Dataset: High-frequency measurements of mussel gaping behavior and dissolved oxygen across flow treatments in laboratory flume experiments at Friday Harbor Laboratories from October to December 2023

Laboratory flume experiments were conducted at Friday Harbor Laboratories (University of Washington, Friday Harbor, WA, USA; 48°32'45.7249" N, 123°00'46.9262"W) in late Fall 2023 to quantify mussel gaping behavior and dissolved oxygen dynamics under controlled flow conditions. Experiments used monolayer aggregations of three mytilid mussel species: Mytilus trossulus, Mytilus galloprovincialis, and Mytilus californianus. The mussels were obtained from three different locations. Both bay mussel species were ordered from commercial suppliers-- M. trossulus (MT) from Penn Cove Shellfish Farm (Coupeville, WA, USA; 48°13'06.5579" N, 122°42'25.4331"W) and M. galloprovincialis (MG) from Taylor Shellfish Farm (Totten Inlet, WA, USA; 47°08'23"N 123°05'26"W). California mussels (MC; Conrad, 1837) were obtained from Cattle Point on San Juan Island, WA (48°26'59.6811"N, 122°57'51.9358”W). All mussels were maintained in sea tables (66 × 135 × 32 cm) at ambient seawater temperature with a constant flow of seawater prior to experimentation.

For each species, a single-species mussel aggregation was assembled on a flat experimental platform (100 × 36 × 2 cm, L × W × H) constructed from a 1.3 cm plexiglass sheet atop a 0.7 cm egg crate louver and placed in the working section of  a recirculating laboratory flume (152 × 38 × 51 cm; Rolling Hills Research Corporation, Model 1520 Water Tunnel). Flow moved unidirectionally across the mussel aggregation before recirculating as the flume operated as a semi-closed system, continuously flushing and cooling the 1520 L recirculating volume with ~2 L·min⁻¹ of seawater from the Friday Harbor Laboratory supply, which kept temperatures between 10–12°C across trials. Aggregations of each species were tested in an independent 10-day trial, and flow treatments were randomized within each trial to avoid confounding flow speed with day-of-experiment. We intentionally did not impose an identical flow sequence across species because doing so would systematically link particular flow speeds with specific trial days (e.g., early vs. late in the experiment), introducing potential order effects such as acclimation, fatigue, or cumulative stress. Our primary design priority was to prevent behavioral responses at a given flow speed from being confounded with its position in the treatment sequence. Randomizing flow within each species ensures that responses reflect the imposed flow speed rather than time-dependent effects within the 10-day trial. While species may differ in their physiological responses to flow, the purpose of this design choice was to minimize within-species sequence effects; interspecific differences in sensitivity to flow are evaluated in the results and are not driven by treatment order.

Flow treatments (F1-10) spanned flow speeds from 1.5 to 30 cm s-1 and were maintained continuously for 24 hours. Between each flow treatment, the flume was flushed by increasing the flow speed to 35 cm s-1 for 15 minutes (flushing periods were not included in the data). Experiments were conducted sequentially by species, with identical flow protocols applied to each aggregation.

At the end of each trial, we counted the mussels in the aggregation and recorded mortality rates. We measured several morphometric parameters for a subset of mussels (all mussels with a gape sensor plus a random sample of 30 mussels, n = 46). Specifically, we measured mussel shell length (l), width (w), and height (h) to the nearest 0.1 mm using calipers, and dry weight of body tissue removed from the shell and dried to a constant weight for 72 hours at 65°C.  Mussel biomass density (g cm⁻²) was calculated as dry weight per planform area, assuming an elliptical shape (for a vertically oriented mussel) with shell width and height as the major and minor axes, respectively (Bell and Gosline,1997). This functional trait of an individual mussel can be readily scaled up to whole bed biomass density (also in g cm-2), a metric of interest to benthic ecologists, by multiplying by π/4 (the ratio of the area of an ellipse to its bounding rectangle).  

We measured both dissolved oxygen (DO) and mussel gaping behavior throughout each species trial. We measured DO at 1-minute intervals during each trial using HOBO loggers (HOBO U26-001; Onset, MA-USA) placed within the interstitial zone (in bed) and upstream of the mussel aggregation. We affixed the loggers within the interstitial zone to the center of the plexiglass platform supporting the mussel aggregation. Before each species trial, we calibrated the sensors using the manufacturer’s recommended methods and corrected offsets between loggers before analysis.

We measured mussel gaping behavior using gape sensors composed of magnetic Hall Effect sensors (Allegro Microsystems A1393, Worcester, MA, USA) and a circular magnet (part number 8195-Radial Magnet Inc. Boca Raton FL). We attached the gape sensor to the posterior end of one valve and the circular magnet to the opposite valve using marine epoxy (Splash Zone, KOP-COAT Inc., Rockaway, NJ, USA). We sampled gape sensor output of the distance between the valves every 5 seconds, a sampling rate our preliminary observations determined to be sufficient to characterize gape behavior dynamics. We equipped a total of 16 mussels per species with gape sensors for each trial, spaced more than 10 mussel body lengths apart to prevent sensor interference. Gape sensors were continuously monitored during each trial using a MusselTracker datalogging system (Miller and Dowd, 2017). At the end of each trial, each gape sensor was calibrated by severing the mussel’s adductor muscle and using calipers to set valve gape from 0 to 20 mm in 1 mm increments while measuring the sensor's voltage output. An exponential function was then fit to this calibration data to estimate gape in millimeters (mm) from the voltage output for each sensor, with a sensor resolution of < 0.1 mm. Because each sensor is built by hand and has its own calibration curve, data for both the calibration and raw voltages are not included here but can be requested by contacting the author.