Study species and handling practices

The eastern oyster, Crassostrea virginica, is a shallow water mollusc native to the North Atlantic. Adults commonly form reefs in the intertidal and sub-tidal zones and are broad- cast spawners, producing planktotrophic, free-swimming larvae with a pelagic duration of 2-3 weeks [101, 6]. The penultimate and final larval stages are the veliger and pediveliger stages, respectively. Larvae in both stages are characterized by a ciliated, retractable velum extending from a calcareous shell, which they use to swim and feed. Larvae become pedi- veligers upon the development of a foot and pronounced eyespot, at which point they are competent to settle out of the plankton and metamorphose [103].

2015. The first spawn (Spawn 1) was obtained from the Martha’s Vineyard Shellfish Group (Martha’s Vineyard, MA, USA) and the second (Spawn 2) from the Aquaculture Research Corporation (Dennis, MA, USA). Both spawns were obtained when the larvae were mature veligers, and maintained prior to experiments in identical culture conditions: 1 𝜇m-filtered, aerated seawater at ambient field temperature (20-22∘ _{C) and salinity (33 psu), in covered}

16 L plastic buckets. Larvae were kept at low densities to minimize interactions (< 3 larvae mL−1_{) and fed daily a suspension of haptophyte Isochrysis sp. (∼ 9 × 10}5 _{cells mL}−1

in filtered seawater). Larvae were maintained in these conditions (∼ 2 − 3 days) until > 50% of larvae exhibited eyespots (an indicator of competency to settle), as determined by microscopic examination of a random sample of larvae. Immediately post-observation, larvae were retained from the sampling flask on a 100 𝜇m mesh filter and subsequently preserved in a solution of 95% ethanol. Shell size measurements and eyespot identification, from microscopic inspection, were recorded for a subsample of specimens within 24 hours of preservation. Larval size was estimated by measuring width (perpendicular to shell hinge) and height (parallel to hinge) and treating larvae as ellipsoids, so that larval size is measured in surface area.

6.3.2 Experimental setup

All experimental observations were conducted in an environmental chamber at a constant temperature of 20∘_{C, in the dark. The observational tank was a 50 mL flat-sided plastic flask}

with an open top, filled with either 1) ambient seawater (20∘_{C, 33 psu) filtered to 1 𝜇m, or 2)}

oyster bathwater at identical temperature and salinity (see below for preparation method). Larvae swam into the flask via a gravity-assisted 1 mL pipette suspended above the flask, with the pipette tip just breaching the water surface; this method of introduction imposed no external downward momentum on larvae. Approximately 40-50 larvae were introduced to the tank for each observational period. The tank was illuminated from behind with a near- infrared LED array light source (Olymstore, 12V, 2A, 850 nm) and a monochrome camera (Hitachi KPF-120) recorded a 4 × 5 cm 2-dimensional field of view, which encompassed a vertical cross-section of the entire flask volume. The bottom of the flask consisted of a smooth plastic surface with no biofilm, a poor settlement surface for oyster larvae [166]. The flask was also allowed to rest prior to the addition of larvae, and maintained at the same temperature as the environmental chamber, to minimize convective currents or other flow

fields from forming.

For both experimental spawns, a subsample of larvae was exposed to the control filtered seawater (SW) and a subsample was exposed to the oyster bathwater (OB). Spawn 1 had larvae exposed at time 0, 4, 8, and 12 hours post-competency, while Spawn 2 had larvae exposed at time 0, 10, 22, 43, and 64 hours post-competency. For both spawns, 𝑡 = 0 hours was defined to occur when >50% of larvae displayed evidence of competency. At each time point, 5 replicate observations were taken for each of the SW and OB treatments, with each replicate entailing five minutes of video observation at 30 frames per second of larval motion in the field of view. Experimental images were saved as high resolution TIFF files (1040 × 1390 pixels) for subsequent larval tracking.

Oyster bathwater was prepared in a similar method to that described in Tamburri et al. [40]. Live, unwashed, adult C. virginica were purchased from a local vendor (The Clam Man, Falmouth, MA, USA) from a recent harvest of farmed oysters in Waquoit Bay (MA, USA). Individuals were measured to estimate shell surface area, and a set of oysters totaling 800 cm2 _{surface area were placed in a sterile plastic bucket with 4 L of aerated seawater}

filtered to 1 𝜇m (ambient 20∘ _{C, 33 psu). The bucket was covered and left undisturbed}

in the environmental chamber for 4 hours. Subsequently, the adults were removed and the bathwater filtered to 0.7 𝜇m through a glass microfiber filter. The bathwater was then divided into 50 mL aliquots and frozen at -20∘ _{C until immediately prior to use.}

6.3.3 Larval tracking

The methods for larval identification and tracking were adopted from Wheeler et al. [28, 29] and briefly summarized here. TIFF images were imported into LabVIEW 2013 (National Instruments) and average background intensity was subtracted. Using a fixed threshold particle size and intensity, larval centroid positions (𝑥, 𝑧) were recorded in the frame which they appeared. Centroid positional data were reconstructed into individual larval trajec- tories using a MATLAB script which tracked a larva from frame to frame according to a subsequent-frame tolerance distance radius set by the user (e.g. Figs. 6-1 a, c). Larval trajectories were truncated by five frames at the terminal points of the trajectories to avoid poor centroid estimates in cases where larvae passed laterally into and out of the focal plane. Positional probabilities were calculated by defining a grid in the (𝑥, 𝑧)-space and determin- ing the number of larvae appearing in each grid box over the total observational period,

normalized by number of observations (e.g. Figs. 6-1 b, d). Instantaneous swimming veloc- ities were computed using a central difference scheme of larval centroid positions in time, so that the velocity is defined centered in time between two adjacent frames. Vertical and horizontal distributions of position and velocity were constructed from mean positions and velocities of individual larval trajectories; these distributions were used for subsequent sta- tistical analysis. Unlike the set-ups described in [28, 29], flow velocities in this small tank were minimal and therefore no effort was made to subtract local flow away from observed larval velocities.