Experimental design
Experiments subjecting juvenile rockfishes to simulated future DO levels were conducted at the seawater aquarium facility at the NOAA Southwest Fisheries Science Center laboratory in Santa Cruz, CA. Juvenile rockfishes were exposed to one of four treatment levels corresponding to conditions that currently occur or are predicted to occur in the future on the central California coast: 100% saturation (8.74 ± 0.03 mg O2 L-1), 68% saturation (6.00 ± 0.04 mg O2 L-1), 46% saturation (4.06 ± 0.04 mg O2 L-1), or 26% saturation (2.25 ± 0.05 mg O2 L-1), with two replicate tanks per treatment level. These levels were maintained for the duration of the experiment to simulate chronic exposure to prolonged hypoxia events and oxygen, pH, and temperature levels in all the treatment tanks were measured daily using a HACH HQ40D multiparameter meter. All experimental trials were conducted under constant temperature (12˚C) and pH (~8.05), with the same DO levels as the rearing tank of the fish being tested.

To manipulate oxygen levels, seawater was first pumped from offshore through a series of settling tanks and sand filters and held in a 2000 L reservoir tank maintained at a constant temperature (12˚C) with aquarium chillers and aerated to approximately 100% air saturation (~9.0 mg O2 L-1). This source water was fed into four 500 L treatment reservoirs, which were used to create desired DO concentrations by bubbling nitrogen (N2) gas to strip O2 from the water. Gas delivery was controlled by the program WitroxView via solenoid valves, and O2 levels were monitored using Loligo Systems optical oxygen probes. Manipulated (or control) water was delivered at a rate of 20 ml s-1 to 80 L experimental tanks in a single-pass, flow-through design. 

Ten randomly assigned juvenile rockfish of each species were introduced into each of two replicate tanks for each treatment (20 fish per treatment per species). All tanks were covered to minimize visual disturbance from investigators. Sections of plastic construction fencing material measuring 90 x 120 cm were bunched together and placed in each tank to simulate kelp habitat structure. The two replicate tanks per treatment were fed on alternating days and used in experiments on non-feeding days, thereby ensuring a 36-48 hour fasting period prior to behavioral or physiological trials. Fishes were fed to satiation with frozen, high protein krill on feeding days. Prior to conducting any behavioral or physiological measurements, fish were allowed a minimum of 5 days to recover in their treatment tanks from any previous trial. See Table 1 of Mattiasen et al. (2020) for the schedule of the various experimental trials.

Escape response
Escape response trials tested the time required for a fish to find the exit of an enclosed chamber (Jutfelt et al. 2013). Escape chambers were composed of a PVC tube measuring 28 cm tall x 9 cm diameter with a 5 cm diameter hole cut in the side. A slit located 8 cm from the top of the chamber allowed a black plexiglass divider to be inserted, retaining fish in the top half of the chamber during the acclimation period. Removal of the divider released the fish into the lower portion of the chamber at the start of the timed trial. Escape chambers were placed in 40 L insulated aquaria on a water table to control temperatures. A total of six replicated escape chambers and aquaria were used, allowing for six simultaneous trials. Individual fish were transferred into the top of the chamber and allowed a 15 min acclimation period. At the end of the acclimation period the divider was removed without visual interference by the investigator. Observers watched a mirror above the tanks and recorded the time at which each fish exited the chamber (defined as the time at which the head of the fish exited the chamber). Trials were terminated after 10 min regardless of whether the fish exited.

Behavioral lateralization
Brain functional asymmetry and behavioral lateralization reflect the bias for left vs. right turning decisions in a detour test. To measure lateralization in response to DO treatment conditions, a detour test was employed with a double T-Maze (Domenici et al., 2007). Individual fish were transferred into one end of the two-way T-Maze (50 x 30 x 25 cm L x W x H aquaria), and allowed to acclimate for 3 minutes. The starting side was alternated for every trial to minimize the potential for side bias. After the acclimation period, the fish was gently coaxed to swim down the center channel (without touching the fish) using a long PVC bar, and when it reached a barrier at the end of the channel, the fish had to decide to turn right or left. The turn direction was recorded and the experiment was repeated 10 times, 5 times in each direction. Each trial took approximately 10 minutes to complete. Absolute lateralization (LA) was calculated as

LA = (|# right turns – # left turns| )/(# right turns + # left turns) x 100

as an index of non-directional turn bias. LA reflects whether turn bias exists at the population level, irrespective of direction. Relative lateralization (LR¬) was also calculated to determine whether the fish in a particular treatment exhibited turning bias for a particular direction (i.e., left or right preference). LR was calculated as

LR = (# right turns – # left turns)/(# right turns + # left turns) x 100

Positive values indicate a right turning bias, while negative values indicate a left turning bias.

Critical oxygen tension (pCrit)
A subset of 8 individuals per species from each treatment was tested for hypoxia tolerance by estimating pCrit using an automated intermittent flow respirometry system (Loligo Systems). Fish were placed in sealed respirometry chambers overnight at their treatment oxygen levels to acclimate to the chambers. Subsequently, pCrit trials were initiated by raising the oxygen saturation of the reservoir to 70% air saturation and three MO2 measurement loops (5 min flush, 10 min wait, 5 min measurement) were recorded at each oxygen level. The DO level was then reduced in a step-wise fashion by 10% air saturation, through the addition of N2 gas until reaching 40% air saturation, below which oxygen saturation was reduced at 5% intervals until reaching 10% air saturation, at which point the trial was terminated. This approach allowed us to obtain a more precise measurement of pCrit and to reduce risk of inadvertent mortality.

Oxygen consumption rate (MO2 in mg O2 kg-1 hr-1) was calculated using the following equation:

MO2 = ∆PO2 V∝M-1 ∆t-1

Where ∆PO2 is the change in water partial pressure of O2 (mmHg), ∆t is the elapsed time (h), V is the volume of the respirometer chamber minus the volume of the fish (cm3), M is the total mass of the animal (kg), and ∝ is the O2 solubility coefficient at the experimental temperature (Boutilier et al., 1988). The respirometry system was cleaned using dilute bleach after each trial to eliminate the influence of microbial respiration on subsequent trials. pCrit was calculated for each fish using the broken stick regression method (Toms and Lesperance, 2003) by computing the oxygen saturation level at which the metabolic rate began to decrease linearly with decreasing DO.

Aerobic scope
Aerobic scope is the difference between the standard (or resting) metabolic rate (SMR) and the maximum metabolic rate (MMR). We measured the SMR on a subset of 8 fish per species per treatment using the intermittent flow respirometry system. Four individuals at a time were weighed and placed into separate respirometer chambers, with MO2 measurements taken over a 12 hr period overnight. SMR was measured during nighttime hours to capture the MO2 at the time where the fish were at their lowest metabolic activity levels. The lowest 10% of MO2 measurements per cycle, excluding outliers (values > 2 standard deviations), were used to calculate the SMR of each individual fish (Clark et al., 2012). MMR was subsequently measured following swimming to exhaustion using a Loligo Systems 10 L swim flume (model #SW10100). Exhaustion was achieved by swimming the fish for 5 minutes at a velocity one-body length per second below the estimated average critical swimming speed of the group (N. Kashef, unpublished data). The fish were then quickly returned to the respirometry chambers and run for one measurement cycle to acquire MMR. Preliminary trials concluded that the highest MO2 values occurred directly following swimming to exhaustion. Aerobic scope was calculated by subtracting the SMR from the MMR.

Ventilation rate
Ventilation rate was measured on a subset of 10 individuals per species per treatment using a specially designed array of 10 experimental chambers (5 x 15 cm, water depth 4 cm), each holding an individual fish. Each chamber received flow-through seawater of the appropriate rearing DO treatment and constant temperature (12˚C). Following a two-hour acclimation period, two GoPro HERO4 video cameras recorded each fish for 30 minutes. Ventilation rate was determined by counting the number of open/closing cycles of the gill operculum within a minute (i.e., ventilations per minute [VPM]). Average VPM for each fish was calculated for 3 randomly selected one-minute measurements.

Problem report: Some individual fish were used in each of the behavioral and physiological trials, while other fish were only tested in a subset of the possible experimental trials. In addition, if any fish died in the course of the experiments, they were replaced by a new fish.