Juvenile Atlantic cod behavior appears robust to near-future CO2 levels

Fish rearing and treatment

The ethical committee for animal experimentation approved all experiments (Gothenburg,
Sweden, ethical permits Jutfelt 100–2010 and Jutfelt 151–2011).

The fish (56 juvenile Atlantic cod) were captured by Kristineberg Marine station (lat.
58.249717, long. 11.445522), Lysekil, Sweden, with hand nets, seine nets and fish
trap cages. Length and weight was measured for each individual at the start of the
experiment. Fish were randomly distributed to four tanks, two for each treatment.
All experiments were conducted in thermally constant rooms that ensured temperature
stability at all times. The light regime was L16:D8 h.

Four dark cylindrical fiberglass 100 L aquaculture tanks with coned bottoms (Strandvik
Plast AS, Strandvik, Norway) were continuously supplied with flow-through deep water
that was continuously pumped from 30 meters depth in the fjord. Each fish tank was
supplied with flow-through water at 3 L per minute from its own 200 L header tank,
which were heavily aerated and in two of the header tanks the CO₂ levels were manipulated. The fish were thus exposed to control water (532 ?atm?±?43
SD) or water with elevated CO₂ levels (1014 ?atm ±76 SD) for the duration of the 30 day exposure period. The exposure
duration was chosen to presumably allow acclimation, as acclimation to other environmental
stressors such as temperature occurs over days to weeks 29]. The fish tanks were covered with clear plastic lids to reduce gas exchange. The
pCO₂ of the fish tanks was measured at least once daily using an infrared pCO₂ measurement system, a Vaisala GM70 (Vaisala, Helsinki, Finland) connected to a gas-permeable
silicone membrane according to 30]-32]. pH-stat computers (Aquamedic, Bissendorf, Germany) controlled the pH of the two
CO₂-manipulated header tanks by bubbling of pure CO₂ (Aga gas, Gothenburg, Sweden) through solenoid valves. The temporal pH variance of
the header tanks was low (0.1 pH units), and the pH variability of the fish tanks
was negligible (0.05 pH units). Temperature and salinity were recorded continuously,
with a mean temperature of 14.4°C?±?0.44 (SD) and a mean salinity of 33.1?±?0.8 (SD)
PSU. The alkalinity of this deep water supply was very stable and titrated weekly.
Oxygen saturation in the fish tanks was always above 90%. The carbonate chemistry
was calculated in CO2calc (Hansen, USGS, USA) using data for pCO₂, salinity, temperature, and alkalinity. The results are shown in Table 1.

Table 1. Water chemistry for the treatments Control and Elevated CO₂

The cod were fed shrimp daily, and any mortality was counted. The fish were not individually
tagged because they were to be released after the experiment. The starting lengths
and weights, as well as the mortality in the tanks, are presented in Table 2. As mortality occurred in all tanks, it was not possible to determine the growth
rates. Half of the mortality was due to cannibalism; a few fish escaped, and the rest
died of unknown causes without obvious pathology. After the behavioral tests, the
fish were kept for chemosensory evaluation in a related experiment, the results of
which have been published in Jutfelt and Hedgärde 31]; thereafter, the fish were released back into the fjord at the site of capture.

Table 2. Tank means of the initial lengths and weights of the fish

From exposure day 12 and onwards, the fish were used in behavioral tests. The test
fish were randomly sampled from the tanks, and most of the fish were used in all trials.
All tests were performed during the daytime, at the treatment water pCO₂, temperature and light conditions to which the fish were acclimated. The water in
the test chambers was taken from the respective treatment tanks and replaced between
trials to maintain water quality and pCO₂. The pH of the trial water did not differ from pH in the treatment tanks. All trial
tanks were visually shielded from disturbance.

Activity trials

The activity trials were performed during day 12 to 19 after exposure initiation.
The activity arena was a 50×50 cm Plexiglas tank with a painted grid bottom forming
nine equal squares, and the water depth was 15 cm. The tank was visually shielded
from disturbance. A video camera positioned above the tank was used to record the
experiments. Single fish were placed in the tank and filmed for one hour, of which
the first 30 minutes were considered the acclimatization time. During the remaining
30 minutes, the activity was measured as the number of lines crossed, similar to the
activity experiment performed by Munday et al. 10]. The data are presented as lines crossed per minute.

Emergence from shelter

The experiment on emergence from shelter was performed after 26 days of exposure.
The emergence-from-shelter test is commonly used to estimate fish boldness 33]. The arena used for the emergence-from-shelter experiment was a 50 × 50 cm plastic
tank with a rock and plastic algae shelter (10 × 10 cm) in one corner, and the water
depth was 15 cm. A video camera positioned above the tank was used to record the experiments.
Each fish was left to acclimatize to the experimental tank for 5 minutes before it
was chased by hand into the shelter. The chasing was short and similar for all fish,
and the fish quickly entered the shelter. The time to emerge from the shelter was
then measured using the video footage, similar to the methods described in Munday
et al. 10].

Lateralization

Lateralization was tested on days 29 and 30. A double T-maze runway was used for the
lateralization trials using the methods described previously 12],13]. The fish were given 5 minutes to acclimatize to the T-maze before starting the trial.
The fish were gently encouraged down the central channel using a plastic rod (approaching
but not touching the fish) and forced to make a turning decision at each T-crossing.
The fish swam spontaneously and did not require much encouragement to follow the central
channel and make turning decisions. The procedure was then repeated in the reverse
direction. In total, 14 decisions per fish were recorded. The relative and absolute
lateralization indices were calculated using:

and

according to 34].

Data analysis

Statistical analyses were performed in SPSS with a significance level of 0.05. Because
all data were normally distributed (Kolmogorov-Smirnov p??0.05), parametric tests
were used. A nested ANOVA (tank nested under treatment) was performed for activity,
shelter, relative lateralization and absolute lateralization. When tank effects were
found, a post hoc test was used to determine which tanks were significantly different.
The data are presented as the mean?±?SEM, unless otherwise noted.