Effects of Simulated Ocean Acidification on the Activity, Escape Response, and Muscle Physiology of Marine Threespine Stickleback (Gasterosteus aculeatus)

ABSTRACT

Rapidly increasing anthropogenic CO₂ can impose physiological challenges for fish species that are thought to be tolerant. We tested the hypothesis that elevated pCO₂ will affect the routine activity and escape response by affecting energy metabolism and/or the muscle physiology of coastal fish. We exposed threespine stickleback (Gasterosteus aculeatus) to pCO₂ of ~ 700 µatm (pH 7.9 representing current levels), ~ 1400 µatm (pH 7.6 representing upwelling events) and ~ 3500 µatm (pH 7.3 representing a future predicted scenario for coastal areas) for 2 weeks. Baseline activity was significantly higher in fish exposed to 1400 µatm compared to the control at both sampling points, while the escape response was lower (p < 0.05). Metabolic rate was not different (p > 0.05), but lactate dehydrogenase activity was significantly higher at 3500 µatm compared to control fish after the first week (p < 0.05), while no difference was found in muscle histology between treatments or time points. Our study demonstrates that the baseline activity and escape responses of adult marine coastal fish were temporarily affected by the current level of ocean acidification, but this was not due to changes in metabolism or muscle function, but potentially neuronal effects of high pCO₂. Our study shows that ocean acidification might affect predator‐prey interactions during current upwelling events and in the future.

Summary

Baseline activity and escape response of marine threespine stickleback were temporarily affected by current and future levels of ocean acidification, which has important implications for understanding predator‐prey interactions of coastal fish.

1. Introduction

Coastal areas are often characterized by a great variation in the partial pressure carbon dioxide (pCO₂) due to oceanic circulation and geographical traits. For example, the pCO₂ of the Pacific Northwest of North America regularly fluctuates between 400 and 1,300 µatm (Bednaršek et al. 2020). It has been predicted that atmospheric CO₂ will increase from 420 to 1135 ppm by 2100 under the business‐as‐usual model (SSP5‐8.5) due to anthropogenic greenhouse gas emissions (Meinshausen et al. 2020). As a quarter of atmospheric CO₂ is absorbed by the ocean, anthropogenic CO₂ will cause ocean acidification, resulting in a similar increase in pCO₂ in the ocean by the end of the century. It was believed that impacts of this magnitude on coastal aquatic organisms might be negligible because the upper level of daily fluctuations in these areas already exceed the projected levels for the open ocean by the end of century (Baumann 2019). However, the last 20 years of research have shown that many aquatic organisms are still vulnerable to the ongoing climate change because they may already be living near their physiological limits in terms of pCO₂ (Hickey and Banas 2003; Feely et al. 2010; Haigh et al. 2015; Schunter et al. 2019; Bednaršek et al. 2021; Huang et al. 2021; Messié et al. 2023).

The behavioral effects of increased pCO₂ have been documented from a variety of species (Ferrari et al. 2011; Munday et al. 2013; Munday et al. 2019), and in particular baseline activity and escape responses have received a great deal of attention due to their ecological importance (Allan et al. 2013; Nasuchon et al. 2016; Domenici et al. 2019; Schunter et al. 2021). In the last decade, there has been a growing body of literature to explain the mechanisms behind these behavioral changes (Domenici et al. 2019; Hamilton et al. 2023; Nagelkerken et al. 2023) such as reduced aerobic metabolism (Pimentel et al. 2014), altered neurological function (Nilsson et al. 2012; Schunter et al. 2021) and impaired olfactory responses (Porteus et al. 2018). These mechanistic approaches have been useful to understand potential physiological underpinnings, the heterogeneity of results in literature have made it difficult to come up with a widely accepted hypothesis (Cattano et al. 2018; Lefevre 2019). Although this might be related to methodological differences in CO₂ exposure (Riebesell et al. 2011; Clark et al. 2020; Munday et al. 2020), this would have likely resulted from the variability of abiotic and biotic factors such as life history, habitat type (e.g. open ocean or coastal areas) and geographical distribution (e.g. tropical and temperate species) (Vargas et al. 2017, 2022). Therefore, it is important to investigate the relationship between the physiological effects of OA and behavioral variation within individual species rather than generalizing the assumption that metabolic effects of OA will be similar across species (Domenici et al. 2019; Munday et al. 2019; Killen et al. 2021b).

In response to elevated CO₂, cardio‐ventilatory responses or respiratory plasticity may be sufficient to ameliorate physicochemical disruptions in blood chemistry (Perry and Gilmour 2006; Esbaugh et al. 2016). Nonetheless, it has been long recognized that OA can produce an allostatic load on metabolism because high CO₂ can mount metabolic costs for homeostasis while limiting the maximum aerobic metabolism (Pörtner and Farrell 2008; Brauner et al. 2019). In addition, metabolic acclimation and compensation in response to high CO₂ can include changes in gill morphology (Perry and Abdallah 2012), muscle histology (Rossi et al. 2018), glycolytic/oxidative phosphorylation enzymes (Michaelidis et al. 2007) and neurophysiology (Nilsson et al. 2012), all of which could significantly influence locomotory performance that determines survival, migration, mating and reproductive success (Lauder 2015; Domenici and Seebacher 2020).

In the present study, we challenged the view that coastal fish are resilient to high CO₂ and tested the hypothesis that future levels of higher CO₂ will constrain energy metabolism and locomotion of coastal fish (Baumann 2019). We chose to expose adult marine threespine stickleback (Gasterosteus aculeatus) to relevant levels of PCO₂ for 2 weeks to mimic seasonal upwelling events (García‐Reyes et al. 2014). We conducted our experiment in late‐July, which is a peak time for upwelling and also a critical time period for spawning for this species. pCO₂ levels were 700 µatm (pH 7.9; the current level at the Pacific Northwest), 1400 µatm (pH 7.6; found during upwelling events) and 3500 µatm (pH 7.3; a predicted future scenario) (Baumann 2019; Bednaršek et al. 2020). Although 3500 µatm has not been predicted from our study site, this level allows us to investigate the physiological mechanism by which coastal aquatic species could adapt in the future as ocean acidification will intensify and last longer during the future upwelling events (Takeshita et al. 2015; Baumann 2019; Bednaršek et al. 2020).

We evaluated the baseline activity, interindividual distance as a proxy for group sociality, and escape response, all of which are key traits in survival and reproduction in adult fish. Then, we measured whole‐body routine metabolic rate and maximum metabolic rate to determine if changes in locomotory traits might be due to changes in metabolism. We also assessed lactate dehydrogenase (LDH; EC 1.1.1.27), which is a glycolytic enzyme that interconverts pyruvate and NADH to lactate and NAD⁺ as well as citrate synthase (CS; EC 2.3.3.1), which is a mitochondrial enzyme that catalyzes the first step of TCA cycle to synthesize citrate from acetyl‐CoA and oxaloacetate. LDH and CS enzyme activities are widely measured to evaluate anaerobic and aerobic capacity, respectively. Additionally, we also measured muscle fiber size and succinate dehydrogenase staining as a measure of muscular metabolism. These measures of muscle metabolism and histology were used to determine if there was a link between these data and changes in baseline activity and escape response (Borowiec et al. 2015; Rossi et al. 2018). We predicted that increasing CO₂ would reduce baseline activity and escape response, and these changes would correlate with a reduced metabolic scope (i.e. difference between maximum and routine metabolic rate) and the aerobic capacity of muscle in a dose–response manner.

2. Materials and Methods

2.1. Animal Husbandry

Marine sticklebacks were caught by minnow traps at the head of Bamfield inlet, Alberni‐Clayoquot, British Columbia, Canada in June 2022 (48°48'55“N, 125°09'23“W). Average length and body mass were 61 ± 3 mm and 1.9 ± 0.3 g (mean ± S.D.), respectively. Fish were immediately transferred to Bamfield Marine Science Centre and housed in one of three 114 L aquaria in groups of 50 fish per tank on a natural photo period. Each aquarium consisted of Top Fin® Sponge Filter, moderate aeration, artificial plants and substrate. The substrate was comprised of ~ 5 mm diameter pebbles reflecting the natural habitat of marine stickleback. Owing to the logistical and spatial constraints, we could not have replicate tanks for each experiment group. While we were fully aware of this limitation, we ensured consistency in environmental factors such as flow rate, aeration, and light exposure to minimize potential tank effects. Importantly, our study in the following year used similar levels of pCO₂ and revealed similar results (data not yet published). Fish were initially fed to satiation once a day with Hikari Bloodworm (Hikari, California, USA) before the exposure experiment and three times a day until the termination of experiment. Our research was approved under the Bamfield Marine Science Centre Animal Utilization Protocol number RS2206 under the requirements established by the Canadian Council for Animal Care.

2.2. CO₂ Exposure

The control level (pH 7.9 ± 0.1) was achieved by bubbling ambient air that removed excess CO₂ using soda lime (Sigma Aldrich, MO, USA). 1400 μatm was achieved by bubbling CO₂ with a pH solenoid system (Pinpoint pH controller, American Marine Inc, CT, USA) set to pH at 7.6 ± 0.1. 3500 μatm was achieved similarly with the pH set at 7.3 ± 0.1. Water chemistry was monitored daily to ensure consistent pH, temperature, salinity and pCO₂. pH and temperature were measured using a handheld pH meter (HI1230B; Hanna Instruments, RI, USA). The pH solenoid system and handheld pH meter were calibrated daily with fresh National Bureau of Standards (NBS) pH buffers (Fisher Chemical, NH, USA) to ensure consistent readings between pH meters. Salinity was measured using a handheld salinometer (HI98319; Hanna Instruments, RI, USA). pCO₂ was measured with a portable CO₂ analyzer (S157‐PCO₂; QUBIT Systems, Kingston, ON, Canada). Unfortunately, our CO₂ analyzer stopped working on the 5th day of exposure, so we could not measure pCO₂ directly afterwards and relied on pH measurements to calculate pCO₂. Importantly, the pH solenoid system and handheld pH meter were calibrated daily as described above while water was sampled three times a week with 4% mercuric chloride in gas tight glass containers stored at 4°C for later analysis of total CO₂ levels. Total CO₂ in these water samples were measured using an Apollo Dissolved Inorganic Analyzer (LI‐COR Inc, Lincoln, NE, USA; Model number LI‐5350). Our poisoned water samples have confirmed the validity of our initial measurements taken with the CO₂ analyzer and that the remainder of the experiment our pCO₂ values remained stable at the target level for each treatment. Carbonate chemistry data are included in Table 1.

Table 1.

Water chemistry parameters for control and elevated CO2 treatments in exposure tanks. pH_NBS, temperature and salinity data are direct measurements. pCO₂ and total alkalinity (TA) were calculated from these parameters in CO2SYS (Pierrot et al. 2021). Data are presented as mean ± SD.

Parameter	700 µatm	1400 µatm	3500 µatm
pHNBS	7.93 ± 0.08	7.61 ± 0.09	7.30 ± 0.10
Temperature (°C)	12.35 ± 0.67	12.23 ± 0.64	12.30 ± 0.60
Salinity (ppt)	31.93 ± 0.63	31.72 ± 0.65	31.70 ± 0.50
TA (µmol/kgSW)	2242.06 ± 183.56	1962.57 ± 283.77	2071.5 ± 95.10
pCO2 (µatm)	669.12 ± 186.00	1432.79 ± 192.74	3534.04 ± 577.53

2.3. Sampling

We chose to stagger exposure treatments every 2 days to facilitate the sampling process, and we sampled fish two times: after 1 and 2 weeks of exposure (i.e. 7 to 8th and 14 to 15th days of exposure) over 2 days. On the first day of sampling, we assessed baseline activity, escape response, and subsequently measured whole‐body metabolic rate of eight fish (for 24 h). On the second day of sampling, we assessed baseline activity and escape response of seven additional fish. On the second day of sampling (on the 8th and 15th days of exposure), 15 fish were killed on each day with an overdose of MS‐222 (Syndel Laboratory, BC, Canada) at the corresponding pH buffered with NaOH. Total length and body mass were measured to the nearest 1 mm and 0.1 g, respectively. White muscle tissue was removed and quickly frozen in liquid nitrogen for later enzyme assay. Also, a 2 mm transverse muscle steak immediately posterior to anus was removed and coated with embedding medium (Tissue‐Plus™ O.C.T. Compound; Fisher Scientific, NH, USA) on a cork board. Then, muscle steak samples were quickly frozen in liquid nitrogen chilled isopentane. All muscle samples were stored at ‐80°C until the analysis was performed. It is worth noting that there was only one mortality in our study at 3,500μatm between week 1 and 2 of exposure.

2.4. Baseline Activity

Baseline activity (n = 14–15) was assessed in a translucent rectangular tank (W 324 mm × L 257 mm × H 108 mm; Nalgene, NY, USA) filled with 4000 mL water from their exposure tank. Then, fish were captured with a cup to avoid air exposure and to minimize handling stress, and placed into the behavioral arena. A black curtain was hung around the tank to remove any visual distractions. Based on our preliminary video analysis, we chose to acclimate fish for 25 min to the behavior arena and measure baseline activity in a group of 3 or 4 fish for 5 min to account for their high sociality. However, we recognized that having different number of individuals in the arena could cause changes in social dynamics; thus, we included the number of individuals in the tank as a random factor in our statistics (see below). The behavioral tank had moderate aeration for the first 15 min to keep pH and dissolved oxygen levels to the same as the CO₂ treatment, but then the aeration was stopped to avoid interference with the tracking of fish during video analysis. Then, baseline activity was recorded for 5 min at 10 frames per second in the resolution of 1920 × 1080 pixels with a camera (U3‐3680XLE; IDS, Obersulm, Germany) mounted with a Tamron 8 mm F1.4 lens (M118FM08; Tamron, Saitama, Japan). After the baseline activity was measured, fish were allowed to recover until the escape response test in a separate part of the exposure tank without air exposure to minimize the handling stress (Shishis et al. 2023). The time to recover was at least 30 min, and our statistical analysis showed that there was no significant effect of recovery time on escape responses in our study (p > 0.1). Videos were analyzed to track fish motion in two dimensions on LoliTrack 5 (Loligo Systems, Viborg, Denmark). Three points were automatically generated on each fish to mark the tip of nose, center of mass, and tip of the tail. We used the tip of the nose of each fish to analyze distance moved (pixels), baseline average speed (pixels · s⁻¹) as well as average interindividual distance (pixels).

2.5. Escape Response

The assessment of escape response was carried out as described previously with some modifications (Allan et al. 2014; Roche et al. 2023). The escape response (n = 14–15; 60.7 ± 3.4 mm, mean ± s.d.) was measured in a white rectangular plastic tank (W 190 mm × L 280 mm × H 105 mm) filled with 2500 mL water from the respective exposure tank. For the calibration of kinematic analysis, a linear scale of 20 mm was drawn on the bottom of the arena. A PVC pipe (OD 32 mm × H 60 mm) was perpendicularly held on the bottom of the tank in the corner where a metal tube was dropped to the bottom to induce a mechano‐acoustic stimulus without a visual stimulus. Preliminary trials ensured that temperature did not change more than 0.5°C throughout the experiment and that dissolved oxygen stayed over 90% while pH did not change during the experiment.

For the escape response test, individual fish were fasted for 12 h before the measurement. Then, fish were captured from the holding tank as described above, avoiding air exposure. A black curtain was hung around the entire setup to avoid any visual disturbance to the fish. Each fish was allowed to acclimate for 15 min. We were aware of the ideal requirements for escape response of fish (Roche et al. 2023), but due to the logistical constraints at the remote field station, we elicited escape response only when fish had sufficient space to successfully perform escape response toward open space. Based on our preliminary trials, it was either when the fish was positioned in the middle of the arena or at least 5 cm away from the wall and with its head toward the center of the tank. This resulted in reliable measurements of escape responses. Our preliminary trials also showed that those fish that did not respond to the second stimulus within an hour of the first attempt would not respond to any subsequent stimuli. Therefore, if fish did not respond to the first stimulus, the fish was retested within an hour of the initial attempt. When fish did not respond to the second stimulus, we assumed that fish would not respond and we excluded it from analysis.

An escape response was induced by a mechano‐acoustic stimulus by dropping a metal tube inside the PVC pipe. The escape response was recorded with a high‐speed camera (Edgertronic SC1; Campbell, CA, USA) positioned perpendicularly above the arena. The camera was mounted with 24–85 mm F2.8‐4 Nikon zoom lens at 1000 frames per second while two LED lights were placed above to homogeneously luminate the entire experimental arena (AOS 35 W LED Light; Dättwil, Switzerland). After each trial, water was changed to remove any metabolites or cues from the previous trial that could influence the future trial.

The video was trimmed for capturing stage 1 and 2 of escape responses and its contrast was adjusted on a free video editing program ShotCut (https://shotcut.org). Videos were then analyzed using LoliTrack 5 (Loligo Systems, Viborg, Denmark). We conducted a preliminary statistical analysis to examine random effects of waiting time and retesting for escape response results within each trial, and they were found to be not statistically significant. Thus, we removed these variables from the model but included the time to the stimuli as a random effect so that we could address the assumption of independence in our linear regression (see below for more details). All behavioral data were presented in pixels as previously recommended (Roche et al. 2023), but data expressed as body length are also available as supplementary figures (Figure S1, S2). Detailed information on the escape response experiment protocol is provided in Supporting Table S1.

2.6. Metabolic Rate

Whole‐body metabolic rate ($\dot{M}{O}_2$; n = 6–8) was measured by an intermittent flow respirometry (Loligo Systems, Viborg, Denmark) following previous guidelines (Chabot et al. 2016; Svendsen et al. 2016; Killen et al. 2021a). The respirometry system consisted of borosilicate glass chambers with oxygen sensor spots (PreSens, Regensburg, Germany) in a 7 L water bath. Temperature and dissolved oxygen were measured at 1 Hz by a temperature probe and fiber optic cables attached to a Witrox 4 Oxygen Meter. Tubing was nongas permeable with the chamber volume of 42.32 ± 1.06 mL and tubing volume of 3.07 ± 0.11 mL; mean ± S.D. The average ratio between the net respirometer volume (mL) and body mass of fish (g) was 24.44 ± 4.56 (mean ± S.D.). Oxygen probes were calibrated with 2% sodium sulfite and fully saturated water for 0% and 100% dissolved oxygen, respectively. Preliminary trials ensured that flow rate and time cycle were programmed to replenish oxygen to saturation during each flush cycle and that oxygen level stayed above 80% during each measurement cycle. A black curtain was hung around the respirometry system to avoid physical disturbance.

Fish were fasted overnight before the measurement of oxygen consumption. At each sampling date, fish (1.86 ± 0.30 g; mean ± S.D.) was haphazardly captured from the holding tank, and a standardized chase protocol was performed by gentle prodding to the tail for 5 min and the fish was placed into the metabolic chambers within 30 s (Zhang et al. 2019). It is worth to note that all fish became exhausted and lethargic to the stimulus after the chase protocol. MMR was measured for two measurement cycles. The first cycle consisted of 60 s waiting followed by 120 s measurement, whereas the second cycle consisted of 360 s flushing, 60 s waiting and 120 s measurement. After 6 h of acclimation, RMR was measured for 15 h with the following measurement cycle: 360 s flushing, 60 s waiting and 300 s measurement while ensuring a linear decrease in oxygen levels over each measurement. Before and after each trial, oxygen consumption was measured for 15 min without fish to estimate background respiration, which was linearly interpolated to correct all $\dot{M}{O}_2$ data points. The average level of background respiration as a percentage of $\dot{M}{O}_2$ was 5 ± 5% (mean ± S.D.). After each trial, the respirometry system was cleaned and rinsed using clean seawater without any chemicals.

Only $\dot{M}{O}_2$ slopes with R ≥ 0.9 were used for data analysis. MMR was the highest $\dot{M}{O}_2$ during the first two measurements, whereas RMR was the average of $\dot{M}{O}_2$ data. Metabolic scope was the difference between MMR and RMR, which is analogous to absolute aerobic scope. Following the measurement of metabolic rate, all fish were killed and then sampled for muscle enzyme assay and histology as described below. We assumed that fish would have fully recovered from the handling stress before sampling for muscle enzyme because our preliminary trials demonstrated a substantial decrease of metabolic rate during the first 6 h of acclimation period and the stable readings afterwards. Detailed information on the respirometry experiment protocol is provided in Supporting Table S2.

2.7. Muscle Enzyme Assay

Muscle enzyme assay (n = 6–7) was performed as previously described (Dalziel and Schulte 2012; Mandic et al. 2013) with some modifications. In brief, muscle tissue was weighed and added to 9 times of ice‐cold homogenization buffer (50 mM HEPES, 1 mM EDTA, 0.1% Triton X‐100, pH 7.4). Tissue was then homogenized with a 7 mL glass dounce homogenizer (Kimble dounce tissue grinder; Sigma‐Aldrich, MO, USA) kept on ice. Then, the homogenate was centrifuged at 16,000 g with 4°C for 10 min, and only the supernatant was used for assays. Final concentrations for each enzyme assay were as follows: LDH (50 mM Tris pH 7.4, 5 mM NADH, 25 mM Pyruvate) and CS (50 mM Tris pH 8, 0.15 mM Acetyl CoA, 0.15 mM DTNB, 0.1% v/v TX‐100, 1 mM Oxaloacetate). Our conditions ensured that the substrate was not limited but fully saturated for measurement of enzyme activities on a 96‐well plate with a microplate reader (BioTek Synergy HT, Agilent Technologies, CA, USA). We measured CS on fresh homogenates within 1 h of homogenization, but LDH was measured on homogenates within 1 h of thawing after being frozen once at −80°C. We chose to measure enzyme activities at 25°C owing to the unstable air temperatures in the lab. We ensured that our approach resulted in maximal enzyme activities based on our preliminary trials. All samples were measured in duplicate with one left blank for measuring background reaction without substrate. Maximal enzyme activity was calculated by linear regression using five data points on the software (Gen 5, Agilent Technologies, CA, USA). Protein content was measured by Bradford Reagent (Bio‐Rad, CA, USA) according to the manufacturer's instruction, which was used to standardize enzyme activity.

2.8. Muscle Histology

Each frozen muscle steak was transversely sectioned at 9 µm using a cryostat (Leica CM3050 S, Leica, Wetzlar, Germany) at −20°C. Sections were mounted on Superfrost Plus slides (Fisher Scientific, MA, USA) and stored at −80°C until staining was performed. For succinate dehydrogenase (SDH) staining each slide was covered with 200 µL of incubation medium (50 mM Phosphate buffer solution pH 7.8, 50 mM Sodium succinate, 0.05% Nitroblue Tetrazolium) for 45 min. The reaction was stopped by dipping slides in tap water and slides were air‐dried in dimmed light at room temperature. After this, slides were mounted with distilled water and cover glass, and images were taken immediately using a compound microscope (Motic X5; Motic, TX, USA) at 40 times magnification. Images were randomly chosen to minimize observation bias and analyzed on ImageJ (http://imagej.nih.gov/ij/). Two slices per individual fish were used for analysis. SDH staining intensity was measured by superimposing a square of 1000 pixels over and below the horizontal septum of oxidative muscle and the integrated density within the square was calculated by ImageJ. SDH staining intensity was measured six times for each slice. The size of red and white muscle fibers was calculated by randomly choosing 10 and 20 fibers, respectively, and hand‐drawing each fiber per slide using freehand selection tool in ImageJ. Muscle fiber size data was normalized with body mass of fish to account for individual size variation (Rossi et al. 2018).

2.9. Statistical Analysis

We used a linear mixed model analysis to address the nested structure of behavior data in which experiments were carried out on more than one occasion for each treatment, which violates the assumption of independence for parametric test (Kuznetsova et al. 2017). pCO₂ levels and exposure duration were considered as discrete factor variables with levels, and they are represented by C (700, 1400 and 3500 μatm) and E (1 week to 2 week exposure), respectively. Length (L) was included as a covariate to account for the body size that could influence the response variable (Roche et al. 2023). ${\alpha }_{\mathit{Trial}}$ represents the intercept of one of two random effects: it is either group size for baseline activity and average interindividual distance experiments or time to stimulation to address the effects of habituation in the behavioral arena for escape response experiment. ε indicates residuals errors. The full model was written as follows:

$$\hat{R}={\alpha }_0+{\beta }_c\mathrm{·}C+{\beta }_e\mathrm{·}E+{\beta }_{c*e}\mathrm{·}C\times E+{\beta }_l\mathrm{·}L+{\alpha }_{\mathit{Trial}}+\epsilon$$

where the response variable ($\hat{R}$), represents either baseline total distance moved (pixels), baseline average speed (pixels· s⁻¹), average interindividual distance (pixels), escape total distance moved (pixels; log transformation), escape average speed (pixels · s⁻¹), escape maximum speed (pixels· s^‐1; log transformation). The result of linear mixed model was summarized by lmerTest (Kuznetsova et al. 2017). When results indicated significance with p ≦ 0.05, post hoc comparisons were performed using emmeans function in the R package emmeans while accounting for the random effect of trials (Hothorn et al. 2008) with significance determined at alpha = 0.05 (see Supporting Tables S3 and S4). The other physiological data were analyzed with two‐way ANOVA with post hoc analysis of Tukey test. Assumptions of normality and homoskedasticity were visually assessed as previously outlined (Zuur et al. 2010). Where these assumptions were not met, data transformation was performed as follows. Log transformation was performed on muscle fiber size data. Tukey's power transformation was done on the CS activity data. Also, rank transformation was performed on MMR, metabolic scope and LDH activity data. We reported estimated marginal mean with standard error using emmeans function from the R package emmeans. All statistical analyses were conducted in R 4.3.3. (R Core Team 2025).

3. Results

3.1. Baseline Activity

Both pCO₂ and exposure time affected the total distance moved, average speed and average interindividual distance (p < 0.05; Supporting Table 3). However, length affected both total distance moved and average speed (p = 0.030; Supporting Table 3), but not average interindividual distance (p = 0.303; Supporting Table 3). Fish exposed to 1400 μatm showed an increased total distance moved and average speed (56686 ± 3360 pixels, 189 ± 11.2 pixels s⁻¹; mean ± S.E.) compared to the control at both sampling points (41765 ± 3370 pixels, 139 ± 11.2 pixels s⁻¹) (p < 0.01; post hoc test), but these were not different from fish exposed to 3500 μatm at both sampling points (p = 0.234, p = 0.5057; post hoc test, Figure 1a,b). In addition, fish exposed to 1400 μatm showed a greater average interindividual distance at the first week of exposure compared to the control (610 ± 16.5 vs 542 ± 16.5 pixels; p = 0.0506, post hoc test). However, there was no difference in the second week of exposure (p > 0.1; post hoc test).

Figure 1.

Total distance moved (a), baseline average speed (b) and average interindividual distance (IID) of baseline activity of marine sticklebacks (Gasterosteus aculeatus) exposed PCO₂ levels of 700 µatm, 1400 µatm and 3,500µatm for one and 2 weeks. Different letters indicate significant difference between groups (p < 0.05). n = 14–15.

3.2. Escape Response

We did not find a large difference in the responsiveness of escape responses between pCO₂ levels after each week, although fish exposed to 700 μatm showed a decrease from the first week to the second. After the first week of exposure, the responsiveness of escape response for 700, 1400 and 3500 μatm were 80% (12/15), 73% (11/15) and 67% (10/15), respectively and after the second week, these were 60% (9/15), 73% (11/15) and 71% (10/14), respectively.

Our initial analysis showed that total distance moved, average speed and maximum speed were significantly affected by pCO₂ level and exposure time (p ≦ 0.05; Supporting Table 3). Also, our analysis showed that length significantly affected distance moved and average speed (p < 0.05; Supporting Table 3), not maximum speed (p = 0.076; Supporting Table 3). Fish exposed to 1400 μatm showed lower average and maximum escape speeds (log transformed; 6.87 ± 0.111 pixels s⁻¹, 8.31 ± 0.244 pixels s^‐1) than those of the control only after the first week (log transformed; 7.46 ± 0.112 pixels s⁻¹, 9.46 ± 0.244 pixels s⁻¹, p ≦ 0.05, post hoc test, Figure 2b,c) while fish exposed to 3500 μatm were intermediated in their response. However, there was no difference between treatments on average and maximum escape speeds after the second week (p = 0.7189; p = 0.9446; post hoc test, Figure 2b,c).

Figure 2.

Total escape distanced moved (a), average escape speed (b) and maximum escape speed (c) during escape responses of marine sticklebacks (Gasterosteus aculeatus) exposed PCO₂ levels of 700 µatm, 1,400µatm and 3,500µatm for one and two weeks. Different letters indicate significant difference between groups (p < 0.05). n = 9–12.

3.3. Metabolic Rate

Routine metabolic rate, maximum metabolic rate and metabolic scope did not significantly differ between groups and time points (p > 0.05; Supporting Table 4; Figure 3a,b,c).

Figure 3.

Routine metabolic rate (a), maximum metabolic rate (b) and metabolic scope (c; b‐a) of marine sticklebacks (Gasterosteus aculeatus) exposed to CO₂ levels of 700 µatm, 1,400µatm and 3,500µatm for one and two weeks. n = 7–8.

3.4. Muscle Enzyme Activity

CO₂ level only affected LDH activity (p = 0.05; Supporting Table 4). Specifically, fish exposed to 3500 μatm had a significantly higher LDH activity (rank transformed; 35.1 ± 4.64 µmol⁻¹ mg of protein ⁻¹ min⁻¹) than those to 700 μatm after the first week (rank transformed; 15.6 ± 4.38 µmol⁻¹ mg of protein ⁻¹ min⁻¹, p = 0.041, post hoc test, Figure 4a), but not the second week (p = 1.000, post hoc test, Figure 4a). Citrate synthase activity and LDH/CS were not significantly different between groups at either sampling point (p > 0.1; Supporting Table 4; Figure 4b,c).

Figure 4.

Enzyme activity of lactate dehydrogenase (LDH; a), citrate synthase (CS; b) and LDH/CS (c) of marine sticklebacks (Gasterosteus aculeatus) exposed to PCO₂ levels of 700 µatm, 1400 µatm and 3500 µatm for one and two weeks. Different letters indicate significant difference between groups (p = 0.05). n = 7–8.

3.5. Muscle Histology

SDH staining intensity, red muscle fiber size and white muscle fiber size were not significantly influenced by PCO₂ and duration of exposure (p > 0.05; Supporting Table 3; Figure 5 a,b,c).

Figure 5.

Succinate dehydrogenase staining intensity in red muscle fiber (a), red muscle fiber size (b) and white muscle fiber size (c) of marine sticklebacks (Gasterosteus aculeatus) exposed to PCO₂ levels of 700 µatm, 1,400µatm and 3,500µatm for one and two weeks. n = 6–7.

3.6. Total Length and Body Mass

The total length and body mass were not significantly different between groups at the initial sampling time point and they did not change during the exposure experiment (p > 0.05; Supporting Table 4, Figure 6a,b).

Figure 6.

Total length (a) and body mass (b) of marine sticklebacks (Gasterosteus aculeatus) exposed to PCO₂ levels of 700 µatm, 1,400 µatm and 3,500 µatm for two weeks. n = 10–15.

4. Discussion

Our research demonstrates that elevated CO₂ could impose a sustained effect on the baseline activity, but a short‐term and transient effect on escape response and group cohesion of adult marine threespine sticklebacks. This could mean that the behavior of marine threespine sticklebacks may be transiently but significantly impacted during current upwelling events; however, this result implies that in the long‐term they could be tolerant to future ocean acidification.

We documented that threespine sticklebacks increased baseline activity at 1,400 µatm, which was not in agreement with physiological data such as metabolic rate, enzyme activity and muscle histology. Therefore, it is possible that pCO₂ level of 1400 µatm may not be severe enough to induce physiological perturbation (Hamilton et al. 2017), but could cause behavioral effects in threespine sticklebacks as previously documented (Allan et al. 2013; Nasuchon et al. 2016). It has been proposed that high environmental pCO₂ can cause anxiety‐related behaviors in some fish due to the side effects of metabolic compensation on neuronal cells (Hamilton et al. 2017; Hamilton et al. 2014, 2023; Nilsson et al. 2012). It is well known that under acidosis, fish compensate with increased blood plasma HCO₃ ^‐, which is catalyzed by carbonic anhydrase. Uptake, retention and excretion of plasma HCO₃ ^‐ rely on several ion channels such as anion exchanger and sodium bicarbonate channel that could decrease plasma Cl^‐. This can affect the flux of these two ions at the neuronal cell membrane of ${\mathit{GABA}}_A$ receptors, leading to depolarization (i.e. excitation) rather than hyperpolarization (i.e. inhibition). In elevated pCO₂, it is possible that fish may become bolder due to changes in the neurons that become excitatory, leading to more exploratory behaviors (Nilsson et al. 2012; Munday et al. 2013), which increased the baseline activity of sticklebacks in our study.

It is worth noting that we sourced wild sticklebacks from the West Coast of North America in which they would experience periodic fluctuations in CO₂ that can typically range from 300 to 1,300 µatm due to upwelling events, algal photorespiration and land derivatives (Bednaršek et al. 2020). Therefore, at the level of 1,400 µatm, it is plausible to expect that marine stickleback may show plastic metabolic adjustments with cardio‐ventilatory or respiratory measures (Lefevre 2019). However, our data suggest that when pCO₂ increases to 3,500µatm, which is beyond the upper level of what sticklebacks would normally experience in the wild, aerobic metabolism might be compromised. This is further supported by our observation of the higher LDH activity at 3500 µatm than the control, which might have masked the behavioral effects of higher CO₂ in our study.

It has been recognized that high pCO₂ can not only act as loading stress on homeostasis, but also it can be limiting stress on aerobic metabolism (Pörtner and Farrell 2008; Brauner et al. 2019). Therefore, it has been hypothesized that high pCO₂ should decrease aerobic metabolism; however, the literature clearly shows the mixed results. Senegalese sole (Solea senegalensis) exposed to 1000 µatm for 75 days showed an increase in standard metabolic rate compared to those at 400 µatm (Oliveira et al. 2023), whereas spiny damselfish (Acanthochromis polyacanthus) exposed to 946 µatm for 17 days showed a decrease in RMR compared to those at 451 µatm (Rummer et al. 2013). In terms of CO₂ effects on enzyme activity, juvenile gilthead seabream (Sparus aurata; ~ 50 g) exposed to pCO₂ of around 5,000 µatm (3.82 mmHg) for 11 days showed increased LDH activity with decreased CS activity of white muscle tissue (Michaelidis et al. 2007). Conversely, rainbow trout (Onchorynchus mykiss) fry exposed to 1000 and 2000 ppm for 30 days showed increased pyruvate kinase activity, but no significant difference in LDH activity compared to fish at the control level of 380 ppm (Chen et al. 2020). The mixed results in the literature may reflect different metabolic sensitivity or CO₂ tolerance from the local acidification conditions (Vargas et al. 2022) and/or strategies driving activity changes for foraging or energy conservation in acclimatization to high CO₂ (Lefevre 2016; Cattano et al. 2018; Yoon et al. 2024). However, the inconsistent effects in the literature should be interpreted cautiously, while metabolic effects of OA can vary with exposure duration, magnitude, and life stage (Lefevre 2016; Munday et al. 2019).

Given the significant implications for predator‐prey interactions, a number of studies have examined the effects of ocean acidification on escape responses. Japanese anchovy (Engraulis japonicus) exposed to 1000 ppm for a month did not show any difference in kinematic parameters of escape responses, such as velocity and acceleration, compared to those at 400 ppm (Nasuchon et al. 2016). When marine medaka (Oryzias melastigma) was exposed to 1160 µatm for 8 days reduced the likelihood of performing c‐start by 30% (Wang et al. 2017). As it is widely recognized that escape response would be anaerobically fueled (Domenici et al. 2019), in our study, it is possible that neurological effects of high CO₂ could have masked the results of escape responses (Allan et al. 2013).

In addition to the GABA hypothesis and its implications on fish behavior, it is well known that ocean acidification could reduce group cohesion in some fish species (Lopes et al. 2016; Fowler 2018; Cattano et al. 2019), but see (Kwan et al. 2017). It is well known that ocean acidification could impair olfaction in marine fish because of altered olfactory receptor neurons as well as changes in biochemical conformation of cues (Porteus et al. 2021). Indeed, a previous study demonstrated that at the lower pH of 6.3 ± 0.1, freshwater threespine sticklebacks (Gasterosteus aculeatus) showed a significant reduction in half‐weight index than those at the high pH (8.4 ± 0.1) due to potentially altered olfaction (Kleinhappel et al. 2019). Thus, in our study, it is possible that high CO₂ could have affected conspecific recognition and thus shoal size. It is worth noting that we used marine sticklebacks, and we saw a reduction in group cohesion at pH 7.6 (1400 µatm), but not at 7.3 (3500 µatm). This may indicate that adaptation to the marine and freshwater environments could result in differential behavior or physiological responses to acidification.

Our findings have important implications for understanding and speculating cohort and population dynamics of marine threespine stickleback during coastal upwelling events and future climate change induced increases in pCO₂. It has been widely accepted that coastal fish should be very resilient to future ocean acidification due to environmental variability of pCO₂ in coastal areas (Baumann 2019). Some studies have assessed the behavioral responses and their underpinning mechanisms in fish experiencing natural fluctuations of CO₂ in coastal areas, highlighting altered reproductive behaviors that and could negatively affect offspring fitness (Milazzo et al. 2016; Spatafora et al. 2021). In line with these studies, our results showed that elevated CO₂ could result in increased activity and a reduced escape response and group cohesion of adult threespine stickleback, which can impose significant challenges for surviving a predator attack and individual fitness, although these impacts would not be likely permanent. Previous studies in marine threespine stickleback demonstrated mixed results on growth and reproductive behaviors in response to elevated pCO₂ (Jutfelt et al. 2013; Schade et al. 2014; Sundin et al. 2017; Devergne et al. 2023). This is interesting because these results could be derived from intraspecific variation that would reflect adaptation to environmental conditions (Morris et al. 2014), but it remains unknown how intraspecific variation and plasticity may play a role in acclimation to rapid climate change (Jutfelt et al. 2013). As we conducted our study in the spawning season, our results imply that future upwelling events may influence reproduction of marine threespine sticklebacks in the Pacific Northwest.

5. Conclusion

Our research demonstrated that 1400 µatm pCO₂ can trigger behavioral changes of adult marine sticklebacks. These were not from changes in metabolism or muscle function, but potentially due to altered neuronal function, which could make them vulnerable to predation during the spawning season. Our study suggested that although marine sticklebacks may not be resilient to elevated CO₂ despite the environmental variability of pCO₂, adaptation could still be facilitated by natural CO₂ fluctuations occurring in natural coastal environments.

Author Contribution

Gwangseok R. Yoon: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, validation, writing – original draft preparation and review and editing. Elissa Khodikian: data curation, investigation, methodology, writing – review and editing. Gary R Junlin: data curation, investigation, methodology, writing – review and editing. Cosima S. Porteus: conceptualization, data curation, formal analysis, funding acquisition, investigation, validation, supervision, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: A checklist of criteria for reporting the methods and results of escape response experiments in fish. TABLE S2: Checklist of 53 essential criteria for the reporting of methods for aquatic intermittent‐flow respirometry. Table S3: Summary of linear mixed model for baseline activity, group behaviour, escape response of marine sticklebacks (Gasterosteus aculeatus) exposed to 700 µatm, 1400 µatm and 3500 µatm for two weeks. Summary statistics was generated from lmerTest. ^#denotes that post hoc analysis did not support the result. Table S4: Summary of two‐way ANOVA for metabolic rate, histology, enzyme activity, total length and body mass of marine sticklebacks (Gasterosteus aculeatus) exposed to 700 µatm, 1,400 µatm and 3,500 µatm for two weeks. Summary statistics was generated from lmerTest. Figure S1: Total distance moved (a), baseline average speed (b) and average inter‐individual distance (IID) of baseline activity of marine sticklebacks (Gasterosteus aculeatus) exposed PCO₂ levels of 700 µatm, 1,400 µatm and 3,500 µatm for one and two weeks. Different letters indicate significant difference between groups (p < 0.05). n = 14–15. Figure S2: Total escape distanced moved (a), average escape speed (b) and maximum escape speed (c) during escape responses of marine sticklebacks (Gasterosteus aculeatus) exposed PCO₂ levels of 700 µatm, 1,400 µatm and 3,500 µatm for one and two weeks. Different letters indicate significant difference between groups (p < 0.05). n = 9–12.

Yoon, G. R. , Khodikian E., Ren G. J., and Porteus C.. 2025. “Effects of Simulated Ocean Acidification on the Activity, Escape Response and Muscle Physiology of Marine Threespine Stickleback (Gasterosteus aculeatus).” Journal of Experimental Zoology Part A: Ecological and Integrative Physiology 343: 1219–1230. 10.1002/jez.70034.

Data Availability Statement

All raw data and R scripts are available in a repository and can be accessed via the following link (https://figshare.com/s/0d8335835005a14691d7).