Hypoxia tolerance is conserved across genetically distinct sub-populations of an iconic, tropical Australian teleost (Lates calcarifer)

We used five genetically distinct sub-populations of Australian barramundi (Lates calcarifer) to examine the extent of intraspecific variability in hypoxia tolerance at typical (26°C) and warm (36°C) temperatures. Critical oxygen tension ([O2]crit) was lower at 26°C than at 36°C, indicating greater hypoxia tolerance at the cooler temperature, but was not different between sub-populations.


Introduction
The availability of dissolved oxygen (DO) in aquatic systems is critical in defining the abundance and distribution of biological communities (Pearson et al., 2003). In freshwater and estuarine systems, oxygen can fluctuate between normoxia and severe hypoxia due to respiration of organisms inhabiting such systems and thermal stratification (Pearson et al., 2003;Brauner and Val, 2005). In tropical regions, initial rains at the onset of the monsoon season can carry high loads of organic material into freshwater and estuarine systems, causing a rapid depletion of oxygen, which can be fatal for resident animal populations (Bishop, 1980;Townsend et al., 1992;Erskine et al., 2005). Furthermore, natural occurrences of hypoxia may be exacerbated by anthropogenic disturbances, such as agriculture and urbanization (Pollock et al., 2007).
Altered weather patterns resulting from climate change have the potential to exacerbate periods of environmental hypoxia. Higher water temperatures reduce the solubility of oxygen in water, while concurrently increasing the metabolic demand for oxygen in aquatic ectotherms (Weiss, 1970;Diaz and Breitburg, 2009). As such, the unprecedented rate of temperature rise predicted under current climate change scenarios may push many aquatic species close to, or above, their upper thermal thresholds (Tewksbury et al., 2008;Morrongiello et al., 2011;Huey et al., 2012). This may be particularly pronounced at very low latitudes near the equator, where species may be adapted to mean temperatures that may span only 1-2°C annually (Tewksbury et al., 2008). Thus, it is likely that the elevated temperatures predicted to occur in the future will progressively increase the prevalence and severity of hypoxia in aquatic systems, with flow-on effects to larger-scale processes, such as population growth, fitness and ecosystem dynamics (Pollock et al., 2007).
Fish can respond physiologically and behaviourally to fluctuating DO, although such responses are highly dependent on species and context. For example, fish that encounter hypoxic conditions may actively avoid areas of low DO (Herbert and Steffensen, 2005;Poulsen et al., 2011), increase gill ventilation volume (Jesse et al., 1967;Fernandes and Rantin, 1989), increase the number of red blood cells and haemoglobin concentration (Wells et al., 1989), and/or induce bradycardia (Randall and Smith, 1967). Exposure to acute or chronic hypoxia can have sub-lethal effects, such as altered behaviour and reduced growth and reproduction, which may be further exacerbated by increasing water temperatures (Vaquer-Sunyer and Duarte, 2008).
The physiological responses to hypoxia are regarded as being dependent on the frequency and severity of the hypoxic event (Farrell and Richards, 2009). Below a certain DO, most fish are unable to regulate their oxygen consumption rate ( ) M O 2 independently of ambient oxygen levels and consequently enter a state of oxygen conformity (Prosser and Brown, 1961;Fernandes et al., 1995;Schurmann and Steffensen, 1997). The level of DO at which this conformity occurs is referred to as the critical oxygen saturation ([O 2 ] crit ) and is commonly used as a measure of the hypoxia tolerance of a species (Prosser and Brown, 1961;Schurmann and Steffensen, 1997).
The prevalence of intraspecific local adaptation (or interdemic variation) may influence the capacity for sub-populations to respond to environmental changes (Grabowski et al., 2009;Tobler et al., 2011;Whitehead et al., 2011). It has previously been proposed that selection pressure for hypoxia tolerance may lead to variation among sub-populations for species with broad habitat ranges (Timmerman and Chapman, 2004). A number of studies to date have found evidence for local thermal adaptation in metabolic traits between sub-populations of species including killifish (Fundulus heteroclitus; Dhillon and Schulte, 2011) and Atlantic cod (Gadus morhua; Sylvestre et al., 2007;Grabowski et al., 2009). Local adaptation to hypoxia between sub-populations of fish species across broad spatial scales, however, is less well understood, and no study has investigated this for a higher-level predator.
To help address this knowledge gap, we investigated [O 2 ] crit and resting M O 2 in five different sub-populations of Australian barramundi (Lates calcarifer) spanning ~12° of latitude ( Fig. 1). Experiments were performed at two temperatures chosen to represent typical (26°C) and warm summer conditions (36°C). The main aims of this study were as follows: (i) to investigate whether sub-populations differ in their [O 2 ] crit and/or resting M O 2 ; (ii) to quantify how different sub-populations respond to the combined challenges of temperature and hypoxia; and (iii) to determine whether hypoxia tolerance is related to latitudinal position. We expected barramundi sub-populations from lower latitudes to be more hypoxia tolerant than their higher latitude counterparts due to the inverse relationship between water temperature and oxygen solubility. We also expected barramundi (irrespective of sub-population) to exhibit higher metabolic rates at warmer temperatures, and consequently, to be less hypoxia tolerant due to the elevated metabolic demand for oxygen.  Hutchinson and Dowling (1991) and are indicated as follows: I, Indian Ocean; II, Kimberley and Arafura Sea; III, Gulf of Carpentaria; and IV, North East Coast. Red lines adjacent to the coast indicate the southerly limits of the species' distribution.

Experimental animals and holding conditions
Barramundi juveniles were obtained from five commercial hatcheries located at Broome (Broome Aquaculture Centre), Darwin (Darwin Aquaculture Centre), Karumba (Barramundi Discovery Centre), Townsville (Mainstream Aquaculture), and Gladstone (Gladstone Water Board Barramundi Hatchery; Fig. 1). All five hatcheries use broodstock sourced from local rivers that are separated by a minimum of 700 km (Fig. 1). Wild sub-populations of Australian barramundi differ genetically (Keenan, 1994;Doupé et al., 1999), and broodstock maintained at these locations have been identified as genetically distinct (Smith-Keune C, unpubublished data). Prior to experimental treatments, fish were grown on to a size of ~200 g over ~10 months in tanks containing fresh water connected to a recirculating system at the Marine and Aquaculture Research Facility Unit of James Cook University (Townsville, Queensland, Australia). All fish were fed ~1% body weight per day using a commercial pelleted feed (Ridley Aquafeed, Narangba, Queensland, Australia), and maintained at 26 ± 1°C under a 12 h-12 h photoperiod. Dissolved oxygen was maintained at >75% saturation in the holding tanks, and water quality was monitored daily. Individual fish were weighed and measured prior to experiments and following respirometry. Mean (±SD) body mass and condition factor (K) of fish after respirometry were 194.16 ± 22.60 g and 1.17 ± 0.09, respectively. Condition factor was calculated according to the following formula: K = 100a × b −3 , where a = weight (in grams) and b = total length (in centimetres; Froese, 2006) .

Experimental design
Temperature treatments were selected based on a review of atmospheric and sea-surface temperatures obtained from Australian government databases (AIMS, 2012;BOM, 2012), and from a comprehensive review of river temperatures from previously published data (Pusey et al., 1998;Stuart and Berghuis, 2002;Webster et al., 2005). Based on this information, the two temperature treatments were as follows: (i) a 'typical' temperature (26°C), which is representative of the annual mean across the species distribution in Australia and the temperature at which the fish had been held long term; and (ii) a 'warm' temperature (36°C), which is regarded to be representative of the upper limit that wild populations experience (Russell and Garrett, 1983), but still within the known tolerance limits for the species (Bermudes et al., 2010). Prior to experiments, fish were removed from their holding tanks and acclimated to one of two temperatures in a separate system where water temperature could be closely controlled. While fish in the 26°C treatment group did not undergo a temperature change relative to prior holding conditions, fish used in the 36°C treatment group were acclimated by increasing the temperature by 1°C per day and then held at 36°C for a minimum of 7 days prior to experimentation. All fish were fed as stated in the previous subsection, but food was withheld for 24 h prior to conducting respirometry to minimize the effect of specific dynamic action on oxygen consumption measurements.

Respirometry
All measurements were performed using intermittent flowthrough respirometry, following best practices outlined in Clark et al. (2013). Respirometers (volume = 10.3 l) were fitted with a small Perspex window to allow fish to be observed during respirometry trials. Each respirometer was connected to two pumps; a single recirculating pump to keep water within the chamber mixed, and a flush pump to supply the chamber with aerated water between M O 2 measurements.
Respirometers and flush pumps were submerged in a shallow 1000 l tank, with vigorous aeration to provide both stable temperature conditions during experiments (26.2 ± 0.5 and 36.4 ± 0.2°C) and to supply the chambers with ~100% saturated water during the flush cycle. Temperature-compensated oxygen concentration (in milligrams per litre) of the water within each chamber was continuously recorded (0.5 Hz) using oxygen-sensitive REDFLASH ® dye on contactless spots (2 mm) adhered to the inside of each chamber and linked to a Firesting Optical Oxygen Meter (Pyro Science e. K, Aachen, Germany) via fibre-optic cables. Oxygen-sensing equipment was recalibrated daily using a one-point 100% saturation calibration and an electrical (factory-calibrated) zero. Following initial measurements of background respiration in the respirometers, individual fish that had been fasted for 24 h were placed in respirometers in the evening and allowed to acclimate to the respirometers for 16 h. During the acclimation period, the flush pumps attached to each respirometer were set to a 30 min-15 min on-off cycle, and M O 2 (in milligrams of O 2 per kilogram per minute) was measured from the decline in oxygen in each respirometer during each 15 min off cycle. Resting M O 2 was calculated for each fish as the mean of the lowest three measurements recorded during the acclimation period.
Following the chamber acclimation period and resting M O 2 measurements, each flush pump was turned off, and fish were permitted to deplete the oxygen within their respective respirometers down to 5% air saturation. The M O 2 was calculated for each consecutive 5 min period during the decline in oxygen (~1.5-4 h depending on temperature). Upon reaching 5% air saturation, each flush pump was turned on to restore oxygen levels to 100% saturation. Preliminary trials conducted using similar-sized fish indicated that depletion to 5% saturation was sufficient to calculate [O 2 ] crit but still above the oxygen levels that induce loss of equilibrium. The following two values were identified from each [O 2 ] crit experiment: (i) pre-hypoxia M O 2 , which was calculated from the mean of the lowest three 5 min measurements recorded between 100 and 75% saturation after flush pumps had been turned off; and (ii) [O 2 ] crit , which was determined using previously established methods (Corkum and Gamperl, 2009;Nilsson et al., 2010 (Ott et al., 1980;Nilsson et al., 2004). After the completion of each trial, additional background respiration measurements were obtained for each chamber in the absence of fish. Any change in background respiration between the start and end of experiments was assumed to be linear when correcting fish M O 2 .

Data analyses and statistics
All oxygen consumption rate measurements (in milligrams of O 2 per kilogram per minute) were calculated using commercial software (LabChart v. 7; ADInstruments, Sydney, NSW, Australia) from the slope of the decline in oxygen concentration according to the formula: where slope a and slope b are the declines in oxygen (in milligrams per litre per second) measured in the presence and absence of fish within the chamber, respectively, and V c and M b are the volumes (in litres) of the chamber and fish, respectively.  (Zar, 2010). Homogeneity of variance and normality were assessed using Levene's test and normal quantile-quantile (Q-Q) plot, respectively. Data are presented as means ± SD, and results were considered statistically significant at P < 0.05.

Results
Irrespective of population, the resting M O 2 of fish acclimated to 26°C was significantly lower than the resting M O 2 of fish acclimated to 36°C (1.47 ± 0.24 vs. 3.10 ± 0.43 mg O 2 kg −1 min −1 for all sub-populations combined; F 1, 110 = 563.57, P < 0.001; Fig. 2). There was a significant interaction between sub-population and temperature for resting M O 2 (P = 0.025). No differences were found for resting M O 2 between sub-populations at 26°C (H 5 = 1.446, P = 0.84). At 36°C, resting M O 2 differed from that at 26°C (H 5 = 11.25, P = 0.0239; Fig. 2), although Dunn's post hoc comparison did not identify differences between individual sub-populations. Pre-hypoxia M O 2 was higher for the Townsville sub-population than for the Darwin sub-population at 26°C (H 5 = 11.90, P = 0.018), and was higher for the Gladstone sub-population than for the Broome or Karumba sub-populations at 36°C (H 5 = 17.02, P = 0.002).
The average Q 10 for resting M O 2 and pre-hypoxia M O 2 across all populations was 2.12 ± 0.30 and 2.12 ± 0.36, respectively.
Barramundi exhibited a common and clear trend in response to decreasing oxygen, with fish maintaining a relatively constant M O 2 above the [O 2 ] crit , followed by a steep decline (Fig. 3). There was no significant interaction between population and temperature for [O 2 ] crit (P = 0.486). Temperature had a significant effect on mean [O 2 ] crit across all populations, with lower [O 2 ] crit at 26 than at 36°C (F 1,114 = 69.97; P < 0.001; Fig. 4). No significant differences were observed for mean [O 2 ] crit between four of the five sub-populations tested at both 26 and 36°C (Fig. 4). The exception was that mean [O 2 ] crit was highest for fish from Darwin at both 26 and 36°C (18.91 ± 2.97 and 23.84 ± 3.49%, respectively). Critical oxygen saturation 4 Figure 2: Resting oxygen consumption rates (in milligrams of O 2 per kilogram per minute; means ± SD) for five barramundi subpopulations at 26°C (blue; n = 48) and 36°C (red; n = 59). Oxygen consumption rates were calculated from the average slope of the decline in chamber [O 2 ] across a 15 min period during the 'closed' cycle of intermittent flow-through respirometry, and measured every 45 min. The continous and dashed horizontal lines represent the mean and standard deviation for all populations at each temperature. measurements varied between individuals, with minimal and maximal values spanning 8.63-23.02% saturation at 26°C (mean = 15.44 ± 3.20% saturation) and 13.47-31.17% saturation at 36°C (mean = 21.07 ± 3.92% saturation). Fish displayed signs of distress, such as erratic movements and pale body colour, below [O 2 ] crit ; however, all fish recovered fully upon returning to 100% saturated conditions.

Discussion
Species inhabiting thermally stable environments (e.g. equatorial or polar) are thought to be physiologically constrained within a narrow range of environmental thresholds, suggesting that they may be particularly susceptible to increasing temperatures and increasing variability in mean temperatures (Tewksbury et al., 2008). Due to the inverse relationship between temperature and oxygen solubility in water (Diaz and Breitburg, 2009), we expected barramundi from lower latitudes to be more hypoxia tolerant than their higher-latitude counterparts. Contrary to expectations, Darwin fish (low latitude) displayed a slightly higher average [O 2 ] crit (less hypoxia tolerant) than fish from the other four populations tested at both control and warm temperatures. However, on   Fig. 1). The continuous and dashed horizontal lines represent mean and standard deviation, respectively, of critical oxygen saturation for all populations at each temperature. Letters indicate significant differences (P < 0.05; a < b and A < B).
the whole, we were unable to find consistent and conclusive evidence for local adaptation in hypoxia tolerance between Australian barramundi sub-populations separated by large longitudinal and latitudinal distances.
Interspecific variation in hypoxia tolerance, as derived from [O 2 ] crit measurements, exists for a number of temperate and tropical fish species at temperatures that are routine or typical (Table 1). Species such as Nile tilapia (Oreochromis niloticus) can regulate M O 2 down to 11 ± 3% saturation, whereas other species, such as Atlantic salmon (Salmo salar), can regulate M O 2 to only 36 ± 3% saturation (Table 1). Schofield and Chapman (2000) reported a [O 2 ] crit of 15% saturation for juvenile Nile perch (Lates niloticus) at 20°C, and also noted the tendency of Nile perch to congregate at the surface layer of water during periods of aquatic hypoxia in an attempt to enhance oxygen uptake from the more aerated surface waters (commonly referred to as aquatic surface respiration). It has been suggested, however, that Nile perch are inefficient at using this technique to obtain oxygen (Schofield and Chapman, 2000). Anecdotal reports exist of farmed barramundi aggregating at the surface during periods of low DO; however, the capacity for barramundi to use aquatic surface respiration remains unclear. No attempt was made to measure aquatic surface respiration in this study; however, future experiments using respirometers with an air pocket at the surface would help to elucidate the potential for aquatic surface respiration in barramundi, and therefore their capacity to exist for extended periods in hypoxic waters below [O 2 ] crit .
Barramundi are featured in mass-mortality events in northern Australia (Townsend et al., 1992), and previous reports indicate that barramundi succumb rapidly to hypoxia in severe conditions. Pearson et al. (2003) reported a 'lethal DO concentration' of ~15% saturation for barramundi living in temperatures ranging from 28 to 30°C; however, the details of the fish and experimental design were unclear, making the results difficult to interpret in the context of the present study.
In contrast, Wu (1990) reported no mortality for 150-250 g barramundi held in saltwater after 8 h of exposure to ~15% saturation at 25°C, but exposure to ~8% saturation resulted in 50% mortality after 6.5 h. Butler et al. (2007) reported an 'acute asphyxiation concentration' of 4% saturation for 200 g barramundi in freshwater at 28°C. Fish were visibly distressed in severe acute hypoxia in the present study, but there was only one mortality during the [O 2 ] crit tests, despite the fact that each individual fish was exposed to DO of ~4-5% saturation in order to obtain precise [O 2 ] crit measurements.
As well as measuring 'acute asphyxiation', Butler et al. (2007) investigated the relationship between gill ventilation volume and frequency in response to declining DO for barramundi at 28°C. Butler et al. (2007) demonstrated that this species increases ventilation rates down to ~15-20% saturation, followed by a steep decline in ventilation rate as DO continues to decline. The maximal ventilation rate observed by Butler et al. (2007)  There is varying evidence in the literature for local adaptation of performance traits in sub-populations of Australian barramundi. Cairns (latitude 16°S) and Burrum River (26°S) barramundi have been found to exhibit no differences in specific growth rate, critical thermal minima, or critical thermal maxima between populations (Rodgers and Bloomfield, 1993;Burke, 1994;Keenan, 2000). Recent research, however, has documented differences in performance traits at high temperatures for Australian barramundi from lower (northern) latitudes compared with those from higher (southern) latitudes through the measurement of critical swimming speed (Edmunds et al., 2010) and time to loss of swimming equilibrium when challenged with acute water heating (Newton 6 for teleost fish, irrespective of species or life history, reflecting increased metabolic requirements at higher temperatures (Fry and Hart, 1948;Brett and Groves, 1979;Clark et al., 2011).
Barramundi elicit a 2-fold increase in resting M O 2 (Q 10 = 2.12) over a 10°C increase in temperature. Higher M O 2 at warmer temperatures might be expected to induce a decrease in hypoxia tolerance in all fish species due to the increased oxygen demands of the fish. Results from previous studies on species such as Nile tilapia (Fernandes and Rantin, 1989;Mamun et al., 2013), rainbow trout (Oncorhynchus mykiss; Ott et al., 1980), and common carp (Cyprinus carpio; Ott et al., 1980) indicate that [O 2 ] crit appears to be less temperature sensitive than what is typically found for resting M O 2 ; however, this is not consistent across all teleosts. Species such as Atlantic salmon (Barnes et al., 2011), Doederlein's cardinal fish (Ostorhinchus doederleini; Nilsson et al., 2010), and Atlantic cod (Schurmann and Steffensen, 1997) elicit a large increase in [O 2 ] crit with increasing temperature, particularly towards upper thermal tolerance limits. Our results indicate that [O 2 ] crit displays only a small increase for barramundi from typical (26°C) to high temperatures (36°C), demonstrating the resilient nature of this species to the synergistic effects of temperature and environmental hypoxia.
Aside from temperature, a number of other environmental parameters can impact the hypoxia tolerance of fish. The ability to adjust to hypoxic conditions through a lowering of [O 2 ] crit following pre-exposure has been reported for the epaulette shark (Hemiscyllium ocellatum; Routley et al., 2002) and goldfish (Carassius auratus; Fu et al., 2011), although Cook et al. (2011) found no differences in [O 2 ] crit between naïve and hypoxia-conditioned silver sea bream (Pagrus auratus). Henriksson et al. (2008) demonstrated that alterations in salinity can increase the [O 2 ] crit by up to 30% in prickly sculpin (Cottus asper) acclimated to freshwater, compared with fish adapted to seawater; however, the same trend was not observed in the closely related Pacific staghorn sculpin (Leptocottus armatus). Barramundi inhabit environments that are prone to acute and chronic hypoxia, and such environments also experience broad fluctuations in salinity and temperature. Currently, there is no understanding of the effects of exposure to repetitive (short-term) or chronic (longterm) hypoxia on the physiology, and consequently, performance of barramundi. However, this topic warrants future research to elucidate more fully the capacity of fishes to tolerate extreme and variable environments.
Elevated temperatures due to climate change are predicted to increase the frequency and severity of hypoxic events through lower O 2 solubility, increased animal respiration rates, and enhanced stratification (Justic et al., 2001;Diaz and Breitburg, 2009). Such conditions have the potential to increase habitat availability at higher latitudes, while leading to the deterioration of habitats at lower latitudes (Graham and Harrod, 2009;Gardiner et al., 2010). Nutrient inputs to aquatic coastal systems from human waste and agriculture, combined with habitat degradation through increased land use, have the potential to exacerbate further the effects of environmental hypoxia on physiological systems. The present study has shown, for the first time, that a tropical euryhaline fish, barramundi, does not display obvious local adaptation in resting M O 2 and hypoxia tolerance. This strongly suggests that all sub-populations of this species in northern Australia can cope equally well in environments with large fluctuations in both temperature and dissolved oxygen.