the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Reviews and syntheses: Biological Indicators of Oxygen Stress in Water Breathing Animals
Abstract. Anthropogenic warming and nutrient over-enrichment of our oceans have resulted in significant, and often catastrophic, reductions in dissolved oxygen (deoxygenation). Stress on water-breathing animals from this deoxygenation has been shown to occur at all levels of biological organization: cellular; organ; individual; species; population; community; and ecosystem. Most climate forecasts predict increases in ocean deoxygenation, thus it is essential to develop reliable biological indicators of oxygen stress that can be used by regional and global oxygen monitoring efforts to detect and assess the impacts of deoxygenation on ocean life. This review focuses on indicators of low-oxygen stress that are manifest at different levels of biological organization and at a variety of spatial and temporal scales. We compare particular attributes of these indicators to the dissolved oxygen threshold of response, time-scales of response, sensitive life stages and taxa, and the ability to scale the response to oxygen stress across levels of organization. Where there is available evidence, we discuss the interactions of other biological and abiotic stressors on the biological indicators of oxygen stress. We address the utility, confounding effects, and implementation of the biological indicators of oxygen stress for both research and societal applications. Our hope is that further refinement and dissemination of these oxygen stress indicators will provide more direct support for environmental managers, fisheries and mariculture scientists, conservation professionals, and policy makers to confront the challenges of ocean deoxygenation. An improved understanding of the sensitivity of different ocean species, communities and ecosystems to low oxygen stress will empower efforts to design monitoring programs, assess ecosystem health, develop management guidelines, track conditions, and detect low-oxygen events.
- Preprint
(1453 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-616', Anonymous Referee #1, 09 May 2024
Roman & Levin et al. provide a thorough and authoritative review of indicators of seawater deoxygenation, which I found very useful and interesting. I think it represents a great deal of work and knowledge that readers of many disciplines will be grateful for. All of my comments are minor but the authors may find them useful, especially to improve clarity in several places.
It depends on other nominations but this synthesis manuscript is certainly of wide appeal to the broad geoscience community and discusses how the importance of seawater deoxygenation remains underappreciated in the scientific, political, and management communities, and the general public.My comments are generally preceded by the line number in the original pdf.
67. Delete 'have'. sounds like the evolution was recent
70. comma after 'temperatures'
93. Maybe a note to caution here that, in general, all indicators can give the wrong impression if they are not thoroughly studied or poorly decided e.g. if the same behaviour may be correlated with other impacts
137-142. It may be best here to supplement this important section (especially "there were also clear interactions among stressors in their biotic effects") with a meta-analysis that considered interactions with deoxygenation e.g. Reddin et al. 2020, which is also not as restrictive in deoxygenation threshold i.e. 'hypoxic event' as in Sampaio et al. 2021; Vaquer-Sunyer & Duarte 2011 though note that this paper does not use a conventional meta-analysis approach, which limits the robustness of its conclusions
Reddin, C. J., Nätscher, P. S., Kocsis, Á. T., Pörtner, H. O., & Kiessling, W. (2020). Marine clade sensitivities to climate change conform across timescales. Nature Climate Change, 10(3), 249-253.
Vaquer-Sunyer, R. & Duarte, C. M. Temperature effects on oxygen thresholds for hypoxia in marine benthic organisms. Glob. Change Biol. 17, 1788–1797 (2011).
Figure1. The benthos seems a bit empty (perhaps it has already suffered from hypoxia) of indicators. e.g. as mentioned later the microbial mat, lack of bioturbation, benthic organisms ‘snorkelling’ the better oxygenated water column. Some of the higher up indicators could show their benthic component. Not clear what the different shades of blue are – simply depth? Purely artistic?
165. Can give a little more information on what constitutes a responsive and less responsive species? Not quite a definition perhaps, but more information would be nice.
Section ‘sensory systems’. The section lacks a summary that the previous section had. A general theme here seems to be that the use of sensory system changes as a deoxygenation indicator is promising but needs much further research before any general application is possible?
221. “levels of growth hormone-insulin-like growth factor”. I am not familiar with this topic but this description is particularly difficult to understand
223. Is this sentence needed? Seems obvious. “Cortisol levels increase when oxygen levels are low enough to cause a physiological stress response”
245. “understood,” rather than “understand”
In general, it would improve clarity to use more commas, where appropriate, to show the main sentence from the components that add information but are grammatically unnecessary e.g. the “where appropriate” above. Alternatively, long sentences could be broken into two or more smaller sentences. There are many instances early on in the MS (it seems not so bad later on) but here is one example (L271-4) of both the above points:
e.g. “The laboratory data were then used to develop a model of the endocrine functioning of vitellogenesis of individual fish (Murphy et al., 2009) to examine how the indicators measured as blood and organ concentrations would vary over time and under exposures not replicated in the laboratory.”
To
“The laboratory data were then used to develop a model of the endocrine functioning of vitellogenesis of individual fish (Murphy et al., 2009). This allowed examination of how the indicators, measured as blood and organ concentrations, would vary over time and under exposures not replicated in the laboratory.”
345. “estimate” rather than “ascertain”
388. Also consider that spawning adults may have greater sensitivities than other stages (though whether this results in mortality or lower ultimate fertility may determine which section this is more appropriate for)
413. Three as written form, rather than ‘3’
446-448. This sentence is unclear: “The resting metabolic oxygen demand of the 447 metabolic indices (Deutsch et al., 2020; Penn et al., 2018) occurs at the onset of mortality 448 or anaerobic metabolism (Deutsch et al., 2015).” The demand occurs at the onset of mortality? This doesn’t make sense to me.
466. DIC should be familiar but doesn’t harm to define.
483-5. Commas are a bit messed up in this sentence.
498. “types of soft-bodied fishes, cnidarians (jellies) and ctenophores”. Maybe give an example of these fishes. Also perhaps be more specific with ‘cnidarians’, since that includes corals, which I don’t think would be a good fit here. ‘Jellies’ is rather informal – can a better/more taxonomic term be found?
504-508. Maybe note that care need taking to account for changing sedimentation or even erosion rates.
536. It would be good to mention that the use of a single indicator taxon may incur error as species niches are not unidimensional, while the use of multiple indicators at once (and multiple threads of evidence in general) should be more robust.
562-7. Please split this long sentence.
718. Perhaps add Simpson’s 1-D. I am not familiar with the ‘ESx’ acronym the authors have here. In my mind, rarefaction is a sampling standardization approach, not a diversity metric in itself, but any of these listed metrics could be based on rarified data. E.g. rarefaction curves are usually of species richness values. Also, species dominance and evenness are two sides of the same coin, rather than separate (usually Simpson’s D or 1-D)
753. Perhaps change “shifts to a dominant that is a hypoxia-tolerant species” to “shifts to dominance by a hypoxia-tolerant species”
759. Salinity may be another important factor in coastal waters
765. These references for ‘fossil forming biota’ all seem to be about foraminifera. Would be nice to have some macrofauna references e.g. Aberhan, M., & Baumiller, T. K. (2003). Selective extinction among Early Jurassic bivalves: a consequence of anoxia. Geology, 31(12), 1077-1080.
766. I would not say alpha diversity is scaled up to beta diversity as one only needs two sites to obtain a beta diversity estimate (i.e. change between two sites). Correct that gamma diversity is the next scale up from alpha diversity.
839. Can be more specific? “than that required to see changes in diversity”. But several diversity metrics use abundance data, so these changes would be visible in those metrics, wouldn’t they? Depends how diversity is defined e.g. if it was species richness, I would agree with the authors.
959. There is a stray colon and then semi-colons later in the sentence in a way that looks intended but is not the usual way these are used. Please write as a normal sentenceDiscussion.
The different aspects of scale that the authors are delineating here are not completely clear to me. Please try to distill a clear summary of each aspect. The first aspect of scaling here appears to be sampling, i.e. relating the scale aspects of the sample to the wider population that the samples are attempting to capture. Second aspect seems to be level in the biological and ecological hierarchy, which are linked mechanistically and thus scales can be crossed not by statistics but via this mechanistic understanding. The third aspect I am unclear about how it differs. It seems an integration of statistical and mechanistic modelling (aspects 1 and 2)?
1071. Please break this sentence down e.g. split it. It is unclear. “Modeling involves significant effort beyond the conceptual modeling that is thus done in 1072 scaling of indicators when quantitative links from the indicator to the system state are 1073 needed.” Especially “that is thus done”
1081. “time and space scales”. Temporal and spatial scales
Table 1. This table did not display well on the pdf and I am not sure the displayed version is the complete version e.g the edges look like they were cut off. The text in many cells is cut off and the font is difficult to read. Better spread table over two pages than miss information. Acronyms and abbreviations need spelling out in a footnote or somewhere. Column ‘Useful/relevant settings’: aren’t all these indicators to be used in the field? Makes this column dubious in its use. ‘Expertise required’: why has ‘basic’ got a question mark? One cell in this column also has a red earmark.
1117. What does “non-sublethal” mean?Citation: https://doi.org/10.5194/egusphere-2024-616-RC1 -
RC2: 'Comment on egusphere-2024-616', Jackson Chu, 24 Jul 2024
Roman & Levin et al. updates the current knowledge of biological effects of ocean deoxygenation and reviews indicators that could be useful if implemented into monitoring programs designed to detect and track ecological shifts as a direct consequence of low oxygen stress.
General comments
There is a great and timely review of the biological response to hypoxia and the large-scale impacts of deoxygenation. Literature examples span multiple ecosystems and are global in representation. The authors target a wide audience and bridges this with examples that span multiple biological levels of organization perspective (and thus interdisciplinary among the biological sciences). As a consequence, they brought to my attention new papers on a topic that I am somewhat familiar with which is an indication that the literature review component alone is immensely valuable. The review furthers the discussion by touching on where deoxygenation may have been overlooked by major research and global science initiatives. This will be a useful guidance document for those who wish to direct resources towards ‘oxygen’ as a key variable in their monitoring and management efforts, and hopefully cite this review as a result.
The majority of my comments focuses on a meat of the review – the summary of indicators of deoxygenation stress. At the onset, the authors introduce key terms like ‘indicator’, ‘monitoring programs’, and cite the Essential Ocean Variables (EOVs) as background context. Readers that already have some buy-in and are tasked with monitoring efforts would predictably ask “what do I measure and how do I measure it?”. These are the start-up questions to address when developing indicators for monitoring programs (See Yoccoz et al. 2001, Reynolds et al. 2016 for more). The term ‘indicator’ could be better defined and applied in the manuscript. In general, ideal indicators have specific criteria when being evalualted for monitoring programs – they should be: quantifiable (measurable with units), sensitive (small change in the system pressure incudes a big change in indicator), responsive (indicators responds in the same time frame as system pressure), specific (not influenced by covariates), and operationally feasible (i.e. costs associated with acquiring enough data points so that a signal can be detected is reasonable)…and be able to detect & track trajectories of change over time (or space) as a consequence of the system pressure (here – deoxygenation).
Several of the examples as presented are uncoupled from this main manuscript theme which sandwiches the beginning and end of the manuscript. Not all the reviewed biological response examples are presented as indicators in the current text but still exist as a recap of their original citation. Additionally, a science evaluation of the criteria for indicator assessment and associated ‘challenges’ of using them in monitoring programs could be useful – this is loosely touched upon in Table 1 (e.g. the specific criteria relates to the table’s confounding factors), but is inconsistent. Some structured text that considers this for all presented examples could be useful in a minor revision. The translation of the expansive theoretical knowledge into operational guidance is often the missing piece that results in the failure to launch of many monitoring programs.
Refs:
Yoccoz et al. Monitoring of biological diversity in space and time (2001) Trends in Ecology & Evolution, 16: 446-453.
Reynolds, J.H., Knutson, M.G., Newman, K.B., Silverman, E.D. and Thompson, W.L., (2016). A road map for designing and implementing a biological monitoring program. Environmental Monitoring and Assessment, 188, pp.1-25.
Specific Line comments -
Many of my specific line comments were written before my general comments section; they will mostly focus on whether the biological response examples could meaningfully be used in a monitoring program designed to track deoxygenation impacts.
L40 – is ‘indicators’ an appropriate term here? Biological response seems a better fit.
L82-L85 – Do the authors suggest that ‘only’ measuring biotic indicators here? Or in companion with existing oxygen monitoring? If the former and with the rest of the manuscript read – it would be challenging to conclude on the causal relationship of any signal in the biotic parameters back to deoxygenation without having the ambient oxygen having been measured in parallel as well.
L85-L89 – it would strengthen the messaging to the intended target audience listed in the previous sentence if a guiding definition of ‘indicator’ is presented.
L89-L92 – subjective and arguable given the evaluation criteria of what makes a good indicator for monitoring programs hasn’t been presented. Changes in animal behaviour, unless specific to the focal question/objective of the monitoring program could be considered poor indicators?
E.g. if the cost of deploying and recovering an oxygen sensor is less than the cost and risks of acquiring and maintaining a population of animals and measuring live animal behaviour in a controlled experimental setting…than the latter would not be a good indicator. How would one then apply any lab-derived data back into the natural system state influenced by deoxygenation stress?
L98-102 – Just a thought about the idea behind the EOVs and how the relate to this review. Core traits of the EOVs are that they are measureable and inclusive, but also ‘non-specific’ to just one system pressure/stessor. The challenge with developing deoxygenation indicators is to have them be ‘exclusive’ – not correlated to other pressures in that they are informative to one very specific stressor.
L107 – « Garcon et al. 2019 examined… » What did the authors conclude ?
L124 – The covariate and scaling issue. Given these are established challenges with interpreting the ecological consequences of deoxygenation/hypoxia – can the authors reflect on the presented list of their indicators and suggest which ones are best?
L143 – The focus of this review
Reading through – are the levels of biological organization intended to present the ‘level’ of organization at which (1) the indicator is measured, or (2) the level of organization at which ‘the biological response’ manifests? The manuscript isn’t quite consistent or clear on this.
L154 – Consider “Individual-level indicators”?
Individual-level indicators – these are described more as biological responses – there lacks a discussion on how these can be translated into monitoring programs?
E.g. in research fisheries surveys – individual specimens are brought up in catch. Length (cm), weight (kg), age (otoliths), are measurable, individual-level data. I can envision tissue & blood subsamples being taken if some of the presented indicators can be quantified post-hoc (e.g. protein & enzyme levels present at sampling), but the behavioral ones, at least as presented, would be extremely challenging to implement in a realized monitoring program.L163 – The indicator “Amount of HIF-alpha protein”
- Indicator sensitivity is species-dependent
L182 – Not clear why HIF may be more difficult to measure? Is it the cost and challenges of the live-animal cultivation requirements to get a data point?
L186-L213 – For sensory systems indicators and the suite of variables that require manipulative lab-based experiments to quantify – would these really be ‘monitoring indicators’?
L198 - Would monitoring programs, which infer natural/field systems…be tracking vision loss or sensory loss in situ? How would fisheries actually measure this within their standard equipment? Table 1 suggests ‘control’ specimens would be required to be brought into the lab, and manipulated via factoral experiments . Unless these biological responses can be calibrated back to a field-based indicator (can behavioural responses like these be calibrated to a fisheries metric like CPUE?) – these feel a bit uncoupled from being useful monitoring program indicators without additional thought.
L186 – Vision – perhaps mention this metric only applies and is limited to mobile animals with eyes.L302-L306 – Pcrit, oxyregulation/oxyconformation and the theory comes from quite a long history of fish physiology research – suggest citing at least one reference here. (e.g. Fry 1933).
L320-L329 – the theory underpinning Pcrit and oxyregulatory ability also relates to ontogeny; this nuance is often lost with the generalization towards ‘size’. The ability to oxyregulate is not present throughout developmental life-stages in many marine ecotherms – the ability to oxyregulate manifests in later life history stages (e.g. fish) which also correlates with larger body sizes.
Pcrit seems out of place here but also is inconsistent with what is / isn’t an indicator – which goes back to the comment on what is actually being measured and how it can be measured as part of a monitoring program. Compare this with the units of the other indicators presented which is quantified through a measurement of a biological parameter / response variable.
A Pcrit is just a single environment oxygen level but determined to be a biological relevant threshold using live-animal respirometry experiments. Time-series of the threshold itself is not what would be generated by a monitoring program as it’s not the parameter that is responding to the deoxygenation system pressure nor would time-series be generated of Pcrits. A realized deoxygenation monitoring program that implements Pcrits into its general framework would realistically be generating standard oceanographic oxygen time-series using CTDs with the interpretation/early warning signal being the where and when those oxygen data show environmental oxygen levels to be at Pcrit levels of interest.
L377 – Lethal hypoxia has also been estimated from field measurements with organism presence/absence.
I would interpret these as more a sublethal hypoxia threshold being estimated from organism presence/absence for at least mobile species; since most will swim away before at levels higher than the actual lethal levels are reached unless the oxygen loss was rapid and companion data on carcasses was also available
L421 – Metabolic Index –high degree of extrapolation, not only from individual-level lab-based experiments but to entire theoretical niches (based on only a few dimensions…O2 and temperature primarily), but also across entire phyla, coupled with the uncertainties of climate-model forecasts.
The primary discussions coming from this metric have been theoretical hypotheses presented as predictions – as noted, the real-world validation through testing the accuracy of M.I. predictions at the regional scenarios – the extant fisheries perspective remain limited.
When considering the metabolic index under the monitoring / indicator science context – the metabolic index is a derivative of Pcrit theory and so the same comments above apply here as well. The metabolic index in its current formulation doesn’t quite seem like an indicator that could be used to track and detect system-state changes.L489 – Suggest clarifying that indicator species is not a hypoxia-specific idea, and they are bioindicators of a set of environmental conditions or state.
L536 – Indicator species presence may be a straightforward way to detect oxygen changes and is easy to interpret.
Potentially – but cited examples are metazoans only.
Empirical evaluation of how well an organism can act as an indicator species is determined by fidelity and specificity. The hypothetical perfect indicator taxa of ‘deoxygenation waters’ would have high fidelity (the species is present at all sites with deoxygenated conditions) and highly specific (that species is only present at sites with deoxygenated conditions). Since there are no obligate anaerobic water-breathing animals – using any metazoan as an indicator species of deoxygenation conditions would always require additional context and would not be straight forward. I.e., all metazoans can inhabit normoxic waters…if they are absent, it would be another dimension of their niche that is restricting them from occupying oxygenated waters.
Considering the above, consider adding context and specifying obligate anaerobes as potentially more ideal use case for ‘indicator taxa’ theory.
L577–L624 are behavioral responses described not actually individual-level responses when being measured?
This section repeats the measurable parameter described in the individual-level response associated with sensory systems.
E.g. L589 – this paragraph describes the consequence of low oxygen exposure to individual fish rather than presenting as an actual measurable indicator. For a measurable parameter indicator a population-level indicator (or biological response); the population-level response here when describing the compression into shallow waters (shoaling) is the decrease in the average depth distribution of the population.
L635 – Population size
As presented – the measurable parameters presented for population size/growth rate/recruitment are standard ones measured by traditional stock assessment monitoring programs.
Can you provide some discussion on whether this would be a ‘good’ indicator given it relies on having a foundation of another species/individual-specific indicator (lethal O2, L641) and how in reality, if this measurable parameter (standardized counts for a species’) is confounded by so many other variables. In general, consider adding some text that clarifies when indicators may or may not be useful in a realized monitoring program.
L711 - Ecosystem Indicators: L713-L782 – Most of the quantifiable parameters only inform on the community-level of biological organization (e.g. species assemblages metrics); suggest clarifying this in the section subheading.
L752 – there are additional ‘diversity indices’ not mentioned that integrate some level of functioning into their formulation by including simple traits / functional levels into the calculations. E.g. ITI (infauna trophic index) and the AMBI – AZTI indeces are often used in aquaculture-impact assessments marine biotic index that includes simple traits/trophic level categories into the calculations. On the topic of and linking back to anthropogenic over-enrichment; the benthic infauna/macrofauna diversity indeces do include ones that integrate a level of functional. While this doesn’t explicitly link back to hypoxia sensitivity, they are inversely correlated to hypoxia (as the are correlated to high sulphide levels), and integrate trophic traits into their formulation, thus may be slightly better at representing the ecosystem level of organization than pure species assemblage derived diversitiy metrics.
L799-L814, L838-851 – Abundance and Biomass
As written – this is discussion individual population-level abundances without a quantifiable metric that captures the interspecies component of the community re-organization in mind. E.g. Given hypoxia sensitive differs among species – there can be interspecific shifts in abundance levels (some increase, some decrease) as some moderate hypoxia thresholds are crossed, but community-wide biomass decreases as oxygen approaches zero
Analogous indices have been developed for infauna to monitor for system degradation as a consequence of anthropogenic enrichment (e.g. poylchaete/amphipod ratios).
L853 – Taxonomic shifts and ratios:
I wrote the above comment before reading this section – which might work better if it shifts before the abundance/biomass paragraphs.
In a system experiencing gradual deoxygenation – the interspecific shift in species abundances usually happens first (among mobile species - sensitive ones leave, tolerant ones invade) before the community-wide decline in abundance occurs. This would be a useful to suggest as the ‘early-warning’ signal, given the shift in relative abundances among species would theoretically happen first before community-wide abundance decline.
Taxonomic shifts could also go by different names discussed earlier in the paper (i.e., beta diversity), so there is some intellectual overlap here with the earlier section as well.
L930-934 – The metric presented sems to be behavioural response and not a measurable indicator of a community-level component. Given this is the ecosystem-level + ecosystem function discussion section, as a reader I would be looking towards quantifiable biological processes capturing the energetics of the system (e.g. biological rates) – this paragraph could use a retool and link the shift in bioturbation behaviour to the measurable ecosystem indicator.
L1020 – Scaling of Indicators
Might be useful to provide an operational definition on what ‘scaling’ means; the types of ‘scaling’ aren’t clearly defined for the reader. The common theme across the summarized ‘scaling’ approaches appears to be ‘multi-level integration across biological responses to deoxygenation’ but not clearly defined in a way that makes the differences obvious. Given ‘spatial-temporal’ is also discussed (L1023) - I was looking for the traditional concepts/definitions of ecological scale, resolution (grain) and extent, to be presented in a discussion that draws analogies to the biological organization structure theme; wasn’t sure if that was the intent. I feel just a slight refocusing of the wording could help with the clarity in this section.
L1085 – “oxygen-stress”, perhaps reword to deoxygenation stress. Review doesn’t discuss hyperoxia effects (oxygen being too high).
L1186-L1188 - Many of the biological indicators of oxygen stress described in this
1187 paper, if tied to specific DO response thresholds,
From the perspective of, “what is needed in the practical launch and implementation of a realized deoxygenation monitoring program” – this might be the key statement to emphasize. Implementation of any of the parameters to produce longitudinal data (time-series) to track biological responses to deoxygenation will require the natural oxygen levels to be monitored in tandem.Technical corrections:
L433 – Howard et al. (2020) missing from References
L1079 – ROMS-BEC – could you define this acronym?
L1087 – Table 1. Difficult to read with some text cut off within the excel table format.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
338 | 122 | 24 | 484 | 15 | 11 |
- HTML: 338
- PDF: 122
- XML: 24
- Total: 484
- BibTeX: 15
- EndNote: 11
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1