The effect of carbon dioxide on growth of juvenile Atlantic cod Gadus morhua L.

Aquatic Toxicology 102 (2011) 24–30

Contents lists available at ScienceDirect

Aquatic Toxicology journal homepage: www.elsevier.com/locate/aquatox

The effect of carbon dioxide on growth of juvenile Atlantic cod Gadus morhua L. D. Moran a,b,∗ , J.G. Støttrup a a b

DTU Aqua, Technical University of Denmark, Charlottenlund, Denmark Department of Biology, University of Lund, SE-22362 Lund, Sweden

a r t i c l e

i n f o

Article history: Received 4 October 2010 Received in revised form 20 December 2010 Accepted 22 December 2010 Keywords: Carbon dioxide Environmental hypercapnia Growth Recirculating aquaculture system

a b s t r a c t A trial was undertaken to investigate how exposure to graded hypercapnia affected the growth performance of juvenile (15–80 g) Atlantic cod. Juveniles were grown at 20‰ salinity and 10 ◦ C for 55 days under three hypercapnic regimes: low (2 ± 0.9 mg L−1 CO2 , 0.6 mm Hg, 1000 ␮atm), medium (8 ± 0.5 mg L−1 CO2 , 2.8 mm Hg, 3800 ␮atm) and high CO2 exposure (18 ± 0.2 mg L−1 CO2 , 6.3 mm Hg, 8500 ␮atm). All water quality parameters were within the range of what might normally be considered acceptable for good growth, including the CO2 levels tested. Weight gain, growth rate and condition factor were substantially reduced with increasing CO2 dosage. The size-specific growth trajectories of fish reared under the medium and high CO2 treatments were approximately 2.5 and 7.5 times lower (respectively) than that of fish in the low treatment. Size variance and mortality rate was not significantly different amongst treatments, indicating that there was no differential size mortality due the effects of hypercapnia, and the CO2 levels tested were within the adaptive capacity of the fish. In addition, an analysis was carried out of the test CO2 concentrations reported in three other long-term hypercapnia experiments using marine fish species. The test concentrations were recalculated from the reported carbonate chemistry conditions, and indicated that the CO2 concentration effect threshold may have been overestimated in two of these studies. Our study suggests that juvenile Atlantic cod are more susceptible to the chronic effects of environmental hypercapnia than other marine fish examined to date. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The effect of environmental hypercapnia on fish has not been extensively studied in marine species because from a toxicological perspective, the sea generally has a low and unvarying concentration of CO2 . However, the development of land-based recirculating aquaculture systems (RAS) means marine species are now being exposed to water CO2 concentrations 10–40 times those encountered in the sea due to the accumulation of respired CO2 in the system. The accumulation of CO2 in RAS is becoming recognized as a water quality variable that can significantly impact fish health and growth (Fivelstad et al., 2003b; Foss et al., 2003; Steffensen and Lomholt, 1988). Removal of CO2 from saltwater RAS is particularly problematic due to the slow dehydroxylation reaction of the carbonate system and the effect of salt on the ionization fraction of inorganic carbon species (Moran, 2010a,b). Compared to other water quality variables such as O2 and NH3 , relatively little work has been carried out in assessing the effects of elevated CO2 on fish growth (Colt, 2006), in particular for marine species. In addition to the applied interest in this topic, there is also an urgent need to

∗ Corresponding author at: Department of Biology, University of Lund, SE-22362 Lund, Sweden. Tel.: +46 46 2229335; fax: +46 46 2224425. E-mail address: [email protected] (D. Moran). 0166-445X/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.aquatox.2010.12.014

assess and understand the impacts of chronic hypercapnia in fish due to increasing atmospheric concentrations of CO2 (Ishimatsu et al., 2008), though the CO2 concentration ranges need to considered when extrapolating findings between research fields. We assessed the effect of chronic exposure to elevated CO2 in juvenile Atlantic cod (Gadus morhua L.), a marine species with a wide geographical range and one which has considerable ecological and economic value. Carbon dioxide has a variety of effects on fish physiology. Increasing the concentration of dissolved CO2 decreases water pH through the formation of carbonic acid (H2 CO3 ), and it is useful to separate the effects of CO2 versus the effects of water acidification. It has been demonstrated that marine fish can cope with moderate levels of water acidification caused by the addition of mineral acids, but have difficulty coping with water acidified by CO2 (Hayashi et al., 2004; Kikkawa et al., 2004). The flow of H+ across the gills and other membranes can be controlled (Heisler, 1984), however, CO2 is highly soluble and can easily diffuse across the gills and within body tissues (Ishimatsu et al., 2005). The resulting acidification of body tissues through the formation of H2 CO3 can have a variety of impacts, from lowering brain pH (inducing anaesthesia, Yoshikawa et al., 1994), to decreasing blood oxygen carrying capacity (Gallaugher and Farrell, 1998) and cardiac output (Lee et al., 2003). While fish can adjust blood chemistry relatively quickly to cope with acute hypercapnia, chronic hypercapnia is

D. Moran, J.G. Støttrup / Aquatic Toxicology 102 (2011) 24–30

known to reduce growth rates, feed intake and feed conversion efficiencies in Atlantic salmon smolts, Salmo salar L. (Fivelstad et al., 1998; Hosfeld et al., 2008), juvenile spotted wolffish, Anarhichas minor Olafsen (Foss et al., 2003) and European seabass Dicentrarchus labrax L. (Lemarié et al., 2000), although the physiological explanations for these effects are not well understood (Ishimatsu et al., 2008). Carbon dioxide induced nephrocalcinosis is a condition often recorded in freshwater fish species and salmonids (Fivelstad et al., 2003a,b), and the study of Foss et al. (2003) suggests this may also occur in at least one marine species. Compared to salmonids, relatively little is known about how Atlantic cod or marine fish in general cope with hypercapnia, and what constitutes a safe CO2 level to raise fish at. When discussing CO2 as a water quality variable it is common to express dissolved CO2 in terms of mg CO2 L−1 . Some aquaculture textbooks and technical reports suggest that keeping CO2 concentrations below 20 mg L−1 (approximately 10000 ␮atm) should have little impact on growth provided adequate oxygenation is available (Losordo et al., 1998; Timmons et al., 2002; Wedemeyer, 1996), although it is recognized that there has been relatively little research into this subject. The Norwegian Ministry of Fisheries and Coastal Affairs recommends a maximum CO2 concentration of 15 mg L−1 (approximately 8600 ␮atm) for land-based aquaculture systems in order to maintain water quality for fish welfare (NMFCA, 2004). Foss et al. (2006) found no consistent difference in growth rate amongst Atlantic cod juveniles raised under different degrees of water reuse which also had differing CO2 concentrations (varying between 2–12 mg L−1 , equivalent to 980–5800 ␮atm), however, there were differences in blood osmolyte concentration indicating physiological compensation for hypercapnia. The study of Foss et al. (2006) was not designed to directly test the effect of CO2 on growth performance as many water quality variables changed with the degree of water reuse, and the CO2 concentrations varied throughout the trial. For these reasons no direct conclusions could be drawn about the effect of CO2 on growth in Atlantic cod. The present study took the commonly employed approach of using graded hypercapnia to assess the effect of CO2 on growth, condition factor and mortality.

2. Materials and methods 2.1. Trial set-up Ethics approval for the study was given by the Technical University of Denmark and the Danish Ministry of Food, Agriculture and Fisheries. The experiment was conducted at Asnæs Fiskeopdræt, Kalundborg, Denmark. At this facility there were six RAS units constructed by Billund Aquakulturservice ApS (Billund, Denmark) in 2006. One of the juvenile rearing units was used for the present study, which is described in detail by Fülberth et al. (2009). The juvenile RAS units consisted of nine 1 m3 cubic-shaped rearing tanks and a water treatment unit, with a total circulating water volume of approximately 21 m3 . Mean water retention time in the rearing tanks was 22 min. Automatic disc feeders provided feed in each tank. Lighting to the tanks was supplied by fluorescent tubes fixed approximately 1 m above the rearing tanks, giving 13 lux light intensity at the water surface. The photoperiod consisted of 6 h dark, 6 h sunrise (light intensity gradually increased), 6 h light, 6 h sunset (light intensity gradually decreased). Food was administered every 30 min during the 10 h period that spanned the last 2 h of the sunrise and first 2 h of the sunset. The feed type used was 1.5, 2 and 3 mm DAN-EX 1582 extruded pellets from Dana Feed A/S (Horsens, Denmark). The diet and photoperiod were selected because this combination gave the best growth rates in a study by Fülberth et al. (2009). The macronutrient profile of the diet was as follows: crude protein 58% mass mass−1 , crude lipid 15%, crude ash 10.8%, crude

25

fibre 1.03%. The daily feed ration was weighed and placed into the automatic feeders every morning. The mass of food administered was based on the estimated individual weight, and was adjusted according to visual observations of the number of uneaten pellets from previous days. The amount of food administered decreased from 2.6% of body weight at the beginning of the trial to approximately 1.3% at the end. The tanks were covered with nets to avoid losses due to individuals jumping out. Each tank was supplied with gaseous oxygen which was monitored and dosed via the RAS computer management system to maintain oxygen levels of 90–100% air saturation. The temperature and water level in each tank was also monitored by the same system. After mechanical filtration, the water flowed into another tank and was cooled by a cooling coil to 10 ◦ C ± 0.6. This temperature was selected on the basis that it was near the optimal growth temperature of juvenile Atlantic cod reported by Björnsson et al. (2007). Sodium hydroxide (NaOH) was added by dosing pump to maintain pH 7.8, and an in-line cascade column degassed most CO2 before the water flowed back to the rearing tanks. Water was exchanged at a rate of 5–10% of the total RAS volume day−1 . The salinity of the incoming water to the RAS facility varied tidally from 15 to 25‰, but the salinity of the water in the experiment was maintained at 20‰ ± 1 via the daily addition of sodium chloride or dilution with fresh water. Ammonia, nitrite, nitrate and phosphorus were measured routinely (1–4 days) using commercially available aquaria test kits. 2.2. Test subjects, sampling and hypercapnic conditions Juvenile Atlantic cod (approximately 1200 individuals 15 g in mean weight) were purchased from Fosen Aquacentre A/S (Stadsbygd, Norway) and transported to the experimental facility on the 5th of December 2007. The broodstock used to produce the juveniles were wild caught fish originating from the Trondheim fjord area in Norway. The fish were allowed to recover for one week before being collected together again for allocation to the trial tanks. Size variation between individuals was high, so the entire group was graded into three size classes and approximately equal numbers of each grade distributed between the 3 replicate tanks of each treatment to achieve a similar biomass density (1.60–1.76 kg m−3 ) and number of individuals (87–109) per tank. After measurement and allocation to the trial tanks, the fish were grown for 37 days before CO2 dosing commenced. Fish were starved one day prior to weight and length measurements being conducted. Fish from each tank were corralled together and 50 individuals randomly selected for weight and length measurements (to nearest lower 0.5 g and 0.5 cm), while the remaining fish were counted. The dosing of CO2 commenced the day after measurement. The concentration of CO2 was manipulated via the direct addition of gaseous CO2 through diffusers into each tank. Gas flow was controlled via 0–1 L min−1 flow meters, and the CO2 concentration of each tank monitored and recorded daily using OxyGuard CO2 Analyzers (OxyGuard International, Birkerød, Denmark). The precision of the OxyGuard CO2 Analyzers is reported by Moran et al. (2010), as is the advantage over commonly employed pH proxy methods of measuring dissolved CO2 . The target treatment concentrations were designated as low (1–2 mg L−1 ), medium (8 mg L−1 ) and high (18 mg L−1 ) concentrations. In this study both the CO2 concentration and PCO2 are presented simultaneously, the latter calculated from the OxyGuard CO2 Analyzer output concentrations using the appropriate solubility constant for 20‰ salinity and 10 ◦ C (Weiss, 1974). The CO2 concentrations were increased from the baseline concentration (2 mg L−1 ) at a rate of 2–3 mg L−1 per day until the target concentrations had been reached. The pH in each tank was also measured daily using a pH meter (type WP-81, TPS Australia Pty Ltd, Queensland, Australia). The CO2 concentration in the highest treatment had a mean (±SD) daily

26

D. Moran, J.G. Støttrup / Aquatic Toxicology 102 (2011) 24–30

concentration of 17.8 mg L−1 ± 0.5 (6.3 mm Hg ± 0.2, approximately 8500 ␮atm), and a mean pH of 7.06 ± 0.02. The medium treatment had a mean daily concentration of 7.9 mg L−1 CO2 ± 0.2 (2.8 mm Hg ± 0.1, approximately 3800 ␮atm) and mean pH of 7.40 ± 0.04. The low treatment experienced a gradual increase in CO2 concentration from 1 mg L−1 to 4 mg L−1 over the first 5 days of dosing as the baseline CO2 concentration of the RAS system increased. Some effort was made to increase the degassing efficiency of the cascade column. However, it was evident that the baseline CO2 concentration of the water could not be maintained at the targeted low concentration. The fish were therefore transferred to a separate but identical RAS unit after 26 days of the trial, where no gaseous CO2 was added to the system. For the remainder of the trial the CO2 concentration of this unit remained at a steady 1 mg L−1 and pH was maintained at 7.8. The mean (±SD) CO2 concentration the low treatment was exposed to during the entire dosing period was 1.8 mg L−1 CO2 ± 0.9 (0.6 mm Hg ± 0.3, approximately 1000 ␮atm), and the mean pH was 7.8 ± 0.0. The CO2 concentrations experienced by fish in the low treatment were not analogous to control conditions in that control concentrations would represent atmospheric CO2 concentrations (equivalent to circa 0.7 mg L−1 CO2 ). However, the range and mean CO2 concentrations the low treatment fish were exposed to would be considered a low CO2 exposure in light of the concentrations used in similar studies (Fivelstad et al., 1998; Foss et al., 2003; Lemarié et al., 2000). Fish length and weight were re-measured 26 days after CO2 dosing started, and again at day 55, when the trial ended. Any mortalities that occurred were collected daily and noted. In some tanks the total number of fish surveyed was less than would be expected from the daily mortality count. It was assumed that the missing fish had been cannibalized or consumed after death. The mortality rate was calculated from the number of fish remaining in each tank at the end of the hypercapnic period. 2.3. Data handling and statistics The median was used as an average measure of weight in each replicate tank as there was considerable variation in size, and some large individuals had a disproportionate effect on the mean weight. The coefficient of variation (CVW % = SD/mean × 100) was used as a relative measure of the variance in weight for each sampling period. Specific growth rate (SGR, % weight gain day−1 ) was calculated from median individual weight using the formula SGR = (expg − 1) × 100, where g = (lnWT1 − lnWT0 )/T1 − T0 . The length (cm) and weight (g) measurements were used to calculate condition factor (K = 102 × W/L3 ). Mortality (% mortality) was calculated from the time the CO2 dosing started. For each of the aforementioned attributes (weight, SGR, K, mortality) a grand mean ± SD was calculated for the 3 replicate trial tanks per treatment. The influence of CO2 treatment on the aforementioned attributes amongst treatments was compared using analysis of variance (ANOVA), and differences between treatment means detected using Tukey’s post hoc test (˛ < 0.05). The mortality data was arcsine transformed before ANOVA as it was proportional data. When plotting SGR over time, the middle time point between sampling days was used to represent the period for which growth was measured. In addition to plotting the change in SGR over time, growth was also plotted as a function of individual weight at each sampling period, the rationale being that SGR changes with size in juvenile Atlantic cod (Björnsson et al., 2007; Imsland et al., 2005). When weight was used as a dependent variable, the geometric mean weight was calculated from the grand mean weights for each treatment at the beginning and end of each measurement period. Linear regressions were fitted to the size-specific growth data to characterize the growth trajectories of each experimental treatment.

2.4. Reanalysis of test CO2 concentrations in other studies A reanalysis was carried out of the test CO2 concentrations reported in the three studies we are aware of that measured the chronic effects of elevated environmental CO2 on the growth of marine fish species. We did this because we were unsure how the authors had arrived at the test concentrations from the carbonate chemistry parameters reported in these studies. The three studies were that of Fivelstad et al. (Fivelstad et al., 1998) on Atlantic salmon smolts, Foss et al. (2003) on juvenile spotted wolffish, and Lemarié et al. (2000) on European seabass. That last study has only been published as a conference abstract, but had been cited regularly enough in the peer reviewed literature (e.g. Foss et al., 2003; Lemarié et al., 2004; Siikavuopio et al., 2007) to warrant investigation. Where the experimental set-up or carbonate chemistry measurements were not explicitly stated in the papers we endeavoured to contact the authors to clarify. Dissolved CO2 concentration can be difficult to measure directly, so most aquatic hypercapnia studies measure two parts of the carbonate system (e.g. pH and alkalinity or total inorganic carbon and pH) and use equilibrium chemistry to infer CO2 concentration (in contrast, the CO2 concentration in the current study was measured directly via infra-red absorption). Water pH was a common measurement parameter in all studies and was assumed to be accurate. The alkalinity of seawater is relatively constant (Almgren et al., 1983), and for the purpose of having a known carbonate parameter to check the other parameters, alkalinity was assumed to be 2.3 mM. In the three studies surveyed, a flow-though system was used to raise fish, where seawater is pumped ashore and passes through the experimental system only once. No additional alkalinity agents were added to the water, so the assumed alkalinity was probably close to the actual value. Our reanalysis of CO2 concentration used temperature- and salinity-specific solubility and dissociation constants, and included the effect of boron. The references for these are given in Table 1. The relevant equations used for carrying out the carbonate chemistry calculations are given in brief in Table 1, or for a more detailed overview of how to execute such calculations readers should refer to the thorough summary given by Dickson (2010).

3. Results The maximum recorded concentrations of biofiltrationassociated chemicals for both recirculation units were as follows: total ammonia nitrogen <0.5 mg L−1 , nitrite 0.8 mg L−1 (as NO2 − ), nitrate 25 mg L−1 (as NO3 − ), phosphate <5 mg L−1 . Elevated CO2 levels led to a decreased median body weight. By the end of the 55 day dosing period, fish in the hypercapnic treatments showed a significant dose-dependent decrease in median body weight, such that individual weights in the medium and high treatments were reduced by approximately 25% and 30% (respectively) compared to fish from the low treatment (Fig. 1a). The high values for CVW reflected the large size variation of fish used in the trial (Fig. 1b). The CVW was not found to vary appreciably amongst treatments during the CO2 dosing period, and remained relatively constant throughout most of the trial (Fig. 1b). The growth rate of fish from the high CO2 treatment was statistically and substantially lower than the growth rates of the medium and low CO2 treatments during the hypercapnic period (Fig. 2a). Although the SGR of the medium treatment was not significantly different to that of the low treatment for any sampling time, there was a trend towards a dose-response effect of CO2 over time (Fig. 2a). When growth rate was plotted as a function of fish weight to account for size-specific growth, the dose-dependent effects of hypercapnia were more evident (Fig. 2b). Linear regressions were fitted to the data in Fig. 2b to approximate the growth trajectories

2.3 (Measured) 7.88 7.00 6.62 6.37 −20.03 −20.92 34‰ Atlantic salmon postsmolt Fivelstad et al. (1998)

15.5 ◦ C

−31.32

−3.27

−13.66

2.37

2.3 (Assumed) 7.9 7.3 6.9 6.5 6.2 −19.81 3.22 −20.59 −13.50 −3.47 −30.66 37‰ 22 ◦ C European seabass Lemarié et al. (2000)

AlkT = total alkalinity. An approximate general alkalinity for seawater is 2.3 mM (Almgren et al., 1983). All three studies utilized a flow-through experimental system with seawater pumped ashore, and alkalinity agents were not added. CT = total inorganic carbon. This was measured by Foss et al. (2003) using a CO2 selective ion probe, however, it appeared to be over-reading. Salinity and temperature density–volume conversion function taken from Martin and Bernwell (1993, pg 11.3). KSW = Ion product of seawater. Salinity and temperature-specific equation for KSW from Millero (1995, eq. 63). K0 = solubility constant for CO2 . Salinity and temperature-specific equation for K0 from Weiss (1974). K1 and K2 = carbonic acid dissociation constants. Salinity and temperature-specific equations for seawater (5–45‰) from Roy et al. (1993). BT = total boron concentration. Estimated from salinity (BT = 0.000416[S/35]), from Millero (1982). KB = dissociation constant for boric acid. Salinity and temperature-specific equation for KB from Dickson (1990). Relevant equilibrium equations: [B(OH)4 − ] = [BT ]KB /([H+ ] + KB ) (Millero, 1995, eq 53); [Carbonate alkalinity, AlkCarb ] = [AlkT ] − [B(OH)4 − ] − [OH− ] + [H+ ]; [CT ] = ([AlkCarb ] − [OH− ] + [H+ ])/(˛1 + 2˛2 ); [CO2 ] = [CT ]˛0 ; PCO2 = [CO2 ]/K0 . Ionization fractions ˛0 , ˛1 and ˛2 calculated according to Stumm and Morgan (1996, pg 15).

1 (0.5) 9 (3.9) 21 (9.4) 37 (16.8) 1.3 11 26 44

0.8 (0.4) 4 (2.0) 9 (5.1) 23 (12.8) 46 (25.7) 0.8 9 18 35 75

0.7 (0.3) 11 (3.8) 21 (7.0) 39 (12.8) 1.1 18 34 59 2.94 3.43 3.76 4.28 2.3 (Assumed) 8.1 6.98 6.71 6.45 −20.33 1.73 −21.34 33‰ Spotted wolffish Foss et al. (2003)

6 ◦C

−32.30

−2.95

−13.90

Reported pH ln KB B(OH)4 − (␮M) ln K2 ln K1 ln K0 ln KSW Salinity Temp. Species Study

Table 1 Recalculation of CO2 treatment concentrations in previously published studies of chronic hypercapnia and marine fish.

AlkT (mM)

CT (mM)

Reported CO2 mg L−1

Recalculated CO2 mg L−1 (and mm Hg)

D. Moran, J.G. Støttrup / Aquatic Toxicology 102 (2011) 24–30

27

Fig. 1. The effect of CO2 on (a) median fish weight and (b) coefficient of variation of weight (CVW ) during the course of the trial. The symbol legend refers to CO2 treatment concentration. The shaded area represents the period of CO2 dosing. Data points with a symbol are significantly different from data points that do not share the same symbol within the sampling period (Tukey’s post hoc test, p < 0.05). Each data point represents the mean (±SD) value from 3 replicate tanks, with 50 individuals measured per tank. Data points have been offset from each other along the x-axis to display variation.

of each experimental treatment. The linear regressions statistics were as follows: low treatment y = 2.121 − 0.006x, R2 = 0.987, p = 0.02; medium treatment y = 2.346 − 0.015x, R2 = 0.918, p = 0.11; high treatment y = 3.541 − 0.046x, R2 = 0.977, p = 0.11. The correlation coefficients and significance statistics indicated that the linear regressions were reasonable approximations of the growth trends. The degree to which hypercapnia affected growth trajectories was assessed via comparison of the regression slopes of Fig. 2b. The sizespecific growth trajectories of fish reared under the medium and high CO2 treatments were approximately 2.5 and 7.5 times lower (respectively) than that of fish in the low treatment. The condition factor of fish was similar between replicate tanks prior to CO2 dosing, however, under hypercapnia there was a dosedependent decrease in condition factor (Fig. 3). Individuals from the low CO2 treatment had the best condition throughout the hypercapnic period, whereas individuals from the high CO2 treatment had significantly poorer condition (Fig. 3). It was incidentally observed that there was a relationship between the colouration of fish and CO2 concentration. Within four weeks of the CO2 dosing commencing virtually all of the fish from the high treatment were dark grey or black. The majority of fish from the medium CO2 treatment were considerably darkened compared to the fish from the low CO2 treatment, which remained the same light grey colour throughout the trial.

28

D. Moran, J.G. Støttrup / Aquatic Toxicology 102 (2011) 24–30

Fig. 2. The effect of CO2 on (a) specific growth rate (SGR) and (b) SGR versus geometric mean individual weight. The dashed vertical lines in (a) represent the sampling times of individual weight, and the SGR for the sampling period is plotted at the middle time point. Data points in (a) with a symbol are significantly different from data points that do not share the same symbol within the sampling period (Tukey’s post hoc test, p < 0.05). Data points in (a) have been offset from each other along the x-axis to display variation. Each data point represents the mean value from 3 replicate tanks (±SD for (a)). See main body text for a description of the linear regressions in (b).

Fig. 3. The effect of CO2 on condition factor (K). Data points with a symbol are significantly different from data points that do not share the same symbol within the sampling period (Tukey’s post hoc test, p < 0.05). Each data point represents the mean (±SD) value from 3 replicate tanks, with 50 individuals measured per tank. Data points have been offset from each along the x-axis other to display variation.

The mortality rate for the test concentrations ranged between 1% and 4% during the hypercapnic treatment period, and there was no significant difference in mortality rate amongst treatments (F2,6 = 1.12, p = 0.39). Most of the mortalities that occurred could be accounted for, and only a small number of fish were assumed to have been consumed via cannibalism or necrophagy. There was little direct aggression observed (e.g. fin nipping or chasing). Reanalysis of the carbonate chemistry in the three similar studies suggests that the CO2 concentrations were over-estimated by a significant margin in two of the studies (Table 1). In the Foss et al. (2003) study, the problem with the reported test concentrations appears to be attributable to erroneous total inorganic carbon (CT ) measurements. At atmospheric CO2 concentrations (i.e. control conditions for hypercapnia studies), seawater should have a CT value slightly higher than the alkalinity, but the CT values reported by Foss et al. (2003) are much higher than the alkalinity for control conditions. This suggests the selective CO2 ion probe used by the authors was malfunctioning. We assumed that the water pH data given by Foss et al. (2003) was reliable given that the calculated dissolved CO2 concentrations under control conditions were the same as that expected for water in equilibrium with atmospheric gas concentrations. The water pH values given by Lemarié et al. (2000) at the different CO2 test conditions were assumed to be reliable, however, it appeared the authors incorrectly calculated the test CO2 concentrations. For example, the authors reported that at a water pH of 6.9 the CO2 concentration was 18 mg L−1 . If one calculates what the water pH should be at a CO2 concentration of 18 mg L−1 (and the water conditions given in Table 1) the pH is closer to 6.6. Unfortunately Lemarié et al. (2003) do not give any details of what dissociation constants were used in their calculations so it is impossible to know where their calculation errors lie. The CO2 values we calculated for the study of Fivelstad et al. (1998) were close to the reported values, and the minor discrepancy is likely due to the use of average pH and water temperature data versus the daily calculation of CO2 by the authors.

4. Discussion The need to transfer the low treatment group to a different system during the CO2 dosing period was an unforeseen complication that arose during the trial, and it is necessary to consider whether this would have impacted on results. The two RAS units used were identical in construction, meaning that there would have been no difference in lighting, feeding method, filtration, rearing tank hydrodynamics, and tank stocking density. In both RAS units, the concentrations of biofiltration-related biochemicals were very low, and reflected the low total system stocking density (2–4 kg m−3 water). The biochemical concentrations were well below that which is generally reported as harmful to fish (Wedemeyer, 1996). The amount of handling stress imposed on the fish during the transfer process was lower than that experienced during the reweighing procedures. For these reasons it is unlikely that the transfer of the low CO2 treatment fish to another recirculation unit had a significant impact on their growth performance. The growth rate, condition factor and mortality data of fish from the low CO2 treatment group indicate at the experimental conditions should were adequate for achieving optimal or near-optimal growth. The growth rate of juveniles in the low CO2 treatment were higher than the growth model predictions of Björnsson et al. (2007) and the experimental data of Foss et al. (2006) for juvenile Atlantic cod reared in RAS, but similar to that recorded in an earlier study on the effects of diet and photoperiod on juvenile Atlantic cod growth conducted at the same research facility (Fülberth et al., 2009). The condition factor of juveniles in the current study were higher than that of juveniles in the Foss et al. (2006) study and comparable to

D. Moran, J.G. Støttrup / Aquatic Toxicology 102 (2011) 24–30

that recorded by Fülberth et al. (2009). The mortality rates of the current study were low and similar to the levels recorded under comparative experimental conditions (Foss et al., 2006; Fülberth et al., 2009). Exposure to chronic hypercapnia had a clear impact on the median weight, condition factor and growth rate of the fish. The reduction in growth and body weight of juvenile Atlantic cod was particularly pronounced under the high CO2 treatment (18 mg L−1 , 6.3 mm Hg). The growth rate of the medium CO2 treatment was not statistically different from the low treatment at the individual sampling times. However, when growth rate was plotted against the median weight to account for the size-dependency of SGR, it was evident that CO2 did indeed affect growth rates at 8 mg L−1 CO2 (2.8 mm Hg). The 2.5 and 7.5 fold reduction in the size-specific growth trajectories exhibited by fish reared under the medium and high treatments, respectively, was striking. The growth rate of juvenile Atlantic cod is known to decrease rapidly over the size range we tested (15–80 g, Björnsson et al., 2007), and it is unclear whether the effects of hypercapnia would be as clearly delineated at larger sizes, where the growth rate is lower. The value of CVW was not found to vary significantly amongst treatments, therefore, growth and mortality rates were assumed to be relatively even across all size grades of fish. While the levels of hypercapnia tested certainly affected growth and fish condition, there was no difference in mortality rate amongst treatments, and the mortality rate during the hypercapnic period was low. The PCO2 levels tested appeared to be within the range that Atlantic cod can physiologically compensate for. Foss et al. (2006) reported that juveniles exposed to varying rearing densities and CO2 concentrations (2–12 mg L−1 ) exhibited physiological compensation for the associated elevated blood PCO2 levels, so clearly this species has the ability to cope with moderate hypercapnia. Other long-term studies of hypercapnia have also found no difference in fish mortality rates over a range of CO2 concentrations (Danley et al., 2005; Foss et al., 2003; Lemarié et al., 2000). The mechanisms by which chronic hypercapnia reduces growth rates are not well understood, however, a review by Ishimatsu et al. (2008) reported that the metabolic rate of fish in a high CO2 environment would likely be higher due to elevated costs of acid–base regulation and increased swimming speeds or ventilation for gill irrigation. Melzner et al. (2009) investigated the effect of long term exposure to elevated CO2 on Atlantic cod, and reported that standard and active metabolic rates, critical swimming speeds and aerobic scope did not differ between individuals exposed to 0.3 and 0.6 kPa CO2 (approximately 1.3 and 14 mg L−1 , or 540 and 5800 ␮atm, respectively). These findings indicate the fish maintained locomotory and respiratory performance despite a significant and sustained level of environmental hypercapnia, therefore, the Atlantic cod in the current study may not have required increased swimming speed for gill irrigation as the oxygen delivery status of the blood may not have been perturbed. At the higher CO2 level Melzner et al. (2009) found elevated Na+ /K+ -ATPase protein expression and enzyme activity, indicative of adaptation to increased acid–base load. This acid–base maintenance mechanism would likely add to the cost of homeostasis in Atlantic cod and reduce the scope for growth, as alluded to by Ishimatsu et al. (2008). Though unclear why, the feed intake and feed conversion efficiencies of fish are also known to be affected by exposure to chronic hypercapnia (Fivelstad et al., 1998; Foss et al., 2003; Hosfeld et al., 2008; Lemarié et al., 2000), which would also reduce growth rate. To our knowledge, only three other studies have investigated the chronic effects of hypercapnia on marine fish species in a dose dependent manner (Fivelstad et al., 1998; Foss et al., 2003; Lemarié et al., 2000). The concentrations of CO2 found to effect juvenile Atlantic cod in the current study were considerably lower than the effect concentrations reported in the aforementioned publications. However, reanalysis of the carbonate chemistry in the three

29

comparative studies suggests that the CO2 concentrations were over-estimated by a significant margin in two of the studies Future studies of hypercapnia need to ensure accurate measurement of the test CO2 concentrations, and the solubility and dissociation constants used to calculate inorganic species should be reported along with the carbonate chemistry parameters measured. Researchers also need to be aware of the relative merits and errors associated with different CO2 measurement techniques. We suggest future studies adopt some of the recommendations made by Riebesell et al. (2010) in a report entitled “Guide to best practices for ocean acidification research and data reporting”. While the concentrations of CO2 tested in almost all fish hypercapnia studies are much higher than those encountered in ocean acidification research, both fields suffer from the same problem of difficulties in accurately dosing and measuring dissolved CO2 . Using the revised CO2 dosing levels of comparable studies given in Table 1, the growth of juvenile spotted wolffish was lower at 39 mg L−1 (13 mm Hg) compared to 21 mg L−1 (7 mm Hg) and below (Foss et al., 2003). For European seabass, growth was considerably lower at 46 mg L−1 (26 mm Hg) compared to 4 mg L−1 (2 mm Hg) (Lemarié et al., 2000). Fivelstad et al. (1998) concluded that the no observed effect concentration (NOEC) of CO2 for Atlantic salmon postsmolts was in the range of 8–12 mg L−1 (3.5–5.4 mm Hg), but negative effects were observed at CO2 concentrations of 21 mg L−1 (9 mm Hg) and above. Although re-analysis of the carbonate chemistry in the comparative studies lowers the apparent effect threshold for CO2 exposure in marine fish species, the results from the current study still indicate that juvenile Atlantic cod are more susceptible to the chronic effects of hypercapnia than other marine fish species tested so far. Probably one of the best studied fish species with respect to hypercapnia is Atlantic salmon smolts in freshwater hatcheries. Parr have been found to have reduced growth at 45 ppm (12 mm Hg) (Fivelstad et al., 2007). In one study smolts were reported not to be affected by CO2 in the range of 6–24 mg L−1 (2–7 mm Hg) (Fivelstad et al., 2003a), but in other studies were found to be affected at 17–18 mg L−1 (6 mm Hg) (Hosfeld et al., 2008; Waagbø et al., 2008). The simultaneous changes in growth and condition factor recorded in the current study indicates that juvenile Atlantic cod are more affected by CO2 than juvenile Atlantic salmon. The growth of juvenile Atlantic cod was significantly reduced at 18 mg L−1 (6.3 mm Hg), and size-specific analysis of growth rate showed lowered growth at 8 mg L−1 (2.3 mm Hg). The dose-dependent reduction in condition factor supports a CO2 concentration effect at 8 mg L−1 (2.3 mm Hg). As fish aquaculture moves towards intensification coupled with a decrease in water exchange rates and higher reliance on water reuse, CO2 is increasingly becoming an issue in terms of fish health and growth and will likely become an important fish welfare issue. In light of the findings of this study and another recent study that found that concentrations of only 18 mg L−1 CO2 (6 mm Hg) affected Atlantic salmon smolt growth (Hosfeld et al., 2008), the widely held view in the aquaculture industry that 20 mg L−1 CO2 is adequate for fish production needs to revisited, or at least qualified. There are probably species-specific thresholds for differing levels of effect, just as there is for hypoxia tolerance (Nilsson and ÖstlundNilsson, 2008). As Hosfeld et al. (2008) and Fivelstad et al. (2007) have demonstrated, there are also likely to be interactive effects between hypercapnia and other water quality variables, which will have to be taken into consideration. The development of a better understanding of the effects of hypercapnia is not only going to improve fish health and productivity. Removing CO2 from water is an energy intense process which generally involves pumping water against gravity (Summerfelt et al., 2000; Timmons et al., 2002), and is particularly difficult in salt or saline waters (Moran, 2010a). Further research into the effects of chronic hypercapnia will help RAS designers scale degassing systems appropriately to meet specific

30

D. Moran, J.G. Støttrup / Aquatic Toxicology 102 (2011) 24–30

CO2 operating targets, thereby improving the economics of landbased fish farming. The other impetus for long term studies into environmental hypercapnia in fish stems from the interest in ocean acidification (Ishimatsu et al., 2008), though the CO2 concentration ranges are an order of magnitude different and the results need to be interpreted accordingly. Acknowledgements We are thankful for the technical and practical support given by M. Fülberth, logistical support given by H. Jarlbæk, B.H Olsen, J. Bregnballe, and T. Samuelsen. We also thank the anonymous reviewers who helped improve this paper. This work was funded by the EU Fisheries Sector Program FIUF and the Danish Ministry of Food, Agriculture and Fisheries. DM was supported by a fellowship from the New Zealand Foundation for Research Science and Technology. References Almgren, T., Dyrssen, D., Fonselius, S., 1983. Determination of alkalinity and total carbonate. In: Grasshoff, K., Ehrhardt, M., Kremling, K., 1 (Eds.), Methods of Seawater Analysis. Verlag Chemie. Weinheim, Germany, pp. 99–123. Björnsson, B., Steinarsson, A., Árnason, T., 2007. Growth model for Atlantic cod (Gadus morhua): effects of temperature and body weight on growth rate. Aquaculture 271, 216–226. Colt, J., 2006. Water quality requirements for reuse systems. Aquacultural Engineering 34, 143–156. Danley, M.L., Kenney, B.P., Mazik, P.M., Kiser, R., Hankins, J.A., 2005. Effects of carbon dioxide exposure on intensively cultured rainbow trout Oncorhynchus mykiss: physiological responses and fillet attributes. Journal of the World Aquaculture Society 36, 249–261. Dickson, A.G., 1990. Thermodynamics of the dissociation of boric acid in synthetic seawater from 273.15 to 318.15 K. Deep Sea Research A 37, 755–766. Dickson, A.G., 2010. Part 1: Seawater carbonate chemistry. In: Riebesell, U., Hansson, L., Fabry, V.J., Gattuso, J.-P. (Eds.), Guide to Best Practices in Ocean Acidification Research and Data Reporting. European Project on Ocean Acidification, Luxemborg, pp. 17–40. Fivelstad, S., Haavik, H., Løvik, G., Olsen, A.B., 1998. Sublethal effects and safe levels of carbon dioxide in seawater for Atlantic salmon postsmolts (Salmo salar L.): ion regulation and growth. Aquaculture 160, 305–316. Fivelstad, S., Olsen, A.B., Åsgård, T., Baeverfjord, G., Rasmussen, T., Vindheim, T., Stefansson, S., 2003a. Long-term sublethal effects of carbon dioxide on Atlantic salmon smolts (Salmo salar L.): ion regulation, haematology, element composition, nephrocalcinosis and growth parameters. Aquaculture 215, 301–319. Fivelstad, S., Waagbø, R., Zeitz, S.F., Hosfeld, A.C.D., Olsen, A.B., Stefansson, S., 2003b. A major water quality problem in smolt farms: combined effects of carbon dioxide, reduced pH and aluminium on Atlantic salmon (Salmo salar L.) smolts: Physiology and growth. Aquaculture 215, 339–357. Fivelstad, S., Waagbø, R., Stefansson, S., Olsen, A.B., 2007. Impacts of elevated water carbon dioxide partial pressure at two temperatures on Atlantic salmon (Salmo salar L.) parr growth and haematology. Aquaculture 269, 241–249. Foss, A., Røsnes, B.A., Øiestad, V., 2003. Graded environmental hypercapnia in juvenile spotted wolffish (Anarhichas minor Olafsen): effects on growth, food conversion efficiency and nephrocalcinosis. Aquaculture 220, 607–617. Foss, A., Kristensen, T., Åtland, Å., Hustveit, H., Hovland, H., Øfsti, A., Imsland, A.K., 2006. Effects of water reuse and stocking density on water quality, blood physiology and growth rate of juvenile cod (Gadus morhua). Aquaculture 256, 255–263. Fülberth, M., Moran, D., Jarlbæk, H., Støttrup, J.G., 2009. Growth of juvenile Atlantic cod Gadus morhua in land-based recirculation systems. Effects of feeding regime, photoperiod and diet. Aquaculture 292, 225–231. Gallaugher, P., Farrell, A.P., 1998. Hematocrit and blood oxygen-carrying capacity. In: Perry, S.F., Tufts, B. (Eds.), Fish Respiration. Academic Press, San Diego, pp. 185–227. Hayashi, M., Kita, J., Ishimatsu, A., 2004. Comparison of the acid–base responses to CO2 and acidification in Japanese flounder (Paralichthys olivaceus). Marine Pollution Bulletin 49, 1062–1065. Heisler, N., 1984. Acid–base regulation in fishes. In: Hoar, W.S., Randall, D.J. (Eds.), Fish Physiology. Academic Press, New York, U.S.A, pp. 315–401. Hosfeld, C.D., Engevik, A., Mollan, T., Lunde, T.M., Waagbø, R., Olsen, A.B., Breck, O., Stefansson, S., Fivelstad, S., 2008. Long-term separate and combined effects of environmental hypercapnia and hyperoxia in Atlantic salmon (Salmo salar L.) smolts. Aquaculture 280, 146–153.

Imsland, A., Foss, A., Folkvord, A., Stefansson, S., Jonassen, T., 2005. The interrelation between temperature regimes and fish size in juvenile Atlantic cod (Gadus morhua): effects on growth and feed conversion efficiency. Fish Physiology and Biochemistry 31, 347–361. Ishimatsu, A., Hayashi, M., Lee, K.-S., Kikkawa, T., Kita, J., 2005. Physiological effects on fishes in a high-CO2 world. Journal of Geophysical Research 110, C09S09. Ishimatsu, A., Masahiro, H., Kikkawa, T., 2008. Fishes in high-CO2 , acidified oceans. Marine Ecology Progress Series 373, 295–302. Kikkawa, T., Kita, J., Ishimatsu, A., 2004. Comparison of the lethal effect of CO2 and acidification on red sea bream (Pagrus major) during the early developmental stages. Marine Pollution Bulletin 48, 108–110. Lee, K.-S., Kita, J., Ishimatsu, A., 2003. Effects of lethal levels of environmental hypercapnia on cardiovascular and blood-gas status in yellowtail Seriola quinqueradiata. Zoological Science 20, 417–422. Lemarié, G., Dutto, G., Le Roux, A., Lemoalle, J., Maxime, V., Person-Le Ruyet, J., 2000. Long-term effects of pH and carbon dioxide on growth and feed efficiency in European seabass. European Aquaculture Society Special Publication 28, 384. Lemarié, G., Dosdat, A., Covés, D., Dutto, G., Gasset, E., Person-Le Ruyet, J., 2004. Effect of chronic ammonia exposure on growth of European seabass (Dicentrarchus labrax) juveniles. Aquaculture 229, 479–491. Losordo, T.M., Masser, M.P., Rakocy, J., 1998. Recirculating Aquaculture Tank Production Systems: An Overview Of Critical Considerations. SRAC Publication. Southern Regional Aquaculture Centre, Stoneville, Mississippi U.S.A. Martin, J.L., Barnwell, T.O.J., 1993. Water Quality. McGraw-Hill, New York, U.S.A. Melzner, F., Göbel, S., Langenbuch, M., Gutowska, M.A., Pörtner, H.-O., Lucassen, M., 2009. Swimming performance in Atlantic Cod (Gadus morhua) following longterm (4–12 months) acclimation to elevated seawater PCO2 . Aquatic Toxicology 92, 30–37. Millero, F., 1982. The thermodynamics of seawater at one atmosphere. Ocean Science and Engineering 7, 403–460. Millero, F.J., 1995. Thermodynamics of the carbon dioxide system in the oceans. Geochimica et Cosmochimica Acta 59, 661–677. Moran, D., 2010a. Carbon dioxide degassing in fresh and saline water I: degassing performance of a cascade column. Aquacultural Engineering 43, 29–36. Moran, D., 2010b. Carbon dioxide degassing in fresh and saline water II: degassing performance of an air-lift. Aquacultural Engineering 43, 120–127. Moran, D., Tirsgård, B., Steffensen, J.F., 2010. The accuracy and limitations of a new meter used to measure aqueous carbon dioxide concentration. Aquacultural Engineering 43, 101–107. Nilsson, G.E., Östlund-Nilsson, S., 2008. Does size matter for hypoxia tolerance in fish? Biological Review 83, 173–189. Norwegian Ministry of Fisheries and Coastal Affairs, 2004. Notes to Regulation 22 December 2004 No. 1785 on the operation of the aquaculture facility (Aquaculture Operations Regulations). Norwegian Ministry of Fisheries and Coastal Affairs. Riebesell, U., Fabry, V.J., Hansson, L., Gattuso, J.-P., 2010. Guide to Best Practices For Ocean Acidification Research And Data Reporting. Publications Office of the European Union, Luxembourg. Roy, R.N., Roy, L.N., Vogel, K.M., Porter-Moore, C., Pearson, T., Good, C.E., Millero, F.J., Campbell, D.M., 1993. The dissociation constants of carbonic acid in seawater at salinities 5–45 and temperatures 0–45 ◦ C. Marine Chemistry 44, 249–267. Siikavuopio, S.I., Mortensen, A., Dale, T., Foss, A., 2007. Effects of carbon dioxide exposure on feed intake and gonad growth in green sea urchin, Strongylocentrotus droebachiensis. Aquaculture 266, 97–101. Steffensen, J.F., Lomholt, J.P., 1988. Accumulation of carbon dioxide in fish farms with recirculating water. In: Rayans, R.C. (Ed.), Fish Physiology, Fish Toxicology and Fisheries Management. U.S. Environmental Protection Agency, Athens U.S.A, pp. 157–161, Environ. Res. Lab. Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry: Chemical Equilibria and Rates in Natural Waters, 3rd ed. Wiley-Interscience, New York, U.S.A. Summerfelt, S.T., Vinci, B.J., Piedrahita, R.H., 2000. Oxygenation and carbon dioxide control in water reuse systems. Aquacultural Engineering 22, 87–108. Timmons, M.B., Ebeling, J.M., Wheaton, F.W., Summerfelt, S.T., Vinci, B.J., 2002. Recirculating Aquaculture Systems, 2nd ed. Cayuga Aquaculture Ventures, Ithaca, New York, U.S.A. Waagbø, R., Hosfeld, C.D., Fivelstad, S., Olsvik, P.A., Breck, O., 2008. The impact of different water gas levels on cataract formation, muscle and lens free amino acids, and lens antioxidant enzymes and heat shock protein mRNA abundance in smolting Atlantic salmon, Salmo salar L. Comparative Biochemistry and Physiology A 149, 396–404. Wedemeyer, G.A., 1996. Physiology of fish in intensive culture systems. Chapman & Hall, New York. Weiss, R.F., 1974. Carbon dioxide in water and seawater: the solubility of a non-ideal gas. Marine Chemistry 2, 203–215. Yoshikawa, H., Kawai, F., Kanamori, M., 1994. The relationship between the EEG and brain pH in carp, Cyprinus carpio, subjected to environmental hypercapnia at an anesthetic level. Comparative Biochemistry and Physiology A 107, 307–312.