase (Fig. 4a and Supplementary Fig. 5a), consistent with elevated cellular ROS levels (Fig. 2a, b) and a weakly stratified water column (Fig. 1e). TEP production was also highest in the Climax phase (Supplementary Fig. 5b), indicative of nutrient stressed or viral infected diatoms^48,49, chlorophytes⁵⁰, and haptophytes^21,28,49. As cells⁵¹ and viruses⁵² in high TEP environments have altered aggregation, sinking, and ascending⁵³ potentials, their physical removal may be altered in this phase.

Coincident with elevated virus concentrations was an increase in seasonally accumulated dissolved organic carbon (DOC_SA). Viral lysis is an ecosystem process that leads to DOC accumulation^50,54,55 and can stimulate bacterial accumulation⁵⁶. Virus and DOC_SA concentrations were greatest during the Decline phase and varied by ~1000% and ~100%, respectively, across the annual cycle (Fig. 4c, e). We note that viruses and bacteria are operationally included in DOC measurements (see ‘Methods’). However, calculations using conversions for average carbon quotas for virus particles (0.2 fg⁵⁷) and bacterial cells (12.4 fg⁵⁸) and their respective concentrations confirmed negligible contributions to total DOC and DOC_SA concentrations (Supplementary Fig. 6; see ‘Methods’). Virus and DOC_SA concentrations were both positively correlated with water column stability and stratification (Fig. 4d, f)⁵⁹. The dramatic increase in virus production from the Accumulation to Decline phases corroborates high levels of cell death in phytoplankton (Fig. 2c, d, f) and may indicate virus-induced losses as a driver of low phytoplankton accumulation rates observed in highly stratified water columns⁶⁰ (Fig. 1c, d).

The high ROS levels in the Climax phase (Fig. 2a, b) may have been linked to virus infection, since positive virus accumulation and elevated virus cell⁻¹ started in this phase of the bloom (Supplementary Fig. 5b, e). Our flow-cytometry-based method (Supplementary Fig. 7) could not differentiate between viruses (and virus-like particles) derived from phytoplankton, heterotrophic bacteria or grazers. It could also not discern if diatoms, which were abundant during the Accumulation and Climax phases^61,62, were producing small ssRNA- or ssDNA-based viruses (20–40 nm diameter)³⁹, for which flow cytometry-based techniques have not been developed to our knowledge. Instead, they require techniques targeting RNA-dependent RNA polymerase (ssRNA) and replicase (ssDNA) genes, which were not performed in this study. Consequently, our flow cytometry-based data paint a partial picture of how virus production changed with mixed layer depths and across the seasonal North Atlantic bloom.

The enrichment of phytoplankton in the surface layer during Winter Transition (Supplementary Figs. 8a and 9a), absent of bacteria and virus enrichment, indicates a decoupling with these microbes during this phase. As the bloom progressed into the Climax and Decline phase, bacteria (Supplementary Figs. 8b and 9b and Supplementary Table 2) and then viruses and DOC became enriched in the upper 25 m (Supplementary Figs. 8c, d and 9c, d). Viral accumulation in the relatively shallow MLDs of Climax and Decline phases (Supplementary Fig. 5e) may set up an ecosystem where predator−prey encounters are amplified. An ecosystem with high viral lysis can lead to elevated DOC accumulation⁵⁰. In contrast, bloom phases with deeper MLDs (Winter Transition and Accumulation) had significantly lower virus cell⁻¹ (Supplementary Fig. 5b) with no surface enrichment of viruses (Supplementary Figs. 8c and 9c); the Accumulation phase also had no surface enrichment of DOC (Supplementary Figs. 8d, 9d).

We related all biomarkers and physical parameters within the mixed layer to each other across bloom phases using cluster analysis with an Optimal Leaf Ordering algorithm (Fig. 5; see ‘Methods’). Each variable/biomarker was normalized to its own distribution (log-transformed distribution for biomarkers already transformed in Figs. 2–4) across cruises on a scale of 0 to 1 prior to clustering since the scales and units of individual measurements were fundamentally different. The Accumulation phase had relatively low values of stress- and cell death-associated biomarkers and possessed relatively high accumulation rates, metacaspase activity, and storage lipids. As water columns became more stratified during the Climax phase, both phytoplankton cell concentration and biomass peaked, and oxidative stress, oxidized lipids, and TEP cell⁻¹ were most prominent. Stable and shallow mixed layers during the summer/autumn Decline phase brought peak signals in storage lipids, virus cell⁻¹, PCD activity, and DOC_SA. Percent compromised membranes peaked into the Winter Transition phase, but the deeper mixing coincided with generally lower stress markers and lower virus cell⁻¹, setting the stage for subsequent positive accumulation (Fig. 5).

These biomarker patterns also held true for populations collected across bloom phases and geographical subregions (Fig. 6). We noted that individual stations within each bloom phase clustered together more than with other stations from similar latitudes and subregions but in different bloom phases (Fig. 6). Indeed, through the lens of these biomarkers, populations within a bloom phase, but deriving from different subregions, resembled each other more than other stations across other bloom phases. This analysis revealed that some biomarkers, such as virus cell⁻¹, OxPC, caspase activity, and TEP cell⁻¹, clustered more strongly within each bloom phase; other biomarkers such as ROS and metacaspase activity appeared to vary within and between bloom phases (Fig. 6). This implies that some biomarkers represent a seasonal community physiological state that is conserved in mixed layers throughout our study area, while other biomarkers are more sensitive to more subtle bloom timings. Similarly, we noted that certain biomarkers clustered into distinct groups throughout the seasonal bloom. For example, compromised membranes, virus cell⁻¹, caspase activity, and DOC_SA consistently clustered with degree of stratification (Fig. 6).

Principal component analysis revealed the most pronounced separation in collective physiological markers during Climax and Decline phases (Fig. 7a, b). Recognizing that the biomarker signatures are derived from distinct phytoplankton communities across bloom phases and subregions, our analysis integrated both pigment-based⁶² and 16S rDNA-based⁶¹ community analyses to infer taxa identities. Incorporating either community-based analysis, our biomarkers and taxa contributed similarly to covariance in PC1 (Fig. 7c, d). On a bloom-wide scale, stratification positively covaried with cyanobacteria, caspase activity, virus cell⁻¹, DOC_SA, TAG ChlA⁻¹ and compromised membranes. These variables negatively covaried with macronutrients, TEP cell⁻¹, ROS, OxPC, and diatoms (Fig. 7c, d). Our analysis suggests that diatoms (Climax) and cyanobacteria (Decline) were enriched in fundamentally different ecosystem/bloom states. It uniquely links these phytoplankton community types to specific cell signatures such as ROS, TEP, cell death, DOC, viruses, and stress lipids.

We further examined biomarker relationships within Climax (Supplementary Fig. 10) and Decline (Supplementary Fig. 11) phases to better understand how biomarkers and community types covaried by bloom phase and stations. TEP cell⁻¹, OxPC, and ROS did not covary with nutrient stress (Supplementary Fig. 10c, d) within the Climax phase. TEP cell⁻¹ negatively covaried with stratification (Supplementary Fig. 10c, d), indicating that high turbulence⁶³ may contribute to elevated TEP cell⁻¹. ROS did not strongly covary with any biomarker or community type (Supplementary Fig. 10b, d), implying that ROS was a general stressor experienced at all subregions, and possibly linked to UV stress. ROS negatively covaried with nutrients in the Decline phase (Supplementary Fig. 11b, d) suggesting that here nutrient limitation may have contributed to higher ROS. Overall, metacaspase activity was not strongly covarying with any one biomarker or group in our analyses (Fig. 7c, d and Supplementary Figs. 10, 11), which is supportive of their reported involvement in a variety of healthy or stress-related functions unique to different taxa^7,46.

Integrating community composition revealed some associations between biomarkers and community type. TEP cell⁻¹ notably clustered with communities in which diatoms were in higher relative abundance by either pigment- or 16S rRNA-based community composition analysis (Fig. 7c, d and Supplementary Fig. 10c). Some diatom species are known to have significantly higher TEP cell⁻¹ compared to other phytoplankton⁴⁹, so they may have been a source of TEP. 16S-based community composition showed that Micromonas and Cryptophyceae also positively covaried with TEP cell⁻¹, implying that a community type, rather than one group of phytoplankton, was associated with high TEP cell⁻¹. Although cyanobacteria positively covaried with compromised membranes, TAGs, caspase activity and viruses in the highly stratified waters of the Decline phase (Fig. 7 and Supplementary Fig. 11), they were likely not the source of these biomarkers for several reasons. First, cyanobacteria did not positively covary with compromised membranes and caspase activity; likewise, they only weakly positively covaried with virus cell⁻¹ (Supplementary Fig. 11c, d). Second, our compromised membrane detection procedure did not include cyanobacteria due to cell size discrimination (Supplementary Table 1). Third, TAGs are not known to be a major component of marine cyanobacteria biomass³². They are more likely the product of eukaryotic phytoplankton^42,43 or zooplankton metabolism³⁴ in a nutrient limited environment. Fourth, caspase/metacaspase orthologs have not been identified in the genomes of unicellular cyanobacteria, such as Prochlorococcus and Synechococcus²⁹. Lastly, lytic virus production and cell death would mechanistically reduce host cell concentrations; we instead observed that higher total virus concentrations positively covaried with cyanobacteria populations. Additional size fractions, high speed cell-sorting, and/or single-cell analyses, combined with nucleic acid-based techniques would provide finer resolution on which taxa experienced specific physiological changes and paint a more comprehensive picture of virus infection dynamics throughout the North Atlantic bloom.

Physiological responses to mixing and stratification on shorter time scales

We further explored our biomarker data between stations occupied during NAAMES as case studies for community physiological responses to different mixing scenarios. This allowed us to discern biomarker patterns over shorter temporal and spatial scales. Climax Station 4 provided an opportunity to assess changes in stress and death biomarkers within the same phytoplankton population in a water column, which transitioned from a deep to a shallow MLD. Upon occupation of the station, we observed a deep MLD (>230 m), which shallowed to <20 m during our 2-day occupation (Fig. 8). This physical transition did not significantly alter⁶¹ community composition. We determined the in situ population physiology over these 2 days and for an additional 5 days using on-deck incubations.

The respective physiological states of phytoplankton communities residing in the upper 25 m of deep (Fig. 8, pink symbol) and shallow (Fig. 8, yellow symbol) mixed layers showed key differences. Re-stratification of Climax Station 4 stimulated both phytoplankton and viral accumulation. Increased phytoplankton accumulation is likely due to the increased daily availability of light and reduced grazer presence⁶⁴, while increased virus L⁻¹ may be indicative of a subset of phytoplankton or bacteria lysing/releasing viruses into the surrounding water during periods of strong stratification. Phytoplankton populations residing within the deeper mixed layer also had notable physiological differences. These populations initially had higher compromised membranes (~20%, Supplementary Fig. 12m), which decreased in subsequent days of occupation (Fig. 8). We point out that this level of compromised membranes is close to the values seen in the relatively healthy Accumulation phase (Fig. 2). Populations residing in the deep mixed layer also had lower TEP cell⁻¹ compared to the communities in more stratified surface waters (Fig. 8). For comparison, the Accumulation phase, which also harbored deeply mixed populations (Fig. 1b), had lower TEP cell⁻¹ and virus cell⁻¹ relative to this phase (Supplementary Fig. 5a, b). The increased turbulence during the episodic deep mixing event at Climax Station 4 may have facilitated the abiotic aggregation⁶³ of dissolved TEP precursors making them large enough for retention on the 0.45 µm pore-size filters (see ‘Methods’). The relatively high TEP cell⁻¹ could have also been derived from production by colonizing bacteria on cellular detritus^51,65 brought to the surface by active mixing. Even though TEP cell⁻¹ decreased, bulk (non-cell normalized) TEP L⁻¹ increased as the mixed layer shallowed (Fig. 8). Increased TEP L⁻¹ is consistent with general increases in phytoplankton biomass^49,66,69,68, or higher daily irradiance⁶⁷, both of which occurred during this period.

We next compared in situ phytoplankton at Climax Station 4 to the community at Climax Station 1. These stations were in the same bloom phase with similar phytoplankton concentrations (8.0–8.3 × 10⁶ cells L⁻¹) but had different mixing histories. Climax Station 1 had a consistently shallower MLD (ranging from 42 to 56 m) two weeks prior to and during our occupation, compared to the recently mixed and stratified Climax Station 4 (from 233 to 17 m). Given the general increase in stress and death markers for these populations after a five day incubation (Fig. 8), we expected cell populations at Climax Station 1 to resemble the communities at Climax Station 4 after stratification/incubation. Accordingly, Climax Station 1 had higher levels of intracellular oxidative stress, virus L⁻¹, metacaspase, and caspase activity (Fig. 9) than freshly mixed populations from Climax Station 4. This implies that populations in more stable water columns are transitioning to ROS-induced stress, increased PCD, and viral infection. Based on pigment composition, both days of Climax Station 4 were classified as diatom community types. In contrast, Climax Station 1 had a higher relative abundance of diatoms on the first day of occupation and dinoflagellates on the second day of occupation. Based on 16S rRNA analysis, the most abundant group at Climax Station 4 for both days of occupation was Ostreococcus (37 and 44% relative abundance, R.A.), while Climax Station 1 contained mainly Micromonas (47% R.A.). Collectively, these data may indicate succession to dinoflagellate or chlorophyte populations as the water column became more stratified during this bloom phase.

Lastly, we compared two stations in the Decline phase that had relatively stable, shallow mixed layers prior to and during occupation, but were characterized by relatively high (Decline Station 6: 2 × 10⁷ cells L⁻¹) or low (Decline Station 2: 3 × 10⁶ cells L⁻¹) phytoplankton concentrations. Decline Station 2 had higher oxidative stress (ROS, OxPC) and TAGs (Fig. 9). These data are consistent with nutrient stress^8,12 and virus infection¹³, both of which could contribute to lower phytoplankton biomass. Since Decline Station 2 had lower virus cell⁻¹ and virus L⁻¹ than Decline Station 6, (Fig. 9 and Supplementary Fig. 13), the elevated TAG and ROS at this station more likely came from nutrient limitation than viral infection. Decline Station 2 was located in the subtropical north Atlantic, which was relatively more nitrate limited (Fig. 6), which agrees with modeling predictions of these subregions⁶⁹. Decline Station 6, on the other hand, was in higher latitude subpolar waters and did not appear to be macronutrient limited (Fig. 6). These differences in nutrient concentration between these subregions in Decline phase may have shaped community composition and its impact on the aforementioned biomarkers (TAG, ROS, OxPC). Pigment analysis showed that Decline Station 6 and Decline Station 2 were respectively haptophyte and cyanobacteria community types for both sampling dates. Cyanobacteria were still the most prominent group in Decline Station 6 based on 16S rRNA analysis but they were less dominant than at Decline Station 2 (30, 32% vs. 86% R.A, respectively). Decline Station 6 contained more Bacillariophyceae (diatoms) (13% vs. 0.1% R.A.) and Prymnesiophyceae (haptophyte group) (~13% vs. 1% R.A.).

Resident populations in Decline Station 6 had significantly higher TEP cell⁻¹, compromised membranes, and caspase activity compared to Decline Station 2 (Fig. 9), consistent with active viral infection and lysis^29,50. Compromised membranes and virus cell⁻¹ both increased during at this station (Supplementary Fig 14), even though total phytoplankton concentration did not decrease, even 6 days after initial occupation (Supplementary Fig. 14). This finding further confirms that Decline Station 6 was not nutrient stressed and maintained a high biomass in the presence of top-down processes such as grazers or viruses. We note that viruses started accumulating in the Climax phase (Supplementary Fig. 5e) and resulted in universally high virus cell⁻¹ in the Decline phase (Supplementary Fig. 5b), regardless of latitude, sub-regional water type or macronutrient variability (Fig. 6). Our analysis suggests that this high virus cell⁻¹ is a persistent ecosystem pressure for phytoplankton, bacteria, or possibly grazers living in stratified water columns in early autumn in the Western North Atlantic.

In summary, our findings provide evidence that seasonal water column stratification is correlated with intracellular stress, PCD, and viral predation in phytoplankton communities. These cellular loss processes are variable and follow water column stratification, which effectively governs biomass accumulation over the annual phytoplankton bloom cycle in the North Atlantic. High phytoplankton accumulation occurs with an absence of physiological stress markers in the deeply mixed early spring. Phytoplankton cells continue to accumulate into the Climax phase as the mixed layer shallows in the late spring. Intracellular ROS, oxidized lipids, and TEP cell⁻¹—all signatures of oxidative stress—became elevated in this phase, signaling a transition towards loss in the Decline phase. The lowest accumulation rates of phytoplankton biomass throughout the bloom occurred within physically stable and shallow mixed layers of the Decline phase. While nutrient availability varied with sub-region, communities in this phase had the highest compromised membranes, cell death protease activity, virus cell⁻¹, DOC_SA, and TAG ChlA⁻¹ across all stations and subregions. These Decline phase biomarkers have been extensively linked to cell death and late stages of viral infection in diverse model systems in culture^29,43,50,70. Deepening of the mixed layer during Winter Transition increased nutrient concentrations and diminished both nutrient stress biomarkers and virus cell⁻¹, setting the stage for renewed phytoplankton accumulation. Community stress responses were also discernable on a shorter temporal scale between locations with different mixed layer histories and in response to a mixed layer shallowing event. The stress patterns we observed on different time scales provide insight to a previously unappreciated coupling of physics and phytoplankton physiology that can help inform our understanding of the underlying processes that govern phytoplankton bloom dynamics across the western North Atlantic.

Source: Ecology - nature.com