Dual benefit of ocean acidification for the laminarialean kelp Saccharina latissima: enhanced growth and reduced herbivory

The laminarialean kelp Saccharina latissima is a common macroalga along rocky shorelines that is also frequently used in aquaculture. This study examined how ocean acidification may alter the growth of S. latissima as well as grazing on S. latissima by the gastropod Lacuna vincta. Under elevated nutrients, S. latissima experienced significantly enhanced growth at pCO2 levels ≥1200 μatm compared to ambient pCO2 (~400 μatm). Elevated pCO2 (≥830 μatm) also significantly reduced herbivory of L. vincta grazing on S. latissima relative to ambient pCO2. There was no difference in grazing of S. latissima previously grown under elevated or ambient pCO2, suggesting lowered herbivory was due to harm to the gastropods rather than alteration of the biochemical composition of the kelp. Decreased herbivory was specifically elicited when L. vincta were exposed to elevated pCO2 in the absence of food for ≥18 h prior to grazing, with reduced grazing persisting 72 h. Elevated growth of S. latissima and reduced grazing by L. vincta at 1200 μatm pCO2 combined to increase net growth rates of S. latissima more than 4-fold relative to ambient pCO2. L. vincta consumed 70% of daily production by S. latissima under ambient pCO2 but only 38 and 9% at 800 and 1200 μatm, respectively. Collectively, decreased grazing by L. vincta coupled with enhanced growth of S. latissima under elevated pCO2 demonstrates that increased CO2 associated with climate change and/or coastal processes will dually benefit commercially and ecologically important kelps by both promoting growth and reducing grazing pressure.


INTRODUCTION
Ocean acidification is changing marine ecosystems. As anthropogenic CO 2 accumulates in the atmosphere and surface oceans, levels of pH, CO 3 2− , and the saturation states of calcium carbonate are de clining ). In addition, many coastal ecosystems can experience partial pressure of CO 2 (pCO 2 ) levels not projected to occur in open ocean systems until the year 2100 (>1000 μatm) due to a multitude of processes, including upwelling (Feely et al. 2008), riverine discharge (Vargas et al. 2016), macrophyte respiration (Wahl et al. 2018), and eutrophication-enhanced microbial respiration (Wallace et al. 2014). These changes in ocean chemistry stand to reorganize the function of coastal ecosystems, as acidification can be inhibitory to some marine animals (Poloczanska et al. 2016), especially calcifiers (Talmage & Gobler 2010, Young et al. 2019), but may benefit other ocean organisms (Koch et al. 2013, Young & Gobler 2016. Saccharina latissima (also known as sugar kelp) is a bladed, cold-water brown macroalga that is widely distributed across the North Atlantic, Pacific, and Arctic Oceans (Brinkhuis et al. 1983, Sivertsen & Bjørge 2015. S. latissima can form dense, highly productive and biologically diverse beds that provide numerous ecosystem services, including nursery habitat, refuge from predators, coastal defense, carbon and nitrogen sequestration, and food sources for other organisms (Norderhaug et al. 2005, Chung et al. 2013, Smale et al. 2013. More recently, the commercial importance and aquaculture of S. latissima and other species of kelp has grown (Marinho et al. 2015). Excessive grazing, however, can significantly lower kelp abundance and ecosystem function (Scheib ling et al. 1999, Norderhaug & Christie 2009). Moreover, global distributions of S. latissima are declining due to increasing temperatures (Filbee-Dexter et al. 2016), eutrophication (Moy & Christie 2012), and overfishing of the predators of kelp grazers (Steneck et al. 2002). However, recent evidence suggests that elevated pCO 2 associated with ocean acidification could benefit some species of kelp, potentially counteracting the processes contributing to their decline (Hepburn et al. 2011).
Beyond climate change, eutrophication may also act to negatively impact kelp communities. In some temperate regions, excessive nutrient loading can initiate a succession whereby kelp forests become overgrown by turf algae (Eriksson et al. 2002, Connell et al. 2008) and this succession may be accelerated by acidification (Connell & Russell 2010). Still, kelp can also exist in ecosystems where turf algae are rare, and kelp can be purposely grown in more eutrophic locales for aquaculture purposes (Kim et al. 2015, Jiang et al. 2020. While climate change and eutrophication are strong environmental drivers in coastal ecosystems, the manner in which nutrients and acidification act, and potentially interact, to directly alter kelp growth is not fully clear. The northern Lacuna snail Lacuna vincta (Gastropoda) is a common grazer in coastal North Atlantic and North Pacific ecosystems (Chavanich & Harris 2002, Janiak & Whitlatch 2012 and is a well-known grazer of macroscopic kelp sporophytes (Brady-Campbell et al. 1984, Johnson & Mann 1986). Grazing by large populations of L. vincta can cause extensive damage to kelp blades in kelp beds (Chenelot & Konar 2007, Krumhansl & Scheibling 2011, consuming up to 10% of the surface area of S. latissima (Molis et al. 2010, Krumhansl & Scheibling 2011. The resulting damage done by the grazer on the kelp blade, meristem, and stipe can significantly lower the tensile strength of the kelp, making it more susceptible to breakage (Chenelot & Konar 2007, Molis et al. 2010. High grazing intensity by L. vincta on the reproductive sorus tissue of S. latissima may exacerbate recruitment limitation and further hinder the recovery of degraded kelp beds (O'Brien & Scheibling 2016). Compared to other macroalgae native to the northern Atlantic Ocean, laminarialean kelps are preferred by L. vincta due to overall lower phlorotanins (anti-grazing defense) and relatively high palatability (Wakefield & Murray 1998). While ocean acidification can disrupt the grazing by L. vincta on some macroalgae (Young et al. 2019), the manner in which L. vincta herbivory on kelps might be altered by this process is unknown.
The overarching objective of this study was to assess how coastal acidification may affect Northwest Atlantic populations of S. latissima. Experiments quantified the growth rates as well as elemental and isotopic content of S. latissima under treatment levels of pCO 2 with and without nutrients. Experiments were also performed to quantify herbi vory rates of L. vincta on S. latissima under normal and elevated levels of pCO 2 . Given that prior re search has determined that the effects of acidification on grazing by L. vincta are dose-dependent with regard to both pCO 2 levels and exposure duration (Young et al. 2019), experiments exploring differing levels of pCO 2 and differing durations of exposure were performed. Given that the effects of acidification on invertebrates can depend on food supply (Melzner et al. 2011, Thomsen et al. 2013, Pansch et al. 2014), exposures to acidification were made following periods of ad libitum and restricted feeding. Finally, the net growth rates (growth minus grazing) of S. latissima were determined under treatment pCO 2 conditions, providing a novel examination of the net effects of ocean acidification on a keystone macrophyte. While studies exploring how ocean acidification affects herbivory have been limited, to our knowledge this is the first study to assess how ocean acidification concurrently alters growth and grazing for a macrophyte.

Collection and preparation of Lacuna vincta and Saccharina latissima
Lacuna vincta used for this study were collected by hand during low tide from Shinnecock Bay, NY, USA (40.85°N, 72.50°W), part of New York's South Shore Estuary Reserve (NYSSER) (Fig. S1 in the Supplement at www. int-res. com/ articles/ suppl/ m664 p087_ supp. pdf). L. vincta are an abundant macroalgae grazer in estuaries of the northeastern USA (Chavanich & Harris 2002, Janiak & Whitlatch 2012 and were identified based on morphology. Saccharina latissima used in the study was cultured from blades collected from Long Island Sound, spawned on line, and grown along horizontal longlines at a commercial oyster farm (i.e. Great Gun Shellfish) in Moriches Bay, NY (40.78°N, 72.78°W), a lagoon contiguous with Shinnecock Bay to the east. Large, well-pigmented blades of S. latissima were selected from samples collected by hand at low tide (Fig. S1). S. latissima were cut from their holdfasts as close to the longlines as possible. Following collection, S. latissima and L. vincta were immediately placed in seawater-filled containers and transported to the Stony Brook Southampton Marine Science Center of Stony Brook University within 30 min of collection. Upon arrival to the facility, L. vincta were placed in a 20 l polycarbonate vessel filled with filtered (0.2 μm) seawater taken from the collection site. The vessel was supplied with air and recently collected S. latissima as a food source until experiments were initiated (Duffy et al. 2014). S. latissima used in experiments were placed in a large, round ~2000 l tank filled with flowing, 1 μm filtered seawater from Shinnecock Bay.
For experiments performed to quantify algal growth rates in response to elevated pCO 2 and/or nutrients, individual S. latissima rectangular sporophytes were prepared by cutting the stipe 2.5 cm below the blade− stipe interface and cutting the blade 5 cm above the blade− stipe interface. This was done to standardize initial kelp blade tissue type and size (Boderskov et al. 2016). S. latissima samples were weighed on a Scientech ZSA 120 digital microbalance (± 0.0001 g) to obtain initial fresh weight in grams. For experiments performed to quantify grazing rates on S. latissima, rectangular sections (2 × 4 cm, length × width) of the algae were cut from large blades with care taken to ensure uniformity of size, shape, and tissue type. Sections from the upper blades of S. latissima were used due to L. vincta's preference for this section of the organism over the lower blade, meristem, or stipe (Molis et al. 2010). All samples were spun in a salad spinner to remove debris and epiphytes, extensively rinsed with filtered (0.2 μm) seawater before being spun again to further remove any remaining debris, epiphytes, and excess seawater (Young & Gobler 2016), and weighed as described above.

Preparation of experiments
Two experiments were performed to assess the effects of pCO 2 and nutrients on the growth rates of S. latissima and 5 experiments were performed to assess the effects of elevated pCO 2 on the herbivory rates of L. vincta on S. latissima. Each experiment was performed in 1 l polycarbonate vessels that were acid-washed (10% HCl) and liberally rinsed with deionized water. All experimental containers were placed in an environmental control chamber set to a temperature (~10−18°C), light intensity (~250 μmol photons m −2 s −1 ), and duration (12 h light:12 h dark cycle) that matched ambient conditions at the collection site to allow for optimal conditions for L. vincta and S. latissima. As temperatures rose through the spring, so did the temperatures used for experiments to minimize thermal shock to the kelp and snails. All containers were filled with filtered seawater, randomly placed within the environmental control chamber, and randomly assigned in quadruplicate to each treatment, which varied based on the experiment performed. For grazing experiments (see below), 2 additional containers were filled with filtered seawater and assigned, without L. vincta, to assess S. latissima residual growth (i.e. additional growth the S. latissima samples experienced during the grazing period). During those experiments, the lights of the environmental control chamber were turned off 24 h prior to the introduction of Lacuna in order to minimize residual S. latissima growth (Nelson et al. 2008, Young et al. 2019. For each grazing experiments, 8 L. vincta (~3 mm) were added to each separate container, mimicking densities found on macroalgae at the collection site (0.5−1 grazer cm −2 ) and reported in the literature (Chenelot & Konar 2007, Dubois & Iken 2012, Young et al. 2019. Dissolved gases were delivered into each experimental container via air diffusers (Pentair) connected to 1 ml polystyrene serological pipettes inserted into the bottom of each vessel and connected via Tygon tubing to an air source. The containers that were subjected to treatment CO 2 conditions utilized multitube gas proportioner systems (Cole Parmer ® Flowmeter) to mix ambient air with 5% CO 2 (Talmage & Gobler 2010). The containers subjected to ambient conditions utilized single tube proportioner systems to introduce only ambient air into the containers. The gases were mixed through gang valves and were de livered at a flow rate of 2500 ± 5 ml min −1 to experimental containers through serological pipettes inserted through plexiglass covers on the containers. The bubbling rates in the containers turned over the volume >1000 times d −1 , and bubbling was initiated at least 2−3 d prior to the initiation of experiments to allow CO 2 levels and carbonate chemistry to reach a state of equilibrium. A Honeywell DuraFET III ion-sensitive field effect transistor-based (ISFET) solid-state pH sensor (± 0.01 pH unit, total scale) was used to measure pH within containers each day of the ex periments. Dissolved inorganic carbon (DIC) within experimental vessels was measured directly from water samples that were collected at the beginning and end of experiments and preserved using a saturated mercuric chloride (HgCl 2 ) solution and stored at ~4°C until they were analyzed on a VINDTA 3D (Versatile Instrument for the Determination of Total inorganic carbon) delivery system coupled within a UIC Inc. coulometer (model CM5017O). During the coulometric ana lyses, all inorganic carbon species were converted to CO 2 gas by the addition of excess hydro gen to the sample, and the evolved CO 2 gas was subsequently carried into the titration cell of the coulometer. The gas then reacted quantitatively with an ethanolamine-based reagent to generate hydrogen ions, which are titrated with coulometrically generated OH − , and CO 2 was measured by integrating the total change required to titrate the hydrogen ions (Johnson et al. 1993). The pCO 2 concentrations (Table 1) were calculated from measured levels of DIC, pH, temperature, salinity, and the first and second dissociation constants of carbon acid in seawater (Millero 2010) using the program CO2SYS (http://cdiac.ess-dive. lbl.gov/ftp/co2sys/). Certified reference material (CRM) provided by Dr. Andrew Dickson (University of California, San Diego, Scripps Institution of Oceanography; Batch 180 = 2021.87 μmol DIC kg −1 seawater) was used as a quality assurance measure, and ana lyses only proceeded when recovery of the CRM was 99.8− 100%. Final DIC concentration of the CRM was 2019.85 μmol DIC kg −1 sea water.

Assessing the effects of elevated nutrients and pCO 2 on S. latissima
To quantify the effects of elevated pCO 2 and/or nutrients on the growth of S. latissima, 2 experiments were performed. In the first experiment (designated as 'Co-effects of pCO 2 and nutrients'; Table 2), the goal was to assess how high and low pCO 2 and nutrients altered the growth rates of S. latissima. Estuaries are dynamic environments where levels of nutrients and pCO 2 vary in time and space (Wallace et al. 2014, R. B. Wallace et al. unpubl. data). Given prior studies (Olischläger et al. 2012, Zhang et al. 2020, we hypothesized that elevated pCO 2 would increase S. latissima growth, but that ele-  (Connell et al. 2008). S. latissima was placed in 1 of 4 treatments established in quadruplicate (for this and all experiments): a control with ambient pCO 2 levels (350− 450 μatm) and no nutrient additions, a treatment with ambient pCO 2 and nutrient additions (50 μM nitrate, 3 μM phosphate), a treatment with elevated pCO 2 levels (~1800− 2100 μatm) and no nutrient additions, and a treatment with elevated pCO 2 and nutrient additions. For this experiment, S. latissima samples were incubated under these conditions for 7 d. The elevated nutrients and pCO 2 concentrations were higher than levels present at the collection site (0− 10 μM nitrate, 0−1 μM phosphate, 400−800 μM pCO 2 ), but consistent with concentrations present in some US East Coast estuaries, including those in the nearby New York region during winter and spring (Gobler et al. 2006, Wallace et al. 2014. For example, total dissolved nitrogen concentrations exceeding 50 μM have been reported in Quantuck Bay (Gobler et al. 2011), the estuary contiguous with and 10 km from the S. latissima collection site, and pCO 2 levels in the range of 1000− 3000 μatm have recently been re ported for Shinne cock Bay and the Peconic Estuary (Wallace et al. unpubl. data), systems contiguous with the S. latissima collection site. Given the ability of future ocean acidification to synergistically depress pH values when coupled with eutrophicationdriven acidification (Sunda & Cai 2012), we expect future climate change scenarios to increase regional pCO 2 concentrations to levels higher than presently observed. For the second experiment ('pCO 2 dose response'; Table 2), the goal was to identify the minimum level of pCO 2 needed to yield enhanced growth rates considering the levels currently present in regional estuaries (Wallace et al. 2014, Wallace et al. unpubl. data) and levels that may be present in the coming centuries due to climate change (IPCC 2014). We hypo thesized that, like other carbon-limited autotrophs (Raven et al. 2020), growth rates would increase linearly with in creasing levels of pCO 2 . S. latissima samples were placed in 1 of 4 treatments: a control with ambient pCO 2 (350−450 μatm), a treatment with 750− 850 μatm pCO 2 , a treatment with 1200− 1500 μatm pCO 2 , and a treatment with 1800− 2100 μatm pCO 2 . While the higher pCO 2 levels used in this and the prior experiment exceed 21 st century climate change projections, they capture the range of levels observed in coastal ecosystems influenced by eutrophication (Wallace et 91  (Feely et al. 2008), macrophyte respiration (Wahl et al. 2018), or riverine discharge (Vargas et al. 2016). Each treatment was supplied with nutrients (50 μM nitrate, 3 μM phosphate), and S. latissima samples were incubated for ~2 wk. At the end of the incubation periods for this and the first experiment, samples were removed from their respective treatments, weighed as described above, and final length and width measurements were made.
Weight-based growth rates were determined by the following formula: (1) where W final and W initial are the final and initial fresh weights, in grams, and Δt is the duration of the experiment in days.

Herbivory by L. vincta on S. latissima
Five herbivory experiments were performed, the first of which (designated as 'Effects of high and low pCO 2 on herbivory') gauged the herbivory rates of L. vincta feeding on S. latissima under ambient (350− 400 μatm) and elevated (1800−2100 μatm) pCO 2 levels (Table 2). We hypothesized that L. vincta herbivory rates would be lower under higher pCO 2 levels (Young et al. 2019). L. vincta were placed in ambient or elevated pCO 2 conditions without S. latissima and starved for 24 h and then placed into containers with S. latissima (never exposed to elevated pCO 2 ) under ambient or elevated pCO 2 for 48 h. At the end of the grazing period, L. vincta and S. latissima were removed from the containers and S. latissima samples were weighed as described above. Herbivory rates were calculated by obtaining the difference in the initial and final corrected weights divided by the number of grazers and the elapsed time of the grazing period and multiplied by 1000 to convert weights to mg (mg grazer −1 d −1 ). Final weights were corrected using residual growth of the additional S. latissima samples grown without L. vincta (see above).
The next experiment ('Direct vs indirect effects of elevated pCO 2 on herbivory'; Table 2) was performed to determine if lowered herbivory rates of L. vincta under high pCO 2 were caused by direct ef fects on L. vincta or indirectly by altering the palatability of S. latissima. Given prior studies of this snail, we hypothesized that L. vincta herbivory rates would be lowered due to direct exposure to high pCO 2 levels and not due to exposure of S. latissima to high pCO 2 (Young et al. 2019). L. vincta were starved under am-bient (350−450 μatm) or elevated (1800− 2100 μatm) pCO 2 levels and feed S. latissima incubated under either ambient or elevated pCO 2 , with the intent of placing one-half of the L. vincta from each pCO 2 group in either the pCO 2 level they were starved in or the opposite pCO 2 group. For this experiment, S. latissima was incubated for ~1 wk under ambient or elevated pCO 2 ; on the final day of the incubation period, L. vincta were starved in separate vessels under ambient or elevated pCO 2 without S. latissima for 24 h. At the end of the 1 wk S. latissima incubation period and concurrent 24 h L. vincta starvation period, the lights of the environmental chamber were turned off, and L. vincta were introduced into the vessels containing S. latissima for a total of 4 treatments: a control with ambient pCO 2 containing L. vincta starved under ambient pCO 2 and allowed to graze on S. latissima incubated under ambient pCO 2 , a treatment with elevated pCO 2 containing L. vincta starved under ambient pCO 2 and allowed to graze on S. latissima incubated under elevated pCO 2 , a treatment with ambient pCO 2 containing L. vincta starved under elevated pCO 2 and allowed to graze on S. latissima incubated under ambient pCO 2 , and a treatment with elevated pCO 2 containing L. vincta starved under elevated pCO 2 and allowed to graze on S. latissima incubated under elevated pCO 2 . Once in their respective containers, L. vincta were allowed to graze for 96 h, with S. latissima samples being replaced every 24 h with the same source of S. latissima (high or ambient pCO 2 exposure). At the end of each 24 h grazing period, S. latissima samples were removed from the containers and were weighed as described above and herbivory was calculated.
Given that the effects of acidification on invertebrates can depend on food supply (Melzner et al. 2011, Thomsen et al. 2013, Pansch et al. 2014, for the next experiment ('Co-effects of elevated pCO 2 and food restriction on herbivory'; Table 2), herbivory rates of L. vincta were quantified on individuals that were either starved or fed under ambient (350− 450 μatm) or elevated pCO 2 (1800−2100 μatm) for 24 h. Given prior studies, we hypothesized that L. vincta exposure to high pCO 2 during the starvation period would yield lowered herbivory rates while exposure during the grazing period would not (Young et al. 2019). There were 4 treatments: starved under ambient pCO 2 , fed at ambient pCO 2 , starved under high pCO 2 , or fed at high pCO 2 . After 24 h, fresh S. latissima blade portions were placed in the containers and L. vincta were allowed to graze for 24 h. At the end of the grazing periods, L. vincta and S. latissima were removed from the containers and S. latissima samples were weighed as described above and herbivory was calculated.
The final experiments were performed to determine how the intensity and duration of CO 2 exposure altered herbivory of L. vincta grazing on S. latissima. These experiments specifically assessed the effective minimum duration of elevated pCO 2 required to alter herbivory rates ('Effective minimum exposure duration'; Table 2) and the effective minimum concentration of pCO 2 to alter herbivory rates of L. vincta ('Effective minimum dose'; Table 2). Given that pCO 2 concentrations can vary diurnally within shallow estuaries (Baumann et al. 2015) and that even short-term exposure to acidification can disrupt herbi vory (Young et al. 2019), the 'effective minimum exposure duration' experiment was designed to identify the minimum duration exposure needed to disrupt herbivory in L. vincta. We hypothesized that exposure to high pCO 2 for less than 24 h would still lower grazing rates (Young et al. 2019). There were 5 treatments in this experiment: a control with ambient pCO 2 (350−450 μatm) during the 24 h starvation period (no dose), a treatment with elevated pCO 2 (1800−2100 μatm) during the entire 24 h starvation period, and treatments with 6, 12, and 18 h of exposure to elevated pCO 2 during the starvation period with 18, 12, and 6 h, respectively, of ambient pCO 2 exposure prior to exposure to elevated pCO 2 . At the end of the starvation period, L. vincta were placed in containers with ambient pCO 2 levels, S. latissima samples were introduced, and L. vincta were allowed to graze for 24 h, after which herbivory rates were calculated. Given that pCO 2 levels can be dynamic in estuaries (Wallace et al. 2014) and that levels of pCO 2 are expected to more than double this century, the 'effective minimum dose' experiment was conducted to assess how levels of pCO 2 found in estuaries today (Wallace et al. 2014) as well as those projected for the future in less impacted regions (IPCC 2014) affect herbivory by L. vincta. Given prior studies with L. vincta, we hypothesized that moderately elevated (>1500 μatm) concentrations of pCO 2 would significantly lower herbivory rates (Young et al. 2019). This experiment established 4 treatments: a control with ambient pCO 2 (350−450 μatm), a treatment with slightly elevated pCO 2 (750−850 μatm), a treatment with moderately elevated pCO 2 (1200− 1500 μatm), and a treatment with high pCO 2 (1800− 2100 μatm). During this experiment, L. vincta were placed in 1 of the 4 treatments and starved for 24 h. At the end of the starvation period, S. latissima samples were introduced, and L. vincta were allowed to graze for 24 h at their respective pCO 2 levels. At the end of the grazing period, L. vincta and S. latissima were removed from the containers and S. latissima samples were weighed as described above and herbivory was calculated.

Post-experimental analyses
For carbon (C) and nitrogen (N) analyses of S. latissima, frozen samples were dried at 60°C for 24 h, and then homogenized into a fine powder with a mortar and pestle. Tissue C, N, and δ 13 C were analyzed using an elemental analyzer interfaced to a Europa 20-20 isotope ratio mass spectrometer at the UC Davis Stable Isotope Facility. The measured δ 13 C levels of the CO 2 gas (−80 ‰) used in experiments and isotopic mixing models (Young & Gobler 2016 were used to identify the relative use of CO 2 and HCO 3 − by S. latissima. Net algal growth rates were calculated using growth rates and total herbivory rates from the 'pCO 2 dose response' and 'effective minimum dose' experiments, respectively. The herbivory rates for each replicate in each pCO 2 treatment in the 'effective minimum dose' experiment were calculated by subtracting the final S. latissima fresh weight from the initial fresh weight. We note that the S. latissima in this herbivory experiment were all raised at ambient pCO 2 . The mean total herbivory rate for each pCO 2 treatment was subtracted from the S. latissima growth rates in the corresponding pCO 2 treatment in the 'pCO 2 dose response' experiment to obtain the net growth rate (growth minus herbivory) for 4 pCO 2 levels.
One-way ANOVAs were performed within Sigma -Plot 11.0 to assess significant differences in herbivory rates in the 'pCO 2 dose response', 'effects of high and low pCO 2 on herbivory', 'effective minimum exposure duration', and 'effective minimum dose' experiments (n = 4 treatment −1 for each experiment). Two-way ANOVAs were performed with SigmaPlot to assess herbivory rates in the 'co-effects of pCO 2 and nutrients', 'direct vs. indirect effects of elevated pCO 2 on herbivory', and 'co-effects of elevated pCO 2 and food restriction on herbivory' experiments (n = 4 treatment −1 for each experiment), where the main treatment effects were pCO 2 and nutrient levels (ambient or elevated for both) for the 'co-effects of pCO 2 and nutrients' experiment, pCO 2 level for the S. latissima incubation and L. vincta starvation periods (ambient and elevated for both) for the 'direct vs. indirect effects of elevated pCO 2 on herbivory' experiment, and pCO 2 level (ambient or elevated) and starvation (starved or not starved) for the 'co-effects of elevated pCO 2 and food restriction on herbivory' experiment. Normality and equal variance were tested via the use of Shapiro-Wilk and Leven tests within SigmaPlot 11.0; assumptions of equal variance and normality were met for all data. For all experiments, if sig ni ficant differences were detected, a Tukey's HSD test using R v.3.4.0 within RStudio v.1.0.143 was performed (R Core Team 2020, RStudio Team 2020).

Effects of elevated pCO 2 and nutrients on Saccharina latissima growth rates
In the 'co-effects of pCO 2 and nutrients' experiment, growth rates of Saccharina latissima were significantly higher in elevated than in ambient pCO 2 by ~70% (2-way ANOVA and Tukey's HSD, p < 0.05; Fig. 1, Tables S1 & S2 in the Supplement). Furthermore, growth rates were significantly higher bỹ 50% under elevated nutrient conditions relative to treatments that did not receive nutrient additions (2-way ANOVA and Tukey's HSD, p < 0.05; Fig. 1, Tables S1 & S2), and there was no interaction between nutrient additions and pCO 2 levels (Fig. 1).

Herbivory rates of Lacuna vincta grazing on S. latissima
During the 'effects of high and low pCO 2 on herbivory' experiment, which exa mined how high and low pCO 2 affected herbivory of Lacuna vincta that were fed S. latissima, grazing rates in the elevated pCO 2 treatment were signifi-cantly reduced by 60% relative to the ambient pCO 2 after 48 h (1-way ANOVA, p < 0.05; Fig. 4, Table S10). For the 'direct vs. indirect effects of elevated pCO 2 on herbivory' experiment, L. vincta herbivory rates were sensitive to pCO 2 levels during the starvation period, but not the pCO 2 conditions that S. latissima were incubated under (Fig. 5). Specifically, for the 24, 48, and 72 h timepoints, herbivory was significantly decreased by~78,~86, and~94% when Lacuna were starved under elevated pCO 2 relative to ambient conditions (2way ANOVA, p < 0.05 for all; Fig. 5, Table S11). In contrast, herbivory was not significantly altered by the pCO 2 conditions in which S. latissima were grown under throughout the experiment (2-way ANOVA, p > 0.05 for all; Fig. 5, Table S11). After 96 h, herbi vory rates of L. vincta ex posed to elevated pCO 2 recovered and were no longer affected by the pCO 2 levels it had been ex posed to during the starvation period (2-way ANOVA, p > 0.05 for both; Fig. 5, Table S11).
In the 'co-effects of elevated pCO 2 and food restriction on herbivory' experiment, pCO 2 levels and being starved or fed prior to grazing was examined in a 2 × 2 experimental design. While herbivory rates of L. vincta on S. latissima during this experiment were not significantly affected by whether snails were starved under ambient pCO 2 (2-way ANOVA, p > 0.05; Fig. 6, Tables S12 & S13), exposure to elevated pCO 2 during the starvation period significantly de pressed herbivory rates of starved snails (2-way ANOVA and Tukey's HSD, p < 0.05; Fig. 6, Tables S12 & S13). Within the elevated pCO 2 treatments, L. vincta that were starved had herbivory rates that were ~85% lower than individuals that were fed throughout experimentation (Tukey's HSD, p < 0.05; Fig. 6, Tables S12 & S13).
During the 'effective minimum exposure duration' experiment, exposure to elevated pCO 2 levels for 6 95 Fig. 3. Mean (± SD) tissue δ 13 C content of Saccharina latissima grown under treatment pCO 2 levels with nutrient additions for the (A) 'Co-effects of pCO 2 and nutrients' and (B) 'pCO 2 dose response' experiments, respectively (  Table 3. Tissue carbon (mg C mg −1 dry tissue), nitrogen (mg N mg −1 dry tissue), and the molar carbon:nitrogen (C:N) ratios of Saccharina latissima grown under ambient or elevated pCO 2 with and without nutrient additions ('Co-effects of pCO 2 and nutrients') or under treatment pCO 2 levels with nutrient additions ('pCO 2 dose response'; Table 2) or 12 h did not alter herbivory rates compared to the control (1-way ANOVA and Tukey's HSD, p > 0.05 for all; Fig. 7, Tables S14 & S15) but 18 and 24 h exposure did, yielding rates significantly lower than the 0, 6, and 12 h exposure treatments (1-way ANOVA and Tukey's HSD, p < 0.05 for all; Fig. 7, Tables S14 & S15). There was no significant difference in herbivory between the 18 and 24 h exposure treatments (1-way ANOVA and Tukey's HSD, p < 0.05; Fig. 7, Tables S14 & S15). Finally, during the 'effective minimum dose' experiment, pCO 2 concentrations reduced herbivory rates of L. vincta feeding on S. latissima in a dose-dependent manner. Herbivory rates within the ~830, 1420, and ~2450 μatm pCO 2 treatments were all significantly lower than in the ambient (~450 μatm) pCO2 treatment (1-way ANOVA and Tukey's HSD, p < 0.05 for all; Fig. 8, Tables S16 & S17). While herbivory rates in the ~830 μatm pCO 2 treatment were significantly higher than in the ~1420 and ~2450 μatm pCO 2 treatments (1-way ANOVA and Tukey's HSD, p < 0.05 for both; Fig. 8, Tables S16 & S17), there was no significant difference in herbivory between the 1420 and 2450 μatm pCO 2 treatments (1-way ANOVA and Tukey's HSD, p > 0.05; Fig. 8, Tables S16 & S17).  Table 2). Statistical analyses: 2-way ANOVA and post hoc Tukey's HSD tests performed for each 24 h grazing period (n = 4 for all treatments). *p < 0.05 between treatments for each 24 h grazing period relative to the control (ambient pCO 2 snail starvation → ambient pCO 2 kelp incubation) Fig. 6. Mean (SD) herbivory rates of Lacuna vincta either starved or not starved for 24 h under ambient or elevated pCO 2 levels and allowed to graze on Saccharina latissima for 24 h ('Co-effects of elevated pCO 2 and food restriction on herbivory'; Table 2). Statistical analyses: 2-way ANOVA and post hoc Tukey's HSD tests (n = 4 for all treatments); significant differences (p < 0.05) between treatments indicated by letters Fig. 7. Mean (SD) herbivory rates of Lacuna vincta starved for various lengths of time under ambient and elevated pCO 2 levels and allowed to graze on Saccharina latissima under ambient pCO 2 for 24 h ('Effective minimum exposure duration'; Table 2). Statistical analyses: 1-way ANOVA and post hoc Tukey's HSD tests (n = 4 for all treatments); significant differences (p < 0.05) between treatments indicated by letters

DISCUSSION
Ocean acidification is restructuring the function of coastal ecosystems. Given the potential for elevated pCO 2 to promote the growth of some autotrophs and to negatively affect some calcifying organisms, interactions between herbivorous calcifiers and macroalgae may be significantly altered by ocean acidification. During the present study, growth rates of the ecologically and economically important kelp species Saccharina latissima were enhanced by elevated pCO 2 levels, while herbivory rates of Lacuna vincta grazing on S. latissima were reduced when exposed to elevated pCO 2 and deprived of food prior to exposure. This study demonstrates, to our knowledge for the first time, the ability of ocean acidification to benefit the growth of a macrophyte by concurrently altering both top-down (herbivory) and bottom-up (growth) processes.
During this study, S. latissima growth was significantly increased under elevated pCO 2 , an outcome consistent with prior studies (Hepburn et al. 2011, Olischläger et al. 2012, Zhang et al. 2020. While these effects could have been temperature-dependent, our use of in situ temperatures from our study site did not allow for detection of such trends. The physiological response of macroalgae to elevated pCO 2 levels is largely dependent on its mode of carbon uptake and the extent to which its inorganic carbon uptake is substrate-saturated at current pCO 2 levels (Koch et al. 2013). When exposed to elevated pCO 2 levels, macroalgae may be relived of C limitation and/or may downregulate carbon-concentrating mechanisms that convert HCO 3 − to CO 2 , allowing for more energy to be available for vegetative growth (Mercado et al. 1998, Koch et al. 2013 Table 2). Statistical analyses: 1-way ANOVA and post hoc Tukey's HSD (n = 4 for all treatments); significant differences (p < 0.05) between treatments indicated by letters Fig. 9. Mean (SD) net growth rates of Saccharina latissima grown under treatment pCO 2 levels following grazing by Lacuna vincta under the same pCO 2 conditions. The net growth rates consider the growth rates of S. latissima and herbivory rates of L. vincta from the 'pCO 2 dose response' and 'Effective minimum dose' experiments, respectively (Table 2). Statistical analyses: 1-way ANOVA and post hoc Tukey's HSD tests (n = 4 for all treatments); significant differences (p < 0.05) between treatments indicated by letters CO 2 (Maberly et al. 1992, Raven et al. 2002, Hepburn et al. 2011). In the present study, S. latissima had δ 13 C signatures that ranged from −16 to −18 ‰ when exposed to normal conditions, suggesting it primarily, but not exclusively, uses HCO 3 − as a C source under ambient pCO 2 conditions (Maberly et al. 1992, Raven et al. 2002, Hepburn et al. 2011). Exposure to elevated pCO 2 significantly lowered δ 13 C signatures (−18 to −29 ‰) of S. latissima, indicating that when exposed to higher pCO 2 levels, this alga obtained a larger fraction of its C from CO 2 , an observation consistent with previous studies of stable carbon isotope discrimination of S. latissima and other laminarian kelp species (Maberly et al. 1992, Fernández et al. 2015. Furthermore, in the present study, growth rates increased under increasing pCO 2 levels up tõ 1200 μatm, suggesting that S. latissima may become substrate-saturated at this concentration. While this level of pCO 2 is not predicted to occur in open ocean regions for more than a century, it can occur regularly in some coastal systems (Feely et al. 2008, Wallace et al. 2014. Be it in the future or due to present day coastal processes, elevated pCO 2 levels may directly benefit the growth of laminarian kelp species such as S. latissima.
Nutrients also increased S. latissima growth rates, although there was no interaction with pCO 2 levels. Eutrophication can initiate phase shifts in coastal ecosystems whereby turf algae benefit from the overloading of nutrients and cover substrate onto which kelp may grow and recruit, causing declines in kelp forests (Eriksson et al. 2002, Gorgula & Connell 2004, Connell et al. 2008. However, kelps including S. latissima can grow robustly using aquaculture approaches in eutrophic ecosystems (Kim et al. 2015, Jiang et al. 2020, and the results shown here demonstrate that, without competition, elevated nutrients can benefit S. latissima via enhanced growth rates. Beyond growth, herbivory on S. latissima also changed under elevated pCO 2 . The effects of elevated pCO 2 and the absence of food on the herbivory rates of L. vincta are consistent with prior observations regarding the manner in which acidification alters herbivory by this gastropod grazing on the green alga, Ulva (Young et al. 2019). In the present study, herbivory by L. vincta on S. latissima was even more sensitive to acidification, with food limitation decreasing herbivory rates at 830 μatm, slightly lower than the level at which L. vincta grazing on Ulva was reduced (850 μatm; Young et al. 2019). Exposure to elevated pCO 2 levels may cause a state of acidosis that can disrupt metabolism and homeostatic function, thus diverting energy from critical functions such shell and somatic growth, shell repair, and gametogenesis (Lindinger et al. 1984, Pörtner et al. 1998, Hendriks et al. 2010. L. vincta exposed to elevated pCO 2 and starved display lower respiration rates (Young et al. 2019), a finding interpreted as metabolic depression within other marine gastropods during periods of food limitation (Maas et al. 2011). The ability of acidification alone to slow the metabolism of gastropods has also been previously reported (Hendriks et al. 2010, Melatunan et al. 2011), which may serve as a survival strategy to match lowered energy supply (Bishop & Brand 2000) and may result in an increased reliance on anaerobic respiration (Pörtner et al. 1998, Melatunan et al. 2011). Reductions in herbivory under elevated pCO 2 were not observed when L. vincta was fed ad libitum, however. Similarly, when provided with an adequate food source, some other calcifying organisms (Melzner et al. 2011, Thomsen et al. 2013, Pansch et al. 2014) and early life stage fish ) are resistant to acidification.
In an ecosystem setting, kelp beds may provide a refuge to grazers by simultaneously buffering carbonate chemistry and providing ample quantities of food. Previous studies have demonstrated the ability of macroalgae to buffer carbonate chemistry and promote the growth and survival of calcifying organisms (Wahl et al. 2018, Young & Gobler 2018. However, such a buffering effect can be species-and/or site-specific (Rivest et al. 2017). Daytime primary productivity within kelp beds has been shown to significantly reduce pCO 2 compared to outside the bed on diel and even seasonal timescales (Delille et al. 2000(Delille et al. , 2009. Furthermore, vertical gradients in pCO 2 can also form in kelp canopies due to enhanced primary productivity in surface waters (Hofmann et al. 2011). The lower quantities of food and relatively higher levels of pCO 2 on the vertical or horizontal margins of the kelp beds may be the regions more likely to facilitate disruption of gastropod herbivory. Similarly, acidified regions with low-density kelp blades could similarly be disruptive to snail grazing. All these conditions would be exacerbated by nocturnal acidification (Wallace et al. 2014, Tomasetti & Gobler 2020. While the pCO 2 levels in the present study are not expected for the open ocean for decades or centuries to come, eutrophication, upwelling, riverine discharge, and other coastal processes may result in the diurnal and/or seasonal accumulation of CO 2 in coastal zones that can produce the pCO 2 and nutrient levels shown here to accelerate the growth of, and reduce grazing on, S. latissima (Feely et al. 2008, Wallace et al. 2014). In the present study, 18 h of exposure to elevated pCO 2 in the absence of food suppressed L. vincta herbivory, with this effect persisting for 72 h. In an ecosystem context, L. vincta and other gastropods and grazers sensitive to elevated pCO 2 may be vulnerable to diurnal and seasonal shifts in pH and pCO 2 . During nighttime and/or low tides, eutrophic estuaries can become strongly net heterotrophic and produce acidified conditions that can persist >18 h (Wallace et al. 2014, Baumann et al. 2015. These conditions are mostly likely to occur in temperate estuaries during the late summer, when microbial respiration rates accelerate acidification, and may persist into the fall (Wallace et al. 2014), which is the beginning of the growing season for numerous kelp species (Kim et al. 2015, Augyte et al. 2017. Aside from the diurnal and seasonal shifts in pCO 2 , acidified conditions often cooccur with low oxygen conditions in estuaries (Tomasetti & Gobler 2020), and the combination of hypoxia and acidification has been shown to additively and synergistically disrupt grazing by L vincta (Young & Gobler 2020). Finally, the elevated nutrients associated with eutrophication may also affect herbivory rates by L. vincta and growth by S. latissima. Nutrient enrichment can reduce the abundance of common estuarine grazers, including Lacuna, due to potentially toxic concentrations of ammonia (Atalah & Crowe 2012). While eutrophication indirectly harms S. latissima via reduced light availability and overgrowth by epiphytic and filamentous algae (Moy & Christie 2012), the alga can directly benefit from the increased nutrient concentrations (Boderskov et al. 2016, this study) and the acidification wrought from eutrophication (this study).
Despite the potential benefits that ocean acidification may provide to some primary producers (Koch et al. 2013, Hattenrath-Lehmann et al. 2015, Young & Gobler 2016, grazing pressure by common herbivores, such as gastropods, may limit the extent of these benefits through top-down controls on growth (Baggini et al. 2015). Among the factors that limit kelp abundance, overgrazing is a major biotic driver of kelp loss via consumption of kelp blades, inhibition of new recruitment, and/or grazing damage causing blade breakage (Steneck et al. 2002, Molis et al. 2010, Filbee-Dexter & Wernberg 2018. As L. vincta is the primary mesograzer of S. latissima macro scopic kelp sporophytes in the Northwest Atlantic (Brady- Campbell et al. 1984, Johnson & Mann 1986, large population increases in the gastropod can consume significant quantities of kelp biomass, which can result in significant losses of kelp canopy biomass due to wave action (Krumhansl & Scheibling 2011). Studies have shown that L. vincta grazing of kelp can remove up to 10% of the total surface area of blades in natural settings (Molis et al. 2010, Krumhansl & Scheibling 2011 and larger amounts (> 40%) in experimental settings (Chenelot & Konar 2007). In the present study, L. vincta displayed significantly reduced herbivory when ex posed to elevated pCO 2 and prior food restriction, while S. latissima growth was significantly increased under elevated pCO 2 . When considering the re sponses of S. latissima across a pCO 2 gradient, net growth rates increased more than 4-fold from under ambient pCO 2 to ~1200 μatm (Fig. 9). L. vincta had the potential to consume 70% S. latissima productivity per day under ambient pCO 2 , but only 38 and 9% at 800 and 1300 μatm, respectively, meaning that as pCO 2 levels rise, L. vincta, and other gastropods that experience reduced herbivory rates under elevated pCO 2 , may become incapable of controlling kelp proliferation.
Beyond growth and grazing, competition with other algae will also influence the fate of S. latissima populations in high-CO 2 environments (Young & Gobler 2017). During the last decade, turf-forming algae have begun to replace kelp beds in many ecosystems, primarily due to warming and eutrophication (Moy & Christie 2012, Filbee-Dexter & Wernberg 2018. In regions with naturally occurring high CO 2 conditions from volcanic vents, however, turf algal communities decline in diversity and abundance with decreasing pH and increasing pCO 2 (Porzio et al. 2011). Hence, beyond the ability of high pCO 2 to foster higher net growth rates in S. latissima, these conditions may also allow it to maintain dominance in regions where it may be competing with turf-forming algae, possibly off-setting losses associated with other anthropogenic processes.
Within the northwestern Atlantic, L. vincta can re produce year-round on kelp in cooler climates (Maney & Ebersole 1990), with a distinctive peak in spawning during January and February (Johnson & Mann 1986), larval recruitment and hatching occurring thereafter, followed by a summer maxima (South gate 1982). Within the same regions, kelp species such as S. latissima begin their growing season during the fall, grow rapidly in spring, and experience rapid decay and mortality during the summer as thermal thresholds are surpassed (Krumhansl & Scheibling 2011, Kim et al. 2015, Augyte et al. 2017. The emergence of marine gastropod grazers with the onset of the warmer summer weather accelerates the seasonal demise of kelp (Johnson & Mann 1986, Krumhansl & Scheibling 2011. As coastal waters sea-sonally warm, accelerating rates of ecosystem metabolism can cause a supersaturation of pCO 2 (Wallace et al. 2014, Baumann et al. 2015) that may enhance the net growth rates of S. latissima and prolong the growing season of the kelp that might otherwise slow due to warmer temperatures and grazing (Wernberg et al. 2010, Filbee-Dexter et al. 2016. In the future, as climate change intensifies acidification, shortened growing seasons due to warming may be offset by increased net growth due to higher pCO 2 . Future experiments should consider the interactive effects of temperature and pCO 2 on the growth of S. latissima.

CONCLUSIONS
Elevated levels of pCO 2 have a dual benefit for the kelp Saccharina latissima. The net growth rates of S. latissima at higher pCO 2 (>1200 μatm) exceeded those at ambient pCO 2 by 4-fold due to both accelerated growth by S. latissima and suppressed herbivory by the gastropod Lacuna vincta. The pCO 2 conditions that elicited reduced herbivory (~830 μatm) and enhanced kelp growth (~1200 μatm pCO 2 ) are seasonally found in many estuaries today and will be come more common in the future as climate change accelerates. Eutrophication associated with coastal acidification may benefit S. latissima directly through nutrient-and CO 2 -enhanced growth rates and may further reduce herbivore grazing due to hypoxia and/or high ammonia levels. Since only 18 h of acidification suppressed herbivory by L. vincta, nocturnal acidification associated with eutrophication (Wallace et al. 2014) may further suppress herbivore grazing. While S. latissima may buffer carbonate chemistry to the benefit of calcifying organisms such as L. vincta, these benefits may only be realized in the center of dense kelp beds as opposed to horizontal and vertical margins where acidification and gastropod starvation may become more common. Lastly, given that turf-forming algae may not directly benefit from increased pCO 2 , the combination of increased growth and suppressed herbivory experienced by S. latissima may allow the alga to maintain dominance facilitated by high-CO 2 conditions.