Feeding frequency does not interact with BPA exposure to influence metabolism or behaviour in zebrafish ( Danio rerio )

Resource limitation can constrain energy (ATP) production, and thereby affect locomotion and behaviour such as exploration of novel environments and boldness. Consequently, ecological processes such as dispersal and interactions within and between species may be influenced by food availability. Energy metabolism, and behaviour are regulated by endocrine signalling, and may therefore be impacted by endocrine disrupting compounds (EDCs) including bisphenol A (BPA) derived from plastic manufacture and pollution. It is important to determine the impacts of these novel environmental contexts to understand how human activity alters individual physiology and behaviour and thereby populations. Our aim was to determine whether BPA exposure interacts with feeding frequency to alter metabolism and behaviour. In a fully factorial experiment, we show that low feeding frequency reduced zebrafish ( Danio rerio ) mass, condition, resting metabolic rates, total distance moved and speed in a novel arena, as well as anxiety indicated by the number of times fish returned to a dark shelter. However, feeding frequency did not significantly affect maximal metabolic rates, aerobic scope, swimming performance, latency to leave a shelter, or metabolic enzyme activities (citrate synthase and lactate dehydrogenase). Natural or anthropogenic fluctuation in food resources can therefore impact energetics and movement of animals with repercussions for ecological processes such as dispersal. BPA exposure reduced LDH activity and body mass, but did not interact with feeding frequency. Hence, behaviour of adult fish is relatively insensitive to disruption by BPA. However, alteration of LDH activity by BPA could disrupt lactate metabolism and signalling and together with reduction in body mass could affect size-dependent reproductive output. BPA released by plastic manufacture and pollution can thereby impact conservation and management of natural resources.


Introduction
Food availability determines the total amount of metabolic fuel that can be converted to cellular energy (adenosine triphosphate, ATP), which is allocated to somatic maintenance, movement and behaviour, growth, and reproduction [1][2][3].Animals may respond to reduced food availability and the resultant energy limitation by reducing maintenance (resting) metabolic rates, and by altering behaviour to either reduce movement and hence energy expenditure or to increase foraging to improve the chances of locating food [4].Feeding frequency can influence the energy available for locomotion [5,6], and reduced food availability can decrease locomotor activity and performance in fish [7,8].However, the relationship between feeding and metabolic rate (oxygen consumption) on the one hand, and ATP production on the other can vary depending on mitochondrial efficiency [9,10].For example, compared to fed fish fasted brown trout (Salmo trutta) and southern catfish (Silurus meridionalis) showed increases or no change in sustained swimming speeds, respectively, indicating no decline in muscle performance despite decreased food availability [11,12].The effects of reduced food availability were offset by an increase in the efficiency of mitochondrial ATP production in trout and zebrafish (Danio rerio) [10,13].Nonetheless, prolonged energy limitation can lead to altered behaviour to prioritize foraging over risk avoidance [14,15].However, it can also result in a decrease in the total distance travelled during exploration trials suggesting an attempt to balance energy use while maximizing foraging [10].Exploration behaviour encompasses anxiety and boldness, which are important for predator avoidance and foraging [15,16], and adjustments in these behavioural phenotypes can be important to mitigate the effects of energy limitation.The interaction between resource availability on the one hand, and physiological and behavioural responses on the other can influence broader ecological processes such as dispersal into novel environments [17,18].
All behavioural and physiological traits are controlled in some manner by the action of hormones.Thyroid and glucocorticoid hormones in particular regulate feeding, metabolism and locomotor performance in fish [19][20][21][22][23]. Anthropogenic release of endocrine-disrupting chemicals (EDCS) from plastic manufacture and pollution can interfere with neuro-endocrine systems that regulate feeding behaviour, locomotion, and energetics [20,[24][25][26][27]. Human activity has now introduced EDCs into most aquatic systems globally [28].Bisphenol A (BPA) is one of the most highly produced industrial chemicals in the world, and acts as an exogenous ligand or blocker of hormone receptors to disrupt normal signalling pathways [28][29][30].BPA can thereby impact energetics and performance in a variety of traits including swimming performance, growth, and metabolism [27].Exposure to BPA can also have neurological effects and alter behaviour independently from energetics [31,32], and it reduced exploration distance in adult zebrafish [32].Hence, both resource availability and BPA exposure can alter behaviour and energetics by themselves, and they can interact by impacting the same processes.These interactions could result in mismatched responses such as increased metabolic rates and exploration in low energy environments.It is important to determine the impacts of these novel, anthropogenically altered environments on animal physiology and behaviour to predict their consequences for animal populations.These insights are essential for effective conservation and management of natural resources [33,34].
Our aim was to test the effects of energy limitation and BPA exposure, and their interaction on metabolism and behaviour.We conducted a fully factorial experiment using adult zebrafish where we manipulated feeding frequency to simulate energy limitation, and BPA exposure to test the hypotheses that 1) BPA exposure and limited resources both have a main effect in reducing metabolic and swimming performance, and exploratory behaviour; 2) there is an interaction between resource availability and BPA exposure so that there will be an increase in anxiolytic behaviors that reduce exploration of a novel environment even if resource levels are high.

Animal husbandry
Adult zebrafish (mean mass = 0.37 ± 0.0090 [s.e.] g; mean standard length = 2.73 ± 0.022 [s.e.] cm) of mixed sex were obtained from a commercial supplier (Livefish, Bundaberg, Australia) and housed in 10 l glass tanks (30×19×21 cm; 1-2 fish l − 1 ) at 23 • C, which is the same temperature fish were maintained at the supplier, and with a photoperiod of 12 h light: 12 h dark.Temperatures were maintained with aquarium heaters (Eheim, Germany) and all tanks contained air stones.After 48 h in these conditions to allow recovery from transport, fish were acclimated gradually to 27 • C over 7 days, which more closely matches temperatures experienced in the wild without being stressful [35].All procedures were approved by the University of Sydney Animal Ethics Committee (approval number 2021/1897).

Experimental design
We conducted a fully factorial experiment with BPA exposure and feeding frequency as independent factors.Following temperature acclimation, fish were haphazardly assigned to treatments of BPA exposure or no-exposure controls.BPA-exposed fish were dosed with a nominal concentration of 10 μg l − 1 BPA (Sigma Aldrich, Castle Hill, Australia) dissolved in DMSO (Ajax Finechem, Taren Point, Australia) in tank water, while non-exposed control fish were given an equivalent dose of DMSO only.The concentration of DMSO in all tanks was 0.0002 % v/v, well below toxicological limits [36].We verified nominal concentrations of BPA in tank water by mass spectroscopy conducted by the Bioanalytical Mass Spectrometry Facility, Mark Wainwright Analytical Centre (University of New South Wales, Australia).75 % water changes were conducted weekly after which BPA was replenished to maintain tank concentrations.Additionally, we added BPA every 3-4 days between weekly water changes because our mass spectrometry data showed that BPA degraded by ~60 % over one week, which provided the background to our protocol and dosage.
Within each BPA treatment, fish were further subdivided into lowfeeding frequency and high-feeding frequency groups following protocols described in [10].Briefly, high-feeding frequency groups were fed fish flakes (Supervit Tropical, Tadeusz Ogrodnik, Chorsów, Poland) three times per day on 5 days per week, while low-feeding frequency groups were fed once per day on four days per week.We pre-weighed food for each feeding event in both treatments to give a total amount of food per tank that was equivalent to 5 mg per fish in each tank.Both feeding time and feeding day were randomized for all treatments.This design resulted in four experimental treatments (BPA+low-feeding, BPA+high-feeding, control+low-feeding, control+high-feeding) with four replicate tanks per treatment (n = 5 fish per tank), and a final sample size of n = 15-20 fish taken from all replicate tanks per treatment.
Phenotypic responses were measured in three consecutive experimental blocks that each spanned 4 days.We measured a subset of fish from all treatments from all tanks in each block so that block and treatment were not confounded.In each fish, we measured behaviour, resting metabolic rates, active metabolic rates, and sustained swimming speed (U crit ) at 27 • C in that order.To track individuals between measurement days within each block, fish were housed in a 1 l cylindrical perforated basket within their home tank [10].Baskets separated individuals while allowing water flow and visual and olfactory cues to pass between individuals.The difference in time spent under treatment conditions between the start of the first block and end of the third was 15 days as treatments were not staggered.After U crit measurements, fish were euthanised by cervical dislocation following anaesthesia in a solution of Aqui-S (Aqui-S, New Zealand) as per manufacturers instruction, weighed on an electronic balance (Sartorius, Australia) and photographed to determine standard body length using ImageJ [37].We calculated fish condition (K) as where mass is in g and length is in cm [38].Skeletal muscle tissue was dissected, weighed, and frozen immediately in liquid nitrogen and stored at − 80 • C for later biochemical analysis of citrate synthase and lactate dehydrogenase activity.All hands-on experimentation was conducted by one of us (AMR), who was not blinded to the treatments.No animals or datapoints were excluded from the experiment.

Behaviour
We measured the latency to leave a refuge and exploration behaviour in circular white opaque plastic arenas (60 cm diameter) following procedures in [10].Each arena was subdivided into a trapezoidal refuge compartment (20 cm x 10 cm x 10 cm) constructed from black opaque plastic corflute and separated by a gate (12 cm x 10 cm) from the exploration compartment.The exploration compartment contained three submerged pyramid-like structures that were coloured red, blue, or green to provide novel landmarks for the fish.Each structure was 2.5 cm in height with an 8 × 8 cm base.Water depth in the arena was 7.5 cm.For each behavioural assay, fish were placed in the refuge compartment for 10 mins after which time the gate was opened remotely without disturbing the fish.Movement of the fish was then recorded for 10 min (using a Hero 6, GoPro, USA camera at a resolution of 1080×720 pixels and filming at 30 frames per second).From each video, we extracted the time it took the fish to leave the refuge compartment (latency) as a proxy for boldness [39].If fish did not leave the refuge during the behavioural trial, they were assigned a maximum latency of 600 s.
Additionally, we counted the total number of times a fish returned from the light arena to the dark refuge during the total 10 min trial; light-dark tests are a common method to evaluate anxiety in fish [40].Lastly, we extracted the total distance and average speed travelled for the first minute after the fish first left the refuge compartment.Videos were analysed in Tracker Video Analysis software (Version 6.0.8, https://ph yslets.org/tracker/).Not all fish left the refuge compartment so that sample sizes were reduced in behavioural variables measured in the exploration compartment (n = 13-20 fish per treatment).

Oxygen consumption
Resting metabolic rates (RMR) and active metabolic rates (AMR) were measured following established protocols [10].RMR was measured via intermittent flow respirometry by placing fish in clear Perspex cylindrical respirometers (15 mm diameter x 100 mm length; 27 ml volume) attached to a peristaltic pump (i150, iPumps, Tewkesbury, UK) which circulated water through each respirometer.All respirometers contained a sensor spot (Pt3, Loligo Systems, Denmark) attached to the inside at the midline of each respirometer, and O 2 concentration inside the respirometers was measured using a fibre optic cable connected to an O 2 metre (Witrox, Loligo Systems, Denmark).Respirometers were housed in a water bath set at 27 • C. Once placed in a respirometer, fish were left undisturbed for 2 h with the pump turned on; we have shown previously that 2 h is sufficient for fish to recover from handling stress [41] and that measurements of RMR are repeatable within individuals using the methodology described here [42].Following the rest period, the pump was turned off remotely without disturbing the fish and the decrease in O 2 was recorded for approximately 20 min.
To measure AMR, individuals were placed in a 130 ml glass respirometer containing a magnetic stir bar separated from the fish by aquarium mesh, and a sensor spot attached to an O 2 metre via fibre optic cable as above.A plastic column attached to the centre of the lid and extending the height of the respirometer helped to reduce turbulence.The respirometer was placed in a temperature-controlled water bath set to 27 • C on top of a magnetic stir plate.The flow speed in the jar was increased by increasing the speed of the stir bar to a point where the fish could no longer hold its position in the respirometer.The speed was then slightly reduced to allow the fish to swim close to maximal [10].The decrease in O 2 consumption was recorded for 10-12 min for each fish.Rates of oxygen consumption (RMR and AMR) were calculated from the slopes of decline in oxygen concentration in the respirometers using the "RespR" package in R [43].Background measurements were taken each day to account for potential microbial respiration, but these were negligible in all cases.We calculated aerobic scope (AS) as the difference between AMR and RMR.

Swimming performance
Sustained swimming performance was measured as the critical sustained swimming speed (U crit ) according to published protocols [44].A cylindrical swimming flume (150 mm length and 26 mm diameter) was tightly fitted over the intake end of a submersible pump (12 V DC, iL500; Rule, Hertfordshire, UK) that drew water through the flume.A plastic grid at the intake end of the pump prevented fish from entering the pump.The pump was connected to a variable DC power supply (NP9615; Manson Engineering Industrial, Hong Kong) and water flow through the flume was regulated by adjusting the voltage input.Water flow was measured in real-time using a flow meter (DigiFlow 6710 M; Savant Electronics, Taichung, Taiwan) attached to the outlet end of the pump.A bundle of hollow straws inserted into the inlet end of the flume helped maintain laminar flow.The flume and pump were submerged in a tank (60 × 40 × 35 cm) filled with aged water at 27 • C (± 0.5 • C).Water temperature was monitored throughout the experiments with a calibrated digital thermometer.Individual fish were introduced into the flume and swam initially at a flow rate of 0.10 m s − 1 for 10 min.After min, the flow rate was increased by increments of 0.06 m s − 1 every 10 min until the fish could no longer hold its position in the water flow and fell back against the plastic grid.The flow was stopped for 10 s to allow the fish to rest briefly, and then increased back to the previous setting.The trial was ended when the fish fell back against the plastic grid for a second time.U crit was calculated as where U f is the highest speed maintained for an entire interval (T i = min), T f is the time until exhaustion at the final speed interval, and U i is the speed increment.U crit is reported as body lengths per second (BL s − 1 ).

Enzyme activities
We measured citrate synthase (CS) and lactate dehydrogenase (LDH) activity in skeletal muscle in duplicate for all individuals using published protocols [45].All reagents were from Sigma Aldrich (Castle Hill, Australia).Muscle tissue, approximately 0.02-0.03g wet mass, was removed from frozen samples and thawed on ice.Thawed tissues were homogenized in 9 vol of extraction buffer (50 mM imidazole, 2 mM MgCl 2 , 5 mM EDTA, 0.10 % Triton X-100, 1 mM glutathione titrated to pH = 7.5) using a Tissuelyser II (Qiagen, Melbourne, Australia).
For CS activity assays tissue homogenate was added to metabolite solution (100 mM Tris-HCl, 0.1 mM acetyl CoA, 0.1 mM DTNB (5,5′dithiobis (2-nitrobencoic acid), pH 8) incubated at 27 • C. A final concentration of 15 mM oxaloacetate was added to each sample to start the reaction.Samples were vortexed and absorbence was read on a spectrophotometer (Ultrospec 2100 Pro: GE Healthcare, Sydney Australia) equipped with a temperature-controlled cuvette holder at 412 nm for min.We conducted control assays omitting addition of oxaloacetate to account for non-CS mediated reduction of DTNB, and absorbence was calculated by subtracting control from sample assays.For measurements of LDH activity, tissue homogenate was diluted to 1:100 and added to the assay solution (100 mM phosphate buffer (KH 2 PO 4 /K 2 PO 4 ), 0.16 mM NADH, 4 mM pyruvate, pH =7.0) starting the reaction.The reaction absorbence was then measured at 340 nm as for CS above.

Statistical analyses
All analyses were conducted in R (Version 4.0.5 "Shake and Throw") and R Studio (Version 1.4.1103)using the "aovp" function in the "lmPerm" package (Wheeler and Torchiano 2016).We chose permutational analyses because they use the data per se for analysis rather than underlying frequency distributions and therefore do not have associated assumptions [46], and we present permutational p-values only [47].For all analyses, we ran identical models that included main effects of BPA exposure and feeding regime, as well as their interaction.We also included a random error of experimental block for all models to account for any differences due to the amount of time fish remained under treatment conditions between blocks.
In addition to reporting p-values, we also calculated effect sizes for main effects as standardised mean differences (Cohen's d) ± 95 % bootstrapped confidence intervals (CIs).Effect sizes provide an estimate of the likelihood of particular responses that cannot be gleaned from pvalue cut-offs, and are therefore helpful in interpreting results [48].We determined 95 % CIs by bootstrapping using the "boot" package in R [49].In interpretations of effect sizes and 95 % CIs both the directionality and degree of overlap with zero are informative in assessing biologically effects, and CIs that minimally or partially overlap with zero indicate the presence of biological effects with relatively high confidence [48].

Increased feeding frequency increased fish mass and condition
Increased feeding frequency increased both fish mass and condition factor (Fig. 1).However, there was no impact of BPA exposure on body condition but BPA reduced mass at a one-tailed significance (Table 1); there was no interaction between feeding frequency and BPA exposure (Table 1).

Behaviour was altered by feeding frequency
BPA exposure did not have a significant effect on any behavioural response (Table 1; Figs. 2 and 3).Latency to leave the shelter was not impacted significantly by feeding (Fig. 2A), but frequently fed fish returned to shelter significantly more often than infrequently fed fish (Fig. 2B).Additionally, frequently fed fish explored a significantly greater distance (Fig. 3A) and at a faster average speed (Fig. 3B) than infrequently fed fish (Table 1).

Resting metabolic rates increased with increased feeding frequency
BPA exposure did not impact any measure of oxygen consumption significantly (Table 1; Fig. 4).Increased feeding frequency significantly increased RMR, but it did not influence AMR nor AS.There was no significant interaction between feeding frequency and BPA exposure for any measure of oxygen consumption (Fig. 4; Table 1).

No impact of feeding frequency or BPA exposure on swimming performance
There was no significant effect of feeding frequency, BPA exposure, or their interaction on U crit (Table 1; Fig. 5).

LDH but not CS activity was decreased by BPA exposure
Feeding frequency did not have a significant impact on enzyme activities, nor was there a significant interaction between feeding frequency and BPA exposure (Table 1).CS activity was not impacted significantly by BPA exposure (Fig. 6A), but BPA exposure significantly reduced LDH activity at a one-tailed significance (Table 1; Fig. 6B).

Effect sizes
Feeding frequency increased fish mass, the number of times a fish returned to refuge, total distance moved and velocity during exploration, and RMR (Fig. 7B).BPA exposure decreased fish mass with a relatively high level of confidence (small overlap of confidence intervals with zero), and it decreased LDH activity (Fig. 7B).

Discussion
Overall, we found little support for our hypothesis that short-term exposure to an environmentally relevant concentration of BPA disrupts whole organism energetics in adult zebrafish, although we did observe that BPA decreased LDH activity indicating there is some level of energetic disruption associated with exposure.BPA tended to reduce body mass, but we found no support for our hypothesis that there is an interaction between BPA exposure and feeding frequency.However, our feeding regimes led to the predicted decreases in fish mass and condition [10], but all fish remained in good condition at the end of the experiment with average condition factors above 1 and below 2. Hence, our manipulation led to positive and negative energy balance in frequently and infrequently fed fish, respectively, while fish remained within a Fig. 1.Fish mass and condition factors.Mass (A) and condition factors (B) are shown for BPA exposed and non-exposed ("Control") fish under high (black bars) and low (grey bars) feeding frequencies.BPA treatment did not have a significant effect but low feeding frequency reduced mass and condition factors (indicated by an asterisk).Data are presented as mean ± s.e. and sample sizes were 15-20 fish per treatment group.

Table 1
Results of permutational analyses.We tested the effects of BPA exposure and feeding frequency and their interaction on size, behavioural responses, oxygen consumption, swimming performance and metabolic enzyme activities.Permutational p-values are shown, and df = 1, 67 in all cases.healthy weight range [50,51].Decreased food availability can decrease investment into growth, somatic maintenance, and energy expenditure [2,52].Periods of resource limitation are common for most species, and physiological responses to food scarcity can reflect adaptative mechanisms that produce compensatory responses to even relatively mild energy limitation [53,54].Responses to the energetic environment and feeding are meditated by endocrine mechanisms [22,24,55] so that BPA and other endocrine-disrupting compounds have the potential to disrupt energetics and feeding.However, the lack of interactions between feeding regime and BPA exposure in our experiment indicates that adult fish are not susceptible to BPA-mediated disruption of responses to energy limitation.Nonetheless, we did observe a decrease in body mass with BPA exposure with relatively high confidence.
A possible pathway leading to decreased mass resulting from BPA exposure is via disruption of energy availability and glucose homoeostasis.BPA can interfere with pancreatic ß-cell function and glucose physiology in rats, altering insulin and blood glucose levels.Consequently, glucose transport into tissues can be disrupted leading to a suppressive effect on skeletal muscle function [56,57].We observed a decrease in LDH activity indicating decreased anaerobic metabolism and glucose processing in the skeletal muscle of BPA-exposed fish.These data confirm previous findings that BPA exposure decreases LDH activity in a temperature-dependent manner in zebrafish [58].Additionally, the BPA analogue bisphenol S (BPS) decreased glycolysis in zebrafish muscles at low concentrations in a sex-specific manner, thereby limiting the total amount of substrates available to LDH [59].
LDH is an important regulatory target for insulin and cortisol [60].At least cortisol can be disrupted by BPA exposure [30], which represents a potential pathway leading to altered LDH activity [60,61].LDH catalyses the reversible reaction converting lactate and NAD + to pyruvate and NADH; it thereby influences mitochondrial bioenergetics by altering substrate concentrations (pyruvate and NADH) and its activity is crucial for lactate metabolism [61,62].Lactate is a metabolic intermediary with Fig. 2. Boldness and refuge seeking behaviour.Latency to leave a shelter (A) the number of times fish returned to the shelter (B) are shown for BPA exposed and nonexposed ("Control") fish at high (black bars) and low (grey bars) feeding frequencies.BPA treatment did not have a significant effect but low feeding frequency reduced the number of times fish returned to the shelter (indicated by an asterisk).Data are presented as mean ± s.e. and sample sizes were 13-20 fish per treatment group.Fig. 3. Exploration of a novel arena.Total distance moved (A) and the average speed (B) decreased with low feeding frequency (grey bars) compared to high feeding frequency (black bars; indicated by an asterisk).BPA treatment (BPA Exposed and non-exposed Control) did not have a significant effect.Data are presented as mean ± s.e. and sample sizes were 13-20 fish per treatment group.a broad range of functions including controlling expression of genes involved in mitochondrial function, and activating AMP-activated protein kinase activity and thereby a suite of metabolic regulatory pathways [63,64].Lactate is also an intermediary in the release of energy from fish white muscle to fuel locomotion and restoration of glycogen stores following exercise [65].There was a reduction in LDH activity following food deprivation in cod (Gadus moruha), which was associated with reduced swimming performance [66].We did not observe these responses in zebrafish, and the decrease in LDH activity was not associated with a reduction in swimming performance.We also did not observe a significant effect of BPA exposure on resting or maximal rates of oxygen consumption suggesting that BPA exposure and the reduction in LDH activity did not lead to a shift toward aerobic metabolism.It is possible that swimming performance is not constrained by ATP supply, but is primarily determined by Ca 2+ cycling dynamics and muscle composition [67].Additionally, whole animal oxygen consumption does not necessarily scale with ATP production because of varying mitochondrial efficiencies [9,10].In contrast, BPS decreased swimming performance in zebrafish indicating that different bisphenols have different physiological consequences [58,68].
There was a pronounced decrease in resting metabolic rate in infrequently fed fish.A decrease in resting oxygen consumption following energy limitation is likely to be advantageous to conserve limited energy supply and stores [2,52].Increased access to food is often associated with an increase in resting rates of oxygen consumption as a result of specific dynamic action [69].Frequently fed fish were likely to have experience greater specific dynamic action resulting in increased oxygen demand.However, there was no effect of feeding frequency on citrate synthase activity, active metabolic rates (AMR), or aerobic scope (AS) indicating that food supply did not affect maximal aerobic capacity and exercise under our experimental conditions.Our treatments were relatively benign and rather than causing starvation shifted energy balance to positive or negative.Hence, even infrequently fed fish may have sufficient access to stored glycogen and lipids to maintain AMR [70], and reduced RMR would have helped to maintain AS.Increased lipid metabolism using stored reserves during even slight fasting conditions can compensate for decreased food availability in changing environments [53].
Resource availability and feeding frequency can alter brain and neuroendocrine function, potentially leading to behavioural differences between fish under different feeding regimes [24].Numerous hormones including thyroid, glucocorticoid, insulin and insulin-like growth factors, growth hormones, and various peptide hormones play regulatory roles in fish behaviours related to feeding, digestion, and energy use [71,72].Energy status in turn influences the willingness of individuals to take risk, and fish in negative energy balance are more likely to take risks and forage in novel environments [73].Individuals with relatively high resting metabolic rates, in particular, show increased boldness (risk taking) when food availability decreases [14,73].On the other hand, low feeding frequency may decrease foraging and exploration as a result of energy constraints.
In our study, feeding frequency did not affect latency to leave refuge, a measure of boldness.In contrast, increased feeding frequency increased the number of times fish returned to shelter, exploration speed, and the distance explored.Increased feeding frequency therefore had an overall positive effect on voluntary exploration in a novel environment, maybe because sufficient energy reserves allowed unrestrained exploration of a new environment.However, increased Fig. 4. Whole animal oxygen consumption.Resting rates of oxygen consumption (MO 2 ; A) decreased with low feeding frequency (grey bars) compared to high feeding frequency (black bars; indicated by an asterisk), but feeding frequency did not affect active rates of oxygen consumption (B) or aerobic scope (C).BPA treatment (BPA Exposed and non-exposed Control) did not have a significant effect on any measure of oxygen consumption.Data are presented as mean ± s.e. and sample sizes were 15-20 fish per treatment group.exploration was accompanied by greater anxiety (frequency to return to shelter).Increased anxiolytic behaviour reflects greater fear of predators [16,40] and represents a trade-off to the increased foraging (exploration) behaviour.Conversely, infrequently fed fish returned to shelter less frequently indicating that these fish prioritized foraging potential over predation risk [15].However, infrequently fed fish swam shorter distances at lower speeds indicating reduced exploration, which may balance the energetic cost of movement with foraging success.
The lack of effect of BPA on behavioural traits is somewhat unexpected.A lower dose of BPA (1.5 μg l − 1 ) led to reduced exploration of a novel arena in adult zebrafish [32].BPA and other bisphenols have non-monotonic dose response curves [74] and it is possible that the effects of BPA on behaviour are relatively greater at very low doses.Environmental exposure to BPA is associated with increases in anxiety-related behaviors and hyperactivity [26,32,75].However, the effects of BPA on physiology and behaviour are most pronounced when individuals are exposed early in development [27].Adult fish are less sensitive to sublethal pollutant exposures because the neural and physiological pathways that are often targets for disruption are already formed [75].There remains a gap in our understanding of the functional consequences of endocrine disruption in wildlife especially regarding how EDCs interact with other anthropogenic and ecological factors at different developmental stages.Overall, our results show a lack of interaction between resource limitation and BPA exposure in adult fish.However, both factors in our experiment did independently impact different aspects of physiology showing that relatively short exposure to mild resource limitation alters physiology and behaviour, and that BPA disrupts some aspects of metabolism.It is worth noting that ectothermic physiology and the action of bisphenols are temperature-dependent [58,76,77].The increasing pollution with EDCs under global warming requires interpretation of functional consequences of endocrine disruption in the context of ecologically relevant factors such as temperature as well as diet and ontogeny.

Fig. 5 .
Fig. 5. Locomotor performance.Critical sustained swimming speed (U crit ) was not affected significantly by feeding frequency (high frequency = black bars; low frequency = grey bars), or by BOPA treatment (BPA Exposed and nonexposed Control).Data are presented as means ± s.e. and sample sizes were 15-20 fish per treatment group.

Fig. 6 .
Fig. 6.Metabolic enzyme activities.Citrate synthase (CS; A) and lactate dehydrogenase (LDH; B) were not affected significantly by feeding frequency (high frequency = black bars; low frequency = grey bars) but BPA exposure (BPA Exposed and non-exposed Control are shown) decreased LDH activity at one-tailed significance (indicated by #) but did not affect CS activity significantly.Data are presented as mean ± s.e. and sample sizes were 15-20 fish per treatment group.

Fig. 7 .
Fig. 7. Effect sizes.Effect sizes (standardised mean difference ± bootstrapped 95 % CI) for the main effects of high-relative to low feeding frequency (open circles; A) and BPA exposure relative to control (filled circles; B).Variable names on the y-axis are as in Figs.1-6.