Electrosensory and metabolic responses of weakly electric fish to changing water conductivity

Anthropogenic activity such as deforestation, dams, mining, industrial waste disposal and agricultural runoff negatively impact the estimated 5000 species of fishes living in the freshwater ecosystems of South America (Reis et al., 2016) by degrading water quality, crating barriers that inhibit aquatic migration, and altering seasonal weather patterns (Castello and Macedo, 2016; Coe et al., 2011; Costa and Foley, 1997). Changes in ionic concentrations, especially salinity, are a common consequence of many anthropogenic disturbances, resulting in corresponding changes in the water's electrical conductivity. Although most freshwater fishes can adapt to changes in water conductivity, experienced as osmotic stress (Kültz, 2015), the South American weakly electric gymnotiform fishes may experience an additional type of stress owing to the potential effects of water conductivity on their electric signals.

Gymnotiform fishes generate electric fields, known as electric organ discharges (EODs), and detect distortions of these electric fields for sensory processes such as navigation, communication and foraging. In all gymnotiform species, the EOD is produced by an electric organ (EO) that extends bilaterally along the body and into the tail. The EO is composed of electrogenic cells, known as electrocytes, that produce the EOD via coordinated action potentials. Electric current, generated by ionic currents through the electrocyte voltage-gated ion channels, is directed along the fish's body and produces an electrical current that flows through the surrounding water (Markham, 2013). Gymnotiform species are categorized either as wave-type fishes that produce EODs at continuous frequencies of ∼100–2000 Hz, or as pulse-type fishes that generate EODs at low intermittent rates of ∼10–100 Hz. In all species examined to date, EOD production incurs significant metabolic costs: as high as 30% of the daily energy budget in wave-type fishes and 20% in pulse-type fishes (Lewis et al., 2014; Salazar and Stoddard, 2008; Salazar et al., 2013).

EOD amplitude (measured as electrical potential in water) likely coincides with the fish's electrosensory range – the distance at which fish can detect objects or communicate with conspecifics (Rasnow, 1996). In accordance with Ohm's law, which states that for a constant electrical current, electrical potential is inversely related to conductivity, EOD amplitude is determined by the fish's EO output and water conductivity. Accordingly, increased conductivity reduces EOD amplitude if EO output is constant, whereas decreased conductivity increases EOD amplitude at a constant EO output. In a shift from low to high water conductivity, some species have shown decreased sensory performance (MacIver et al., 2001), suggesting that increased water conductivity may degrade a fish's ability to forage or detect prey.

Many species of electric fish modulate EO output on a circadian rhythm and in response to social encounters (Markham et al., 2009; Markham and Stoddard, 2005), which suggests a potential mechanism to compensate for the effects of increased water conductivity on EOD output by increasing EO output accordingly. This led us to investigate whether fish increase EOD amplitude to compensate for the effects of increased water conductivity. If fish do compensate by increasing EO output, an important second question that follows is whether the increase in EO output requires additional metabolic investment in EOD production. Given the high metabolic costs of EOD production under normal circumstances, any additional metabolic investment could make EOD production physiologically unsustainable forcing a decrease in sensory performance, or a metabolic trade-off with physiological functions such as immunity and reproduction.

To address these questions, we studied the electrosensory and metabolic responses of pulse- and wave-type gymnotiform species to rapid increases in water conductivity. This comparative approach allowed us to evaluate whether species with different EOD rates and different life histories might exhibit different response strategies. We measured EOD amplitude as a metric of electrosensory performance and measured metabolic responses by intermittent flow respirometry. In increased water conductivity, we hypothesized that if the fish increases EO output to compensate for the reduced EOD amplitude, this response will cause an associated increase in metabolic rate. Conversely, if the fish does not increase EO output, it will maintain a constant metabolic rate, likely forcing a trade-off in electrosensory performance.

The fish used for this study were the wave-type gymnotiform Eigenmannia virescens (Valenciennes 1847) and the pulse-type gymnotiform Brachyhypopomus gauderio Giora and Malabarba 2009. Eigenmannia virescens were obtained through tropical fish importers and B. gauderio were obtained from an on-campus breeding colony at the University of Oklahoma. Fish were housed in an indoor recirculating aquarium system with a 12 h:12 h light:dark light cycle, at 26°C, and fed live oligochaete black worms ad libitum. Water for the animals was prepared from reverse-osmosis purified and deionized water. Water conductivity was controlled using a concentrated pH-buffered saline solution (Walter's solution) consisting of deionized water containing (mmol l⁻¹): CaSO₄·2H₂O (732), MgSO₄ (83), KCl (107), NaH₂PO₄·H₂O (17) and FeC₆H₅O₇ (7). Fish care and experimental protocols were approved by the Institutional Animal Care and Use Committee of the University of Oklahoma.

EOD recordings and respirometry experiments were performed separately but under standardized water conductivity parameters. Fish were acclimated to the low conductivity condition (150±50 μS) for a minimum of 7 days before data collection began to eliminate stress related to changing water conditions during baseline measurements. Measurements were collected within the low conductivity condition for 2 days (day –2 and day –1) before Walter's solution was added on day –1 to raise conductivity to the high conductivity condition (350±50 μS), which was stable 24 h later (designated as day 0). Measurements then continued from days 0 through 6 in the high conductivity condition. The control condition (respirometry experiments only) followed the same experimental timeline, but fish remained in the low conductivity condition throughout the experiment.

The EODs of seven B. gauderio and seven E. virescens were measured continuously during experimental days −2 through 6. Fish had constant access to food throughout the experiment.

Procedures for measuring calibrated EODs in free-swimming fish followed standard methods reported previously (Stoddard et al., 2003). Individual fish were placed in a 285-liter glass aquarium (120×44×44 cm) that was electrically shielded with grounded aluminum mesh screen. The aquarium was divided into three compartments with fiberglass screen panels, and a mesh tube in the center compartment connected the outer two compartments such that the fish could swim between the outer compartments by passing through the center tube. The fish remained in the aquarium when Walter's solution was added to increase water conductivity. A custom-built amplifier detected when the fish was centered in the tube, which then triggered a real-time digital processor (Tucker-Davis Technologies RP8, Alachua, FL, USA) to digitize the EOD at 48 kHz across a different pair of nichrome wires at opposite ends of the tank. EODs were amplified at 500× gain and low-pass filtered at 500 kHz (Cygnus FLA-01, Cygnus Tech, Delaware Water Gap, PA, USA).

The mass-specific respiration rate (mmol l⁻¹ O₂ g⁻¹ min⁻¹) of 14 E. virescens and 14 B. gauderio was measured once during the low conductivity condition (day –1) and twice in the high conductivity condition (days 1 and 6). Under the control condition, respiration was measured in additional E. virescens (n=14) and B. gauderio (n=14) following the same schedule as the experimental group (days –1, 1 and 6). Respiration rate scales allometrically with body mass in gymnotiform fish (Julian et al., 2003). In the present experiments, we assessed mass-specific respiration rate because our focus was to compare changes in respiration rate within rather than between individuals.

Fish were acclimated to individual open flow respiration chambers for a minimum of 18 h before respiration measurements to ensure that fish were in a post-absorptive state so digestion would not affect their metabolic rate. Fish were acclimated in the same conductivity as the treatment condition (low conductivity or high conductivity). The prolonged acclimation period also served to prevent transient changes in respiration rate caused by initial stress when fish are first placed in the chamber. Each fish's length was recorded once at the start of the experiment and mass was recorded after each respiration measurement. EOD frequency (Hz) was recorded for the E. virescens, once, at the start of the experiment. To control for background oxygen consumption caused by the chamber's microbial activity, background respiration was recorded immediately after each fish's respirometry measurement from the empty chamber after the fish was removed, and the mass-specific respiration measurements were corrected according to recommended practices (Svendsen et al., 2016).

The respiration chamber was a translucent acrylic cylinder, 24 cm in length and 5 cm in diameter, with threaded polyoxymethylene caps on each end (AlphaCool, Model 15719, Braunschweig, Germany). The end caps were fitted with quick-connect receptacles (9.5 mm inside diameter; Colder Products Company, Roseville, MN, USA). During acclimation periods, the quick-connect receptacles were open, leaving a 9.5-mm opening on each side – large enough to allow water exchange through the chamber but small enough that fish could not swim out of the chamber. During respiration measurements, quick-connect fittings attached to aquarium tubing and a 9 V DC aquarium pump were connected to the chamber's quick-connect receptacles, creating closed circulation through the chamber.

During the open circulation phase of the respiration measurements, oxygen saturated water from a large outer tank was pumped through the respiration chamber, while during closed circulation the pump circulated water through the respiration chamber without introducing new water from the outer tank. Open and closed circulation were controlled by connecting and disconnecting the tubing from the inflow side of the pump. The outflow side of the pump passed water through a custom-constructed acrylic plastic measurement chamber, where water oxygen concentration was measured with a NeoFox phase fluorometer (Ocean Insight, Orlando, FL, USA), from a RedEye FOSPOR oxygen indicator patch (Ocean Insight) adhered to the interior of the measurement chamber. Similar fluorometric technology has been successfully used previously for respirometry with weakly electric fish (Julian et al., 2003) and the performance of RedEye FOSPOR indicators in freshwater applications is highly accurate and reliable across a broad range of temperatures (10–50°C) and oxygen saturations (0–100%) (Ocean Insight specifications: accuracy ±0.1%; resolution <0.1%; drift <0.001% h⁻¹). The RedEye oxygen indicator patch was calibrated using a two-point calibration method with the first calibration point taken at a dissolved oxygen level of zero, achieved by bubbling nitrogen into aquarium water, and the second point taken from aquarium water at full atmospheric oxygen saturation. Oxygen concentration was recorded from the NeoFox unit on a laptop PC using NeoFox Viewer software (Ocean Insight).

For both B. gauderio (n=7) and E. virescens (n=7), EOD amplitude was measured from the negative minimum to the positive maximum of the EOD waveform (Fig. 1). Additional EOD parameters for B. gauderio included the amplitude of the EOD's positive first phase (P1a) and the amplitude of the second negative phase (P2a), measured from zero volts to the positive peak or negative minimum, respectively. The durations of P1 and P2 were measured as the duration of each phase at 50% of peak amplitude (Fig. 1).

Change in EOD waveform parameters over time and between species were analyzed by Student's t-test or univariate ANOVA as appropriate. Comparisons between before and after rapid conductivity change assessed the immediate effects of increased water conductivity on EOD parameters. Comparisons between day 0 and day 6 assessed changes in the same EOD characters during continued exposure to high conductivity. All analyses were performed with MATLAB (MathWorks, Natick, MA, USA). Data are presented as means±s.e.m.

Respiration rate was derived from a linear regression fit to the decline in water oxygen concentration during the closed circulation periods. Data were considered valid only if r_­² for the regression exceeded 0.99. The corrections for background respiration and the calculation of mass-specific respiration of each fish were calculated according to accepted standard practices (Svendsen et al., 2016). A two-way ANOVA was then used to compare the mass-specific respiration rates across species and across water conductivities. Significant omnibus tests were further analyzed by post hoc pairwise comparisons with experiment-wise alpha maintained at 0.05 by Tukey's HSD. A linear regression was also used to test for correlations of respiration rate with body condition index [BCI; mass (g) divided by length (cm)] in both species and a separate linear regression tested for a correlation of BCI with EOD frequency in E. virescens.

In the low conductivity condition, EOD amplitude was higher at baseline for B. gauderio (3.21±0.48 mV cm⁻¹) than for E. virescens (1.14±0.31 mV cm⁻¹; Fig. 2). After the addition of Walter's solution on day –1, water conductivity stabilized by mid-day on day 0. The increased water conductivity resulted in a reduction of EOD amplitude for B. gauderio, from 3.21±0.45 to 1.7±0.30 mV cm⁻¹, and produced a small but not statistically significant decrease from 1.15±0.29 to 0.89±0.18 mV cm⁻¹ for E. virescens (two-way ANOVA species×conductivity: species, F_1,24=14.74, P=0.0003; conductivity, F_1,24=5.36, P=0.018; interaction, F_1,24=2.65, P=0.087; post hoc pairwise comparisons via Tukey's HSD).

After the initial decline of EOD amplitude in high water conductivity, EOD amplitude increased over the course of 6 days in B. gauderio by 20.2±4.3% but did not increase in E. virescens, changing by only −0.05±6.1% (Fig. 3), a statistically significant difference in percentage change after 6 days between species (t=2.58, d.f.=12, P=0.024). We further evaluated changes in B. gauderio EOD amplitude by comparing the 6-day changes in additional EOD amplitude and duration parameters (e.g. Fig. 1). In B. gauderio, all three amplitude parameters (EOD amplitude, P1a and P2a) increased in tandem during the 6 days in high conductivity, whereas P1 duration and P2 duration did not change (Fig. 4; omnibus ANOVA, F_4,30=11.14, P<0.0001, with post hoc pairwise comparisons via Tukey's HSD).

Resting respiration rate was not correlated with BCI in E. virescens (R²=0.054, P=0.39), whereas resting respiration rate had a weak positive correlation with BCI in B. gauderio (R²=0.198, P=0.04). Because respiration rate was mildly correlated with BCI in only one species, we did not control for BCI during subsequent analysis.

In all experimental conditions, the respiration rate was higher for E. virescens than for B. gauderio (species main effect: F₁₁₇₅=141.4, P<0.0001; Fig. 5). After the addition of Walter's solution to increase water conductivity on day –1, water conductivity stabilized by mid-day on day 0. A significant species by conductivity interaction (F₃₁₇₅=4.93, P=0.0025, with post hoc pairwise comparisons by Tukey's HSD) supports that respiration rate in E. virescens decreased immediately after the change to the high conductivity condition (P=0.0034), whereas the respiration rate of B. gauderio remained constant after the change to high conductivity (P=1.0000). The respiration rates for both E. virescens and B. gauderio did not change over 6 days in high conductivity. In the control condition, respiration rates for both species remained constant throughout the experiment.

Following a rapid increase in conductivity, EOD amplitude decreased in accordance with Ohm's law only in the pulse-type B. gauderio, whereas the wave-type E. virescens unexpectedly showed only a small decrease in EOD amplitude. During the following 6 days of stable high conductivity, B. gauderio began recovering EOD amplitude within 48 h, ultimately increasing EOD amplitude by approximately 20% after 6 days. Surprisingly, during the rapid conductivity increase and the six subsequent days in high conductivity, there was no change in metabolic rate for B. gauderio, even as EOD amplitude increased, whereas metabolic rate decreased after the rapid increase in water conductivity for E. virescens even as EOD amplitude remained constant.

Because EOD amplitude determines the fish's electrosensory range (Assad et al., 1999; Nelson and Maciver., 1999), these results suggest that B. gauderio prioritizes the maintenance of sensory range during periods of increased water conductivity through some mechanism of recovering EOD amplitude that does not increase total metabolic rate. Under the same conditions, the EOD amplitude of E. virescens did not decrease as expected when measured several hours after the addition of a saline solution to the water, suggesting perhaps a rapid physiological mechanism that adjusts EOD amplitude to changes in conductivity faster than our measurements could detect. The decreased metabolic rate observed in E. virescens may or may not be associated with any such mechanism. Several factors, alone or in concert, could account for the widely divergent responses of these species to disruptions in electrosensory and electric communication performance after a sudden increase in water conductivity.

Differences in habitat offer one potential explanation for the observed differences between species. Eigenmannia virescens inhabits deep waters with historically stable water conditions (Silva et al., 2003), whereas B. gauderio inhabits a variety of aquatic environments including riverbanks, slow-moving creeks, and floodplains (Giora and Malabara, 2009). Thus, B. gauderio may be adapted to habitats where rapid changes in water conductivity are common, whereas E. virescens may be suited for less variability than B. gauderio and is instead well adapted for a limited range of change.

Divergent reproductive life histories could also explain why B. gauderio restores EOD amplitude whereas E. virescens does not. Semelparous species, such as B. gauderio, have one breeding period before death, whereas iteroparous species, such as E. virescens, have multiple breeding periods over the species' lifespan. Urgency to mate, such as a terminal investment in reproduction, could in part explain why B. gauderio responds to increased conductivity by directing effort into EOD amplitude recovery, thus prioritizing communication. A similar pattern of preserving communication during stress occurs when B. gauderio males increase both EOD amplitude and EOD duration in response to food deprivation, again perhaps as a terminal investment in reproduction (Gavassa and Stoddard, 2012).

In contrast, the decrease in the metabolic rate of E. virescens could be a characteristic of iteroparity, possibly suggesting an effort to conserve energy and wait for conditions to improve. However, this is inconsistent with earlier reports of that behavior where E. virescens reduced energetic costs under hypoxia by diminishing EOD amplitude, but without an associated reduction in metabolic rate (Reardon et al., 2011). Furthermore, when under metabolic stress from food deprivation, E. virescens reduces EOD amplitude over several days, possibly as a reproductive strategy to conserve energy until conditions improve (Sinnett and Markham, 2015).

Contrary to our expectations, B. gauderio showed no increase in metabolic rate associated with their gradual increase in EOD amplitude. This outcome can only be explained in one of two ways: B. gauderio recovers EOD amplitude either by increasing metabolic investment in EO output while reducing metabolic investment in other physiological processes, or by a mechanism that requires no additional metabolic investment.

If B. gauderio invests more energy in EOD production by diverting metabolic resources from other physiological processes, this would represent a potentially harmful trade-off between electric signaling and other life processes. These types of metabolic trade-offs are common among animals (Stearns, 1989; Zera and Harshman, 2001; Moore and Hopkins, 2009). For example, when exposed to increased water temperatures, the resting metabolic rate (RMR) of the Antarctic fish Trematomus bernacchii increases initially in a temperature-dependent fashion. The RMR then returns to baseline levels over the course of 9 weeks, but that happens with an associated decrease in body mass (Sandersfeld et al., 2015). Another example can be seen in the trade-off between energy conservation and increased predation in the mourning dove, Aristolochia macroura. During the winter months, when food availability is low, A. macroura uses regulated nocturnal hypothermia to conserve energy. However, this energy saving behavior comes at the expense of increased predation owing to an associated reduction in flight ability (Carr and Lima, 2013).

We believe a more likely explanation is that B. gauderio is recovering EOD amplitude by a mechanism that does not require additional metabolic investment. In all studies published to date where B. gauderio increases EOD amplitude by increasing electrocyte power output and metabolic investment in EOD production, increased EOD amplitude is accompanied by corresponding increases in EOD duration, P1 duration and P2 duration (Markham and Stoddard, 2005, 2013; Salazar and Stoddard, 2008). This was not the case in the present study, wherein B. gauderio increased EOD amplitude with no accompanying changes in EOD duration, P1 duration and P2 duration, suggesting a different physiological mechanism is at work. If B. gauderio is instead increasing EOD amplitude without additional metabolic investment in EO output, the most feasible mechanism for doing so is impedance matching. When increased water conductivity reduces the impedance of the water (the load impedance), the fish may be reducing the internal resistance of the EO (the source impedance) to maximize power transfer from the EO into the water in the new conductivity condition. One mechanism for reducing the internal resistance of the EO would be to decrease the electrocyte membrane resistance, which is known to occur in response to stress and social activity in other gymnotiforms (Markham et al., 2009), thereby reducing the internal resistance of the EO without requiring additional metabolic investment. To test this, the internal resistance of the EO can be derived by recording EODs from electrodes positioned directly on the fish's skin while a portion of the fish is out of water (Cox et al., 1945). This technique, also known as the ‘air gap’ technique (Caputi et al., 1989, 1993), also allows for systematic variation of the external load resistance across the electrodes. The resulting measurements permit derivation of the internal voltage and resistance of the EO before and after a shift in water conductivity, enabling a direct assessment of whether B. gauderio is using impedance matching to compensate for the effects of high water conductivity on their EOD amplitude.

In accordance with Ohm's law, EOD amplitude should be reduced by half when water conductivity is increased by a factor of two, but surprisingly, for E. virescens, EOD amplitude decreased very little after water conductivity was doubled. In this experiment, the effects of conductivity on EOD amplitude were assessed by comparing EOD amplitude before addition of the saline solution with that measured after conductivity had stabilized several hours later. Measured in this manner, the EOD amplitude of E. virescens did not decrease significantly. One possible explanation is that E. virescens employs a rapid mechanism to compensate for changes in conductivity within a matter of minutes or hours. Such changes would not have been detected with the methodology used in the present study.

Another surprising finding of this study was that E. virescens showed a large decrease in metabolic rate after water conductivity increased. One possible explanation is that this decrease in metabolic rate is the result of lower activity levels in electrosensory areas of the brain. Electrosensory processing in the brain incurs high metabolic costs (Sukhum et al., 2016; Nilsson, 1996), driven in part by large populations of neurons firing 1:1 with the EOD rate (Salazar et al., 2013). Reduced EOD amplitude in high water conductivity could reduce activity levels in these neuronal populations, thereby reducing overall metabolic demand in peripheral and central neural systems that encode and process electrosensory information (Sukhum et al., 2016). However, because the decrease in EOD amplitude was not significant for E. virescens, diminished electroreceptor activation by reduced EOD amplitude would not account for a decrease in metabolic rate. Perhaps instead, a reduction in electroreceptor sensitivity as a response to increased conductivity would then lead to lower metabolic costs of central electrosensory processing.

An important limitation of this work is that we exposed fish to only two different water conductivities. Perhaps the shift from 150±50 to 350±50 μS was not large enough to elicit a metabolic response from B. gauderio or an EOD amplitude response in E. virescens. Another variable not considered within the study design was the normal water conditions in each species' habitat. Instead, the conductivities were standardized for both species, creating the possibility that one or both conductivity phases could have unintentionally been more favorable to one species. We did not measure EOD rate or assess fish movement during the respirometry experiments, an important limitation because both factors have been shown to affect respiration rate in weakly electric fish (Julian et al., 2003; Salazar and Stoddard, 2008). These variables will be explored in future studies.

Further research is needed to address how weakly electric fish respond to anthropogenic disturbances in water quality, and should include a larger diversity of species and a wider range of water conditions. This includes examining response differences between species who differ significantly in EOD frequencies, species whose EO is derived from muscle tissue (myogenic) versus neural tissue (neurogenic), and comparisons between Neotropical gymnotiforms and Afrotropical mormyrid electric fish. Further investigation involving different magnitudes of change in water conductivity is also needed to determine how E. virescens responses to increased water conductivity. Additionally, the effects of rapid transitions from high to low conductivity (opposite to this study), and prolonged exposure to altered water conductivity should be explored.

Continued examination is necessary to determine what sensory mechanism in B. gauderio initiates the increase of EOD amplitude in high conductivity. Possible mechanisms include the direct sensing of water osmolarity or conductivity, or electrosensory detection of the reduction in EOD amplitude. It will be important to assess the nature and extent of how changes in EOD amplitude affect sensory performance with respect to navigation, object detection and prey detection/capture. Additionally, future research should examine whether communication or social interactions are altered or impaired after increases in water conductivity.

The negative impacts of anthropogenic activity on South American freshwater ecosystems are expected to accelerate (IPCC, 2022) if preventative actions are not immediately implemented (Kuemmerlen et al., 2022). As a result, electric fishes may be disproportionately harmed relative to other freshwater fishes as pollution harms physiology in general, but also potentially degrades their primary sensory and communication modalities (Markham et al., 2016). The findings of the present study suggest that weakly electric fish may display widely divergent responses to changes in water conditions, and emphasizes that a more comprehensive and comparative analysis of how electric fish respond to disturbances in aquatic habitats will be essential for predicting and perhaps mitigating the consequences of anthropogenic activity to neotropical aquatic habitats.

We thank Rosalie Maltby for laboratory support and fish care. We are grateful to Caryn Vaughn and Ricardo Betancur-R. for comments on an earlier version of this manuscript. The data presented in this paper were reported in a master's thesis submitted by S.D.W. to the University of Oklahoma, available at: https://shareok.org/handle/11244/336889.