Oyster spat were exposed to one of the following cues in individual aquaria: blue crab urine (mud crab or oyster diet), trigonelline (24.6 µM), homarine (15.1 µM), or trigonelline (24.6 µM) + homarine (15.1 µM). Additionally, we used natural seawater (settled for 3 days to remove sediment) as a negative control and predator water as a positive control.

Diploid oyster spat were purchased from the Auburn University Shellfish Laboratory and settled onto 4.5 cm × 4.5 cm marble tiles. For one week after settlement, spat on tiles were caged and kept in 1250 L mesocosms with natural flowing seawater from Mobile Bay at a flow rate of 20 L/min. We then ensured each tile had at least 15 oyster spat and used high-density polyethylene fishing line to tie tiles back-to-back. Three tile pairs were placed in each aquarium, ensuring that every tile pair was upright to maintain good water flow around the spat. An intact, sun-bleached adult oyster shell was also placed into each aquarium for spat tile pairs to lean on so that they maintained an upright position. Aquaria were filled with 2 L of natural seawater (with the exception of the predator water control, which received 1.5 L seawater + 0.5 L predator water) that had settled for at least 3 days to remove sediment particles. Seawater was supplemented with either Instant Ocean salt or deionized water to reach 20 ppt (± 2 ppt). Each aquarium contained filtered air bubblers for oxygenation. Aquaria were covered with lids to reduce evaporation and stored outside under a covered pavilion in a water bath containing ambient flow-through seawater to regulate temperature. Spat were fed Instant Algae Shellfish Diet 1800 (Reed Mariculture). At the start of the experiment, spat were fed 0.5 mL twice daily, but we increased this amount to 1 mL twice daily as spat grew larger. Complete water changes and aquarium cleanings were performed twice weekly, and 1 mL of predator chemical cues (i.e., blue crab urine and pure chemical compounds) were added to the aquaria immediately after water changes. Predator water was created by housing 6 blue crabs (carapace width 12-17 cm), in a 238 L volume mesocosm, and feeding each crab an adult oyster (length 6-7 cm). After 3-5 hours, 500 mL of this water was added to each positive control aquarium.

Homarine (15.1 µM) and trigonelline (24.6 µM) solutions used to induce oyster spat were prepared at natural concentrations found in urine of blue crabs fed an Eastern oyster diet (SI Appendix, Fig. S1 in Roney et al. (submitted)). Homarine (Santa Cruz Biotechnology Company) and trigonelline (Toronto Research Chemicals) were dissolved in deionized water at 27x and 97x (respectively) of the final concentrations used for the bioassay. These stock solutions were aliquoted and frozen at –80 °C. On days designated for cue addition, stock solutions were diluted to 1x concentration and 1 mL was added to each corresponding replicate aquarium. To prepare the trigonelline + homarine treatment, 2x concentration homarine and trigonelline solutions were mixed together in a 1:1 ratio to form a 1x concentration mixture.

Predator urine was collected from blue crabs fed two different diets, oysters or mud crabs (Panopeus herbstii). Blue crabs were collected from crab pots near Dauphin Island, AL, USA and housed in 1250 L mesocosms flowing with natural seawater. Crab size ranged from 11-19 cm, measured from spine to spine on the widest part of the carapace. Crabs were starved for 2-3 days and then were each fed either one adult oyster (~6-7 cm length) or ~5 g of frozen mud crabs every 48 hours to standardize diets. Mud crabs used for feeding were collected from either Priest’s Landing, Skidaway Island, GA (31°57'44.89"N, 81° 0'48.22"W) or the North Inlet Estuary, SC (33°21'52"N, 79°10'03"W). Crabs were maintained on these diets for one week prior to urine extraction to ensure all extracted metabolites were from the specified diets. Urine was collected from individuals twice a week. Crabs were cooled to quiescence, then a 23 gauge-needle was inserted approximately 2 mm into the nephropore and urine was extracted with gentle vacuum suction into clean glass vials. Urine used for the experiment was clear or yellow in color and foamy; urine was discarded if it appeared cloudy or bluish-gray in color, as this indicated contamination with hemolymph. Urine was frozen at –80 °C immediately after collection. Oyster diet urine was collected from a total of 161 crabs, each crab provided on average 4.08 ± 3.11 mL per collection, and was extracted from 2.2 ±1.0 times on average. Mud crab diet urine was collected from approximately 102 crabs; each crab provided on average 5.04 ± 4.07 mL per collection and was used for extraction 2.0 ± 1.0 times on average. We later combined the urine of different individuals into 12 mixtures using the fewest individuals possible, where each mixture was considered a biological replicate (Table S1 in Supplemental File PDF). These mixtures were then partitioned into 1 mL aliquots and stored at ‑80 °C until use.

In this bioassay, all treatments and controls had 12 replicates, and all treatments were run simultaneously. However, the replicates were divided among two blocks, n=3 per treatment (21 total aquaria) in the first block and n=9 per treatment (63 total aquaria) in the second block. The blocks began approximately 4 weeks apart and both experimental blocks were planned for 8 weeks, however, the first block of the experiment ended early (June 5, 2020 through July 6, 2020, total 31 days) due to high levels of mortality and the second block of the experiment was disrupted (July 9, 2020 through August 23, 2020, total 48 days) due to heavy storms in Dauphin Island, AL.

At the completion of the experiment, spat from each aquarium were randomly selected for assessment of shell strength. Approximately 20 oysters were crushed from each aquarium (distributed across tile pairs), except for one predator water replicate (15 oysters) and one trigonelline replicate (19 oysters) due to high mortality. Individual spat width was measured for each crushed oyster to 0.01 mm using a Vernier digital caliper. The force required to crush oysters was measured using a Kistler 5995 charge amplifier and Kistler 9207 force sensor following standard protocol (Robinson et al., 2014). Crushing force was divided by spat width to produce a size-standardized metric of shell strength (i.e., standardized crushing force, N/mm) because larger individuals typically have a stronger shell as a byproduct of their size. Standardized crushing force measurements for oysters within the same aquarium were averaged, however data were removed for outliers, dead oysters, and unreliable measurements resulting in fewer spat contributing to an aquarium average. Due to a batch effect (i.e., all seawater replicates in block one were statistical outliers), all block one samples were excluded from further statistical analyses. Additionally, two predator water replicates, and one trigonelline + homarine replicate were excluded from statistical analyses due to high mortality. As a result, there were 9 replicate aquaria per treatment with the exception of predator water (n=7) and trigonelline + homarine (n=8) treatments.

Instruments:
Individual spat width was measured for each crushed oyster to 0.01 mm using a Vernier digital caliper. The force required to crush oysters was measured using a Kistler 5995 charge amplifier and Kistler 9207 force sensor following standard protocol (Robinson et al., 2014).

Problems/Issues:
Only live oysters were crushed for the experiment. If an oyster was crushed and found to be dead (no soft tissue within shell), that was noted on the data file to be excluded from the analysis. Due to this, some treatments had fewer replicates than others, but this was accounted for in data analysis. Data were only included in the statistical analysis if there was no note in the life_status column or if it was specifically labeled as alive. All of Block 1 was later excluded from the statistical analysis for this publication due to this high mortality (more detail in the analysis description of the methods above).

life_status notations and meanings:

"a" = Spat was alive and it was not excluded from initial analysis (though block 1 data was excluded from final analysis for publication regardless of life status)
"?" = Unclear if the spat was alive or dead. These rows were excluded from the statistical analysis
"most dead" = each row in the data refers to a single spat. This means the spat in this row was dead
"lots of dead" = each row in the data refers to a single spat. This means the spat in this row was dead
"d" and "dead" = each row in the data refers to a single spat. This means the spat in this row was dead
"tiles looked bad/dead, algae growing" = treatment excluded from analysis