Critical oxygen levels and metabolic suppression in oceanic oxygen minimum zones

A recent study demonstrated that 50% of coastal marine animal species die after exposure to oxygen levels less than ∼70 μmol kg^–1 (Vaquer-Sunyer and Duarte, 2008). Equivalent oxygen partial pressures (P_O2) are known to cause physiological impairment in humans as well, for example at the tops of mountain peaks (Huey and Eguskitza, 2001). Yet, life thrives in deep-ocean oxygen minimum zones (OMZ), where oxygen values lower than this are pervasive (Levin, 2002; Childress and Seibel, 1998) (Fig. 1). The horizontal and vertical extent of the OMZ, as well as the intensity of the hypoxia found there, varies considerably between ocean basins (Fig. 1, Fig. 2A), but as much as 8% of the volume of the entire ocean contains less than 20 μmol kg^–1 oxygen (Paulmier and Ruiz-Pino, 2009). The oxygen content of a water mass is dependent on several factors, including the oxygen content when it was last in contact with air, the time elapsed since that point, and the rate of biological oxygen consumption. These factors, in turn, depend on temperature, air–sea gas exchange, ocean circulation and biology (Keeling et al., 2010; Paulmier and Ruiz-Pino, 2009). OMZs are most pronounced in regions of intense upwelling, such as the Eastern Tropical Pacific (ETP), that support high productivity and high subsurface oxygen demand (Wyrtki, 1962; Kamykowski and Zentara, 1990; Olson et al., 1993; Karstensen et al., 2008).

Substantial loss of dissolved oxygen and shoaling of hypoxia throughout much of the ocean, including the Subarctic and ETP, has been reported (Emerson et al., 2004; Stramma et al., 2008; Bograd et al., 2008; Whitney et al., 2007). For example, Bograd et al. showed that the oxygen content at 200–300 m has declined by 11–33 μmol kg^–1 in the California Current since 1984 (Bograd et al., 2008). This was predicted to occur with changing climate (e.g. Keeling and Garcia, 2002; Sarmiento et al., 1998; Bopp et al., 2002; Oschlies et al., 2008) because (1) oxygen is less soluble in warm water and (2) global warming is increasing upper ocean stratification, which increases productivity in surface waters, fueling increased oxygen demand at depth and simultaneously reducing oxygen supply to greater depths (Keeling et al., 2010). The hypoxia characterizing the North Pacific is already among the most intense in the global ocean such that further decreases in oxygen content may have severe consequences for resident biota. In some instances, expanding hypoxia (and the associated acidification due to respiratory carbon dioxide production) has encroached onto continental shelves where organisms are not adapted to such conditions (Chan et al., 2008; Grantham et al., 2004; Feely et al., 2008; Bograd et al., 2008).

The adaptations of animals for low oxygen tolerance are well-known and are driven by strong selective pressure to maintain aerobic metabolism via enhanced oxygen extraction from hypoxic water (Childress and Seibel, 1998). However, in more intense OMZs, the minimum oxygen level may be less than 5 μmol kg^–1 across depths of 100–900 m (Fig. 1). Such low values strongly influence the vertical distribution of resident metazoans, with most of them migrating to shallower, more oxygenated waters at night (i.e. diel vertical migration). The physiological adaptations that permit transient occupancy of these intense OMZs have been explored in only a few species but, for these few, it appears that an oxygen limit to the evolution of oxygen extraction capacity is surpassed, and the routine metabolic rate cannot be supported fully by aerobic pathways. These vertical migrators and their metabolic adaptations are of central focus here.

Three prominent strategies for tolerance of hypoxia during daytime forays into oxygen minimum zones are apparent in diel migrators: (1) aerobic metabolic ATP production is enhanced to the extent possible via adaptations for oxygen extraction and transport or, more rarely, oxygen storage; (2) anaerobic ATP production is upregulated via enhanced glycolytic capacity and alternative pathways with improved efficiency; and (3) total energy consumption is suppressed by shutting down energy-intensive cellular processes including locomotor activity (Childress and Seibel, 1998; Richards, 2011). Many organisms utilize a combination of these strategies to accommodate their highly variable and complicated vertical distribution patterns and local oxygen profiles. The integration of these three strategies is discussed below (Fig. 3). Metabolic suppression by oceanic migrators during daytime excursions into OMZs is a strategy that is little appreciated by oceanographers, yet one that has profound consequences for the ecology of the oceans. Anthropogenic changes in oxygen content may affect the distribution, abundance, performance and survival of marine organisms as well as the biogeochemical cycles dependent on organismal function (Seibel and Dierssen, 2009). This highlights the importance of defining critical oxygen levels based on a mechanistic understanding of organismal tolerance.

The term ‘hypoxia’ is most appropriately used to denote an oxygen (O₂) concentration less than some arbitrary value perceived as normal. While hypoxia sometimes specifies a perceived deficiency in O₂ for some physiological or biogeochemical process, evidence of O₂-limited function is usually lacking. Within the oceanographic community, hypoxia is typically defined by O₂ concentration, again usually without knowledge of the effect of that particular O₂ level on organism or ecosystem function. For example, Kamykowski and Zentara define hypoxia as a concentration less than 0.2 ml l^–1 (∼9 μmoles) (Kamykowski and Zentara, 1990) while Diaz and Rosenberg define it as less than 2.0 ml l^–1 (90 μmoles) (Diaz and Rosenberg, 1995). Defining hypoxia by O₂ concentration is also problematic because it is the P_O2, not the O₂ concentration, that drives O₂ provision to organismal tissues and determines the effect of hypoxia on organism function. Moreover, organismal O₂ transport systems and biochemical pathways are adapted to manipulate the P_O2 internally or to adjust the affinity for O₂ to drive ATP synthesis across the range of ambient O₂ concentrations under which a species has evolved (Connett et al., 1990). Thus, any definition of hypoxia related to the survival or performance of organisms is necessarily species-specific and will depend on the particular physiological requirements of the species and on the O₂ environment to which the species is adapted.

Brewer and Peltzer recently proposed a ‘respiration index’ to define critical levels of O₂ and CO₂ (Brewer and Peltzer, 2009). They suggest that the free energy (ΔG) obtainable from the oxidation of organic matter is limited by the ratio of reactants and products (O₂ and CO₂) as if in a closed system moving toward equilibrium (e.g. a test tube). However, they ignore the role of evolution in shaping organismal tolerances to environmental variability (Seibel et al., 2009). Living organisms are open systems that acquire free energy from their surroundings and operate at an approximate steady state. The concentrations of substrates and products in living systems are actively maintained at disequilibrium to ensure a constant ΔG and, thus, thermodynamic drive, for forward flux through the pathways of energy metabolism. Importantly, the concentrations of reactants (e.g. O₂ or CO₂) inside biological cells can be maintained at concentrations far higher than those found in the surrounding seawater (Seibel et al., 2009). Critical gas levels represent species-specific adaptations in O₂ uptake and acid–base regulation that have evolved within specific habitats. Critical gas levels for most species are far from the ultimate limits that can evolve in extreme habitats such as the OMZ. While the physiological community has also struggled with the ambiguity in terminology (Connett et al., 1990; Farrell and Richards, 2009), they recognize that an O₂ concentration that is stressful to one species or under one set of conditions may not be to other species or under other circumstances.

Both oceanographers and physiologists have developed a variety of terms, well-defined only in the unique context of their own disciplines or for specific studies. For example, the terms dysoxia and suboxia are used by oceanographers to refer to particular O₂ concentrations (Levin, 2002) that often, but not always (Canfield and Thamdrup, 2009), correspond to a transition in microbial processes (e.g. denitrification). However, to some physiologists, these terms refer to O₂-limited cytochrome turnover, a state that can only be defined by measuring intracellular P_O2 and the production of, and demand for, ATP (Connett et al., 1990). A host of other terms, including microxic, dysaerobic, suboxic, oxic and normoxic, have been inconsistently defined and applied. I believe these terms cannot be defined in a useful way and that inconsistent usage leads to confusion in the literature. The constraints that have been used to bound and define OMZ and O₂ dead zones are also not consistent across studies nor are they biologically relevant in many cases. I recommend abandoning all terms except hypoxia, meaning relatively low O₂, and anoxia, the complete absence of O₂ (although this term may be merely theoretical as our ability to measure low concentrations of O₂ is imperfect). Similarly, an OMZ should not be defined by a particular O₂ level but rather, vaguely, as a zone bracketing the depth or depth range of the lowest O₂ concentration in the water column. By this definition, OMZs may be more or less pronounced and occupy more or less of the water column. As long as there is a subsurface minimum O₂ level, there is an OMZ. Whether or not an OMZ has biological relevance depends on the processes and species being addressed. Rather than try to classify O₂ levels, I suggest that authors refer to specific O₂ partial pressures as they influence specific biological processes.

In part because of the ambiguity in terminology, a framework for predicting the biological response to changing O₂ levels remains elusive. The critical limitation for developing such a framework is the identification of O₂ thresholds critical (P_crit) for organismal survival and performance. Useful criteria for establishing limits for O₂ tolerance must rely on measures of physiological or ecosystem performance and recognize interspecific variation in O₂ tolerance and the capacity of organisms to adjust O₂ provision and demand to compensate for O₂ limitation.

The critical P_O2 (P_crit) may be identified in numerous ways, each relying on a different set of assumptions and with slightly different ecological and physiological implications. Typically, it is defined by (1) the P_O2 at which the rate of aerobic metabolism (O₂ consumption) can no longer be maintained independent of P_O2 or (2) the onset of anaerobic metabolism, as indicated by the accumulation of metabolic end-products (e.g. lactate) (Hochachka and Somero, 2002). P_crit describes the lower O₂ limit to an organism's ability to make physiological adjustments that regulate its metabolic rate at a constant level independent of the ambient P_O2 (i.e. oxyregulation) (see Richards, 2011). Possible adjustments include increased heart rate, enhanced gill surface area via remodeling (Mitrovic et al., 2009), increased blood pigment levels, adjustment of enzyme systems, and elevated rates of ventilation. Some of these adjustments are triggered by hypoxia-mediated upregulation of gene expression patterns (Flück et al., 2007) while others are regulated physiologically. By contrast, animals whose metabolic rate drops with declining P_O2 are termed ‘oxyconformers’. Practically, oxyconformation merely describes the metabolic response of organisms to O₂ levels below their P_crit. This is because there must always be an upper limit to metabolism beyond which additional O₂ will have no influence. There may be species with very high critical O₂ levels that conceal the upper limit to metabolism.

Connett et al. provide a framework for understanding the influence of P_O2 on metabolic rate and the requirement for anaerobic metabolism and metabolic suppression (Connett et al., 1990) (Fig. 3). In their model, the need to actively regulate metabolism, via physiological adjustments, as P_O2 decreases (arrows in Fig. 3) is influenced by total energy demand (ATP), as is a species' critical P_O2 (P_crit,1; blue line in Fig. 3). As energy demand increases, the P_crit also increases, and anaerobic metabolism may be required to contribute to total ATP flux (white region, Fig. 3). For example, an animal operating at its basal (resting, fasted state) level can maintain ATP demand at O₂ levels below the P_crit by increasing the rate of anaerobic (e.g. glycolytic) ATP production. At a still lower O₂ level (P_crit,2), anaerobic metabolism is insufficient to maintain basal ATP flux. Metabolism must be depressed below that point to avoid impaired function [termed ‘dysoxia’ by Connett et al. (Connett et al., 1990)]. Sustained exposure to O₂ levels below the P_crit,2 will eventually result in death.

The critical P_O2 is species-specific and depends on the regional O₂ concentration to which a species is adapted as well as the species' O₂ demand. As such, a broad regional comparison of P_crit values is required to determine levels of hypoxia that are harmful to ecosystems generally. With some simplifying assumptions, the model of Connett et al. (Connett et al., 1990) can be applied across species to understand the physiological strategies that determine critical O₂ thresholds in species with divergent metabolic rates and from regions with varying O₂ concentration. From Fig. 4, it is apparent that, in a given region, the P_crit increases with metabolic rate. Thus, animals with low O₂ demand are more tolerant of hypoxia, all else being equal. It also becomes apparent that, across the range of O₂ demand, organisms from more hypoxic regions have evolved greater hypoxia tolerance due to the mechanisms reviewed below.

Here, I identify two O₂ thresholds that may be important for the development of policy relating to climate-induced changes in O₂ content. In contrast to the two critical O₂ levels discussed above (Fig. 3), these O₂ thresholds are not species or situation specific. They are, however, temperature dependent. The first describes the O₂ level below which specific physiological adjustment or adaptation in O₂ uptake ability is required to sustain routine rates of metabolism in virtually all exposed species. The second describes what may be a physical constraint on O₂ extraction; it is the P_O2 below which no further adjustment or adaptation in aerobic O₂ utilization appears possible. These thresholds are apparent by plotting P_crit for diverse species as a function of the minimum O₂ level to which the species are adapted (Fig. 2B). In this plot, a close correspondence is observed, at least within an intermediate range of P_O2 values, between P_crit and the minimum P_O2 to which a species is exposed. Above this range, P_crit plateaus, indicating that further increases in O₂ concentration have no effect on the capacity of animals to regulate metabolism. In other words, virtually all marine animals can regulate their metabolism to at least this O₂ level. Below this threshold, organisms that are not specifically adapted to low O₂ will suffer physiological stress and eventual death. This first threshold occurs at approximately 5 kPa in the present dataset but will increase with temperature. The second threshold is evident where further decreases in O₂ concentration fail to select for lower P_crit. It corresponds approximately to O₂ levels below which the biomass of permanent-OMZ residents is reduced and transient residents (e.g. diel migrant biota) reduce metabolic expenditure to facilitate brief (hours) forays into hypoxic water (see ‘Metabolic suppression in OMZs’).

The first O₂ threshold corresponds closely to the mean LC₅₀ (O₂ value resulting in death of 50% of individuals tested) value of 70 μmol l^–1 reported for coastal benthic organisms (Vaquer-Sunyer and Duarte, 2008). Higher measurement temperatures, greater activity (metabolic) levels and synergistic stressors will all result in greater sensitivity to hypoxia. Most of the P_crit represented in Figs 2 and 4 were measured at 5°C. Regardless, the suggested boundaries put constraints on the expected hypoxia-induced shifts in species' vertical and horizontal distributions and provide a framework within which reductions in biodiversity and expansion of dead zones may be predicted and assessed. For example, the lower O₂ threshold falls just below the minimum O₂ level present along the California coast. Thus, any expansion of the OMZ in this region has the potential to dramatically alter the species composition, vertical distribution of organisms and functioning of the ecosystem as a whole.

The vertical distribution of oceanic organisms is highly complex (Robinson et al., 2010) and difficult to summarize within the scope of this review. However, a few general trends are informative. Firstly, there is little effect of low O₂ levels in the OMZ on the biomass of midwater organisms over a very wide range of O₂ concentrations (down to ∼10 μmol l^–1). However, different species have their own unique upper and lower depth limits and unique O₂ signatures. The biomass of permanently deep-living zooplankton and micronekton is quite high throughout the depth range in regions with moderate O₂ minimum zones, such as off the coast of California (Childress and Seibel, 1998) (Figs 1, 2). The modest effect of the OMZ on organismal vertical distribution is especially interesting considering that these O₂ concentrations are far below what is normally considered hypoxic in shallow and coastal habitats (e.g. Vaquer-Sunyer and Duarte, 2008).

At O₂ concentrations below ∼10 μmol l^–1, there are pronounced effects on both species' distributions and biomass. Hypoxia of this extreme level characterizes OMZs in the ETP (Fig. 1) and Arabian Sea. Within such regions, there are two dominant patterns of zooplankton distribution. (1) The permanent zooplankton biomass, including copepods, amphipods, chaetognaths and a variety of gelatinous species, is typically reduced within the core of the OMZ, and there is a secondary biomass peak at the lower edge of the OMZ, where O₂ content is rising (Wishner et al., 2008). This group effectively avoids the lowest O₂ concentrations and is adapted to stable hypoxia, like permanent residents in more moderate OMZs (Childress and Seibel, 1998). (2) The few species that do breach the core of the most intense OMZs are typically diurnal vertical migrators that enter shallower, more oxygenated waters at night (for reviews, see Fernandez-Alamo and Färber-Lorda, 2006; Wishner et al., 2008). This group is dominated by myctophid (lantern) fishes, which make up the bulk of the sound-scattering layer in many cases, and euphausiids (krill), although a variety of other taxa are also part of this mass migration. A third group, consisting of highly mobile, nektonic predators such as large squids and fishes, enters the OMZ but is usually tied closely to the surface by physiological constraints (see below).

A large proportion of animal species in all regions of the open ocean migrate on a diel basis between well-lit surface waters and the more dimly lit mesopelagic zone. This is the largest mass animal migration on Earth (Robison, 2009) and is driven predominantly by predator avoidance (although some organisms gain considerable energetic advantage by migrating into cool, hypoxic waters) (Rosa and Seibel, 2010; Svetltichny et al., 2000). Diel vertical migrators living in areas of oceanic OMZs migrate through a gradient of relatively stable O₂ environments and so possess adaptations for enhanced anaerobic metabolism and metabolic suppression similar to those from unstable O₂ environments (Flück et al., 2007; Richards, 2011). By contrast, organisms from stable O₂ environments tend to rely more on aerobic adaptations for extraction of O₂ sufficient to maintain routine metabolic rates (Childress and Seibel, 1998).

The aerobic adaptations of animals to O₂ minimum layers have been extensively reviewed (Childress and Seibel, 1998) and are summarized here. The problem of hypoxia is one of a reduced gradient to drive diffusion of O₂ from the ambient seawater, across gas-exchange surfaces, to the mitochondria. Because the energy yield of aerobic metabolism is high relative to anaerobic metabolic pathways (Hochachka and Somero, 2002), there is strong selection for mechanisms to take up O₂ effectively from very low ambient partial pressures (i.e. to increase the effective gradient). Accordingly, most permanent residents of moderate OMZs are able to regulate their O₂ consumption rate to the minimum O₂ levels encountered in their environments (Fig. 2) (reviewed in Childress, 1995; Childress and Seibel, 1998). In other words, their P_crit are generally lower than the minimum O₂ level they encounter. Specific adaptations for O₂ extraction in crustaceans, fishes and cephalopods living in O₂ minimum layers include: (1) high ventilatory volumes and circulatory capacity (Belman and Childress, 1976); (2) thin blood-to-water diffusion distances across the gills and high gill surface area (e.g. Antezana, 2002; Childress, 1969); and (3) respiratory proteins with a high affinity for O₂ (high affinity=low P₅₀, the O₂ concentration resulting in 50% O₂ saturation of respiratory proteins) (Seibel et al., 1999) and high sensitivity to pH (e.g. a large negative Bohr coefficient) (Seibel and Walsh, 2003). Interestingly, gelatinous zooplankton (e.g. medusae and ctenophores) also appear capable of regulation to very low P_O2 despite an apparent lack of O₂ transport systems or specific gas exchange surfaces. In such cases, weak diffusion gradients may simply be sufficient to meet low O₂ demand in species that are 95% water and whose actively metabolizing tissues are located towards the exterior of the animal (Thuesen et al., 2005).

From a mammalian perspective, ‘breath-holding’ is an attractive option for periodic forays into hypoxic environments. However, very few, if any, fishes or invertebrates are known to store O₂ in any comparable way. A few possible exceptions have been hypothesized recently. The fish Bregmacerous cantori was shown to undergo dramatic changes in the volume and tension of the O₂-filled swimbladder during its vertical migration into the anoxic Cariaco Basin (Love et al., 2004). The conclusion was that the bladder volume decreased at depth as the fish utilized the O₂ stored within to support aerobic metabolism in otherwise anoxic waters. O₂ stored for buoyancy in the chambers of the Nautilus shell (Cephalopoda) may be used to support a suppressed rate of aerobic metabolism during exposure to hypoxia (Boutilier et al., 1996). Thuesen et al. recently demonstrated that gelatinous organisms (e.g. medusae) take up O₂ (passively, as far as is known) into the mesogleal gel that can later be used to support aerobic metabolism in active tissues when ambient O₂ levels are too low (Thuesen et al., 2005). It was shown that this O₂ storage is sufficient to support short migrations into subsurface hypoxia in shallow estuaries. The utility of this strategy for O₂ provision during migrations over greater depths into OMZs would be limited.

The ability of a species to regulate its rate of O₂ consumption to low P_O2 is facilitated by low O₂ demand. The evolved demand for O₂ (energy) varies by nearly 1000-fold, independent of body size and temperature, among species across a depth range. Efforts to explain this variation typically invoke limitations on resource availability (e.g. O₂ or food) in the deep sea. This implies that deep-living species would evolve a higher energy consumption rate if only more resources were available. However, without specific energetic demand, an increase in consumption is wasteful and presumably not favored by natural selection. In other words, organisms do not consume O₂ and produce ATP they don't need. Thus, reduced metabolic rates of permanently deep-living and gelatinous species do not result from resource limitation. Regional comparisons have shown that deep-sea species of some taxa have low metabolic rates even in regions with higher O₂ concentrations. Furthermore, species from high O₂ regions have higher P_crit despite metabolic rates as low as those from hypoxic regions (Fig. 4) (Childress and Seibel, 1998). Thus, low rates are not a specific adaptation to hypoxia. Rather, low rates reflect sluggish habits that result from relaxed selection for strong locomotory abilities for visual predator–prey interactions in the poorly lit bathypelagic waters (Childress, 1995; Seibel and Drazen, 2007).

Nevertheless, many vertically migrating zooplankton living in OMZs cannot support their routine energy demands aerobically at depths across the entire O₂ gradient encountered (Fig. 2). For example, the large copepod Gaussia princeps has a P_crit of 1.6 kPa but lives part of the time at P_O2 values as low as 0.8 kPa (off the Californian Coast). It has a substantial capacity to survive entirely without O₂ (10–14 h) (Childress, 1977) whereas some other copepod species are excluded from intense OMZs (Auel and Verheye, 2007). Similar strategies have been described for the vertically migrating cephalopod Gonatus onyx (Hunt and Seibel, 2000) and the vertically migrating fish Stenobrachius leucopsaurus (Torres et al., 1979). Gonatus onyx has a P_crit of only 4.0 kPa but spends its daytime at a P_O2 of <0.8 kPa. Similarly, S. leucopsaurus has a P_crit of 1.6 kPa despite spending its daytime hours in relative hypoxia. The jumbo squid, Dosidicus gigas, has among the highest metabolic rates of any animal on the planet, yet is closely tied to regions of low O₂ (Seibel, 2007). Despite respiratory adaptations for hypoxia tolerance, D. gigas cannot support its routine metabolic rate across the entire depth range it occupies (Gily et al., 2006; Rosa and Seibel, 2010). D. gigas has a P_crit of 1.6 kPa but routinely experiences a P_O2 below 0.4 kPa (L. Trueblood and B.A.S., unpublished data).

The elevated metabolic rates of vertically migrating and high-performance nektonic species preclude aerobic survival in the most pronounced OMZs by virtue of the extraordinary ventilation volumes required to meet routine O₂ demand. In the absence of hypoxia-induced metabolic suppression (see below), a 100 g D. gigas must pass more than 2 liters of OMZ core water, at 10 μmol l^–1 O₂ and assuming complete extraction of all available O₂, across its gills every minute to meets routine needs (Rosa and Seibel, 2010). Because ventilation and locomotion are tied in many active squids and fishes (Wells, 1988), the demand for O₂ increases as ventilation rates increase. This positive feedback effectively precludes most top predators from OMZs. As a result, the depth distribution of most high-performance fishes, such as tunas and swordfish, is limited by O₂ concentrations less than ∼150 μmol l^–1 (Brill, 1994). In fact, swordfish and marlin were shown to reduce the extent of their vertical excursions by more than 500 m off the Californian Coast relative to more oxygenated waters in the Atlantic (Carey and Robison, 1980; Prince and Goodyear, 2006). Both white sharks and mako sharks appear to be limited in waters with O₂ concentrations less than ∼50 μmol l^–1 (Vetter et al., 2008; Nasby-Lucas et al., 2010) although both occasionally experience waters with as little as 20 μmol l^–1 O₂. The great white shark, Carchardodon carcharias, was similarly shown to limit its depth range in hypoxic regions and to make shorter and shallower dives where O₂ was low (Nasby-Lucas et al., 2010).

Epipelagic predators appear to have reached an upper limit to O₂ extraction. Increased metabolism and decreased P_O2 both select for increased gill diffusion capacity. Permanent residents of the OMZ already have gill diffusion capacities similar to those of active fishes and squids (Childress and Seibel, 1998; Wegner et al., 2010) despite a demand for O₂ that is orders of magnitude lower than that of shallower-living species. Presumably, both groups are near a lower limit for gill membrane thickness and perhaps an upper limit for gill surface area. Thus, organisms with higher metabolic rates may be constrained in terms of the structural and biochemical capacities required for O₂ extraction in the OMZ.

There are some high-performance predators that have managed to minimize the trade-offs inherent between high performance and hypoxia tolerance. In bigeye tuna (Thunnus obesus), for example, the onset of cardio-respiratory adjustments during acute hypoxia occurs at lower ambient O₂ levels (∼40 μmol l^–1) than in other species from more oxygenated regions (Lowe et al., 2000). In this case, greater hypoxia tolerance appears to be due to a hemoglobin with a high affinity for O₂. However, the utility of a high-affinity respiratory protein is constrained in active animals by the need to release O₂ at high rates to metabolically active tissues. Even in sluggish permanent residents of OMZs, an enhanced pH sensitivity (e.g. Bohr coefficient) is required to facilitate the release of O₂ to the tissues at rates sufficient to support even a low O₂ demand (Childress and Seibel, 1998). In the case of T. obesus, this constraint is overcome because the enhanced Bohr effect is coupled with unique adaptations that facilitate greater off-loading as blood is warmed in the core of the tuna's heterothermic body (Lowe et al., 2000). Similarly, D. gigas and other vertical migrators have high-affinity respiratory proteins that are both temperature and pH sensitive, facilitating O₂ uptake in cold, deep waters where O₂ is scarce and enhancing O₂ release in warmer surface waters where O₂ and O₂ demand are elevated. However, even these adaptations are insufficient for survival in the core of the OMZ. Blood hemocyanin–O₂ saturation falls below 20% at around 300 m depth in D. gigas (Fig. 5). The P_crit corresponds to ∼60% hemocyanin–O₂ saturation at in situ temperatures. This level is reached as shallow as 160 m throughout much of the squid's natural range. If not for an upregulation of anaerobic ATP production and the pronounced metabolic suppression described below, vertical migrators would still largely be excluded from well-developed OMZs.

Aerobic metabolism may be supplemented with anaerobic metabolic ATP production (e.g. fermentation) in order to maintain a routine metabolic rate under hypoxic conditions in the OMZ. However, the yield of ATP per mole of fermentable substrate is always modest during anaerobic glycolysis compared with oxidative metabolism (Hochachka and Somero, 2002). Some anaerobic metabolic pathways (those resulting in succinate and propionate accumulation, for example) are more efficient than others, although still nowhere nearly as efficient as oxidative metabolism. Therefore, utilization of anaerobic metabolic pathways for anything but short bursts of activity requires a substantial supply of food or large stores of fermentable substrate (i.e. glycogen). Depletion of reserves during anaerobic metabolism minimizes the advantages gained by migration to the rich food supplies in shallow water. A further consequence of anaerobic metabolism is the accumulation of deleterious end-products. Anaerobic glycolysis results in the accumulation of protons as well as organic compounds such as lactate and octopine (Hochachka and Somero, 2002), so consideration must be given to acid–base disturbances (Hochachka and Mommsen, 1983). Some anaerobic pathways also produce volatile end-products that can be excreted directly across the gills, allowing their continued use without risk of acid–base disturbance (Stecyk et al., 2004). However, the remaining energy contained in the volatile end-products (e.g. ethanol) is lost and cannot be recycled. Despite these disadvantages, there is evidence for elevated anaerobic metabolic capacity as well as the accumulation of anaerobic end-products following hypoxic exposure in some diel vertical migrators living in regions with pronounced OMZs.

Torres et al. (J. J. Torres, M. D. Grigsby and M. E. Clarke, manuscript in preparation) demonstrated higher activities of alcohol dehydrogenase, an enzyme involved in glycolytic ATP production with the concomitant formation of ethanol, in myctophid fishes from the Arabian Sea compared with regions with higher O₂ concentrations. This exciting finding suggests that cyprinids are not the only fishes that produce ethanol as an anaerobic end-product and represents a novel physiological adaptation to OMZs. Ethanol is a volatile end-product that can be excreted directly across the gills, along with the CO₂ that is simultaneously produced, thereby avoiding several of the disadvantages of anaerobic metabolism (Stecyk et al., 2004). However, the energy contained in the excreted ethanol is lost. Similarly, Gonzales and Quinones showed that krill that migrate into an OMZ have very high activities of lactate dehydrogenase compared with a copepod that lives permanently in the well-oxygenated waters above (Gonzales and Quinones, 2002). However, this comparison is confounded by body size and phylogeny and the conclusions are in contrast to most studies of enzymatic activities in OMZs. For example, most studies find no suggestion of increased anaerobic capacity in the enzymatic profiles of fishes, cephalopods or copepods living in OMZs (Childress and Somero, 1979; Vetter and Lynn, 1997; Thuesen et al., 1998; Seibel et al., 2000). Vetter et al. actually found that lactate dehydrogenase (LDH) activity in the Dover sole (Microstomus pacificus) decreased in response to low O₂ (Vetter et al., 1994).

The few studies that have measured accumulation of anaerobic end-products under hypoxic conditions suggest that glycolysis is upregulated in the OMZ but that it accounts for only a small fraction of the total routine energy budget. Dosidicus gigas spends hours each day at O₂ levels below its P_crit (Rosa and Seibel, 2008; Rosa and Seibel, 2010). Under laboratory conditions that mimic its daytime habitat (10°C, ∼10 μmol l^–1 O₂), octopine accumulated to a level that amounts to only 7% of the total ATP demand under those conditions (Rosa and Seibel, 2010). Similarly, the euphausiids Euphausia eximia (Fig. 6B) and Meganyctiphanes norvegica both accumulated lactate under hypoxic conditions. However, E. eximia, which lives in an extreme OMZ in the ETP, survived treatment with 0.8 kPa P_O2 (J. Schneider and B.A.S., unpublished observation) whereas O₂ below 1.8 kPa resulted in 100% mortality in M. norvegica (Spicer et al., 1999), a species that lives in the North Atlantic but sometimes experiences moderate hypoxia in fjords. Anaerobic metabolic ATP production may simply support activity levels above the basal metabolic rate for short periods of time, as opposed to wholly supporting metabolic expenditure throughout the period of hypoxic exposure.

Suppression of total metabolism appears to be a prerequisite for the survival of prolonged bouts of O₂ limitation (Hochachka and Somero, 2002; Bickler and Buck, 2007; Guppy and Withers, 1999). Total metabolism consists of the energy produced via oxidative (aerobic) metabolic pathways (estimated from the O₂ consumption rate, ) and that produced by anaerobic metabolic pathways (McDonald et al., 1998), typically estimated from the accumulation of anaerobic end-products. Metabolic suppression is achieved primarily by shutting down expensive cellular activities (Hand, 1998), such as biosynthesis and ion pumping (Buck and Hochachka, 1993). The resulting decrease in ATP consumption is matched by a decrease in ATP production primarily achieved through down-regulation of glycolytic enzymes (Lutz and Nilsson, 1997; Brooks and Storey, 1997). Thus, a ubiquitous feature of hypoxia- and anoxia-tolerant organisms is the ability to maintain nearly constant levels of ATP (Hochachka and McClelland, 1997; Suarez et al., 1987).

Not surprisingly, the duration of anoxia/hypoxia tolerance among organisms is inversely related to the extent of their metabolic suppression (Hand, 1998). Brine shrimp (Artemia spp.) can shut down metabolism to immeasurable levels and survive complete anoxia for at least 4 years (Hand, 1998). Such complete metabolic suppression has not been reported in marine species, but extreme suppression is thought to occur during diapause in some oceanic copepods (Auel et al., 2005). More commonly, organisms suppress metabolism by 50–95% and supplement the remaining energy demand using a combination of available O₂ and anaerobic metabolic pathways. Such organisms survive anoxia for hours or days.

Metabolic suppression during diel migrations in the OMZ has been demonstrated in only a few oceanic species. Dosidicus gigas maintains its rate of energy consumption independent of the ambient P_O2 until a critically low level (P_crit=1.6 kPa) is encountered (Gilly et al., 2006) (L. Trueblood and B.A.S., unpublished data). Below the P_crit (at depths below ∼150 m), the of D. gigas sharply declines and anaerobic metabolism is enhanced to partially offset the reduction in aerobic ATP synthesis. Squids use the glucose–opine anaerobic pathway, which entails a reductive condensation of pyruvate to form octopine with the net synthesis of three ATPs per unit glycogen catabolized (1.5 units ATP per unit octopine produced). Octopine is accumulated to levels 10-fold higher under hypoxic conditions relative to controls. Despite this dramatic upregulation, glycolytic flux accounts for only 7.0% and 2.8% of the energy deficit at 10°C and 20°C, respectively. This represents a substantial suppression of total ATP production in D. gigas (∼87% at 10°C, under the hypoxic conditions characteristic of its daytime habitat depth) (Rosa and Seibel, 2010) (Fig. 6A). Some vertically migrating euphausids, which are a common prey item for D. gigas, similarly suppress total ATP production (∼40%) under hypoxic conditions characteristic of their deeper daytime habitat. The accumulation of lactate in euphausiids (see ‘Anaerobic adaptations to the OMZ’) is significant but, as in D. gigas, represents a fairly small fraction of the total metabolic demand (Fig. 6B).

Several behavioral observations are consistent with the idea that vertically migrating species suppress metabolism during daytime forays into OMZs. In one of the earliest reports of midwater animal behavior in situ, Barham described the orientation and activity levels of several species of vertically migrating midwater fishes in the California Current (Barham, 1971). At their deeper, daytime depths, most individuals were ‘passively drifting’ whereas at shallower, night-time depths, they were ‘actively swimming’. Barham attributed this ‘lethargy’ to the low O₂ levels in the deep-sea (Barham, 1971). When alerted to the submersible, all fishes swam quickly away. Similar in situ behavior was reported for the vertically migrating mesopelagic squid G. onyx. Hunt and Seibel observed that individuals in shallower water were quick to dart away when approached by the submersible whereas individuals in deeper water held their positions and escaped only after persistent harassment by the investigators (Hunt and Seibel, 2000). Some species are quick to escape from submersibles while others are attracted to submersible lights. Thus, in situ determinations of activity levels can be misleading. For example, Cowles reports equal activity levels during both day and night off the Californian Coast for the vertically migrating shrimp Sergestes similis (Cowles, 1994). However, he also reported that shrimps observed by the submersible were always swimming directly downward. This was explained as positive geotaxis in response to the lights of the submersible. His laboratory analysis, by contrast, revealed a reduction in O₂ consumption and swimming speed with declining O₂ concentrations (Cowles, 2001).

Svetlichny et al. found that the copepod Calanus euxinus reduced its basal metabolism, as well as its scope for activity, under laboratory conditions that mimic its daytime habitat in the Black Sea (P_O2 ∼1.5 kPa, 8°C) relative to air-saturated conditions at 18°C (Svetlichny et al., 2000). The frequency of locomotory acts decreased more than 3-fold. Hypoxia was shown to provide an important energetic advantage during diel migrations in this species.

Antezana found no difference in , swimming or feeding activity under low O₂ conditions in the vertically migrating krill Euphausia mucronata, living in the OMZ in the ETP (Antezana, 2002). This finding contrasts with those for other krill species (J. Schneider and B.A.S., unpublished) (Spicer et al., 1999; Teal and Carey, 1967) (see above). Antezana did find elevated gill surface area relative to species living in more oxygenated regions, suggesting enhanced O₂ extraction capabilities in E. mucronata (Antezana, 2002).

I suggest that metabolic suppression is a common strategy for survival during daytime forays into pronounced OMZs. The prevalence of suppression within the vertically migrating community in any given region will depend on the local O₂ distribution as well as temperature and the particular species complex. The extent of suppression for any particular species will depend on its metabolic demand, phylogenetically available biochemical pathways, its vertical distribution and capacity for O₂ transport. Thus, the consequences for organismal contributions to biogeochemical cycles and ecosystem dynamics (see below) will be difficult to predict and requires direct measurement for the dominant species.

The survival of oceanic organisms in OMZs depends on total O₂ demand and the capacities for O₂ extraction and transport, anaerobic ATP production and metabolic suppression. These parameters are, within a range of P_O2, closely tuned to the O₂ level to which a species has evolved. In other words, the minimum O₂ level experienced by a particular species is often very close to its evolved P_crit. This fact facilitates the identification of two O₂ thresholds, as discussed above, that may prove useful for policy makers in the near future. Between these thresholds, specific physiological adaptations are required of virtually all species. Climate-induced changes in O₂ concentrations may dramatically alter the ecosystem, especially if O₂ concentrations are pushed below the lower threshold. This threshold represents a limit to adaptation that will certainly result in an ecological transition, from one dominated by a diverse midwater fauna to one dominated by diel migrant biota that must return to surface waters at night.

In the event of such a transition, the ecology of the region will be fundamentally altered. For example, interactions between top predators and their prey may change as the habitable night-time depth range is compressed. Anaerobic metabolism and metabolic suppression are not sustainable due to substrate limitation, end-product accumulation (i.e. O₂ debt) and the likely constraint imposed on foraging at depth by metabolic suppression. These factors dictate a return to surface waters during some portion of the day. If deep-water hypoxia expands due to climate change, the depth to which migrating organisms must return to burn off an O₂ debt at night will shoal. However, surface waters also impose O₂ stress due to elevated O₂ demand at high temperature and constrained O₂ transport due to ocean acidification (Rosa and Seibel, 2008) and high temperature (Ekau et al., 2010). Thus, the synergistic impact of these three climate-related variables (temperature, ocean acidification and declining O₂) is to compress the habitable night-time depth range of vertically migrating species (Fig. 7). Because vertical distributions and environmental tolerances are species-specific, habitat compression will affect each species differently and may result in altered ecological interactions. Some species may additionally alter their geographical range as a result of climate variation. For example, a recently reported invasive expansion of the range of D. gigas in the Eastern North Pacific (Zeidberg and Robison, 2007) has been attributed to, among other things, a climate-induced shoaling of the OMZ within the California Current System (Bograd et al., 2008; Vetter et al., 2008).

Expanding OMZs will also affect biogeochemical cycles. Of particular interest here is the influence of OMZs on the contribution of mesopelagic animals to respiratory carbon flux. Diel migrators consume carbon near the surface and then transport it to depth, where it is excreted as CO₂ (Angel 1989; Longhurst et al., 1990). This process is responsible for a large fraction of the carbon exported from surface waters in some regions (Ducklow et al., 2001). The OMZ influences this flux by altering the biomass, vertical distribution and physiology of resident organisms. In particular, if most migrators are suppressing metabolism at depth, then the amount of CO₂ released is similarly suppressed, thus decreasing the efficiency of this ‘biological carbon pump’. In other cases, such as for the myctophid fishes discussed above, a reliance on anaerobic ethanol production may contribute novel substrates for microbial metabolism. These considerations have not been considered in any quantitative analysis of carbon flux. Animal physiology and, in particular, the response of animals to expanding hypoxia, is a critical, but understudied, component of biogeochemical cycles and oceanic ecology (Seibel and Dierssen, 2009).

I thank R. Rosa, L. Trueblood, J. Schneider, T. Towanda, A. Maas, L. Birden and A. Nyack for stimulating ideas, discussion and data as well as their assistance at sea. I thank J. Childress, B. Robison, J. Torres and W. Gilly for helpful discussions and for sharing data. S. Haddock made unsolicited, but helpful, comments on an earlier draft of this manuscript while I was sleeping.

 
• ETP

  Eastern Tropical Pacific

 
• 

  oxygen consumption rate in molar units

 
• OMZ

  oxygen minimum zone

 
• P₅₀

  oxygen partial pressure resulting in 50% saturation of respiratory proteins

 
• P₉₀

  oxygen partial pressure resulting in 90% saturation of respiratory proteins

 
• P_crit

  critical oxygen partial pressure

 
• P_O2

  oxygen partial pressure

 
• ΔG

  free energy