A major challenge in biodiversity research is the incorporation of species interactions into frameworks describing population and community response to global environmental change (GEC). Mutualisms are a type of species interaction especially sensitive to changing environmental conditions, and the breakdown of facilitative species interactions could amplify the negative impacts of novel climate regimes on focal species. Here, we investigate how reciprocal interactions between two coastal foundation species, the eastern oyster (Crassostrea virginica) and eelgrass (Zostera marina) shift in sign and magnitude in response to ocean warming (+1.5C) and acidification (-0.4 pH) via a manipulative co-culture experiment in mesocosms. Under ambient environmental conditions, oysters facilitated eelgrass leaf growth and clonal reproduction by 35% and 38%, respectively. Simultaneously, eelgrass increased oyster condition index (the ratio of tissue to shell biomass) by 35%, indicating greater allocation of energy to production of soft tissues instead of shell at ambient conditions. Varying sensitivity of each species to ocean warming and/or acidification treatments led to complex shifts in species interactions that were trait dependent. As such, community outcomes under future conditions were influenced by species interactions that amplified and mitigated species response to environmental change.

Synthesis: Given that species interaction effect sizes were similar in magnitude to effect sizes of warming or pH treatments, our results underscore the need to identify key species and interaction types that strongly influence community response to GEC. Specifically, for macrophyte-bivalve interactions, understanding how physiological limitations on growth are impacted by environmental heterogeneity and co-culture will support successful restoration of natural populations and the rapid expansion of aquaculture.

Species Collections and Preparation: We haphazardly collected 600 eelgrass (Zostera marina) ramets (non-reproductive shoots with rhizomes and roots) from a 200 m² area encompassing the low intertidal and shallow subtidal zones in Brewer Cove, Orr’s Island, ME (43º 47’ 33.5” N, 69º 57’ 14.8’’ W) on June 27th, 2021. Simultaneously, we collected sediment from the adjacent mudflat, which we sieved through 5 mm mesh and homogenized. We standardized eelgrass ramets to one terminal shoot (length ± SD, 90 ± 23 cm) with 3 cm of rhizome and held standardized shoots in a cooler with seawater until planting them in mesocosms on the same day.

We obtained 480 triploid eastern oysters (Crassostrea virginica) from a local oyster aquaculture facility, the Quahog Bay Conservancy, on June 30th, 2021.  Oysters were approximately the same size (mean ± SD, 5.9 ± 2.2 cm²) and drawn at random from hatchery provided bags. We used Gluemasters Industrial Strength Cyanoacrylate Adhesive to attach six oysters equally spaced on to a 15 x 15 cm PVC plate (80 plates total).  Each plate was numbered and photographed with a camera facing down at a fixed height to measure initial projected surface area for each oyster using the software ImageJ. Oyster plates were transferred to mesocosms within a few hours of being photographed.

Mesocosm Experiment: Our goal was to understand how reciprocal interactions between oysters and seagrass changed across warming and pH treatments. We built a system of 60 outdoor flowthrough-seawater mesocosms to culture “oyster monoculture”, “seagrass monoculture”, and “oysters + seagrass co-culture” under ambient and three future ocean treatments consisting of warming-only, low pH (elevated pCO₂)-only, as well as warming and low pH combined. Therefore, our fully factorial experimental design consisted of 12 unique culture and ocean treatment combinations, with five mesocosm replicates (N=5) for each of the 12 combinations. Spatially, we arranged mesocosms in a 5 x 12 grid. Each column of 5 mesocosms was fed by a single seawater manifold, therefore the same temperature and pH treatment combinations were applied to all mesocosms with a column.  We systematically alternated temperature and pH treatments across the 12 columns.  Along each manifold, we systematically alternated the position of “oyster monoculture”, “seagrass monoculture”, and “oysters + seagrass co-culture” mesocosms.  Thus, despite mesocosms experiencing temporary shading from nearby trees or buildings throughout the day, replicates for each of the 12 treatment combinations were spread evenly throughout the grid of mesocosms. Mesocosms were 27 L plastic buckets that measured 30 cm diameter x 50 cm high. We added approximately 10 cm of sieved and homogenized Brewer Cove sediment to each mesocosm. We planted 15 standardized eelgrass shoots (see Species Collection and Preparation) within each “seagrass monoculture” and “oysters + seagrass co-culture” mesocosm (equivalent to 205 shoots m^-2, average shoot density in Brewer Cove during Summer 2021 was 145 shoots m^-2 in the intertidal zone to 250 shoots m^-2 in the subtidal zone). We suspended two oyster plates (or 12 oysters total, see Species Collection and Preparation) back-to-back in the water column (fully submerged) within the center of each “oysters monoculture” and “oysters + seagrass co-culture” mesocosm.  Plates without any oysters attached were suspended in “seagrass monoculture” mesocosms to control for any potential shading effects of the plates.  We ran unfiltered seawater into all mesocosms at a rate of approximately 1.2 L min^-1 (seawater turnover rate was approximately three times per hour).  We checked seawater flow rates and removed epiphytes from seagrass and oyster surfaces approximately once a day.

We allowed seagrass and oysters to acclimate to mesocosms for approximately a week before beginning warming and pH treatments on July 6th, 2021. We plumbed ambient seawater directly into ambient ocean treatment mesocosms.  For future ocean treatments, we used three 200 L sump tanks which we either heated and/or bubbled with CO₂. In both the warming only and the warming and low pH sumps we installed two Process Technologies 1000 W titanium aquaculture heaters.  We installed twelve Onset Temperature Pendants within the mesocosms, one in each of the twelve species monoculture or co-culture, temperature, and pH combinations.  We continuously controlled pH in both the low pH-only and the warming and low pH combined sumps using an APEX Aquacontroller system to bubble CO₂, which was outfitted with APEX lab-grade pH probes (precision ±0.1 pH units), APEX temperature probes (precision ± 0.5°C) and the APEX Fusion software. We set a goal treatment pH of 7.6, turning on a CO₂ exchanger to bubble CO₂ into sumps when pH was above 7.6.  Temperature treatments averaged 19.06 ± 0.01°C in warmed mesocosms compared to 17.78 ± 0.01°C in ambient mesocosms (Fig 1 a).  For comparison, the long-term average summer sea surface temperature for the Gulf of Maine is 15°C, with Summer 2021 being the hottest on record with an average sea surface temperature of 17.23°C (Kemberling 2023). Solar heating and tidal cycles caused daily temperature fluctuations within mesocosms; for example, in early August the warmed mesocosms ranged daily from 17.02 - 21.61°C and ambient mesocosms ranged daily from 16.53 - 19.38°C (Fig. 1 b). Throughout the summer, average temperatures increased by approximately 2°C. Daytime pH within the mesocosms differed between our ambient and low pH treatments (7.91 ± 0.01 and 7.53 ± 0.04, respectively, Fig. 1 c). However, the 24-hour pH profile of a single mesocosm from our warm and low pH treatment containing both eelgrass and oysters demonstrates that pH fluctuated greatly within the mesocosms, with pH ranging from a maximum of 8.17 during the day and a minimum of 7.47 at night (Fig. 1 d). There was no impact of environment or species co-culture on daytime total alkalinity (average TA range: 2095 – 2118, see supplemental Fig. S1). Because of tidal cycles and daily fluctuations, treatments are more accurately described relative to ambient conditions (+1.5°C and -0.4 pH units relative to ambient, see Fig. 1). We ran the experiment for 34 days.

One week prior to the end of the experiment we tagged three eelgrass shoots in each “seagrass monoculture” and “oysters + seagrass co-culture” mesocosm with flagging tape and we used a hypodermic needle to mark the leaves of these shoots to determine leaf growth rates (see “hole-punch method” Dennison 1987). At the end of the experiment, we collected all eelgrass from each “seagrass monoculture” and “oysters + seagrass co-culture” mesocosm. We measured leaf growth rates on the tagged shoots. To assess changes in eelgrass clonal reproduction (addition of new clonal side shoots), aboveground and belowground biomass, we first rinsed all eelgrass from each mesocosm in freshwater, counted the total number of shoots, and then divided the eelgrass into aboveground tissue (shoots) and belowground tissue (rhizomes + roots). We dried eelgrass tissue at 60°C and weighed dry biomass for both tissue types. We rephotographed all oyster plates and then calculated total shell growth over the experimental period as initial projected shell area minus final projected shell area using the software ImageJ (Ricart et al. 2021a). We froze all oysters at -80°C.  To determine changes in oyster tissue biomass and condition index, we dissected the remaining oysters and dried oyster tissue and shells separately at 60°C. We calculated oyster condition index, a common proxy for energy allocation and stress response in bivalves, using the following equation (Rainer and Mann 1992, Ricart et al. 2021a).

CI = Tissue Dry Mass / Shell Dry Mass * 100

Additional seawater chemistry monitoring: We monitored pH and temperature continuously in all treatment sumps and in one mesocosm (oysters + seagrass co-culture, warmed and low pH combined) with APEX temperature and lab-grade pH probes. We also assessed daytime water chemistry within mesocosms by taking discrete water samples and spot measurements using a YSI sonde (YSI 556 multiparameter handheld). Once a week, we took water samples from twelve haphazardly selected mesocosm (representing each of the twelve culture/co-culture, temperature, and pH combinations). We filtered all water samples through 0.2 µm GF/F filters into borosilicate glass containers and fixed them with 100 µL of HgCL2 (Dickson et al. 2007). Two bottles were collected for each mesocosm at each sampling interval, one for pH and one for Total Alkalinity (TA). We partitioned each bottle sample into two cuvettes, which were each analyzed in triplicate. We analyzed water sample pH spectrophotometrically (Cary 300 UV-Vis Spectrophotometer) using the m-cresol purple sodium salt (mCP: 10mM) method (Dickson et al. 2007, Dickson 2010). We quantified total alkalinity via titration using a Metrohm 905 Titrando (Dickson et al. 2007). In-situ pH was calculated using the ‘pHinsi’ function in the seacarb package in R (Gattuso et al. 2015). We took pH measurements within all 60 mesocosms using the YSI during one mid-July water sampling session. For this sampling date, YSI pH readings (mV) were regressed against in-situ pH values as a means of confirming the efficacy of our YSI and APEX probe-based pH measurements (R² = 0.893, p < 0.001). Because our sampling design for water chemistry did not replicate measurements across treatment combinations within a sampling timepoint and there were rapid changes in water chemistry over the 4-hour sampling period (from 10:00 AM – 2:00 PM), these measurements are only sufficient for confirming the efficacy of the pCO₂ treatments and not for assessing species effects on daytime pH within mesocosms.