(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .
Climate warming erodes tropical reef habitat through frequency and intensity of episodic hypoxia [1]
['Noelle M. Lucey', 'Smithsonian Tropical Research Institute', 'Balboa Ancon', 'Curtis A. Deutsch', 'School Of Oceanography', 'Department Of Biology', 'University Of Washington', 'Seattle', 'Washington', 'United States Of America']
Date: 2023-04
Climate warming threatens marine life by increasing metabolic oxygen demand while decreasing oxygen availability. Tropical species living in warm, low oxygen environments may be most at risk, but their tolerances and exposures to these stressors remain poorly documented. We evaluated habitat restrictions for two brittle star species from Caribbean coral reefs by integrating field observations, laboratory experiments and an ecophysiological model. The absence of one species from the warmest reefs results from vital activity restrictions during episodic low oxygen extremes, even though average conditions are well within physiological tolerance limits. Over the past decade, warmer temperatures have been significantly correlated with a greater frequency and intensity of hypoxic events. Continued warming will progressively exclude hypoxia-tolerant species, even if average oxygen remains constant. A warming-driven increase in frequency or intensity of low oxygen extremes could similarly accelerate habitat loss across other marine ecosystems.
Funding: NL was supported by a fellowship awarded by Smithsonian Tropical Research Institute (STRI) and funding from M&J Bytnar. NL and RC were supported by a grant from the National Science Foundation (BIO-OCE 2048955) during manuscript preparation. A Smithsonian Institution Competitive Grant supported RC, and the Smithsonian Women’s Committee award #53, STRI and the Smithsonian Marine Global Earth Observatory supported MDJ. The Natural Sciences and Engineering Research Council of Canada Discovery Program grant RGPIN-2015-06500 and RGPIN-2020-05627 and Canada Foundation for Innovation grant was awarded to PC, and supported PC, MHC and FV, as well as some of the research costs. The Quebec Center for Biodiversity Science Excellence Award supported MHC. FV was supported by the mobility grant from the University of Quebec in Rimouski foundation (FUQAR). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Copyright: © 2023 Lucey et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Here we investigate how climate is restricting aerobic habitat in a major tropical reef. We do this by calibrating Φ with laboratory experiments on two reef–dwelling brittle star species and episodic temperature and oxygen extremes over a decadal time scale. The niche partitioning of the two species is only explained when the physiological and ecological components of the species’ energetic demands are accounted for, and when temperature and oxygen effects are considered in combination. These combined effects explain current patterns of habitat exclusion and predict a rapidly increasing rate of tropical biodiversity erosion [ 16 ].
Oceanic conditions that are extreme, even if brief and/or infrequent, may also play an important role in mediating climate’s impact on species fitness and habitability. Acute low oxygen events and ocean heatwaves both induce short-lived yet dramatic ecosystem changes [ 12 – 14 ]. But little is known about their co-occurrence [ 15 ] or their long-term consequences for species geography. This is because there is little sustained high-frequency data on the covariation of oxygen with temperature, or knowledge of the temperature-dependent hypoxia tolerances of resident species. Filling these data gaps is essential to establish mechanistic links between the extreme conditions of ocean ‘weather’ events and their long-term biological impacts.
As temperatures increase and oxygen levels decline, organisms’ ability to fulfill their oxygen demand decreases [ 8 ], reducing energy available for critical ecological activity and long-term survival. The limits to metabolically viable habitat can be quantified by the Metabolic Index [ 9 ], a measure of the ratio of oxygen supply to demand that reflects both species physiological traits and environmental conditions. The Metabolic Index (Φ) has predicted species geographic range boundaries, and the loss of habitat from climate-induced changes in temperature and oxygen on seasonal, decadal, and geological timescales [ 8 – 11 ]. But the impact of short-term fluctuations in Φ on long-term environmental habitability is not known.
Tropical ecosystems are especially vulnerable to climate warming due to the prevalence of species living near their thermal optima, with low tolerance for temperatures outside the narrow range of tropical variability [ 1 , 2 ]. The relationships between thermal tolerance, temperature variability, and the response to warming trends are well documented for terrestrial taxa, but their salience for marine species is unclear for two related reasons. First, rising ocean temperatures are linked to declining global oxygen content [ 3 , 4 ], and while these stressors can act synergistically on physiological functions [ 5 , 6 ], tropical species account for only 12.5% of those of which quantitative traits governing temperature-dependent oxygen tolerance have been measured [ 7 ]. Second, while temperature variability is low in both tropical marine and terrestrial environments, the relationship between oceanic oxygen and temperature variability from diel to decadal time scales are under sampled. A better understanding of both the tolerance for and exposure to increasing temperatures and declining oxygen in tropical oceans is therefore key to evaluating the vulnerability of the world’s most productive and biodiverse ecosystems, including coral reefs, mangroves, and seagrass meadows.
(a, b) Exposure to hypoxia for both brittle star species, computed as a fraction of time with Φ < FAS in the current climate (dots) and under three different climate warming scenarios (curves). Observed relationship between temperature and pO 2 extrapolated to warmer conditions ( Fig 4 ) yields a large increase in hypoxia exposure (red line), that disappears when increased pO 2 variance is neglected (green and blue lines). (c) Hypoxia exposure in the inner bay for a projected warming of 3°C across the range of active hypoxia tolerance and temperature sensitivity traits (E o and A o ) for known marine species globally (black dots) [ 7 ], and newly measured O. cinereum (orange circle) and O. wendtii (blue circle). Exposure on the reef would currently exclude most known species at a 5% exclusion threshold (black line). If future pO 2 variance is constant, exclusion of additional trait space from exposure is minimal (dashed red curve), whereas the observed trend toward increased pO 2 variance erodes habitat for a broad swath of currently viable ecotypes (solid red curve), including O. cinereum.
Hypoxia exposure on the reef can be evaluated across all combinations of active hypoxia tolerances and temperature sensitivities (i.e., species traits) based on the environmental conditions in the inner bay. Most known species trait combinations fall below the exclusion threshold exhibited by O. wendtii, indicating that current conditions on this Caribbean reef are not metabolically habitable for most species sampled to date ( Fig 5C ; black points in yellow area). Yet there is a broad swath of viable trait space in this environment laying between that of O. cinereum and O. wendtii. Other tropical species may occupy this space, but their traits have not yet been measured. Even if this is the case, the frequency and intensity of low pO 2 would exclude all species within this range of physiological limits under a 3°C warming scenario ( Fig 5C ; shifting exclusion thresholds from the black to red line).
This observed temperature-pO 2 variance relationship is then used to project hypoxia exposure under scenarios of climate warming ( Fig 4B , S4 Fig ). If mean temperatures increase by 1°C, the frequency and severity of pO 2 conditions also increase ( Fig 4C ). As warming approaches 3°C, as projected for this region by Earth System Models by the end of this century, the frequency and severity of low pO 2 substantially increase [ 23 ] ( Fig 4D ). Such increases in hypoxia exposure will limit the activity of both brittle star species, as conditions will regularly fall below their respective exclusion thresholds (Φ < Φ crit , Fig 4C and 4D , the increase of conditions occurring under both species exclusion thresholds). This future warming will increase the hypoxia exposure of O. cinereum, surpassing the exposure level that already excludes O. wendtii.
Orange coloration represents the environmental space of the inner bay, with panel (a) showing all measurements made from each temperature and oxygen combination during the 10-y sampling period. This is statistically modelled in panels b-d ( Eq 2 , S4 Fig ). Light yellow colors indicate low occurrences or rare conditions and red colors indicate common, frequent (i.e., average) conditions. The ecophysiological exclusion thresholds (Φ = Φ crit ) are drawn over the environmental space for both O. cinereum (cyan lines) and O. wendtii (blue lines). Exposure to conditions that limit activity due to hypoxia are represented by the fraction of time that conditions fall below the species-specific thresholds (Φ < Φ crit ). (c, d) Projected future bivariate frequency distributions assuming temperatures increase by 1 or 3°C, respectively. The frequency of stressful conditions (i.e., hypoxia exposure, Φ < Φ crit ) below the cyan and blue lines becomes more frequent as temperature rises.
We analyzed the relationship between temperature and pO 2 over the past decade to find out if hypoxia exposure increased with temperature. We find no significant relationship between temperature and pO 2 . However, temperature is both positively and significantly correlated to pO 2 variance, such that the frequency and magnitude of both high and low pO 2 extremes increase with temperature, yielding a bivariate frequency distribution with a clear wedge-shaped pattern ( Fig 4A ). We then determined each the species’ hypoxia exposure by mapping exclusion thresholds over the environmental ‘wedge’ using their respective Φ crit values. The frequency of environmental conditions occurring under the species’ exclusion thresholds determine how frequently they are exposed to metabolically limiting hypoxia (i.e., hypoxia exposure). The current hypoxia exposure of O. wendtii is ~8 times greater than O. cinereum, as shown by the higher frequency of conditions occurring under its exclusion threshold.
We determined how much time each species was exposed to environmental conditions that did not meet either their resting or active oxygen demands (i.e., conditions that limit activity). Hypoxia exposure times are identified when the Φ of each species falls below a threshold value of Φ crit , where Φ crit = 1 in a resting state, and Φ crit = FAS in an active state. Hypoxia exposure is then calculated as the total amount of time Φ < Φ crit within a measurement period. Hypoxia exposure was minimal for both species in the outer bay, regardless of energy state or time frame. At night, in the inner bay, hypoxia exposure was minimal for O. cinereum (0.03% during hourly 37-d diel timescale). However, hypoxia exposure of O. wendtii occurred 17% of the time during the same period ( Fig 3C and 3E ). Moreover, daytime hypoxia exposure during the last decade occurred ~0.02% of the time for O. cinereum, compared to 10% for O. wendtii. This suggests that infrequent episodic events of low Φ, reflecting fluctuating temperature-dependent oxygen, are not meeting the active metabolic requirements of O. wendtii, and that they account for the long-term exclusion of this species from the inner bay despite mean conditions being adequate to maintain maximum energetic demand.
To quantify how often the environmental conditions in the inner bay met both species’ metabolic demands, we calculated Φ for O. cinereum and O. wendtii from diel and annual timescale data ( Fig 3 ). During the 24 h diel period, Φ decreased nightly and increased daily for both species at both sites, primarily from lower pO 2 , however, the magnitude of this diel cycle was greater in the inner bay ( Fig 3C and 3E ). In both sites, the average nightly Φ ( Fig 3C and 3E , solid lines) was sufficient to meet both species oxygen demands, whether at rest or under peak activity (dashed black lines). However, the average daily cycle obscures the importance of episodic extremes of low Φ: extreme low pO 2 occurred approximately weekly in the inner bay and limited the aerobic scope of O. wendtii (i.e., Fig 3E —red shaded portion falling under dashed line; Φ < FAS), but not of O. cinereum (i.e., Fig 3C —all Φ values are above the dashed line that depicts O. cinereum FAS). Over the last ten years in the inner bay, daytime conditions also limited the aerobic scope of O. wendtii, providing further support that extreme Φ conditions render the inner bay unsuitable for O. wendtii ( Fig 3F ).
The Metabolic Index is computed from spatial and temporal observations of temperature and pO 2 (see Fig 1 ) and the temperature-dependent hypoxia traits for each species: O. cinereum (c, d) , O. wendtii ( e, f ). The solid lines show average values and shading denotes the standard deviation around the mean. The Factorial Aerobic Scope (FAS) of each species (dashed lines) denotes the value of Φ below which maximum metabolic rate at the observed temperature would be limited by local oxygen availability. Base map from:
https://stridata-si.opendata.arcgis.com/datasets/SI::bathymetry-of-the-republic-of-panama .
For both species, the values of Φ vary strongly throughout the bay ( Fig 3A and 3B ), with lower values in the inner bay due primarily to the lower pO 2 . The spatial gradient of Φ is amplified by temperature for O. cinereum, but weakened for O. wendtii, according to the temperature sensitivity of their P crit . For both species, daytime Φ generally exceeded the value needed to support maximum metabolic activity across the bay.
To evaluate the consequences of species’ physiological traits on species distributions, we used the Metabolic Index (Φ), an ecophysiological framework based on the ratio of oxygen supply to metabolic demand at any environmental temperature and oxygen level. For environmental oxygen supply to sustain an organism’s metabolic demand at rest, the Φ of an individual must be at least one (Φ ≥ 1) and, to enable maximum sustained activity, it must equal or exceed the factorial aerobic scope (Φ ≥ FAS). The parameters of Φ correspond to two species traits—hypoxia tolerance and its temperature sensitivity ( Eq 1 )—and are calibrated from experimental P crit data and its thermal variation ( Fig 2 ).
We determined the oxygen demand of each species under low oxygen and normoxic conditions, both at rest and after physical activity. The ratio of the maximum to minimum metabolic oxygen demand, termed Factorial Aerobic Scope (FAS) [ 22 ], was 2.5 times greater for O. wendtii than for O. cinereum under normoxic conditions: 2.44 ± 1.93 compared to 1.05 ± 0.26, as shown by a significant species effect ( Fig 2C , p = 0.002, S2 Fig , Table C in S1 Text ). The FAS of the two species did not differ under low oxygen, but the FAS of O. wendtii did decrease considerably under low oxygen compared to normoxia (p = 0.027). This decrease in the aerobic scope with strenuous activity under stressful conditions indicates that O. wendtii needs far more environmental oxygen to fully perform ecologically important activities such as reproduction, foraging and/or evading predation compared to O. cinereum. Additional assays, including behavioral escape trials and low oxygen survival rates, provide further support that O. cinereum is more hypoxia tolerant compared to O. wendtii ( Fig 2E and 2F , S3 Fig ).
The physiological tolerance traits measured in both species, coupled with the oxythermal gradient, eliminate some simple hypotheses for the observed distribution. First, while the inner bay is ~1.0°C warmer than the outer bay, the CT max of O. wendtii is ~1.5°C higher than that of O. cinereum, indicating thermal tolerance alone cannot explain the different distributions. Second, resting critical oxygen tolerances (P crit ) are not substantially different between the species, at least at the ambient temperature of 29°C. At the warmer temperatures of the inner bay O. wendtii has a higher tolerance for low oxygen relative to O. cinereum. Thus, low oxygen tolerance under resting conditions also fails to explain the exclusion of O. wendtii in the inner bay, even after accounting for temperature dependence.
As the inner bay is consistently warmer, the upper thermal limit of O. wendtii might explain its exclusion from the inner bay. To test this, we measured both species’ Critical Thermal Maximum (CT max ), a widely used dynamic method to assess organisms’ upper thermal limit ( Fig 2B ). Surprisingly, under normoxia, O. wendtii had a significantly higher upper thermal limit of 40.1 ± 1.1°C compared to O. cinereum at 38.5 ± 1.6°C (p = 0. 027). Low oxygen, or hypoxia, significantly reduced the upper thermal limit of O. wendtii by 1.0°C (to 39.1 ± 1.8°C) and increased that of O. cinereum by 0.5°C (to 39.1 ± 1.2°C) ( Fig 2B ; p = 0.024, Table C in S1 Text ). The two species’ mean CT max values were hence equivalent under low oxygen conditions.
Species sensitivity to low oxygen was measured as the critical oxygen pressure (P crit ) that can sustain resting aerobic metabolism [ 20 , 21 ]. At ambient temperature (29°C), both species had a comparable P crit : 5.04 ± 1.72 kPa (mean ± SD); Fig 2A . At the highest temperatures recorded in the bay (32°C) the P crit of O. cinereum increased marginally (5.84 ± 2.14 kPa) but decreased in O. wendtii (2.77 ± 1.50 kPa). This lowered P crit indicates O. wendtii is more tolerant to low oxygen–and therefore cannot account for its exclusion from the less oxygenated inner bay.
(a) Low pO 2 limits (P crit ) under ambient (29°C) and high temperatures (32°C). (b) Non-lethal upper thermal limits (CT max ) and (c) Factorial Aerobic Scope (FAS) under normoxia or low oxygen/hypoxia, 19.5 or 6.5 kPa, respectively. (d) Time to reach physical exhaustion in either normoxic conditions (light colors) or hypoxic conditions (dark colors) for each species. (e) Percentage of individuals that ‘escaped’ from experimental hypoxic conditions (< 6.5 kPa) by climbing out of a low oxygen layer. (f) Percentage of living individuals after experiencing low pO 2 (1.0 kPa) while resting, with all O. wendtti individuals dying after a 3.1 h duration. This time period is equivalent to one extreme low pO 2 event on the reef. Significantly different mean values (p < 0.05) in different treatments for the same species are indicated by capital letters, while significantly different mean values in the same treatment among different species are indicated by lowercase letters. Pairwise comparisons were conducted using the Estimated Marginal Means test with Least Significant Difference test correction. Error bars indicate ± SE.
Hourly oxygen and temperature measured with sensors deployed directly within the reef matrix reveal the high-frequency variability associated with these spatial patterns ( Fig 1 ). While diurnal temperature ranges are relatively small (~1°) and similar across the bay, the daily oxygen fluctuations have a large amplitude, reaching a diurnal range of ~3x the mean value in the inner bay. The dynamic range of oxygen reflects the extremely high net primary productivity and rapid respiration at night, especially in the inner bay. The high productivity of the inner bay, coupled with lower abundance of the dominant brittle star predator [ 19 ] argues against food web dynamics being responsible for the exclusion of O. wendtii (Table B in S1 Text ).
A decade of weekly sampling at sites in the inner and outer bay confirm that these regional differences are persistent over time ( Fig 1 ). The outer bay sites are consistently ~1°C cooler than the inner bay throughout the year, while oxygen is generally higher except during June-July when differences disappear ( S1 Fig , Table A in S1 Text ). Temperature and oxygen are spatially and temporally negatively correlated, with the inner bay being slightly warmer, having lower mean oxygen and greater oxygen variability, compared to the outer bay ( Fig 1 , top-right).
The oxygen and temperature throughout the bay exhibit regional patterns resembling that of the brittle star abundance ratio, indicating a potential causal link. The lowest oxygen and highest temperatures co-occur near the mainland, at sites dominated by O. cinereum and where O. wendtii is excluded ( Fig 1 ). Other environmental parameters (pH, salinity, chlorophyll) also varied throughout the bay, but variation was minimal relative to the changes in oxygen and temperature [ 17 , 18 ].
(Left column) Pie charts show proportion of individuals from two species of brittle star, Ophioderma cinereum (orange) and Ophiomastix wendtii (blue), found at each survey site. Animals were collected for laboratory experiments from 3 m at the inner bay (red asterisk) and outer bay (blue triangle) reef sites. (Center column) Daytime temperature (top) and pO 2 (bottom) surveyed on September 25 th , 2017, at 3 m across 83 sites. (Right column) Time series of temperature and pO 2 over the mean seasonal cycle (upper panels) from a decade of weekly measurements 2010–2020 and the mean diurnal cycle (lower panels) from 37 d of hourly data collected between Oct.–Nov. 2019 (light blue shaded box). Solid colored lines in time series show average values and shading denotes the standard deviation. Base map from:
https://stridata-si.opendata.arcgis.com/datasets/SI::bathymetry-of-the-republic-of-panama .
We mapped the distribution of temperature, oxygen, and abundance of brittle stars in a large coral reef system on the Caribbean coast of Panama ( Fig 1 ). Among the two species, not a single individual of O. wendtii was found close to the mainland, whereas O. cinereum dominated these inner bay sites. The relative abundance of O. wendtii increased on reefs closer to the outer bay, until there was an equal or higher abundance of O. wendtii.
Finally, an increase in oxygen variability as temperatures rise will dramatically expedite reductions of available aerobic habitat. This is especially important in shallow tropical marine environments where Earth System Models and long-term observations indicate a relatively stable, or even increasing, O 2 as climate warms [ 24 , 25 ]. Models that fail to reproduce sub-seasonal extremes and their response to climate change will greatly underestimate how quickly local extirpations are occurring in shallow environments, where oxygen variability rises sharply with warming. An expanded effort to measure temperature-dependent hypoxia tolerances of more tropical ectotherms, and to determine high-frequency covariation of environmental temperature and oxygen on other shallow ecosystems is required to evaluate how pervasive this mechanism is in accelerating the climate-driven loss of habitat and biodiversity.
Our results have important implications for the role of climate in the maintenance and erosion of tropical marine biodiversity. First, the link between fine-scale spatial patterns of biogeography and aerobic habitat restrictions here supports similar conclusions based on large scale range limits across latitude and depth [ 7 ]. Second, the metabolic constraints on habitat occupancy become evident only when the role of temperature on oxygen variance is accounted for, and when both organismal physiological and ecological energetic demands are included. Neither temperature alone, nor the combined thermal and hypoxia tolerances in a resting state provide sufficient information to explain niche partitioning in the tropical marine organisms studied here. Third, survival in an aerobically demanding environment requires distinct adaptive strategies and energetic trade-offs among traits governing tolerance to temperature-dependent hypoxia. This is seen in the widely divergent active hypoxia tolerances and temperature sensitivities of both brittle star species found at the edges of habitable trait space for this reef ( Fig 5 ). Given the high biological diversity currently making up tropical coral reefs, a greater sampling of these traits among reef biota is needed.
Materials and methods
Species distributions We used two ophuiroid species associated with tropical coral reefs: Ophioderm cinereum and Ophiomastix wendtii (Fig 1; MiAmbiante Permit SE/AP-20-18). To establish their relative abundance, we surveyed 10 coral reef sites throughout the bay between 3–5 m depth by collecting at least 30 individuals of either species. From this data, we determined the ratio of O. cinereum to O. wendtii at each site. We collected animals via SCUBA and snorkel by inspecting coral matrix for brittle star arms and then coaxing the animals out of the reef using a string and bait. Collection methodology was consistent across all sites to avoid potential bias.
Mapping a coral reef Spatial analysis. We determined the spatial patterns of partial pressure of oxygen (pO 2 ) and temperature throughout the study area (i.e., Fig 1), by placing a YSI multiparameter sonde (YSI EXO2 and EXO optical dissolved oxygen (DO) Smart Sensor–accuracy: 0.1 mg L-1, Conductivity and Temperature Sensors–accuracy: 0.01°C, Yellow Springs, USA) at 3 m in the water column at each of the 83 sites on 26 September 2017. Two boats were used to make all measurements between 08:00 and 17:30. 19-y hourly timeseries. We determined if temperature differed between the inner and outer bay sites by analyzing a 19 y, hourly continuous temperature record from loggers deployed directly on a concrete block nestled within the reef matrix at a depth of 3 m (Hobo Onset loggers; Dec. 2000 –May 2019; datasets provided by the STRI Physical Monitoring Program; S1 Fig). 37-day hourly timeseries. To monitor the joint variation in temperature and pO 2 through time at both the inner and outer bay sites, we deployed temperature, DO, and conductivity loggers at a depth of 3 m directly within the coral reef matrix (U26-001 HOBO DO Logger–accuracy: 0.2 mg L-1; U24-002 Conductivity Logger; Onset, MA). Loggers were 10 cm from living coral and reef structure (i.e., logger was at the same height at the coral, with coral surrounding the logger). They were deployed between October and November 2019 and took measurements hourly. pO 2 measurements were calculated from DO, conductivity and temperature. Measurement precision was based on comparisons to measurements made in the field before and after calibration checks with a YSI Pro2013 multi-parameter handheld (accuracy: DO 0.2 mg L-1, temperature 0.35 °C, conductivity 0.5%). 10-y weekly timeseries. Weekly data was collected during the day from Jan 2010 to March 2020 with a handheld YSI multiparameter sonde which was lowered to a depth of 10 meters in the water column above the benthos (YSI EXO2 and EXO optical DO–accuracy 0.1 mg L-1; Conductivity and Temperature Sensors–accuracy 0.01°C, Yellow Springs, United States; STRI Physical Monitoring Program). Sensors were calibrated monthly following the manufacturers’ instructions. The weekly monitoring sites are approximately 10 m from the benthos and provide comprehensive overview of the oxygen conditions occurring in the area, as well as represent conservative estimates for the oxygen fluctuations occurring on the reef (datasets provided by MarineGEO and the STRI Physical Monitoring Program). GPS coordinates for physical data collection sites and species abundance collection sites are provided in the raw datasets (S1 Text). A distance of ~ 10 km separates the outer bay reef site (37-d hourly record) from the outer bay weekly monitoring site (10-y weekly record, ‘Colon’), however discreet measurement calibrations between these outer bay sites validated outer bay site pairing.
Experimental details For all experiments, we hand collected O. cinereum from both the inner and outer bay sites, and O. wendtii from the outer bay sites (Fig 1). Within 2–4 h of collection, they were transported back to the Bocas Research Station (BRS) in temperature-stable coolers and were immediately transferred to 40 L tanks and maintained with flowing seawater at ambient temperature until the start of the experiments (within ~10 d of collection). Prior to the experiments, individuals were fed small pieces of fish in the laboratory; they were starved for 48 h before the SMR trials and not fed anything thereafter. To maintain/establish oxygen treatments before for the SMR, MMR, CT max trials, individuals were moved to 4 L aquarium tanks that were either bubbled with pure nitrogen to reduce oxygen levels (i.e., hypoxic treatment; pO 2 , 6.5 kPa; DO, 2 mg L-1), or bubbled with air to maintain normoxic conditions, pO 2 , 19.5 kPa; DO, 6 mg L-1. pH was not experimentally modified and did not decrease during these experiments. Treatment acclimation lasted for 4 h prior to trials. During this time aquaria were kept in water baths with aquarium heaters to ensure temperature remained constant at 29°C and salinity at 34 ppt. Filtered (0.45 μm) seawater was used for all trials.
Physiological tolerances P crit experiments were conducted on 42 O. cinereum and 23 O. wendtii individuals. Individuals were gently rinsed with well-oxygenated filtered seawater and put into closed respirometry chambers. Equal numbers of trials were performed at 29 and 32°C. For trials at 32°C, the seawater was ramped from ambient 29 to 32°C in 1 h in an isolated tank before being placed in the respiratory chamber. The initial pO 2 value was 19.5 ± 0.29 kPa for all trials. pO 2 in the chamber was recorded with fiber-optic sensor technology (Fibox 4 and Pst-3 sensors, PreSens, Regensburg, Germany), every minute until the concentration fell to 1.0 kPa (0.3 mg L-1). Temperature in the water bath was recorded simultaneously with the Fibox 4, as well as a thermocouple thermometer (HH802U, OMEGA, Canada—accuracy 0.1°C). Upon reaching the oxygen depletion target in the chamber, each individual was promptly moved into ambient, oxygenated water. No individuals died immediately after P crit trials. Wet weight was measured using a precision balance (PL6001E, Mettler Toledo–accuracy 0.1 g) and volume was measured via displacement using a graduated cylinder and beaker. After taking morphometric measurements individuals were then moved to well-oxygenated recovery tanks with flowing, ambient temperature seawater for 24 h. Survival was documented after the 24 h recovery period, i.e., Fig 2F. We determined the P crit for each individual using broken stick regression which fits two regression lines through the data using the respR package [26]. The intersection of the two regression lines is the critical point and defines P crit . Acute upper thermal limits (CT max ) were measured in O. cinereum (32 indiv.) and O. wendtii (16 indiv.) under hypoxic (6.5 kPa, 2 mg L-1) and normoxic (19.5 kPa, 6 mg L-1) conditions, using the onset of spasm as the endpoint (defined below in more detail). Trials were conducted following exposure to two oxygen treatments representative of a single nightly drop in oxygen (4 h hypoxia; 6.5 kPa, temp. 29°C, salinity 34, 2.0 mg L-1) or normoxic conditions on the reef (4 h normoxia; >19.5 kPa, temp. 29°C, salinity 34, 6.0 mg L-1). They were then transferred in 300 mL glass bowls containing water kept at their corresponding pO 2 treatment and left to rest for 10 min at 29°C. Water temperature was then increased at a rate of 1°C min-1 using a thermostatic water bath (Cole-Parmer ® StableTemp® Digital General-Purpose Baths, Cole-Parmer Canada, Canada) and individuals were observed continuously until they reached the endpoint (see S1 Text for video A link to CT max trial). The corresponding temperature was identified as their CT max temperature and measured using a thermocouple thermometer (HH802U, OMEGA) [27, 28]. Given the scarce literature available for CT max measurements on echinoderms, initial trials were performed to identify a repeatable behavioral response, assayable for both species, that would correspond to the equivalent of onset of spasm for brittle stars. From these preliminary trials, we define CT max as the non-lethal temperature at which animals rapidly curl their arms together and become stiff and completely unresponsive to touch. Immediately after reaching the endpoint, individuals were removed, and their wet weights were measured before transferring to an ambient temperature recovery tank. Animal mass was used as a covariate in the statistical models to rule out any statistical differences in size. Survival was assessed after 24 h and only surviving individuals were included in CT max dataset. Note that CT max and P crit measurements are strictly designed to define and compare the relative physiological performances, behavioral responses, and tolerance levels for species’ comparison (S1 Text).
Oxygen demand Prior to the upper thermal limit measurements (CT max ), oxygen consumption rates (MO 2 ) at rest (i.e., standard metabolic rate; SMR) and after/at exhaustion (i.e., maximum metabolic rate; MMR) were measured in O. cinereum (32 indiv.) and O. wendtii (16 indiv.), under either hypoxic or normoxic conditions. We assayed metabolic rates using closed cell respirometry. Oxygen partial pressure (in μmols) inside the respirometry chambers were measured every 15 min using fiber-optic sensor technology (Fibox 4 and Pst-3 sensors, PreSens, Regensburg, Germany), for two hours or until the oxygen levels in the chambers were reduced to 80% of the initial saturation level. Sensors were individually calibrated just before the experiment, using two-points (0–100% air saturation) with the manufacturer’s method at ambient temperature (29°C). Each chamber was sealed and included a mesh screen that divided a magnetic stir bar from the main chamber area. Chambers were placed over a multi-channel magnetic stirrer (Mix 15 Eco Stir) to ensure gentle mixing throughout the trial and prevent oxyclines from forming inside the chambers. Standard metabolic rates (SMR) were measured under resting conditions. We acclimated an animal to the selected treatment for 3 h before placing them into their respirometry chamber for an additional acclimation period of 1 h while continuing their hypoxic or normoxic treatment. After the SMR trial, animals were then returned to separate maintenance tanks for 2 d before undergoing MMR measurements. Individual MMR was measured following a post-exercise methodology that was adapted for the ophiuroids [29, 30]. Trials started with 4 h treatment acclimations, individuals were moved to a new tank with filtered seawater (47 mm GF/F glass microfiber filter, Whatman®, Buckinghamshire, United Kingdom) matching their oxygen treatment. Each individual’s maximum metabolic rate (MMR) was then measured by exhausting the individual by manually flipping the animal onto its back and repeating when it righted, until they remained upside-down without any attempt to right themselves for more than 1 min (i.e., exhaustion, Video B in S1 Text shows the exhaustion method). The time it took each individual to reach exhaustion was recorded. At this point of exhaustion, we moved individuals into respirometry chambers and immediately began measuring oxygen consumption (MO 2 ) as in SMR trials above, obtaining the MMR. To minimize microbial respiration, all respirometry chambers were transferred and closed in a clean tank filled with filtered seawater (Whatman ®). We monitored microbial respiration by including two chambers filled only with water from the clean tank, for each trial, or blanks, where six individuals were undergoing measurements. Oxygen consumption rates (in μmol h-1 mg-1) were individually calculated from blank corrected animals’ fluxes (μmol L-1 h-1) standardized by their wet mass (mg). Microbial respiration was effectively contained to marginal fluxes and the animals’ oxygen consumption was constant (i.e., linear decrease in oxygen concentration during trials). We corrected the chambers’ volumes by subtracting the stirrer’s and the individuals’ volumes from their total capacity. The aerobic scope–both absolute (AAS) and factorial (FAS) of each individual was calculated in absolute terms: MMR–SMR and factorial terms: SMR: MMR [29]. FAS values < 1 were replaced by 1 as aerobic scope must be a positive number, after determining no statistical changes between unmodified vs. modified FAS. After the MMR trial, individuals were rested in a maintenance tank overnight, and CT max tested the following day (see above).
Escape behavior To determine if O. cinereum and O. wendtii exhibited hypoxia escape behaviors, we simulated hypoxia shoaling in an oxygen-gradient chamber. The chambers consisted of a clear plexiglass 50 cm tall column (S3 Fig). Freshwater was mixed with seawater to slightly reduce the salinity in the top normoxic water layer, while deoxygenated seawater was pumped in from below to create the bottom hypoxic layer. The salinity differed by 2–3 ppt between water layers. We tested a total of 55 individuals, 35 O. cinereum and 20 O. wendtii. Individuals were placed in an oxygen-gradient chamber and allowed to rest for 1 h before the trial with an escape ladder inserted into the chamber so that it was reachable to the individual. This allowed individuals to explore the chamber, discover the ladder and acclimate to the new surroundings and salinity. After the resting period, deoxygenated seawater was slowly added to the bottom of the chamber to avoid mixing with the top layer, until the bottom half of the chamber was filled with hypoxic water (1.13 ± 0.41 mg L-1), while the top half of chamber remained normoxic (6.34 ± 0.23 mg L-1). This was equivalent to pO 2 of ~3.7 kPa in the hypoxic layer and 20.0 kPa in the normoxic later. Trials began as soon as hypoxic water was introduced and lasted 1 h or until the individual climbed to the surface. We recorded the success of each individual to ‘escape’ from hypoxic conditions by climbing up the ladder into the normoxic area. The time to escape (min), as well as any effort made to escape was also recorded. No individuals died during respiration or behavior trials.
Habitat exclusion by episodic hypoxia waves The metabolic index (Φ) is calculated using the following equation: (1) where A O is the ratio of rate coefficients for O 2 supply capacity relative to resting metabolic rate at a reference temperature (T ref = 15°C), Bε is the body mass scaling, pO 2 in ambient O 2 pressure, E O is the temperature dependence of the O 2 supply/demand ratio, k B is the Boltzmann’s constant, and T is temperature [9]. The physiological parameters of Φ are obtained from fitting a regression to the natural log of the measured P crit values of each species at the reference temperature. The slope and intercept are set to E O and A O , respectively. In this study, we calculated Φ for both species, through time, from temperature and pO 2 taken from the 37-d hourly timeseries (diel timescales) as well as the 10-y weekly timeseries (annual timescales), at both sites (physical parameters = pO 2 and temperature).
Climate warming increases hypoxia exposure To project the frequency of hypoxic extremes as the reef warms, we modeled the bivariate frequency distributions of local temperature and pO 2 . The probability of a given combination of environmental temperature (T) and pO 2 was modeled as a product of separate probability distributions functions (pdf’s) for temperature (n) and pO 2 (m): The pdf for temperature is modelled as normally distributed, with statistics (mean, standard deviations) that are assumed independent of pO 2 . In contrast, based on the long ‘tail’ of low pO 2 , its pdf is modeled as a lognormal distribution. We use a log-transformed version of pO 2 , where the normalization (0.3 atm) is chosen to be above the maximum observed pO 2 , so that varies from 0–1 and is approximately normally distributed. Based on the wedge-shaped bivariate histogram of T and pO 2 (Fig 4A), the statistics of pO 2 are allowed to change linearly with T. Under these conditions, we can write the joint probability of observing an environment with a given combination of T and pO 2 as: (2) The mean and variance of are assumed to be linear functions of T, so that and . The joint pdf (i.e., f) thus has 7 parameters: α 1 , μ T , σ T , b 1 , c 1 , b 2 , c 2 , which we determined by finding the best fit to the binned 2D histogram of T and pO 2 . We chose measurements in the inner bay in the 10-y weekly daytime record (Fig 4B), because this decadal timescale includes the relationships between variation of T and pO 2 most relevant to climate change. The fits are highly significant (p < 1 e-30, R 2 = 0.56, d.f. = 118), including for the slope of pO 2 variance versus temperature, c 2 = 0.078913 (s.e. = 0.019927, p = 1 e-4).
[END]
---
[1] Url:
https://journals.plos.org/climate/article?id=10.1371/journal.pclm.0000095
Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.
via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/
