Discussion Session 5: What approaches are available to conduct manipulative experiments on whole communities and assemblages? Does the group contain transferable skills and techniques that could be used in collaborative experiments?

Chair: Jim Barry. Rapporteur: Silvana Birchenough.


What strategy could be most successful in advancing understanding of the consequences of OA (and other factors) for the structure, dynamics, and function of natural ecosystems?

1) Exploit natural variability (pH gradients in space or time, events).

  • Depth, latitude, estuaries, upwelling / non-upwelling sites, seasonal, tidal…
  • Might be most successful by leveraging existing long-term studies of plots or transects in natural communities along or among gradients in carbonate chemistry. 1) CalCOFI (California Cooperative Fisheries Investigations) data sets gradients- changes in oxygen and temperature.  Long term, large spatial scale measurements of hydrographic (T,S,O, PO2, SiO2, NO2, NO3, but not pH), biological parameters (primary productivity, macozooplankton biomass, sample archived, seabird abundance).  Data from the 1950’s onwards. 2) Diablo Canyon power plant thermal outfall. Data collected includes physical observations (temperature, S, maybe O2), and monitoring of community structure in intertidal and subtidal plots from 1978 to present (thermal outfall started in 1984).  Sites include impact area and control sites in region.  May not have much pH variation, but comprehensive data set. 3) LTER sites (Larry Crowder) can be used to look at changes and trends. There are potential sites that can be coupled with long term data sets (Santa Barbara, Antarctic / Palmer Peninsula). 4) Hopkins Marine Station Long Term Studies - Hewatt (1931-1933) transect resampled by Barry et al. 1995 and Sagarin et al. 1998. – Surveys continue today (Fio Micheli). 5) Connell’s long term coral data set in Australia (who is continuing this? Joe Connell, Terry Hughes?). 6) Dove Time series and some opportunistic data sets in the North Sea are available. 7) Intertidal plots at Tatoosh Island (Tim Wooten). 8) Hebrides terrace there is a long- term hydrographic time series and there is not carbonate chemistry. But there are other data sets that will go up to the North Atlantic centre. 9) Deep-sea Monterey Bay data


Most of the time series have not measured pH, but it may be possible to back calculate some of the carbonate chemistry from CTD data, though this may also be impossible in surface systems?

  • Yuri - some of the mean pH values can be calculated, but these are based on the mean variability and there are not directly based on the environment.
  • DIC/alkalinity, is may be possible to back-calculate the carbonate system, but there will likely be significant uncertainty.
  • For organisms reaching 100 years old, there is relationship of studies where pH has been monitored and there could be some other climate proxies.
  • Coralline algae (e.g.Clathromorphum nereostratum) are long-lived and record variation in temperature ands stable isotopes (C,N), but what about carbonate composition, pH? / Nick yes but there is validation to be done and can be used
  • Cold water corals can be very long-lived (millennia), and have been used as proxies for climate or intermediate water composition. Can a pH record be reconstructed? - issues with the proxies and there is oscillation. Maybe info from the ocean drilling programme can be used and inform the programme, including archival material for additional analyses.

What approaches to use in either these existing sites or new research sites?

  • Natural experiments: Gradients, Events
  • Manipulative experiments: Laboratory, Field
  • Focus? Single / multiple factors, Direct effects, Interactions/indirect effects, Population bottlenecks ( e.g. recruitment)

2) Experimental Approaches - FOCE

  • Shallow Water-FOCE is one approach (photos of FOCE at Heron Island and installation at Hopkins Mar. St. (2012)


  • Steve: open vs closed field experiments will give different results
  • Mark: need to make this distinction clear
  • Jim: more of a spectrum as the field mesocosms are more realistic than the lab versions
  • FOCE experiments are open treatments but are subject to supply of gametes, etc. from outside untreated areas. So need to be mindful of negatives associated in this type of experiment.
  • Piero: cannot extract all problems and look for consistencies in data from multiple approaches, integrate experiments.
  • Fiorenza: need to translate field patterns in abundances, demographic rates, mortality, growth, combine with experiments and then with models project the population trajectory etc. Identifying bottlenecks in lifecycles of species of interest and then scale-up with models can inform management.
  • Jim: counter-argument, population models and understanding demography are key to understanding our system, but manager and the public are likely to be more worried about resource extraction - e.g. how big are the commercial fish going to be?
  • Larry: whole lake manipulations did work, what about marine?  E.g. saline lagoons. This would be at the systems scale and could be very powerful. 
  • Jim: similar suggestions have been made for fjords, either by acidification or alkalinization.
  • Shallon: are there any thoughts to alkalinize any of the natural CO2 vents?
  • Kristy: have thought about transplants along pH gradients at vents, but not about changing the chemistry
  • Piero: need to be careful as in these areas you may have already lost genetic variability.  Transplants have ethical issues. Can lose all of the genetic diversity within the original patch if the transplanted animals do well.
  • Jennifer: shell addition to sediments by shellfish growers have increased DIC in sediments. Experiment on shell plots, many more clams in the shell plots, added plain rock and again these had many more clams. This is likely a predator effect rather than an abiotic effect – good to put out different types of structures.
  • Giulio: if there is a direct effect on the target species i.e. abalone, specific lab experiments of exposure concentration and duration and look at responses of key life stages and biological responses.  If there is an indirect effect e.g. predation, it may not be possible to replicate these in mesocosms so it becomes more complex. May not have an a priori answer of the best way to tackle this. Make an association between particular stressors and responses.
  • Jim: are you interested in the trajectory at the system level i.e. coral reefs, or reduced level? With limited funds, how should plots be allocated along a gradient? How do we decide what to spend on experimental plots versus research to determine the mechanisms (e.g. physiological studies?).
  • Giulio: If change is seen in one or two species the field, then it is worth bringing these into the lab to test hypotheses concerning what factors have caused the observed changes in the field.
  • Steve: long-term time-series do encompass time-scales over which OA is occurring, but if we want to look at resilience we need experiments to push communities to see how quickly they respond, not mimicking a future high CO2 scenario -  just force the system and take them to areas with natural variability. Use as a tool to challenge communities of different complexity, challenge what happens to and between species. FOCE is a great asset.
  • Helen: forcing a system also looks at to what degree it can bounce back, i.e. resilience as well as resistance/vulnerability. These can often by the type of data that modelers can use.
  • Jim:  unsure how useful FOCE would be for this. Intertidal systems turn over in around a decade. Maybe look at shorter timescales,.
  • Shallon: FOCE – what can you infer from the results given that it is embedded in a larger, non-acidified system?  How do you interpret your results accurately?
  • Jim:  need to be specific with the question/hypothesis.How does OA mediate recruitment of barnacles under competition, predation, or both?  This might be assessed by deploying barnacle settlement plates in the FOCE system with and without grazers and predators, plus newly settled barnacles at different ages.
  • Mark: why does it have to be only 2 treatments with 3 replicates?
  • Jim: Cost – we hope to increase the number of FOCE chambers in the future.  Presently, this will be minimally scientific.
  • Why not use regression (with no replicates) instead of ANOVA (replicates)?
  • Martin: Emmet Duffy had same problem with seagrasses so pumped into a mesocosm, acidified and then pumped back into the chambers in the ocean.
  • Question can we raise an organism in the lab and then explore the performance in the field?
  • Jim: Are there key experiments that we want to do at different sites?
  • Steve: would like to see more pan scourer-type experiments from different ecosystems. Can start to combine responses and functions at a basic level (balance CaCO3 and non-calcifying organisms) and then bringing them into the lab and then challenging them with different pH treatments.
  • Piero: tiles within mussel beds to collect the assemblage within a biogenic reef.  Identify an item that provides the main ES then we can standardize experiments.
  • Jim:  good way to look at recruitment and then test in lab, but also important to look at recruitment of dominant species, barnacles, mussels, recruitment panels to monitor them.
  • Exploit the gradients to see how these organisms are recruiting along a gradient.
  • Martin: need to standardize across sites, factor in T, season, depth etc.
  • Jim: we need to make a list of priorities, costs, manpower etc. as this will be expensive
  • Kristy: if we use mesocosms as well need to ensure the conditions are actually the same as some systems are better than others.
  • Martin: do the different systems affect the response? Standardize the scourers, etc. to all systems and see if you get the same response?
  • Martin: interested in thought experiment – what are the effects of OA on different services? Little ‘test’ for the group.
  • Silvana: are some of the traits across the systems common enough to look at as well as the actual data. 
  • Fiorenza: clearance, recruitment, recovery experiments, frequency of monitoring. 

Useful Plan: exploit local to regional gradients in similar ways at a network of global sites, so that the network can provide a global overview.

Strategy focus:

  • Demographic bottlenecks
  • Recruitment, survival, fecundity, reproduction
  • Recovery / succession
  • Develop projects around existing long-term study sites ( CalCoFI, various intertidal, MCO etc.)


  • Brian: likes our hypotheses as they are system-independent therefore could be powerful if a general theme emerges across different systems.
  • Jim: Ecosystem Services (ES) not easy to cost
  • Piero:  what is the effect of OA in the context of ES?  Then this is mis-focused or doesn’t answer the questions.
  • Jim:  we need to define the ES in relation to or resulting from the processes/sp. we are measuring.
  • Martin:  ecologists are ‘uncomfortable’ with the term ES.  Can’t measure coastal protection but can measure habitat complexity etc.   Measure factors contributing to a service and then package this as a service.
  • Nova:  can we use PML’s economists and the MBA’s ES people into assisting if we need to know this?
  • Steve:  We should communicate with this group of people to ensure that what we do is appropriate but we should be attacking the ecological questions that are valuable. Mel Austen is running a workshop to look at ecosystem valuation at the UK/US level so we can always link up with these groups.  But this shouldn’t detract from the science.
  • Jim:  summary; be mindful of the ES but not to let it drive our science.
  • Jennifer:  FOCE – what can they learn from the FACE experiments?  FACE probably did lab experiments and free air experiments and someone probably compared the two, most of the FACE experiments were press perturbations over long time-frame, initial responses were then modified over time as organisms adapted.  Combined with other manipulations e.g. diversity, density.  We know species interactions can be very strong,
  • Lorenzo:  we are restricted to a few FOCE systems at the moment.
  • Jim:  For now, there is only 1 chamber at the Villefranche site, PML is also becoming involved.  Another FOCE being install here at HMS (initially 2 chambers in late 2012), and now (2013, funding has been secured for an Antarctic FOCE installation at the Australian base.
  • Characterization of the carbonate system during experiments is important: 1) Two systems that most people are working in, rocky intertidal and soft sediment, need alkalinity and DIC samples from these sites to characterize carbonate chemistry. 2) SeaFET pH Loggers. $5-10K Satlantic intertidal loggers.  To build it costs around $1200 (maybe).  Logger will run without drift for up to 6 months.  Hydrolab mini logger – 4-5 sensors, larger logger with lots of sensors, but only has a 2 week lifespan.
  • Many of us are already doing recruitment studies in relation to OA. 
  • Many of us can identify foundation or habitat forming/structuring features.
  • What is driving the observed dynamics of these foundation/keystone species.
  • Kristy:  likes the tuffy idea but it doesn’t target the foundation species that are important to community structure and functioning.
  • Fiorenza:  likes the idea of using mussels as a low hanging fruit.
  • Clearance/settlement for recruitment characterizing the habitat, take photos and analyze later, coupled with tuffies.
  • Steve:  we have agreed on a set of hypotheses, a set of questions based on what we do and what we hope to achieve, these should spawn a series of experiments and observations.
  • Shallon:  are the hypotheses just for FOCE or general?  How would you explain to managers how these experiments relate to OA?
  • Steve: the ‘resilient’ hypothesis is about understanding how changes in CO2 concentrations will affect community and ecosystem responses to other perturbations?
  • Fiorenza: If you expose species that come from a naturally more variable system they are likely to be more resilience to the pH change. The exposure to OA affects the resilience of the community to other perturbations via a loss of function (resilience in this instance).
  • Ted:  how would you do this? Some kind of common garden experiment?
  • Fiorenza:  have an additional disturbance, ideally along a gradient.  Functional services from the system will decrease as systems become less diverse and less resilience.  A combination of field and lab experiments to take spp from along the gradient and test physiological parameters will help identify sensitivities.
  • Murray:  how is productivity going to change around the UK?  How are deep water systems which are affected my intermediate water masses going to recover via refugia – if the properties of the intermediate water masses change then this may not occur.  These haven’t been tackled yet in this workshop.
  • Jim: how can we connect inshore with deep sea?
  • Nova:  biogenic habitats (does this really mean deep-sea corals / sponges?)
  • Murray:  connectivity with shallow ecosystems via circulation of water masses.
  • Jerry:  lack of data to valuate such models of depth profiles of all environmental/physical/chemical etc. 
  • Jerry: new models emerging which have variable resolution 5km @ coast up to 10km further out – new but do exist.  Need info on gradients from shore to deep and to make the connections with what we want to look at.
  • Shallon:  hypotheses seem cross-cutting between benthic, pelagic, coastal and offshore systems.
  • Silvana:  pressure in the UK for priority habitats BAP – may be similar questions across different systems, applicable to managers.  MCZs, MPAs, etc.
  • Larry: biophysical better understood offshore whereas biology better understood inshore and look for biologists who are interested in offshore and modelers who are interested in inshore.
  • Yuri:  in a system with a lot of variability, the mean isn’t that relevant, the exceptional value is not driven by the long-term signal. 
  • Jim:  change in the mean is a significant component of the variability, can be up to 50%.
  • Jerry:  seasonality of undersaturation is going to increase, so when is the exposure threshold exceeded? 
  • Jim:  how does pH affect behavior?  Energy availability?  How is energy intake mediated if performance is depressed?  Maybe you just need to know the relationship between change and performance rather than identifying mechanisms.
  • Brian:  scale of change in mean aragonite saturation is much larger in relation to the variability compared to temperature.
  • Jim:  What is the daily variability in the N Atlantic for aragonite saturation?  Aragonite saturation boundary in the eastern Pacific is ~150m, ~2400m in Atlantic, shoaling by ~0.4m per year in the Pacific, but ten times that in the Atlantic.  Thus, both oceans are changing rapidly in response to increasing CO2, but the Atlantic will undergo more drastic changes in deep water chemistry.

Additional Issues to Consider

  • How will single versus multiple stressors affect ecosystem structure and function?
  • Will direct or indirect effects be more important in shaping ecosystem structure and services in response to OA and other stressors?
  • Are there key focal areas that will advance understanding of ecosystem responses to OA? Population studies (demography), Energy flow, Population bottlenecks (i.e. life stage sensitivities).


The group explored a range of ideas and approaches that were considered to be fruitful areas of research and collaboration to advance our understanding of the effects of ocean acidification on the structure and function of marine ecosystems.  The major thrust of the discussion focused on combining new research with established time series or data sets that provide some context, at least for the structure and variability of selected ecosystems.  Some priorities or important considerations for future OA Ecosystem research included;

  • Studies of population and ecosystem dynamics along gradients in pH (or other factors)
  • Combination of field and laboratory studies that exploit new methods (e.g. FOCE) and examine species specific sensitivities or responses in focused physiological studies.
  • Greater focus on population demography for taxa important for the structure and function of ecosystems.
  • Goals would benefit from an integrated approach that: Quantifies patterns of abundance of key species, and rates of growth and demography; Uses focused experiments to understand mechanisms (or underlying physiological responses); Projects population trajectories and identifies key population bottlenecks, and Scales up this assembly of information using models that will inform resource managers (i.e. address or predict variation in ecosystem services).
  • The cascading direct and indirect effects of OA may make interpretation of field studies more complex.
  • Early FOCE systems with limited replication may have limited power to detect the effects of OA.  Important to consider design constraints of any experimental approach.
  • A global network of research sites using similar approaches (e.g. pan scourer-type experiments) in different ecosystems would help understand if there are fundamental similarities in the response of natural ecosystems to OA, such as shifts in the balance of calcifying and non-calcifying species.
  • Laboratory experiments can supplement field ecosystem studies by testing hypotheses concerning the specific responses of particular tax to OA.
  • It is useful to consider the impacts of OA on ecosystem services, but OA studies are more likely to be focused on other, more easily measured metrics of population and community responses.
  • OA ecosystem research should exploit the approaches and lessons learned by the FACE research community.
  • Characterization of the carbonate chemistry will be important for ecosystem studies.
  • Foundation species are high priorities for ocean acidification studies.
  • How does the research community link nearshore community dynamics and models with deep-water systems?
  • How will natural communities respond to single and multiple stressors?