We measured red cell space with 51Cr-labeled red blood cells, and dextran space with 500 kDa fluorescein-isothiocyanate-labeled dextran (FITC-dextran), in two groups of yellowfin tuna (Thunnus albacares). Red cell space was 13.8±0.7 ml kg−1 (mean ± S.E.M.) Assuming a whole-body hematocrit equal to the hematocrit measured at the ventral aortic sampling site and no significant sequestering of 51Cr-labeled red blood cells by the spleen, blood volume was 46.7±2.2 ml kg−1. This is within the range reported for most other teleosts (30–70 ml kg−1), but well below that previously reported for albacore (Thunnus alalunga, 82–197 ml kg−1). Plasma volume within the primary circulatory system (calculated from the 51Cr-labeled red blood cell data) was 32.9±2.3 ml kg−1. Dextran space was 37.0±3.7 ml kg−1. Because 500 kDa FITC-dextran appeared to remain within the vascular space, these data imply that the volume of the secondary circulatory system of yellowfin tuna is small, and its exact volume is not measurable by our methods. Although blood volume is not exceptional, circulation time (blood volume/cardiac output) is clearly shorter in yellowfin tuna than in other active teleosts. In a 1 kg yellowfin tuna, circulation time is approximately 0.4 min (47 ml kg−1/115 ml min−1 kg−1) compared with 1.3 min (46 ml kg−1/35 ml min−1 kg−1) in yellowtail (Seriola quinqueradiata) and 1.9 min (35 ml kg−1/18 ml min−1 kg−1) in rainbow trout (Oncorhynchus mykiss). In air-breathing vertebrates, high metabolic rates are necessarily correlated with short circulation times. Our data are the first to imply that a similar relationship occurs in fishes.
The relationships between metabolic rate and the respiratory, circulatory and cellular mechanisms governing oxygen transport from the respiratory medium to the tissues in air-breathing vertebrates have been the subject of intensive scrutiny over the last two decades (e.g. Taylor et al. 1996 and the related papers in the same volume). Many of the lessons learned from these studies appear to be applicable to fishes (Mathieu-Costello et al. 1992, 1995, 1996; Moyes et al. 1992), although a complete set of supporting physiological and morphometric data for a given fish species is often wanting. Regardless, it is clear that tunas (family Scombridae, tribe Thunnini) are high-energy-demand teleosts and that both their standard and maximum rates of oxygen consumption exceed those of other active teleosts (e.g. salmonids) by at least fourfold (Brill, 1987; Boggs and Kitchell, 1991; Dewar and Graham, 1994; Korsmeyer et al. 1996a). Adaptations supporting the high metabolic rates of tunas include large gill surface area, high cardiac output, elevated hemoglobin concentrations and the ability to maintain muscle temperature significantly above ambient (Bushnell and Jones, 1994; Dickson, 1996). Other adaptations that could potentially allow tunas to achieve high rates of oxygen consumption are either a significantly elevated blood volume (Korzjynew and Nikolskria, 1951) or a short circulation time (i.e. blood volume/cardiac output) (Coulson, 1986). Using radio-iodinated bovine serum albumin, Laurs et al. (1978) reported blood volumes of albacore (Thunnus alalunga) ranging from 82 to 197 ml kg−1. In contrast, blood volumes for 25 species of elasmobranch, teleost and holostean fishes listed by Tort et al. (1991) range from 18 to 80 ml kg−1, although blood volumes of most teleosts lay within a more restricted range (30–70 ml kg−1, Itazawa et al. 1983; Olson, 1992). In albacore, elevated blood volumes would extend circulation times even in the face of the elevated cardiac outputs of tuna (Lai et al. 1987; White et al. 1988; Bushnell and Brill, 1992; Jones et al. 1993), implying that the former is more important than the latter for achieving exceptionally high metabolic rates. However, based on the dilution of iodocyanin green dye (‘cardio-green’), Bushnell (1988) estimated blood volumes in skipjack (Katsuwonus pelamis) and yellowfin (Thunnus albacares) tunas to be approximately 50 ml kg−1. If correct, the impact of these data on our concepts of circulation time, oxygen and substrate delivery, and their relationship to metabolic rate is considerable because of the interrelationship between these variables (Coulson et al. 1977; Coulson and Herbert, 1984; Coulson, 1986; Bushnell and Brill, 1992; Korsmeyer et al. 1996a,b).
Olson (1992), in his excellent and comprehensive review on the blood volumes of fishes, identified two trends. First, Osteichthyes have the lowest blood volumes of any vertebrate and, second, estimated blood volumes depend on the techniques employed. Larger blood volumes are usually measured with plasma volume indicators (e.g. Evans Blue or radio-iodinated albumins) than are obtained with 51Cr-labeled red blood cells. Problems associated with dyes or protein markers arise because they may not remain within the primary circulatory system, instead entering the secondary circulatory system (a parallel ‘lymph-like’ circulatory system in teleosts; Steffensen and Lomholt, 1992; Olson, 1996), the interstitial space or even be excreted. The use of 51Cr-labeled red blood cells can, however, also result in an overestimate of red cell space and blood volume because labeled red blood cells may be selectively sequestered by the spleen (Duff et al. 1987). Moreover, the accuracy of both techniques is dependent on measured hematocrit because these data are needed to calculate blood volume from the measured plasma volume or red cell space (Jones, 1970; Fairbanks et al. 1996). It is generally assumed that whole-body hematocrit and hematocrit measured at the sampling site (usually a large blood vessel) are the same, although they can differ considerably (Albert, 1971; Gingerich and Pityer, 1989; Olson, 1992; Fairbanks et al. 1996).
Consequently, we decided to quantify red cell space and plasma volume of yellowfin tuna using two independent techniques. The former was measured directly using 51Cr-labeled red blood cells and the latter by dilution of large (500 kDa) fluorescein-isothiocyanate-labeled dextran (FITC-dextran). We chose 500 kDa FITC-dextran as a marker in an attempt specifically to reduce complications associated with non-binding of a dye to plasma proteins and extravasation (Gingerich and Pityer, 1989; Tort et al. 1991).
Materials and methods
Yellowfin tuna Thunnus albacares (Bonaterre) were purchased from local commercial fishermen and maintained at 25±1 °C in circular outdoor tanks (approximately 8 m in diameter and 1.5 m deep) at the Kewalo Research Facility (National Marine Fisheries Service, Honolulu Laboratory). Fish were fed chopped squid daily, but were not fed for approximately 20 h before use in an experiment to allow sufficient time for gut clearance (Magnuson, 1969).
Anesthesia and surgical procedures were as described previously (Bushnell and Brill, 1992). Briefly, fish were dip-netted from their holding tank and immediately placed in a plastic bag containing 1 g l−1 of tricaine methanosulfonate (MS222) buffered with an equal molar concentration of sodium bicarbonate. After initial anesthesia, fish were rushed into the laboratory, placed ventral-side up in a soft chamois sling and ventilated with oxygenated sea water (temporarily chilled to 21–22 °C) containing 0.1 g l−1 of buffered tricaine methanosulfonate. A 20 gauge, 3.2 cm Instyle Vialon intravenous catheter (Becton Dickinson Vascular Access, Sandy, Utah, USA) was introduced into the ventral aorta under manometric guide and connected to a 20 cm length of polyethylene tubing (PE 160). Fish were turned upright, a 20 gauge hypodermic needle was placed into the neural canal immediately posterior to the skull and used to inject 0.1–0.2 ml doses of 4 % lidocaine. This procedure blocks spinal motor nerves and prevents excessive tail movements, but leaves all cranial nerves and cardiorespiratory function intact (Bushnell and Brill, 1991). Fish were then placed in front of a pipe delivering approximately 30–35 l min−1 of oxygenated sea water and were thus able to set their own ventilation volume by adjusting mouth gape. Throughout the duration of the experiment, fish were kept sedated by repeated 0.1–0.3 ml intramuscular injections of the steroid anesthetic Saffan (alphaxalone, Glaxovet, Harefield, Uxbridge, UK), a highly effective anesthetic in fish (Oswald, 1978). At the conclusion of an experiment, animals were killed with an overdose of sodium pentobarbital, injected via the indwelling ventral aortic catheter, and weighed. Data were obtained from seven fish using 51Cr-labeled red blood cells (body mass 0.750–2.420 kg) and from seven fish using 500 kDa FITC-dextran (body mass 0.865–3.325 kg).
Measurement of red cell space
Red blood cells labeled with 51Cr (Amersham Canada Limited, Oakville, Ontario, Canada) were used to measure red cell space. To avoid any complications due to blood clotting, fish were given 1000 i.u. of sodium heparin (0.1 ml of 10 000 i.u. ml−1) via the ventral aortic catheter approximately 1 h after catheterization. Ten minutes later, 2 ml of blood was removed. The blood was mixed with 10 ml of tuna saline (1.17 % NaCl), centrifuged (approximately 50 g for 2 min), and the supernatant discarded. The red blood cells were resuspended in 2 ml of saline, 1.85×106 Bq (50 μCi) of 51Cr was added, and the cells were then gently swirled for 2 h at room temperature (approximately 25 °C). The labeled red blood cells were centrifuged (approximately 50 g for 2 min), washed three times with 14 ml of saline each time, and finally resuspended in 1.5 ml of saline for injection into the same fish from which they were originally withdrawn. The precise volumes injected (mean ± S.E.M., 1.57±0.14 ml) were determined by mass. A sample of the final rinse solution was taken to determine whether any unincorporated 51Cr remained in the red blood cell suspension. The hematocrit of the final injectate was measured, and a 300 μl sample was taken for counting. Because of the difficulties in placing and maintaining a second patent catheter in the dorsal aorta of yellowfin tuna (Brill et al. 1987; Jones et al. 1986; Bushnell, 1988), the ventral aortic catheter was used for both injection of labeled red blood cells and blood sampling. To minimize contamination of blood samples with labeled red blood cells from the initial injectate, approximately 2–3 ml of blood was withdrawn prior to injection of the labeled cells, and this blood used to flush the catheter immediately following injection. Blood samples (500 μl) were withdrawn at 30, 60 and 300 min post-injection and hematocrits were measured. Subsamples were centrifuged (8000 g for approximately 1 min) and the plasma was retained to test for hemolysis of labeled red blood cells. Duplicate 50 or 100 μl whole-blood and plasma samples were mixed with 500 μl of distilled water, and their radioactivity was measured in a Beckman 5500 gamma counter (Beckman Instruments, Fullerton, California, USA).
Measurement of FITC-dextran space
500 kDa FITC-dextran (Sigma Chemical Co., St Louis, Missouri, USA) was used to determine dextran space. Approximately 60–75 mg (12.9±0.9 mg kg−1 body mass) was dissolved in 2–4 ml of saline, and a weighed volume (2–3 ml) was injected via the ventral aortic catheter. To minimize any increase in circulatory system volume caused by the injectate, an equal volume of blood was removed and centrifuged (as above), and the red blood cells were reinjected after the dye. Blood samples (200 μl) were taken at timed intervals (nominally 15, 30, 45, 60, 90, 120, 150, 180, 210, 240, 300, 360 and 420 min post-injection). Samples were centrifuged in micro-hematocrit tubes, 40 μl subsamples of plasma were diluted with 960 μl of saline, and the dye concentration was measured spectrophotometrically (Coleman Junior 6/20A, Perkin Elmer, Maywood, Illinois, USA) at 496 nm.
51Cr-labeled red cells
Red cell space was calculated from the volume of cells injected multiplied by the ratio of 51Cr activity in the injected red blood cells to 51Cr activity in the red blood cells sampled from the fish. 51Cr activity in the red blood cells was calculated by dividing the activity measured in the whole blood by the hematocrit of that sample (expressed as a decimal fraction). Blood volume was obtained by dividing red cell space by the hematocrit (expressed as a decimal fraction). Plasma volume was calculated as the differences between blood volume and red cell space.
The first exponential decay term describes the movement of the marker between the compartments, and the second exponential decay term describes movement of the marker out of the vascular space. Because of the high protein permeability of fish capillaries (Wardle, 1971; Hargens et al. 1974), and lacking any data to the contrary, we initially assumed that 500 kDa FITC-dextran would mix into and leak from the circulatory system of yellowfin tuna in a manner similar to the behavior of plasma-borne markers used in other fishes. Consequently, we initially fitted plasma dye concentration data to the double-exponential decay equation (equation 1).
The dextran space of the ‘fast-mixing’ compartment was calculated by dividing the quantity of FITC-dextran injected by the concentration of FITC-dextran in the plasma at t=0 (i.e. cf→s+cout, see Fig. 1). The combined dextran space of the ‘fast-mixing’ plus ‘slow-mixing’ compartments was calculated by dividing the quantity of dye injected by the dye concentration at t=∞ (i.e. cout, see Fig. 1). The dextran space of the ‘slow-mixing’ compartment is the mean of the difference between the two.
Data were analyzed using a one-way analysis of variance (ANOVA) for treatment groups (red cell space and blood volume calculated from samples taken 30, 60 and 300 min post-injection) or by pairwise comparisons (plasma volume, red cell space and blood volume measured with 51Cr-labeled red blood cells and their equivalents measured with 500 kDa FITC-dextran) (Sigma Stat, Jandel Scientific, San Rafael, California, USA), and 5 % was taken as the fiducial limit for a significant difference. All data presented in the text and tables are means ± standard error of the mean (S.E.M.).
51Cr-labeled red cells
Body mass, hematocrit, red cell space, blood volume and plasma volume data are summarized in Table 1. Mean 51Cr activity in the final rinse solution was less than 1 % of the activity in the injectate, showing that no significant amount of unincorporated label was given to the fish. Plasma samples also showed no significant amount of radioactivity at any sampling time, indicating no hemolysis of the labeled red blood cells within the fish. Other investigators have observed hemolysis of labeled red blood cells, presumably because blood from one fish was injected into a another fish (Gingerich et al. 1987). We avoided this problem by re-injecting 51Cr-labeled red blood cells into the same fish from which they were withdrawn.
There were no significant differences between mean red cell spaces calculated from samples taken 30, 60 and 300 min post-injection when the data were averaged across fish (see Fig. 1). The 30, 60 and 300 min data were therefore combined, and a single value of red cell space, blood volume and plasma volume was calculated for each fish.
When the FITC-dextran data were fitted to the double-exponential decay equation (equation 1), Rout (i.e. the term representing the rate of loss from the vascular space) was not significantly different from zero (6.24×10−5±5.03×10−3 min−1). The half-time for loss from the vascular space (ln2×Rout−1) was 5.27×1018±3.49×1018 min. These data imply that the marker did not leave the vascular compartment even when plasma concentrations were followed for more than 400 min.
The data from the FITC-dextran experiments are summarized in Table 2. A representative trace of the mixing-in kinetics for 500 kDa FITC-dextran is shown in Fig. 1. In contrast to the 51Cr-labeled red blood cells, which were completely mixed some time before the first sample (30 min post-injection), the half-time (calculated as ln2×Rf→s−1; Riggs, 1963) for the mixing-in process of FITC-dextran was 75±23 min.
The dextran spaces of the ‘fast-mixing’ and ‘slow-mixing’ compartments were 21.6±2.2 ml kg−1 and 15.4±2.4 ml kg−1, respectively. If the dextran space of the ‘fast-mixing’ compartment is assumed to be equivalent to plasma volume within the primary circulatory system, then mean blood volume is the dextran space divided by one minus the hematocrit (expressed as a decimal fraction) or 31.2±3.1 ml kg−1, a value significantly less than the blood volume calculated using the 51Cr-labeled red blood cells (Table 1). If, however, the dextran space of the ‘fast-mixing’ and ‘slow-mixing’ compartments is assumed to be equivalent to plasma volume within the primary circulatory system, then blood volume is 53.7±5.4 ml kg−1, a value not significantly different from the blood volume calculated using the 51Cr-labeled red blood cells.
51Cr-labeled red blood cells
Mean red cell space and blood volumes were not significantly different at 30, 60 or 300 min post-injection (Fig. 1). These data therefore imply either that the spleen of yellowfin tuna did not selectively remove any 51Cr-labeled red blood cells or that removal of labeled cells ceased prior to our first sample. Duff et al. (1987) found that removal of 51Cr-labeled red blood cells by the spleen of rainbow trout (Oncorhynchus mykiss) continued for up to 240 min post-injection, something we did not observe. The relatively short circulation time of the yellowfin tuna (see below) might, however, allow the splenic sequestration of 51Cr-labeled red blood cells to be completed rapidly. Why the spleen selectively accumulates labeled red blood cells is not clear, but it is possible that the incubation, centrifugation and resuspension procedures used during labeling of red blood cells with 51Cr slightly damages a fraction of the cells. It may be these ‘damaged’ cells that the spleen selectively removes from the circulation. More importantly, Duff et al. (1987) calculated that in rainbow trout blood volume would be overestimated by only approximately 15 % as a result of splenic accumulation of 51Cr-labeled red blood cells.
Can the 51Cr-labeled red blood cell method, therefore, be regarded as the ‘gold-standard’ for blood volume determination? Perhaps only up to a point, because any assessment of blood volume is also at the mercy of the measured hematocrit (Fairbanks et al. 1996). Our calculations assume that the hematocrit was the same throughout the vascular space as at our sampling point in the ventral aorta (i.e. whole-body hematocrit = large-vessel hematocrit). In mammals, the hematocrit in the large vessels is greater than that within the remainder of the circulatory system (whole-body hematocrit/large-vessel hematocrit ≈ 0.9) and the difference may be even larger in fish (whole-body hematocrit/large-vessel hematocrit ≈ 0.8) (Albert, 1971; Gingerich and Pityer, 1989; Olson, 1992). If this is also true in yellowfin tuna, then we may have underestimated the actual blood volume by as much as approximately 25 %.
There is good evidence that commonly used plasma-borne indicators (e.g. Evans Blue, radio-iodinated albumin) are lost from the vascular compartments of fishes (Laurs et al. 1978; Duff and Olson 1989; Gingerich and Pityer, 1989; Olson, 1992). Indeed, Nichols (1987) found that in rainbow trout albumin exchanged with the extravascular compartment at a rate of up to 48 % h−1. Extravasation of the marker, dilution into vascular compartments not accessed by labeled red blood cells (e.g. owing to plasma skimming or plasma movement into the secondary circulatory system) and the necessity of extrapolating plasma concentration decay curves back to time zero increase inaccuracies in the calculation of plasma volume and usually result in estimated blood volumes larger than those measured using 51Cr-labeled red blood cells (Olson, 1992). The difference in measured plasma volumes has been used as an estimate of the volume of the secondary circulatory system, a volume to which plasma-borne markers have access but red blood cells do not (Steffensen and Lomholt, 1992).
We therefore expected the dextran space of the ‘fast-mixing’ compartment (Table 2) to have been at least as large as, if not larger than, the plasma volume calculated from the 51Cr-labeled red blood cell data (Table 1) and were surprised when we found just the opposite. What are the anatomical counterparts of the ‘fast-mixing’ and ‘slow-mixing’ compartments in tuna? Given that our data show that 500 kDa FITC-dextran remains within the vascular space, there are two plausible answers. First, the former is equivalent to the plasma volume within the primary circulatory system and the latter to the plasma volume within the secondary circulatory system; second, both compartments are within the primary circulatory system.
Analysis of similarly shaped plasma concentration decay curves of radio-iodinated albumin in rainbow trout obtained by Steffensen and Lomholt (1992) suggests that the dextran spaces of the ‘fast-mixing’ and ‘slow-mixing’ compartments are indeed equivalent to the plasma volumes within the primary and secondary circulatory systems in tuna. Therefore, our FITC-dextran data show that blood volume in yellowfin tuna is 31.2±3.1 ml kg−1 and that the volume of the secondary circulatory system is 15.4±2.4 ml kg−1, or approximately 50 % of blood volume. In contrast, the volume of the secondary circulatory system in rainbow trout has been estimated to be 150 % of blood volume (Steffensen and Lomholt, 1992). Dewar et al. (1994) have described a secondary circulatory system associated with the central vascular heat exchangers in skipjack tuna. Although this system has been well described anatomically in other fishes (e.g. Vogel and Claviez, 1981; Steffensen et al. 1986), its volume and function(s) remain uncertain (Wardle, 1971; Steffensen and Lomholt, 1992; Olson, 1996).
If, however, the dextran spaces of the ‘fast-mixing’ and ‘slow-mixing’ compartments are equivalent to the total plasma volume within the primary circulatory system, then plasma volume and blood volume measured using 51Cr-labeled red blood cells (Table 1) and using FITC-dextran (Table 2) are not significantly different. When analyzed under this set of assumptions, our data imply that the secondary circulatory system in yellowfin tuna is so small that its volume is unmeasurable by our methods and that there is no significant loss of 500 kDa FITC-dextran from the vascular system of yellowfin tuna. If there were loss of the indicator into the interstitial space, then the dextran space would have significantly exceeded the plasma volume calculated from the 51Cr-labeled red blood cell data. In contrast, when 500 kDa FITC-dextran was used in an attempt to measure plasma space in rainbow trout and cod (Gadus morhua), plasma concentrations continued to decline over the 12 h for which they were followed and never reached equilibrium (Bushnell et al. 1997). It must be noted, however, that the mixing-in of 51Cr-labeled cells was complete some time prior to the first sample (30 min post-injection), whereas the mixing-in of the 500 kDa FITC-dextran required significantly longer (half-time 75±23 min). The difference in equilibration kinetics between the 51Cr-labeled red blood cells and FITC-dextran and yet the agreement between plasma volumes measured with the two indicators are unexpected if both the ‘fast-mixing’ and ‘slow-mixing’ compartments are indeed within the primary circulatory system. At present, we have no explanation for this apparent paradox.
Nevertheless, our estimates of yellowfin tuna blood volume (31–47 ml kg−1) are at the middle to lower end of the range reported for the majority of teleosts (30–70 ml kg−1, Itazawa et al. 1983; Olson, 1992) and well below the blood volumes reported for albacore (mean 132 ml kg−1, range 82–197 ml kg−1; Laurs et al. 1978). The reasons for the differences in the blood volumes of the two tuna species are unclear but may, in part, be due to species differences, methodological differences or both. Laurs et al. (1978) employed radio-iodinated bovine serum albumin and thus had to extrapolate indicator dilution curves back to time zero. They were also forced to work on board ship using animals immediately after they had been landed by hook and line. The elevated hematocrit (53–58 %) compared with normal values in rested tunas (30–41 %, Bushnell and Jones, 1994) imply the fish were still recovering from the stress of capture.
Relationship between blood volume and metabolic rates
After measuring blood volumes in albacore ranging from 82 to 197 ml kg−1, Laurs et al. (1978) concluded that large blood volumes are characteristic of high-energy-demand teleosts. Sleet and Weber (1983), however, found that the blood volume of buffalo sculpin (Enophrys bison) was 70–90 ml kg−1, which is at the upper end of the range of blood volumes recorded in fishes, except albacore. Because the buffalo sculpin is a relatively inactive benthic species, they concluded that blood volumes in teleosts are not necessarily correlated with activity. Moreover, Coulson et al. (1977), Coulson and Herbert (1984) and Coulson (1986) present data showing that metabolic rates in air-breathing vertebrates are directly correlated with short circulation times (i.e. blood volume/cardiac output), rather than with elevated blood volumes or cardiac outputs per se. In other words, rates of energy production are set by rates of metabolic substrate delivery which are, in turn, set by the number of circulatory cycles occurring per unit time (Coulson, 1986).
Our estimate of blood volume in yellowfin tuna, therefore, allows us to determine circulation time and to compare this with estimates for other active teleosts. In 1 kg fish and measured under similar circumstances, circulation time is approximately 0.3–0.4 min (31–47 ml kg−1/115 ml min−1 kg−1) in yellowfin tuna compared with 1.3 min (46 ml kg−1/35 ml min−1 kg−1) in yellowtail (Seriola quinqueradiata) (Yamamoto et al. 1980, 1981) and 1.9 min (35 ml kg−1/18 ml min−1 kg−1) in rainbow trout (Kiceniuk and Jones, 1977). The circulation time of yellowfin tuna is, therefore, approximately one-third to one-sixth that seen in these other active teleosts, a ratio similar to that seen for their metabolic rates (Yamamoto et al. 1981; Bushnell and Jones, 1994). These data imply that, when examined over reasonable ranges of temperature and blood oxygen-carrying capacities (i.e. excluding species such as the Antarctic hemoglobinless icefish Chaenocephalus aceratus), short circulation times are correlated with high metabolic rates in teleosts as they are in air-breathing vertebrates.
This paper was funded in Cooperative Agreement NA37RJ0199 from the National Oceanic and Atmospheric Administration with the Joint Institute for Marine and Atmospheric Research, University of Hawaii, by a research grant from NSERC of Canada to D.R.J., and the National Marine Fisheries Service (Honolulu Laboratory, Southwest Fisheries Science Center). The views expressed herein are, however, those of the authors and do not necessarily reflect the views of NOAA or any of its subagencies. The authors thank Dr Gordon Grau (Hawaii Institute of Marine Biology, University of Hawaii) for generously providing access the the gamma counter and an anonymous reviewer for providing criticisms and helpful suggestions on an earlier draft of this manuscript.