METABOLISM, HYPOXIA TOLERANCE AND HEAT SHOCK RESPONSE OF AMPHIPODS, EMPHASIZING THE HYPERIID AMPHIPOD PHRONIMA SEDENTARIA

This work investigates the ecophysiology of marine amphipods. Amphipods are an important part of the zooplankton community in the pelagic environment. Amphipods are a food source for a variety of fishes and also have a role in carbon cycling. Little is known about their physiology and how they have adapted to environmental variation. The Intergovernmental Panel on Climate Change (IPCC) reports that global warming is causing temperatures to rise throughout the world’s oceans, a trend that will continue with rising human carbon emissions. As temperature and CO2 levels increase, oceanic oxygen levels are predicted to decrease, and as a result, oxygen minimum zones will expand. Ocean general circulation models have shown that the detectable decrease in dissolved oxygen concentrations is driven by increasing ocean surface temperatures and enhanced stratification. Low oxygen concentrations and high temperatures affect physiological performance and, consequently, vertical distribution and ecology of marine organisms. Vertically migrating amphipods living in the Eastern Tropical North Pacific currently experience temperature changes of 15 degrees Celsius or more and changes in oxygen concentration from saturation to near anoxia. Metabolic depression is the reduction in total metabolic rate, including aerobic and anaerobic ATP consumption, to below the basal metabolic rate. This happens in response to environmental stress such as extreme temperatures, desiccation, anoxia and food deprivation. The tolerance of an organism to low oxygen is inversely related to the extent of their metabolic. Organisms subjected to physiological stress, such as stresses that cause proteins to denature, will respond by producing heat shock proteins (hsps). Hsps act as molecular chaperones and are able to prevent/reduce denaturing of proteins and target proteins that are irreversibly denatured for removal from the cell via the ubiquitin-proteosome pathway. No previous studies have been done on midwater amphipods to see if the temperature gradient they experience during diel vertical migration induces a stress response. Chapter 1 examines how temperature and oxygen gradients affect the physiology of the amphipod Phronima sedentaria by quantifying the aerobic and anaerobic metabolic rates at oxygen levels consistent with those experienced across Phronima’s vertical range in tropical regions. Total ATP production (metabolic rate) was compared in specimens subjected to night time surface conditions (oxygenated) and day time conditions (hypoxia). In Chapter 2 protein concentrations of hsp 70 were measured in specimens subjected to a range of temperatures within and above what they typically experience. Understanding the adaptations of pelagic amphipods to their current environment will help predict the physiological impacts of global warming for amphipods and their predators. One adaptation for living in hypoxia is metabolic depression. Metabolic rates of organisms are affected by a number of variables, particularly by temperature, body mass and ecology. Metabolic rate typically doubles or triples for every 10°C change in body temperature. Routine oxygen consumption rates of most vertically migrating, visually oriented, midwater crustaceans decline with depth primarily due to temperature, but also due to the low light and consequential lack of visual cues which reduces locomotion needs. Transparent organisms in epipelagic regions would be relieved of this selective pressure because they are hidden from their visual predators. Hyperiid amphipods are the only group of crustaceans that are truly dominated by transparency. The influence of transparency on metabolic rate has not been examined in amphipods. Chapter 3 sought to determine what environmental and ecological factors influence the rate of metabolism in marine amphipods by examining a broad data set from polar to tropical environments, and including transparent specimens. The data set for this study was obtained from the literature and original data. Recent molecular work allowed us to look at hyperiid metabolism in a phylogenetic context. Understanding patterns of pelagic and deep sea metabolism is important for further understanding of global carbon flux and the consequences of climate change on migration strategies.

Organisms subjected to physiological stress, such as stresses that cause proteins to denature, will respond by producing heat shock proteins (hsps). Hsps act as molecular chaperones and are able to prevent/reduce denaturing of proteins and target proteins that are irreversibly denatured for removal from the cell via the ubiquitin-proteosome pathway. No previous studies have been done on midwater amphipods to see if the temperature gradient they experience during diel vertical migration induces a stress response.
Chapter 1 examines how temperature and oxygen gradients affect the physiology of the amphipod Phronima sedentaria by quantifying the aerobic and anaerobic metabolic rates at oxygen levels consistent with those experienced across Phronima's vertical range in tropical regions. Total ATP production (metabolic rate) was compared in specimens subjected to night time surface conditions (oxygenated) and day time conditions (hypoxia).
In Chapter 2 protein concentrations of hsp 70 were measured in specimens subjected to a range of temperatures within and above what they typically experience.
Understanding the adaptations of pelagic amphipods to their current environment will help predict the physiological impacts of global warming for amphipods and their predators. One adaptation for living in hypoxia is metabolic depression. Metabolic rates of organisms are affected by a number of variables, particularly by temperature, body mass and ecology. Metabolic rate typically doubles or triples for every 10°C change in body temperature. Routine oxygen consumption rates of most vertically migrating, visually oriented, midwater crustaceans decline with depth primarily due to temperature, but also due to the low light and consequential lack of visual cues which reduces locomotion needs. Transparent organisms in epipelagic regions would be relieved of this selective pressure because they are hidden from their visual predators.
Hyperiid amphipods are the only group of crustaceans that are truly dominated by transparency. The influence of transparency on metabolic rate has not been examined in amphipods.
Chapter 3 sought to determine what environmental and ecological factors influence the rate of metabolism in marine amphipods by examining a broad data set from polar to tropical environments, and including transparent specimens. The data set for this study was obtained from the literature and original data. Recent molecular work allowed us to look at hyperiid metabolism in a phylogenetic context.

Introduction
In some regions of the oceans, at intermediate depths, biological oxygen use exceeds the rates of oxygen replenishment via the processes of advection and diffusion (Packard et al. 1988) leading to zones of low oxygen. These oxygen minimum zones (OMZs) occur in areas of high primary productivity such as the Eastern Tropical North Pacific, where organic matter from the surface sinks and decays, adding to the oxygen demand at intermediate depths (Levin 2002, Fiedler & Talley 2006. The OMZ in the Eastern Tropical North Pacific (ETNP) is remarkable for both its size and degree of hypoxia (Kamykowski & Zentara 1990). This OMZ extends vertically from 50m to 1200m (Fernández-Álamo & Färber-Lorda 2006). Below 300m oxygen levels vary, but levels can be less than 2 µM (0.15kPa, 0.04ml l -1 ) (Wishner et al. 2013). The California current has a less severe OMZ, with oxygen levels reaching a minimum of 13.4 µM (0.8kPa, 0.3 ml l -1 ) (Childress & Seibel 1998).
OMZs are predicted to expand both vertically and horizontally as a result of the changing world climate (Bograd et al. 2008, Stramma et al. 2008, Keeling et al. 2010, Deutsch et al. 2011). Most of the oxygen decrease is attributed to increased stratification, which limits the mixing of oxygenated surface waters with subsurface waters and reduces the subsurface oxygen concentrations .
Increasing global temperatures will warm ocean surface waters, leading to a decrease in oxygen content because oxygen is less soluble in warm water. Oxygen levels influence vertical distribution and ecology of marine animals , Wishner et al. 2013. The effects on crustacean zooplankton are particularly important because of the role of zooplankton as a link between marine primary producers and upper trophic levels (Ekau et al. 2010). Understanding how oxygen concentrations affect crustacean physiology is important because expanding OMZs may cause fluctuations in species' vertical and horizontal habitat ranges. Those fluctuations could, in turn, change ecosystem trophic structures due to alterations in predator-prey interactions as well as affecting carbon cycling , Doney et al. 2012).
Most studies on hypoxia tolerance of marine animals have been conducted in OMZs where dissolved oxygen levels are relatively higher than in the OMZ of the Eastern Tropical North Pacific. Organisms found in the California Current OMZ are often able to remain aerobic (Childress 1977), so there is little effect of low oxygen on organism distribution. This ability to extract oxygen from hypoxic water is due to a variety of adaptations including: increased ventilation and circulation capacity, high gill surface area, short blood to water diffusion distances, and respiratory proteins with high oxygen affinity and cooperativity (Sanders & Childress 1990, Childress & Seibel 1998). The distribution in such moderate OMZs is dominated by permanent deepliving zooplankton and micronekton throughout the depth range , Childress & Seibel 1998, Robinson et al. 2010).
At oxygen concentrations less than ~ 10 µM in the ETNP, there is a reduction in biomass at depth. Most organisms either live at the upper or lower OMZ interfaces (zones of steep oxygen gradients), or vertically migrate to more oxygenated waters at night (Vinogradov & Voronina 1962, Wishner et al. 1990, Saltzman & Wishner 1997, Wishner et al. 2013. However, organisms accustomed to variable and transient hypoxia, such as that experienced by diel vertical migrators, will often depress their total ATP consumption rate to limit the accumulation of harmful anaerobic end products (e.g., H + ) and to conserve fuel stores. For example, the euphausiid Euphausia eximia reduces its oxygen consumption rate by more than 50% at 1% oxygen (0.8 kPa), and the contribution from anaerobic pathways was insufficient to make up the energy deficit . Thus, total metabolism was depressed.
Similarly, metabolic depression has been suggested for the copepods Gaussia princeps (Childress 1977) and Subeucalanus subtenuis (Cass 2011). Metabolic depression (also known as metabolic suppression) is a common response among marine animals to environmental stressors such as desiccation, food deprivation and low oxygen (Dymowska et al. 2012). The duration of animal survival in anoxia is inversely related to the extent of their metabolic depression (Hand 1998).
Hyperiid amphipods are the third most abundant type of marine zooplankton in the crustacean subphylum, after euphausiids and copepods (Bowman & Gruner 1973, Diebel 1988. The hyperiid amphipod Phronima sedentaria (Forskal, 1775), in particular, has a worldwide distribution (Shih , 1991 and is abundant in the pronounced oxygen minimum zone of the Eastern Tropical North Pacific. P. sedentaria is a diel vertical migrator, spending nighttime near the surface (0-25 m) and living at depths of 350-600 m during the day . Like most hyperiid amphipods, P. sedentaria often lives parasitically on tunicates or siphonophores, using them as a food source and a brood chamber , Laval 1978, Diebel 1992, Gasca & Haddock 2004, Bishop & Geiger 2006. Phronimids eat the internal tissue of their host leaving the remaining gelatinous matrix in a barrel shape (Hirose et al. 2005) that is propelled through the water with the urosoma (tail) half out the back . Childress and Seibel (1998) suggested that amphipods may be especially tolerant of low oxygen because their gelatinous host provides a substrate that can fuel extended anaerobic metabolism.
This study was conducted to determine whether, and to what extent, P.
sedentaria depresses metabolism to survive migration into a pronounced oxygen minimum zone and to what extent it relies on anaerobic metabolism. To test this, total metabolic rate was estimated from the accumulation of anaerobic end-products and the rates of oxygen consumption under hypoxia and normoxia. Metabolic enzyme activities were also measured as indicators of the capacity for aerobic and anaerobic metabolic rate in P. sedentaria from regions with varying oxygen levels.

Materials and methods
Collection:  . The net was opened and closed using a MOCNESS-type step motor (Wiebe et al. 1985) and equipped with temperature and pressure sensors. Specimens from the California Current (between 33 and 34ºN, 118 and 119ºW) were collected in November 2012 using a 505µm mesh bongo net and a 1 m 2 MOCNESS net with 332 µm mesh. A CTD (conductivity, temperature, density) cast was conducted daily at each station to obtain water profile information (Figure 1).
Specimens from each location were used for metabolic rate experiments.
Enzyme activities were compared between specimens collected from: the ETNP, a region with a pronounced OMZ; the California Current, where the oxygen levels are higher than in the ETNP; and the North Atlantic, which does not have a strong OMZ.
Only female specimens were used for this study because they were more abundant than males in all locations. A low male to female ratio in hyperiid amphipod populations has been demonstrated in specimens from the North Pacific Central Gyre, with a mean ratio of 1 male to 2 females for 49 distinct species. P. sedentaria in the North Pacific Gyre study had a male to female ratio of 2.17 , but the total number caught was only 19 specimens. The lower abundance of males in the collection for this study with a higher sample size indicates the ratio may be closer to the mean ratio of 0.5 found in other species of hyperiid amphipods.

Total metabolism:
Glycogen stores have been shown to be an important energy store in gammariid crustaceans (Foucreau et al. 2013). Assuming glycogen stores are also used by hyperiids amphipods as substrate during anaerobic metabolism, 1.5 moles of ATP are produced per mole lactate accumulated. 6 moles of ATP are produced per mole O 2 consumed during aerobic metabolism (McDonald et al. 1998). Combining these components provides a measure of the total ATP produced (total metabolic rate). ATP production in normoxic conditions is considered to be the stable pool of lactate for an organism's function. This stable pool of lactate was subtracted from the total ATP produced in normoxic and hypoxic conditions. Metabolic depression was then calculated from the reduction in total ATP produced when exposed to hypoxic conditions.

Metabolic rate (MO 2 ):
All respiration experiments were conducted at sea. After collection, specimens found parasitizing tunicates or siphonophores were gently removed from the host before acclimation. For acclimation to laboratory conditions specimens were individually transferred to filtered seawater and allowed to recover for at least 12 hours, ensuring they were in a post-absorptive (starved) state. Filtered (0.2 μm demicap filter, Fisher scientific, USA) and treated (25 μmole l -1 each of streptomycin and ampicillin) seawater was poured into a water jacketed gas-equilibration column, which was connected to a temperature controlled circulating water bath (Lauda, Germany  (Marsh & Manahan 1999). The electrode was calibrated prior to measurements using air-and nitrogen-saturated seawater. The oxygen consumption rate of each specimen was calculated by subtracting the final oxygen concentration in the experimental chamber from final concentration in the blank chamber. At the end of incubations, all specimens were immediately blotted dry, frozen in liquid nitrogen, then transferred to a -80ºC freezer. Weights were determined from frozen specimens in the lab for all specimens except for those collected in the Gulf of California. Specimens from the Gulf of California were weighed on a shipboard balance system ) and frozen in liquid nitrogen. Metabolic rate was determined per hour incubation per gram body weight for each individual.
A temperature coefficient, or Q 10 (= (R 2 /R 1 ) ((T2-T1)/10) , R= oxygen consumption rate, T= temperature), quantifies the factorial change in metabolic rate with 10°C change in temperature and typically falls in the range of 2-3 (Hochachka & Somero 2002). Q 10 was calculated from the average mass specific routine metabolic rate at each temperature.

L-Lactate measurements:
To determine reliability of handheld lactate meters, measurements of lactate standards were compared using the traditional spectrophotometric method by specimens of P. sedentaria were collected from each trawl and immediately frozen in liquid nitrogen and transferred to a -80ºC freezer. Any specimens found on a tunicate or siphonophore, were gently removed prior to freezing. All specimens were alive and in good condition. The CTD data from that day ( Figure 1) indicate that the oxygen concentration at the depth specimens were collected was between 1.6 and 10.6 μM oxygen for the deep trawl and between 48.9 and 195.3 μM oxygen for the shallow trawl. Specimens were weighed in the lab prior to L-Lactate measurement.  (Bergmeyer et al. 1985).

Statistics:
Statistics were performed using the software SAS version 9.2 (SAS institute inc. USA). One-tailed Students t-tests were used to compare metabolic rates scaled to a common body size. One-way Analysis of Variance (ANOVA) and one-way Analysis of Covariance (ANCOVA)s were used to compare differences between treatments.
Linear regression was used to test the relationship between body mass and metabolic rate. Mass-specific metabolic rate (MO 2 ) and enzymatic activities typically decline with increasing body mass (M) according to a power equation where a is a normalization constant and b is a scaling coefficient that describes the slope of the relationship. The relationships of metabolism and enzymatic activities versus mass were linearly regressed on a log scale using KaleidaGraph version 4.1 (synergy software, USA) to obtain the power equation.

Total metabolism:
In the species P. sedentaria from the ETNP, total metabolism (in ATP resulted in a 64% reduction in metabolism compared to normoxic oxygen concentrations at 20ºC (Figure 2).

Field study:
There was no significant difference in lactate accumulation for specimens collected in the shallow trawl versus the deep trawl (t-test: t(19)=-1.52; p=0.1461, Figure 7). Based on CTD data from the day of collection (ETNP station 2, Figure 1), specimens from the deep trawl were collected at oxygen concentrations between 1.581 and 10.637 μM, close to and below P. sedentaria's critical partial pressure (P crit, the oxygen partial pressure at which an organism's aerobic metabolic rate can no longer be maintained,  of 28 µM at 10°C  . The shallow trawl collected specimens at oxygen concentrations above the P crit , between 48.9 and 195.3 μM. Specimens from the deep trawl had an average L-lactate accumulation of 22.56 ± 1.38 µmol g -1 (n=10). Shallow trawl specimens had an average L-lactate accumulation of 26.019 ± 1.78 µmol g -1 (n=11). CTD data from that same day and station recorded that the oxygen levels for the deep trawl were between 1.6 and 10.6 μM oxygen and the shallow trawl were between 48.9 and 195.3 μM oxygen ( Figure   1).
Field caught specimens of P. sedentaria had significantly higher accumulation of lactate than any of the specimens used in laboratory experiments (t-test: t(55)=-11.47, p<0.001), and a significantly higher lactate accumulation than specimens for normoxia treatment experiments (t-test: t(40)=-17.30; p<0.0001, figure 7). Specimens from the two trawls had a combined average lactate accumulation of 24.29±1.58 µmol g -1 . Specimens used in normoxia experiments in the lab had an average lactate accumulation of 3.60±0.67 µmol g -1 .

Enzymatic activity:
For whole specimens from the ETNP, CS activity was an average of 1.11± 0.07 units g -1 (range in mass 0.07-0.47 g). CS activities were plotted on a log axis to obtain the regression equation CS=1.3609x 0.157 ( Figure 8A). Specimens from the  Figure 8A). Enzyme activities were then scaled to a common weight of 0.15g (using the regression equations in Figure 5A) to eliminate weight as a factor in the comparison; regressions could not be compared due to differences in slopes. There was a significant effect of location on scaled CS activity, (one-way ANOVA between subjects design, F(2,50)=30.23; p<0.0001). Mean scaled CS activity was 1.05±0.06, 1.25±0.07 and 2.133±0.16 units g -1 CS for the ETNP, California Current and North Atlantic respectively ( Figure 9A). Tukey's Honestly Significant Difference (HSD) test showed that specimens from the North Atlantic had significantly higher CS activity than specimens from the ETNP and California current ( Figure 9A; p<0.05). There were no significant differences between the ETNP and California Current.
In ETNP specimens, LDH activities scaled positively with body mass, with a regression equation of LDH =49.073x 0.727 ( Figure 8B Figure 8B). Enzyme activity was then scaled to a common weight of 0.15g using the regression equations from Figure 5B. There was no significant effect of location on LDH activity (one-way ANOVA between subjects design: f(2,48)=2.17; p<0.1251). Mean scaled LDH activity was 13.87±1.47 units g -1 for the ETNP, 11.08±1.67 units g -1 in the California Current and 9.89±1.67 units g -1 in the North Atlantic ( figure 9B).

Total metabolism:
When Phronima sedentaria was exposed to conditions matching those to which they are exposed during their daytime migrations into the OMZ, total metabolism was depressed by 78% relative to normoxic conditions at surface temperatures ( Figure 2). Anaerobic metabolism, estimated from lactate accumulation, did increase in hypoxic conditions, but was not enough to compensate for the decrease in aerobic ATP production during hypoxic exposure. Hypoxic conditions alone reduced total metabolism by 35% compared to normoxia at the same temperature.
In pronounced OMZs, where oxygen concentrations are commonly below 5% of air saturation (1% O 2 , ~15 µM ), metabolic depression is anticipated to be a widespread mechanism allowing energy conservation during daytime forays into hypoxia  Dosidicus gigas, reduced total metabolism by 82%; decreased routine metabolic rate from 8.9 to 1.6 umol O 2 g -1 hr -1 and increased mantle octopine production from 0.5 to 5.24 umol g -1 at 10ºC (Rosa & Seibel 2010). Euphausia eximia exhibits a 45% reduction in total metabolism . Studies conducted in the ETNP have demonstrated metabolic rate depression in response to hypoxia in the copepod S.
subtenuis (Cass, 2011;exposed  One study demonstrated that Phronima sedentaria will swim actively only at low light levels (below 3 cd m -2 ), and suggested that this is a mechanism to remain at a constant light level (isolume) and therefore maintain the desired depth in the water column . Thus, it is not possible to conclude definitively that low oxygen is driving the reduced locomotion at depth in OMZs. Regardless, reduced activity in response to low light at depth represents an adaptation that facilitates survival in low oxygen regions.
P. sedentaria feeds more readily at night. Passage of salps through the gut of Phronima at night required, on average, 4 hrs 46 min and during the day more than 14 hours (Diebel 1988). This suggests that P. sedentaria may be able to decrease metabolism by reducing feeding and digestion rates at depth. In addition to reduction of feeding, digestion and movement, P. sedentaria is able to regulate biochemical pathways to accomplish metabolic rate depression. This is evident because the current study eliminated feeding and digestion as factors by the long acclimation period, and movement was minimized by keeping specimens in darkness. Therefore, metabolic depression exhibited by hypoxia-treated specimens compared to the control specimens must have been accomplished by the shutdown of cellular processes. The arrest of cellular processes as potential mechanisms for rate reduction has not yet been examined in hyperiid amphipods, but may include reduced protein synthesis, reduced transcription/translation or ion transport (reviews by: Hand, 1998;Storey and Storey, 2004).
In the OMZ of the California current, some migrating crustaceans are able to regulate their routine metabolism down to the lowest oxygen level they experience during the day, and therefore remain aerobic. These species have very low critical partial pressures (P crit ), at which anaerobic metabolic pathways are upregulated (Pörtner & Grieshaber 1993. At oxygen concentrations below the P crit , anaerobic pathways may be used as a supplement to oxidative phosporylation for ATP production. The crab Pleuroncodes planipes is an example of a pelagic crustacean that is able to remain aerobic when migrating into the OMZ of the California Current. The P crit of P. planipes decreases with temperature, allowing it to have a very low P crit of 0.26 kPa (3.53 µM) at 10°C when migrating into the OMZ (Quetin & Childress 1976). P. planipes is more abundant in the ETNP, which has lower oxygen levels than the California Current. In the ETNP, P. planipes' low P crit at 10°C is most likely sufficient to remain aerobic in the lowest oxygen exposure. The lophogastrid Gnathophausia ingens is a permanent resident of the California Current OMZ that is able to remain aerobic. This species has a large gill surface area compared to crustaceans living at higher oxygen partial pressures, as well as the ability to efficiently ventilate the gills, and a high ventilatory volume in low oxygen . Euphausiids in OMZs have enlarged gill surface areas to increase oxygen uptake from the water so they can continue using aerobic metabolism in addition to supplementing it with anaerobic metabolism. At least one species, Euphausia mucronata, actively swims and feeds while in the OMZ. This is based on equal probability of finding fed animals from day and night collection, and finding plant material in the guts of surface specimens and animal material in those collected at depth (Antezana 2002).
The copepod Gaussia princeps cannot remain aerobic at the lowest oxygen concentrations experienced in its vertical distribution near California, but it can tolerate hypoxic conditions (0.2 ml l -1 , 8.93µM) for approximately 12 hours, presumably by using anaerobic metabolism and a lower metabolic rate during the day at depth (Childress 1977). In more pronounced OMZs, such as the one in the ETNP, it is uncommon for organisms to remain fully aerobic at depth because the oxygen levels are below the P crit for most species. Seibel (2011)  sedentaria is 2.11 kPa (28 µM at 10°C) which was determined from two specimens from the California Current . For this study, specimens from the ETNP were able to survive 6 hours at 0.8 kPa at 10°C (13.4 µM) but accumulated 4.51±1.23 µmol g -1 lactate. Assuming the P crit is the same for the ETNP as the California Current, Phronima sedentaria is adapted to survive below its critical partial pressure by depressing total metabolic rate and increasing anaerobic metabolism.

Metabolic rate:
Thermal effects on respiration are often quantified by the Q 10 , the factorial rise in a rate process for a 10°C increase in temperature (Hochachka & Somero 2002).
The Q 10 for respiration is often between 2 and 3. Outside an organism's normal temperature range, the Q 10 may be elevated (Hochachka and Somero, 2002). Hence, the Q 10 s of 1.26-1.9 reported in this study imply that P. sedentaria is probably within its normal physiological temperature range when vertically migrating. The mean oxygen consumption rate for P. sedentaria when compared between regions and normalized to the same temperature is significantly different between the ETNP, Gulf of California and North Atlantic ( Figure 5). The average rate for the ETNP is approximately 20% higher than the Gulf of California. As shown in Table 1, the rates for the ETNP and California Current both fall within the range of most literature values. The sample size for the North Atlantic is small (4 total). Future work in the North Atlantic on a larger sample size would clarify if rates are higher in this region, The difference in oxygen consumption rate between the ETNP and North Atlantic is relatively small and may also be due to differences in regional productivity at the time of collection. The variation in metabolic rate could be due to differences in food availability in the regions when the studies were conducted, or local adaptations. Bishop and Geiger (2006) reported a mean MO 2 of 13.07 µmol g -1 hr -1 at 20 °C.
The size range overlapped with the present study (range ~0.04-.45g). This value is 10 times higher than literature values and the rates from the current study (Table 1). We suspect these rates are exaggerated by stress as their specimens were only acclimated to laboratory conditions for one hour before measurements were started, which is probably not adequate time to allow for gut clearance or for animals to become accustomed to respirometry chambers. Also, it was not mentioned if experiments were conducted in darkness. These differences in methodology may result in rates for the Bahamas specimens to be elevated relative to the rates reported here. Two specimens from the current study in the North Atlantic were not used in the analysis because they were both brooding females and had very high rates, 18.67 and 9.74 µmol g -1 hr -1 at 20 °C. The higher of the two was very active in the chamber, had been used for photographs prior to incubation, and is representative of an extremely stressed organism. Bishop and Granger (2006) concluded that the metabolic rate of Phronima is not lower than other pelagic crustaceans or pelagic amphipods. However, they compared Phronima's rate to intertidal gammarid amphipods and various epipelagic shrimp species. They did not compare it to any pelagic amphipod species or to the results from Childress 1975.
In fact, the rates of oxygen consumption are close to the relatively low rates of many mesopelagic dwelling organisms. Respiratory rates in some midwater groups decrease with increasing depth of occurrence .  examined the respiratory rates of some midwater crustaceans at temperatures characteristic of their minimum depth of occurrence near southern California.
Excluding P. sedentaria, the range of rate for epipelagic species (Minimum depth of occurrence, MDO, 0-100m) was 17.32-3.47 µmol g -1 hr -1 . The range in rate for mesopelagic (MDO 400-900m) species in the same study was 2.4-0.924 µmol O 2 g -1 hr -1 . P. sedentaria from Childress's study at 10°C had a respiration rate of 1.69 µmol O 2 g -1 hr -1 , which falls in the range of mesopelagic specimens from that region despite the fact that its minimum depth of occurrence is shallow (25 m). Low metabolic rates in mesopelagic zooplankton are hypothesized to be related to the decreasing selection for locomotory capacity because low light levels limit predator-prey interactions among visually oriented organisms , Seibel & Drazen 2007. The low rate in Phronima may be related to its transparency, as this limits their visibility to predators and prey even in well-lit surface waters. Cephalopods, being highly visual predators, exhibit a decline in oxygen consumption with increasing minimum habitat depth similar to the crustaceans. However, squids from the family Cranchiidae have low metabolic rates despite occupying shallow water for at least part of their life history. It has been suggested that transparency relieves them from selective pressures on locomotion and metabolism associated with predator-prey interactions (Seibel & Carlini 2001). Phronima is highly transparent (Johnsen 2001), as is the salp barrel they are housed in. In fact, hyperiid amphipods are the only group of pelagic arthropods that are truly dominated by transparent forms (Johnsen 2001).

L-Lactate:
In laboratory experiments, whole specimens of P. sedentaria exposed to 1% oxygen concentrations (0.8 kPa, 14 µM at 10°C) had a significantly higher accumulation of lactate (4.51±1.23 µmol g -1 ) than specimens exposed to normoxic conditions (3.07±0.58 µmol g -1 , Figure 6). Increasing temperature significantly elevates the amount of lactate accumulated in hypoxic exposed specimens, but did not have a significant effect on normoxic exposed specimens ( Figure 6). The lactate accumulation in hypoxic conditions is lower than reported concentrations for other crustaceans considered to be relatively hypoxia intolerant. The intertidal prawn Palaemon elegans subjected to slightly more hypoxic conditions (0.66 kPa, 8.96 uM at 10°C) for a similar duration accumulated a higher amount of lactate in the whole body (13.1±0.25 µmol g -1 , (Taylor & Spicer 1987)). Lactate accumulation in normoxic conditions was similar to this study: 3.4 µmol g -1 . Palaemon elegans experiences hypoxia in high shore tide pools but is not able to survive environmental anoxia (Taylor and Spicer, 1987). The nordic krill, Meganyctiphanes norvegica, is a diel migrator that has poor anaerobic capacity but occasionally encounters hypoxia when there is poor bottom water exchange in the Nordic fjords. In these conditions, oxygen concentrations at their daytime depth is close to or below their P crit of 4-5 kPa at 8°C. Prolonged exposure (18hr) at 6 kPa PO 2 resulted in haemolymph lactate concentrations of 9.91±1.68 mmol l -1 . There was 100% mortality at 1.8kPa. 18 hours exposure to oxygen of 14.9 kPa, well above their P crit, , led to 3.01±1.05 mmol l -1 lactate (Spicer et al. 1999). Although these lactate values are for haemolymph and not whole organisms, 9 mmol l -1 is higher than P. sedentaria whole body values ( Figure 6) when subjected to oxygen concentrations below its own P crit of 28 µM (2.11 kPa at 10°C; . The low levels of lactate accumulated during hypoxic exposure in P. sedentaria are possible because total metabolism is depressed, an ability that many of these other species apparently lack. It has been proposed that the scope for total lactate production may be correlated with the duration of periods of environmental exposure to hypoxia or anoxia (Pritchard & Eddy 1979). The prawns P. elegans and P. serratus have a low capacity for lactate accumulation, indicating they cannot survive long periods of hypoxia. Immediately after death, maximum lactate concentrations in tissue are 16.7 and 9.6 µmol g -1 for P. elegans and P. serratus respectively (Taylor & Spicer 1987).
Anoxia tolerant crustaceans have been found to have much higher maximum levels of lactate.
For example, burrowing shrimp species Upogebia pugettenisis and Callianassa californiensis exposed to 12 hours of anoxia had levels of 22.1±5.6 and 11.3±0.6 respectively, with maximum levels of 60 µmol g -1 lactate for Upogebia and 20 µmol g -1 lactate for Callianasse. Callianasse can survive up to 60 hours of anoxic condition (Zebe 1982). Live P. sedentaria specimens frozen directly from the trawl had an average lactate accumulation of 24.29±1.58 µmol g -1 , with the highest being 34.76 µmol g -1 (Figure 7). This value suggests a high capacity for lactate accumulation, similar to intertidal species. The laboratory P. sedentaria exposed to 0.8 kPa water had an average accumulation of 4.51±1.23 µmol g -1 lactate, which is much lower than the highest levels from organisms collected in the trawl (mean 24.29±1.58 µmol g -1 , Figure 7). The relatively low levels of lactate accumulated after exposure to oxygen partial pressures below their P crit is consistent with the idea that metabolism is depressed and the requirement for anaerobic metabolism is minimized.
The relatively high capacity for lactate accumulation, as evidenced by the trawl caught specimens, may be related to locomotory and metabolic activity above the routine level such as that required for predator-prey interactions or the migration in low oxygen itself.
There was no significant difference in lactate accumulation for specimens collected in the shallow versus deep trawl ( Figure 7). This may be due to the stresses of capture in the cod end of the net including: crowding, containment, temperature, and pressure changes, among others. This possibility would also explain the higher lactate concentrations in organisms frozen directly from the trawl compared to those acclimated in the laboratory for experiments. This finding indicates the importance of allowing specimens to acclimate to laboratory conditions before conducting physiology experiments.

Enzymatic activity:
The metabolic enzyme CS is an indicator of aerobic potential and LDH is an indicator of anaerobic glycolytic potential. Both of these enzymes have been previously measured in Phronima specimens from Exumas Sound, Bahamas (Bishop & Geiger 2006) where there is not an oxygen minimum zone. The average CS activity of P. sedentaria from the Bahamas was 3.00±1.90 (mean size ~ .25g, range ~0.04-.45g). The CS activity of Phronima in the Bahamas is higher than specimens from all three locations used in this study (mean 1.57 units g -1 , Figure 8B). This difference may be an artifact of the size distribution of the specimens used by Bishop and Geiger, for which we have only the range. The size ranges for their study and ours overlapped but if the distribution is skewed toward large or small specimens, the mean enzymatic activity will be similarly skewed. Their mean CS activity falls within the range of values reported here.
In the North Atlantic, the average CS activity scaled for a 0.15g organism using the measured scaling coefficient of -0.10845, is 2.133±0.16 units g -1 . The California Current average activity for the same size is 1.25±0.07 units g -1 and the ETNP enzyme activity is an average of 1.05±0.06 units g -1 . Specimens from the North Atlantic had a significantly higher CS activity than the other two locations (one-way ANOVA between subjects design, F(2,50)=30.23; p<0.0001).; Figure 8A). Nutritional status contributes to differences in metabolic enzyme activities in copepods, with activity decreasing in unfed specimens (Clarke & Walsh 1993). Similarly, CS activity in the hepatopancrease of two deep sea crabs was significantly lower after one month of food deprivation, although activity in muscle tissue was not affected (Company et al. 2008). CS activity in the North Atlantic was 0.68 units g -1 higher than those measured in the ETNP and California Current. The higher aerobic capacity is consistent with the higher average metabolic rate in the North Atlantic than the other locations ( Figure 6). This variation could be due to differences in food availability in the regions when the studies were conducted.
Gonzalez and Quiñones (2002) hypothesized that LDH activity would be elevated in organisms adapted to low oxygen environments. Evidence in the literature for increased LDH activity in organisms, particularly crustaceans, adapted to hypoxia is mixed. A study comparing enzymatic activities of different copepods species found that epipelagic copepods have a lower LDH activity, and are therefore less reliant on glycolytic energy sources than mesopelagic and bathypelagic copepods. The meso-and bathypelagic species may use glycolysis as an energy source for burst swimming in low oxygen (Thuesen et al. 1998 (2002). C. chilensis is a non-migrator that remains in oxygenated waters and is much smaller in maximum body size than the vertically migrating E. mucronata. Given that C. chilensis and E. mucronata are not only different taxa, but also ecologically distinct, this comparison does little to answer the question at hand.
To test the hypothesis of elevated LDH activity relating to survival in hypoxia, the same, or closely related species should be compared from regions with and without OMZs. This type of comparison would avoid confusion from variation in ecology and life history.
The LDH activity of P. sedentaria from the Bahamas measured at 20º C was 3.00±2.00 units g -1 (mean size ~ 0.25g, range ~0.04-0.45 g, Bishop & Geiger 2006), which is lower than activities for all locations in this study ( Figure 9B). Similar to the difference in CS activity between the present study and Bishop and Geiger, the lower LDH value may be an artifact of the size distribution of the specimens, or variation in nutritional status. P. sedentaria is expected to use anaerobic glycolysis for burst swimming as well as metabolic demand while migrating into regions of low oxygen.
Anaerobic glycolysis may be an important strategy for burst swimming when manoeuvring the salp barrel they live in (Bishop and Geiger 2006). In the current study, P. sedentaria mean scaled LDH activity for a 0.15 g organism measured at 20º C was not significantly different between specimens collected from regions with oxygen minimum zones versus the oxygenated Atlantic Ocean ( figure 9B). This study adds to the growing support that LDH activity is not related to survival in low oxygen environments.
Body size in relation to oxygen availability: Chapelle and Peck ( 1999;2004) propose that the concentration of oxygen in the water limits the maximum potential size in aquatic amphipods. Because oxygen solubility increases at cold temperatures, this finding may explain polar and deep-sea gigantism, and the increase in body mass with latitude among interspecific, but closely related species (Blackburn et al. 2008).
The conclusions of Chapelle and Peck 1999 were questioned by Spicer and Gaston who argued that oxygen partial pressure, not its concentration, would determine the restrictions on size (Spicer & Gaston 1999). They propose that the oxygen partial pressure gradient across the gills is what drives the movement of oxygen. Oxygen partial pressure does not change with latitude, but it does change with depth. Spicer and Gaston (1999) postulated that temperature is the more likely factor for the correlation of latitude and size than oxygen concentration or partial pressure. At the lower latitudes, the influence of increasing temperature on body size may be explained by oxygen limitation from reduced dissolved oxygen in water and increased respiratory rate (Atkinson 1995). In the pelagic environment, some groups of zooplankton show specific adaptations of the circulatory system to enhance oxygen extraction and overcome the limitations of oxygen partial pressure (Childress & Seibel 1998). These circulatory adjustments allow organisms to obtain the oxygen they need, allowing routine metabolism to continue and therefore not impacting body size.
Oxygen does not become limiting until a P crit has been reached. Virtually all marine organisms can regulate their metabolism to at least 68 µM oxygen at 5°C (5kPa). This critical level will increase with temperature  Moving forward, this new approach may achieve a better understanding of body size patterns for some taxa, but it has yet to be tested along gradients of oxygen found in OMZs.
The current study did not set out to address the ongoing discussion of how oxygen concentration, partial pressure or a combination of the two drives patterns in body size of aquatic ectotherms. However, if oxygen concentration is the limiting factor in maximum body size, then this trend would also be seen across the gradients of the water column such as in OMZs. We examined a single species of amphipod from four different locations, each with varying oxygen concentrations; from a severe OMZ in the ETNP to no OMZ in the North Atlantic. Due to this range, we felt it relevant to address the ongoing debate by addressing the size range of specimens from our collection. As noted previously Phronima has not been examined for adaptations for enhanced oxygen extraction from the water. Hyperiid amphipods that have been examined do not have oxygen binding pigments to enhance oxygen extraction from the water (Spicer & Morritt 1995).
Reported literature values for maximum size are in length (reported maximum size of female P. sedentaria is 42 mm ). Length was not measured in this study because specimens were frozen in liquid nitrogen at the end of experiments. Length measurement and microscope light needed for a digital measurement had the potential to increase stress on the specimens which could elevated lactate levels. Some of the specimens collected at each location were brooding eggs, or had juveniles on the inside of the barrel they were removed from.
This indicates that the size range sampled includes size at maturity. In the ETNP, mean size was 0.249 ± 0.018 g and range was 0.0068-0.562g, n=79. In the Gulf of California, mean size was 0.261±0.025g, range 0.037-0.474g, n=27. In the California Current, mean size was 0.123±0.007, from 0.039-0.393g, n=21. North Atlantic specimens mean size was 0.224 ± 0.042g from 0.058-0.497g, n=8. When the 95% largest organisms from each location were compared for weight, location was not a significant factor (one-way ANOVA f(3,5)=4.31;p<0.0748). Oxygen concentration decreases with depth in regions with OMZs. The ETNP has the most pronounced OMZ with the lowest levels of oxygen and the North Atlantic, which has no OMZ, has the highest oxygen levels ( Figure 1). The lack of a significant difference in maximum size between these locations indicates environmental oxygen concentration does not limit maximum size in this species of amphipod.

Significance:
The expansion of hypoxic zones due to global climate change may cause changes in zooplankton distribution which has ecological implications including: altered species composition of an area, changes in prey availability, prey size or predation risk (Ekau et al. 2010), and/ or changes in trophic dynamics due to shifts in  (Vinogradov & Voronina 1962, Somero 2002. Anaerobic metabolism and metabolic depression are not sustainable for long periods of time due to substrate limitations and end product accumulation. Organisms must return to oxygenated surface waters for part of the night to burn off accumulated end products. The combination of increasing temperature and decreasing oxygen supply will compress the habitable nighttime depth range of vertically migrating species , Wishner et al. 2013).
In the southern California Current region, a > 60% decline in some mesopelagic fishes is likely due to the decline of midwater oxygen levels. Oceanic transport of carbon is known as the biological carbon pump. Diel migrating zooplankton play a significant role in this interaction (Ducklow et al. 2001).
Zooplankton consume phytoplankton near the ocean surface at night and migrate down during the day where they metabolize ingested food, release carbon in the forms of dissolved organic carbon (DOC), sinking faecal material, and CO 2 , therefore transporting carbon to depth (Longhurst et al. 1990, Ducklow et al. 2001). Respiration and metabolic activity are among the most important components of carbon flux . To depress metabolism, Phronima will decrease feeding, digestion and respiration. This depression will result in a reduction of faecal pellet production and CO 2 excretion at depth, leading to an overall decrease in the species' contribution to carbon flux. If metabolic depression is common to vertically migrating zooplankton, the decreased carbon input at depth would reduce the efficiency of the biological carbon pump in regions with pronounced OMZs . Burd and colleagues (2010) have noted a problem of imbalances in estimates of organic carbon sources (biogeochemical) and sinks (ecological) below the photic zone. An overestimate of metabolic activity at depth is one potential reason for this imbalance ). Metabolic depression may be one of the reasons for the overestimate of zooplankton contribution to carbon sources at depth since they are reducing their respiration and feeding rates while in the OMZ.
Remineralisation of particulate organic carbon sets the concentration of deep ocean nutrients, which are then returned to the surface via upwelling, providing a feedback loop for the strength of primary productivity. If the carbon pump is reduced, this remineralisation will also decrease (Buesseler et al. 2007). Reduction in carbon transport to the deep sea would lead to amplification of the positive feedback on climate change and reduce total anthropogenic carbon sequestration in the ocean (Sarmiento et al. 1998, Buesseler et al. 2007.

Conclusions:
In the ETNP, the species P. sedentaria is adapted for diel exposure to critical oxygen partial pressures by depressing metabolism while migrating into the OMZ.
LDH activity of P. sedentaria did not increase with decreasing environmental oxygen concentrations. This indicates that the anaerobic enzyme LDH is not used to increase anaerobic potential for P. sedentaria to survive migration into hypoxic conditions. As global warming continues, oxygen minimum zones may expand and Phronima sedentaria may change daytime depth to avoid hypoxic waters. This could affect predator-prey interactions in the region as well as carbon cycling   Figure 2 : Total metabolism of P. sedentaria. Striped: ATP produced from anaerobic metabolism, L-Lactate µmol g -1 . Grey: ATP produced from aerobic metabolism, µmole Oxygen g -1 hr -1 . At 10ºC the combine aerobic and anaerobic ATP production is reduced by 35% in hypoxic compared to normoxic conditions. At 20 ºC total ATP production is reduced by 64% in hypoxic conditions. The migration from normoxic, 20ºC conditions to 1% 0 2 , 10ºC results in a 78% reduction in total metabolism.   Rates were normalized to a common temperature of 20°C using Q 10 s from this study when possible, or assuming a Q 10 of 2. There is a significant different in average Mo 2 between the ETNP, Gulf of California and the North Atlantic: figure 6, ANCOVA: F(3,89)=21.88; p<.0001. Figure 6: L-Lactate accumulation in whole specimens of Phronima sedentaria from the Eastern Tropical North Pacific. Striped: hypoxic, Dark grey: normoxic. Lactate accumulation was significantly higher at higher temperatures for hypoxic conditions (ANOVA, f(2,11)= 4.92; p<0.0297). There was no significant effect of temperature on lactate accumulation in normoxic conditions. For 10°C n= 5 in normoxia and 3 in hypoxia, for 16°C n=10 in normoxia and 9 in hypoxia, for 20°C n=8 in normoxia and 4 in hypoxia. All values shown are means ±SE. Oxygen concentration at the depth the specimens were collected at was between 1.581 and 10.637 μM oxygen for the deep trawl and between 48.852 and 195.333 μM oxygen for the shallow trawl. All values are mean ± se. This indicated organisms frozen directly after collection in the field (trawl samples) were more stressed than organisms allowed to acclimate to laboratory conditions (normoxia samples). Specimens from the trawls had a significantly higher lactate accumulation than specimens acclimated to the lab and used for normoxia experiments (t-test: t(40)=-17.30;p<0.0001). The average lactate accumulation for the two trawls was 24.29±1.58 µmol g -1 . The average lactate accumulation for normoxia experiments in the lab was 3.60±0.67 µmol g -1 . Weight (g) Figure 8A: Mass specific activity Citrate synthase (CS) and 8B: Mass specific activity Lactate dehydrogenase (LDH) in whole specimens of Phronima sedentaria shown on a log scale. N=25 for CS from the ETNP, N=23 for LDH Eastern Tropical North Pacific, N=8 for CS and LDH from the North Atlantic, N=21 CS for the California current, 20 LDH California current. Regression equations are shown on the graphs. CS is an indicator of aerobic potential and LDH is an indicator on anaerobic potential. Location has a significant effect on LDH activity (ANCOVA: f(5,38)= 4.40; P<0.003), error bars represent standard deviation. The x in both plots represents the mean activity level from a previous measurement done in the Bahamas, in the Central Atlantic by Bishop and Geiger 2006, the x axis error bar represents the size range for that study, the y error bar represents the range in activity for their study.

Introduction
Phronima sedentaria is an abundant species of hyperiid amphipod, rumored to have inspired the design of the monster in the 'Alien' movies. Its enormous eyes and claws belie its small size. Phronima parasitizes pelagic tunicates, turning their barrelshaped bodies into a brood chamber Laval, 1978;Diebel, 1988). While most pelagic species of amphipods "hitch-hike" on gelatinous zooplankton that serve as physical and metabolic substrate , Gasca & Haddock 2004, the relationship between Phronima and its parasitized host is unique in that the host is transformed by the parasite . Phronimids eat the internal tissue of their siphonphore or tunicate host leaving the remaining gelatinous matrix in a barrel shape (Hirose et al. 2005) that is propelled through the water with the urosoma (tail) half out the back .
Phronima sedentaria (Forskål, 1775) is found throughout the world oceans and, like many zooplankton, is a diel vertical migrator, spending the day in deeper colder waters and nighttime foraging near the surface , Voznesensky et al. 2004. P. sedentaria may encounter a temperature change of 15°C during its diel vertical migrations, experiencing surface temperatures approaching 30°C in some regions. Such wide temperature variation within the natural range of a species can be stressful (Hofmann & Somero 1995).
Furthermore, the maximum habitat temperatures of many warm-adapted organisms (such as those found in the tropics) are near their maximum thermal limits. Additional increases in temperature due to climate change may not be tolerated by such organisms (Somero 2010).
Oceanic temperatures have increased over the past century as a likely result of anthropogenic carbon dioxide emissions (Trenberth et al. 2007). Increasing environmental temperatures are predicted to affect the physiological performance, and consequently the vertical distribution and ecology of marine organisms (Saltzman & Wishner 1997, Somero 2002, Doney et al. 2012. Organisms subjected to thermal stress typically respond by expressing heat shock proteins (hsps) (DuBeau et al. 1998) and may become oxygen-limited resulting in the upregulation of anaerobic metabolism (Pörtner, 2002). Hsps act as molecular chaperones that are able to prevent/reduce denaturing of proteins and target those that are irreversibly denatured for removal from the cell via the ubiquitin-proteosome pathway. Hsp 70 is one of the most highly conserved heat shock proteins, especially noted for its role in recovery from thermal stress (reviewed by (Feder & Hofmann 1999).
We have quantified the critical temperature for a tropical population of Phronima sedentaria from the Eastern Tropical North Pacific. The expression of hsp 70 and the production of the anaerobic end product, lactate, were quantified at temperatures spanning the range experienced by Phronima sedentaria (Forskål, 1775) across their vertical distribution. We tested the hypothesis that the highest temperatures experienced within the natural range can induce a stress response that would result in an increase in synthesis in heat shock protein 70, and a shift to anaerobic metabolism.

Collection:
Phronima sedentaria (Figure 1)  aboard the R/V Knorr (Woods Hole Oceanographic Institute). Collection was done using a tucker trawl with a thermally insulated cod end .
Specimens were identified according to published taxonomic keys (Shih 1991. Physical vouchers to confirm the identification were preserved in formaldehyde and housed in the Seibel lab at the University of Rhode Island. Specimens were collected from two separate trawls on January 1 st and 2 nd 2009 in discrete tows between the depths of 250 and 300m with a speed of 1.5-2 knots. The Sightings from blue water SCUBA diving, and other trawls has shown that this species can be found near the surface at night in water at temperatures of 23-25° Celsius (personal observations).

Thermal stress:
Specimens were sorted immediately after retrieval and identified quickly under a microscope to reduce stress. Specimens were placed in chilled filtered seawater until experimentation. The specimens in good condition (intact with no injuries) were separated into containers with 0.2 micron filtered sea water and exposed to their approximate nighttime temperature (23°C) for 3, 9 or 24 hours before raising or lowering the experimental temperature. This acclimation of 3, 9 or 24 hours allowed enough time for specimens to recover from any trawl-related stress as indicated by lactate measurements (Elder & Seibel In Prep  ) to make sure water remained above the published critical oxygen partial pressure of 2.11 kPa (28 µM at 10°C) ). Specimens were then taken out of the vial with feather forceps and blotted dry before being immersed in liquid nitrogen and stored at -80 degrees Celsius.
Approximately one half of the specimens used for experiments were frozen at 0100 local time. The remaining experiments were ended at 1300 local time.

Lactate:
Individual whole frozen specimens were ground on ice in a prechilled glass tissue homogenizer (Kimble Chase, USA) using a 1/3 dilution with grinding buffer,

Statistics:
Statistics were performed using the software SAS version 9.3 (SAS institute inc. USA). One-way Analysis of Variance (ANOVA), with between subjects design were conducted to compare differences in lactate accumulation or hsp 70 concentration between treatments.

Results:
At the time of collection surface temperatures of the ETNP were between 23 and 25°C. The maximum surface temperature recorded in the ETNP during this cruise was 27°C. Based on published distribution for Phronima sedentaria, temperatures at the deepest range of daily migrations are between 8 and 10°C. This indicates Phronima sedentaria may experience a temperature change of 13-17°C in the ETNP during diel migration in the ETNP (Figure 3).
There was no significant difference in mortality for 3, 9 or 24 h exposure to nighttime temperature (23°C). Mortality data for those exposure times is combined for subsequent analyses. There was no mortality of specimens between 10 and 20°C.
At 23°C 13% of specimens died (1 out of 7) and at 25°C 30% of specimens died (2 out of 8); Table 1, Figure 4). The most significant mortality occurred at 29°C, at which temperature 50% of the experimental specimens died (4 out of 8 total specimens at that temperature; Table 1, Figure 4). Dead organisms had a significantly higher accumulation of lactate, and so are not included further.
There was no significant difference in hsp 70 concentration or lactate accumulation between specimens frozen at 0100 local time and specimens frozen at 1300 local time. No further evaluation of diel rhythms was conducted.
No significant differences in lactate accumulation were observed between specimens exposed to the night time temperature of 23°C for 3, 9 or 24 hours previous study on P. sedentaria found that specimens frozen immediately after collection had very high levels of lactate (≥20 µmol g -1 ) indicating use of anaerobic metabolism in oxygenated conditions, which is thought to be a result of capture stress (Elder & Seibel In Prep). The low values measured here at temperatures below 29°C indicated that acclimation time was sufficient to recover from collection stress.
For western blot analysis using an antibody for hsp 70, one band occurred at approximately 70kDa ( Figure 6). Dead specimens were not used for western blot analysis. Samples subjected to 23 and 25° C after the 9 hour incubation at 23°C were combined and designated 24°C. Specimens acclimated to 23°C for only 3 hours had significantly lower hsp70 levels than either 9 or 24 hour acclimated specimens

Discussion:
For this study we assessed mortality, lactate and hsp 70 accumulations in specimens exposed to night time temperature (23°C) for varying durations to determine if temperatures at the surface induce a stress response. During daily migrations Phronima sedentaria experiences a temperature change of ~15°C ( Figure   3) with sustained upper temperatures near 23°C at night. Phronima migrates between the surface and 200-350m in during diel migration , Shih 1991.
This temperature change when migrating through the thermocline would be rapid, with a change of up 10°C degrees across 50m (Figure 3). The prediction for this study was that the temperatures routinely experienced by P. sedentaria within its natural range would induce a stress response. If this stress response occurred, it would result in a shift to anaerobic metabolism due to oxygen-limitation (discussed below; Pörtner, 2002) which can be measured by a increase in lactate production under oxygenated conditions. In addition a stress response would result in an increase in hsp 70 concentrations. In contrast to this prediction, we found that both lactate and hsp70 levels were consistent across a temperature range of 10-25°C. There was an increase in both lactate and hsp70 occurred at 29°C, a temperature not experienced during our expedition (December) in the Eastern North Tropical Pacific. However, we further postulated that exposure time could be an important factor in the expression of a stress response. In support of this latter hypothesis, we demonstrated a modest elevation in hsp70 expression in specimens that were pre-exposed to 23°C for 9 or 24 hours relative to those pre-exposed for only 3 hours (Fig. 7). The 13% mortality at 23°C and 30% mortality at 25°C (Figure 4) indicates some amount of stress at night time temperatures. In all acclimation treatments, including 3-hour specimens, subsequent exposure to 25°C for five hours did not result in significant hsp70 expression. The less than 30% mortality and lack of an increase in hsp 70 suggests that Phronima is somewhat tolerant of nighttime temperatures for at least 8 hours, equivalent to its nightly exposure duration before returning to cooler depths. Pörtner (2002) has suggested that upper critical temperatures are related to a mismatch between oxygen supply and demand. This is supported by an elevation in lactate at 29°C. However, lactate levels did not increase at temperatures below 29°C at any exposure duration. This suggests that the heat-shock response in the 9 and 24hour pre-exposures is independent of oxygen stress. In addition temperatures below 23°C did not result in a reduced amount of lactate production or hsp70 concentrations ( figure 5 and 7), indicating that the low lactate levels measured were a true "basal" level. Total mortality was 13% at 23 and 25°C and there was no mortality at temperatures below 23°C (Figure 4). This suggests that the modest heat-shock response at temperatures below 29°C was successful at protecting the organism from detrimental effects of thermal stress.
At 29°C P. sedentaria had a significant increase in lactate production ( Figure   5), a significant increase in hsp 70 concentration (Figure 7), and a 50% mortality rate ( Figure 4). This indicates that the critical temperature range for Phronima sedentaria in the ETNP is between 26 and 29°C, which is slightly higher than the ambient surface temperatures during our expedition.
The increase in lactate production at 29°C represents the onset of anaerobic metabolism. At their critical temperature, organisms experience a failure of ventilatory or circulatory systems to meet elevated oxygen demand, which results in reduced aerobic scope and a transition to anaerobic metabolism under oxygenated conditions. This loss of system function is thought to reflect the earliest level of thermal stress and is known as oxygen and capacity limited thermal tolerance (Pörtner 2010). Our measurements suggest that thermal stress begins earlier than this critical or "pejus" temperature but that protective mechanisms are effective, at least for short periods of time. Although we did not test heart or ventilatory function directly, the onset of anaerobic metabolism in aerobic conditions indicates this mismatch in oxygen supply, which is caused in part by the inability to deliver enough oxygen to the body (Somero 2005). Survival beyond critical temperature leads to a decline in performance and is time limited due to low ATP yield from anaerobic glycolysis (Pörtner 2002, Pörtner 2010. The pejus range (Latin for 'turning worse') is the range when organisms are past optimum conditions but can still survive with reduced aerobic activity (Jost et al. 2012). During the pejus range, there is an increase in ventilation rate with temperature to compensate for increasing oxygen demand with temperature. At the upper pejus temperature ventilation rate becomes relatively constant and Po 2 begins to decrease (Frederich & Pörtner 2000). Oxygen supply to tissues and overall aerobic scope, is suspected to be linked to fitness and functioning at the ecosystem level (Pörtner 2010), although this relationship has yet to be conclusively determined .
Lactate accumulation at 29° C in this study (20.5 ±4.52 µmol g -1 , Figure 5) is similar to the lactate level of 17.15±4.75 µmol g -1 in the same species subjected five hours of environmental hypoxia levels (1% oxygen) at the intermediate temperature of 20°C. Lactate concentrations at 25°C and below were comparable to levels in the previous study when exposed to normoxic conditions 2.85±0.40 µmol g -1 (Elder & Seibel In Prep). This indicates that specimens were experiencing tissue level hypoxia at 29°C despite the oxygenated conditions of the water they are in. This tissue level hypoxia could be due in part to failure of ventilatory or circulatory systems, but factors other than oxygen transport can also be thermally limited and potentially cause decline in performance that leads to tissue hypoxia .
A critical temperature of approximately 30°C is found in several other crustacean species. The spider crab Maja squinado from Roscoff France has a critical temperature close to 30°C, which was indicated by accumulation of anaerobic end products succinate and lactate. This coincided with very low arterial PO 2 values (Frederich & Pörtner 2000). The critical temperature range at which anaerobic metabolism begins in the intertidal crabs Carcinus maenas and Cancer irroratus is 34°C and 30°C, respectively. Interestingly, hsp70 was not detected in either of these crabs, but it may be due to the experimental design, which included a rapid rate of temperature increase (Jost et al. 2012). Our results suggest a 14-30 hour lag in the onset of hsp70 expression following exposure to stressful temperatures.
In addition to anaerobic metabolism, survival beyond critical temperatures is supported by protection of proteins and membranes by heat shock proteins and antioxadative defense (Pörtner & Knust 2007, Pörtner 2010. Hsp 70 is a biochemical indicator for the degree of protein unfolding in a cell and therefore an indirect indicator of protein damage (Hofmann 2005).
The majority of studies on heat shock response in ectothermic invertebrates have focused on intertidal organisms, especially mussels. A theme from these studies is the plasticity of hsp expression, where past thermal history has an impact on induction temperature (Hofmann et al. 2002, Hofmann 2005 (Pennington et al. 2006).
Induction of hsp70 by heat shock has been shown in several studies on freshwater gammarid amphipods, primarily from Lake Baikal in Russia (Bedulina et al. 2010, Shatilina et al. 2011, Bedulina et al. 2013. High constitutive levels of hsp 70 are thought to provide a general protective mechanism against heat shock, and possibly other stresses, in freshwater amphipods (Bedulina et al. 2013). There was a stronger hsp response in intertidal amphipods from a variable habitat (sublittoral) versus a less variable habitat (supra-littoral) (Bedulina et al. 2010). This may indicate that the heat-shock response is critical for tolerating natural temperature fluctuations, even below critical extremes.
Rythmic pre-synthesis of hsps to prepare for potential heat stress, such as prior to low tide, has not been found in rocky intertidal organisms (Hofmann et al. 2002).
The dependable timing of diel migration compared to the variability of low tide levels, suggest that vertical migrators would be more likely to have an anticipatory increase in hsp production than intertidal organisms. For this study half the organisms were frozen at 1am, and half were frozen at 1pm. At 1 am diel migrators would have been at the surface for a few hours, while at 1pm they would have arrived at depth several hours prior. If P. sedentaria were producing hsp in anticipation of vertical migration, one would expect lower levels of hsp in the group subjected to the same temperature frozen at 1pm compared to the group frozen at 1am. However, there was no significant difference in the hsp concentrations or level of mortality between the two freezing times.

Implications/conclusions
23°C is stressful for Phronima sedentaria, but is somewhat tolerated via heatshock response at longer exposure times. Five hour recovery at lower temperature does not result in reduced hsp concentrations. An escalation in hsp 70 concentrations occurred at 29°C (Figure 7) , which coincided with a 50% mortality rate ( Figure 4) and a significant increase in anaerobic metabolism under oxygenated conditions ( Figure 5). In the ETNP P. sedentaria tolerates its normal migration to the surface, but is close to an upper temperature limit and approaching a maximum duration of exposure. Though P. sedentaria experiences a large temperature fluctuation during vertical migration, the consistency of the surface temperatures has allowed P.
sedentaria to adapt to the change, which can be seen by the lack of stress response at surface temperatures of 23-25°C.
Upper thermal tolerance limits are correlated with the maximum habitat temperatures in intertidal organisms (Stillman & Somero 2000). Although limited exposures to current maximum temperatures of the region do not induce a full stress response, the critical temperature of 29°C may be reached in summer and, due to global warming (Deser et al. 2010), during future winters. The Eastern Tropical Pacific warms by approximately 0.8-1.0°C per century (Deser et al. 2010). If organisms are already close to their critical temperatures, global warming will cause some species to be over their thermal limits and may quickly affect their biogeographic range. Increasing temperature and decreasing oxygen supply will compress the night time habitat of vertically migrating species (Seibel 2011, Elder & Seibel In Prep). This change will have important impacts on zooplankton physiology, ecology, and vertical distribution as well as carbon cycling (Vinogradov and Voronina 1962;Somero 2002).

Acknowledgments:
Thanks to Gretchen Hofmann for numerous insightful discussions.    (Shih, 1991;.  10  3  15  3  20  3  25  3  29 3 1 Table 1: Thermal stress experimental setup for initial duration of acclimation to night time temperature of 23°C for 3, 9 or 24 hours and a follow-up five hour exposure to the designated follow-up temperature. n is number of individuals kept at those conditions. n deceased at end is the number deceased at the end of each experiment. Follow-up temperature °C * Figure 5: There was no significant difference in lactate accumulation between the different night time exposure durations (ANOVA: f(2,50)=2.35; p<0.106). Therefore accumulation of lactate in µmol g -1 was averaged among the specimens acclimated to night time temperature (23°C) and then incubated for five hours at the designated follow-up temperatures. All values are mean ±se. There was a significantly higher accumulation of lactate at 29°C (one way ANOVA, F(5,15)=8.26; p=0.0025). Mean hsp 70 concentration ±se for specimens exposed to night time temperature of 23°C for 3(gray), 9 (black) or 24 hours (white) followed by exposure to the designated temperatures. * indicates Hsp 70 concentration at 3 hours was significantly lower than 9 and 24 hour: ANOVA: f(2,47)=7.82; p<0.0012. ** indicates there was a nearly significant increase in hsp 70 concentration in organisms acclimated to 23°C for 24 hours before a five hour incubation at 29°C (ANOVA: F(5,24)=2.57, P<0.0535). Hsp 70 was not quantified for specimens that were dead at the end of the incubation.

Abstract:
This study sought to determine what environmental and ecological factors influence the rate of metabolism in marine amphipods by examining a broad data set from polar to tropical environments, and including transparent specimens. The data set for this study was obtained from the literature and original data. Recent molecular work allowed us to look at hyperiid metabolism in a phylogenetic context.
Understanding patterns of pelagic and deep sea metabolism is important for further understanding of global carbon flux and the consequences of climate change on migration strategies. hyperiid amphipods from the mesopelagic clade Physosomata had significantly lower metabolic rates than species in the epipelagic clade Physocephalata. Transparent species also had significantly lower metabolic rates than nontransparent species. The lower rates in mesopelagic and transparent amphipods support the visual interactions hypothesis: decreasing selection for locomotory capacity limits predator-prey interactions among visually oriented organisms and results in lower metabolic rates. Lower rates of metabolism in benthic and mesopelagic gammarids may also be attributed to the visual interactions hypothesis.

Introduction:
Amphipods are a highly diverse group of peracaridan crustacean. Peracaridans (which include amphipods, mysids, cumaceans, isopods and tanaids) bear eggs in a thoracic brood pouch, where juveniles hatch, rather than having true larval stages (Bousfield 1973, Dick et al. 1998). The synapomorphic character uniting amphipods is the arrangement of the three pairs of posterior appendages known as the uropods (Browne et al. 2007) into a fan-like structure. Pelagic amphipods, though low in relative abundance compared to copepods and euphausiids, are an important part of the epi-and meso-pelagic communities. Hyperiidae and gammaridae are two suborders of amphipods that are found in the pelagic environment.
Gammarids comprise nearly 85 percent of all amphipod species, and are primarily benthic (though there are some pelagic species). Gammarids can be found in almost every environment that has at least some moisture, including: marine, brackish and fresh water, sediment burrows and tubes, caves, as well as humid terrestrial environments (Bousfield 1973). Hyperiids are the younger suborder that evolved from gammarid amphipods (Bousfield 1973, Martin & Davis 2001. Recent phylogenetic work indicates hyperiids are polyphyletic, with two independent radiations from gammarids: the Physocephalata and the Physosomata (Browne et al.

2007
, Hurt et al. 2013). Hyperiids are exclusively marine, primarily found in the open ocean, although there are a few coastal species (Bowman & Gruner 1973). They are commensally or parasitically associated with gelatinous zooplankton at some time in their lives , Laval 1980, Gasca & Haddock 2004. Females are thought to spend the majority of their lives on the host, while male hyperiids may be largely free swimming, spending time on gelatinous zooplankton as a food source during development and as the place to locate females for reproduction . Hyperiid amphipods are known for their large, image forming eyes, with many having double eyes, which may be used for downward scanning of objects that reflect light from above , although eyes of some deep sea members of the clade Physosomata are reduced or absent (Bowman & Gruner 1973). Hyperiid amphipods are dominated by transparent forms (Johnsen 2001). They are also a food source for larger predators and contribute to carbon flux and other biogeochemical cycling via their strategy of diel vertical migration (Vinogradov et al. 2004, Yamada et al. 2004, Ikeda 2013).
Diel migrating zooplankton play a significant role as a link between marine primary producers and upper trophic levels (Ducklow et al. 2001, Ekau et al. 2010).
Zooplankton consume phytoplankton near the ocean surface at night and migrate down during the day where they metabolize ingested food, release carbon in the forms of dissolved organic carbon (DOC), sinking faecal material, and CO 2 , therefore transporting carbon to depth (Longhurst et al. 1990, Ducklow et al. 2001). Estimates of the contribution of migrating zooplankton to this "biological Pump" depend on accurate assessment of the rates of metabolism of species involved. Respiration and metabolic activity are among the most important components of carbon flux ). Burd and colleagues have noted a problem of imbalances in estimates of organic carbon sources and sinks below the photic zone that may result from imprescise or inaccurate metabolic rates used in calculations ).
Metabolism can vary intarspecifically with temperature (Donnelly & Torres 1988), body mass (Glazier 2005(Glazier , 2006, as well as with environmental differences such as oxygen levels (Seibel 2011, Elder & Seibel In Prep) and interspecifically due to ecological habits and phylogenetic influences (Seibel & Carlini 2001). Metabolic rates in visually-orienting pelagic organisms, including crustaceans, are documented to decline with depth (Seibel & Drazen 2007), primarily due to reduced demand for energy for predator-prey interactions in the light-limited deep sea (Childress & Mickel 1985, Seibel & Drazen 2007. Transparency may relieve the demand for strong swimming abilities and high metabolism in such organisms (Seibel & Carlini 2001, Elder & Seibel In Prep). Torres et al. (1994) noted that Antarctic pelagic crustaceans (when normalized to the same mass and temperature) have similar or lower metabolic rates compared to temperate species, suggesting no metabolic temperature compensation. Torres et al (1994) also noted that the representation of amphipods within the micronektonic crustacea community decreases with increasing sea surface temperature. Due to the abundance of amphipods in colder environments, previous work on their metabolism has primarily been conducted in polar coastal waters. Metabolic rates of pelagic crustaceans, including both gammarid and hyperiid amphipods, decrease with increasing depth of occurrence in the Antarctic (Ikeda 1988, Torres et al. 1994. A recent analysis found that metabolic rates of pelagic amphipods decrease with increasing depth of collection in all regions (Ikeda 2013); however, depth of collection is not always ecologically relevant and therefore it is difficult to interpret this result.
Rates of hyperiid amphipods were higher on average than rates of gammarids in the Antarctic, partially due to the smaller body size of hyperiids in that region (Torres et al. 1994) or to their association with the benthos, providing greater opportunities for refuge from predators.
This study aimed to determine what environmental and ecological factors influence the rate of metabolism in marine amphipods by examining a broad data set from polar to tropical environments and including transparent specimens.
Specifically, we tested the effect of temperature, oxygen, body mass, habitat (benthic versus pelagic) and transparency on the metabolic rates of diverse amphipods across a latitudinal and depth range. The data for this study was obtained from the literature and original data. Recent molecular work (Hurt et al. 2013) allowed us to look at hyperiid metabolism in a phylogenetic context. Understanding patterns of pelagic and deep sea metabolism is important for further understanding of global carbon flux and the consequences of climate change on migration strategies.

Collection:
Collection locations and methods are listed in table 1. For most locations specimens were collected with a Mother Tucker Trawl, a modified opening-closing Tucker Trawl equipped with a 30 l thermally insulated cod-end .
The net was opened and closed using a MOCNESS-type step motor (Wiebe et al. 1985) and equipped with temperature and pressure sensors. Immediately after the codend reached the surface, the contents were placed in a large container full of chilled seawater. At McMurdo station, Antarctica, specimens were collected by hand along shore. The species Abyssochromeni was also collected using a bait bag filled with dead fish and left overnight in a hole on the sea ice just south of McMurdo station. For all collection methods, upon retrieval, specimens were individually transferred to filtered seawater and allowed to recover for at least 12 hours, ensuring they were in a post-absorptive (starved) state. Specimens were identified using published keys  and representatives of each species were preserved in 4% formaldehyde, or 70% ethanol to confirm identification.

Metabolic rate (MO 2 ):
All respiration experiments were conducted at sea or, in the case of Antarctica, at the lab in McMurdo station. Depending on the size of the organism, either 25 ml or 50ml glass gas-tight syringes were used as respiration chambers. The chambers were filled with filtered (0.2 μm demicap filter, Fisher scientific, USA) and treated (25 μM liter -1 each of streptomycin and ampicillin) seawater, and a single specimen was immediately placed in the chamber using feather forceps. A blank chamber with no specimen was filled with identically treated water and processed simultaneously to monitor background respiration of microbes. The chambers were sealed and incubated in a temperature controlled circulating water bath (Lauda, Germany). All experiments were carried out in darkness. Experiments were conducted for 5-27 hours. The size, volume and metabolic rate of individuals determined the duration needed to provide sufficient time for measureable changes in oxygen saturation.
Water was removed from incubation chambers using a 500 microliter syringe (Hamilton, USA). Oxygen concentrations of the water in incubation chambers were measured at the end of the experiment using a Clark-type oxygen electrode ) connected to a Strathkelvin Instruments 782 Oxygen Interface (Strathkelvin Instruments, United Kingdom). The oxygen electrodes were maintained in a thermally jacketed electrode holder (MC100 Microcell, Strathkelvin Instruments, United Kingdom) attached to a water bath at the appropriate experimental temperature (Marsh & Manahan 1999). The electrode was calibrated prior to measurements using air-and nitrogen-saturated seawater. The oxygen consumption rate of each specimen was calculated by subtracting the final oxygen concentration in the experimental chamber from final concentration in the blank chamber. At the end of incubations, all specimens were immediately blotted dry, frozen in liquid nitrogen, then transferred to a -80ºC freezer. Specimens were weighed on a motion compensated shipboard balance system  and frozen in liquid nitrogen. Metabolic rate was determined per hour incubation per gram body weight for each individual.
A temperature coefficient, or Q 10 (= (R 2 /R 1 ) ((T2-T1)/10) , R= oxygen consumption rate, T= temperature), quantifies the factorial change in metabolic rate with 10°C change in temperature and typically falls in the range of 2-3 (Hochachka & Somero 2002). Q 10 was calculated from the average mass specific metabolic rate at each temperature. Mean metabolic rates of hyperiid amphipod species were normalized to the same temperature (10°C) and mass (0.25g) using measured Q 10 and scaling coefficient and plotted as a function of the minimum depth of occurrence (MDO) ( Figure 2). Routine metabolic rate data was summarized from our studies and those in the literature.

Rates from Literature:
Oxygen consumption values were adjusted to 10°C, which was chosen because it was the approximate median temperature for the range of data collected. Published temperature coefficients were used where available, or we assumed a Q 10 of 2. For comparison between regions and across depth gradients, species rates were adjusted to a common body size of 0.25g using published scaling coefficients, the coefficient derived from the present study for data original to this study, or assuming a scaling coefficient of -0.25 for mass specific metabolism. Quarter-power scaling is not universal (Glazier 2006) but is commonly used when scaling coefficients are not available because of its prevalence; thus we considered it a safe assumption (Schmidt-Nielson 1984). Mean oxygen consumption rates were used for some species because published papers often do not give all the data for individual specimens but give mean values and size range.
Literature values of oxygen consumption were used only if they met the following requirements: Measurements were made in the absence of food, in darkness, and at temperatures within the natural environmental range for the particular species.
All rates in the present study are presented as oxygen consumed per unit wet body mass. Thus, only studies providing wet mass, or the data necessary to calculate wet mass (e.g. dry mass and % water), were included.

Statistics:
Statistics were performed using the software SAS version 9.3 (SAS institute inc. USA). One-way ANOVAs were used to compare differences in metabolic rates between groups. Linear regressions were also used to compare the relationship of metabolic rate with mass or minimum depth of occurrence. Mass-specific metabolic rate (MO 2 ) and enzymatic activities typically decline with increasing body mass (M) according to a power equation (MO 2 = aM b ), where a is a normalization constant and b is a scaling coefficient that describes the slope of the relationship. For linear regression against habitat depth, species were assigned a minimum depth of occurrence (MDO), the depth below which 90% of the individuals of a given species have been captured ). However, due to the uncertainty of the species' vertical distributions, we also used a one-way ANOVA to assess differences in metabolic scaling curves for species believed to live above or below a particular depth. For pelagic species, a depth of 100m was used, since the majority of the decline in metabolism occurs in the upper 100 m in other visuallyorienting pelagic taxa. For benthic species, a depth of 500 m was used following the protocol of Torres et al. (1994). Benthic organisms typically display little or no decline in metabolic rate with depth (Seibel & Drazen 2007).
A linear regression was used to compare the relationship of metabolic rate and environmental temperature in different species of the hyperiid amphipod Themisto. A one tailed t-test was also used to compare the mean metabolic rate of cold water (less than 1°C) versus warmer water (5°C or above) species of Themisto.

Results:
Oxygen consumption rates of eight species of hyperiid amphipod and two species of gammarid amphipods were measured (Table 2) from six study locations (Table 1). Representative study specimens are pictured in Figure 1. MO 2 values for all species of hyperiid amphipods combined declined significantly with body mass (MO 2 =1.1074M -0.38 , r 2 =0.54; Figure 2). Regression equations for individual species are also provided in the figure legend. Metabolic rates of 25 additional hyperiid and 7 gammarid amphipod species were taken from the literature and used with data original to this study for the remaining analysis (Tables 3 and 4).
Mass-and temperature-normalized metabolic rates declined significantly with increasing depth of occurrence according to the regression equation MO 2 =1.074 -0.38 (p<0.0087, Figure 3). The previously published regression equation for decline in MO 2 with increasing depth of occurrence in pelagic crustaceans  visually appears to be outside of the 95% confidence limits for the regression equation for hyperiid amphipods below 150m (Figure 3).
Only five species of hyperiid amphipods from the literature and this study had a minimum depth of occurrence below 150 meters (Table 3). All these specimens are in the clade Physosomata. Species with a minimum depth of occurrence deeper than 150 meters had a mean metabolic rate of 1.54±0.67 µmol O 2 g -1 hr -1 (n=5), which is significantly lower than the mean metabolic rate of 7.27±0.85 µmol O 2 g -1 hr -1 (n=28) for species with a minimum depth of occurrence above 100 meters (ANOVA: F(1,31)=7.46, P< 0.0087, Figure 4).
Metabolic rates were available for five species in the genus Themisto from five different locations (Table 3). T. japonica has been studied at two locations. The environmental temperature ranged from -0.1°C in the Barents Sea (North of Russia, connects to the Arctic Ocean) to 10°C in the North Atlantic. Metabolic rates of these specimens were compared to see if rates from polar species were higher than species from warmer climates once normalized to temperature (10°C) and scaled to a body mass of 0.25g. Metabolic rates were not significantly related to environmental temperature, according to the regression equation MO 2 =5.63-0.048x, R 2 =.09, Figure 6. MO 2 is not significantly different between specimens from cold water regions (less than 1°C) and specimens from regions with temperature of 5°C or higher (t-test: t(4)=-1.84; P<0.1617).
Rates of gammarid amphipods for the species Epimeriid sp. and Abyssochromeni plebs (pictured Figure 1G) were adjusted to 10°C. In both species, the MO 2 scales significantly with body mass (Figure 7). The scaling relationship published by Seibel and Drazen (2007) for benthic crustaceans (MO2=1.9031m -0.27 ) is also plotted (Figure 7). This work on benthic crustaceans included gammarid amphipods from tropical to polar environments. Neither the slopes nor the elevations of the regressions for the two Antarctic species was significantly different than the regression line from Seibel and Drazen (2007, Figure 7). However there is a substantial interspecific variation ampng the species studied by Seibel and Drazen (2007). It may be worth noting that Epimeriid sp. falls entirely above the regression line.

Discussion:
This study increases the number of amphipod species for which metabolic rates are available by nearly 50% and more than doubles the number of pelagic species measured. The mean metabolic rates for marine hyperiid amphipods were lowest in the deep-living Physosomata, with a mean metabolic rate of 1.54±0.67 µmol O 2 g -1 hr -1 and highest in the shallow-living Physocephalata with and mean metabolic rate of 7.27±0.85 µmol O 2 g -1 hr -1 ( Figure 5). This 3-4-fold variation in metabolism is attributable to physical, morphological, ecological, and phylogenetic differences between species, as well as size differences within and between species. Specifically, habitat depth, temperature, transparency, and body mass are determinants of metabolism, while their relative association with the benthos or latitude had little effect on metabolism. Within the Physocephalata metabolic rates of transparent species had lower metabolic rates (mean of 3.74±0.88 µmol O 2 g -1 hr -1 ) than nontransparent species (mean12.00±1.38 µmol O 2 g -1 hr -1 ) ( Figure 5). This difference in metabolic rate may be attributed to the reduced susceptibility to visual predation and therefore less reliance on strong locomotion in transparent species. Transparency is achieved, in part, by increased water content of tissues that further reduces wet-mass specific MO 2 .

Habitat Depth
Respiratory rates in many midwater groups decrease with increasing habitat depth , Seibel & Drazen 2007. Low metabolic rates in mesopelagic zooplankton are hypothesized to be related to the decreasing selection for locomotory capacity because low light levels limit predator-prey interactions among visually oriented organisms (Cowles et al. 1991, Seibel & Drazen 2007. Visually-limited groups (those without image-forming eyes or those found in constant darkness) have low, but variable, metabolic rates at all depths (Seibel & Drazen 2007).  examined the respiratory rates of midwater crustaceans at temperatures characteristic of their depth of occurrence near southern California. In Childress 1975 the range of metabolic rates for epipelagic species (MDO 0-100m) was 17.32-3.47 µmol g -1 hr -1 . The range in rates for mesopelagic (MDO 150-900m) species in the same  study was 2.4-0.924 µmol O 2 g -1 hr -1 . Childress' study included one gammarid and two hyperiid amphipods. Data from those specimens were included in the present study (Table 3).
Studies on the vertical distribution of hyperiid amphipods are scarce, but most species in the clade Physocephalata have a minimum depth of occurrence of 0-50m, well within the epipelagic zone (Table 3). The hyperiid amphipod clade Physosomata is generally found in the mesopelagic and bathypelagic zones , Hurt et al. 2013. Eyes are often small, inconspicuous or absent in the Physosomata (Bowman & Gruner 1973. In the mesopelagic, vision is used to see dim, downwelling daylight and bioluminescence, which is often point-source flashes not relevant for protracted predator-prey interactions (Warrant & Locket 2004). In this study, Physosomata with inconspicuous or absent eyes are represented by the genera Lanceola, Scina, and Chuleola. The eyes of Scina crassicornis function like other mesopelagic zooplankton, with a spectral sensitivity and response latency that allows them to capture the maximum amount of light in the mesopelagic zone (Cohen & Frank 2007). Megalanceola spp. have eyes that are narrow, kidney shaped and relatively large . All of these Physosomata genera have significantly lower metabolic rates than hyperiids that have a minimum depth of occurrence above 100m (Figure 3). The rate for these species (1.54 ± 0.67 µmol O 2 g -1 hr -1 ) is within the range of rates for other mesopelagic crustaceans examined by . The mean for epipelagic non-transparent hyperiids in this study (12.00 ±1.38 µmol O 2 g -1 hr -1 ) is within the range of other epipelagic crustaceans. The only other Physosomata species in this study was Paraphronima spp. (Figure 1 E), which, along with Cystosoma spp., was only recently assigned to this clade (Hurt et al. 2013). Paraphronima spp. is a vertical migrator with a much shallower minimum depth of occurrence than the other Physosomata represented but does reach depths of 500m or deeper during the day (Brusca 1967. Since Paraphronima spp. spends some time in the photic zone, the low metabolic rate of this species may be due to its nearly complete transparency (see below), more than its presence in the dimly-lit mesopelagic zone.
Deep living gammarids had lower rates (1.25 µmol O 2 g -1 hr -1 ) than shallow living gammarids (4.01 µmol O 2 g -1 hr -1 ) at 0.5°C (Torres et al. 1994). Data from the Torres study is included in the present study (Table 4). Data collected for this study had a similar trend; deep living gammarids (500m and below) had an average metabolic rate of 3.18 µmol O 2 g -1 hr -1 when normalized to 10°C and 0.25g.
Epipelagic gammarids, from 100m and above had an average metabolic rate of 6.16 µmol O 2 g -1 hr -1 . Torres et al (1994) noted that this difference in metabolic rate may not be due to reduced activity levels, since the ratios of maximum to minimum rates are not significantly different between shallow and deep living specimens. Deep living species of micronekton crustaceans are able to alter their activity levels and metabolic rates equivalently to shallow species (Torres et al. 1994). Their respiratory rate is lower probably because their body composition is different (reduced protein and lipid levels, higher water levels ) than epipelagic amphipods.
Some studies examine metabolic rates as a function of depth of capture rather than minimum depth of occurrence. In these studies organisms are captured in net tow where the net remains open between the deepest depth and the surface. So epipelagic organisms are also captured in tows that start in the mesopelagic because the tow collects continuously from the mesopelagic up to the surface. The mean metabolic rate of all individuals from a fixed collection depth interval is plotted against the middle of the depth interval. As noted by Childress et al. (2008), this method results in the same species represented in more than one depth interval. For example Ikeda (2013) reports the species Phronima sedentaria with both deep and shallow capture depths, 2 m and 750 m. This deeper depth is not ecologically significant since Phronima sedentaria is not found below 600m . Also because Phronima sedentaria is a vertical migrator, the minimum depth of occurrence is a more accurate depiction of the ecological pressures relevant to metabolic demand and visual predator-prey interactions. Comparing this epipelagic species to deep sea species that do not spend time in the photic zone does not test the visual interactions hypotheses (Childress et al. 2008). In addition, this misleading depth range results in inaccurate estimates of the contribution of these organisms to carbon flux.

Transparency
Several species of epipelagic amphipods had rates that were low relative to other epipelagic crustacean species. Phronima sedentaria ( Figure 1D), like Paraphronima spp., was in the range of mesopelagic specimens from Childress's study (1975) on midwater crustaceans, despite the fact that its minimum depth of occurrence is shallow (25 m). Elder and Seibel (in prep.) hypothesized that the low rate in P.
sedentaria is related to its transparency, which should limit its visibility to predators and prey even in well-lit surface waters. Similarly, squids from the family Cranchiidae have low metabolic rates despite occupying shallow water for at least part of their life history. It was suggested that transparency relieves them from selective pressures on locomotion and metabolism associated with predator-prey interaction (Seibel and Carlini, 2001). Other Cephalopods, being highly visual predators, exhibit a decline in oxygen consumption with increasing minimum habitat depth similar to the crustaceans (Seibel et al. 1997). Transparency is one of the few means of camouflage from visual predators available to oceanic organisms. The distribution of transparency in primarily photic aquatic environments overlaps with the habitat of visual predators (Johnsen 2001), providing evidence that this is an adaptation to avoid visual predators.
Paraphronima is the only transparent species of Physosomata for which metabolic rates have been measured, but in line with the low activity of Physosomata, the metabolic rate is lower than any of the transparent Physocephalata. Transparent members of Physocephalata, including many Oxychechalidae ( Figure 1B and H for examples), are slender and elongate (Bowman & Gruner 1973), improving their swimming efficiency due to their streamline shape. All transparent species from the family Oxycephalidae had higher metabolic rates than the other transparent amphipods species. This is most likely due to their reported strong swimming abilities and relatively active behavior.
The only two species of transparent hyperiid amphipods that have been previously studied for metabolic rate are Phronima sedentaria , Mayzaud et al. 2005, Ikeda 2012) and a single specimen of Oxychephalus clausi (Ikeda & McKinnon 2012). Rates for Phronima sedentaria were comparable with the data reported from our previous work (Elder & Seibel In Prep). Oxychephalus clausi from the coral sea had an average rate of 5.5 µmol O 2 g -1 hr -1 (Ikeda & McKinnon 2012). Oxychephalus clausi has been described as a predatory species that can rapidly devour salps, has a greater developed pleon and urosome (tail region), and can accelerate for burst swimming . The well developed pleon and urosome would assist Oxychephalus clausi in achieving faster swimming speeds than species with more reduced structures. This is in line with the other activity levels of the other species of Oxychephalidae. The ecological reasons selecting for this more active mode in Oxychephalidae compared to other hyperiids is not known.

Temperature and Metabolic Cold Adaptation
Metabolic cold adaptation is the outdated concept that polar ectotherms have compensated for the depressing effect of temperature on metabolic rate (Clarke 1980).
Metabolic rates represent the sum of numerous energetic expenditures, so elevating metabolic rate and the amount of ATP required is energetically costly. It has been noted that there is no selective advantage in arbitrarily increasing metabolic rate in order to achieve compensation for temperature (Clarke 1993). Therefore metabolic cold adaptation as a concept is largely abandoned. However, polar species are able to adjust or compensate some physiological processes (e.g. enzyme-mediated reactions (Kawall et al. 2002)) for living at low temperatures so that they attain greater activity levels than similar organisms from a warmer habitat acclimated to the same low temperature (Crockett & Sidell 1990). There is an advantage to increasing metabolic rate to a level consistent with that required for ecologically relevant activity levels ). For example, in pelagic pteropods, active predator-prey interactions select for high maintenance of wing-beat frequencies, the cost of which is reflected in the whole-animal metabolic rate. Polar and temperate species need to swim at similar rates to capture prey, despite differences in habitat temperature, so energy consumption in support of swimming is similar across a latitudinal range ).
Previous work on amphipods has found no support for metabolic cold adaptation. Torres et al (1994) compared rates of micronekton from studies in the Antarctic, temperate and tropical studies across a depth range. The mass and temperature corrected regressions for Antarctic representatives are either below or similar to regressions from lower latitude locations (Torres et al. 1994). Cowles et al. (1991) also found a lack of support for metabolic cold adaptation when comparing crustaceans from California and Hawaii. Decline in metabolism with depth in isothermal waters support the conclusion that this decline is related to relaxation of pressure from visual predators, not from temperature (Torres et al. 1994).
To examine if temperature compensation of metabolism is present in hyperiids, the genus Themisto was compared. For this genus there are five species from five different locations, with two representatives of one species ( Figure 1C pictures Themisto abyssorum). The hyperiids in the genus Themisto are numerous in abundance and high in biomass in the world oceans and are primarily found in high latitude seas (Yamada et al. 2004). Themisto are predominantly carnivores and are well studied because of their abundance and importance as an intermediate between primary producers and higher trophic levels (reviewed by (Auel & Werner 2003). For this comparison, three species of Themisto were from regions with environmental temperatures of 0.5°C or below, two were from the western subarctic pacific where their habitat temperature is 5°C (Yamada & Ikeda 2003), and one species was collected from the North Atlantic Ocean where habitat temperatures were 10°C (Table   3). When normalized to the same temperature and body mass, the metabolic rates of these species were not significantly different ( Figure 6). This indicates that metabolic cold adaptation is not present in this genus. If the species from the 0.5°C waters had a significantly higher rate than the species from the 10 and 5°C waters, than it would indicate metabolic cold compensation.
Two Antarctic gammarid amphipods from this work, Abysssochromene plebs ( Figure 1G) and Epimmeriid sp. were normalized to 10°C and compared to the regression equation for benthic crustaceans from Seibel and Drazen (2007), also normalized to 10°C (Figure 7). That regression includes an extensive review of benthic amphipods from polar to tropical and temperate waters. There was no evidence of metabolic cold adaptation in the polar organisms compared to crustaceans from other latitudes (Seibel & Drazen 2007). The rates of gammarid amphipods from Antarctica in this study were not significantly different than the Seibel and Drazen (2007) regression, further substantiating the lack of evidence for metabolic cold adaptation in crustaceans.

Ecology
Classification and phylogenetic confusion in hyperiids results from convergent evolution of traits correlated with their pelagic life history and parasitic relationship with gelatinous zooplankton (Hurt et al. 2013). They are known descendants of gammarid amphipods, which are predominantly benthic. Hyperiids are exclusively pelagic, but are commonly thought of as living a benthic-like existence because they live on gelatinous zooplankton (Laval 1980). Juveniles are deposited on the host from the brood pouch of the female. Juveniles are unable to swim on their own and therefore would be unlikely to encounter a host on their own. Females do not produce large numbers of offspring (Gasca & Haddock 2004), as many broadcast spawning pelagic species do to increase chance of progeny survival. This indicates that the gelatinous host acts as a secondary brood pouch for further development of the larva (Laval 1980). The importance of gelatinous hosts for males of pelagic amphipods is not clear. There have been far fewer observations of males on hosts ), but as previously noted it has not yet been fully examined if males are more independent from hosts than females (Gasca & Haddock 2004).
The time hyperiids spend on substrate (hosts) does not appear to be equivalent to living an entirely benthic lifestyle, because of their relatively high metabolic rates compared to benthic gammarids and to behavioral observations. The degree of dependence on hosts varies according to species (Ohtsuka et al. 2009). Many species of amphipods have been noted to be strong swimmers and relatively active, especially shallow living ones, and have been observed without a gelatinous host. Pelagic amphipods remain in the water column by either active swimming or a combination of active swimming and resting on floating substrates (gelatinous zooplankton) (Ikeda 2013). Some species, such as Themisto occur in swarms in gatherings of up to 38 individuals have been observed from submersibles, though the reason for this behavior is unknown (Vinogradov 1999). Based on gut content analysis Themisto pacifica and Cyphocaris challengeri preyed on copepods, amphipods, ostracods and cladocerans (Haro-Garay 2004). Examination of the mouth parts of T. pacifica determined they can feed on gelatinous tissue and hardier organisms that possess an exoskeleton. This indicates that T. pacifico has raptorial capacity. Strong palps and sharp incisors on C.
challengeri indicate they are predatory by nature, and are able to eat exoskeleton covered small planktonic invertebrates. These morphological conclusions were supported by gut content analysis. C. challengeri is an active exclusively carnivorous predator; T. pacifica is a more passive predator relying on microphagy/canivory (Haro-Garay 2004).
Some pelagic gammarid amphipods are also adapted to associate with gelatinous zooplankton (Vader 1972 Benthic gammarids have a lower metabolic rate than nontransparent pelagic hyperiids and gammarids. Their lower rate may be explained by the lack of energy expenditure for vertical migrations or finding a host. Epipelagic amphipods  are known to vertically migrate on a diel cycle. While they may be traveling on their gelatinous host, their high activity levels indicated they likely spend some time swimming for this migration. Rather than remaining immobile on the gelatinous host, adult hyperiids wander from host to host, and use them as a platform for attaching other prey (Laval 1980). Transparent species have a lower metabolic rate than other mesopelagic species because of the relief from pressure for locomotory performance to avoid visual predators discussed above. Pelagic deep sea hyperiids have the lowest rates of any amphipods. These species are known to have a less hydrodynamic shape (Bowman & Gruner 1973) than gammarids and epipelagic amphipods, as well as visual systems that function well for a lifestyle of sitting and waiting for prey when they are not on gelatinous zooplankton (discussed above in habitat depth section). Deep sea hyperiid amphipod species do not conduct large energy costly vertical migrations on a diel cycle; this also reduces their selection for strong swimming abilities.
Respiration rates of amphipods are similar to euphausiids (krill) but greater than copepods (Ikeda 2013). Euphausiids, which are entirely pelagic, are considered active swimmers with well-developed eyes. Copepods lack image-forming eyes and are documented to spend less time swimming, alternating between a swimming mode and a cleaning mode (Schmitt et al. 2006 However some amphipods also have high ash content relative to copepods and euphausiids (mean 25% of dry mass) due to their robust exoskeletons (Ikeda 2013).
This high ash content would decrease buoyancy. Negative buoyancy may result in amphipods actively swimming to maintain position in the water column unless "hitchhiking" on gelatinous zooplankton, though no direct studies have been. Amphipods may use the host jellies as a resting place and refuge between forays, and a food source at times. The species that approach neutral buoyancy are likely to spend less energy swimming than negatively buoyant species. However, previous work found there was no significant relationship between relative buoyancy and overall respiratory rate . Relative buoyancy in pelagic crustaceans decreases with increasing depth of occurrence , but no specimens from the clade Physosomata were included. This clade may be closer to neutral buoyancy because their globular appearance indicates a high water content. Also they are known to be more inactive than pelagic species.

Conclusions
This study demonstrates that the differences in amphipod metabolism are associated with physical, morphological, ecological, and phylogenetic differences between species. The species in the hyperiid amphipod clade Physosomata had significantly lower metabolic rates than species in the clade Physocephalata. The clade Physocephalata is typically more streamline in morphology, and has a minimum depth of occurrence in the epipelagic zone. Transparent species are found in both Physosomata and Physocephalata, although metabolic rate was only available for one species of Physosomata (Table 3). The lower rates in mesopelagic and transparent amphipods support the visual interactions hypothesis: decreasing selection for locomotory capacity (by transparency or remaining below the photic zone during the day) limits predator-prey interactions among visually oriented organisms and results in lower metabolic rates (Childress & Mickel 1985, Cowles et al. 1991, Seibel & Drazen 2007 Table 2. Experimental specimens original to this study, including samples sizes, temperature at which experiments were conducted, and corrected metabolic rates. MO 2 =metabolic rate. ETNP= Eastern Tropical North Pacific. Q 10 is the temperature coefficient that quantifies the factorial change in metabolic rate with 10°C change in temperature.  Table 3: Hyperiid amphipod data collected from the literature and this study normalized to a common temperature (10°C) and body mass (0.25g). 11 of the 33 species are original to this study (see table 2). Clade is based on molecular phyolgenetic analysis from Hurt et al. 2013. MDO is minimum depth of occurrence, which is provided in the reference study, or other sources as noted. * distribution from Shulenberger 1977, + distribution from , ~ distribution from Coyle and Pinchuk 2005. When specific distribution values were not available for that species, the MDO is based on known distribution of other species in the same genus based on the cited reference. Q 10 is the temperature coefficient that quantifies the factorial change in metabolic rate with 10°C change in temperature. Q 10 from table 2 was used, or literature values when available, otherwise a Q 10 of 2 was assumed. A scaling coefficient of -.25 was assumed, except for the scaling of the genus Phronima, for those rates the regression equations from were used: MO 2 =3.92M (-0.263) in the North Atlantic, MO2=1.907M (-0.25) in the Gulf of California and MO2=2.45M (-0.208) in the Eastern Tropical North Pacific (Elder and Seibel in Prep).  Table 4. Gammarid amphipod data collected from the literature and orignial to this study. 2 of the 12 species are original to this study (see Table 2). Abbreviations are as in Table 3. ~ distribution from Coyle and Pinchuk 2005 .  Table 2 has further details on collection location etc. All photos by L. Elder. Figure 2. MO 2 for hyperiid amphipods from this study by weight on a log scale. All rates were adjusted to 10C using calculated Q 10 s or assuming a Q 10 of 2 where necessary (see  Figure 3: MO 2 for pelagic hyperiid amphipods compared to minimum depth of occurrence (MDO) plotted on a semi log scale. MO 2 was scaled to a common weight of 0.25g and a common temperature of 10°C. Each point is the mean for one species of amphipod (see table 3 for raw data). There is a significant decline in MO 2 with increasing depth of MDO (F(1,31)=7.84;p<0.0087). The gray line is for pelagic crustaceans from , the regression equation for that line is MO 2 =23.02MDO -0.59. The black regression line is for hyperiid amphipods in this study, regression equation: MO 2 =7.22MDO -0.29 , r 2 0.28. Top and bottom dotted lines are the 95% confidence limits(cl) for the hyperiid amphipod regression. The regression equation for the lower cl is MO 2 =5.649MDO -0.31 . The regression equation for the upper cl is MO 2 =8.901MDO -0.31 .  Figure 4: Pelagic hyperiid amphipod mean scaled metabolic rate (MO 2 ) in µmol 02 g -1 hr -1 by minimum depth of occurrence (MDO) for species living at 150m and below versus species living above 100m. Specimens with a MDO of 150m or deeper had a significantly lower MO 2 than species with a shallow MDO: ANOVA:F(1,31)=7.46,p<.0087. The mean rate for shallow specimens was 7.27±0.85 µmol 02 g -1 hr -1 , n=28. The mean MO 2 for deep specimens was 1.54±0.67 µmol 02 g -1 hr -1 , n=5. Mass (g) Figure 5: Mean MO 2 values of hyperiid amphipods in µmol O2 g -1 hr -1 comparing Physocephalata that are non transparent (closed black circles) or transparent (open black circles) and Physosomata that are non transparent (closed gray diamonds) or transparent (open gray diamond, one species). All MO 2 s are corrected to 10°C, and reported on a log scale. Specimens are from both this study and literature (Table 3). Transparent amphipods had a significantly lower MO 2 than non-transparent species in the clade Physocephalata (ANOVA: F(1,26)=13.16; p<0.0012). Non-transparent Physocephalata had a sample size of 20, and the mean MO 2 was 12.00±1.38 µmol O 2 g -1 hr -1 . The transparent amphipods had a sample size of 8 and a mean MO 2 of 3.74±0.88 µmol O 2 g -1 hr -1 . Physosomata had a significantly lower metabolic rate (mean 1.38±0.48 µmol O 2 g -1 hr -1 , n=6) than all Physocephalata (mean 7.29±0.83, n=27) (ANOVA: f(1,31)=10.14, p<0.0033).  Table 3. The linear regression is not significant Mo2=5.63-0.048x, R 2 =.09. MO 2 is not significantly different between specimens from cold water regions (less than 1C) and specimens from regions with temperature of 5C or higher (t-test: t(4)=-1.84;P<0.1617). Therefore, there is no metabolic cold adaptation in the genus Themisto. Table 3 has environmental temperatures and location for each species.