Introduction

Freshwater ecosystems vary widely with respect to dissolved carbon dioxide (CO2) and are often supersaturated (Cole et al. 1994). With continually rising atmospheric CO2 levels and other environmental changes (e.g., higher precipitation rates; Butman and Raymond 2011), some freshwater ecosystems are becoming even more hypercarbic and are acidifying (Weiss et al. 2018); however, which water bodies and the rate of change has not been determined (Hasler et al. 2016). The impacts of high dissolved CO2 and weak acidification on aquatic biota are largely unknown (Hasler et al. 2016, 2018); however, some studies on freshwater fishes and invertebrates have been completed. Broadly, high CO2 is known to have several behavioral and physiological effects on freshwater fishes (e.g., Heuer and Grosell 2014; Tierney 2016), but the exact outcomes vary and likely are dependent on CO2 level and length of exposure (Kates et al. 2012). Lab studies have shown several trends, including loss of predator awareness (Tix et al. 2017), altered behavior (Cupp et al. 2016), and loss of equilibrium (Kates et al. 2012) when exposed to elevated CO2. Due to the suite of physiological and behavioral effects that hypercarbia has on fishes, the use of plumes of intentionally elevated CO2 to prevent the movement of invasive fishes (e.g., bigheaded carp) has been explored (Noatch and Suski 2012; Treanor et al. 2017).

Though there is variation in how fish respond to prolonged exposure to high CO2, avoidance of high CO2 seems to be robust across freshwater fishes (Cupp et al. 2016; Dennis et al. 2016a; Kates et al. 2012) and across magnitude of exposure (Donaldson et al. 2016), likely as fish seek out improved water quality to avoid potential costs from CO2 exposure. For this reason, injection of CO2 into canal locks is being considered as a method to limit the distribution of invasive species in the Mississippi River and elsewhere (Treanor et al. 2017), particularly bigheaded carp (Hypophthalmichthys spp.; Noatch and Suski 2012). Additionally, this potential technology could be used for applications with native fishes as well. Fish possess chemoreceptors localized on the gills that sense increasing levels of ambient CO2 (Perry and Abdallah 2012), and different species (invasive carp and natives) often react similarly to CO2 exposure (Cupp et al. 2016; Kates et al. 2012). Specifically, Kates et al. (2012) showed that silver and bighead carp (H. molitrix and H. nobilis, respectively), largemouth bass (Micropterus salmoides), and bluegill (Lepomis macrochirus) all showed similar responses to elevated CO2 including elevated stress hormone levels, modified ion balance, increased ventilation, and loss of equilibrium all occurring at similar levels across different species. Thus, injecting plumes of CO2 into fish habitat will adversely affect both native and non-native fishes.

Exposure to CO2-rich plumes intended to prevent the movement of invasive fish may have impacts on non-target species, including economically important sport fish. A possible outcome of exposure to high CO2 may be reduced swimming performance due to lack of coordination (Yoshikawa et al. 1994), a central phenotype in determining fish fitness (Plaut 2001). Loss of equilibrium and swimming coordination is thought to be caused by CO2 moving into the brain and affecting energy processing (Yoshikawa et al. 1994). As swimming is either fueled by oxygen (aerobic) or glycolytic products (anaerobic), the effect CO2 has on swimming performance may be context-dependent, though ultimately controlled by the tolerance of the brain to maintain coordination. Furthermore, understanding how fish not only experience altered behavior but also altered swimming performance in hypercarbic environments can have ecological and management implications. However, a thorough investigation of the effects of elevated CO2 levels on various swimming strategies (here, burst and sprint) in freshwater fishes has yet to occur. This information is essential in understanding how fish will interact with an increasingly hypercarbic environment and has high potential to steer future management strategies and develop tools to prevent the spread of invasive fish.

To address these shortcomings, this study aimed to define (1) the interactive effects of pCO2 and exposure duration on burst swimming performance in a freshwater fish and (2) the effects of elevated pCO2 on sprint swimming performance, with an emphasis on barrier development in a flowing water environment. In assessing the impacts that elevated pCO2 may have on both mid-range burst swimming and short-distance sprint swimming performance, an understanding of how exposure to a carbon dioxide barrier may impair the swimming performance of fishes attempting to traverse a CO2 plume can be developed, and management recommendations can be made to maximize the effectiveness of a CO2 barrier (including factors such as the size and velocity a barrier would need to be and target pCO2 required to impair movement through it). Despite both burst and sprint swimming being primarily anaerobically fueled, we predict that the incorporation of some aerobic metabolism in a long (~ 30 min) burst swimming activity may cause burst swimming performance to decrease as pCO2 increases, while sprint swimming will remain unaffected. This may occur due to a combination of the anesthetic properties of CO2, as well as the potential impacts of pCO2 on oxygen uptake and delivery in fish (Portner et al. 2004). However, this is largely unknown as the root drivers (central nervous system or muscles) leading to a breakdown in swimming abilities under hypercarbic conditions are still unclear.

Methods

Fish holding and husbandry

In November 2016, sub-adult largemouth bass (Micropterus salmoides) were purchased from a hatchery (Keystone Hatcheries, Richmond, IL; pond-raised, pellet-fed) and transported to the Aquatic Research Facility at the University of Illinois in Urbana, Illinois, where they were acclimated for 1 week in a 1200-L indoor holding tank. During this time, fish were fed pelleted commercial food (Purina Aquamax, St. Louis, MO) to satiation daily; however, food was withheld for 24 h prior to trials to ensure a post-absorptive state was reached before swimming trials (Roche et al. 2013). Water was sourced from a nearby earthen-bottom pond with natural vegetation, and 10% water changes were performed daily to maintain water quality. Both mechanical (Fluval 406 Canister Filter, Mansfield, MA) and UV filters (Vecton-4: V2 400 15 Watt UV Filter), along with supplemental aeration, were used to maintain optimum water quality. Dissolved oxygen (YSI ProODO, Yellow Springs Instruments, Irvine, CA, USA), temperature (YSI ProODO, Yellow Springs Instruments, Irvine, CA, USA), pH (WTW pH 3310 meter, SenTix probe, Germany), ammonia (LaMotte Company, Ammonia Nitrogen Kit no. 3351-02, Chestertown, MD, USA), total alkalinity (Hach Company, Titrator 16,900, kit 2272700, Loveland, CO, USA), and pCO2 (GMT220 Infrared CO2 meter, Vaisala, Vantaa, Finland) were monitored in the holding tank daily (Table 1). Largemouth bass were selected for this study due to the limitations in collecting and housing appropriately sized Asian carp. Expected similarities in CO2 responses between these fishes are detailed in the discussion.

Table 1 Water quality parameters measured once daily in the 1200-L fish holding tank through the duration of experimentation. All values shown are mean ± 1 standard error
Full size table

Uburst swimming

To quantify the effects of pCO2 exposure on Uburst swimming, a swim tunnel (30 L swim tunnel respirometer, Loligo Systems, Viborg, Denmark) was used to perform a constant acceleration test wherein fish were forced to swim against an elevated current velocity in water treated with CO2 (Reidy et al. 2000). The swim tunnel was calibrated using an in-line flow meter (HFA, Höntzsch GmbH, Waiblingen, Germany), which allowed us to relate motor speed to water velocity. After appropriate lab acclimation and starvation periods of 24 h, a single fish was netted from the holding tank, loaded into the test chamber of the swim tunnel, and acclimated for 10 min at a velocity of 0.5 body lengths per second (BL s−1) (Gregory and Wood 1998) in water at ambient pCO2. Concurrent with this, water in an external holding tank was treated to a target pCO2 using the common method of bubbling compressed CO2 gas through an airstone (Tix et al. 2016). The pCO2 treatments used were < 400 μatm (ambient), 10,000 μatm, 50,000 μatm, and 100,000 μatm. The ambient treatment of 400 μatm falls within normal expected ranges for inland lakes and rivers (Cole et al. 1994). Additionally, similarly elevated pressures can induce physiological and behavioral responses in both largemouth bass and bigheaded carp, and also span the range of pressures that would be targeted for use in a non-physical fish barrier (Cupp et al. 2016; Dennis et al. 2016a; Donaldson et al. 2016; Kates et al. 2012; Noatch and Suski 2012), as well as in areas downstream of barrier applications where CO2 may have dissipated relative to application sites. The use of multiple pCO2 levels is useful for identifying mechanisms and stimulating hypotheses about how pCO2 may impact fish biology (Heuer and Grosell 2014).

Following the acclimation period, water from the external holding tank at the pre-defined pCO2 was pumped into the test chamber, and the swimming trial was initiated 15 s later. Pumping CO2-rich water from the external holding tank caused water at ambient pCO2 in the swim tunnel to be displaced, and the excess water exited the swim tunnel via an overflow tube in the top of the tunnel. The propeller in the swim tunnel ensured mixing within the swim tunnel, and preliminary trials indicated that water in the swim tunnel reached target pCO2 after 15 s. Once target pCO2 was reached in the swim tunnel, largemouth bass were forced to swim in elevated pCO2 for either 30 s, 120 s, or 600 s, after which “fresh” water at ambient pCO2 was flushed through the swim tunnel. These durations were selected because fish exposed to elevated pCO2 for similar times have displayed both behavioral and physiological changes (Dennis et al. 2016a; Kates et al. 2012). Together, all possible combinations of the above CO2 pressures (< 400 μatm [ambient]; 10,000 μatm; 50,000 μatm; and 100,000 μatm) and exposure durations (30 s, 120 s, and 600 s) were tested in a 4 × 3 full-factorial design for a total of 12 distinct treatments (n = 8 fish per treatment).

At the start of the swimming trial, water velocity was increased at a linear and continuous rate of 0.2 BL s−1 every minute (adapted from Reidy et al. 2000). Fish maintained swimming under these conditions for anywhere from 10–25 min. The swimming trial was concluded when the fish was no longer able to maintain swimming position in the current, evidenced by it becoming pinned against the downstream grate of the test chamber, at which point the water velocity was recorded as its Uburst swimming velocity (Gregory and Wood 1998; Reidy et al. 2000). Fish that lost equilibrium were easily distinguished from fish that were simply exhausted as they rolled and inverted just before contacting the rear grate, whereas fish that became exhausted before losing equilibrium remained vertically oriented and struggled to continue swimming after contacting the grate. If equilibrium loss occurred, this was noted and the water velocity was similarly recorded as an endpoint for Uburst swimming velocity. At either of these two endpoints, the motor was turned off and the fish was allowed to recover before being removed from the swim tunnel. All fish were measured and size did not differ across treatments (n = 8 fish per treatment, ANOVA: total Length, F12,103 = 0.66, p = 0.784). Water quality data for each swimming trial was also measured from the source header tank providing water to the swim tunnel (Table 2). Trials were always performed during daylight hours (typically between 08:00 and 16:00) at an average rate of 3 fish per day.

Table 2 Water quality parameters measured during the Uburstswimming trials (A) (n = 8) and during the sprint swimming trials (B) (n = 10) in the external header tank that provided water to the swim tunnel. Water from the header tank was pumped into the swim tunnel for swimming trials, and header tank water replaced swim tunnel water in approximately 15 s (confirmed in preliminary trials using the same probe to measure pCO2 of water exiting the swim tunnel outflow). Data shown are mean ± 1 standard error
Full size table

One additional experiment was performed during which fish were forced to swim in the tunnel without pumping ambient water in from a separate header tank. The purpose of this experiment was to define any impacts of pumping water into the swim tunnel on swimming performance. These methods and results are detailed in Supplementary Material and indicated no impact of pumping water from an external header tank into the swim tunnel on fish swimming performance, such that any changes to swimming performance can be attributed to the increase of CO2.

Sprint duration

The impacts of exposure to elevated pCO2 on sprint duration were quantified using the same swim tunnel and conditions as described above (fish were not re-used from the burst swimming trials). For these trials, however, following the 10-min acclimation to the swim tunnel at a velocity of 0.5 BL s−1, water velocity was immediately increased to 4 BL s−1 to transition the fish into a sprinting gait (detailed in Reidy et al. 2000; Tierney 2011). At this time, CO2-treated water at the same targeted pressures was again pumped into the swim tunnel using methods identical to those described above. CO2-treated water was pumped into the swim tunnel 15 s before the trial was initiated and the velocity was increased, to ensure that the target level was reached in the swim tunnel before the trial began. Time to exhaustion for this sprint test was recorded when the trial ended, which was defined by one of two endpoints: when the fish was no longer able to maintain swimming position and became pinned against the downstream grate of the test chamber, or when the fish lost equilibrium and ceased swimming activity, again. At this point, the motor was turned off and the fish was able to recover. All fish were measured and size did not differ across treatments (n = 10 fish per treatment, ANOVA: total length, F3,39 = 1.10, p = 0.361). Water quality data were also measured from the source header tank providing water to the swim tunnel (Table 2).

Data analyses

A linear model was used to quantify the fixed effects of pCO2, exposure duration, and the interaction between pCO2 and exposure duration on Uburst swimming (a Tukey test was used post hoc). Similarly, the effect of pCO2 on sprint duration was assessed using a one-way analysis of variance (ANOVA), with pCO2 as the independent variable. All data were log-transformed prior to analyses to ensure that residuals from fitted models passed visual inspection for both normality and homogeneity of variances, and significance (α) was tested at the 95% confidence level (Zar 1984). Logistic regressions were used to analyze the effects of pCO2 and exposure duration on loss of equilibrium during the Uburst swimming protocol and to analyze the effects of pCO2 on loss of equilibrium during the sprint duration test. All statistical tests were completed in JMP Pro 11 (SAS Institute, Cary, NC).

Results

Uburst swimming velocity did not differ for largemouth bass exposed to pCO2 levels of either 10,000 or 50,000 μatm relative to swimming velocity at 400 μatm (ambient) (Fig. 1a). However, when forced to swim in water treated to 100,000 μatm pCO2, largemouth bass experienced a significant 30% decrease in Uburst swimming velocity relative to water at ambient and 10,000 μatm. The decrease in Uburst swimming velocity occurred independent of exposure duration (Fig. 1a; Table 3). In addition, fish lost equilibrium at a significantly higher rate as pCO2 increased, and this loss of equilibrium was independent of exposure duration (Fig. 1b; Table 4). More specifically, Uburst swimming in water at either ambient pCO2 or in water treated to 10,000 μatm did not cause any loss of equilibrium, and all fish were able to swim until exhaustion. When exposed to water at 50,000 μatm for 600 s, 15% of fish lost equilibrium prior to reaching exhaustion, while water treated to 100,000 μatm resulted in loss of equilibrium at rates of 12.5%, 25%, and 50%, respectively, for the different exposure durations (30 s, 120 s, and 600 s) (Fig. 1b; Table 4).

Fig. 1
figure 1

Uburst swimming velocity (measured in BL s−1) (a) and proportion of individuals that lost equilibrium (b) for largemouth bass (M. salmoides) exposed to a range of pCO2 (< 400; 10,000; 50,000; or 100,000 μatm) for either 30 s, 120 s, or 600 s (n = 8 fish per treatment). Dissimilar letters (a and b) represent significant differences across pCO2 treatments

Full size image
Table 3 Results from a linear model used to analyze the effects of pCO2 (ambient < 400 μatm; 10,000 μatm; 50,000 μatm; and 100,000 μatm) and exposure duration (30 s, 120 s, and 600 s) on Uburst swimming performance in largemouth bass. Significant results are shown in italics
Full size table
Table 4 Results from a logistic regression used to analyze the effects of pCO2 and exposure duration on loss of equilibrium (LOE) during Uburst swimming, and the effects of pCO2 on loss of equilibrium during sprint swimming. Significant results are highlighted in italics
Full size table

Largemouth bass experienced no significant changes in sprint duration (measured as time to exhaustion) across any of the pCO2 treatments (Fig. 2a; ANOVA, F3,39 = 1.12, p = 0.296). However, the proportion of fish that lost equilibrium significantly increased with pCO2, similar to results from the Uburst swimming test. More specifically, none of the largemouth bass lost equilibrium during the sprint test at 400 μatm (ambient) or 10,000 μatm pCO2, only 20% of fish tested lost equilibrium during the sprint test at 50,000 μatm, and 70% lost equilibrium at 100,000 μatm (Fig. 2b; Table 4).

Fig. 2
figure 2

Time to exhaustion during sprint swimming trials (measured in seconds) (a) and proportion of individuals that lost equilibrium (b) for largemouth bass during sprint swimming trials when exposed to a range of pCO2 (< 400; 10,000; 50,000; or 100,000 μatm) (n = 10 fish per treatment)

Full size image

Discussion

Largemouth bass only experienced an impairment in burst swimming performance at the highest levels of CO2 tested, and this was independent of exposure duration. Specifically, Uburst swimming performance decreased by about 30% at 100,000 μatm relative to ambient pCO2. When fish operate in areas where environmental pCO2 is increased, fish blood is at risk of becoming acidified due to the influx of protons (Janssen and Randall 1975). Fish are generally efficient acid-base regulators, however, and they can often maintain appropriate blood pH by either exporting protons or importing base (bicarbonate) (Evans et al. 2005). If acid-base regulating mechanisms are unable to maintain homeostasis under hypercarbic conditions, this can result in an impaired ability to transport oxygen due to reductions in hemoglobin efficiency explained by the Bohr and Root effects (Brauner and Randall 1996), which may impair certain swimming activities. While the present study tested burst swimming which is primarily anaerobically fueled, fish swam for as long as 30 min and may have employed some aerobic metabolism. The fact that we saw no impairment of burst swimming at the low pCO2 treatment can indicate that largemouth bass may be able to successfully regulate acid-base status to maintain oxygen delivery to tissues, but are unable to maintain homeostasis above 100,000 μatm. Largemouth bass are robust at maintaining oxygen delivery to tissues in the face of external challenges with Furimsky et al. (2003) demonstrating that largemouth bass can maintain oxygen delivery to tissues despite ambient dissolved oxygen falling by 60%. Alternatively, if the effects of hypercarbia are being derived in the central nervous system, this may have a similar outcome as effects manifested in the muscles. Together, results from this study clearly show that elevated CO2 impairs the ability of fish to burst swim, but only at pCO2 above 50,000 μatm.

Largemouth bass also did not show any changes in sprint swimming performance when exposed to a range of CO2 pressures. Specifically, sprint duration while swimming at 4 BL s−1 did not differ between ambient (< 400 μatm) and 100,000 μatm pCO2. Sprint swimming, which is powered by white muscle, is often employed by fish for short duration, high-intensity behaviors such as capturing prey and avoiding predation (Domenici and Blake 1997), or making upstream movements through high-velocity flows. Sprint swimming is fueled by anaerobic glycolysis with energy substrates coming from stores within the white muscle (Dobson et al. 1987; Garenc et al. 1999). While Randall and Brauner (1991) suggest that sprinting activities are largely unaffected by changes in environmental conditions unless they alter muscle efficiency or energy stores, other work suggests that factors such as temperature (Batty and Blaxter 1992) and hypercarbia (Dennis et al. 2016a) can affect sprint swimming performance. More specifically, Dennis et al. (2016a) showed that sprint swimming performance by juvenile largemouth bass decreased when exposed to 120 mg/L CO2. However, this decrease was after an exposure of 2 h that could have potentially led to partial anesthesia or altered acid-base physiology and hence a reduced swimming performance. Results from this study demonstrate that elevated CO2 environments do not impair the ability of largemouth bass to perform sprint swimming.

During both burst and sprint swimming activities, largemouth bass lost equilibrium with increasing frequency as pCO2 increased. The proportions of fish that lost equilibrium during trials were similar during burst and sprint swimming activities, occurring around 20% of the time at 50,000 μatm and up to 70% of the time at 100,000 μatm (depending on exposure duration). CO2 is known to have anesthetic properties on fish and has previously been used as such in fisheries work (Gilderhus and Marking 1987). As environmental pCO2 increases, the gradient between a fish’s external and internal CO2 is reduced or reversed, rendering the fish unable to excrete CO2 (Janssen and Randall 1975). Fish are able to counteract this internal acidification by exporting protons or gaining bicarbonate, but this typically fails above 10,000 μatm (reviewed by Heuer and Grosell 2014). While the exact mechanism that causes fish to lose equilibrium is not well understood, movement of CO2 across the blood-brain barrier and disrupting energy processing is likely the cause (Yoshikawa et al. 1994). In our study at pressures below 100,000 μatm, largemouth bass appear to have the ability to maintain homeostasis as little equilibrium loss occurred. Once pressures of 100,000 μatm were reached, however, there was a significant increase in equilibrium loss independent of exposure duration, indicating that this pressure likely overwhelmed acid-base mechanisms. Together, results from this study clearly show that a CO2-induced loss of equilibrium will frequently occur at 100,000 μatm pCO2 during both burst and sprint swimming in largemouth bass.

Research from static ponds and tanks indicate carp will swim away from a CO2 barrier, giving it potential to act as a “push-type” barrier that can repel fish from an area (Noatch and Suski 2012). As such, CO2 is being considered as a barrier to deter the movements of invasive bigheaded carp in Illinois, with flowing water as a potential application site (United States Army Corps of Engineers 2014). Target pCO2 for the barrier based on previous findings has been estimated to be around 30,000–50,000 μatm (Cupp et al. 2016; Dennis et al. 2016b; Donaldson et al. 2016). Deployment strategies will likely involve injecting CO2 into a lock or creating a “wall” of CO2 at a chokepoint such as a culvert or a lake inlet, with elevated pCO2 presumably dissipating as water flows downstream. Our study shows that if fish decide to challenge a barrier and swim through it using either burst or sprint swimming, there would be little if any impairment to their swimming ability at pressures previously demonstrated to induce avoidance. However, a barrier could still impede passage of a portion of fish by inducing a loss of equilibrium, but this goal would necessitate a target of 50,000–100,000 μatm and would need to be paired with other technologies for complete deterrence. Future deployments could either rely on the barrier acting as a push-type deterrent at 50,000 μatm or raise the target pressure to 100,000 μatm to achieve equilibrium loss, thereby increasing the likelihood of immobilization. Thresholds for equilibrium loss need to be defined in other target fish species in flowing water, with a priority on potentially invasive fishes (e.g., bigheaded carp, ruffe (Gymnocephalus cernuus), and round goby (Neogobius melanostomus)).

Aquatic invasive species are problematic due to their negative impacts on ecosystem services and commerce and their ability to outcompete native species (Ehrenfeld 2010; Smith et al. 2015). More importantly, it is more difficult to eradicate invasive species from an area following establishment than to prevent them from spreading (McDermott et al. 2013). To date, prototype CO2 barriers have proven effective as push-type detractors that exclude fish from target areas, and results from our study show that a carbon dioxide barrier will likely not impair swimming performance unless target pressures of 50,000–100,000 μatm are reached. Efforts should be made to better define the impairment to target species such as Asian carp and to begin scaling deployment efforts from lab to field trials in hopes of protecting the aquatic ecosystems that are at risk from invasive species.