Temperature-induced limitations on the capacity of the cardiorespiratory system to transport oxygen from the environment to the tissues, manifested as a reduced aerobic scope (maximum minus standard metabolic rate), have been proposed as the principal determinant of the upper thermal limits of fishes and other water-breathing ectotherms. Consequently, the upper thermal niche boundaries of these animals are expected to be highly sensitive to aquatic hypoxia and other environmental stressors that constrain their cardiorespiratory performance. However, the generality of this dogma has recently been questioned, as some species have been shown to maintain aerobic scope at thermal extremes. Here, we experimentally tested whether reduced oxygen availability due to aquatic hypoxia would decrease the upper thermal limits (i.e. the critical thermal maximum, CTmax) of the estuarine red drum (Sciaenops ocellatus) and the marine lumpfish (Cyclopterus lumpus). In both species, CTmax was independent of oxygen availability over a wide range of oxygen levels despite substantial (>72%) reductions in aerobic scope. These data show that the upper thermal limits of water-breathing ectotherms are not always linked to the capacity for oxygen transport. Consequently, we propose a novel metric for classifying the oxygen dependence of thermal tolerance; the oxygen limit for thermal tolerance (PCTmax), which is the water oxygen tension (PwO2) where an organism's CTmax starts to decline. We suggest that this metric can be used for assessing the oxygen sensitivity of upper thermal limits in water-breathing ectotherms, and the susceptibility of their upper thermal niche boundaries to environmental hypoxia.
Marine ectotherms largely occupy the extent of latitudes tolerable within their thermal limits (Sunday et al., 2012), and are therefore expected to shift their latitudinal distribution ranges poleward with climate warming (Pörtner et al., 2014; Sunday et al., 2012). The critical thermal maximum (CTmax), typically determined as the temperature at which animals exhibit a loss of equilibrium (LOE) (Beitinger et al., 2000), defines the upper limit of a species' fundamental thermal niche and is the temperature where animal function ceases due to the collapse of one or more vital physiological functions (Pörtner, 2010).
Oxygen supply capacity refers to the maximum ability of the cardiorespiratory system to supply oxygen to the tissues by increasing ventilation, cardiac output and blood oxygen carrying capacity. Over the last two decades, laboratory studies have reported reduced heart and ventilation rates, diminished cardiac output and blood oxygen content, and an accumulation of anaerobic metabolites in water-breathing ectotherms exposed to acute temperature increases that approach their upper thermal limits (Frederich and Pörtner, 2000; Mark et al., 2002; Melzner et al., 2006; Wittmann et al., 2008). This led to the hypothesis that fishes and other water-breathing ectotherms, at temperatures approaching their CTmax, are unable to maintain sufficient oxygen supply for basic metabolism as a result of temperature-induced cardiorespiratory constraints. The temperature where animals become reliant on unsustainable anaerobic metabolism for vital physiological functions is termed the critical temperature (Tcrit) (Farrell et al., 2009; Pörtner, 2010). Above Tcrit, extreme hypoxemia develops and the time until terminal ATP deficiency (i.e. CTmax) becomes progressively shorter with increased warming (Pörtner, 2010). In species where the upper thermal limits are determined by insufficient oxygen supply for vital physiological functions, any reduction in oxygen availability should reduce both Tcrit and CTmax. The upper thermal limits of these species can, thus, be described as oxygen dependent.
In fishes, reduced cardiorespiratory performance and oxygen supply capacity at supra-optimal temperatures have been correlated with reductions in population abundance (Johansen et al., 2015; Pörtner and Knust, 2007) and migration performance (Eliason et al., 2011). It has also been suggested that oxygen-dependent upper thermal limits identified in the laboratory represent those in the field (Giomi et al., 2014; Pörtner and Knust, 2007). However, the link between oxygen supply capacity and upper thermal limits has recently been questioned, as other studies have reported that aerobic scope (AS, the difference between the maximum metabolic rate, MMR, and the standard metabolic rate, SMR) and cardiorespiratory performance are maintained in a number of fish and crustacean species experiencing ecologically relevant thermal extremes (Brijs et al., 2015; Clark et al., 2013; Ern et al., 2015, 2014; Gräns et al., 2014; Healy and Schulte, 2012; Jost et al., 2012; Norin et al., 2014). It has, therefore, been suggested that some species possess a more thermally resistant cardiorespiratory system (Ern et al., 2014; Jost et al., 2012), and that insufficient tissue oxygen supply is not the primary determinant of upper thermal limits in all water-breathing ectotherms (Brijs et al., 2015; Wang et al., 2014).
Thermal tolerance can be described as the capacity to maintain the performance of a single physiological function (e.g. heart rate) or a group of related functions (e.g. cardiorespiratory oxygen transport) when temperatures change. In contrast, thermal limit is described as the temperature where the performance of such functions begins to decline, or is reduced to zero. CTmax is a widely used metric for evaluating the thermal tolerance and thermal limits of animals (Terblanche et al., 2011, and references within), but involves an acute increase in temperature, representative of relatively few aquatic environments (e.g. shallow tropical estuaries, tide pools, desert streams). Furthermore, because the performance capacity of physiological functions is influenced by both the rate of temperature change and the duration of exposure to new temperatures, the ecological relevance of such measurements must take into account these two parameters (e.g. Somero, 2010; Terblanche et al., 2011; Farrell, 2016), especially if the rates of change greatly exceed those encountered by the species in the wild. Nonetheless, CTmax and an organism's thermal breadth (CTmax−CTmin, the critical thermal minimum) have been applied widely in both experimental studies and meta-analyses on the impacts of climate warming on aquatic ectotherms, including in predictions of their global (re)distribution (e.g. Sunday et al., 2011; Sunday et al., 2012; Magozzi and Calosi, 2015; Vinagre et al., 2016). Investigating the oxygen dependence of thermal tolerance using CTmax is, therefore, highly relevant for continued research efforts aimed at understanding how climate change will impact ecological physiology and species' distributions. This is especially true as accelerated climate change involves not only increasing (and more extreme) temperatures but also an increase in the frequency and severity of aquatic hypoxia (Altieri and Gedan, 2015; Diaz and Rosenberg, 2008). This latter condition is likely to play a key role in shaping the distribution of species if their upper thermal limits are oxygen dependent.
aerobic scope (MMR−SMR) (mg O2 h−1 kg−1)
critical thermal maximum (°C)
critical thermal minimum (°C)
loss of equilibrium
maximum metabolic rate (mg O2 h−1 kg−1)
oxygen consumption rate (mg O2 h−1 kg−1)
critical oxygen tension (mmHg)
oxygen limit for thermal tolerance (mmHg)
water oxygen tension (mmHg)
routine metabolic rate (mg O2 h−1 kg−1)
standard metabolic rate (mg O2 h−1 kg−1)
critical temperature (°C)
From a physiological perspective, aquatic hypoxia can be defined as any water oxygen tension (PwO2) that reduces MMR and AS, with the critical oxygen tension (Pcrit) characterized as the PwO2 where AS is zero and any further reductions in PwO2 result in a proportional decrease in the rate of oxygen consumption below that required to sustain baseline metabolism (i.e. SMR) (Farrell and Richards, 2009). Thus, in species with oxygen-dependent upper thermal limits, Tcrit should be reduced at any level of hypoxia that causes a significant reduction in MMR and, thus, AS (Fig. 1A,B). Because upper thermal limits of these species are caused by terminal ATP deficiency, CTmax should also decline accordingly. However, in species where the upper thermal limits in normoxia are not determined by insufficient oxygen supply, a portion of AS would remain at CTmax in normoxia (Fig. 1C) and no change in CTmax would be expected upon exposure to progressive hypoxia until the PwO2 where MMR is reduced down to SMR, and AS reaches zero (Fig. 1D). In such a case, the upper thermal limits in normoxia can be described as oxygen independent.
To investigate whether acute upper thermal tolerance (i.e. CTmax) in fishes is determined by insufficient oxygen supply capacity, we actively constrained AS by lowering water oxygen availability (i.e. induced hypoxia), and then measured CTmax at a wide range of PwO2, from normoxia to severe hypoxia. This allowed us to investigate whether, or when, the fish's upper thermal limit (i.e. CTmax) became oxygen dependent. We used two fish species with different thermal niches and ecologies: the athletic, free-swimming, eurythermal red drum (Sciaenops ocellatus; approximate natural temperature range, 15–26°C) and the more sluggish, cold-water lumpfish (Cyclopterus lumpus; approximate natural temperature range, 3–11°C) (Aquamaps.org, 2015).
MATERIALS AND METHODS
Animals and maintenance
Red drum, Sciaenops ocellatus (Linnaeus 1766), were reared at 22±1°C (salinity 35±1 ppt) at the University of Texas at Austin, Marine Science Institute (Port Aransas, TX, USA), whereas lumpfish, Cyclopterus lumpus Linnaeus 1758, were reared at 10±1°C (salinity 32±1 ppt) at the Ocean Sciences Centre, Memorial University of Newfoundland (St John's, Newfoundland, Canada). Both species were fed with pelleted food every day, but were fasted for 24 h prior to experiments to avoid the metabolic effects of digestion on respirometry measurements. Individual fish were used only once in the experiments described below (i.e. no fish were tested more than once).
Experimental procedures were performed in accordance with policies of the Institutional Animal Care and Use Committee of the University of Texas at Austin (red drum) and the Animal Care Committee of Memorial University of Newfoundland (lumpfish; protocol no. 15-88-KG). Studies on lumpfish also followed the guidelines of the Canadian Council on Animal Care.
SMR and Pcrit
SMR and Pcrit were measured at 10 and 16°C for lumpfish and at 24 and 30°C for red drum (N=8 for each temperature and species). For each experimental group, the fish were placed inside respirometry chambers submerged in a tank with a flow-through supply of fully aerated seawater at 10°C for the lumpfish or 24°C for the red drum (the starting temperature for red drum was slightly higher than their acclimation temperature because of a limited capacity for cooling in the experimental setup), and allowed 24 h to settle inside the respirometry chambers. For the 10°C (lumpfish) and 24°C (red drum) groups, ṀO2 recordings were started immediately after the settling period. For the 16°C (lumpfish) and 30°C (red drum) groups, water temperature was gradually increased to these temperatures after the settling period, at a rate of 2°C h−1, upon which ṀO2 recordings were initiated. ṀO2 recordings were then continued for 12–17.5 h, which produced 95–100 (lumpfish) or 160–200 (red drum) individual ṀO2 measurements. For lumpfish, the durations of each intermittent measurement cycle (flush/wait/measure) were 300/120/240 s and 300/60/180 s at 10 and 16°C, respectively, and for red drum were 90/30/120 s and 120/30/120 s at 24 and 30°C, respectively. Oxygen levels in the respirometry chambers did not fall below 90% air saturation during ṀO2 measurements. SMR was determined by first taking the mean of the lowest 10% of ṀO2 measurements over the 12–17.5 h respirometry period, excluding outliers that were ±2 s.d. from the mean (no more than two data points were identified as outliers for any fish), and finally calculating the mean of the remaining ṀO2 measurements (Clark et al., 2013). This method ensured that only ṀO2 measurements recorded when the fish were quiescent were included in the SMR calculations.
Pcrit measurements were made immediately following the determination of SMR, by turning off the flush pumps and allowing the fish to consume the oxygen in the sealed respirometry chambers, thereby gradually exposing them to increasing hypoxia. At a PwO2 of ∼15 mmHg, the flush pumps were turned back on and the fish were returned to a recovery tank. The obtained ṀO2 measurements were then plotted against the declining PwO2, and Pcrit was calculated as the PwO2 where ṀO2 first decreased below SMR. It should be noted that the ṀO2 of the fish during the Pcrit experiment was often elevated above SMR as a result of spontaneous movements under hypoxia, and is therefore designated as routine metabolic rate (RMR) (see Fig. 2).
To account for background microbial respiration, ṀO2 recordings were also made after removal of the fish. ṀO2 from bacterial respiration was not detectable in any of these background measurements, likely due to the thorough cleaning of the respirometry equipment between measurements.
MMR and AS
In separate experiments, MMR was measured at 10 and 16°C for lumpfish and 24 and 30°C for red drum (N=8 for each temperature and species) in both normoxia and hypoxia (PwO2 of 70–76 mmHg; Table 1). For all measurements, the fish were placed in a circular experimental tank (50 l) containing fully aerated seawater at 10°C (lumpfish) or 24°C (red drum) and left overnight. The following morning, water temperature was either maintained at 10 or 24°C or heated to 16°C (lumpfish) or 30°C (red drum) at a rate of 2°C h−1. The fish were then individually exercised in normoxic water for 2 min (until exhaustion) by manual hand-chasing by the experimenter, and immediately transferred to a respirometry chamber kept at the appropriate temperature and PwO2. ṀO2 recordings commenced within 20 s of the cessation of chasing and MMR was determined over a 2 min period during the first ṀO2 recording, which was always the highest. The PwO2 (±2.5 mmHg) in the tank containing the respirometry chambers was regulated using a solenoid system that bubbled air or nitrogen into the water.
To calculate the percentage change in AS with changing PwO2, a 3-parameter power function was fitted to MMR in normoxia and in hypoxia, SMR at Pcrit and RMR below Pcrit. The reductions in AS (from 0% in normoxia to 100% at Pcrit) were then calculated as the difference between the corresponding MMR on the regression line and SMR in normoxia (see Fig. 2A,B,D,E). Given that MMR was not measured at oxygen levels between ∼50% and 100% air saturation, the calculated change in AS with PwO2 does not account for a potential ‘zone of hypoxia insensitivity’ near 100% air saturation, where MMR may have been unaffected by mild reductions in dissolved oxygen. Consequently, the values of AS at high PwO2 may be subject to a slight underestimate, which diminishes as PwO2 approaches Pcrit.
CTmax and the oxygen limit for thermal tolerance (PCTmax)
Finally, in both species, CTmax was measured in normoxia and at multiple levels of hypoxia (see below; N=8 in all groups). For all measurements, the fish were placed inside respirometry chambers submerged in a 40 l tank containing fully aerated seawater at 10°C (lumpfish) or 24°C (red drum) and left overnight. The following morning, PwO2 was either maintained at normoxia or acutely reduced to 100, 80, 67, 60, 40 or 21 mmHg (lumpfish) or 76, 47, 35, 23 or 11 mmHg (red drum) by bubbling the reservoir supplying water to the chambers with nitrogen. The reduction in PwO2 took 5–30 min depending on the target PwO2, and this level (±2.5 mmHg) was maintained during the entire CTmax measurement by use of the solenoid system. Once the target PwO2 was reached, the water temperature was elevated continuously at a rate of 2°C h−1 until the fish exhibited LOE. The temperature where LOE occurred was taken as CTmax. Because of the morphology of the lumpfish (flattened ventral surface), they did not necessarily fall over at the point of LOE, and were therefore gently prodded with a cotton swab at regular intervals as CTmax was approached. The temperature where the fish no longer righted themselves after being gently prodded was taken as CTmax for this species. Once CTmax was reached for individual fish, they were removed from their chambers and returned to a recovery tank. No fish were used more than once.
Both Pcrit and the hypoxia-induced reduction in AS increased with temperature from 10 to 16°C (lumpfish; Fig. 2A,B) and from 24 to 30°C (red drum; Fig. 2D,E). Therefore, the percentage change in AS at the PwO2 levels where CTmax was measured (see Fig. 2C,F) was calculated, as described above, from the ṀO2 data at 16°C (lumpfish) or 30°C (red drum). Consequently, the calculated percentage change in AS is a minimal (conservative) estimate (i.e. CTmax occurred at temperatures higher than those for which the changes in AS were calculated, meaning that AS would have been reduced even more at those temperatures).
PCTmax was determined from the CTmax data by fitting a piecewise, two-segmented linear regression through the CTmax values that were not significantly different from CTmax in normoxia, and the CTmax values that were significantly different from CTmax in normoxia (see Fig. 3).
Student's t-tests were used to examine the effect of temperature on MMR, SMR, AS and Pcrit, as well as the effect of hypoxia on MMR. All P-values were corrected for multiple comparisons using false discovery rate (FDR) correction. FDR cutoff values were 0.0383 and 0.00802 for lumpfish and red drum, respectively. One-way ANOVA was used to test the effect of PwO2 on CTmax. All assumptions of these statistical tests were met. Statistical analyses were conducted using SigmaPlot (Systat Software, Inc., Chicago, IL, USA), and the significance level for all tests was P<0.05.
MMR decreased significantly in both lumpfish and red drum exposed to hypoxia (P<0.001 for both species and temperatures), and as a result, AS fell with PwO2, as expected, until the Pcrit where AS was zero (Fig. 2A,B,D,E). At the higher temperatures (16 versus 10°C for lumpfish, 30 versus 24°C for red drum), tolerance to hypoxia was reduced (i.e. Pcrit was significantly increased; P=0.038 for lumpfish and P=0.008 for red drum) and the hypoxia-induced decline in AS was more pronounced (Fig. 2B,E versus Fig. 2A,D). However, CTmax did not show a gradual decrease with falling PwO2 (Fig. 2C,F). For lumpfish, CTmax was maintained at 21.9–22.3°C down to a PwO2 of 80 mmHg (P>0.05), at which point AS was reduced by at least 72% (Fig. 2A–C). At a PwO2 of 67 mmHg, slightly above the Pcrit at 16°C (63.8 mmHg), CTmax was significantly reduced to 20.9°C (P<0.05 compared with CTmax in normoxia), and this reduction continued with decreasing PwO2 (Fig. 2C). For red drum, a similar pattern was observed, with no significant change in CTmax (36.1–36.5°C) down to a PwO2 of 47 mmHg (P>0.05), despite AS being reduced by at least 89% (Fig. 2D–F). In red drum, the first sign of reduced CTmax occurred at 35 mmHg (P<0.05 compared with CTmax in normoxia), slightly below the Pcrit at 30°C (38 mmHg).
The oxygen limit for thermal tolerance (PCTmax), as calculated from the CTmax data, was 72.2 mmHg for lumpfish (Fig. 3A) and 35.8 mmHg for red drum (Fig. 3B). These values were significantly different (P<0.001).
All results, including fish body mass, are presented as means with associated s.e.m. in Table 1.
This study tested the hypothesis that the upper thermal limits of fishes are determined by insufficient oxygen supply for vital physiological functions and, therefore, affected by oxygen availability. More specifically, we tested the assumption that the fish's CTmax cannot be maintained under aquatic hypoxia where oxygen supply capacity (i.e. AS) is reduced (see Fig. 1A,B). Contrary to this assumption, we found that the CTmax of both lumpfish and red drum was independent of oxygen availability over a wide range of water oxygen levels. In fact, the AS of lumpfish and red drum could be reduced by more than 72% and 89%, respectively, before CTmax was affected (Fig. 2). This shows that the upper thermal limits of water-breathing ectothermic animals are not always determined by oxygen supply capacity (see Fig. 1C,D).
In our experimental approach, we initially verified that MMR and AS were constrained in both lumpfish and red drum as PwO2 was lowered at a constant temperature, both within (10°C for lumpfish, 24°C for red drum) and slightly above (16°C for lumpfish, 30°C for red drum) the species' typical temperature range (Fig. 2A,B,D,E). This was simply a proof of concept, as the constraining effect of hypoxia on AS is a well-known phenomenon in fishes (Claireaux and Chabot, 2016; Claireaux et al., 2000; Farrell and Richards, 2009; Fry, 1971; Lefrançois and Claireaux, 2003). We then defined CTmax at water oxygen levels ranging from air saturation (normoxia) to below the fish's critical oxygen tension (Pcrit). In determining CTmax, we employed a 2°C h−1 warming protocol, which approximates the maximum heating rate that fish experience under natural conditions (Fangue et al., 2011; Gamperl et al., 2002; Loong et al., 2005). This rate is also similar to the rate of temperature change (1–4°C h−1) used in other key studies investigating the role of oxygen limitations on the thermal tolerance of ectotherms (Eliason et al., 2011; Frederich and Pörtner, 2000; Giomi et al., 2014; Giomi and Pörtner, 2013; Melzner et al., 2006; Wittmann et al., 2008).
In lumpfish, the first sign of a reduced CTmax occurred slightly above Pcrit at 16°C, whereas in red drum, CTmax was not reduced until slightly below Pcrit at 30°C (Fig. 2C,F). Furthermore, the PCTmax of lumpfish was closer to normoxia than the PCTmax of red drum. Although the two species were acclimated to different temperatures, as befitted the species, these data indicate that the lumpfish retained a lower oxygen supply capacity at CTmax in normoxia when compared with red drum. It has been suggested that species occupying a wide thermal range (niche) have been evolutionarily selected for a more thermally resistant cardiorespiratory system (Ern et al., 2014; Jost et al., 2012). The differences in oxygen supply capacity at CTmax, observed here for lumpfish and red drum, may therefore be related to the ecology of the two species. The red drum is an athletic, free-swimming, eurythermal species inhabiting environments such as estuaries and coastal swamps that can fluctuate greatly in temperature over short temporal scales, whereas the lumpfish is a relatively sluggish, cold-water species with a more thermally stable niche (it lives in the coastal zone and in the open ocean). Despite these differences, the CTmax of both lumpfish and red drum decreased when PwO2 levels fell below 80 and 47 mmHg, respectively, in accordance with the theory that survival below Pcrit is determined by the capacity for anaerobic metabolism, and is, therefore, time limited.
Our findings add to a series of recent studies that have suggested that upper thermal limits in a number of water-breathing ectotherms are not determined by insufficient oxygen supply capacity. These studies, performed on a range of species, found that AS is maintained in normoxia at environmentally relevant temperature extremes (Ern et al., 2015; Gräns et al., 2014; Norin et al., 2014), or showed that experimentally induced anemia (internal hypoxia) has very little effect on CTmax (Brijs et al., 2015; Wang et al., 2014). The present study bridges these earlier studies, and confirms their findings, by showing an uncoupling of AS and CTmax when fish are exposed to environmental (ambient) hypoxia. Our findings are also supported by previous studies which have reported reduced CTmax under severe hypoxia (Weatherley, 1970; Rutledge and Beitinger, 1989; Ellis et al., 2013; Healy and Schulte, 2012) and maintained CTmax under moderate hypoxia (Weatherley, 1970; Ellis et al., 2013). However, direct comparisons with our results are difficult because these studies only measured CTmax at one (Rutledge and Beitinger, 1989; Healy and Schulte, 2012) or two (Weatherley, 1970; Ellis et al., 2013) levels of hypoxia, applied extremely high rates of temperature increase (18–90°C h−1; Weatherley, 1970; Rutledge and Beitinger, 1989; Healy and Schulte, 2012), allowed PwO2 to drift from normoxia to hypoxia during measurements (Ellis et al., 2013), or were performed on an air-breathing species with surface access (Rutledge and Beitinger, 1989).
Based on the species-specific dependence of CTmax on oxygen availability observed here (i.e. red drum were more resilient to hypoxia than lumpfish), we propose a novel metric for classifying the oxygen-dependent thermal sensitivity of different species; the oxygen limit for thermal tolerance (PCTmax). By measuring CTmax at a range of PwO2 levels from normoxia to below Pcrit, an oxygen-dependent breakpoint can be established where CTmax is significantly affected by hypoxia (Fig. 3). As PCTmax defines the PwO2 below which the upper thermal limits of water-breathing ectotherms becomes constrained, the breadth of the thermal niche under hypoxic conditions will increase with the distance between PCTmax and the PwO2 where AS starts to be reduced by hypoxia. If the latitudinal distribution ranges of marine fishes are shaped by their upper thermal limits (Sunday et al., 2012), then populations of species exhibiting oxygen-independent upper thermal limits at environmentally relevant PwO2 levels should be more resilient to the occurrence of environmental hypoxia and oxygen-poor ‘dead’ zones within their thermal niche. In addition to assessing the oxygen sensitivity of upper thermal limits in water-breathing ectotherms, and the susceptibility of their upper thermal niche boundaries to environmental hypoxia, we also suggest that the PCTmax metric can be used to assess the effects of anthropogenic pollutants and other climate change-related stressors (e.g. acidification and salinity changes) on the hypoxia sensitivity of upper thermal limits.
In conclusion, the CTmax of lumpfish and red drum was unaffected by hypoxia exposure down to PwO2 levels of 80 and 47 mmHg, respectively, despite more than 72% and 89% reductions in AS. Thus, constrained oxygen supply capacity did not affect the CTmax of these species under conditions of reduced oxygen availability that are routinely encountered by many fishes in the wild. However, increases in the frequency and severity of aquatic hypoxia may cause some species to experience PwO2 levels sufficiently low for CTmax to be reduced. The proposed PCTmax metric can help identify such vulnerable species. In general, our results show that oxygen and upper thermal limits are not intimately linked, thus reinforcing the idea that there are other important physiological constraints that determine the upper thermal limits of fishes.
The authors thank Dr Keng Pee Ang (Cooke Aquaculture Inc.) and Mr Danny Boyce (JBARB, MUN) for providing the lumpfish used in this study.
Conceptualization: R.E., T.N. and A.J.E.; Methodology: R.E., T.N. and A.J.E.; Formal analysis and investigation: R.E. and T.N.; Writing - original draft preparation: R.E. and T.N.; Writing - review and editing: A.K.G. and A.J.E.; Funding acquisition: R.E., T.N., A.K.G. and A.J.E.; Resources: A.K.G. and A.J.E.; Supervision: A.K.G. and A.J.E. All authors gave final approval for publication.
We gratefully acknowledge financial support from the National Science Foundation grant to A.J.E. (EF 1315290), from the Natural Sciences and Engineering Research Council of Canada (NSERC) to A.K.G., and from the Carlsberg Foundation to R.E. through their Internationalisation Fellowship program (CF15-0321). T.N. gratefully acknowledges financial support from the Danish Council for Independent Research (Det Frie Forskningsråd) during the writing of this paper (Individual Post-doctoral Grant and Sapere Aude Research Talent Grant; DFF-4181-00297).
The authors declare no competing or financial interests.