Mesozooplankton grazing minimally impacts phytoplankton abundance during spring in the western North Atlantic

The impacts of grazing by meso- and microzooplankton on phytoplankton primary production (PP) was investigated in the surface layer of the western North Atlantic during spring. Shipboard experiments were performed on a latitudinal transect at three stations that differed in mixed layer depth, temperature, and mesozooplankton taxonomic composition. The mesozooplankton community was numerically dominated by Calanus finmarchicus at the northern and central station, with Calanus hyperboreus also present at the northern station. The southern station was >10 °C warmer than the other stations and had the most diverse mesozooplankton assemblage, dominated by small copepods including Paracalanus spp. Microzooplankton grazing was detected only at the northern station, where it removed 97% of PP. Estimated clearance rates by C. hyperboreus and C. finmarchicus suggested that at in-situ abundance these mesozooplankton were not likely to have a major impact on phytoplankton abundance, unless locally aggregated. Although mesozooplankton grazing impact on total phytoplankton was minimal, these grazers completely removed the numerically scarce > 10 µm particles, altering the particle-size spectrum. At the southern station, grazing by the whole mesozooplankton assemblage resulted in a removal of 14% of PP, and its effect on net phytoplankton growth rate was similar irrespective of ambient light. In contrast, reduction in light availability had an approximately 3-fold greater impact on net phytoplankton growth rate than mesozooplankton grazing pressure. The low mesozooplankton grazing impact across stations suggests limited mesozooplankton-mediated vertical export of phytoplankton production. The constraints provided here on trophic transfer, as well as quantitative estimates of the relative contribution of light and grazer controls of PP and of grazer-induced shifts in particle size spectra, illuminate food web dynamics and aid in parameterizing modeling-frameworks assessing global elemental fluxes and carbon export.

157 (Pall). Each dilution treatment was prepared in a carboy, and gently siphoned through silicone 158 tubing from the carboys into 1.2 L polycarbonate bottles. Duplicate bottles of each dilution were 159 amended with a final concentration of 10 μM of both nitrate and silicate, and 1 μM of phosphate. 160 An additional set of duplicate 100% WSW bottles was prepared without adding nutrients to serve 161 as a nutrient control. 162 Mesozooplankton grazing was measured in incubations paired with the dilution assays 163 ( Figure S1). The mesozooplankton experiments were performed using either additions of discrete 164 numbers of handpicked copepods (S1 and S2) or a whole mesozooplankton assemblage (S4). 165 Copepods for discrete copepod additions were collected at midnight from the upper 15 m of the 166 water column with a vertical net tow using a 1 m diameter ring net fitted with 220 µm mesh and 167 a non-filtering cod-end. To minimize stress on the animals, as soon as the copepod net was 168 retrieved, the content of the cod-end was diluted into buckets containing unfiltered (S1) or 169 filtered (S2) surface seawater. Individual undamaged copepods were selected under a dissecting 170 microscope using wide-bore pipettes and were placed into 30 mL vials until all incubation bottles 171 were filled. Actively swimming copepods were transferred to 1.2 L polycarbonate bottles 172 containing <200 µm WSW. The remainder of the sample from each net tow was preserved in 173 ethanol (10% final concentration). All fixed samples were later sorted under a stereo dissecting 174 microscope to identify species, estimate abundances, sex, and stage distribution of copepods, 175 accounting for those that were removed for the experiments. Animals were identified to the 176 lowest possible taxonomic level (species for Calanus spp., genus for other copepods, various 177 levels for other taxa). 178 Taxa used in discrete experiments were determined based on copepods observed 179 abundance and ease of sorting. In order to measure detectable feeding rates and reduce the 180 possibility of the copepods clearing all particles from the whole bottle, we chose the number of 181 copepods per bottle based on expected per capita clearance rates, such that the copepods in the 182 bottles would not be expected to clear more than 30-40% of the volume in the bottles (Gifford, 183 1993). A range of copepod concentrations was used to bracket the potential variability in feeding 184 rates. At S1, feeding experiments were conducted using C. hyperboreus despite the species not 185 being the most abundant, because based on its reported distribution range, it was expected that its 186 prevalence would decrease at more southern stations, preventing further investigation. Adult 187 individuals, known to range in size from 5-7 mm (Leinaas et al., 2016), were added to the 188 incubation bottles at concentrations of 1, 3, or 5 individuals per bottle. At S2, adult individuals of 189 C. finmarchicus were used. Based on this species' smaller size (2.5-2.7 mm; Leinaas et al., 190 2016), 10, 20, and 40 individuals L -1 were added. At S4, net tows revealed the dominance of 191 small (<1mm) metazoa that would be difficult to sort by hand, as done for S1 and S2. Hence 192 some whole seawater was collected without the 200 µm screening mesh so as to include 193 mesozooplankton. Incubations of this unscreened whole seawater served as the experimental 194 treatment representing the mesozooplankton assemblage at concentrations occurring in situ.

195
Discrete copepod additions were incubated in duplicate for each copepod concentration, 196 and unscreened WSW including the mesozooplankton assemblage used at S4 was incubated in 197 triplicates ( Figure S1). In all experiments, no nutrients were added to the bottles containing the 198 mesozooplankton. Incubations for all experiments lasted 24 h so as to encompass grazers' 199 possible diel feeding cycle. Bottles were placed in 250 L on-deck plexiglass incubators kept at 200 ambient temperature using an open-circuit flow of surface seawater. Neutral density screening 201 was used to maintain incubations at light intensities targeted to correspond to in situ light 202 intensity at the collection depth, approx. 10% of surface irradiance at S1 and S2, and 40% of 250 whether grazing was significantly different from zero (α≤ 0.05), and to yield the standard error of 251 the grazing rate estimate. Phytoplankton in situ specific growth rates (µ, d -1 ) were determined as 252 the sum of the grazing rate (g, d -1 ) and the mean net growth rate of the non-amended <200 µm 253 WSW bottles. Negative values of grazing rates, which result when the phytoplankton apparent 254 growth rate (k) is lower in the diluted than in the undiluted treatments, indicate a violation of a 255 central assumption of the dilution method (Landry & Hassett, 1982). Thus in case of statistically 256 significant negative grazing rates, losses were undetermined, and in the absence of a grazing loss 257 estimate, μ was equated to the net growth rate (k) in the undiluted non-amended bottles. 258 For S1 and S2, copepod clearance rates for C. hyperboreus and C. finmarchicus based on 259 bulk Chl a measurements were calculated as described previously (Leising et al., 2005b). Briefly, 260 the net phytoplankton growth rates (k) in the copepod additions and the microzooplankton-only 261 control were regressed against the concentration of copepods in each treatment, with the 100% 262 <200 µm seawater treatment from the microzooplankton dilution experiments used as ''zero'' 263 copepod treatment. The slope of the regression line is equivalent to the per capita daily grazing 264 rate of copepods on Chl a. Negative slopes of this regression indicate removal of Chl a through 265 copepod feeding. Specific grazing rates were converted to clearance rates using the relationship 266 of F = Vg/N, where F is the clearance rate (mL individual -1 d -1 ), V is the volume of the 267 experimental container (mL), g is the grazing rate (d -1 ), and N is the number of copepods (Frost, 268 1972). Copepod viability was verified for all incubation bottles at the end of the experiments. 269 Corrections for copepod mortality were not necessary, because all individuals except one C. 270 hyperboreus in a bottle with 5 copepods were found to be actively swimming at the end of the 271 incubation period.

272
For S4, total phytoplankton mortality due to the combined effects of micro-and 273 mesozooplankton grazing was calculated by subtracting the net phytoplankton growth rate (k) in 274 the unscreened WSW bottles from µ. Mesozooplankton grazing rates were then calculated as the 275 difference between total and microzooplankton-induced mortality rates.

276
In order to estimate grazing effects on particle size abundance spectra, particle 277 abundances between 3 and 60 µm equivalent spherical diameter (ESD) were determined from 2 278 mL samples collected from initial stock and from all bottles at the end of the incubation period 279 and analyzed with a Multisizer TM 3 Coulter Counter® (Beckman Coulter, USA). A FSW sample 280 was also analyzed and served as a correction blank. Blank controls contained a negligible 281 amount of particles, with a coefficient of variation of 1%. Significant difference in particle size 282 abundance spectra for the copepod treatments relative to the microzooplankton-only control was 283 assessed with a 2-sample Kolomogorov-Smirnov test at a significance level of p <0.05.

287
Each station had distinctive physical characteristics reflective of its latitudinal location 288 and the mesoscale feature it occupied (Figure 1). Mixed layer depth was 56 m, 41m, and 22 m at 289 S1, S2, and S4 respectively. Mixed-layer average water temperature ranged from 3.9 °C at S1 to 290 15.5 °C at S4 (Figure 2). Mixed-layer average salinity ranged from 34.58 at S2 to 36.13 at S4 291 ( Figure 2). Chl a concentration in the source water, i.e. in situ concentration at sampling depth 292 for the incubation experiments, ranged from 1.4 (± 0.09) µg L -1 at S4 to 2.7 (± 0.11) µg L -1 at S2 293 (Table 1). Collection depth for source water and copepods was within the surface mixed layer, 294 well shallower than the mixed layer depth (Figure 2). Copepod community composition differed 295 between stations ( Figure 3). C. finmarchicus was present at all stations, and was numerically 296 dominant at S1 and S2 (Table S1). C. hyperboreus was only present at S1. The most diverse 344 depth of sample collection, resulted in a 14% reduction in k relative to the <200 µm control, 345 corresponding to a removal of 14% of PP (Table 1). There was no significant interaction between 346 light intensity and grazer size fraction (p= 0.793). 347 348 Particle removal by micro-and mesozooplankton. 349 An effect of copepod grazing on the size abundance spectrum of particles in the 3-60 µm 350 size range was observed, indicating a shift in size spectra mediated by different grazer 351 communities. Although few particles >10 µm were generally observed, a larger proportion of 352 >10-15 µm particles were removed by mesozooplankton grazers at all stations over the course of 353 the 24 h incubations, in comparison with microzooplankton-only incubations ( Figure 6). 354 Measurably higher particle abundances in the >10 µm, and particularly in the 10-20 µm size 355 range were observed in microzooplankton-only incubations compared to incubations that 356 contained mesozooplankton grazers. The effect of copepod grazing on particle size distribution 357 was greatest at S2 (Figure 6b), where at concentrations of 20 and 40 C. finmarchicus per bottle, 358 ca. 5-fold lower particle abundances in the 10 to 20 µm size range remained after incubation. In 359 contrast, at that same station there was no measurable change in the relative contribution of each 360 size category to total particle abundance in the microzooplankton-only incubations, and size 361 spectra of the initial and final samples were indistinguishable ( Figure 6b). This observation is 362 consistent with the lack of detectable microzooplankton grazing at that station. 363 At S1, the difference between treatments in particle size distribution (PSD) was not 364 significant, either for the entire spectrum (Kolmogorov/Smirnov test, p= 0.62) or for >5 µm (p= 365 0.56) and >10 µm (p> 0.25) particles. However, the analysis was done for the treatment with 366 only one copepod added per bottle, which was likely insufficient to manifest a significant change 367 in particle concentration. At S2, PSD in the copepod treatment was significantly different from 368 the microzooplankton-only treatment (p= 0.04) for particles >5µm when 40 copepods per bottle 369 where present, and for the ≥ 5µm to 20 µm size range when only 20 copepods were added. At 370 S4, although variable among size bins, the net rate of change in the abundance of particles in the 371 microzooplankton treatment was similar to the Chl a based net growth rate from the dilution 372 experiment, with an average of 0.6 d -1 for particles 3-30 µm, a size fraction representing 99% of 373 the total counts. Although an increase in particle abundance was observed for most size bins 374 irrespective of the type of grazer, the number of >5µm particles in the mesozooplankton 375 treatment were reduced relative to the microzooplankton control ( Figure 6c) and the difference in 376 PSD between the two treatments was significant (p= 0.04). 377 378 Discussion 379 380 In many systems, the degree of grazer-induced phytoplankton mortality and the type of 381 grazer largely drive the fate of primary production (Steinberg & Landry, 2017). Whether organic 382 matter is shuttled to higher trophic levels, recycled within the photic zone, or exported to depth 383 via fecal pellets or other mechanisms, depends on the relative grazing impact of large metazoa 384 and herbivorous protists. Zooplankton grazing is also considered to play a role in the 385 development of phytoplankton blooms ), including the formation of the 386 North Atlantic spring bloom (Behrenfeld & Boss, 2014). In this study, we observed a spatial 387 switch in the relative grazing impact of micro-vs. mesozooplankton. At the most northern S1, 388 microzooplankton were responsible for a high removal of phytoplankton production, whereas 389 larger, mm-sized copepods had a low grazing impact, suggesting a relatively greater rate of 390 biomass recycling in the surface compared to export facilitated by larger specimen that produce 391 significant fecal pellets and/or vertically migrate (Ducklow, Steinberg & Buesseler, 2001;392 Schnetzer & Steinberg, 2002). At the more southern S2 and S4, grazer-induced losses of primary 393 production were low and essentially due to mesozooplankton. At S2, grazing by 394 microzooplankton was undetectable, and although grazing was undetermined at S4, the high net 395 growth rates of a doubling per day obtained at that station both from Chl a and particle counts 396 suggest that microzooplankton grazing was minimal. Remarkably, the expected decrease in net 397 phytoplankton growth rates under reduced light at S4 was approximately three times larger than 398 the effect of grazing. The observations presented here may well be restricted to the specific 399 conditions encountered during the study. However, the biological and physical characteristics of 400 the three stations span a considerable dynamic range and the results provide concrete rate 401 estimates and testable hypotheses linking physical features, grazing pressure, light availability, It has been argued that copepod grazing impact has commonly been underestimated due 453 to the use of nets >200 µm that leave out smaller species and stages of copepods (Turner 2004). 454 In our experiments at S1 and S2, copepods were collected using a 220 µm net, thus the grazing 455 impacts presented here are based on the species investigated, assuming their prevalence in the 456 >220 µm mesozooplankton community, and may represent underestimates of the impact on 457 phytoplankton by metazoan grazers as a whole. Some of the smaller <200 µm metazoan 458 specimen are part of the microzooplankton community (Calbet 2008), however the dilution 459 method used to quantify microzooplankton grazing does not allow to distinguish between the 460 small metazoan grazing impact and that of protistan grazers.

461
The low impact of grazing by copepods on phytoplankton globally has been interpreted 462 as an indication that copepods likely rely on other food sources to supplement their diet (Calbet ). High grazing rates and selective feeding by 465 mesozooplankton on microzooplankton in incubations would release microzooplankton grazing 466 pressure on phytoplankton, and thus could lead to an increase in net phytoplankton growth 467 compared to the microzooplankton-only treatments and to an underestimation of copepod 468 grazing rates (Nejstgaard, Gismervik & Solberg, 1987;. Here, however, 469 we did not observe the induction of such trophic cascades where higher net phytoplankton 470 growth rates are observed in the presence of copepod grazers, which would have suggested 471 preferential feeding on herbivorous protists. Indeed our calculation of clearance rates indicated 472 direct uptake of phytoplankton for both S1 and S2, which scaled with the concentration of 473 copepods. It is possible that copepods feeding on both phytoplankton and microzooplankton 474 would offset the individual effect on net phytoplankton growth rates of each feeding mode, but at 475 S1, the only station where microzooplankton grazing was significant, such offset would fail to 476 explain the large differences in net phytoplankton growth between the copepod treatments and 477 the control. It is likely that copepod feeding was omnivorous, and microzooplankton and 478 phytoplankton were grazed by mesozooplankton at proportions similar to their relative 479 abundances (Barquero et al., 1998;Halvorsen et al., 2001).

480
Despite their low feeding impact on the phytoplankton community as a whole, copepods 481 had a remarkable impact on the particle size spectrum. Although few particles were observed in 482 the >10 µm size range, these particles were removed only in the presence of copepods. This 483 observation supports the often-observed feeding preference of Calanoid copepods for larger 484 particles (e.g. Frost, 1972;Gifford et al., 1995;Levinsen et al., 2000). Particles <5 µm tend to be 485 too small for these copepods, although very large particles such as diatoms with spines may be 486 too large (Campbell et al., 2009). Copepods, however, demonstrate extensive flexibility in their 487 diet (Kleppel, 1993) and may feed size-selectively when food is abundant, and non-selectively 488 when food becomes scarce (Cowles, 1979). It is possible that if copepods in our incubations fed 489 preferentially on large prey types, these may have been depleted before the end of the 490 experiment, potentially biasing feeding rate estimates. However, due to the low in situ abundance 491 of large particles, feeding on such particles would also be limited in the field. As observed here, 492 size-selective copepod grazing can induce shifts in the size distribution of the plankton 493 community, which may have important implications regarding the potential for particles 494 gravitational sinking and the associated vertical flux of carbon (Stemmann & Boss, 2012). 495 In most instances when comparisons between copepod and microzooplankton grazing are 496 possible, the grazing impact by the unicellular herbivores typically exceeds that of the 497 mesozooplankton component at least 2-3 fold (e.g. Morales et al., 1991;Burkill et al., 1993;498 Dam, Miller & Jonasdottir, 1993;Weeks et al., 1993;Gifford et al., 1995). In our study, 499 however, the relative grazing impact of both types of grazers varied spatially. At S1, 500 microzooplankton grazing removed 97% of PP, exceeding the ~66% estimated global average of 501 the proportion of PP removed by herbivorous protists (Calbet & Landry, 2004). Perhaps due to 502 their diverse feeding modes, microzooplankton can indeed have a considerable impact on PP.  (Gifford et al., 1995). Estimates of the 507 aforementioned studies on daily PP removal by microzooplankton ranged from 15-242 %, with 508 the highest average impact (81 %) being attributed to after-bloom conditions (Gifford et al., 509 1995). 510 In contrast to the high microzooplankton grazing impact at S1, no microzooplankton 511 grazing was observed at S2 and remained undetermined at S4. Absence of grazing is not 512 uncommon, and has been recorded in all ocean ecosystems, from estuaries to the coastal and 513 open ocean, and in all ocean basins at all latitudes (Schmoker, Hernàndez-Leòn & Calbet, 2013 514 and references therein). Lack of grazing could suggest that grazers' biomass had not sufficiently 515 accumulated to have a detectable effect on PP as hypothesized by . Such 516 biomass buildup can be delayed if grazers have been exposed to prolonged periods of starvation 517 The 525 negative microzooplankton grazing rates obtained at S4 prevented any estimation of 526 microzooplankton grazing impact. It has been suggested that the presence of chloroplasts in 527 mixotrophs can artificially increase the Chl a-based apparent growth rate in the undiluted 528 treatment, resulting in a positive slope (Landry, Constantinou & Kirshtein, 1995;Calbet et al., 529 2012). Unfortunately the impact -if any-that this process may have had on grazing rate 530 estimation at S4 is difficult to evaluate, as assessing the proportion of mixotrophs in a plankton 531 assemblage that engage in phagotrophy remains challenging (Beisner, Grossari & Gasol, 2019). 532 Although our experimental design controlled for light effects on mesozooplankton-533 grazing rates, no such effect was observed in our experiments at S4. Many copepods exhibit diel 534 periodicity in their feeding (Durbin et al., 1990; that would negate some effects of light, 535 however, under some conditions copepods may feed during daytime (Atkinson et al., 1992). 536 Nonetheless, our observations contrast with previous studies that have shown significantly higher 537 copepod feeding rates at lower irradiances, indicating that light cues may influence both the 538 timing of grazing and the gut fullness in certain copepod species (Stearns, 1986; Cieri & Stearns, 539 1999). In contrast, as expected, there was a significant effect of light on phytoplankton growth 540 rates, both in the presence and absence of copepods. Due to differences in phytoplankton growth 541 in response to different light intensity, the mesozooplankton grazing impact on PP estimated as 542 the fraction of total growth removed was almost three times greater at lower light. Net 543 phytoplankton growth rates at S4 were not adjusted for photoacclimation. Thus at the lower light 544 intensity, net phytoplankton growth rates may have been over-estimated, if a decrease in light 545 resulted in phytoplankton increasing their cellular pigment content. This lack of adjustment, 546 however, does not change our conclusions that light did not influence mesozooplankton-grazing 547 rates. Indeed any photoacclimation process would have affected net phytoplankton growth rates 548 in all treatments similarly, and thus differences between the copepod treatments and the controls 549 with or without adjustment would remain unchanged. On the other hand, if the net phytoplankton 550 growth rates were overestimated at the low light intensity, then the difference in net growth rates 551 between the high and the low light treatments may have been underestimated, magnifying the 552 greater effect of light than of grazing on phytoplankton accumulation. The differential effect of 553 light on growth and grazing rates observed here suggests that in the ocean, the exponential 554 decrease of light with depth acts to decouple growth and grazing, resulting in a vertically 555 increasing grazing impact on PP. 556 557 Conclusions 558 559 Concurrent investigation of mesozooplankton and microzooplankton grazing impact on 560 phytoplankton primary production in the western North Atlantic suggested a generally limited 561 potential control of phytoplankton biomass by herbivorous copepods, sometimes but not always 562 exceeded by microzooplankton grazing. Although the observations reported here were acquired 563 opportunistically and are limited in sampling scope, they represent a diverse range of 564 environmental conditions and species composition from a poorly sampled region. The generally 565 low mesozooplankton grazing impact across stations despite the spatial diversity of species and 566 conditions, as well as the good agreement with similar studies in other regions, provide important 567 constraints to quantifying grazer-induced phytoplankton mortality rates for this region and the 568 NAAMES spring campaign. There was no indication of preferential feeding on 569 microzooplankton by larger copepods. Remarkably, when compared, light effects on net 570 phytoplankton growth rates exceeded mesozooplankton grazing effects by at least 3-fold. 571 Together these coupled biological and environmental effects provide insights into the transfer 572 and recycling of recently fixed carbon. Specifically, they suggest limited vertical export of 573 phytoplankton production as mediated by large zooplankton. Such concurrent environmental and Phytoplankton growth and grazing rates in the western North Atlantic in May 2016.
Chlorophyll a (Chl a, µg L -1 ) concentration in source water used for the experiments (± SD), light intensity used in the incubations (% of incoming surface irradiance), and estimated rates of in situ phytoplankton instantaneous growth (µ NoN), microzooplankton (mzpkt) grazing (g), net phytoplankton growth rates in nutrient amended <200 µm seawater (k +N), non-amended <200 µm seawater (k NoN), and copepod additions (k Cop). Rates are given per day ± one standard deviation of duplicate or triplicate treatments. For copepod treatments, A, B, and C represent the different number of copepods added (A=1, B= 3, C=5 for C. hyperboreus, A=10, B= 20, C=40 for C. finmarchicus). n/d means rate was undetermined.