Hatchery produced and wild-collected bivalves were used for experiments. Mercenaria mercenaria were produced by the Stony Brook University shellfish hatchery in Southampton, NY (40.89° N, 72.44° W) using broodstock collected from Shinnecock Bay, NY, USA. Crassostrea virginica were provided by the Aeros Cultured Oyster hatchery, NY, USA (41.05° N, 72.40° W) using broodstock from the Peconic Estuary, NY, USA. Mytilus edulis were collected from wild populations in Old Fort Pond, NY, USA (40.89° N, 72.44° W) and Moriches Bay, NY, USA (40.78° N, 72.78° W; Supplementary Figure S1 ). S. latissima used for all experiments were grown on horizontal longlines at a commercial oyster farm (Great Gun Shellfish) in Moriches Bay, NY, USA (40.78° N, 72.78° W; Supplementary Figure S1 ), a lagoon contiguous with Shinnecock Bay to the east, using kelp seedstock derived from wild populations in the Long Island Sound, CT, USA. S. latissima was grown throughout winter and spring 2021, and was collected from early March to late May, during which laboratory and field experiments were performed. Densities of kelp on this shallow (~2 m) farm have previously estimated to range from 0.6 g L^-1 during high tide in the early growing season to 4.4 g L^-1 during low tide at the end of the growing season (Sylvers and Gobler, 2021). For experiments, large, well-pigmented blades of the macroalga were chosen from samples collected by hand at low tide. S. latissima blades were cut from their holdfasts as close to the longlines as possible and were immediately placed in seawater-filled containers and transported to the Stony Brook Marine Science Center within 30 minutes of collection. Upon arrival to the facility, S. latissima were placed in a large, round ~2000 L tank filled with flowing, 1 µm filtered seawater from Shinnecock Bay.

Six laboratory experiments were performed to assess the effects of elevated pCO₂ and the presence of S. latissima on the growth rates of M. mercenaria, C. virginica, and M. edulis. Prior to the laboratory experiments with bivalves and S. latissima, a preliminary experiment was performed to assess the ability of S. latissima to alter pH in experimental containers. All experiments were performed in 1 L polycarbonate containers, which were acid-washed (10% HCl) and liberally rinsed with deionized water prior to use. Experimental containers were placed in an environmental control chamber set to a consistent temperature (18.6 ± 0.3°C), light intensity (~200 µmol photons m^-2 s^-1) and photoperiod (14 h: 10 h light:dark cycle). The light intensity and photoperiod of the environmental control chamber mimicked conditions observed at the S. latissima collection site during the time of collection (see above) while temperatures were indicative of the latter half of the growing season (May). Containers were filled with filtered (0.2 µm polysulfone filter capsule, Pall^©) seawater to maximum capacity and covered with plexiglass with pre-drilled holes to allow delivery of dissolved gases and minimize interaction with the environment. Dissolved gases were delivered into each experimental container by aeration via a 3.8 x 1.3 cm air diffuser (Pentair) connected to a 1 mL, polystyrene serological pipette inserted through the pre-drilled holes in the plexiglass covers to the bottom of each container and connected via Tygon tubing to an air source. The containers were then randomly assigned, in quadruplicate, to each treatment (see below). Containers subjected to elevated (~2,000 µatm) CO₂ levels utilized a multitube gas proportionator system (Cole Parmer^® Flowmeter system, multitube frame) that mixed ambient air with 5% CO₂ gas (Talmage and Gobler, 2010). The pCO₂ levels in the elevated treatments were higher than levels present at the S. latissima collection site but are consistent with levels present in eutrophic US East Coast estuaries (Wallace et al., 2014; Baumann et al., 2015; Wallace and Gobler, 2015) including those near the oyster farm (Wallace and Gobler, 2021; Wallace et al., 2021). Containers subjected to ambient (~400 µatm) CO₂ levels utilized a single tube proportioner system to introduce only ambient air into containers. The gases were mixed through gang valves and were delivered at a flow rate of 1000 ± 5 mL min^-1 to experimental containers through serological pipettes inserted through the plexiglass covers. The bubbling rates in the containers turned over the volume ~100 times d^-1, and bubbling was initiated at least 2 – 3 d prior to the beginning of experiments to allow CO₂ levels and carbonate chemistry to reach a state of equilibrium. Continuous and discrete measurements of pH were made using an Orion Star A321 Plus electrode ( ± 0.001 pH unit, NBS scale) calibrated prior to each use using National Institute of Standards and Technology (NIST) traceable standards.

Dissolved inorganic carbon (DIC) within experimental containers was measured directly from water samples collected at the beginning and end of experiments, which were preserved using a saturated mercuric chloride (HgCl₂) 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) as per Young and Gobler (2018). Total DIC, total alkalinity, pCO₂, HC O 3 − , Ω_aragonite, and Ω_calcite ( Table 1 ) were calculated from measured DIC levels, pH, temperature, and salinity, as well as the first and second dissociation constants of carbonic acid in seawater (Millero, 2010) using the program CO2SYS (http://cdiac.ess-dive.lbl.gov/ftp/co2sys/). For quality assurance, DIC levels and pH within certified reference material (provided by Andrew Dickson of the University of California, San Diego, Scripps Institution of Oceanography; batches 180 = 2021.87 µmol kgSW^-1) were measured during analyses of each sample set. Analyses of samples only continued when complete recovery (>99.9%) of certified reference material was attained.

For all experiments, individual S. latissima blades were prepared by cutting the stipe 2.5 cm below the blade-stipe interface. Cuts were made above the blade-stipe interface in order to obtain the desired fresh weight, depending on the experiment being performed (see below). This was done to standardize initial S. latissima blade tissue type and size (Boderskov et al., 2016). All samples were extensively rinsed with filtered (0.2 µm) seawater, spun in a salad spinner to remove debris and epiphytes and excess seawater, rinsed and spun again, and weighed on a Scientech ZSA 120 digital microbalance ( ± 0.0001 g) to obtain initial fresh weight in grams.

An initial experiment was performed to assess the ability of S. latissima to alter pH in experimental conditions similar to the later laboratory experiments with bivalves and S. latissima. For this experiment, there were two treatments: a control with increased CO₂ without S. latissima and a treatment with increased delivery of CO₂ and S. latissima.

Four experiments (Experiments 1 – 4; Table 2 ) were performed to assess the effects of elevated pCO₂ and the presence of S. latissima on the growth of C. virginica, M. edulis, and M. mercenaria. For the first experiment, C. virginica (~2.6 mm) were placed, in quadruplicate, in one of two treatments: a control with elevated pCO₂ (~1700 – 1800 µatm) without S. latissima and a treatment with elevated pCO₂ and S. latissima added at a level within the range of densities found on the kelp farm (1.0 g L^-1; Sylvers and Gobler, 2021). The second, third, and fourth experiments placed C. virginica (~2.8 mm), M. edulis (~2.3 mm), and M. mercenaria (~1.1 m), respectively, in quadruplicate, in one of four treatments: a control with ambient pCO₂ (~360 – 500 µatm) without S. latissima, a treatment with ambient pCO₂ and S. latissima (1.0 g L^-1) added, a treatment with elevated pCO₂ (~2,000 µatm) without S. latissima, and a treatment with elevated pCO₂ and S. latissima (1.0 g L^-1) added. Two experiments (Experiments 5 and 6; Table 2 ) were performed to assess the minimum S. latissima biomass required to alter the growth of M. edulis and C. virginica when grown under elevated pCO₂. For the fifth experiment (Experiment 5; Table 2 ), M. edulis (~5.7 mm) were placed, in quadruplicate, in one of four treatments: a control with elevated pCO₂ (~2,700 µatm) without S. latissima, a treatment with elevated pCO₂ and 0.3 g L^-1 S. latissima, a treatment with elevated pCO₂ and 0.7 g L^-1 S. latissima, and a treatment with elevated pCO₂ and 1.0 g L^-1 S. latissima. For the sixth experiment (Experiment 6; Table 2 ), M. edulis (~6.5 mm) and C. virginica (~5.1 mm) were placed, in quadruplicate, in one of four treatments: a control with elevated pCO₂ (~3,500 µatm) without S. latissima, a treatment with elevated pCO₂ and 0.5 g L^-1 S. latissima, a treatment with elevated pCO₂ and 1.0 g L^-1 S. latissima, and a treatment with elevated pCO₂ and 2.0 g L^-1 S. latissima. Across all experiments, bivalves were fed a mixture of Isochrysis galbana and Chaetoceros muelleri ad libitum (4 x 10⁴ cells mL^-1 d^-1; Helm et al., 2004). In the preliminary experiment, the same microalgal mixtures were added to treatments with and without S. latissima and demonstrated that microalgae did not have a discernable impact on pH levels in the experimental containers. Microalgal cultures were maintained in exponential phase growth in f/2 media using standard culturing conditions (Helm et al. (2004).

Experiments began with the introduction of bivalves and S. latissima into polycarbonate containers fitted with continuous pH electrodes prepared as described above. Initial DIC samples were collected and discrete measurements of pH and temperature were taken and were made daily for the duration of experiments. Each experiment persisted for ~two weeks, with the exception of the preliminary experiment and Experiment 1, which persisted for ~one week and ~three weeks, respectively. At the beginning of each experiment, 15 – 20 individuals from each bivalve cohort were placed in each experimental container, in quadruplicate, with a set of 15 – 20 individuals from the same cohort put aside to obtain initial measurements of shell height (for C. virginica and M. edulis) or length (for M. mercenaria) and tissue weight. Bivalve dimensions were determined by analysis of digital images using the software ImageJ (Young and Gobler, 2018). At the end of each week, a complete water change was performed for all containers using water bubbled in 20-L polycarbonate containers with the same gas mixtures for ambient and elevated pCO₂ treatments as described above to ensure that bivalves were exposed to the target pCO₂ levels for the duration of the experiment. At the end of experiments, final pH, temperature, and salinity measurements were made and final DIC samples were collected and analyzed and bivalves were collected on a 500 µm sieve, transferred to a graduated laminated grid sheet, and digitally imaged for shell measurements.

Two field experiments (Experiment 7; Table 2 ) were conducted to assess the impacts of S. latissima on bivalves. For Experiment 7, the growth of C. virginica was monitored at the Great Gun Shellfish farm in Moriches Bay, NY, USA (40.78° N, 72.78° W), where S. latissima was collected for the laboratory experiments (Experiments 1 – 6). S. latissima was cultivated along four 30m horizontal longlines that were staked ~0.4 m above the bay bottom and spaced ~2 m apart ( Supplementary Figure S2 ). The spacing of the lines in shallow water depths (~0.6 m MLW; ~2.0 m MHW) has been employed in other seaweed farms (Mongin et al., 2016).

Experimental oysters were placed at one of three locations that differed in proximity to the lines of S. latissima: in-kelp, near-kelp, and outside-kelp sites ( Supplementary Figure S2 ). The in-kelp location was positioned in the center of the four-line kelp array, the near kelp location was positioned ~3 m away from the end of the kelp lines, and the outside kelp location was positioned ~50 m away from the kelp lines. At each location triplicate cages were each stocked with 100 experimental oysters (~3.0 mm). For each replicate, oysters were placed in a spat bag (1.0 mm mesh) housed in a floating oyster cage composed of a standard oyster ‘grow-out’ bag (36” x 18” x 3”) made of semi-rigid polyethylene mesh (4 mm mesh size), with plastic, air-filled, cylindrical floats (3” diameter) attached to each long side. The bags were secured directly to the kelp lines in the ‘in-kelp’ treatment with a bridle-clip system and were attached to lines without kelp in the ‘near-kelp’ and ‘outside-kelp’ treatments. Prior to deployment, a subset (n = 100) of oysters were digitally photographed for measurement of shell height and width using ImageJ (see above), and then frozen for later analysis of tissue weight (see below).

YSI EXO3 multi-parameter sondes were deployed to continuously monitor pH (NBS scale), which internally logged data every 10 min. Calibration of the pH sensors was performed by use of three NIST pH buffers with known pH levels of 4, 7, and 10. While not shown here, temperature (°C) and conductivity (used to measure salinity) were also continuously monitored by the EXO3. Temperature was measured with a highly stable and aged thermistor that requires no prior calibration, while conductivity was calibrated using a conductivity standard with a recommended standard of 1 mS cm^-1 (1000 µS cm^-1) for the highest stability. Sondes were deployed at the outside-kelp and in-kelp sites. At both sites, triplicate DIC samples were collected at the beginning, middle, and end of the field experiment. The collection, preservation, and analysis of DIC samples for the field experiment followed the same procedure as for the laboratory experiments, described above.

At the conclusion of experiments, the initial and final shell height and shell width of experimental bivalves was measured and averaged to obtain the mean shell height and width for each experimental container of the lab experiments (n = 4) or cage of the field experiment (n = 3). Shell-based growth (mm d^-1) was determined from the changes in shell dimensions during the experiment. Tissue weights were obtained by weighing bivalves after drying at 60°C for 72 h, combusting them at 450°C for 4 h, weighing them again, and subtracting the combusted weight from the dry weight. All individuals from each experimental container or cage were combined for drying and combustion to obtain collective weights for each replicate. Tissue-based growth (mg bivalve^-1 d^-1) was determined by subtracting the initial tissue weight (from the initially-collected subset of bivalves) from the final tissue weight and dividing by the duration of the experiment in days, for each individual from each replicate container, whereby tissue weight was determined by subtracting the combusted weight from the dry weight. One-way ANOVAs were performed with SigmaPlot 11.0 to assess significant differences in growth rates (Experiments 1, and 5 – 7) and M. edulis abundances (Experiment 7) between treatments. Two-way ANOVAs were performed to assess significant differences in growth rates (Experiments 2 – 4) where the main treatment effects were pCO₂ level (ambient or elevated) and S. latissima (with and without the macroalga). Normality and equal variance were tested via the use of Shapiro-Wilk and Levene tests within SigmaPlot 11.0; assumptions of equal variance and normality were met for all data. If significant differences were detected, a Tukey’s Honest Significant Difference (HSD) test using R v.3.4.0 within RStudio v.1.0.143 was performed.

An initial experiment was conducted to assess how Saccharina latissima would alter pH levels within experimental vessels that had been bubbled with CO₂ and began with a pH of ~7.2 ( Figure 1 ). In the treatment with S. latissima, pH levels increased during the light portion of the photoperiod by ~0.4 units for the first two days, ~0.3 units the next two days, and more slowly thereafter reaching a maximum of ~8.8 on day 7 ( Figure 1 ). In contrast, pH levels in the treatment without S. latissima increased more slowly and steadily, peaking at only 7.5 on the seventh day ( Figure 1 ).

During the first experiment with bivalves using C. virginica ( Table 2 and Figure 2 ), pH values in the treatments with and without S. latissima ranged 7.44 – 8.48 and 7.02 – 7.69, respectively, and pH was, on average, 7.94 ± 0.17 and 7.27 ± 0.19, respectively ( Table 1 and Figure 2A ) and Ω_aragonite was 1.9 ± 0.3 and 0.4 ± 0.1 ( Table 1 and Figure 2B ). Shell-based growth rates of C. virginica were significantly and 40% higher in the presence of S. latissima than the treatment without S. latissima (One-way ANOVA; p < 0.05; Figure 2C and Supplementary Table S1 ). Tissue-based C. virginica growth rates were significantly higher by 50% in the treatment with S. latissima compared to the treatment without S. latissima (One-way ANOVA; p < 0.05; Figure 2D and Supplementary Table S1 ).

In the second bivalve experiment with C. virginica ( Table 2 and Figure 3 ), pH levels in the elevated pCO₂ treatments were 8.25 ± 0.15 and 7.34 ± 0.32, respectively ( Figure 3A ) while Ω_aragonite in the ambient pCO₂ treatments with and without S. latissima were 5.2 ± 0.3 and 2.0 ± 0.3, respectively, and were 3.1 ± 0.7 and 0.5 ± 0.1 in the elevated pCO₂ treatments with and without S. latissima, respectively ( Table 1 ; Figure 3B ). There was a significant interaction in this experiment as the effects of pCO₂ on C. virginica (~2.8 mm) shell-based growth rates were dependent upon the presence of S. latissima (Two-way ANOVA; p < 0.05; Figure 3C ). Comparing specific treatments, shell-based growth rates were significantly lower in the elevated pCO₂ treatment without S. latissima compared to the elevated pCO₂ treatment with S. latissima and the ambient pCO₂ treatments with and without S. latissima by 64, 60, and 58%, respectively (Tukey HSD; p < 0.05 for all; Figure 3C and Supplementary Table S3 ). There were no significant differences in tissue-based growth rates in this experiment ( Figure 3D and Supplementary Table S2 ).

During the experiment with Mytilus edulis ( Table 2 and Figure 4 ), average pH levels in the ambient pCO₂ treatments with and without S. latissima and the elevated pCO₂ treatments with and without S. latissima were 8.88 ± 0.16, 8.31 ± 0.12, 8.14 ± 0.28, and 7.36 ± 0.10, respectively ( Figure 4A ). Average Ω_aragonite in the ambient pCO₂ treatments with and without S. latissima was 6.7 ± 0.4 and 2.0 ± 0.1, respectively, and was 3.0 ± 0.1 and 0.5 ± 0.1 in the elevated pCO₂ treatments with and without S. latissima, respectively ( Figure 4B ). There was a significant interaction between S. latissima and pCO₂ on shell growth for M. edulis (Two-way ANOVA; p < 0.05). Comparing specific treatments, shell-based growth rates were significantly lower in the elevated pCO₂ treatment without S. latissima than the elevated pCO₂ treatment with S. latissima and the ambient pCO₂ treatments with and without S. latissima by 36, 29, and 31%, respectively (Tukey HSD; p < 0.05 for all; Figure 4C and Supplementary Table S5 ). Shell-based growth rates were not different between the elevated pCO₂ treatment with S. latissima and ambient pCO₂ treatments with and without kelp (Tukey HSD; p > 0.05 for all; Figure 4C and Supplementary Table S5 ). Tissue-based growth was significantly higher in the presence of S. latissima by ~47% (Two-way ANOVA; p < 0.05; Figure 4D and Supplementary Table S4 ), and there was no significant effect of pCO₂ on tissue-based growth and no interaction detected (Tukey HSD; p > 0.05; Figure 4D and Supplementary Table S5 ).

During the experiment with Mercenaria mercenaria ( Table 2 and Figure 5 ), average pH levels in the ambient pCO₂ treatments without and with S. latissima and the elevated pCO₂ treatments without and with S. latissima were 8.45 ± 0.19, 8.79 ± 0.35, 7.55 ± 0.07, and 8.18 ± 0.40, respectively ( Figure 5A ). Ω_aragonite in the ambient pCO₂ treatments with and without S. latissima was 5.7 ± 0.8 and 2.1 ± 0.2, respectively, and was 3.8 ± 0.6 and 0.5 ± 0.1 in the elevated pCO₂ treatments with and without S. latissima, respectively ( Table 1 and Figure 5B ). There was a significant interaction between pCO₂ and S. latissima for shell- and tissue-based growth rates of hard clams (Two-way ANOVA; p < 0.05). Shell-based growth rates of M. mercenaria (~1.1 mm) were significantly faster in the elevated pCO₂ treatment with S. latissima, and the ambient pCO₂ treatments with and without S. latissima than in the elevated pCO₂ treatment without S. latissima by ~100, ~110, and ~90%, respectively (Two-way ANOVA and Tukey HSD; p < 0.05 for all; Figure 5C and Supplementary Tables S6, S7 ); no other significant differences were detected. between treatments for shell growth. Tissue-based growth rates were significantly (two-fold) higher in the elevated pCO₂ treatment with S. latissima than the elevated pCO₂ treatment without S. latissima (Two-way ANOVA and Tukey HSD; p < 0.05; Figure 5D and Supplementary Tables S6, S7 ). In the ambient pCO₂ treatments with and without S. latissima, tissue-based growth was significantly higher by 100 and 150%, respectively, than in the elevated pCO₂ treatment without S. latissima (Two-way ANOVA and Tukey HSD; p < 0.05 for both; Figure 5D and Supplementary Tables S6, S7 ). There were no significant differences in tissue-based growth between the ambient pCO₂ treatments and the elevated pCO₂ treatment with S. latissima (Two-way ANOVA and Tukey HSD; p > 0.05 for all; Figure 5D and Supplementary Tables S6, S7 ).

During the fifth experiment growing M. edulis with four levels of kelp ( Table 2 and Figure 6 ), pH levels in the 0.0 (control), 0.3, 0.7, and 1.0 g L^-1 were, on average, 7.44 ± 0.10, 7.74 ± 0.15, 8.01 ± 0.24, and 8.42 ± 0.33, respectively ( Figure 6A ), while Ω_aragonite values were 0.5 ± 0.1, 1.5 ± 0.1, 1.9 ± 0.1, and 4.6 ± 0.2, respectively ( Table 1 and Figure 6B ). Shell-based growth rates were significantly higher in the 0.3, 0.7, and 1.0 g L^-1 treatments than in the control by ~50, ~55, and ~70%, respectively, but there were no differences in shell-based growth between the 0.3, 0.7, and 1.0 g L^-1 treatments (One-way ANOVA and Tukey HSD; p < 0.05 for all significant differences; Figure 6C ; Supplementary Tables and S8 and S9 ). Tissue-based growth rates were significantly higher in the 1.0 g L^-1 treatment than in the control, 0.3, and 0.7 g L^-1 treatments by ~45, ~40, and ~40%, respectively; there were no other significant differences among tissue-based growth rates (One-way ANOVA and Tukey HSD; p < 0.05 for all significant differences; Figure 6D ; Supplementary Tables and S8 and S9 ).

During the sixth and final experiment that grew C. virginica and M. edulis with four levels of kelp ( Table 2 and Figure 7 ), pH levels in the 0.0 (control), 0.5, 1.0, and 2.0 g L^-1 treatments were, on average, 7.37 ± 0.10, 7.66 ± 0.40, 8.19 ± 0.51, and 8.53 ± 0.39, respectively ( Figure 7A ), while Ω_aragonite values were 0.4 ± 0.1, 1.5 ± 0.1, 5.3 ± 0.7, and 6.0 ± 0.3, respectively ( Table 1 and Figure 7B ). Shell-based growth rates of C. virginica (~5.1 mm) were significantly higher in the 0.5, 1.0, and 2.0 g L^-1 treatments compared to the 0.0 g L^-1 treatment by ~140, ~280, and ~290%, respectively (One-way ANOVA and Tukey HSD; p < 0.05 for all; Figure 7C and Supplementary Tables S10, S11 ). Additionally, C. virginica shell-based growth rates were significantly higher in the 1.0 and 2.0 g L^-1 treatments than in the 0.5 g L^-1 treatment by ~60% but did not differ between the former two treatments (One-way ANOVA and Tukey HSD; p < 0.05 for all significant differences; Figure 7C and Supplementary Tables S10, S11 ). Tissue-based growth rates for C. virginica were significantly higher in the 1.0 and 2.0 g L^-1 treatments than in the control by ~160 and ~150%, respectively, but there were no differences detected between other treatment combinations (One-way ANOVA and Tukey HSD; p < 0.05 for all significant differences; Figure 7D and Supplementary Tables S10, S11 ). For M. edulis (~6.5 mm), shell-based growth rates were significantly higher in the 0.5, 1.0, and 2.0 g L^-1 treatments than in the control by ~90, 100, and ~130%, respectively (One-way ANOVA and Tukey HSD; p < 0.05 for all; Figure 7C and Supplementary Tables S12, S13 ). There were no differences in M. edulis shell-based growth between the 0.5, 1.0, and 2.0 g L^-1 treatments (One-way ANOVA and Tukey HSD; p > 0.05 for all; Figure 7C and Supplementary Tables S12, S13 ). Tissue-based growth rates for M. edulis were significantly higher in the 0.5, 1.0, and 2.0 g L^-1 treatments than in the control by ~135, ~110, and ~150%, respectively, but there were no significant differences in M. edulis tissue-based growth rates between the 0.5, 1.0, and 2.0 g L^-1 treatments (One-way ANOVA and Tukey HSD; p < 0.05 for all significant differences; Figure 7D and Supplementary Tables S12, S13 ).

For the one-month field experiment (Experiment 7; Table 2 and Figure 8 ), pH levels at the outside-kelp and in-kelp sites were, on average, 7.71 ± 0.16 and 7.86 ± 0.18, respectively ( Figure 8A ) while average Ω_aragonite values were 2.1 ± 0.2 and 2.3 ± 0.2, respectively ( Table 1 and Figure 8B ). pH differences between within and outside of the kelp were pronounced for the first two weeks of the experiment, small for the third week, and near absent for the final week (Fig, 8A). C. virginica (~3.0 mm) shell-based growth rates were significantly faster in the in- and near-kelp sites than at the outside-kelp site by ~270 and ~90%, respectively (One-way ANOVA and Tukey HSD; p < 0.05 for both; Figure 8C and Supplementary Tables S14, S15 ). Shell-based growth rates were significantly faster at the in-kelp site than at the near-kelp site by ~90% (One-way ANOVA and Tukey HSD; p < 0.05; Figure 8C and Supplementary Tables S14, S15 ). Tissue-based growth rates were also significantly higher at the in-kelp site than at the outside-kelp and near-kelp sites by 160 and ~55%, respectively (One-way ANOVA and Tukey HSD; p < 0.05 for both; Figure 8D and Supplementary Tables S14, S15 ). Additionally, tissue-based growth was significantly higher at the near-kelp site than at the outside-kelp site by 70% (One-way ANOVA and Tukey HSD; p < 0.05; Figure 8D and Supplementary Tables S14, S15 ).