Bacterioplankton response to physical stratification following deep convection

Dissolved organic carbon (DOC) is produced as a result of numerous food web processes (Carlson and Hansell, 2015). However, the magnitude of annual DOC production is ultimately constrained by autochthonous primary production rates. Global net primary production (NPP) produces in excess of approximately 50 Pg C y^–1 (Dunne et al., 2007; Westberry et al., 2008; Silsbe et al., 2016) and is the main source of DOC production (approximately 21 Pg C y^–1) in the oceans (Carlson and Hansell, 2015). The oceanic DOC pool stores approximately 662 Pg C, which is comparable to the inventory of CO₂ in the atmosphere and the amount of carbon in terrestrial biomass (Hansell et al., 2009). Heterotrophic bacterioplankton rapidly recycle DOC for energy generation and biomass synthesis (Azam et al., 1983), and those living in the ocean’s surface waters are responsible for remineralizing a majority (50%–90%) of newly produced DOC (Williams, 1984; Azam et al., 1993; Ducklow, 1999; Carlson and Hansell, 2015). DOC that resists rapid microbial degradation can accumulate in surface waters, generating vertical gradients in DOC (Carlson et al., 1994; Carlson and Hansell, 2015).

Why and how DOC accumulates is not well understood, but a number of hypotheses have been proposed (see reviews in Benner and Amon, 2015; Carlson and Hansell, 2015; Dittmar, 2015). Some DOC compounds accumulate because they are intrinsically resistant to or slowly degraded by heterotrophic utilization (Hansell, 2013; Shen and Benner, 2020). Phytoplankton have been observed to directly produce recalcitrant compounds (Aluwihare et al., 1997) or precursors to recalcitrant compounds (Arakawa et al., 2017). Additionally, labile DOC can be altered to recalcitrant compounds by phototransformation (Kieber et al., 1997; Benner and Biddanda, 1998) or by heterotrophic microbes, including archaea and bacteria (Ogawa et al., 2001; Gruber et al., 2006; Jiao et al., 2010; Guerrero-Feijóo et al., 2017; Goto et al., 2020). Alternatively, the molecular diversity of organic compounds that comprise the bulk DOC pool can limit unique compounds from existing at detectable concentrations or from encountering microbes with the appropriate uptake mechanisms. Thus, the accumulated DOC pool reflects the sum of these highly diverse but individually low-concentration compounds (Kattner et al., 2011; Arrieta et al., 2015; Dittmar, 2015). DOC may also accumulate because metabolic costs may outweigh the benefits of oxidizing particular compounds (Carlson et al., 2009; Treusch et al., 2009; Giovannoni, 2017; Landry et al., 2017; Saw et al., 2020). Possibly, all of these mechanisms contribute simultaneously to the accumulation of DOC compounds in surface waters, where their functional recalcitrance establishes a mismatch between their supply and consumption by microbes (Zakem et al., 2020). However, accumulated DOC that is resistant to rapid microbial utilization at one geographical location or depth can be readily utilized by heterotrophic bacterioplankton at another (Carlson et al., 2011).

In regions like the western North Atlantic where convective mixing extends from the euphotic zone into the mesopelagic zone, particulate organic carbon and accumulated DOC in surface waters can be delivered to the deep ocean physically by the deepening of the mixed layer (e.g., mixed layer pump; Copin-Montégut and Avril, 1993; Carlson et al., 1994; Gardner et al., 1995; Dall’Olmo et al., 2016; Lacour et al., 2019). Deep convection obliterates vertical density gradients and redistributes suspended and dissolved organic matter homogeneously throughout the mixed layer. DOC concentrations become reduced in the epipelagic and enriched in the mesopelagic as a portion of the accumulated euphotic-zone DOC is mixed to depth. As the water column stratifies and the mixed layer shoals to the surface, the exported constituents (DOC and suspended particulates, including microbial biomass and communities) become isolated in the mesopelagic zone below the mixed layer (Carlson et al., 1994; Hansell and Carlson, 2001). However, only approximately 11% of the surface-accumulated DOC exported to the deep ocean persists to depths greater than 500 m (Carlson et al., 2010; Hansell et al., 2012), due to remineralization within the upper mesopelagic zone. Field and experimental studies have provided evidence for the degradation of surface-accumulated DOC by mesopelagic heterotrophic microbes following deep convection (Copin-Montégut and Avril, 1993; Carlson et al., 1994; Ducklow et al., 1995; Carlson et al., 2004; Carlson et al., 2010). Within a warm-core eddy in the Gulf Stream, bacterioplankton abundance was observed to increase throughout the surface 150 m after the water column had been mixed to 450 m and later restratified. The sub-euphotic increase in bacterial abundance was presumed to be a response to organic matter delivered to depth (Ducklow, 1986). At the Bermuda Atlantic Time-series Study (BATS) site (31.67°N, 64.17°W), a portion of the DOC exported out of the euphotic zone following winter deep convection was observed repeatedly to be removed in the upper mesopelagic on a timescale of weeks (Hansell and Carlson, 2001; Goldberg et al., 2009; Liu et al., 2022). At the same time, absolute bacterioplankton abundance in the upper mesopelagic at BATS was observed to increase following deep convection (Carlson et al., 2009), with increases in relative abundance by specific bacterioplankton taxa like SAR11 subclade II, OCS116, SAR202, and marine Actinobacteria (Morris et al., 2005; Carlson et al., 2009; Treusch et al., 2009; Vergin et al., 2013; Liu et al., 2020a). These observations suggest that after surface accumulated DOC was exported to the upper mesopelagic zone, some mesopelagic bacterioplankton were capable of responding to the exported DOC on a timescale of weeks to months (Morris et al., 2005; DeLong et al., 2006; Landry et al., 2017).

Metagenomic evidence supports the claim that some mesopelagic microbial lineages are well adapted to utilize compounds that are apparently recalcitrant to epipelagic populations (Saw et al., 2020). Indeed, mixed culture experiments conducted at BATS demonstrated multi-week increases in the relative abundance of slow-growing mesopelagic bacterioplankton (e.g., SAR202, Methylophaga, Hyphomonadaceae, and Alcanivoracaceae) after the introduction of recalcitrant carboxyl-rich alicyclic (CRAM) proxy compounds (Liu et al., 2020b). While these experiments advance understanding about what could happen in the days and weeks following deep convection, the measurements of in situ microbial responses in the hours to days following deep convection could provide insights regarding the fate of surface-accumulated DOC after it is vertically exported from the epipelagic and potentially remineralized in the mesopelagic zone.

During the second of four field cruises of the North Atlantic Aerosols and Marine Ecosystems Study (NAAMES) in May 2016, we had the unique opportunity to sample a hydrostation in the subtropical region of the western North Atlantic (47.46°N, 38.72°W) during a deep convective mixing event and follow the subsequent water column restratification over a 3-day period (Behrenfeld, 2010; Della Penna and Gaube, 2019). At this hydrostation (NAAMES 2 Station 4, hereafter N2S4), we sought to describe changes in biogeochemistry and in bacterioplankton production (BP) and community structure throughout the water column, with a particular focus on the upper mesopelagic zone (100–200 m). We hypothesized that the surface-accumulated DOC delivered by deep convective mixing into the upper mesopelagic zone would stimulate in situ heterotrophic production by specific mesopelagic-associated lineages, like SAR11 subclade II and SAR202. This stimulation would then lead to the differentiation of bacterioplankton community structure between the euphotic and mesopelagic zones. Here, we describe the observations made at N2S4 and compare them to those made in stratified water columns that were sampled during NAAMES 3 (September 2017). Three of these stratified stations (NAAMES 3 Stations 3, 3.5, and 4) were in the subtropical region of the western North Atlantic, within 1.5˚ latitude and longitude of N2S4 (Della Penna and Gaube, 2019). A fourth station (NAAMES 3 Station 6) was in the subpolar region but was included in this analysis as a representative of a well-stratified station that was also occupied for multiple days (Della Penna and Gaube, 2019).

Of the four cruises of the NAAMES study, we focus on the two where specific hydrostations were occupied for more than two days. The data from the two NAAMES cruises presented here are from NAAMES 2 (May 2016) and NAAMES 3 (September 2017). Each involved a ship transect aboard the R/V Atlantis between 39°N and 56°N latitude and 38°W and 47°W longitude (Figure 1).

On May 24, 2016 (NAAMES 2), the R/V Atlantis occupied a water mass (hydrostation N2S4, 47.46°N, 38.72°W) within the core of a moderately retentive anticyclonic eddy in the subtropical region of the western North Atlantic (Della Penna and Gaube, 2019). Upon occupation, Surface-Velocity-Program (SVP) drifters and a Seabird Navis BGCi float (float ID number n0647) were deployed. The drifters consisted of a spherical surface float tethered to a weighted holey sock drogue extending to 15 m. For this specific analysis, we selected only the drifters that transmitted for more than 24 hours after deployment, resulting in the trajectory observations of one drifter for N2S4 (platform ID number 300234062417670). Conductivity-temperature-depth (CTD) profiles from both the R/V Atlantis and the float revealed deep convective mixing at the time of occupation with a mixed layer of approximately 230 m. The sampling stations of the R/V Atlantis followed the drifter track during the period May 24–26 and then the float track during May 26–27. The drifter and float deployed at N2S4 were retained within the anticyclonic eddy core for 3 days before being transported to the eddy edge by surface currents. The float continued to profile the eddy for a total of 6 days (until May 30; Della Penna and Gaube, 2019; Figure 1a). To provide visually spatial context for our observations, we downloaded the maps of satellite altimetry (for May 25, 2016) from the Copernicus Marine Service Information portal, as well as chlorophyll a (Chl a; for May 23–27, 2016) from the Visible Infrared Imaging Radiometer Suite sensor aboard the Suomi-NPP satellite. A full analysis of the satellite altimetry data set for NAAMES can be found in Della Penna and Gaube (2019). Sea-level anomaly maps are available daily and are gridded on 1/4°.

In September 2017 (NAAMES 3), the R/V Atlantis sampled 3 subtropical stations within 1.5° latitude and longitude of the coordinates of N2S4, including N3S3 at 47.03°N, 40.11°W; N3S3.5 at 48.04°N, 39.24°W; and N3S4 at 48.64°N, 39.13°W (Figure 1b). N3S3 was located within the core of an anticyclonic mode-water eddy, and N3S4 was located at the periphery of a cyclonic eddy (Della Penna and Gaube, 2019). Each 24-hour occupation included a CTD cast to 200 m and a Seabird Navis BGCi float deployment (ID number n0850 at N3S3 and ID number n0849 at N3S4). Station 3.5, not associated with an eddy, was occupied for approximately 2 hours and did not include a float deployment.

On September 13, 2017, during the third NAAMES cruise, the R/V Atlantis arrived on a subpolar hydrostation (N3S6, at 53.38°N, 39.54°W) located in a quiescent subarctic region of the northwest Atlantic, which was characterized by large-scale wind-driven circulation and relatively more uniform currents (Della Penna and Gaube, 2019). This hydrostation was not associated with an eddy and was tracked and sampled via SVP drifters (platform ID numbers 300234065316880 and 300234065316710), Seabird Navis BGCi floats (float ID numbers n0846 and n0847) and ship. The drifters and floats deployed at this station were transported eastward by wind-driven circulation (Della Penna and Gaube, 2019; Figure 1b). The water column at N3S6 remained strongly stratified over a 4-day occupation (mixed layer depth ≤ 31 m).

We recognize that advective features can impact our observations. Because the focus of this study is on the biogeochemical and microbial dynamics in the upper mesopelagic zone (100–200 m) of N2S4, we limit the impact of advective features to our observations at N2S4 by including in our analyses only float and CTD profiles (and associated discrete water samples) with the same temperature, salinity, and density properties at and below 50 m (Figure S1). Anomalous profiles based on temperature-salinity diagrams were also omitted from further analysis for the NAAMES 3 stations (Figures S2 and S3).

Hydrographic ship data were collected up to 5 times each day of station occupation from Sea-Bird Scientific SBE-911+ CTD profiler sensors attached to a rosette with 24 × 10 L Niskin bottles while on station. Additional sensors included those for chlorophyll fluorescence (WET Labs ECO-AFL/FL), beam transmission (WET Labs C-Star), turbidity (WET Labs ECO), and oxygen (SBE43). These data are available through National Aeronautics and Space Administration’s Ocean Biology Distributed Active Archive Center (OB.DAAC) SeaWiFS Bio-optical Archive and Storage System (NAAMES, 2018). Data were averaged over 1-m bins for the present analysis.

Data from Seabird Navis BGCi floats were retrieved and processed by the University of Maine’s In-situ Sound & Color (MISC) Lab. Float profiles were collected up to 7 times each day of station occupation. The L2 processed data used in the present analysis are available via the University of Maine In-situ Sound & Color Lab FTP (NAAMES Floats, 2018) as well as the Argo Global Data Assembly Center (Argo, 2022). Float data were averaged over 1-m bins for the present analysis.

Following previous studies of the NAAMES stations, mixed layer depths for each CTD and float profile were determined as the depth below 5 m at which the Brunt–Väisālä buoyancy frequency (N²) was greater than its standard deviation (Graff and Behrenfeld, 2018; Morison et al., 2019). The maximum depth of the euphotic zone was determined from float data as the depth where irradiance was 1% of surface photosynthetically active radiation (PAR). The maximum euphotic zone depth for N2S4 and NAAMES 3 was estimated to be 77 and 57 m, respectively. Apparent oxygen utilization (AOU, µmol O₂ L^–1) is defined as the depletion of oxygen relative to saturation and was calculated for each 1-m binned CTD profile as the difference between the oxygen saturation concentration and the observed oxygen concentration. Oxygen saturation was estimated from potential temperature and salinity following Weiss (1970).

The Niskin bottles collected seawater samples at nominal depths of 5, 10, 25, 50, 75, 100, 150, and 200 m for the analysis of Chl a, phytoplankton cell abundance, nitrate + nitrite (N + N), bacterioplankton cell abundance (BA), bacterioplankton leucine incorporation rate, DOC, total dissolved amino acids (TDAA), and bacterioplankton community composition via 16 S rRNA amplicon sequencing. Replicate samples for DOC and bacterioplankton leucine incorporation were collected at all stations, while all other variables are represented by single sample analyses. These data are available through OB.DAAC SeaBASS (NAAMES, 2018), as well as the Biological and Chemical Oceanography Data Management Office (Carlson, 2020). All processed data, analyses, and code presented here are available on GitHub (Baetge, 2022).

Chl a concentrations (µg L^–1) were measured from single discrete samples by fluorometric analysis using the acidification technique and a Turner Designs 10AU digital fluorometer (Mueller et al., 2003). For each sample, 500 mL of raw seawater were filtered through a 25-mm 0.45-µm (Millipore HA) nitrocellulose filter, which was then extracted in 90% acetone for 48 hours at 0°C. Fluorescence was measured before and after acidification to determine Chl a concentration. Phytoplankton abundances (cells L^–1) of single samples for cells of < 64 µm diameter were determined within hours of sample collection using a BD Influx Sorting Flow Cytometer and categorized into four major groups: Prochlorococcus, Synechococcus, picoeukaryotes, and nanoeukaryotes (see methods in Graff and Behrenfeld, 2018). Total phytoplankton cell abundance was calculated as the sum of the abundances of all the groups. NPP (µmol C L^–1 d^–1) was determined from ¹⁴C-sodium bicarbonate uptake incubations. Specifically, water was collected from 3 depths predawn, inoculated with ¹⁴C sodium bicarbonate, and incubated in on-deck incubators at light levels corresponding to the collection depths. Water collected at the surface was separated and incubated at 6 light levels (approximately 1%, 10%, 20%, 40%, 60%, and 100% transmission of surface light). NPP was estimated at 1-m bins through the water column to 200 m using the Photoacclimation Productivity Model (PPM) from Fox et al. (2020), which accounts for physiological adjustments in intracellular Chl a concentration in response to light and nutrient availability (Behrenfeld et al., 2016). There was strong agreement between NPP estimates from the PPM and available measurements of 24-h ¹⁴C-uptake conducted at all NAAMES stations (Fox et al., 2020).

BA (cells L^–1) was determined for unfiltered seawater samples preserved with certified American Chemical Society formalin to a final concentration of 1% (vol: vol). Within 36 hours of collection, cells from each sample were filtered onto a blackened 0.2-µm polycarbonate membrane filter, stained with DAPI (4′,6-diamidino-2-phenylindole dihydrochloride, 5 µg mL^–1), mounted onto a slide with high viscosity immersion oil (Thermo Scientific Richard-Allan Scientific Resolve) and stored at –20°C until enumeration. DAPI-stained cells were enumerated manually using an Olympus BX51epifluorescence microscope with ultraviolet excitation at 1,000× magnification. Twelve fields of view were counted for each slide, and on average, 50–60 cells were counted for each field-of-view (Porter and Feig, 1980).

Bacterioplankton ³H-leucine incorporation rates (pmol ³H-Leu L^–1 h^–1) were estimated using a modified version of the microcentrifuge method (Smith and Azam, 1992). For each depth, a killed control (fixed immediately with 100 µL of 100% trichloroacetic acid [TCA]) and replicate 1.6-mL seawater samples were spiked with ³H-Leu (L-[4,5- 3H(N)] at 20 nM; specific activity 50.2–52.6 Ci mmol^–1; Perkin Elmer, Boston, MA). After incubating in the dark for 2–3 hours at ± 2°C of in situ temperature (15.3°C–25.7°C at N2S4, 15.5°C–19.3°C at NAAMES 3 Station 3, 10.4°C–16.0°C at NAAMES 3 Station 3.5, 7.2°C–16.6°C at NAAMES 3 Station 4, and 3.9°C–11.6°C at NAAMES 3 Station 6), incubations were terminated with 100-µL cold 100% TCA. Cells were pelletized after a series of microcentrifugation steps, washed with 5% TCA, 80% ethanol (vol/vol), and resuspended in Ultima Gold scintillation cocktail as described in Ducklow et al. (2001). Radioactivity was measured using a Hidex 300 Scintillation Analyzer and was corrected for quenching using an external gamma source and a quench curve. The coefficient of variation (CV) between measurements from replicate incubations was generally 1%–15%. However, CVs for the upper mesopelagic samples increased up to 40% because of the lower incorporation rates at those depths. The range in CVs for the upper mesopelagic zone was 1%–25% for N2S4, 10%–20% for NAAMES 3 Station 3, 1%–3% for NAAMES 3 Station 4, and 12%–40% for NAAMES 3 Station 6.

N + N concentrations (µmol N L^–1) at each depth were determined from single discrete samples that were gravity-filtered directly from the Niskin bottles through 47-mm 0.8-µm polycarbonate (Millipore) filters and stored at –20°C in sterile 50-mL conical centrifuge tubes. Samples were analyzed using a Lachat QuickChem QC8500 automated ion analyzer (University of Rhode Island Graduate School of Oceanography Marine Science Research Facility), which measures N + N with a precision of approximately 0.3 µmol N L^–1 and has a detection limit of approximately 0.8 µmol N L^–1.

Replicate DOC samples at each depth were gravity-filtered directly from the Niskin bottles through pre-combusted 47-mm 0.7-µm GF/F filters (Whatman, first flushed with 100 mL of sample water) into pre-combusted 40-mL certified Environmental Protection Agency (EPA) borosilicate glass vials. Samples were acidified to a pH of ≤ 3 by adding 50-µL DOC-free 4 N HCl and stored at 14°C in a volatile organic-free environmental chamber until analysis. DOC concentrations (µmol C L^–1) were measured in batches of ≤ 35 samples using the high-temperature combustion method (Carlson et al., 2010; Halewood et al., 2022) on Shimadzu TOC-V or TOC-L analyzers. Concentrations were quantified using standard solutions of glucose and low carbon ultrapure water. Samples were systematically referenced against surface and deep seawater calibrated with consensus reference material (Hansell SSR Lot#08-18), which were run every 6–8 samples. Precision of the Shimadzu analyzers was within 2% CV.

TDAA were determined from single samples that were gravity-filtered directly from the Niskin bottles through 47-mm 0.7-µm GF/F (Whatman, first flushed with 100 mL of sample water) into acid-washed 60-mL high-density polyethylene bottles and stored at –20°C. Samples were processed following Liu et al. (2020a), where samples were hydrolyzed in 6 N HCl under nitrogen for 20 hours at 110°C, neutralized using nitrogen evaporation, derivatized with ortho-phthalaldehyde, and measured using a Dionex ICS5000+ high performance liquid chromatography (HPLC) equipped with a fluorescence detector (excitation = 330 nm and emission = 418 nm). The molecular formula of each amino acid resolved by the HPLC was used to calculate its carbon concentration in µmol C L^–1. Thus, TDAA represents the sum of the carbon concentrations of all the individual amino acids. DOC-normalized yields of TDAA were calculated as the percentage of total DOC measured as amino acid carbon.

Net BP (nmol C L^–1 d^–1) was estimated for each depth by applying a leucine conversion factor of 1.5 kg C mol Leu^–1 incorporated (Simon and Azam, 1989). Phytoplankton cell abundance, NPP, AOU, BA, BP, DOC, and TDAA profiles were all integrated for the euphotic and the upper mesopelagic zones using the trapezoidal rule. The euphotic zone includes samples and data collected above the maximum depth at which PAR was equivalent to 1% surface PAR, which for N2S4 was 77 m and for NAAMES 3 stations 3, 3.5, 4, and 6 was 57 m. The 1% light level ranged between 27 and 77 m (n = 28) at N2S4. The mesopelagic zone includes samples and data collected between the depth where PAR was 1% of its surface value and 200 m. The integrated stocks and rates for both depth horizons were normalized to the integrated depth interval to provide mean volumetric concentrations and rates for the respective depth zones. The upper bound for the error (E) incurred by a trapezoidal integration of a depth profile was estimated by the error bound formula:

where a is the minimum depth, b is the maximum depth, n is the number of trapezoids used in the integration approximation, and k is the upper bound of the absolute value of the second derivative of a polynomial function fitted to the depth profile.

Nucleic acid sample collection, extraction, purification and 16 S rRNA gene amplification, library preparation, sequencing, and amplicon assignment are detailed in Bolaños et al. (2021). A single 4-L sample from each depth was collected by filtration through a 0.22-µm Sterivex filter (polyethersulfone membrane, Millipore) and preserved in sucrose lysis buffer (SLB, 20-mM ethylenediaminetetraacetic acid [EDTA], 400-mM NaCl, 0.75-M sucrose, and 50-mM Tris-HCl, pH 9.0) at –80°C. Nucleic acids were extracted and purified following the phenol: isoamyl alcohol: chloroform (25:1:24) protocol of Giovannoni et al. (1996). The V1–V2 region of the 16 S rRNA gene was amplified via polymerase chain reaction (PCR) using the 27F (5′-AGAGTTTGATCNTGGCTCAG-3′) forward and 338 RPL (5′-GCWGCCWCCCGTAGGWGT-3′) reverse primers, each with respective Illumina overhang adapters (Bolaños et al., 2021). The 25-µL PCR reactions consisted of 2.5 µL (5 ng μL^–1) of genomic DNA template, 2.5 µL of each primer (1 µM), 12 µL of 2× KAPA HiFi HotStart ReadyMix, and 5.5-µL PCR water. Thermocycling conditions of the PCR reactions were 3 minutes at 95°C, 25 cycles of 30 seconds at 95°C, 30 seconds at 55°C, 30 seconds at 72°C, and 5 minutes at 75°C. Libraries for each amplicon reaction product were constructed by attaching dual indices and Illumina sequencing adapters with the Nextera XT Index Kit using a second PCR amplification. Purified libraries were pooled in equimolar concentrations and sequenced using the Illumina MiSeq platform (v2, 2x250 PE lane) at the Center for Genome Research and Biocomputing (Oregon State University, Corvallis, OR, USA). Amplicon sequence data sets can be found in the National Center for Biotechnology Information Sequence Read Archive (NCBI SRA) database under the BioProject identifier PRJNA627189.

Primer sequences were removed from demultiplexed FASTQ files using the CutAdapt algorithm (Martin, 2011). All subsequent sequence analyses were performed using the R software environment (v 4.0.0). Using the package dada2 (v 1.2.0; Callahan et al., 2016), trimmed FASTQ files were quality filtered, dereplicated, and merged to create an amplicon sequence variant (ASV) table, from which potential chimeras were removed de novo. ASVs were taxonomically assigned using the SILVA database (v 123, Quast et al., 2013) and PhyloAssigner (v089, Vergin et al., 2013). SAR116 was reclassified from order Rickettsiales to order Puniceispirillales according to the latest version of SILVA (v. 138.1). Phylogenetic databases are available on GitHub (Bolaños, 2020). Using the function rarefy_even_depth in the package phyloseq (v 1.32.0; McMurdie and Holmes, 2013), sequence read counts were subsampled to the minimum sample read depth (9,702 reads) with replacement to standardize for sampling effort. The Chao1 and Shannon alpha diversity indices were estimated using the phyloseq function estimate_richness. The Shannon evenness index was calculated by dividing the Shannon diversity index by the total number of ASVs observed (Clarke, 1993). Pairwise comparisons of these alpha diversity indices were performed with t tests using the function compare_means of the package ggpubr (v 0.3.0). A one-way permutational multivariate analysis of variance (PERMANOVA) test was used to assess whether there were statistically significant differences between groups (e.g., stations, depth horizons, or time) using the function adonis in the R package vegan (v 2.5-6; Anderson, 2001). Bray–Curtis dissimilarities were calculated using the vegdist function of the package vegan (v 2.5-6; Oksanen et al., 2013). Nonmetric multidimensional scaling (NMDS) ordinations with Bray–Curtis dissimilarities were computed with the phyloseq function ordinate. Similarity percentages breakdown (SIMPER) using the simper function of the package vegan was used to identify major ASVs whose differences over time were an important contribution to the Bray–Curtis dissimilarities in the upper mesopelagic of N2S4. For all statistical analyses, p values of > 0.05, ≤ 0.05, and ≤ 0.01 are described as not significant, significantly different, or highly significantly different, respectively.

For each of the euphotic and upper mesopelagic zones, the relative abundance of a taxonomic group to the family level was calculated by summing its ASV counts across all depths in the depth horizon and then dividing the summed counts by the total counts in the depth horizon. An approximation of cell abundance for each taxonomic group was estimated using the product between the relative abundance of its sequences and the total cell abundance, normalized to its gene copy number, hereafter termed “16 S rRNA-estimated cell abundance.” Bacterial ribosomal RNA copy number information was retrieved from the ribosomal RNA operon copy number database (rrnDB; Stoddard et al., 2015). If copy number information was not available for a phylogenetic order, then the copy number for the next higher taxonomic rank was used.

The mixed layer depth in the anticyclonic eddy of N2S4 was 227 m upon occupation on May 24, 2016, and shoaled during station occupation over the next 4 days to < 26 m (average 11 ± 8 m, n = 9), with sustained thermal stratification through May 29, 2016 (observations after May 27 were from float n0647; Figure 2). On the first day of occupation of N2S4, the biological and chemical parameters N + N, Chl a, phytoplankton cell abundance, BP, AOU, and DOC were distributed homogeneously throughout the surface to 200 m, reflecting the deep mixed layer (Figure 3a–f). BA and TDAA were not distributed homogeneously through depth on the first day of occupation of N2S4 when the mixed layer was deepest (Figure 3h and i). However, BA increased at most depths over the station occupation, while TDAA displayed an increasing and decreasing trend in the euphotic and upper mesopelagic, respectively (Figure 3). Many of the measured variables displayed different trends between the euphotic and mesopelagic zones that can be more clearly discerned from depth-integrated changes, as detailed below (Figure 4).

In the euphotic zone (0–75 m samples), increases in depth-integrated Chl a and phytoplankton cell abundance corresponded to decreases in N + N and AOU (Figure 4a–d). Between May 24 and May 27, euphotic zone Chl a increased from 0.50 ± 0.03 µg L^–1 to 0.85 ± 0.34 µg L^–1 (t test, p = 0.05), phytoplankton cell abundance increased from 1.5 × 10⁷ ± 3.3 × 10⁴ to 4.7 × 10⁷ ± 3.5 × 10⁷ cells L^–1 (t test, p = 0.1; Figure 4a and b). N + N in the euphotic zone showed a decreasing trend (t test, p = 0.03; Figure 4c). Euphotic zone AOU decreased from May 24 to May 27 from 16.6 ± 0.3 to 11.3 ± 6.2 µmol O₂ L^–1 (t test, p < 0.0001; Figure 4d). Both BA and BP increased in the euphotic zone over time (Figure 4e and f). From May 24 to 26, BA increased from 1.2 × 10⁹ ± 8.9 × 10⁷ cells L^–1 to 1.7 × 10⁹ ± 4.1 × 10⁸ cells L^–1 (t test, p = 0.1; Figure 4e), while BP increased from 21.2 ± 2.5 nmol C L^–1 d^–1 to 52.5 ± 2.5 nmol C L^–1 d^–1 (t test, p < 0.0001; Figure 4f). As a result of the competing photoautotrophic and heterotrophic processes, we were not able to resolve significant changes in the bulk DOC concentrations in the euphotic zone during our station occupation (t test, p = 0.2). Bulk DOC concentrations within the euphotic zone averaged 53.4 ± 0.2 µmol C L^–1. However, the contribution of TDAA to the bulk DOC pool increased from 1% to 1.4% in the euphotic zone (t test, p = 0.2; Figure 4g and h).

In the upper mesopelagic zone (100–200 m), Chl a decreased from 0.44 ± 0.05 µg L^–1 to 0.27 ± 0.05 µg L^–1 (t test, p = 0.02), and phytoplankton cell abundance declined from 1.5 × 10⁷ ± 4.1 × 10⁵ cells L^–1 to 9.9 × 10⁶ cells L^–1 ± 7.1 × 10⁵ (t test, p < 0.01; Figure 4a and b). There were small increases in upper mesopelagic AOU (1 µmol O₂ L^–1, t test, p < 0.0001) and N + N (0.2 µmol N L^–1, t test, p = 0.2) corresponding to increases in BA (7.4 × 10⁸ ± 1.8 × 10⁸ cells L^–1 to 1.5 × 10⁹ ± 1.0 × 10⁸ cells L^–1, t test, p = 0.007) and BP (20.0 ± 4.7 nmol C L^–1 d^–1 to 31.0 ± 3.0 nmol C L^–1 d^–1, t test, p = 0.02; Figure 4c–f). The average mesopelagic BP rate during the well-stratified late summer (i.e., NAAMES 3 stations) was 2.1 ± 1.6 nmol C L^–1 d^–1. We observed that the upper mesopelagic BP rates upon occupation of N2S4 were 10-fold greater than those of the NAAMES 3 stations, averaging 20.0 ± 4.7 nmol C L^–1 d^–1 (n = 3) and increasing to as much as 35.2 ± 5.6 nmol C L^–1 d^–1 (n = 3) over the 3 days of occupation of N2S4 (mean upper mesopelagic BP after the first day of occupation was 32.0 ± 3.6 nmol C L^–1 d^–1, n = 18). The increase in BA in the upper mesopelagic (by 7.6 × 10⁸ cells L^–1) accounted for 63% of the total cell abundance increase throughout the water column (1.2 × 10⁹ cells L^–1; Figure 5e). Additionally, the upper mesopelagic increase in BP (by 11 nmol C L^–1 d^–1) accounted for 35% of the total increase in BP throughout the upper 200 m of the water column (31 nmol C L^–1 d^–1; Figure 4f).

Bulk DOC concentrations displayed a systematic decreasing trend within the upper mesopelagic zone over the occupation of station N2S4 (t test, p = 0.04; Figure 4g). The contribution of TDAA to the bulk DOC pool in the upper mesopelagic zone decreased from 1.3% to 0.9% (t test, p = 0.1; Figure 4h).

In contrast to N2S4, the water masses occupied during NAAMES 3 were thermally stratified. Hydrostations N3S3, N3S3.5, and N3S4 were in the subtropical region of the western North Atlantic and were all occupied for less than 24 hours with up to two CTD casts per station. Hydrostation N3S6 in the subpolar region was occupied for four days with up to 4 CTD casts per day. N3S6 remained thermally stratified throughout the occupation with little change in its mixed layer depth (mean of 24 ± 3 m, n = 31) and little variability in its biological and biogeochemical properties (Figures 5 and S4).

All NAAMES 3 stations presented here exhibited strong vertical gradients in N + N, Chl a, phytoplankton cell abundance, NPP, AOU, BA, BP, and DOC (Figure 5). Inorganic N + N increased over depth at all stations and displayed a latitudinal trend of increasing mesopelagic concentrations northward. Euphotic zone N + N concentrations were lower at all 4 of these stations relative to N2S4, averaging 1.9 ± 2.7 µmol N L^–1 compared to 5.2 ± 0.1 µmol N L^–1 at N2S4. Except for N3S3, the NAAMES 3 stations had higher N + N concentrations in the upper mesopelagic, averaging 12.4 ± 4.2 µmol N L^–1 compared to 5.4 ± 0.1 µmol N L^–1 for N2S4 (Figures 4a and 5a).

Chl a decreased over depth at N3S4 and N3S6 and a subsurface chlorophyll maximum was apparent at 50 m for N3S4. In the euphotic zone, Chl a averaged 0.5 ± 0.2 µg L^–1 for these stations, which was lower than the average concentration at N2S4 (0.8 ± 0.4 µg L^–1). Chl a was also lower in the upper mesopelagic of N3S4 and N3S6 (0.02 ± 0.02 µg L^–1) relative to N2S4 (0.3 ± 0.1 µg L^–1; Figures 4b and 5b).

At the NAAMES 3 stations, phytoplankton cell abundance decreased with depth, with higher euphotic zone concentrations at subtropical stations N3S3, N3S3.5, N3S4 than subpolar station N3S6. Phytoplankton cell abundance averaged 2.0 × 10⁸ ± 1.3 × 10⁸ cells L^–1 in the euphotic zone, which was higher than average the cell abundance at N2S4 (2.7 × 10⁷ ± 1.7 × 10⁶ cells L^–1). In the upper mesopelagic, phytoplankton cell abundance averaged 1.7 × 10⁷ ± 5.7 × 10⁶ cells L^–1, comparable to the average cell abundance at N2S4 (1.2 × 10⁷ ± 2.6 × 10⁶ cells L^–1; Figures 4c and 5c).

NPP decreased with depth at N3S3, N3S4, N3S6, with higher surface rates at subpolar station N3S6 relative to subtropical stations N3S3 and N3S4. The euphotic zone of the NAAMES 3 stations had a comparable average (0.9 ± 0.8 µmol C L^–1 d^–1) and lower range (0.1–2.7 µmol C L^–1 d^–1) of NPP than N2S4 (average 1.1 ± 1.0 µmol C L^–1 d^–1, range = 0–4.2 µmol C L^–1 d^–1; Figures 4d and 5d).

AOU was variable between all NAAMES 3 stations but decreased over depth at all stations, with an average in the euphotic zone of 14.5 ± 12.3 µmol O₂ L^–1 comparable to that at N2S4 (14.1 ± 3.4 µmol O₂ L^–1). Average and median AOU values for the upper mesopelagic zone at the NAAMES 3 stations were 47.0 ± 17.6 µmol O₂ L^–1 and 43.0 µmol O₂ L^–1, respectively. These values were substantially higher than the average AOU for the upper mesopelagic zone at N2S4 (17.8 ± 0.9 µmol O₂ L^–1; Figures 4e and 5e).

BA decreased with depth at all NAAMES 3 stations, without any clear latitudinal trends between them. Over the euphotic zone, BA was higher at the NAAMES 3 stations relative to N2S4, averaging 1.5 × 10⁹ ± 5.0 × 10⁸ cells L^–1 compared to 1.3 × 10⁹ ± 3.3 × 10⁸. The NAAMES 3 stations, however, had lower BA (5.8 × 10⁸ ± 3.6 × 10⁸ cells L^–1) in the upper mesopelagic zone relative to N2S4 (1.1 × 10⁹ ± 3.6 × 10⁸ cells L^–1; Figures 4f and 5f).

At N3S3, N3S4, and N3S6, BP decreased to background values of 2.4 ± 1.9 nmol C L^–1 d^–1 by 75-m depth and did not exhibit latitudinal trends between the stations. BP was comparable on average in the euphotic zone at the NAAMES 3 stations (41 ± 9 nmol C L^–1 d^–1) relative to N2S4 (40 ± 10 nmol C L^–1 d^–1). In the upper mesopelagic zone, however, BP values were on average 10-fold higher on the first day of occupation of N2S4 (20 ± 4.7 nmol C L^–1 d^–1) than the average of those at the NAAMES 3 stations (2.1 ± 1.6 nmol C L^–1 d^–1; Figures 4g and 5g).

DOC concentrations decreased with depth at N3S3, N3S3.5, N3S4, and N3S6, with elevated surface concentrations at subtropical stations N3S3, N3S3.5, and N3S4 relative to subpolar station N3S6. In the euphotic zone, DOC concentrations averaged 63.4 ± 4.4 µmol C L^–1 for the NAAMES 3 stations, which was greater than the average concentration of 53.4 ± 0.6 µmol C L^–1 at N2S4. The average DOC concentration in the upper mesopelagic zone was slightly lower at the NAAMES 3 stations (51.4 ± 2.0 µmol C L^–1) compared to N2S4 (53.1 ± 0.5 µmol C L^–1; Figures 4h and 5h).

Bacterioplankton profiles were analyzed over the surface 200 m to assess community composition shifts during rapid restratification of the deeply mixed water column at N2S4 and to compare these dynamics with properties of a persistently stratified systems (i.e., N3S3, N3S3.5, N3S4, and N3S6). Bacterioplankton ASV evenness was not significantly different between the depth horizons at N2S4 (Shannon evenness, t test, p = 0.82), nor at the NAAMES 3 stations (Shannon evenness, t test, p = 0.16). Bacterioplankton ASV richness was also not significantly different between the depth horizons at N2S4 (Chao1, t test, p = 0.31) but was significantly greater in the upper mesopelagic zone (100–200 m) relative to the euphotic zone (Chao1, t test, p = 0.00013) at the NAAMES 3 stations. ASV richness at N2S4 was significantly higher than observed at the NAAMES 3 stations for both the euphotic (t test, p = 0.0006) and the upper mesopelagic zone (t test, p = 0.04). The average Chao1 index for the water column of 5–200 m at N2S4 was 679 ± 135 (n = 21) compared to 585 ± 102 (n = 11) for the upper mesopelagic of the NAAMES 3 stations. ASV evenness was significantly lower at N2S4 than at the NAAMES 3 stations for both the euphotic (t test, p < 0.0001) and the upper mesopelagic zones (t test, p = 0.04; Figure 6). NMDS ordination based on Bray–Curtis dissimilarities (stress = 0.05) showed that the N2S4 samples clustered tightly together and far from the NAAMES 3 samples (Figure 7a). A PERMANOVA test revealed that community structure differed significantly by cruise (r² = 0.37, p < 0.01) and depth horizon (r² = 0.04, p = 0.03), without a significant interaction effect (r² = 0.02, p < 0.17). Within the samples collected at N2S4, the community over the entire water column clustered more closely over time (NMDS stress = 0.09). A PERMANOVA test showed that the N2S4 samples were significantly different based on the main effect of time (PERMANOVA r² = 0.17, p < 0.01) and the interaction between time and depth horizon (PERMANOVA r² = 0.18, p < 0.01), but not on depth horizon alone (PERMANOVA r² = 0.04, p = 0.4; Figure 7b). There were decreasing trends in Shannon evenness and increasing trends in Chao1 richness for both the euphotic and upper mesopelagic communities over the occupation of N2S4, but these changes were not statistically different (Figure S5).

During the restratification at N2S4, subclades (approximately family level) of SAR11 dominated in relative abundance. During the occupation of N2S4, SAR11 relative abundance increased by approximately 7% in the euphotic zone to represent approximately 45% of the bacterial community, whereas in the upper mesopelagic, it increased by approximately 18% to represent approximately 47% of the community (Figure 8). The relative abundance of SAR 11 subclade 1a was greatest of all SAR11 subclades and increased by approximately 9% and 16% to represent approximately 35% and 36% of the total community in the euphotic and upper mesopelagic zones, respectively, throughout the station occupation (Figure 8). Of the remaining SAR11 subclades (I, IB, II, IV), subclades I, IB, and II showed increases (> 12%) in relative abundance in the upper 200 m, individually comprising less than or equal to 5% of the bacterioplankton community in each of the euphotic and mesopelagic zones (Figure 8). Members of the Oceanospirillales (of the Gammaproteobacteria) were the next most relatively abundant members of the total bacterioplankton community, particularly ZD0405, SAR86, and Oceanospirillaceae; altogether, they represented 17% and 18% of the community in the euphotic and mesopelagic zones, respectively. Apart from Oceanospirillaceae, the relative abundance of these families did not change significantly in the euphotic zone. However, in the upper mesopelagic zone, the relative abundance of Oceanospirillales decreased by approximately 19% with families ZD405 decreasing by approximately 19%, SAR86 decreasing by approximately 15%, and Oceanospirillaceae by approximately 32% (Figure 8). Flavobacteriales (8% ± 1%, including families Flavobacteriaceae, NS9_marine_group, Cryomorphaeceae, NS7_marine_group), Rhodobacterales (4% ± 1%, family Rhodobacteraceae), and Rhodospirillaceae (3% ± 1%, family Rhodospirillaceae) were the next most relatively abundant orders across both depth horizons, contributing to greater than or equal to approximately 2% of the total bacterioplankton community. These bacterial lineages showed < 12% changes in relative abundance over time (Figure 8). The change in relative abundance for all remaining orders and families in both depth horizons over the station occupation of N2S4 was up to 50%, but these taxa contributed to < 2% of the total bacterioplankton community (Figure 9).

The total bacterioplankton community abundance increased in the upper mesopelagic (Figures 3h and 4e); thus, if this increase in cell abundance is considered with the changes in relative abundance of each taxonomic family, 16 S rRNA-estimated cell abundances reveal increasing trends for many bacterioplankton families, even those that did not increase in relative abundance. Indeed, as total community abundance increased throughout the surface 200 m of the water column, 87% and 88% of the families represented in the euphotic zone and upper mesopelagic zone, respectively, increased in 16 S rRNA-estimated cell abundance (Figure 8b).

Similarity percentages breakdown (SIMPER) analysis identified 127 families of 75 orders in the upper mesopelagic zone at N2S4; their variability between May 24 and May 26 was primarily responsible for Bray–Curtis dissimilarities (Table S1). Of these groups, six contributed to at least 1% of the total community and showed at least a doubling in 16 S rRNA gene-estimated cell abundance between May 24 and May 26, as shown in Figure 9. These taxa represent the orders Rhodobacterales (110% increase), Puniceispirillales (100% increase), Flavobacteriales (140% increase), and the SAR11 clade (140%–170% increases). These groups showed relative abundance changes ranging from –5% to 29% (Figure 9).

The energy from storm events can erode thermal stratification and lead to convective mixing of accumulated organic matter from the euphotic zone to the upper mesopelagic zone. As storms subside, water column restratification traps the exported particulate and dissolved organic matter within the mesopelagic zone, supporting the carbon and energetic demands of heterotrophic microbes. Capturing one of these mixing events as well as the subsequent restratification is rare for ship-based sampling programs. Here, we have reported the biogeochemical and microbial changes measured during and shortly following a mixing event in the subtropical western North Atlantic. We have demonstrated that the organic matter delivered into and trapped within the upper mesopelagic zone stimulates rapid bacterial production and community structure shifts. However, the change in bacterioplankton community composition was subtle and more gradual than the changes in bacterioplankton stocks and secondary production rates on a timescale of days. We hypothesize that the continued availability of labile DOC, whether delivered from the euphotic zone or produced in situ by particle solubilization, temporarily supports the growth of euphotic zone-associated bacterioplankton at mesopelagic depths.

Over the 3-day occupation of N2S4, NPP coincided with N + N utilization and the reduction of AOU in the euphotic zone (0–75 m; Figures 3a, e, and g and 4c and d). Increases in relative abundance of phytoplankton were attributed primarily to prasinophytes, diatoms (Bolaños et al., 2020) and picoeukaryotes (Graff and Behrenfeld, 2018).

BA and BP in the euphotic zone also responded positively to mixed layer shoaling following deep convective mixing (Figures 3d and h and 4e and f). Despite NPP outpacing BP in the euphotic zone during the occupation of N2S4, the accumulation of bulk DOC was not resolvable over the timescale of station occupation (Figure 3d and g). Previous work on initial bloom conditions following nutrient entrainment by deep mixing demonstrated that a greater fraction of NPP was partitioned as particulate organic carbon compared to DOC (Carlson et al., 1998; Baetge et al., 2021). Despite our inability to resolve changes in the bulk DOC pool, increases in both the concentration and relative contribution of TDAA to the bulk DOC pool (Figures 3i and 4j) suggest that the quality of the accumulated DOC pool changed to a composition that was less altered diagenetically (i.e., fresh and less recalcitrant) within the euphotic zone throughout station occupation. Amino acids comprise a large proportion of labile DOC and are preferentially and rapidly consumed by heterotrophic bacterioplankton (Keil and Kirchman, 1993; Cowie and Hedges, 1994; Davis and Benner, 2005; Guerrero-Feijóo et al., 2017). The elevated contribution of TDAA to the bulk DOC pool is thus indicative of newly produced DOC (Davis and Benner, 2005; Kaiser and Benner, 2012; Stephens et al., 2020), which can result from a variety of pathways including direct release from phytoplankton (Granum et al., 2002) or archaea (Bayer et al., 2019), microzooplankton egestion (Nagata and Kirchman, 1991), mesozooplankton sloppy feeding, egestion, and excretion (Lampert, 1978; Maas et al., 2020), particle solubilization (Smith et al., 1992), and viral lysis (Middelboe and Jørgensen, 2006). The change in DOC composition supports the hypothesis that the flux of labile DOC at the nmol L^–1 level partially supported the enhanced BP in the euphotic zone.

Chl a, phytoplankton cells, and DOC were exported to mesopelagic depths by deep mixing (Figure 3b, c, and f). The net effect of this redistribution is a reduction in the amount of organic matter in the euphotic zone as lower concentrations from depth are mixed to the surface. In turn, organic matter becomes enriched in the mesopelagic zone as a portion of the organic matter accumulated in the euphotic zone is mixed to depth. Subsequent thermal restratification of the upper 200 m then traps the exported organic matter within the mesopelagic zone, where it fuels the mesopelagic food web and metabolic demand (Carlson et al., 1994; Dall’Olmo et al., 2016; Lacour et al., 2019). The observed decreases in Chl a, phytoplankton cell abundance, and DOC in combination with increased AOU in the upper mesopelagic are consistent with enhanced heterotrophy over the 3-day occupation of N2S4 (Figure 3b, c, e, and f).

During the occupation of station N2S4, BA and BP increased throughout the upper mesopelagic as Chl a and phytoplankton cell abundance declined (Figures 3b–d and h and 4a, b, e, and f). While the exact source of available organic matter supporting the elevated BP in the upper mesopelagic is unknown, the increased BP led to an increase in BA and consumption of available organic matter. The extant bacterioplankton community may have been responding to labile DOC delivered from the euphotic zone during mixing. The loss of Chl a and phytoplankton cells suggests that phytoplankton were not yet acclimated to the deeper depths of the upper mesopelagic and may have been grazed upon or solubilized. Grazing or solubilization of organic particles delivered into the upper mesopelagic by deep convective mixing may have been additional potential sources for the freshly produced labile DOC needed to support increased BP (Cho and Azam, 1988; Smith and Azam, 1992; Smith et al., 1995). Significant increases of ASVs in the Flavobacteriaceae family within the upper mesopelagic of N2S4 suggest the potential for particle solubilization by heterotrophic microbes. Taxa from the phylum Bacteroidetes (including Flavobacteria) have been linked with particle-associated lifestyles throughout the global bathypelagic oceans (Salazar et al., 2015; Liu et al., 2018), potentially due to their capability to degrade complex polymeric organic matter (Cottrell and Kirchman, 2000). The presence of members of these bacterial clades in the upper mesopelagic at N2S4 supports the hypothesis of microbially driven organic particle solubilization, thus providing a potential source of labile DOC to the free-living microbial community (Cho and Azam, 1988; Smith et al., 1992; Smith et al., 1995).

Although most changes in bulk DOC and TDAA over the course of station N2S4 occupation were not statistically significant, there was a decreasing trend in bulk DOC, TDAA concentration, and percent contribution to bulk DOC in the upper mesopelagic zone (Figures 3i and 4h). This suggests that the DOC in the upper mesopelagic zone may have started to become qualitatively altered to more recalcitrant material as heterotrophic bacterioplankton preferentially consumed more labile components of DOC. This observation is consistent with the argument that heterotrophic bacterioplankton can alter DOC concentration and quality once isolated in the mesopelagic (Goldberg et al., 2009; Kaiser and Benner, 2009; Guerrero-Feijóo et al., 2017; Liu et al., 2022).

While the data presented here do not provide unequivocal evidence for the degradation of surface accumulated DOC by mesopelagic heterotrophic microbes, they do demonstrate that microbes respond rapidly to the introduction of less diagenetically altered organic matter from the euphotic zone (export of DOC or solubilization of exported suspended organic particles) as the water column physically stratifies following deep convection.

In the water columns of the NAAMES 3 stations, bacterioplankton communities were stratified over depth, with euphotic communities differentiating from upper mesopelagic communities (Figures 6 and 7a). These observations, reported previously for the NAAMES region (Bolaños et al., 2021), are consistent with studies in other oceanic systems (Giovannoni et al., 1996; Field et al., 1997; DeLong et al., 2006; Sunagawa et al., 2015). Taxonomic richness assessed by the Chao1 diversity index increased with increasing depth (Figure 3), a pattern also observed throughout the global ocean (Sunagawa et al., 2015) that reflects the range of niches resulting from heterogeneity in organic and inorganic matter quality and availability. Processes that result in patchiness of DOC concentration and quality include the spatial variability of excretory products of resident and vertical migrating zooplankton (Maas et al., 2020) and DOC heterogeneity associated with the solubilization of sinking particles (Azam and Long, 2001; Azam and Malfatti, 2007). In contrast to the vertical structure observed at the NAAMES 3 stations, the microbial community at N2S4 was homogenized throughout the surface 200 m during and shortly following deep convective mixing. The median alpha diversity indices of the N2S4 euphotic zone samples were higher relative to those of the NAAMES 3 stations, but comparable to the mesopelagic zone samples of N2S4 (Figure 6). This observation suggests that deep convection mixed the euphotic and mesopelagic microbial communities, thus enhancing community richness in the euphotic zone, while not simultaneously reducing richness in the upper mesopelagic. Similar patterns of enhanced euphotic zone richness due to the entrainment of mesopelagic microbial lineages following deep convective mixing have also been reported on seasonal timescales in studies off the coasts of Northwest Spain and California, USA (Cram et al., 2015; Hernando-Morales et al., 2018).

Previous work has reported shifts in mesopelagic microbial community structure on seasonal timescales following physical or chemical perturbations. For example, following deep convective mixing at the BATS site, shifts in upper mesopelagic bacterioplankton community composition have been observed with increases in the relative abundance of SAR202, Acidimicrobiales clade OM1, Salinisphaeraceae, OCS116, and SAR11 subclade II (Morris et al., 2005; Carlson et al., 2009; Treusch et al., 2009; Giovannoni and Vergin, 2012; Liu et al., 2022). Some of these mesopelagic bacterioplankton are hypothesized to retain the metabolic capability to degrade functionally recalcitrant DOC that accumulates in surface waters on a timescale of weeks (Morris et al., 2005; DeLong et al., 2006; Landry et al., 2017; Saw et al., 2020). SAR11, a group of free-living aerobic heterotrophic alphaproteobacteria that are ubiquitous in the ocean, is comprised of a variety of ecotypes adapted to different ocean niches (Brown et al., 2012; Vergin et al., 2013). Over a 7-year time series in the NW Mediterranean Sea, Salter et al. (2015) observed that the diversity of SAR11 ecotypes in the upper 5 m increased following physical mixing, with the elevated contributions of subclades Ib and II interrupting the predominance of subclade Ia during periods of stratification, low phytoplankton biomass, and severe phosphate limitation. Similarly at BATS, the members of the SAR11 subclade II become prevalent in the upper mesopelagic zone annually as the water column stratifies in late spring, after semi-labile DOC from the surface has been delivered to the upper mesopelagic by winter deep convection (Carlson et al., 2009). Throughout the subtropics in the NAAMES study region, Bolaños et al. (2021) showed that members of the SAR11 subclade II dominated the upper mesopelagic zone following deep convective overturn in the late winter and early spring. Taken together, these studies indicate that the upper mesopelagic bacterioplankton community responds to perturbations over the course of weeks to months.

As N2S4 thermally stratified over the station occupation, we expected that the microbial populations and organic matter trapped at depth would lead to a succession in bacterioplankton community structure, whereby the upper mesopelagic population would differentiate from that of the euphotic zone. Specifically, we hypothesized that the upper mesopelagic community would become more populated by lineages that generally predominate in the mesopelagic zones in the subtropical regions of the western North Atlantic, including SAR202 and SAR11 subclades Ic and II (Bolaños et al., 2021). While we observed statistically significant differentiation of bacterioplankton community structure throughout the water column over the 3-day occupation at N2S4 (PERMANOVA r² = 0.17, p < 0.01), the bacterioplankton communities did not diverge between the euphotic and upper mesopelagic on time scales of hours to days (PERMANOVA r² = 0.04, p = 0.4; Figure 7). These data suggest that on these short timescales, organic substrate availability (i.e., DOC flux or recently solubilized POM) resulting from episodic deep mixing events (e.g., via storm-induced winter convection) can supply enough labile organic matter to continue supporting the “euphotic zone” communities that were also mixed into and trapped within the mesopelagic zone (Hernando-Morales et al., 2018). Community composition changes at N2S4 may reflect a transient phase in microbial community structuring that, if observed over a longer period, may transition from a mixed system to a persistently stratified water column with vertical partitioning of euphotic- and mesopelagic-like communities similar to the vertical structure of the NAAMES 3 water columns (Figure 7).

Despite observing rapid changes in BA and BP, the duration of our occupation of N2S4 was simply not long enough to resolve a clear separation in community composition between the euphotic and mesopelagic zones. However, changes in community structure were beginning to take place throughout the water column (PERMANOVA r² = 0.17, p < 0.01). In the upper mesopelagic zone, notable changes in relative abundance were attributed to the SAR11 clade, particularly subclades Ia, Ib, and II, which increased by 15%, 29%, and 19%, respectively (Figures 8 and 9). While there was little change, if any, in the relative abundance of other bacterioplankton orders and families, classifying an individual taxon as being a “non-responder” can be misleading. In a community that increases in cell density and biomass, an individual taxonomic group that does not change in relative abundance would increase in 16 S rRNA-estimated cell abundance and can be considered a responding taxon to new or changing environmental conditions. For example, the relative abundance of taxonomic groups such as Alteromonadaceae, Alcanivoracaceae, and Rhodobacteraceae did not change significantly in the upper mesopelagic zone throughout the occupation of N2S4, but all at least doubled in 16 S rRNA-estimated cell abundance (Figure 9, Table S2). Indeed, increases in 16 S rRNA-estimated cell abundance throughout the upper mesopelagic zone were observed for 88% of the community members, suggesting that nearly the entire upper mesopelagic community contributed to the increase in BP (Figure 8). We note, however, that our observations relied on 16 S rRNA gene amplification using V1–V2 primers that only target bacterioplankton and plastids. As such, our analysis excluded archaeal responses (Wear et al., 2018), which could have been important given that deep-sea archaea have the capacity to degrade complex carbon compounds (Ouverney and Fuhrman, 2000; Li et al., 2015; Bergauer et al., 2018) and have been observed to assimilate phytoplankton-derived exopolymeric substances (Boutrif et al., 2011). We further caution that our 16 S rRNA-estimated cell abundances cannot be taken as absolute values as they reflect only the relative abundances of bacterioplankton (and not archaeoplankton) applied to total prokaryotic counts. Even so, bacterioplankton families that did not change in relative abundance would still be revealed as responders because of the increase in total prokaryotic abundance over the station occupation.

Increases in the relative and/or 16 S rRNA-estimated cell abundances of bacterioplankton taxa in the upper mesopelagic zone that are typically predominant in the euphotic zone, such as SAR11 subclade Ia and SAR116 (Morris et al., 2005; Bolaños et al., 2021), suggest that some of the organic matter exported to depth by deep convection was labile in character and could support the growth of these taxa that are typically associated with euphotic zone lineages. Such rapid responses of euphotic zone lineages at mesopelagic depths would be difficult to capture in coarsely resolved time-series sampling (e.g., monthly sampling at BATS). Intriguing is that SAR11 subclade Ia increased in relative and 16 S rRNA gene-estimated cell abundances in the upper mesopelagic zone over the short occupation of N2S4, as it is associated with life in the surface ocean under nutrient-limited conditions (Salter et al., 2015; Giovannoni, 2017). Furthermore, this subclade typically blooms in the surface mixed layer during summer when UV irradiation is most intense (Zafiriou et al., 2008). SAR11 subclade Ia has been shown to decline in the mesopelagic following annual deep convection at BATS, being supplanted by more mesopelagic-associated taxa like members of SAR11 subclades Ib and II (Carlson et al., 2009; Vergin et al., 2013), which were also observed to increase in relative and 16 S rRNA gene-estimated cell abundance in the upper mesopelagic zone of N2S4. Members of the SAR11 clade possess a high abundance of transporters targeting low-molecular-weight DOC, such as amino acids and osmolytes, and have C1 pathways for oxidizing volatile compounds (Giovannoni, 2017); thus SAR11 subclade Ia may have benefited from scavenging hydrolysis products and leaked compounds released by other bacterioplankton, such as attached Flavobacteriaceae and Pseudoalteromonadaceae, that can solubilize exported POM (Baty et al., 2000; Williams et al., 2013; Fontanez et al., 2015; Salazar et al., 2015; Liu et al., 2018). As such bacterioplankton solubilized particles, they may have released amino acids and other DOC that fueled production of SAR11 subclade Ia (Giovannoni, 2017; Arnosti et al., 2018). Indeed, the presence and increase of Flavobacteriaceae and Pseudoalteromonadaceae in the upper mesopelagic zone at N2S4 may suggest that they were mixed into mesopelagic depths with the export of suspended and sinking particles and subsequently could facilitate the release of labile organic matter through particle solubilization (Figure 9, Table S2). Given more time as well as sustained water column stratification, we hypothesize that a clear vertical separation of community structure would become evident as the sources of labile DOC and particles are utilized and become limiting. We further hypothesize that as conditions become unfavorable to euphotic zone- or particle-associated bacterioplankton taxa, such as SAR11 subclade Ia, SAR116, Flavobacteriaceae, and Pseudoalteromonadaceae, they will be outcompeted in the upper mesopelagic by taxa that are better adapted to utilizing surface-derived, but more recalcitrant DOC. Although the data presented here do not capture an extensive time series of community structure shifts following deep convection, they do provide unique evidence for the rapid response of heterotrophic bacterioplankton in the mesopelagic zone to the delivery of surface-derived organic matter. These data illustrate how deep convective mixing can impact the mesopelagic food web by stimulating BP on a timescale of days.

The generated data sets for this study can be found in the SeaWiFS Bio-optical Archive and Storage System (SeaBASS; https://seabass.gsfc.nasa.gov/) as well as from the Biological & Chemical Oceanography Data Management Office (BCO-DMO, DOI: HTTPS://DOI.ORG/10.26008/1912/bco-dmo.824623.1). Amplicon sequence data sets can be found in the National Center for Biotechnology Information Sequence Read Archive (NCBI SRA) database under the BioProject identifier PRJNA627189. Seabird Navis BGCi float data are available via the MISC Lab FTP (ftp://misclab.umeoce.maine.edu/experiments/NAAMES/floats/) as well as the Argo Global Data Assembly Center (ftp://ftp.ifremer.fr/ifremer/argo). Phylogenetic databases are available at https://www.github.com/lbolanos32/NAAMES_2020. All processed data, analyses, and code presented here are available on GitHub (https://github.com/nbaetge/naames_multiday).

The supplemental files for this article can be found as follows:

Figures S1–S6. docx

Table S1. csv

Table S2. csv

We wish to thank the entire North Atlantic Aerosols and Marine Ecosystems Study team and the captains, officers, crews, and marine technicians of the R/V Atlantis. We thank Elisa Halewood for dissolved organic carbon sample processing and logistical support. Alyson Santoro provided thoughtful feedback that improved this manuscript. This manuscript also benefited from helpful discussions with Craig Nelson, Anna James, Jacqui Comstock, Brandon Stephens, and Chance English. Finally, we are grateful to the editors and reviewers for their time and comments that helped to strengthen this manuscript.

This project was supported by the National Science Foundation and National Aeronautics and Space Administration (Award Numbers NSF OCE-157943 and NASA 80NSSC18K0437 to CC). North Atlantic Aerosols and Marine Ecosystems Study was supported by the NASA Earth Venture Sub-Orbital program (EVS-2, NNX15AAF30G to MJB). Analysis was partially supported by the Simons Foundation International BIOS-SCOPE program.

The authors have no competing interests to declare.

This manuscript has not been published elsewhere.

Collected the samples: NB, CAC, LB, JRG.

Processed the samples: NB, CAC, LB, JRG, SL, KO.

Analyzed the data: NB, CAC.

Assisted with data reduction and contributed to the revision and editing of the final manuscript: All authors.

All authors are aware of and accept responsibility for this manuscript and have approved the submitted manuscript.

How to cite this article: Baetge, N, Bolaños, LM, Della Penna, A, Gaube, P, Liu, S, Opalk, K, Graff, JR, Giovannoni, SJ, Behrenfeld, MJ, Carlson, CA. 2022. Bacterioplankton response to physical stratification following deep convection. Elementa: Science of the Anthropocene 10(1). DOI: https://doi.org/10.1525/elementa.2021.00078

Domain Editor-in-Chief: Jody W. Deming, University of Washington, Seattle, WA, USA

Associate Editor: Jeff S. Bowman, Integrative Oceanography Division, Scripps Institution of Oceanography, San Diego, CA, USA

Knowledge Domain: Ocean Science