Ion Replacement as a Buoyancy Mechanism in a Pelagic Deep-Sea Crustacean

Pelagic marine invertebrates use a variety of methods to reduce their densities and achieve neutral or positive buoyancies, thereby providing these organisms with energetically conservative means of maintaining their positions in the water column. The more common adaptations found in pelagic invertebrates include reducing overall density via an increase in water content, a reduction in exoskeleton calcification, an increase in lipid content and the use of gas-filled spaces. Another approach, found in pelagic squids (Clarke et al. 1979), dinoflagellates (Kahn & Swift, 1978), diatoms (Gross & Zeuthen, 1948) and tunicate eggs (Lambert & Lambert, 1978), is the reduction of fluid density by ion replacement. In one group of pelagic cephalopods, the cranchiid squids, NH₄⁺ replaces virtually all the Na⁺ present in body fluids, and results in a lift of more than 17 mg ml⁻¹ (Denton et al. 1969; Clarke et al. 1979). Although ion replacement has not been documented for any crustacean, it has been suggested for the pelagic deep-sea shrimp Notostomus sp. (Oplophoridae) and the deep-sea ostracod Gigantocypris sp. (Herring, 1973, citing Denton, 1971).

Although pelagic crustaceans are generally streamlined and negatively buoyant, the positively buoyant shrimp Notostomus gibbosus is a dramatic exception (Fig. 1). N. gibbosus possesses a bulky, dorsally expanded carapace which is fluidfilled. Although one species of Notostomus has a high water content (91·3%) and reduced exoskeleton calcification, the lipid content is very low (Childress & Nygaard, 1974), indicating that buoyancy is achieved by some other mechanism. Our initial measurements of the carapace fluid using ammonia electrodes (HNU model ISE-10-10-00) showed high concentrations of NH₄⁺ (up to 500mmoll⁻¹). This, together with the unusual morphology of N. gibbosus, led us to examine the possibility that the greatly expanded carapace acts as a buoyancy chamber, with reduced fluid density providing lift to the animal via ion replacement.

Although the potential contribution of ion partial molal volume to density has been mentioned by several authors (Denton et al. 1969; Denton, 1971; Mac-L Donald, 1975; Kahn & Swift, 1978), it has not generally been discussed in depth.

As a result, textbook discussions of buoyancy via ion replacement have centred on decreasing fluid density by the use of ‘lighter’ ions (i.e. NH₄⁺) to replace ‘heavier’ ions (i.e. Na⁺), leaving the assumption that density changes are due primarily to the differences in ion mass (Marshall, 1979; Hainsworth, 1981; Schmidt-Nielsen, 1983). Our analysis of the very low-density carapace fluid of Notostomus gibbosus, however, revealed the presence of trimethylamine (Me₃NH⁺), a relatively large, heavy ion which replaces more than 25% of the Na⁺ in this fluid. Because of the large mass of Me₃NH⁺, there is little change overall in solute mass in the carapace fluid (33·2 ‰) relative to sea water (34·1 ‰); rather, the dramatic decrease in fluid density is primarily a result of the large, positive partial molal volumes of Me₃NH⁺ and NH₄⁺, and not a function of solute mass. Partial molal volume differences result in changes in the volume of a fluid via the specific ion effects on the water molecules. Ions having positive partial molal volumes disrupt the structure of the water molecules, thereby increasing fluid volume and reducing density. This is the first report of an animal that uses Me₃NH⁺ to reduce fluid density, and the first report of the use of ion replacement for density reduction in a crustacean.

Notostomus gibbosus and other deep-sea pelagic shrimps (Oplophoridae), as well as the mysid Gnathophausia ingens (Dohrn), were collected during July 1987 off the leeward side of the Hawaiian island of Oahu (2TT5-37’ N, 158°15-38’ W) from depths of 400-800m. The deep-sea crustaceans Gigantocypris agassizii (Müller) (Ostracoda) and Cystisoma sp. (Amphipoda) were collected in January 1987 from San Nicolas Basin off southern California (32°50’-33°10’N, 119°00-20’W) from depths of 400-1000m. These animals were captured with a modified Tucker trawl using a thermally insulated cod end (Childress et al. 1978) which protected the animals from surface temperatures. Animals were maintained alive on board ship in individual containers in a refrigerated storage room at 5°C. The mysid Gnathophausia longispina (Sars) was captured with an Isaacs-Kidd midwater trawl at Southeast Hancock Seamount (29°50’N, 179°10’E) from depths of 100-150 m, and live specimens were bled immediately after capture. Only samples from live, undamaged animals were used in these experiments.

To examine the possibility of density reduction via ion replacement, we first estimated the fraction of body mass present as carapace fluid. Live Notostomus gibbosus were weighed at sea on a precision shipboard balance system (Childress & Mickel, 1980), the carapace was then cut at the dorsal margin to drain the fluid, and the animal was reweighed.

Relative buoyancy represents the weight which 1 g wet mass of animal would have to support to maintain its position in the water column (Childress & Nygaard, 1974). Negative values for relative buoyancy indicate that an animal will float. To determine the relative buoyancies (mg wet mass in sea water/g wet mass in air) of intact Notostomus gibbosus at 1 atmosphere (101-3 kPa), three live animals were anaesthetized with N₂ gas, weighed in sea water at 5°C (34·1‰, 1 atmosphere), then weighed in air. After draining the carapace fluid, the animal was reweighed in sea water and in air.

To determine the pH of carapace fluid and blood of Notostomus gibbosus, samples were removed from the animals with a syringe and immediately introduced into a water-jacketed chamber containing a Markson electrode (model GM 989) calibrated with precision buffers (Radiometer S1500 and S1510). The electrode was stabilized with carapace fluid or blood for several minutes before each pH measurement was made. Carapace fluid was sampled by inserting a syringe into the carapace chamber at the dorsal margin; blood samples were obtained from the ventral sinus for all shrimps. Discrete samples of carapace fluid and blood of N. gibbosus, N. elegans (A. Milne-Edwards) and Meningodora mollis (Smith) (a representative of the genus most closely related to Notostomus), and blood samples from the following shrimps were frozen for 1-3 months at −70°C for later analysis: Acanthephyra curtirostris (Wood-Mason), A. acutifrons (Bate), A. smithi (Kemp), Oplophorus gracilirostris (A. Milne-Edwards) and Systellaspis debilis (A. Milne-Edwards) (all oplophorids), and the mysids Gnathophausia ingens and G. longispina. Body fluids from the pelagic deep-sea ostracod Gigantocypris agassizii and the amphipod Cystisoma sp. were sampled via syringe and stored in a similar fashion.

Ion concentrations were analysed with single-column ion chromatography (Wescan cation and anion columns) in an HPLC system with conductimetric detection for the following ions: Na⁺, K⁺, NH₄⁺, Me₃NH⁺, Ca²⁺, Mg^2-1-, SO₄^2- and Cl⁻. Samples from individual animals were deproteinated with methanol (1:1), and then centrifuged; a 1:10 dilution of the resultant supernatant was analysed for ion composition. Na⁺, NH₄⁺, K⁺ and Me₃NH⁺ were separated on a cation column using a 2·3 mmol 1⁻¹ nitric acid eluent (Ultrex HNO₃, Baker Chemical Co.) pumped at 2·0 ml min⁻¹; Ca²⁺ and Mg²⁺ were measured on the same column with an ethylene diamine eluent (0·033mil⁻¹) at pH6-l, pumped at 2 ml min⁻¹. Cl⁻ and SO₄^2- were determined using an anion column with a 0·004 mol I⁻¹ potassium hydrogen phthalate eluent at pH 4-7, pumped at 2 ml min⁻¹.

The presence of high levels of Me₃NH⁺ in chromatograms from carapace fluid and blood of several species was unexpected, and the identity of this peak was verified by mass spectrometry. A deproteinated, diluted sample of Notostomus gibbosus carapace fluid was analysed for Me₃NH⁺ using a Vacuum Generator Instruments model VG70-250HS double-focusing mass spectrometer.

Carapace fluid density was calculated for Notostomus gibbosus using the mean ion concentration from nine individually measured samples (Table 1), together with the mass and partial molal volume of each ion (Table 2). Additionally, a pooled sample of carapace fluid taken from three N. gibbosus was weighed on an electronic balance to measure directly fluid density relative to sea water. The ion concentration of this pooled sample was measured by single-column ion chromatography and the fluid density then calculated from these data for comparison with the density of the mean of nine individual samples.

The fluid within the expanded carapace of Notostomus gibbosus represented 42·9 ±4·6% (S.D., N = 9) of the animal’s wet mass, and the range was 39·6-52·9%. There was no apparent relationship between animal size and the percentage of body mass as carapace fluid. At 5°C, the pH of the carapace fluid was very low (6·59 ± 0·08, N = 8) and the blood pH (7·52 ± 0·06, N = 8) was also unusually low for crustaceans (Mangum, 1980). At these low pH levels, virtually all the ammonia would be present in the ionized NH₄⁺ form.

The relative buoyancies of three Notostomus gibbosus are shown in Fig. 2. These animals were all slightly buoyant when intact (relative buoyancies = −1·36, −1·59 and −2·65mgg⁻¹), and sank when their carapace fluid was removed (relative buoyancies = 7·15, 7·50 and 6-24mgg⁻¹). The difference in relative buoyancy between intact and drained animals represents the amount of lift which the fluid must provide to account for the observed buoyancy. Values for drained animals, however, represent a conservative estimate because of the difficulty of removing all the trapped air bubbles once the carapace is drained.

The ion compositions of fluids are presented in Table 1. Although the carapace fluid and blood of Notostomus gibbosus were both isosmotic with sea water, their ion compositions differed dramatically. Reduced concentrations of Na⁺ and high concentrations of NH₄⁺ and Me₃NH⁺ were present in both fluids but were most evident in the carapace fluid. In addition, the levels of all divalent ions were greatly reduced. Ion compositions of the carapace fluids and bloods of N. elegans and Meningodora mollis were also very different from sea water, and demonstrated ion replacements similar to those in N. gibbosus. The remaining oplophorid shrimps and the mysids had bloods similar in ionic composition to sea water, although very low levels of NH₄⁺ and Me₃NH⁺ were present in some (Table 1).

Using the mean ion composition for Notostomus gibbosus carapace fluid from nine individual samples (Table 1), the density of this fluid (1·0103 g ml⁻¹) was calculated (Table 2). The high concentrations of NH4⁺ provided the greatest contribution to lift (9-1 mg ml⁻¹ fluid relative to sea water having a density of l·028gml⁻¹), and Me₃NH⁺ gives an additional lift of 1·8 mgml⁻¹ carapace fluid. The reduction in divalent ions provided 5·4 mg ml⁻¹ lift, and the total lift contribution calculated for carapace fluid was 17·7 mg ml⁻¹. The density of a pooled sample of carapace fluid (l-Ollgml⁻¹) was also measured directly by weighing a known volume and represented a lift of 17·0 mg ml⁻¹ relative to sea water. For comparison, the density calculated from the ion composition of this pooled sample was l·0112gml⁻¹ (a lift of 16·8mgml⁻¹), compared with 1·0103 g ml⁻¹ calculated from the mean carapace fluid given above. These three separate analyses of carapace fluid are in excellent agreement, varying by only 0·9 mg ml⁻¹. The ion composition of the blood of N. gibbosus results in a lift of 12 mg ml⁻¹ when calculated from the mean ion composition given in Table 1.

Marine crustacean bloods are isosmotic with sea water and are generally similar to sea water in ionic composition, although levels of divalent ions are often reduced (Mangum, 1983; Bridges et al. 1984; Sanders et al. 1988). In Notostomus gibbosus, body fluids are indeed isosmotic with sea water, but ion compositions differ dramatically (Table 1). For example, the Na⁺ concentration in the carapace fluid of N. gibbosus is nearly 90% lower than in sea water, and very high levels of NH₄⁺ and Me₃NH⁺ are present (Table 1). Similarly, our methods detected no Ca²⁺ in this fluid, and very low levels of Mg²⁺ and SO₄^2- (Table 1). These changes in ion composition influence density in two ways: first, a change in density can result from the differences in solute masses alone. Second, and most significant here, is the change in solution volume and resultant change in fluid density due to the differences in the partial molal volumes of the replacement ions (partial molal volume, V^o = cm³mol⁻¹; Millero, 1972). When ions with a negative partial molal volume are added to a solution, the total volume of the solution is less than that of the separate ions and solvent. The addition of ions with a positive partial molal volume results in a solution having a greater volume than the sum of its components. This volume increase occurs because ions with positive partial molal volumes (‘structure breakers’) disrupt the structure of water in their vicinity, whereas small ions with negative partial molal volumes (‘structure makers’), especially those with multiple charges, cause a constriction in the water structure around them (Home, 1969). For example, if only solute masses are examined in relation to carapace fluid density, the fluid (33-2 ‰) would be nearly as dense as sea water (34·1 ‰) due to the relatively large mass of Me₃NH⁺ (Table 2). The partial molal volumes at 5 °C of Me₃NH⁺ and NH₄⁺, however, are large and positive, and in the carapace fluid result in a lift of 10·9 mgml⁻¹, with an additional 5·4 mg ml⁻¹ lift from a reduction in the concentrations of Ca²⁺, Mg^24- and SO₄^2- (Table 1). Even in an organism that completely replaces Na⁺ with NFL,^−1-, and where no Me₃NH⁺ is present, such as in the cranchiid cephalopods (Denton et al. 1969), only 2·4mgml⁻¹ lift would be explained by solute mass differences alone. (It is possible that Me₃NH⁺ was present in cephalopod fluids, but the methods previously employed to measure NH₄⁺ did not discriminate between the amines present, Denton et al. 1969; Clarke et al. 1979.) In N. gibbosus carapace fluid, the total lift from solute mass and partial molal volume (Table 1) is 17·7 mg ml⁻¹ (Table 2). Similarly, N. gibbosus blood has a reduced level of Na⁺, high levels of NH₄⁺ and Me₃NH⁺, and a reduction in the levels of divalent ions relative to sea water. The calculated lift contribution of the blood in 12 mg ml⁻¹.

The mean relative buoyancy (Fig. 2) of Notostomus gibbosus (-1·87 mgg⁻¹) is lower than that observed in 14 other midwater crustaceans (relative buoyancies of -0·7 to +40·0mgg⁻¹) (Childress & Nygaard, 1974), and a negative relative buoyancy indicates that the animal is positively buoyant. The total lift required to account for the observed relative buoyancy in N. gibbosus can be calculated using the mean relative buoyancies for intact (-1·87mgg⁻¹) and drained (6·96mgg⁻¹) animals, and assuming that sea water has a density of l·028gml⁻¹. A 10-g animal having 42·9% of its wet mass as carapace fluid would weigh 5·71 g drained, with 4·29 g carapace fluid. The volume of 4·29 g of carapace fluid (1·0103 g ml⁻¹, Table 2) is 4·25ml. The overall change in mean relative buoyancy (8-83mgg⁻¹) multiplied by the animal’s mass without carapace fluid (5·71 g) is the total lift (50·4 mg) that must be accounted for in a 10-g animal. Therefore, the carapace fluid (4·25 ml) must be less dense than sea water (1·028gml⁻¹) by 12 mg ml⁻¹ to provide 50-4mg of lift. This requires that the density of the fluid be l·016gml⁻¹

The observed carapace fluid density (l·0103gml⁻¹, Table 2), however, gives a lift of 17·7 mg ml⁻¹, and provides the 10-g animal with 75·2 mg of lift. Using these numbers, the lift contribution of the carapace fluid alone is more than sufficient to explain the observed differences in the relative buoyancies of intact and drained N. gibbosus. This discrepancy, however, may be due to a conservative mass estimate for drained animals. When drained animals are resubmerged to measure their mass in sea water, air bubbles can remain trapped in the carapace making precise mass estimates difficult. High levels of NH₄⁺ and Me₃NH⁺ were also found in the bloods and carapace fluids of the congener Notostomus elegans as well as in Meningodora mollis. No evidence of ion replacement was found in Gigantocypris agassizii or Cystisoma sp. (Table 2), and both these animals have a very high water content (Childress & Nygaard, 1974). Cystisoma sp. fluid, however, is apparently slighly hyposmotic relative to sea water (Table 1). The small amounts of NH₄⁺ and Me₃NH⁺ detected in the bloods of the other oplophorid shrimps may indicate a predisposition of this family for ion replacement as a buoyancy mechanism. The other oplophorid species examined in this study, however, are all negatively buoyant and do not have enlarged carapaces.

The high NH₄⁺ and Me₃NH⁺ concentrations in Notostomus gibbosus might present toxicity problems if the fluids were not confined to specific body compartments which prevented direct contact with other tissues. Evidence for isolation of the carapace fluid in a discrete chamber comes from direct observations in preserved specimens of a membrane which contains the fluid in the bulky, dorsally expanded region of the carapace. The pH values were low for both fluids (blood pH = 7·52, carapace fluid pH −6·59) and were sufficiently acidic to maintain virtually all the ammonium ions in the NH₄⁺ ionized form. Additionally, although both the carapace fluid and blood have low pH values, these values are significantly different from one another (99% confidence interval, paired Student’s t-test); the ionic compositions of the carapace fluid and blood also differ significantly for the monovalent cations, Mg²* and SO₄^2- (95% confidence interval, paired Students’s t-test), and these differences support the separate nature of these two fluids.

Although other extracellular body fluids were not measured, both the carapace fluid and the blood of Notostomus gibbosus have unusual ion compositions. High levels of ammonia were not found in the blood of cranchiid squids and other cephalopods (Denton et al. 1969; Denton, 1971; Clarke et al. 1979), and ammonia concentrations as high as those found in coelomic fluids and other body compartments would be sufficient to impair nervous function in these animals if these fluids were not isolated in separate chambers. The ammonia levels found in the blood of N. gibbosus (Table 1) would interfere with the ability of nerve fibres to conduct nerve impulses if found in a squid (Denton, 1971); we do not know how this shrimp handles the potential toxicity of the high ammonia concentrations found in the blood.

Animals which are negatively buoyant must either swim or sink; a positively buoyant animal such as Notostomus gibbosus must swim to stay at depth. This raises the question of the adaptive significance of positive buoyancy in a midwater crustacean. It may be a predator avoidance strategy in that most deep-sea crustaceans respond to a predator by rapidly swimming to escape, or by ceasing swimming and sinking. In maintenance conditions, when startled, N. gibbosus generally stops swimming immediately and slowly floats upwards. This novel behaviour may well enable the shrimp to escape predators which would ‘expect’ disturbed prey to sink or swim away.

The use of the ammonium ion to provide buoyancy for Notostomus gibbosus is the first report of a crustacean that uses ion replacement to reduce fluid density. Whereas high concentrations of NH₄⁺ (a relatively small ion) have been found to contribute to positive buoyancy in other pelagic organisms (Kahn & Swift, 1978; Lambert & Lambert, 1978; Clarke et al. 1979), the use of a large ion with a large positive partial molal volume (Me₃NH⁺) to replace Na⁺ is unique to this group.

This research was supported by NSF OCE-8500237 to JJC and an NSF Predoctoral Fellowship to NKS. We thank the captain and crew of the R/V New Horizon for their help in animal collection, H. Webb for mass spectrometry, and K. Johnson and B. Robison for critical review of this manuscript.