Research articlesEndocrine disruption from plastic pollution and warming interact to increase the energetic cost of growth in a fish
Abstract
Energetic cost of growth determines how much food-derived energy is needed to produce a given amount of new biomass and thereby influences energy transduction between trophic levels. Growth and development are regulated by hormones and are therefore sensitive to changes in temperature and environmental endocrine disruption. Here, we show that the endocrine disruptor bisphenol A (BPA) at an environmentally relevant concentration (10 µgl−1) decreased fish (Danio rerio) size at 30°C water temperature. Under the same conditions, it significantly increased metabolic rates and the energetic cost of growth across development. By contrast, BPA decreased the cost of growth at cooler temperatures (24°C). BPA-mediated changes in cost of growth were not associated with mitochondrial efficiency (P/O ratios (i.e. adenosine diphosphate (ADP) used/oxygen consumed) and respiratory control ratios) although BPA did increase mitochondrial proton leak. In females, BPA decreased age at maturity at 24°C but increased it at 30°C, and it decreased the gonadosomatic index suggesting reduced investment into reproduction. Our data reveal a potentially serious emerging problem: increasing water temperatures resulting from climate warming together with endocrine disruption from plastic pollution can impact animal growth efficiency, and hence the dynamics and resilience of animal populations and the services these provide.
1. Introduction
Animal growth is an essential component determining population dynamics [1]. Individual growth is achieved by converting energy obtained from the environment to new biomass [2]. The energy required to synthesize a unit of biomass (cost of growth) determines the efficiency of energy transfer between trophic levels [3,4]. Any environmental parameter that alters the cost of growth can therefore have important ramifications for ecosystems and the services these provide. Fisheries and aquaculture, for example, rely on energy conversion from lower trophic levels into growth of target species [5].
Somatic growth is centrally regulated by hormones, such as growth and thyroid hormones [6]. In ectotherms, hormone action is modulated by changes in temperature, and in fish changes in water temperature can have pronounced effects on hormone-mediated processes including growth [7,8]. Alarmingly, human activity has now also released endocrine (hormone)-disrupting chemicals (EDCs) into the environment on a massive scale [9]. Plastic production and plastic waste release biologically active chemicals such as bisphenols into the environment [10]. Exposure to EDCs early in life is of particular concern, because early developmental stages are most sensitive to endocrine disruption [11], and disruption at an early life stage may affect growth trajectories and biomass production [12,13].
Bisphenol A (BPA) is one of the most abundantly produced industrial chemicals worldwide [9], and we therefore focused on it for our experiments. BPA binds to nuclear receptors, including thyroid and glucocorticoid receptors [14,15], and can thereby disrupt hormone signalling important in regulating metabolism and growth. For example, thyroid hormones can interact with the growth axis in vertebrates including fish and induce growth hormone and insulin-like growth factors [6]. Elevated thyroid and glucocorticoid levels increase energy (adenosine triphosphate, ATP) availability in cells to meet changing environmental demands via genomic and peripheral action [6,16,17]. In addition to direct effects on metabolism and growth, these hormones also influence the efficiency of energy (ATP) production in mitochondria by altering proton leak relative to substrate oxidation [18,19]. Decreased efficiency of mitochondria to convert food-derived energy to ATP can constrain growth [20] and thereby provides another potential pathway by which EDCs disrupt growth. Additionally, an increase in temperature can reduce mitochondrial efficiency by increasing the amount of oxygen used per ATP produced (ATP/O ratio) [21] and increase the energetic cost of growth [22]. Hence, the energy transfer between trophic levels may decrease at higher temperatures. EDCs can further modify these relationships by their effects on hormones that regulate mitochondrial function [18,23,24], and because their effects can be temperature specific. For example, the impacts of bisphenols (bisphenol A, bisphenol F, bisphenol S) on physiological performance of zebrafish (Danio rerio) were temperature dependent, and bisphenols modified the effects of acute temperature changes and thermal acclimation on metabolic and muscle function, albeit in a trait-specific manner [25,26]. BPA exposure decreased swimming performance and citrate synthase activity in 18°C-acclimated zebrafish but had no effect or increased it in 28°C-acclimated fish, respectively [25,26].
Chemical pollution and climate warming represent a novel, anthropogenic environmental context that is out of step with evolved responses to natural temperature variation. The direction of the interactions between novel combinations of multiple drivers can be nonlinear and unexpected and thereby exert novel selection pressures and disrupt populations [27]. Hence, the combination between the release of EDCs and climate warming represents an emerging problem that could disrupt animal growth. Here, we aimed to determine whether this is the case by using a model species (zebrafish, D. rerio) as a proof-of-concept to test whether the interaction between temperature and BPA alters how energy is invested towards growth (cost of growth). We hypothesized that (i) exposure to BPA will increase the cost of growth, and its effects will be disproportionately greater at higher temperatures and (ii) mitochondrial efficiency will be lower in the presence of BPA, particularly at higher temperatures, which potentially can explain changes in cost of growth.
2. Materials and methods
(a) Study animals and treatments
We used zebrafish as a model because the species is well suited for laboratory experiments; it has a relatively short developmental period, and breeding protocols are well established [28]. We bred experimental fish from a breeding population of 75 adult zebrafish (D. rerio) distributed across five tanks (8 l each, 15 individuals of mixed sex per tank) maintained at 27°C water temperature. Female zebrafish release eggs into the water column where they are fertilized externally. The night before egg collection, we placed a grid that covered the floor of each tank and permitted passage of eggs but prevented access to the eggs by adults. Eggs were collected the next morning, and eggs from all tanks were mixed together to ensure sufficient genetic diversity of experimental fish. Eggs were placed for 24 h into Petri dishes containing E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, 0.0003% methylene blue, 0.01% penicillin–streptomycin) at pH 7.6 (20 eggs per Petri dish, three replicate dishes per treatment). The eggs were allocated to four rearing treatments with temperature (24°C and 30°C, which we refer to as cold and warm, respectively) and BPA exposure (0 and 10 µg l−1, which we refer to as control and BPA, respectively) in a factorial design (24°C/control; 24°C/BPA; 30°C/control, and 30°C/BPA). The cold and warm treatments were within the temperatures experienced by zebrafish in their native habitats, which range from 17°C to 34°C [29], and eggs develop normally between 24°C and 30°C [30]. The exposed group was given 10 µg l−1 of BPA (Sigma Aldrich, Castle Hill, Australia), dissolved in DMSO (final content in tank = 0.0005%), and the control treatment was given a vehicle control of DMSO only (0.0005% final content); 10 µg l−1 is an ecologically relevant concentration of BPA that is commonly found in natural waterways, and fish take up BPA via their gills [31,32]. Experimental fish were exposed to the different treatments from eggs until adults were euthanized at 140 days post fertilization (dpf; see below).
We determined mortality at 24 h post fertilization, and survival within the first 24 h was 75–95% in all treatments (electronic supplementary material, table S1), which is usual for laboratory-reared zebrafish [33]. Surviving eggs were placed in 8 l tanks with filtered aged water at their experimental temperature and BPA treatment conditions (dispersed across three tanks per treatment). After hatching, larvae were fed cultures of Paramecium sp. twice daily. After 11 and 9 dpf for the 24°C and 30°C treatments, respectively, larvae were fed until satiated twice daily with fish flakes (Tetramin, VA, USA), supplemented with live Artemia salina nauplii every 3–4 days. Water temperature was monitored (every 30 min) with water temperature loggers (HOBO MX2201, OneTemp Pty Ltd, Adelaide, Australia) throughout the rearing period and was maintained within ± 0.5°C of the desired rearing temperatures. The light cycle was 14 h light : 10 h dark. We checked water quality every second day and performed a 50% water change once a week (starting at 14 dpf) for the first month, after which we performed 80% water changes twice per week because of greater waste production from larger fish. We replenished BPA and the vehicle control after each water change to maintain the desired nominal concentrations. Note that the half-life of BPA in water is 38 days [31], so that treatment concentrations would have remained relatively stable in between water changes. We acknowledge that if there was degradation of BPA in between water changes, the actual concentration in the tank water would have been lower than 10 µg l−1, which would make our results more conservative.
Before the growth experiment, we tested the repeatability of oxygen consumption (MO2) measures within individuals in a group of zebrafish not used elsewhere in the experiment. We showed that MO2 is a repeatable, intrinsic trait that is suitable to test the effects of experimental treatments (electronic supplementary material, figure S1).
(b) Growth and morphometrics
After 28 dpf, we started weekly measurement of MO2, standard length and body mass. To track individuals throughout the growth experiment and to conduct repeated measures, we placed individuals into 1 l cylindrical baskets that permitted water flow, and hence chemical and visual contact between fish, but prevented egress of individuals [34]. Within treatments, there were six baskets per tank in each of three tanks per treatment [34] to give a total n = 18 fish per treatment, and in the statistical analysis, we used ‘individual’ as the unit of replication.
Standard length was measured in ImageJ [35] from photographs taken of the fish in a transparent container (64 × 40 × 16 mm) against a white background. Body mass was measured on an electronic balance (Sartorius, Australia) by placing fish into a weighing dish with sufficient water to submerge the fish; the dish plus water was tared before the fish was introduced, and excess water was removed from the fish before weighing. While measuring fish each week, we also noted onset of sexual maturity. Age at maturity (around 90 dpf) was determined following morphological indicators described in [36].
At 140 dpf, fish were euthanized to measure mitochondrial bioenergetics (see below). While dissecting muscle for mitochondrial analyses, we also removed livers from all fish and egg masses from females. We counted eggs as an estimate of reproductive investment and weighed liver and eggs to determine hepatosomatic index (HSI) and gonadosomatic index (GSI) as additional measures of growth patterns [37]. Note that we could not distinguish between sexes at the start of the experiment, so that all experimental fish were of mixed sex, and samples sizes for males and females were not equal. At 140 dpf, the sex ratios for the different treatments were (male/female): 24°C control = 1.4; 24°C BPA = 1.3; 30°C control = 1.1; 30°C BPA = 3.0, and total sample sizes were 17, 18, 15, 17 fish for the different treatments, respectively. As a result of the heavily skewed sex ratio, the sample size (n = 4) of females in the 30°C BPA-exposed treatment is particularly low, and these results should be viewed as preliminary.
(c) Oxygen consumption
We used intermittent-flow respirometry to measure whole animal resting rate of oxygen consumption (MO2). For each individual, resting MO2 and growth rates were measured weekly between 28–98 dpf and also at 126 dpf. Each fish was fasted for 24 h before measurements and placed into a cylindrical, clear Perspex respirometer (27 ml volume) submerged in a tank (30 × 25 × 25 cm). Respirometers were connected to a peristaltic pump (i150, iPumps, Palmington, UK, and BT100–1 L, Longer Precision Pump Co., Hebei, China) to provide flow and supply oxygen when switched on, and pumps could be turned on and off remotely without disturbing the fish. Oxygen sensor spots (Loligo Systems, Tjele, Denmark) were fixed inside the respirometers at their midpoint. A fibre-optic cable was attached with a Velcro strap (Loligo Systems) to the outside of the respirometers and connected to an oxygen meter (Witrox 4, Loligo Systems) to record the O2 concentration inside the respirometers. After at least 2 h of undisrupted rest in the chambers, which is sufficient to recover from handling stress [38], the pump was turned off remotely, and the decrease in oxygen concentration was recorded on WitroxView 1.0.2 (Loligo Systems) until there was a steady decline in concentration (approx. 30 min of recording). The background respiration was recorded in an empty chamber simultaneously in every trial to adjust for microbial respiration. Water temperature was maintained (±0.05°C) by a temperature-controlled bath circulator (BL-30, Thermoline Scientific, Australia) during the measurement period. Resting MO2 was measured at the treatment temperature of each fish and calculated from the slope in the decline (approx. 5 min of stable recording) in O2 concentration in the respirometers during the closed phase in the R package respR [39]. Oxygen concentrations inside the respirometers did not drop below 85% saturation in any trial. After measurements of MO2, body mass (g) and standard length (cm) were measured.
(d) Mitochondrial bioenergetics
We measured mitochondrial bioenergetics at the end of the growth experiment (140 dpf) to determine the effects of temperature and BPA on mitochondrial bioenergetics and efficiency. Sample sizes were 18 fish for the 24°C BPA treatment group, 17 fish each for 24°C control and 30°C BPA treatments and 14 fish for the 30°C control group. Fish were euthanized via cervical dislocation. Skeletal muscle was dissected and homogenized in a Potter–Elvehjem glass tissue homogenizer with nine volumes of homogenization buffer (140 mM KCl, 20 mM HEPES, 5 mM MgCl2, 2 mM EGTA, 1 mM ATP, 0.5 g l−1 BSA-FAF, pH 7.0) and centrifuged at 1400 r.p.m. for 5 min. The supernatant containing mitochondria was spun at 9000 r.p.m. for 9 min at 4°C. The pellet was resuspended in a respiration buffer (110 mM sucrose, 60 mM KCl, 0.5 EGTA, 3 mM MgCl2, 20 mM taurine, 10 mM KH2PO4, 20 mM HEPES, 0.5 g l−1 BSA-FAF, pH 7.1) at a dilution of 200 µl buffer per 0.1 g of wet tissue.
Measurements were conducted at the experimental rearing temperature of each individual in a respiration chamber (Mitocell MT200, Strathkelvin Instruments, North Lancashire, UK) with a microelectrode (model 1302, Strathkelvin Instruments) connected to an oxygen meter (model 782, Strathkelvin Instruments) as described previously [40]. We used malate (final concentration in respirometry chamber 5 mM), pyruvate (2.5 mM) and succinate (10 mM) as substrates for mitochondrial measurements. Maximal substrate oxidation rates (state 3 respiration) were induced by the addition of ADP (0.1 mM final concentration). Natural uncoupled (state 4) respiration rate was determined when rates of MO2 stabilized after the addition of ADP. State 4 rates were confirmed with the addition of 2–4 µg ml−1 oligomycin, which blocks the ATP synthase, so that any oxygen used is due to proton leak. Finally, we added 0.5 µM p-trifluoromethoxy carbonyl cyanide phenyl hydrazone (FCCP) to verify the integrity of the mitochondrial membrane. All rates were normalized to the concentration of mitochondrial protein in the chamber determined via Bradford assay (Sigma Aldrich, Castle Hill, Australia) with BSA as the standard. We estimated mitochondrial ATP production efficiency by two standard measures, the respiratory control ratio (RCR) and the P/O ratio [41]. The RCR expresses maximal substrate oxidation rates as a multiple of uncoupled respiration rates, and the P/O ratio estimated the amount of oxygen used for each ADP converted to ATP. Hence, RCRs were calculated as the ratio between state 3 and 4 rates, and P/O ratios were calculated as the amount of oxygen used to phosphorylate the added ADP as described previously [40]. Note that calculations of P/O ratios assumed that all added ADP (0.1 mM) was phosphorylated.
(e) Energetic cost of growth
We estimated the energy expended to synthesize a unit of biomass or energetic cost of growth (Em), from theoretical relationships [22]. Energetic cost of growth can be calculated as,
(f) Statistical analysis
We analysed all data with permutational analyses in the R package lmPerm [44]. Permutational analyses do not make assumptions about underlying distributions but use the data per se to infer significant differences [45]. Hence, permutational p-values are not associated with any particular theoretical distribution and may be referred to as ‘pseudo-p’-values [46,47]. We conducted fully factorial analyses, retaining all interactions, because these were relevant to test our hypotheses.
In analyses of mass, length, MO2 and cost of growth, we used temperature (24°C and 30°C), BPA exposure (exposed and control) and day post fertilization as independent factors. We included individual fish ID as a random factor to account for repeated measures of the same individual in different weeks. As post hoc analysis of three-way interactions between temperature, BPA exposure and age, we calculated effect sizes as Cohen's d (meanBPA−meanControl/s.d.) [48] to resolve the effects of BPA on the different traits at 24 and 30°C. We determined bootstrap 95% CIs [49] of effect sizes in the ‘boot’ package in R. All measures taken at the end of the experiment (140 dpf; age at maturity, HSI, GSI, number of eggs, and mitochondrial bioenergetics) were analysed with temperature and BPA exposure as factors and sex as an additional factor where relevant (age at maturity, HSI). Sex was not significant for any of the mitochondrial measurements, so that we present analyses with temperature and BPA treatment as independent factors only. GSI and number of eggs were analysed in females only.
(g) Global temperature, plastic waste, and fisheries
We extrapolated our finding to predict potential global ‘hotspots’ where fisheries may be impacted by both warming water temperatures and high levels of plastic pollution. We extracted (from Bio-ORACLE v. 2.1 [50]) average annual sea surface temperatures (SSTs; °C) for the current climate (2000–2014) and for two future climate predictions assuming no major climate change mitigations (RCP 8.5 for 2040–2050 and 2090–2100). We also extracted the annual coastal plastic pollution as mass of river plastic flowing into oceans (metric tonnes per year; t yr−1) from [51], since plastic waste in aquatic systems is positively correlated with BPA pollution [11]. Spatial data layers for SST and plastic pollution were superimposed at 1 decimal degree resolution and arbitrarily categorized into four risk groups: (i) high risk, which are areas with SST greater than 30°C and greater than 100 t yr−1 of coastal plastic pollution, (ii) medium risk, which are areas with SST 25–30°C and greater than 1 t yr−1 of coastal plastic pollution, (iii) low risk, which are areas with less than 1 t yr−1 of coastal plastic pollution and (iv) no risk in areas with no plastic pollution (0 t yr−1). The ‘risk’ layer was overlaid on the 2015 global marine fisheries catch rate (metric tonne per square kilometre per year; t km−2 yr−1; [52]), which includes illegal and unreported fishing.
3. Results
(a) Growth
Growth in length (electronic supplementary material, figure S2) and mass (figure 1a,d; electronic supplementary material, figure S3 shows individual trajectories) were determined by a three-way interaction between temperature × BPA treatment × age (days post fertilization) (table 1). Fish exposed to BPA at high temperatures (30°C) were smaller than control fish, while fish exposed to BPA at 24°C were larger than controls (figure 1g). Differences in mass increased with increasing age.
Figure 1. Growth, metabolic rates and cost of growth. All responses were modified by a significant three-way interaction between temperature (a–c 24°C [blue symbols] and d–f 30°C [red symbols]), age (days post fertilization) and BPA exposure (open symbols = control, filled symbols = 10 µg l−1 BPA). Gompertz models were fit to growth data (a,d; solid line = control; broken line = BPA exposure). Metabolic rates (b,e) were normalized to mass, and energetic cost of growth (Em; c,f) was estimated from theoretical relationships taking both changes in mass and metabolic rates into account; see text for more information. Means ± s.e. are shown in a–f, and n = 15–18 fish per treatment; lines in b,c,e,f show trends only. Statistical results are summarized below the legend, and post hoc comparisons of the effects of BPA (g) are shown as effect sizes (±95% CIs) of BPA on each of the individual traits (mass, length, MO2, cost of growth (CoG)) at 24°C (blue inverted triangles) and 30°C (red triangles). (Online version is in colour.)
| source | mass | length | MO2 | Em |
|---|---|---|---|---|
| age | <0.0001 | <0.0001 | <0.0001 | <0.0001 |
| temp | 0.034 | <0.0001 | <0.0001 | <0.0001 |
| TRT | 0.64 | 0.90 | 0.93 | 0.96 |
| age × temp | <0.0001 | <0.0001 | 0.0054 | <0.0001 |
| age × TRT | 0.42 | 0.11 | <0.0001 | 0.46 |
| temp × TRT | 0.17 | 0.10 | <0.0001 | 0.0046 |
| age × temp × TRT | <0.0001 | <0.0001 | <0.0001 | <0.0001 |
(b) Metabolic rates and cost of growth
Similar to growth, mass-specific rates of MO2 (figure 1b,e; electronic supplementary material, figure S4) and estimated energetic cost of growth (figure 1c,f) were determined by a three-way interaction between temperature, BPA treatment and age (table 1). MO2 and cost of growth were higher in BPA-exposed fish compared to controls at 30°C, but BPA-exposed fish had lower MO2 and cost of growth than controls at 24°C (figure 1g). Differences in MO2 and cost of growth between treatments decreased with increasing age.
(c) Mitochondrial bioenergetics
State 3 maximal substrate oxidation rates did not change with temperature or BPA treatments (figure 2a; table 2). However, uncoupled state 4 rates (figure 2b) were higher at 30°C, but BPA treatment decreased S4 rates at 30°C while increasing them at 24°C (temperature × treatment interaction; table 2). Both RCRs (RCR; figure 2c) and P/O ratios (figure 2d) indicate that mitochondria were less efficient at 30°C (main effect of temperature; table 2). An interaction between BPA treatment and temperature at a one-tailed significance hints that BPA decreased RCR at 24°C but increased RCR at 30°C (table 2).
Figure 2. Mitochondrial bioenergetics. Maximal substrate oxidation state 3 rate (a) did not change significantly in response to temperature or BPA treatment (grey bars = 10 μg l−1 BPA; open bars = control). An interaction between temperature and BPA exposure determined uncoupled state 4 rates (b), and both indicators of efficiency, respiratory control ratios (RCR; c) and the ratio of ADP used per oxygen consumed (P/O ratios) (d) indicate that mitochondria were less efficient at 30°C (main effect of temperature). Means ± s.e. are shown, and n = 14–18 fish per treatment. Significant interaction is shown underneath the panel label, and bars with different lettering indicate significant main effects. (Online version in colour.)
| source | mitochondria |
morphometrics |
||||||
|---|---|---|---|---|---|---|---|---|
| S3 | S4 | RCR | P/O | matur. | HSI | GSI | egg# | |
| temp | 0.45 | 0.014 | 0.013 | 0.0016 | <0.0001 | 0.025 | 0.21 | 0.010 |
| TRT | 0.99 | 0.98 | 0.48 | 0.98 | 0.71 | 0.64 | 0.037 | 0.16 |
| sex | 0.020 | <0.0001 | ||||||
| temp × TRT | 0.98 | 0.049 | 0.062 | 0.62 | 0.030 | 0.43 | 0.99 | 0.093 |
| temp × sex | 0.94 | 0.88 | ||||||
| TRT × sex | 0.98 | 0.75 | ||||||
| temp × TRT × sex | 0.024 | 0.11 | ||||||
(d) Morphometrics
A three-way interaction between temperature × BPA treatment × sex significantly affected age at maturity (figure 3a; table 2). In males, age at maturity was lower at 30°C compared to 24°C. In females, BPA exposure decreased age at maturity at 24°C but increased it at 30°C. HSI was lower in males than in females (main effect of sex) and lower at 30°C compared to 24°C (main effect of temperature; figure 3b; table 2). The GSI in females was lower in BPA-exposed fish compared to control fish (main effect of BPA treatment; figure 3c; table 2), and the number of eggs per female was lower at 30°C compared to 24°C (main effect of temperature; figure 3d; table 2).
Figure 3. Morphometric effects. Age at maturity (a) was influenced by a significant three-way interaction between temperature (24°C = blue symbols; 30°C = red symbols) × BPA treatment (control = open symbols, BPA = filled symbols) × sex (females = left panel, males = right panel). Hepatosomatic index (b) was determined by main effects of temperature and sex, while gonadosomatic index of females (c) was affected significantly by BPA treatment. Total number of eggs per female (d) was lower at 30°C than at 24°C. (Online version in colour.)
4. Discussion
The most striking result from this study was the discovery that energy use and cost of growth during development increased with BPA exposure at elevated temperatures. These environmental conditions—increased pollution and warming—are dominant global trends currently. Each in its own right can have pronounced impacts on ecosystems, and both have been identified as among the most important conservation risks that require global cooperation to prevent potentially severe impacts on natural systems [53,54]. Our discovery that these drivers can interact to produce novel impacts on vertebrates escalates the problem considerably.
Increased resting metabolic rate per se can have negative fitness consequences for individual ectotherms. In a modelling study analysing metabolic cost of growth in the context of Darwinian selection, fitness of individuals decreased as resting energy expenditure increased [55]. Applying the same logic, it may be predicted that the observed BPA and temperature-induced increases in energy expenditure should also lead to a decrease in fitness as a result of an increase in metabolic cost at least during growth. In our zebrafish, these negative consequences would be exacerbated in females by a BPA-mediated increase in age at maturity at warm temperatures, reduced investment into reproductive structures and a reduction in the number of eggs at warm temperatures. Together with smaller asymptotic sizes, these negative effects on individual growth and fitness have the potential to reduce population sizes [56–58].
Beyond individual fitness, our results have important ecological ramifications because increased cost of growth can reduce energy transfer efficiency between trophic levels [3] and thereby change biomass distributions across different trophic levels [4]. The efficiency of energy transfer from resources to consumers determines the biomass structure of food webs. Transfer efficiency denotes the proportion of resource production converted into consumer production [59]. On average, transfer efficiency is roughly 10% in marine systems, and even small changes in efficiency at a particular node can be amplified across trophic cascades to affect biomass at high trophic levels [59,60]. Climate change-induced warming has now decreased biomass across all trophic levels, but the decline is particularly pronounced at upper trophic levels, which includes most commercial fish species [61].
Transfer efficiencies can be modified at different levels of cellular organization. At the level of the individual, a proportion of ingested energy is assimilated and converted metabolically to new biomass (anabolism) or energy (ATP) production (catabolism) [2,62]. However, the conversion between different forms of energy is inefficient. For example, up to 30% of mitochondrial substrate oxidation at rest can be lost to proton leak rather than being converted to ATP [63]. ATP may be directed towards growth by supplying the energy necessary to build complex macromolecules from monomer precursors, such as proteins and nucleic acids [2,62]. Again, there are energetic inefficiencies in assembling macromolecules [64] that can contribute to the overall cost of growth. Our data show that increasing temperatures decreased the efficiency of mitochondrial ATP production, which at least partly explains the higher cost of growth at elevated temperatures. BPA modified mitochondrial proton leak in our zebrafish, but not in a way that could explain the higher metabolic rates and cost of growth when BPA exposure coincided with increased temperatures. Mitochondrial efficiency is therefore unlikely to be the principal cause explaining the increased cost of growth resulting from the temperature–BPA interaction. BPA and temperature modify numerous endocrine systems, including those associated with metabolism and growth, such as thyroid hormone, growth hormone and glucocorticoids [15,25,65–67]. Regulation of metabolism by thyroid hormone and glucocorticoids can be temperature specific [8,68]. It is possible therefore that the thermally sensitive effects of BPA on metabolism and growth are mediated via these endocrine axes, which it is known to affect [69]. Investigating the mechanisms underlying the thermal dependence of BPA-mediated effects on hormone and growth signalling pathways [70,71] would be a promising future line of investigation.
Anthropomorphic modification of environments can potentially reset the energetics of ecosystems because evolved responses to environmental change are not matched to novel anthropogenic signals. We now show that EDCs interact with temperature to disrupt the allocation of metabolic energy to growth in zebrafish. The amount of plastic produced, and its attendant plastic pollution, is steadily increasing [10,72]. Similarly, human-induced warming is continuing to accelerate, with pronounced effects on aquatic ecosystems [73,74]. EDCs and climate warming are therefore likely to persist into the future. These anthropogenic drivers thereby provide a novel evolutionary context that may render hitherto evolved responses to changing environments ineffective. Interactions between EDCs and temperature like the one we demonstrated here for zebrafish therefore have the potential to modify energy flow through ecosystems and cause a disproportional decrease in biomass at upper trophic levels [4,75].
Commercial fisheries target mainly fish that are positioned at relatively high trophic levels [76]. A reduction in biomass at upper trophic levels may therefore have economic impacts and threaten food security. For example, 71% of the world's coastlines are warming and most are likely to be subjected to high levels of BPA exposure from upstream manufacturing plants and poorly managed waste [11,77]. We modelled risk of warming and likelihood of plastic pollution of global coastal areas superimposed on current fishing intensity (figure 4, electronic supplementary material, figure S5). We predict that tropical coastal areas across multiple exclusive economic zones (electronic supplementary material, figure S6) in Southeast Asia have the highest risk of biomass reduction resulting from warming and pollution. These areas also have some of the most intensive global fisheries. Few coastal areas globally are not currently exposed to the combination of rising SSTs and plastic pollution (figure 4a). Predicted future warming is likely to decrease the relative proportion of low risk areas while increasing medium to high risk areas in the absence of mitigation (figure 4b–d). Our predictions indicate that with current levels of plastic pollution, future warming will increase the risk around the globe and particularly in Southeast Asia, Central America and equatorial west and east Africa (figure 4b,c).
Figure 4. Spatial distribution of coastal areas at risk. Risk was defined as the concurrent increase in sea surface temperature with high levels of plastic pollution under (a) the current climate scenario (2000–2014) and under the RCP 8.5 future projections for the periods of (b) 2040–2050 and (c) 2090–2100. Risk was categorized as ranging from low (blue areas; no plastic pollution) to high (red areas; high temperatures and high levels of pollution). Risk layers are superimposed on fisheries catch rates for the same areas (grey shading in metric km−2 yr−1). (d) Change in risk from the current scenario (2000–2014) to the two future projections (2040–2050 and 2090–2100); data indicate grid cells (resolution 1°) occupying each categorized risk score. See main text for more details. (Online version in colour.)
Our results present a proof-of-concept that the combination between EDCs and increasing temperature can disrupt growth trajectories of fish. We predict that these changes in individual growth can have socio-economic and food security impacts at regional and global scales. These predictions should be verified for other species, particularly those targeted by fisheries or aquaculture to assess the sustainability of fisheries and food production. It is likely, however, that other species of fish are similarly affected, considering that all biochemical rates are sensitive to changes in temperature, and that metabolic regulators and the endocrine systems affected by BPA and other EDCs are highly conserved among fish [6,78,79].
Ethics
All experiments were approved by the University of Sydney Animal Ethics Committee (approval no. 2018/1139).
Data accessibility
All data are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.v6wwpzgxm [80].
Authors' contributions
N.C.W.: investigation, writing—review and editing; A.M.R.: investigation, writing—review and editing; F.S.: conceptualization, formal analysis, funding acquisition, methodology, project administration, resources, supervision, writing—original draft.
All authors gave final approval for publication and agreed to be held accountable for the work performed therein.
Competing interests
We declare we have no competing interests.
Funding
This research was funded by Australian Research Council Discovery Grant no. DP190101168 to F.S.
Acknowledgements
We thank Stephanie Bamford for logistical support.
