Mussels (M. trossulus) were collected in May 2012 from Argyle Creek, San Juan Island, WA (48.52˚ N, 123.01˚ W) and held in a mesh box submerged under the dock at Friday Harbor Laboratories (FHL), San Juan Island, WA for up to 14 d. Mussels were placed in experimental mesocosms in the Ocean Acidification Experimental Laboratory (OAEL) at FHL as described in O’Donnell et al. (2013) and Timmins-Schiffman et al. (2012). Briefly, manipulations of pH were made by bubbling CO2 into a 150 L temperature-controlled seawater reservoir, that supplied water to eight 3.5 L chambers at a turnover rate of 50 ml min-1. Air was bubbled into the reservoir to maintain 100% oxygen saturation and submersible pumps (model number P396, Annex Depot, Sacramento, CA) provided mixing in the chambers at 3.8 L min-1. The bottom of each chamber was lined with autoclaved pebbles, collected from an FHL beach, to provide a substrate for byssal thread attachment. pH and temperature were monitored continuously in each water reservoir with a Durafet pH and temperature probe and the full carbonate chemistry of the system evaluated with DIC and Total alkalinity measurements once during each trial. Mussels were acclimated to their treatment temperatures in ambient pH (~7.8) over 9 d, ramping temperature up no more than 2˚C per day, and fed a maintenance level of Shellfish Diet 1800 (6 g l-1 day-1, Reed Mariculture, Campell, CA, USA).

The twelve independent temperature x pCO2 treatments spanned the range of local marine conditions (Newcomb, 2015; George et al., 2019; temperature at 10˚C, 18˚C, or 25˚C and pCO2 at 400, 750, 1200, or 2500 µatm). Each mussel was trimmed of external byssus before placement in an experimental treatment for 3 d, sufficient time to produce new mature byssal threads (Bell & Gosline 1996) while minimizing the effect of treatment on mussel condition. Mussels were starved during the 3 d trials to minimize changes in chamber water chemistry due to food addition and to reduce fouling. Three trials were conducted in succession to replicate treatments over time, increasing sample size (n=8 x 3) for each temperature*pCO2 treatment.

At the end of each trial, mussels and the rocks to which they had attached with byssal threads were removed from the chambers. The entire byssus was dissected from each mussel and stored air-dried for up to 20 days. Byssus was rehydrated in seawater prior to testing, a method that does not alter the mechanical properties of the byssal threads (Brazee, 2004). The number of byssal threads each mussel produced was counted, and one thread was haphazardly chosen for mechanical testing following the procedure of Bell & Gosline (1996). Briefly, an individual thread was clamped with submersible pneumatic grips on either end by holding the proximal byssal stem between cardstock with cyanoacrylate glue and affixing the distal plaque with attached rock to an aluminum T-bar with epoxy. An Instron 5565 tensometer (Norwood MA, USA), extended the thread at a rate of 10 mm min-1 in a temperature-controlled water bath (3130-100 BioPuls Bath, Instron, Norwood, MA, USA) until failure. The tensometer measured force (±10-3 N) and extension (± 10-3 mm) at 10 Hz. Tests were performed in seawater with a pH of 7.8 and the relevant treatment temperature.

Pull to failure mechanical tests provided estimates of thread breaking force, yield force, extensibility, initial stiffness and failure location (Bell & Gosline 1996). Yield, due to quasi-plastic deformation in the distal region, was identified as the point where the initial slope of the force-extension curve decreased by 40%. Extensibility was calculated by dividing thread extension at failure by initial length and initial stiffness was determined from the initial slope of the force extension curve. The location of failure (proximal, plaque, and/or distal region) was noted and threads were retested to quantify the breaking force of each remaining region. Tests that broke at the grips were considered underestimates and were discarded.

The cross-sectional area of the proximal region was measured to evaluate morphological differences among treatments. The elliptical area was estimated from measures of the major and minor axes (+ 1 um using a dissecting microscope (Brazee & Carrington 2006). Proximal breaking stress (N mm-2), a material property, was calculated as proximal breaking force divided by proximal area. Thread surface structure was examined using a scanning electron microscope (FEI Sirion XL30 SEM, Hillsboro, OR).

Whole mussel attachment strength was estimated using two mathematical models developed by Bell & Gosline (1996). Each model assumes a mussel is anchored with a constant thread number (n=50) arranged in a circle. The normal model estimates dislodgment force perpendicular to the substrate (e.g, lift); all threads are engaged and extend until they reach their maximum force. The parallel model estimates dislodgement force for an animal pulled parallel to the substrate (e.g., drag); threads on the upstream side are the first in tension, yield and extend until they reach maximum force and break, while more threads are recruited into tension until they have all broken. Additionally, we modified each model to incorporate the variation in thread production across treatments. Because thread production was measured for only three days, treatment means were scaled to a maximum value of 50 threads.

Detailed methods and results are provided in Newcomb, 2015 and Newcomb et al., 2019

Location: Friday Harbor Laboratories, Friday Harbor WA