From single to multiple drivers
Experiments manipulating climate-related variables have provided valuable insights into the wide range of biological responses to projected alteration of oceanic conditions, for example ocean acidification (Gattuso and Hansson, 2011; Hutchins et al. 2013) or warming (Boyd et al., 2013). The design and interpretation of these single-driver manipulation experiments, in which a range of altered conditions – such as 550, 750 or 1000 μatm pCO2 - are compared and contrasted with a control treatment (present day 400 μatm CO2) – have been relatively straightforward. Since 2010, increased awareness across the marine science community of the complexity of the many concurrent changes to future ocean conditions (Doney, 2010) has resulted in more studies manipulating several environmental drivers concurrently. For example, one third of the 225 papers at the 2012 SCOR-sponsored symposium on “The Ocean in a High-CO2 World” which reported on the biological response to Ocean Acidification (OA) also manipulated at least one other property (Cooley, 2012). Cooley reported a wide range of permutations of multi-driver perturbation experiments, for example pH and temperature, or CO2 and nutrient manipulations. Figure 1 provides estimates of the number of studies which looked at multiple drivers (acidification plus at least another one) and how this trend has developed with time.
There has been a growing realisation that the experimental outcomes of such multi-driver experiments may not simply be additive and some are therefore highly non-linear, so their interpretation is exponentially more challenging than for single driver experiments (Figure 2).
Moreover, the results of the warming and acidification manipulation study on the sea urchin presented in Figure 2 highlight several important issues that have both direct scientific and wider policy ramifications. First, the effects of multiple drivers can offset or magnify one another, and so provide a different outcome than could be predicted from the results of a single-driver experiment. Second, the outcome of a multiple-driver experiment depends heavily on the selection and magnitude of the individual drivers being combined. Third, accurate communication and predictions of the collective effects of multiple drivers on marine life to policy makers requires consensus (in experimental trends) across a representative number of multiple-driver experiments.
Hence, to provide more reliable estimates of how marine biota will respond to the cumulative effects of multiple drivers requires that we develop comprehensive approaches/studies that progress from single to multiple environmental drivers.
From organisms to ecosystems
The findings from even sophisticated multiple-driver experiments on organisms, such as phytoplankton, that occupy a single trophic level in a foodweb cannot be used to predict how entire ecosystems will respond to complex ocean change (Boyd et al., 2010; Caron and Hutchins, 2013). The components within a foodweb, such as predators and their prey, may respond in very different ways to the same changing ocean conditions. For example, the physiology of microzooplankton (grazers) is more responsive than that of their prey (phytoplankton) to warming (Rose et al., 2009). Hence, as is evident for the previous theme, there has also been progress in the last five years in transitioning from an organismal to an ecosystem-level view of how marine life responds to global change (Brose et al., 2012). There has been increased use of mesocosms (large volume, 1000 L or more, enclosures, Figure 3) to examine marine pelagic ecosystems in coastal and most recently oceanic waters, which has provided valuable information on the responses of the organisms that occupy trophic levels across foodwebs (Calbet et al., 2014). These mesocosm studies provide unprecedented detail on how ecological and biogeochemical processes will be altered by ocean change. This approach has also opened the door for implementing experimental evolutionary biology approaches in natural systems (Scheinin et al. 2015). Other ecosystems, such as those in benthic nearshore waters (from the tropics to the polar oceans) have also been examined via mid-term (months) deployments of innovative large volume (1000 L) experimental chambers such as Free Ocean CO2 Enrichments (FOCE) (Gattuso et al., 2014). Both mesocosms and FOCE enable multiple large-scale multi-disciplinary marine manipulation experiments that detail both ecological and biogeochemical responses to environmental change (Figure 4).
Although these large volume holistic approaches are advancing this theme, they do have limitations, such as the logistical challenges presented in manipulating more than a single driver (Figure 3). This illustrates the need to build strong cross-links with theme 1 which can more readily tackle the effects of multiple drivers. Thus, an approach such as modelling that facilitates integration oforganism to community and ecosystem levels responses is urgently needed.
From Acclimation to Adaptation
Virtually all manipulation experiments, whether based on single- or multiple-driver experiments with organisms, communities, or ecosystems, have not considered the potential for adaptation to influence the outcome of the study (Schaum et al., 2014). In order to detect a measurable response to environmental manipulation, such experiments are primarily conducted using climate change projections for the year 2100, and thus represent a quasi- instantaneous alteration of environmental conditions, for example, increasing pCO2 from present day (400 μatm) to 750 μatm (projected in some climate change IPCC scenarios for year 2100) on a timescale of hours to days. Such an experimental design cannot take into account the abilities of the study organisms to acclimate (days to weeks) or adapt (longer timescales) to alterations of oceanic conditions that occur incrementally over years or decades. Adaptation via micro-evolution for rapidly reproducing organisms such as microbes has been shown to occur on shorter timescales (<1000 generations, years) than previously thought (Lohbeck et al., 2013, Hutchins et al. in press), revealing the ability and indeed the need to consider evolutionary responses in global change experimental design (see Figure 4).
Hence, failure to more accurately mimic the ability of organisms to respond to environmental change in manipulation experiments may give a series of misleading experimental outcomes which could skew predictions of how organisms, communities and/or ecosystems will response to changing oceanic conditions. Thus, this third theme must be interwoven into themes 1 and 2, such that a subset of experiments considers adaptation in their design.