Samples were collected in the N. Pacific at Ocean Station Papa (Station P; 50° N and 145° W) from August 8 to September 3, 2018 aboard the R/V Roger Revelle (RR1813) and in the N. Atlantic at PAP (49° N, 15° W) from May 5 to 31, 2021 aboard the RRS James Cook (JC214) ( Figure 1 ) while following Lagrangian floats (Siegel et al., 2021; Johnson et al., 2023). Sample water was collected from 10-L Go-Flow (N. Pacific) or 20-L Niskin (N. Atlantic) bottles attached to a CTD rosette directly into 1-L polycarbonate bottles (acid-washed with 10% HCl). Rosette casts were deployed between 2 to 5 am GMT in the N. Atlantic and variable times in the N. Pacific. At a minimum for each sampling cast, a “surface” depth (<10 m in the N. Pacific; 5 m in the N. Atlantic) was collected. In the N. Pacific, depth profiles to 200 m were opportunistically taken, and in the N. Atlantic, a “sub-euphotic” depth (collection depth of 75 m or 95 m; euphotic zone range of 1% surface irradiance from 37 to 59 m) near the base of the mixed layer (range 15 m to 197 m, 0.03 kg m^-3 potential density differential) were collected (Johnson et al., 2023). Additional physical context methodology is referenced in the ancillary data below.

Whole or size-fractionated seawater incubations were performed in 60-mL glass BOD bottles with O₂ sensor spots (SP-PSt-3-NAU, PreSens) mounted to the interior of the bottle wall. The BOD bottles were washed with 5% HCl for 3–6 h and triple rinsed with 18.2 MΩ Milli-Q water before use. For size-fractionation, seawater samples were poured from the collection bottle into acid-washed polycarbonate filter towers. The tower and filter were rinsed with 100 mL of seawater sample, which was gravity filtered, using gentle (>-5 in Hg) vacuum pressure to dry the filter between rinses to help avoid contamination from unfiltered droplets during the sample transfer. The rinsate was discarded, and then 300 mL of the sample was similarly filtered. Size fractions in the N. Pacific included an unfiltered fraction for community respiration and <5 µm small plankton respiration (polycarbonate, Millipore Sigma). The N. Atlantic had these fractions, as well as <3 µm microplankton and <1.2 µm nanoplankton respiration fractions (polycarbonate, Millipore Sigma). The BOD bottles were then filled with either unfiltered or size-fractionated seawater samples by first triple rinsing with c.a. 20 mL of sample water poured directly from the collection bottle (unfiltered) or filter tower (size-fractionated) and overfilled. Bubbles were removed from the BOD bottle over one minute by gently tapping the side, then the bottle was capped with an air-tight glass stopper and carefully inspected for any bubbles.

The BODs were submerged in a dark, temperature-controlled water bath at c.a. in situ temperature for incubation. Incubation times ranged from 16 to 72 h based on detecting a robust decrease in O₂ concentration (Vandermeulen and Chaves, 2022). In the N. Pacific, O₂ measurements were made non-invasively through the bottle wall at 4 h intervals by manually holding the optical fiber to the BOD wall at the sensor spot (OXY1-SMA, PreSens). The N. Pacific average incubation time was 38 h with 8 to 10 discrete O₂ measurements per assay. Over the 30-day occupation, 214 O₂ consumption assays were conducted at Station Papa with 94 unfiltered and small size-fraction assays in the mixed layer (avg. 29 ± 4.5 m) (Siegel et al., 2021). In the N. Atlantic, O₂ measurements were acquired continuously every 30 s by directly mounting the polymer optical fibers to the BOD bottles over the sensor spot. Measurements were made with OXY4-SMA (precision ± 1.4 µmol L^-1 O₂ at 283 µmol L^-1 O₂; PreSens), with an average incubation time of 30 h and 3,000 discrete O₂ measurements per assay, totaling 41 unfiltered and 66 filtered assays. The bottles were inspected at the end of the incubation for bubbles and none were noted.

O₂ measurements were automatically temperature- and salinity-corrected using the PreSens Measurement Studio2 based on direct, continuous temperature measurements of the water bath and a defined in situ salinity and pressure (35 psu, 1013 hPa). A Milli-Q control BOD bottle was kept in the water bath for the duration of the cruise as a secondary temperature control and a reference for sensor drift. Further temperature correction of the samples was performed by subtracting the change in O₂ concentration of a Milli-Q control bottle from the sample bottles for each point. The corrected sample O₂ values were then plotted against incubation time.

For the N. Pacific campaign, only one fiber optic cable was available for reading the optode spots; thus optode measurements were manually acquired from all BOD bottles every few hours, generating relatively few O₂ measurements per experiment (n=8 to 20), as is precedent in Winkler assays and prior optode studies (Edwards et al., 2011; Wikner et al., 2013). The O₂ consumption rates were calculated as a Model I regression of the measured points, as is standard for O₂ consumption assays. Apparent O₂ outliers (>1 µmol L^-1 O₂ difference from the surrounding points) were identified and removed within each assay (avg. 1.2 points removed). Assays with rates <0.2 µmol O₂ L^-1 d^-1 were considered below detection (n=9), and rates greater than 3 µmol O₂ L^-1 d^-1 were considered anomalously high for summer at the N. Pacific site and excluded from analysis (n=3).

In the N. Atlantic, we added 12 more sensor channels and cables, which enabled continuous, operator-independent monitoring for a substantially increased set of point measurements. The first 2.5 h of data were initially removed to compensate for abiotic acclimation of the PSt3 sensor spot and BOD bottles to the water bath conditions. Plots were then visually inspected to trim additional nonlinear components at the beginning or end of the incubation, similar to the effects discussed in Wikner et al. (2013). To calculate the rate over c.a. 3,000 discrete O₂ measurements in a single assay, a rolling mean with a window size of 12.5 min was applied to the trimmed data, and a linear regression was calculated based on the rolling mean points. Only for the purpose of comparing respiration rates with other studies was a respiratory quotient of 1.4 O₂ to CO₂ applied to the O₂ measurements. This respiratory quotient was determined experimentally in the N. Atlantic campaign (Stephens et al., 2023b) and matches that used by Sherry et al. (1999) in the N. Pacific.

Respiratory stain (INT) and O₂-based incubations were conducted independently on the N. Atlantic expedition and were used here as a check for the optode-derived rates. Sample water for INT incubations was collected at midday casts, whereas O₂ consumption samples were collected from pre-dawn rosette casts. For each depth, 3 live samples and 2 poisoned (2 w/v formaldehyde) control incubations were conducted. Following methods detailed by Martínez-García et al. (2009), 8 mmol L^-1 INT was added to a final concentration of 0.2 mmol L^-1 and incubated in the dark at in situ temperature for 2 h. The samples were then filtered under gentle vacuum pressure through stacked (Sartorius), in-line filters (45 mm 3.0 µm and 47 mm 0.2 µm, polycarbonate, Millipore Sigma) to capture size-fractionated cells. The absorbance of the extracted propanol-formazan solution was measured at 485 nm. Linear regressions of the INT concentrations to absorbance were made using 3 unique 12-point dilution-based standard curves of freshly prepared Fomazan dye (Fisher Scientific) ( Supplementary Figure S1 ). After the killed control INT absorbance was subtracted from the live sample, the absorbance units were converted to INT concentrations using the mean slope measured from the known standards (0.0174 absorbance units per µmol INT L^-1). Values were then converted into oxygen units using the previously identified INT to oxygen relationship for samples collected from a nearby site in García-Martín et al. (2019a): Log O_{2 =} 0.80Log(INT) + 0.45. To compare the O₂ consumption rates to the INT-derived rates, assays were selected from both methods where sample collection occurred within 16 h of each other.

Care was taken to limit carbon contamination in respiration assays by acid washing and sample water rinsing surfaces, as well as using gravity filtration to reduce cell shear on rinsed filters, with gentle vacuum pressure used to dry the filter. All respiration assays were monitored for anomalously high rates (>3 µmol O₂ L^-1 d^-1). Intermittent samples were collected for total organic carbon (TOC) measurements to monitor for carbon contamination during the sample handling process. Samples were taken directly off the rosette cast and compared to the BOD bottle samples at the start of the respiration incubation assays. TOC was quantified by the high-temperature combustion method as described by Stephens et al. (2020).

Chlorophyll-a and net primary production were performed in the Pacific and N. Atlantic as reported in Meyer et al (Meyer et al., 2022, 2023). Briefly, net primary production (24 h ¹³C uptake incubation) was estimated based on samples collected at the surface on the same days as respiration assays in the N. Pacific and at the same rosette casts as respiration assays in the N. Atlantic.

Linear regressions were performed on each respiration assay as described above using a general linear regression model ( Supplementary Figure S2 , Supplementary Figure S3 ). For the N. Atlantic, the residual standard error (RSE) was calculated across the multiple size fractions and unfiltered samples. One outlier was identified in the sub-euphotic <5 µm fraction using the ROUT method (Q=1%) and removed. Welch’s ANOVA and Dunnett T3 multiple comparisons test (α=0.05) were used to identify significantly different (p<0.001) RSE pairings across size fractions and depth ( Supplementary Figure S4 , Supplementary Figure S5 ). Normality was determined using the Shapiro-Wilk test with α=0.05. The test statistic (W) for all size fractions and depths was >0.89, except for the unfiltered sub-euphotic (W=0.76, p=0.001). Thus, normality is assumed for all fractions but deviates slightly for the unfiltered sub-euphotic samples, though they were still best represented by normal distributions. Pearson’s correlation coefficients were calculated on the means and standard deviations of the N. Atlantic biological variables, assuming Gaussian distributions to identify significant relationships (n=8 to 18, p<0.05) ( Supplementary Figure S6 ). In the N. Pacific, unfiltered respiration was compared to coinciding program measurements over a depth profile taken on September 3, 2018. Spearman’s correlation was used with one-tailed p-tests and a 95% confidence interval.

Physical parameters were defined by Siegel et al. (2021) in the N. Pacific and Erickson et al. (2023) and Johnson et al. (2023) in the N. Atlantic. Bacterial abundance was enumerated via epifluorescence, as described in Stephens et al. (2023a). To evaluate bacterial activity, ³H-leucine incorporation rates are used as a proxy for bacterial production (BP) as described by Stephens et al. (2023a). Phytoplankton were enumerated via flow cytometry following Graff and Behrenfeld (2018). Additional ancillary measurements from the EXPORTS program and their respective methods can be found on the NASA SeaBASS data repository (doi:10.5067/SeaBASS/EXPORTS/DATA001).

Station Papa is a high nitrate, low chlorophyll-a region where primary productivity was low and representative of typical conditions during the sampling ( Figure 1 ) (Boyd and Harrison, 1999; Meyer et al., 2022). The mixed layer was shallow, stable, and above the euphotic zone (1% surface irradiance) for the majority of our Lagrangian study (Siegel et al., 2021). Over the 30-day occupation, 214 O₂ consumption assays were conducted at Station Papa ( Figure 1 ), with 94 whole water and small (<5 µm) size fraction assays in the mixed layer (avg. depth of 29 ± 4.5 m) (Siegel et al., 2021). Over the two-day assay incubation period, typically 8 to 10 O₂ measurements were obtained for each assay by manually holding the optic fiber over the sensor spot. Linear fits of consumption over time were variable, ranging from strong (r² >0.8, n=25), to moderate (r^{2 =} 0.5 to 0.8, n=33), to weak (r² <0.5, n=12) ( Figure 2B-D , Supplementary Figure S2 ). Surface respiration rates averaged 1.2 µmol O₂ L^-1 d^-1 for both the whole water (range 0.34 to 2.25 µmol L^-1 O₂ d^-1) and small fractions (range 0.55 to 2.41 µmol L^-1 O₂ d^-1) and were relatively stable during the study ( Figures 2A , 3 ). The respiration rates often showed no significant statistical difference between size fractions ( Figures 2 , 3 ). These rates are similar to the mean 1.56 µmol O₂ L^-1 d^-1 total respiration rate derived from Winkler titrations in the summer of 1997 at Station Papa (Sherry et al., 1999).

Vertical profiling in the N. Pacific revealed a pattern in whole water respiration rates: a decrease from the surface down to 50 m, a near-maximum level at 95 m followed by a decline to levels below reliable quantification at 200 m ( Figure 4 , Supplementary Figure S2 ). Simultaneous measurements of BP did not show an increase at 95 meters, despite the relatively high respiration observed at this depth. Instead, BP consistently attenuated from the surface to deeper waters ( Figure 4 ) (Stephens et al., 2023a).

While we successfully estimated respiration rates in the N. Pacific, suggesting that small cells are the primary contributors to community respiration, our capacity to distinguish trends among size classes was hindered by the infrequent resolution of O₂ measurements. To improve on this in the N. Atlantic PAP site ( Figure 1 ), we increased the number of O₂ meter channels and directly mounted the polymer optical fibers to the BOD bottles ( Supplementary Figure S7 ), enabling simultaneous O₂ measurements every 30 s for 12 sample bottles and one Milli-Q control. As a result, the number of O₂ measurements increased from c.a. ten in the N. Pacific to c.a. 3,000 measurements per assay in the N. Atlantic ( Figure 5A ). Within an assay, a rolling mean was applied to the large number of point measurements, which was then used to calculate the rate ( Figure 5A ). The high-resolution O₂ measurements made it clear if portions of the consumption curve needed to be trimmed: the first c.a. 2.5 to 5 h were generally removed ( Supplementary Figure S8 ) as they had a sharp slope consistent with sensor acclimation (as demonstrated in the Milli-Q control; Supplementary Figure S9 ), or if the curve became non-linear at the end of incubation or due to sensor slips ( Supplementary Figure S3 ). These improvements allowed us to calculate O₂ consumption rates with greater accuracy and precision (mean residual standard error of respiration rates <0.1 µmol O₂ L^-1 d^-1; Supplementary Figure S4 , Supplementary Figure S5 ) than the point resolution of the instrumentation (± 1.4 µmol L^-1 O₂), and reduce the incubation time required to obtain a reliable, linear consumption signal in this environment to under 20 h. Sensor drift was negligible (5- and 10-fold less than the average N. Atlantic respiration rates for the small and unfiltered fractions, respectively). With this approach, we resolved respiration rates <0.2 µmol O₂ L^-1 d^-1 ( Figure 5D-I , Supplementary Figure S3 ).

Organic carbon contamination can be a significant challenge in obtaining accurate respiration rates in oligotrophic systems, considering the low in situ labile TOC concentrations and carbon limitation of heterotrophic bacteria. TOC concentrations were higher in the BOD than samples obtained directly from the rosette (ranging from 0 to 8 µmol C L^-1 increase; Supplementary Figure S10 ). However, there was no consistent pattern in TOC enrichment in whole water versus size-fractionated bottles, and the variability in TOC enrichment was not reflected in the variation of respiration rates.

To validate the accuracy of our assay setup and optode-derived respiration rates, we compared data with independently-derived INT measurements available in the N. Atlantic campaign. INT respiration rates (<3 µm size class) ranged from 0.55 to 0.80 µmol O₂ L^-1 d^-1, closely aligning with our optode-derived ranges from the same dates (0.42 to 0.84 µmol O₂ L^-1 d^-1; Figure 6 ). The consistency (<0.2 µmol O₂ L^-1 d^-1 difference on May 18^th, May 28^th, and May 26^th/27^th) between the INT and optode-based methods further support the accuracy of the high-throughput optode assays.

Just as in the N. Pacific, in the N. Atlantic we measured respiration rates among two size classes, whole water and small cells, at the surface and below the euphotic zone. In addition, for six surface samples, we included two additional size classes: micro (<3 µm), and nano (<1.2 µm) ( Figures 5D–I ). Only whole and small (<5 µm) size classes were collected below the euphotic zone due to limited sensor availability. At the surface, N. Atlantic respiration rates averaged 1.7 µmol O₂ L^-1 d^-1 for whole water and 0.9 µmol O₂ L^-1 d^-1 for the small size fraction ( Figure 5B ). The small-, micro-, and nano-size classes averaged 45% of this whole water surface respiration ( Figures 5B, C ), with differences between these three classes often not statistically distinguishable ( Figures 5D–I ). Considering the nested nature of the size fractions, this suggests that nanoplankton were the major contributors to small-fraction respiration at the surface. Additionally, over half of the total community respiration was accounted for by size classes larger than 5 µm.

N. Atlantic respiration rates were substantially lower at depths below the euphotic zone (mean 0.6 µmol O₂ L^-1 d^-1 whole water, 0.3 µmol O₂ L^-1 d^-1 small fraction; Figures 5 , 7B, C ). In contrast to the surface where the small fraction accounted for 53% of total respiration, they contributed 90% of the total in the sub-euphotic zone ( Figure 5C ). This demonstrates a shift towards small cells being the predominant active microbial community size classes with depth, particularly as large chlorophyll-a decreased substantially in the second half of the cruise ( Figure 7D ).

Assay uncertainty tends to increase with a decrease in the O₂ consumption signal as the abundance of respiring cells decreases (such as through fractioning samples or moving from the surface to the mesopelagic) (del Giorgio and Williams, 2005). Given the robust nature of our N. Atlantic measurements, we examined the assay model fit to understand the potential sources of variability across such parameters. Variation between replicate respiration assays demonstrated a dependency on both depth and size fraction. For surface samples, variability was relatively low for the size-fractionated samples and significantly higher for the whole water sample (RSE analysis, Dunnett’s T3 multiple comparisons test, p<0.001). Additionally, surface whole water replicates had significantly higher coefficients of variation (66.2% CV, F*_{5.000, 27.38 =} 16.65) compared to the size-fractionated assays (23% small, 26% micro, 16% nano.; Supplementary Figure S5 ). The increased variability in surface whole water samples likely stems from the stochasticity in capturing large plankton and particles in the upper water column samples. In the deeper, sub-euphotic samples, no significant differences in replicate variability or % CV were observed between whole water and size-fractionated samples ( Supplementary Figure S5 ), fitting with respiration being dominated in the sub-euphotic by smaller cells that are more consistently captured in the BOD setup.

In contrast to the relatively stable N. Pacific system, the N. Atlantic survey exhibited greater dynamics, encompassing a phytoplankton bloom decline with multiple storms and associated mixing events ( Figure 7 ). Though the N. Atlantic campaign was Lagrangian, following a retentive, anticyclonic eddy with a warm core, project sampling above the eddy core was quasi-Lagrangian, as it was impacted by mixing events (Erickson et al., 2023; Johnson et al., 2023). Consequently, we divided the temporal trends of the N. Atlantic campaign at PAP into six periods (the initial system state, four mixing events, and the final system state) to better understand the drivers of respiration between size classes and depth within this water mass.

At PAP, we initially observed a declining, large phytoplankton bloom, marked by reduced primary production and chlorophyll-a levels (Meyer et al., 2023). During this period, surface whole water and small-fraction respiration decreased ( Figure 7B ). A storm event then deepened the mixed layer, corresponding with a maximum in surface whole water respiration ( Figures 7A, B ). During this period, total respiration and chlorophyll-a were dominated by large cells (>5 µm) ( Figure 7D ). Further wind events continued to shift the mixed layer, and the phytoplankton bloom steadily dissipated with a marked decrease of large phytoplankton. This phytoplankton biomass was mixed below the surface with the storm events. In general, whole water surface respiration declined, following chlorophyll-a trends.

Below the euphotic zone, the small size fraction had relatively stable respiration rates. However, whole water respiration rates had two significant increases, corresponding to mixing during the first storm and first wind periods. Notably, a peak in deep water respiration rates was observed at the end of the first wind period, coinciding with a major flux event and an increase in particulate carbon (Meyer et al., 2023).

Utilizing these optode assays in the N. Pacific and N. Atlantic, we observed varying contributions of different microbial size classes to community respiration in two contrasting oceanic provinces. The Northeast Pacific sampling (Station Papa) represented the low productivity endmember where our surface, small size-fraction rates were relatively low and similar to historical measurements in the same region by Sherry et al. (1999). Both biomass and rate measurements were dominated by small cells (Meyer et al., 2022), and small-fraction respiration rates were often statistically indistinguishable from the whole water samples ( Figures 2A , 3 ). When converted to carbon units, total surface respiration (average 0.86 µmol CO₂ L^-1 d^-1 using a respiratory quotient of 1.4 O₂ to CO₂) is elevated compared to the project’s respiration estimates derived from remineralization assays (0.19 µmol CO₂ L^-1 d^-1) (Stephens et al., 2020). This speaks to the importance of size-fractionated rate assays, as our small size fraction (<5 µm) encompasses a larger portion of the microbial community than that of the <3 µm dilutions used in remineralization experiments. Additionally, the size-fractionated respiration rates build on Niebergall et al. (2023) study of several independent ship-based, autonomous, and satellite-based EXPORTS program estimates of net community production at the site, finding it to be slightly net autotropic with low export potential. Overall, these results imply that the bulk of microbial community respiration in the mixed layer can be attributed to small cells, including bacterioplankton, cyanobacteria, and small eukaryotes and was in close balance with primary production (McNair et al., 2021; Meyer et al., 2022).

A comparison of total respiration rates to coinciding program depth profile measurements (Meyer et al., 2022) revealed a significant negative correlation to small-cell chlorophyll-a (Spearman correlation, r=-1.00, p=0.042, n=4), and weak positive correlations with total particulate carbon and net primary production (r=0.80, p=0.17, n=4). These findings reflect the low biomass, low productivity, and highly recycled nature of the Station Papa ecosystem in late summer (Meyer et al., 2022; McNair et al., 2023; Niebergall et al., 2023). The vertical respiration profile decreases with depth ( Figure 4 ), consistent with minimal particle flux occurring at this time (Durkin et al., 2021; Estapa et al., 2021), then respiration rebounds just below the deep chlorophyll maximum. Interestingly, there is no concurrent increase in BP or bacterial abundance observed at the base of the deep chlorophyll maximum. In fact, respiration between 65 to 95 m displays an inverse relationship with BP and small phytoplankton abundance. This pattern suggests a spatial decoupling of microbial respiration from BP and plankton abundance.

Kim et al. (2017) previously found a decoupling between BP and bacterial respiration in downwelling systems where nutrients are low, proposing that a shift in metabolism towards cellular maintenance rather than growth was driving decoupling. We postulate that bacteria around the 100 m depth horizon utilized solubilizing phytodetritus as dissolved organic matter became more recalcitrant (Baetge et al., 2021), leading to an increase in respiration despite the divergence from the BP trend. The 95 m respiration increase occurs just below the base of the euphotic zone, the deep chlorophyll maximum, and the phytodetritus maximum, which presumably originates from high densities of cyanobacteria and picoeukaryotes ( Figure 4 ). Simultaneous work at Station Papa by Stephens et al. (2020) shows a decrease in bacterioplankton growth efficiency with depth, indicating a higher proportion of respiration to BP. It is also possible that the inherently flexible metabolism of mixotrophs, such as abundant Synechococcus, are not well-captured by the BP assay, contributing to the decoupling trend. Below 100 m, respiration, BP, bacterial abundance, and phytodetritus concentrations decreased to rate or concentration minimum as expected.

In the N. Atlantic (PAP), we improved our respiration rate estimates with higher frequency O₂ measurements and additional size fractions to further resolve microbial community member contributions. We observed substantial dynamics in N. Atlantic respiration rates driven by the decline of a phytoplankton bloom and multiple mixing events. Here, large phytoplankton were more prevalent than in the N. Pacific (Meyer et al., 2023). Accordingly, our N. Atlantic whole water respiration rates were, on average, 1.4-fold higher than the N. Pacific and small size-fraction rates were, on average, 1.3-fold lower. Unlike the N. Pacific, small-fraction respiration strongly correlated with small net primary production (p=0.05, Pearson’s correlation; Supplementary Figure S6 ). Small-fraction respiration also moderately correlated with large and total chlorophyll-a (p=0.11, r=0.57 and p=0.09, r=0.60, respectively). However, BP decoupled from small respiration with depth, as observed in the N. Pacific (p=0.01, r=0.83 at the surface and p=0.53, r=-0.26 in the sub-euphotic, Pearson’s correlation; Supplementary Figure S6 ).

While on average, small-fraction respiration in the N. Atlantic was responsible for 45% of the total respiration (similar to prior estimates for this system) (Fasham et al., 1999; Aranguren-Gassis et al., 2012; Binetti et al., 2020), we observed substantial temporal dynamics in the size class contributions. At the surface, initially the small fraction contributed most of the respiration. After the first storm, large size chlorophyll-a decreased, yet the respiration in the whole water was the dominant signal of all size fractions, likely from particle-associated respiration. The largest contribution to respiration then gradually shifted back to the small fraction in the mid to late stages of the observation period, coincident with a decline in chlorophyll-a. As large cells aggregated into phytodetritus, storms mixed the particles below the euphotic zone (Estapa et al., 2023; Clevenger et al., 2024). There were notable increases in respiration in the whole water treatment within the sub-euphotic zone. Otherwise, sub-euphotic respiration was low with a markedly higher contribution (80%) of the small size fraction to total respiration ( Figures 5C , 7 ). After major wind events in the second half of the study, the three filtered size fractions displayed little difference in magnitude at the surface, coinciding with the second bloom phase where chlorophyll-a had shifted from large to small cells (Meyer et al., 2023) and small-fraction respiration accounted for a greater portion of community respiration ( Figure 7B, D ).

Several studies have documented similar dynamics in the N. Atlantic during spring blooms, including the Biogeochemical Ocean Flux Study Spring Bloom Experiment, the N. Atlantic Bloom Experiment, and recent research near PAP (Bender et al., 1992; Savidge et al., 1992; Binetti et al., 2020). These studies all demonstrate that this is an energetic system, net autotrophic annually, with frequent switching between net autotrophic and heterotrophic states. Our findings reveal that these dynamics can occur on relatively short timescales (days) and drive interactions between respiration rates in the surface and sub-euphotic zones. These active systems influenced by regular redistribution of biological and chemical stocks, taken with our findings, suggest that infrequent respiration sampling may miss critical, large shifts in the system and lead to a misrepresentation of regions as net auto- or heterotrophic.

This work sampled two well-studied oceanographic stations for several weeks. At both stations, we found that respiration often followed chlorophyll-a trends, though respiration also unexpectedly decoupled from BP with depth. Facilitated in part by advancements in respiration measurement techniques, like those detailed here, our findings reveal that significant fluctuations in respiration rates can occur on the order of days. This rapid metabolic response can considerably affect incubation-based estimates of net community production within a system. More extensive sampling is required across both temporal and spatial scales to better resolve carbon cycle components, enabling a more accurate assessment of respiration’s role in ocean ecosystem metabolism.