Mammalian metabolic rates in the hottest fish on earth

The Magadi tilapia, Alcolapia grahami, a small cichlid fish of Lake Magadi, Kenya lives in one of the most challenging aquatic environments on earth, characterized by very high alkalinity, unusual water chemistry, and extreme O2, ROS, and temperature regimes. In contrast to most fishes which live at temperatures substantially lower than the 36–40 °C of mammals and birds, an isolated population (South West Hot Springs, SWHS) of Magadi tilapia thrives in fast-flowing hotsprings with daytime highs of 43 °C and night-time lows of 32 °C. Another population (Fish Springs Lagoon, FSL) lives in a lagoon with fairly stable daily temperatures (33–36 °C). The upper critical temperatures (Ctmax) of both populations are very high; moreover the SWHS tilapia exhibit the highest Ctmax (45.6 °C) ever recorded for a fish. Routine rates of O2 consumption (MO2) measured on site, together with MO2 and swimming performance at 25, 32, and 39 °C in the laboratory, showed that the SWHS tilapia exhibited the greatest metabolic performance ever recorded in a fish. These rates were in the basal range of a small mammal of comparable size, and were all far higher than in the FSL fish. The SWHS tilapia represents a bellwether organism for global warming.

knowledge summarized above, we hypothesized that both temperature tolerance and metabolic performance of the former would be exceptional. Indeed the current data show that both upper critical temperature and aerobic metabolic capacity of the SWHS tilapia surpass those of any other teleost fish.

Results and Discussion
Habitat Characteristics. We characterized the diurnal temperature, dissolved O 2 , and pH profiles at the two sites (FSL, Aug. 1; SWHS, Aug. 5,2013) in the coolest month of the year, taking care to place the probe in locations where the fish were present at mid-day. At the SWHS site where the very shallow (< 12 cm), fast-flowing water is under direct sunlight, temperature varied dramatically over the day from 43 °C at 14:00 to 32 °C during the night (Fig. 1). Additionally, we routinely observed fish moving through thermal spring plumes at 43.7-44.2 °C, though not staying there. Fish were never observed in a 44.8 °C plume. Dissolved O 2 varied from super-saturation during daylight hours to hypoxia during the night and anoxia at dawn, whereas water pH was stable at ~9.40. At FSL, where the current is negligible and the water is deeper (~1 m) but again under direct sunlight, conditions were much more constant with diurnal temperature variations of 33-36 °C, dissolved O 2 variations of 30-80% saturation, and stable pH ~ 9.75 (Fig. 2). Other aspects of water chemistry were also very different at the two sites, with much higher ions, osmolality, total CO 2 , and titratable alkalinity at SWHS (Table 1), yet lower pH, in accord with a previous report 5 . Far more avian predators were also seen at SWHS. Critical Temperatures. Field Ct max , measured at lakeside using a standardized protocol 16 in SWHS fish freshly caught from water at 40-41 °C, was 45.6 ± 0.1 °C (N = 8), the highest ever recorded for a fish (Table 2). FSL fish were also very temperature tolerant: the comparable field Ct max for FSL fish caught from 33 °C water was 43.6 ± 0.1 °C (N = 8). In both cases, field Ct min values were approximately 30 °C lower than Ct max . After  Routine MO 2 was also measured in the field on freshly caught fish using Tusker chambers immersed in the lake, to hold ambient temperature. Routine field MO 2 in SWHS fish was 87.7 ± 12.0 (15) μ mol g −1 h −1 at a water temperature of 41 °C. These routine rates for SWHS fish in the field are truly exceptional; a similarly sized small mammal, the 6-g adult pygmy mouse (Baiomys taylori), has an identical routine metabolic rate (87.1 ± 9.8 μ mol g −1 h −1 ) at an air temperature of 30-33 °C (thermoneutral zone) 18 . Routine MO 2 in FSL fish in the field was 4-fold lower, 22.2 ± 1.5 (10) μ mol g −1 h −1 at 33 °C. Routine M Urea-N in the field scaled in proportion: 21.5 ± 3.3 (15) μ mol g −1 h −1 at 41 °C in SWHS fish versus 5.8 ± 0.6 (10) μ mol g −1 h −1 at 33 °C in FSL fish.
Swimming Performance and Metabolism. Critical swimming speed 19 (U crit , 7.5-9.5 body lengths sec −1 ) in Blazka-style swim tunnels 20 was high and independent of temperature (25, 32, 39 °C) in the SWHS fish, but significantly lower at both 25 °C and 39 °C in FSL fish, collapsing to only about 2 body lengths sec −1 at 39 °C ( Fig. 4). In both populations MO 2 increased approximately linearly with swimming speed, but the rates associated with any swimming speed at any temperature were higher (by 1.1-2.5-fold) in the SWHS fish (Fig. 5). In contrast to MO 2 , M Urea-N did not increase with swimming speed in either population, though again rates were generally higher in SWHS fish ( Supplementary Fig. S1). The lack of increase in M Urea-N with increased aerobic metabolic rate during swimming was seen previously in FSL tilapia after exhaustive exercise 9 ; possible explanations include either a switch away from amino-acid based fuels and/or inhibition of urea synthesis to reduce costs during exercise.
Anecdotally, we noticed in both populations, that if an air bubble was provided in the swimming respirometer after exhaustion, the fish would continuously breathe air for up to 40 min, presumably to help satisfy the demands of excess post-exercise O 2 consumption.
Q10 values for MO 2 during swimming, calculated for individual speeds, averaged 2.00 ± 0.26 (8 comparisons across 3 temperatures) in SWHS fish indicating a continuing ability to increase metabolism with temperature in the face of exercise, whereas in FSL fish the mean Q10 value during swimming was only 1.35 ± 0.07 (6 comparisons across 3 temperatures), indicating limited capacity. Mean Q10 values for M Urea-N during swimming were 2.15 ± 0.53 (8) for SWHS and only 1.30 ± 0.21 (6) for FSL, indicating similar limitation. Notably these Q10 patterns contrasted with those for routine metabolism reported above which increased more quickly with temperature in FSL fish than in SWHS fish.
Values for MO 2(min) (calculated rate of O 2 consumption at zero activity), MO 2(max) (calculated rate of O 2 consumption at maximum sustainable swimming activity) and aerobic scope (the difference between these two values) (Fig. 6) illustrate the dramatic differences between the two populations, and the exceptional capacity of the SWHS fish. MO 2(min) , MO 2(max) , and aerobic scope values for FSL tilapia did not differ across temperatures ( consumed much of the aerobic scope of the FSL fish, explaining their poor swimming performance at this temperature. In contrast MO 2(max) , and aerobic scope steadily increased with temperature in SWHS fish (Fig. 6), while routine MO 2 (Fig. 3B) remained close to MO 2(min) , accounting for only a small fraction of the aerobic scope at all temperatures. Interestingly, MO 2(min) only differed from the FSL values at 39 °C, but MO 2(max) , and aerobic scope were significantly greater at all three temperatures. Factorial aerobic scope in SWHS tilapia was about 6 at 25 °C and 32 °C, falling to 4 at 39 °C. MO 2(max) at 39 °C in SWHS tilapia was 175.4 ± 34.6 (6) μ mol g −1 h −1 . It is impressive that this fish can sustain an MO 2(max) equal to about one-third of the MO 2(max) (507 μ mol g −1 h −1 ) of the pygmy mouse 18,21 . We believe that these rates for SWHS tilapia are indicative of the greatest metabolic performance ever recorded in a fish of comparable size. The obvious comparison is with tunas, but there appear to be no published data for 3-g tunas, and they are not endothermic at this size. It is also doubtful whether even adult tuna ever reach body temperatures of 39 °C. The smallest tuna studied appears to be a 24-g kawaka (Euthynnus affinis) which exhibited an MO 2(max) of about 94 μ mol g −1 h −1 at a Ucrit of about 4 body lengths sec −1 at 24° 22 ; application of a standard teleost allometric scaling coefficient 17 would raise this to 145 μ mol g −1 h −1 for a theoretical 3-g tuna. The previous record for highest MO 2(max) in a teleost appears to be 164 μ mol g −1 h −1 in 0.019-g larvae of the damselfish Chromis atripectoralis swimming at a Ucrit > 30 body lengths sec −1 at 30 °C 23 ; this would scale to only 57 μ mol g −1 h −1 for a theoretical 3-g larva.
Perspectives. At least in part, this exceptional metabolic performance is explained by the very high area, thin diffusion distance, and high morphometric diffusing capacity for O 2 (D O2 ) in the gills of Alcolapia grahami; D O2 is twice as high in the SWHS fish as in the FSL fish, and is second only to that in tuna on a mass-specific basis 14 . Mitochondrial respiration rates are also considerably higher than in other fish, but only to the extent expected because of the higher temperature 15 . Other key parts of the O 2 delivery system (cardiac output, blood O 2 characteristics, tissue vascularization) have yet to be examined. The metabolic differences observed could be    ± 1 SEM (N = 5-7). The overall effects of population, temperature, and swimming speed (3-way ANOVA) are all significant (P < 0.05); interaction effects are not significant. Also shown (as stars) are the routine MO 2 values measured in the field for freshly caught fish (SWHS, N = 15, at 41 °C; FSL, N = 10, at 33 °C). phenotypic, genotypic or both. At present, it is unclear whether the FSL and SWHS populations are genetically distinct [5][6][7] . With respect to the current debate about the O 2 -and capacity-limited thermal tolerance (OCLTT) hypothesis 24,25 , the pattern of aerobic metabolism in FSL fish (Fig. 6) does not fit the idealized OCLTT model. Nevertheless, encroachment of routine MO 2 on aerobic scope coupled with declining performance at the highest temperature (39 °C) is in accord with the theory. The pattern for SWHS fish may or may not fit the model because aerobic scope continued to increase to 39 °C (Fig. 6); experiments closer to lethal temperatures ( Table 2) will be required to see if aerobic scope eventually declines as Ct max is approached. The SWHS tilapia lives in a fluctuating thermal environment which comes within 3 °C of this Ct max on a daily basis (Fig. 1, Table 2), and is clearly at the mercy of ambient air temperatures. Already classified by IUCN 26 as "threatened (vulnerable)", it will be a bellwether organism for studying the impacts of climate change. All these are exciting areas for future investigation on this unique teleost athlete.

Methods and Materials
Fish Capture and Habitat Characterization. Research was performed under a research and ethics clearance permit (NCST/RR1/12/1/MAS/99) from the National Council for Science and Technology (NCST Kenya), and with the permission of the Magadi Soda Foundation. Experiments were carried out in accordance with the approved guidelines of NCST Kenya and experimental protocols were approved by the Animal Use and Ethics Committee of the Faculty of Veterinary Medicine, University of Nairobi. All foreign researchers were licensed by NCST and formally appointed as visiting researchers at the University of Nairobi. Collections were made under permission from the Dept. of Fisheries, Ministry of Fisheries (Kenya). All surviving fish were returned to their original collection sites.
Adult Magadi tilapia, Alcolapia grahami (2.97 ± 0.14 g, 6.41 ± 0.12 cm fork length) were collected by beach seine from two sites at the edge of Lake Magadi, Kenya, in July and August, 2013: Fish Spring Lagoons (FSL) (GPS coordinates = 1°53′ 30.2″ S, 36°18′ 09.9″ E) and South West Hot Springs (SWHS) (2°00′ 04.0″ S, 6°13′ 55.2″ E). Water chemistry at the two sites (Table 1) was determined by previously described methods 5 . The daily cycle of water temperature, pH, and dissolved O 2 was continuously recorded for 24 h at the two sites (FSL, Aug. 1; SWHS, Aug. 5, 2013) using a multi-parameter meter and probe system (HI 9828, Hanna Instruments, Woonsocket, RI, USA). Fish were studied either shortly after capture at lakeside or after transport by vehicle (FSL = 15 min; SWHS = 90 min), together with water from the two sites, to a field laboratory set up in a classroom of Magadi Secondary School.

Measurements of Critical Temperatures.
Upper (Ct max ) and lower (Ct min ) critical temperatures were determined by a standardized protocol 16 in which temperature was either increased or decreased at a linear rate of 0.3 °C min −1 and loss of equilibrium was the endpoint. Temperature was monitored continuously by a Symphony SP70P probe and meter (VWR, Radner, PA, USA) referenced to a precision thermometer serial 210620, traceable to NIST standards (H-B Instrument, Trappe, PA, USA). Ct measurements were made either on freshly collected fish or on fish which had been held for 4 days at 33 °C in their source water in the laboratory, and were designated as "field" or "lab" respectively. During the 4-day holding, the fish were fed to satiation daily with US Sera Goldy Colour Spirulina (Sera North America Inc., Montgomeryville, PA, USA). Water was changed daily.

Respirometry.
After capture, fish for respirometry experiments were held overnight without feeding in their appropriate water at 33 °C in the laboratory. Gut clearance occurred during this time. Experiments commenced the next morning, and were always performed using water from the appropriate source. In these trials, O 2 was  5-7). The overall effects of population (2-way ANOVA) are significant (P < 0.05) for all three parameters, whereas those of temperature are significant only for MO 2(max), but there are significant interaction effects for MO 2(min) and aerobic scope. Asterisks indicate significant differences (P < 0.05) between FSL and SWHS fish at the same temperature. measured using a portable WTW Oxi325 Oximeter (Weilheim, Germany); urea-N determinations and calculations of O 2 consumption (MO 2 ) and urea-nitrogen excretion (M Urea-N ) rates were performed as described previously 10,11 . This species is capable of supplementary air-breathing via a physostomous swimbladder 4,12,13 , but access to air was not provided in any of the respirometry experiments. Temperature coefficients (Q10 values) 17 for MO 2 and M Urea-N were calculated based on group means. Fish weights and body lengths (nose to fork of caudal fin) were measured after completion of respirometry experiments.
Routine rates of MO 2 and M Urea-N were measured in amber 500-ml bottles ("Tusker chambers") 10 fitted with aeration devices. In the laboratory, fish were allowed to settle for 1 h in the chambers at the chosen test temperature (25, 32, or 39 °C), then a water sample was taken for urea-N analysis. A final sample was taken after 0.3-3 h, depending on temperature. The respirometer was sealed in the middle of the flux period for a sufficient time to allow depletion of O 2 to no less than 60% saturation, and then aeration was resumed. Blanks without fish were run simultaneously. In the field, a similar procedure was used to measure routine rates, but freshly collected fish were employed, and the respirometers were incubated directly in the lake so as to maintain ambient temperature, which was 33 °C in FSL and 41 °C in SWHS at the time of these measurements.
Swimming respirometry was performed using the 3.2-L Blazka-style swim tunnels described by Wilson et al. 20 which were calibrated using an Onicon flowmeter, model F-1100 (Clearwater, Florida, USA). The four respirometers were submerged, in pairs, in two temperature-controlled baths (200-L) filled with the appropriate water. In each run, three were used for fish swimming, and the fourth as a blank at the same velocities. Each was fitted with an aeration device. Fish were allowed to settle for 1 h at about 3 cm sec −1 , and then the respirometer was flushed (0.25 h), and the first test velocity was set to 7 cm sec −1 for a 1-h swimming period. The respirometer was then flushed again (0.25 h), during which time the velocity was increased to 14 cm sec −1 , with subsequent parallel 1-h swimming periods at 21 and 28 cm sec −1 , each followed by 0.25-h flushes. Thereafter, further progressive speed increments of 7 cm sec −1 were applied for 0.5-h swimming periods until the fish exhausted. These 7 cm sec −1 increments represented about 1.1 body lengths sec −1 (typical fish length = 6.0-6.7 cm). Exhaustion was defined as failure of the fish to leave the rear screen and start swimming again, after the velocity had been briefly stopped and restarted three times. The exact time of the failure was recorded and used in the calculation of critical swimming speed (U crit ) by the method of Brett 19 . Only the first four 1-h periods were used for respirometry, as this duration was necessary for accurate determination of M Urea-N . In the middle of each period, aeration was suspended and the respirometer was sealed for MO 2 determination, using a duration in which O 2 depletion occurred to no less than 60% saturation. The subsequent periods were used only to assess swimming performance. This protocol was adopted to fit into a 9-h period (daylight hours only) in light of security concerns (bandits) when travelling to and from our laboratory. Data Analysis. For each fish, MO 2 was plotted against swimming speed. As with other high performance fish swimming at high temperature 22 , the relationships were better described by linear rather than exponential relationships for most fish (33 out of 37), so linear regression was used throughout to predict MO 2(min) at 0 body lengths sec −1 and MO 2(max) at U crit , with the difference representing aerobic scope. Data have been expressed as means ± 1 SEM (N). Data were transformed as required to achieve normal distribution and homogeneity of variance prior to analysis by three-way ANOVA (population x temperature x swimming speed) or two-way ANOVA (population x temperature) as appropriate, followed by the Holm-Sidak post hoc test (P < 0.05) to identify individual differences.