Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

A Synergistic Approach for Evaluating Climate Model Output for Ecological Applications

A Synergistic Approach for Evaluating Climate Model Output for Ecological Applications ORIGINAL RESEARCH published: 26 September 2017 doi: 10.3389/fmars.2017.00308 A Synergistic Approach for Evaluating Climate Model Output for Ecological Applications 1 1 1 1 Rachel D. Cavanagh *, Eugene J. Murphy , Thomas J. Bracegirdle , John Turner , 1† 2 3 1 Cheryl A. Knowland , Stuart P. Corney , Walker O. Smith, Jr. , Claire M. Waluda , 1 4, 5 2, 6 7 Nadine M. Johnston , Richard G. J. Bellerby , Andrew J. Constable , Daniel P. Costa , Edited by: 8 1 1 9 Eileen E. Hofmann , Jennifer A. Jackson , Iain J. Staniland , Dieter Wolf-Gladrow and Elvira S. Poloczanska, 1, 10 José C. Xavier Alfred-Wegener-Institut für Polar- und Meeresforschung, Germany 1 2 British Antarctic Survey, Cambridge, United Kingdom, Antarctic Climate and Ecosystems Cooperative Research Centre, Reviewed by: University of Tasmania, Hobart, TAS, Australia, Virginia Institute of Marine Science, College of William and Mary, Gloucester Diego M. Macias, Point, VA, United States, SKLEC-NIVA Centre for Marine and Coastal Research, East China Normal University, Shanghai, 5 6 European Commission. Joint China, Norwegian Institute for Water Research, Bergen, Norway, Australian Antarctic Division, Australian Commonwealth Research Center, Italy Department of Environment and Energy, Kingston, TAS, Australia, Department of Ecology and Evolutionary Biology, Nova Mieszkowska, University of California, Santa Cruz, Santa Cruz, CA, United States, Center for Coastal Physical Oceanography, Old Marine Biological Association of the Dominion University, Norfolk, VA, United States, Alfred-Wegener-Institut Helmholtz Zentrum für Polar- und United Kingdom, United Kingdom Meeresforschung, Bremerhaven, Germany, Departamento das Ciências da Vida, Marine and Environmental Sciences Centre, Universidade de Coimbra, Coimbra, Portugal *Correspondence: Rachel D. Cavanagh rcav@bas.ac.uk Increasing concern about the impacts of climate change on ecosystems is prompting Present Address: ecologists and ecosystem managers to seek reliable projections of physical drivers of Cheryl A. Knowland, change. The use of global climate models in ecology is growing, although drawing King’s College London, Graduate School, London, United Kingdom ecologically meaningful conclusions can be problematic. The expertise required to access and interpret output from climate and earth system models is hampering progress Specialty section: in utilizing them most effectively to determine the wider implications of climate change. To This article was submitted to Global Change and the Future Ocean, address this issue, we present a joint approach between climate scientists and ecologists a section of the journal that explores key challenges and opportunities for progress. As an exemplar, our focus Frontiers in Marine Science is the Southern Ocean, notable for significant change with global implications, and on Received: 13 June 2017 sea ice, given its crucial role in this dynamic ecosystem. We combined perspectives Accepted: 08 September 2017 Published: 26 September 2017 to evaluate the representation of sea ice in global climate models. With an emphasis Citation: on ecologically-relevant criteria (sea ice extent and seasonality) we selected a subset of Cavanagh RD, Murphy EJ, eight models that reliably reproduce extant sea ice distributions. While the model subset Bracegirdle TJ, Turner J, Knowland CA, Corney SP, shows a similar mean change to the full ensemble in sea ice extent (approximately 50% Smith WO Jr, Waluda CM, Johnston decline in winter and 30% decline in summer), there is a marked reduction in the range. NM, Bellerby RGJ, Constable AJ, Costa DP, Hofmann EE, Jackson JA, This improved the precision of projected future sea ice distributions by approximately Staniland IJ, Wolf-Gladrow D and one third, and means they are more amenable to ecological interpretation. We conclude Xavier JC (2017) A Synergistic that careful multidisciplinary evaluation of climate models, in conjunction with ongoing Approach for Evaluating Climate Model Output for Ecological modeling advances, should form an integral part of utilizing model output. Applications. Front. Mar. Sci. 4:308. doi: 10.3389/fmars.2017.00308 Keywords: IPCC, CMIP5, climate models, Southern Ocean, marine ecosystems, climate change, sea ice Frontiers in Marine Science | www.frontiersin.org 1 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology INTRODUCTION (Stock et al., 2011), and the time frames of human activities (such as fishing) and political decision-making. At these shorter Natural variability in the climate system and anthropogenic temporal scales it is difficult or impossible to distinguish between climate change result in a complex array of physical and natural variability and the background climate change signal biological responses. Marine ecosystems are inextricably (Macias et al., 2013). Furthermore, many key ecological processes connected to the climate system and significant changes to their occur at regional (i.e., tens to hundreds of kilometers in structure and function are both observed and expected (Doney extent) or smaller scales, hence biological responses to change et al., 2012; Blois et al., 2013; Sydeman et al., 2015). Effects also vary at these scales (Helmuth et al., 2006; Clarke et al., may be direct (e.g., temperature changes affecting physiological 2009; Peck, 2011; Chave, 2013). Similarly, resource conservation processes such as growth, reproduction, consumption and and management is also often concerned with regional or respiration), or indirect (e.g., those resulting from changes smaller scales (e.g., subareas, divisions or subdivisions in the to primary productivity, which in turn can influence species case of fishing areas) that may contain relatively discrete abundance, distributions and interactions; Constable et al., populations of certain species (Stock et al., 2011; Sydeman et al., 2014). Biological feedbacks to the climate system such as the 2015). role of biology in carbon sequestration are recognized (Hauck Here we focus on the Southern Ocean as an exemplar region and Völker, 2015; Hickman, 2015) although difficult to quantify for exploring the use of IPCC-class models in studies of ecological (Passow and Carlson, 2012). change. The region was the subject of much attention during Concern about the effects of climate change on marine IPCC Fifth Assessment Report (AR5), owing to its important ecosystems is growing (Hoegh-Goldberg and Bruno, 2010; role in the global climate system and the significant changes Cheung et al., 2013; Halpern et al., 2015) and is a major focus already observed (Le Quere et al., 2007; Böning et al., 2008; of global environmental research programmes and targets such Hauck et al., 2013; Kennicutt and Chown, 2014; Larsen et al., as Future Earth, the Scientific Committee on Oceanic Research 2014; Landschützer et al., 2015). One of the most notable changes (SCOR) and the United Nations Sustainable Development Goals. in this region has been the increase in westerly wind speeds Ecologists are increasingly being asked to provide advice in associated with stratospheric ozone depletion (Turner et al., 2009; this regard to support conservation and management decisions Thompson et al., 2011; Meijers, 2014). There is strong evidence (Hollowed et al., 2013; Barange et al., 2014; García Molinos that this has had important effects on the Southern Ocean, in et al., 2015). Understanding and predicting these effects requires particular sea ice (Ferreira et al., 2015). However, major issues knowledge of the processes that determine the distribution remain with regard to reproducing the observed physical state and abundance of species, the structure and functioning of of the Southern Ocean in global climate models, and these have ecosystems within which they occur, and their past and present implications for the reliability of predicted responses to future dynamics (Southward et al., 1995; Mieszkowska et al., 2006; Lima climate forcing (Turner et al., 2013; Meijers, 2014). et al., 2007; Doney et al., 2012; Murphy et al., 2012). This is Considerable progress has been made with qualitative coupled with the need to understand the key physical drivers assessments of the effects of physical changes on Southern of change and for reliable projections of them. As such, the Ocean ecosystems, revealing key drivers (including sea ice, use of Intergovernmental Panel on Climate Change (IPCC)- wind, various water mass properties and mixed layer depth) class climate models (Box 1) is rapidly gaining momentum in and complex responses throughout the food web (e.g., Murphy ecological studies of change (Hunter et al., 2010; Bopp et al., 2013; et al., 2007; Massom and Stammerjohn, 2010; Gutt et al., 2012; Jenouvrier et al., 2014; Piñones and Fedorov, 2016). Constable et al., 2014; Hunt et al., 2016; Xavier et al., 2016). Although these models are designed to provide realistic Broadly, the effects are likely to include impacts on habitat, representations of the climate system and projections of climate physiology, distribution, population densities, phenology, variables (e.g., atmospheric and ocean temperatures, sea ice and community interactions (Trathan and Agnew, 2010). extent and winds) that are known to influence ecological Quantitative assessments are more difficult. Relevant biological processes, applying them to ecological problems is challenging data are often patchy or imprecisely known, while effects are (Stock et al., 2011; Snover et al., 2013; Harris et al., 2014). often indirect and multifaceted (Bednarek et al., 2011; Murphy No model is perfect and none reproduce all current and past et al., 2012; Melbourne-Thomas et al., 2013; Gutt et al., 2015). climates, so projections need to be used with care. There are This makes the challenges of assessing future change even greater important caveats that users must consider and these issues (Hill et al., 2013; Kawaguchi et al., 2013; Melbourne-Thomas are likely to be compounded in the ecosystem context where et al., 2016). additional considerations, such as differences in spatial and Despite the challenges, quantitative assessments of the temporal scale, may be significant. Most experiments using effects of change are urgently required, not least because the IPCC-class models are designed to provide projections on time conservation and management of the Southern Ocean must scales of the order multiple decades (>30 years) and tend to account for the dual realities of climate-driven ecosystem change be run with spatial grids of around 200 km. On such temporal and growing demand for fishery resources (Murphy et al., 2008; scales, climate change signals should be clearly distinguishable CCAMLR, 2009; Trathan and Agnew, 2010; Hill et al., 2013). from natural variability. However, from an ecological perspective, Input from the scientific community is critical in providing shorter (<30 years) time frames are key due to biological policy-relevant information on climate change impacts for use in processes, including life cycles, generation lengths and phenology decision making. Frontiers in Marine Science | www.frontiersin.org 2 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology BOX 1 | IPCC-class climate models. By “IPCC-class models” we refer to full complexity coupled atmosphere-ocean-sea ice- climate and earth system models. This type of model is a fundamental tool for quantifying how the environment may change in the future. They are increasingly able to take account of complex processes, including the “ozone hole,” carbon uptake in the ocean, and atmospheric aerosols. Because of the complexity, non-linearity and small horizontal scale of many meteorological and oceanographic processes, these models require intensive computer processing power and are therefore run at only a few dozen climate research centres. They can be run with “pre-industrial” concentrations of greenhouse gases (GHG) before anthropogenic forcing is introduced (from mid-nineteenth century) to quantify the impacts of the known increase in greenhouse gas concentrations and the development of the ozone hole. In addition, possible trajectories of twenty- first century climate change can be derived from climate model simulations run under different (GHG) emission scenarios and recovery of stratospheric ozone amounts. Future GHG emissions will be the product of complex dynamic systems, determined by a combination of political decisions, demography, socio-economic development and technological change. Their evolution is highly uncertain and a range of plausible scenarios have been defined to provide a common set of future emission outcomes that can be used to help compare results across different climate models. The set of scenarios of future change in climate forcings that were used in the IPCC Fifth Assessment Report (AR5) are termed Representative Concentration Pathways (RCPs) (Meinshausen et al., 2011). There are four RCPs (RCP8.5 which corresponds to the pathway with the highest GHG emissions, followed by RCP6, RCP4.5, and RCP2.6), each one represents a different trajectory and cumulative emission concentration to 2100. A major initiative that provided much of the climate model data for the most recent IPCC report is the Coupled Model Intercomparison Project Phase 5 (CMIP5, managed by the World Climate Research Programme) (Taylor et al., 2012). The CMIP5 dataset is a comprehensive set of outputs from approximately 60 of the world’s most sophisticated climate models from major climate modeling centres (such collections are often referred to as climate model ensembles). The CMIP model intercomparisons have historically been synchronized with the preparation phases of IPCC reports, hence the models (and model setups) are often referred to as “IPCC-class” models. The most up-to-date climate models are currently being prepared for CMIP6 in advance of the IPCC Sixth Assessment Report (AR6), which is scheduled to be released around 2021-2022. IPCC-class models are essential tools for quantitatively of the Southern Ocean sea ice extent (SIE) cycle (e.g., late- integrating knowledge of the climate system and making summer minimum and amplitude of the annual cycle), we used projections. There are general recommendations for ecologists selection criteria following the methodology of the Arctic sea on their use (e.g., Snover et al., 2013; Harris et al., 2014), ice analysis of Wang and Overland (2009), adjusted to reflect but information specifically for climate scientists about the the timing of austral seasons. This specifies that simulated requirements of ecologists is lacking. Furthermore, while sea ice must fall within ±20% of satellite data for mean generic guidelines are useful, detail specific to particular minimum SIE and mean seasonality (where seasonality is regions is also important, and expert-agreed benchmarks the annual difference between the maximum and minimum would be valuable. Here, with our focus on the Southern SIE). Ocean, and on sea ice, given its crucial role in this dynamic system (Box 2), we ask how these models can be used most effectively to understand, and manage, the responses Data of species, communities and ecosystems to change. Our For the analysis presented here monthly-mean sea ice study represents a first step in bringing members of the concentration data was used (CMIP5 variable name “sic”), Southern Ocean climate and ecological community together to which is the proportion of each model grid cell that is covered by jointly explore key challenges and consider opportunities for sea ice. Output from two different types of model experiments progress. were retrieved. The CMIP5 data can be downloaded from the Earth System Grid https://pcmdi.llnl.gov/projects/esgf- MATERIALS AND METHODS llnl/. Firstly “historical” simulations, which are run with observed concentrations of greenhouse gases and other Capabilities and Limitations of IPCC-class known important climate drivers between the mid-nineteenth Climate Models century to the present. Second, global warming simulations The first step in this study was the compilation of a set of following a scenario of high emissions of greenhouse gases key questions that Southern Ocean ecologists would ideally like (RCP8.5). Future climate conditions are taken from the late to address through the use of climate models. To begin to twenty-first century in the RCP8.5 simulations (2079-2099) consider the challenges of using IPCC-class models to address and these are compared against baseline (i.e., as a reference such questions, we focused in detail on one important driver, sea period from which to define twenty-first century change and ice, owing to its crucial role in the Southern Ocean (Box 2) and over which to compare observations against climate model identified properties of particular ecological significance. output) conditions taken from the period 1979-1999 in the historical simulations. After first eliminating the models for Sea Ice in CMIP5 Models which both monthly sea ice concentration data and data for the We evaluated the representation of Southern Hemisphere sea ice RCP8.5 high emissions scenario were not available, a total of in CMIP5 models with an emphasis on extent and seasonality 35 different CMIP5 models were identified (see below). These due to the crucial ecological roles of these properties (see baseline conditions were also used for comparing models and below). To assess which models most closely reproduced aspects observations in the sub-setting procedure. The dataset used was Frontiers in Marine Science | www.frontiersin.org 3 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology BOX 2 | Southern Hemisphere sea ice. Southern Hemisphere sea ice has a range of influences in the Earth System including on global heat and carbon cycles, and sea-level rise, and is a sensitive indicator of change in the Polar Regions; therefore, understanding its variability in the climate system is critical. It forms one of the largest, most dynamic marine ecosystems on Earth, with marine biota being highly adapted to its presence, seasonality and properties over evolutionary timescales (Clarke et al., 2007; Massom and Stammerjohn, 2 2 2010; Constable et al., 2014). Overall, sea ice extent around the Antarctic ranges from a late winter peak of ∼19 million km to a minimum of ∼3–4 million km in summer (Massom and Stammerjohn, 2010). Since the late 1970s the annual mean total Southern Hemisphere sea ice extent has increased by ∼3%, albeit with considerable regional contrast and variability (Turner et al., 2013; Gagné et al., 2015; Simmonds, 2015). However, as the impacts of shorter term variability (associated with natural variability and ozone depletion) recede over longer time frames, this overall increasing trend is projected to reverse in direction (Bracegirdle et al., 2008; Turner et al., 2013). the Bootstrap version 2 product described in Comiso (2000; Supplementary Table 3 also includes information for other key updated 2015). drivers, detailing aspects of these that are particularly important ecologically, providing a baseline for future work. Sea Ice Diagnostics and CMIP5 Model Selection Sea Ice in CMIP5 Models The basic parameter used for evaluation was total Southern The criteria we applied reduced the available 35 models Hemisphere sea ice extent (SIE), which was calculated from (Supplementary Table 4) (referred to hereafter as the full monthly mean sea ice concentration data (both the CMIP5 and ensemble) to a subset of eight models (referred to hereafter as the satellite Bootstrap 2 datasets). It was defined as the area enclosed subset). The importance of this sub-setting is presented below. by the 15% concentration contour of the sea ice concentration fields over the Southern Hemisphere. Effect of Sub-Setting on Historical and Future Further details are provided in Supplementary Material CMIP5-Derived Sea Ice Distributions Methods. Late twentieth century sea ice distributions from the full ensemble are shown both for austral summer (December- February) (Figure 1A) and winter (June-August) (Figure 2A). RESULTS These highlight the large ranges in the summer and winter Capabilities and Limitations of IPCC-class distributions in the full ensemble for the 1979-99 period. Large ranges are also seen in projected sea ice distributions in the late Climate Models Considering the key questions identified by Southern Ocean twenty-first century (Figures 1B, 2B). The inter-model range of sea ice in the historical runs is ecologists alongside current capabilities and limitations of the models, revealed fundamental differences in perspectives and much smaller for the subset (Figures 3A, 4A). This is expected since those with large deviations have been removed. However, approaches between the two disciplines. These include methods; research interests and priorities; requirements of the models Figures 3B, 4B illustrate that this narrower range is maintained in the projected sea ice spatial distribution in the late twenty- (Supplementary Table 1); and terminology (Supplementary Table 2). Issues around temporal and spatial scales were noted in first century. This emphasizes that model-projected future sea ice distributions (and associated variables) are highly dependent particular, as was the need to resolve detailed features such as the Marginal Ice Zone (MIZ) (i.e., the transition area between open on the match between simulated and observed climatological SIE conditions (Risbey et al., 2014; Bracegirdle et al., 2015). ocean and sea ice). In terms of ensemble mean change in sea ice concentration, Sea Ice As an Exemplar Variable it is evident that the subset gives more clearly defined regions Focusing on sea ice as an exemplar variable, information was of ice reduction with larger changes in many regions. This is compiled on properties of particular ecological relevance. Sea particularly apparent in winter (Figures 2C, 4C). These regional ice forms an essential surface habitat for resting, breeding and differences between the full ensemble and subset would likely feeding, and sub-surface habitat for food and refugia, it also be more distinct in evaluations of sea ice at specific locations or influences water column properties, and affects the reproductive within specific sectors. cycles, recruitment and foraging behavior of a wide range of The mean twenty-first century change in SIE is similar both species. It plays a pivotal role in Southern Ocean biogeochemical in the full ensemble and the subset, with respective changes of 49 cycles, and influences fisheries, not only through its impact on and 52% in summer and 31% for both in the winter. However, the the distribution and potentially the abundance of target species, range in projected CMIP5 SIE change is smaller across models but by affecting the access of vessels to fishing grounds. It is that more closely reproduce the observed SIE climatology, with a clear that the seasonal advance and retreat of sea ice (timing standard deviation from 12 to 8% in winter and a reduction for and extent) is a major driver of the structure and functioning of the summer from 27 to 19% (Supplementary Table 5). It should the Southern Ocean pelagic ecosystem, and this formed the basis be noted that proportional changes are not as strongly influenced for our evaluation of the models (see Materials and Methods). by weighting, since, for example, models with too little sea ice Supplementary Table 3 summarizes the information collated on produce smaller absolute changes that are a similar proportion of sea ice. Recognizing the range of multiple stressors involved, their own too-small SIE. Frontiers in Marine Science | www.frontiersin.org 4 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology FIGURE 1 | Austral summer (DJF) sea ice distribution for all available 35 models from historical simulations over the period 1979-1999 (A) and RCP8.5 (2079-2099) (B) and the RCP8.5-historical difference (C). On each plot the filled contours show the all-model ensemble mean sea ice concentration, the red lines show the ice edge for the model with most ice and the green lines show the ice edge for the model with least ice. The black lines on (A) show the observed extent (for 1979-1999 period) from the Comiso Bootstrap 2 dataset (see Materials and Methods). FIGURE 2 | As Figure 1 above, but for winter (JJA). DISCUSSION biased projection) but, not sufficient (i.e., an accurate baseline climatology will not necessarily produce an accurate projection Sea Ice in IPCC-class Models of future change) condition for producing reliable projections The first step in the study highlighted the importance of of future change. Despite the large differences between observed understanding the nature and capabilities of models and the and simulated climatological sea ice extent in many climate need for careful communication and interpretation of their models (Turner et al., 2013), previous projections for Southern outputs. Crucially, we have demonstrated the importance of Hemisphere sea ice have either treated all CMIP models equally model evaluation as a means to improve projections of change. (Collins et al., 2013) or used weightings based on other variables For both communities, it is imperative that models reproduce (Bracegirdle et al., 2008). the historical climate satisfactorily. In the case of sea ice this is Our analysis shows that applying criteria, motivated by a necessary (i.e., a biased baseline climatology will produce a ecological considerations, to select a subset of the available Frontiers in Marine Science | www.frontiersin.org 5 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology FIGURE 3 | As Figure 1 above, but for the subset of 8 models. FIGURE 4 | As Figure 2 above, but for the subset of 8 models. CMIP5 models dramatically reduces the range in projected subset and associated projected sea ice changes presented above. late-twenty-first century sea ice distribution and twenty-first However, interpretation of these must be accompanied by an minus twentieth century absolute sea ice extent change. By understanding of the associated uncertainty and caveats. Key better representing Southern Hemisphere sea ice extent and issues include the subjectivity in defining thresholds for sub- seasonality, the subset provides more ecologically meaningful setting (e.g., the 20% threshold used in this study), and the results (Figures 1–4). Box 3 captures some of the potential possibility that the “good” models share common biases and consequences for projecting ecosystem change based on poorly “get the right answer for the wrong reason.” Nevertheless, it is represented sea ice extent and seasonality. clear that a necessary condition for capturing realistic changes in ice-edge environments would be a satisfactory representation of Key Challenges and Recommendations past and current conditions (Bracegirdle et al., 2015). The sub- Therefore, as a first order assessment of the utility of IPCC-class setting technique used here improved projections by taking this climate models in Southern Ocean ecosystem research, and in into account. The difficulty comes in determining the reliability particular sea ice, we propose that ecologists use the eight-model of changes exhibited in sub-sets of “better” models. As the Frontiers in Marine Science | www.frontiersin.org 6 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology BOX 3 | Implications of using models that poorly represent Southern Hemisphere sea ice to project ecological responses to change. Efforts to incorporate biology into physical models tend to include only components of the lower levels (e.g., phytoplankton) (Murphy et al., 2012). However, physical changes can affect all trophic levels. Sea ice has complex multi-directional effects throughout the food web that influence the structure, function and dynamics of Southern Ocean ecosystems (Figure 5, Supplementary Table 3). These can be direct effects (D) as with the provision of crucial habitat and its impact as a physical barrier. They can also be indirect (I) through effects on physical conditions such as upper ocean temperature, irradiance and vertical mixing that in turn influence key ecological processes, food type and availability, and species composition, distributions and abundance. To capture some of the consequences of using models that poorly represent sea ice here we present a simplified view of some of its key effects in a Southern Ocean pelagic food web (for examples of detailed Southern Ocean food webs see, e.g., Ducklow et al., 2007; Murphy et al., 2007, 2012, 2013; Smith et al., 2007). Consider a model(s) that has sea ice melting a month early or a month late. The details of confounding factors and interactions aside, poorly-representing sea ice (in this case, melting too early or too late) will significantly impact phytoplankton and hence the overall productivity of the system: Physical factors—One possible (greatly simplified) outcome is that ice melting too early exposes the water column to increased light, as well as increased air-sea interaction. The opposite would be the case for late melting. Phytoplankton (a)—The early melting would provide increased irradiance for phytoplankton growth, altering bloom timing, composition and extent (D). Productivity would increase under these conditions. Conversely, productivity is likely to decrease if melting is late. There will also be indirect effects, e.g., on vertical flux (I). Zooplankton (b and c)—Phytoplankton dynamics markedly influence zooplankton dynamics. If the ice melt is too early or too late, this will affect food availability and phytoplankton-zooplankton coupling (I). The effects of melt timing on sea ice as a physical habitat are also important, for example this may affect essential overwintering habitat for zooplankton larvae and disrupt life history patterns (D). Fish and air-breathing predators (d and e)—Given the above effects, the timing of ice melt also influences the availability of prey (e.g., krill) for fish and higher predators (whales, seals and penguins) (I). Many Southern Ocean fish and higher predators are directly dependent on sea ice as a habitat for feeding, breeding and haul-out for which timing is crucial (D). Its presence or absence can also affect access to breeding and feeding grounds (D). While there will always be associated uncertainty, by reducing the range and thereby better representing sea ice (Figures 1–4), the more confident we can be in the model output, and in this case, in generating ecologically-relevant results (Figure 6). understanding of how to deal with these and other caveats for Antarctic krill (CCAMLR, 2015). The timing of regional (Supplementary Table 1) progresses within the climate modeling ice arrival, duration and retreat significantly influences habitat community, it will be important for those interested in climate availability, food type and availability, species distributions and change impacts on ecosystems to remain actively involved. vessel access (Supplementary Table 3). Furthermore, the MIZ is Challenges remain in Southern Ocean modeling, and many an area of great ecological importance, yet this and other detailed of these are common to areas other than the Antarctic (Murphy features such as eddies are very difficult to capture in climate et al., 2012; IPCC, 2013; Giorgi and Gutowski, 2015). Some of models due to their small spatial scale (Supplementary Table 1). these are particularly pressing for ecologists and those involved By beginning to collectively understand and address these in conservation and management decision-making. Due to the needs (and others documented in Supplementary Table 1), timescale of many ecological processes, projections of change for information from climate models can be more usefully applied, the next two to three decades are required (Trathan and Agnew, and future research priorities can be jointly determined and 2010). Such timescales are also important for those studying addressed. In the meantime, if the sea ice projections are to sea ice and other aspects of the climatology. However, on such be used, for example, to understand the changing dynamics of decadal time scales future change will be dominated by natural individual species at regional or even local spatial scales, then variability of the climate system which is challenging to predict reconciling the information from global climate models requires with the current climate modeling tools (Meehl et al., 2009; careful interpretation. However, in some situations, particularly O’Kane et al., 2013; Risbey et al., 2014). This results in a mismatch for more immediate (urgent) and high-level decision-making, in temporal scale between what the models can deliver and a high degree of complexity may not always be required. the relevant time-window for ecological considerations (Massom For example, the projections could be used to help identify and Stammerjohn, 2010; Macias et al., 2013; Supplementary Table where particular research efforts should be concentrated, or in 1). highlighting the main trends, at least in terms of direction and Given the variability in sea ice around the continent, and range of change. Large-scale projections can also help to focus the fact that many ecological processes, and conservation and more detailed studies. Smaller-scale regional studies and models management decisions, take place at relatively small spatial are invaluable and continue to be developed for the Southern scales, improving the regional performance of climate models Ocean (e.g., Pinkerton et al., 2010; Smith et al., 2014; Graham is another clear need (Supplementary Table 1). As a case in et al., 2016), and IPCC model projections may be considered point, sea ice is highly variable in the Antarctic Peninsula and as boundary forcing for downscaled, regional or local fields. Scotia Sea region (Murphy et al., 2014; Turner et al., 2016) Regional climate downscaling is a growing field of research, with which is also the location of the main commercial fishery CORDEX providing an example (Dosio et al., 2014; Katragkou Frontiers in Marine Science | www.frontiersin.org 7 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology FIGURE 5 | A simplified representation of a Southern Ocean pelagic food web, where the trophic levels are represented by phytoplankton (a), small zooplankton (copepods) (b), large zooplankton (krill) (c), fish (d), and air-breathing predators (seals, whales, penguins) (e). Sea ice exerts direct (D) (black arrow) and indirect (I) (dotted lines) effects at each trophic level. et al., 2015), although this needs to be carefully applied with an also Nicol et al., 2000; Thorpe et al., 2007; Atkinson et al., 2008; appreciation of its strengths and weaknesses (Grose et al., 2012; Loeb et al., 2008; Murphy et al., 2013; Melbourne-Thomas et al., Corney et al., 2013). 2016), large-scale information is also valuable. Similarly, regional In essence, a combination of large-scale and regional information is important in the design and implementation of information is key for the conservation and management of marine protected areas (MPAs)—this might include ensuring Antarctic marine resources. The circumpolar nature and high appropriate protection for vulnerable areas or those identified as connectivity of the Southern Ocean, including to the global likely to be most resilient to change - with large-scale information ocean, means that a large-scale view is crucial in understanding needed to ensure that regions are not considered in isolation. change (Murphy et al., 2008, 2012). Both regional and large-scale information is required for ecosystem-based management such Future Work as that of the Antarctic krill fishery in the Scotia Sea. Regional This work initiates an iterative process, alongside climate information is needed to provide an understanding of how model development for IPCC AR6, to provide model outputs harvested populations, dependent species, and access to fishing (projections) that can serve as a common resource for use in areas may change over time (CCAMLR, 2016). However, because ecological studies. This evaluation was concerned with processes conditions further afield are known to influence the population at the ocean surface whereas different life stages of species dynamics of krill and their predators (see Supplementary Table 3 can often be associated with deeper waters, thus our approach Frontiers in Marine Science | www.frontiersin.org 8 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology FIGURE 6 | Schematic illustrating that the deviation between observed (actual) and predicted state (in this case sea ice extent) has the potential to amplify as ecological complexity increases. The pale central area and white arrows represent reductions in uncertainty by using the best available models (in this case the eight-model subset). may not be optimal for whole system studies. In building a be predicted and appropriate conservation and management more comprehensive understanding of Southern Ocean change decisions can be made. it must also be recognized that physical and biogeochemical changes do not act in isolation (Gruber, 2011; Bopp et al., AUTHOR CONTRIBUTIONS 2013), although understanding and modeling the interactive and cumulative effects of multiple stressors presents a significant RC, EM, JT, TB, and NJ conceived the study; CK and TB analysed challenge (Murphy et al., 2012; Boyd et al., 2014; Gutt et al., 2015). the data; TB, RC, EM, and JT interpreted the results; RC led the Progress can now be made in extending this work to evaluate writing of the paper together with EM, JT, and TB; CK, SC, WS, model outputs for other key physical variables (e.g., winds, and CW contributed to the writing of the paper; all other authors various water mass properties, mixed layer depth, temperature, reviewed the manuscript; all authors approved the final version. and biogeochemical changes, see Supplementary Table 3) from an ecological perspective. Comparison of model results and performance with existing and emerging observations, ACKNOWLEDGMENTS data products and technology (e.g., new GLODAPv2 (Global Ocean Data Analysis Project Version 2), satellite products This paper builds on discussions that took place at a for chlorophyll, primary production, data from the Southern multidisciplinary workshop convened by the Integrating Climate Ocean Continuous Plankton Recorder and Southern Ocean and Ecosystem Dynamics in the Southern Ocean programme Observing System, FISH-MIP (Fisheries and Marine Ecosystem (ICED) and hosted by the British Antarctic Survey (BAS). Model Intercomparison Project) and others (Rintoul et al., We thank all the workshop participants. The study (and 2012; Constable et al., 2016; Olsen et al., 2016) will be an specifically RC, EM, NJ, JT, CK) was supported by ICED under essential aspect of future work. In addition, more research is a Natural Environment Research Council (NERC) International needed on the mechanisms that link the physical variables to Opportunities Fund Grant NE/I029943/1, together with NERC ecological processes. Beyond that there is a need to consider core funding to BAS. Additional workshop funding was provided ecological scenarios, including the effects of the recovery of by Integrated Marine Biosphere Research (IMBeR). SC and AC over-exploited baleen whale populations (Branch et al., 2004; were supported by the Australian Government’s Cooperative Noad et al., 2011; Tulloch et al., 2017) toward fully integrated Research Centres Programme through the Antarctic Climate and system understanding. Future fishery scenarios are also required Ecosystems Cooperative Research Centre. JX was supported by to explore how demand may change over time. the Investigator FCT program (IF/00616/2013) and this study We advocate the need for synergy between the climate benefited from the strategic program of MARE, financed by modeling community and a range of disciplines, including FCT (MARE- UID/MAR/04292/2013). We acknowledge Tony but certainly not limited to, marine ecologists, as state-of- Philips (BAS) for downloading and managing local copies the-art models continue to advance. We stress there is no of the required CMIP5 data. We acknowledge the World prescriptive method, rather a combination of approaches is Climate Research Programme’s Working Group on Coupled required. By refining large scale climate model output and Modelling (responsible for CMIP), and we thank the climate carefully using large-scale projections, together with specific modelling groups (Supplementary Table 4) for producing and regional downscaling, the likely impacts of change on biota can making their model output available. For CMIP the U.S. Frontiers in Marine Science | www.frontiersin.org 9 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology Department of Energy’s Program for Climate Model Diagnosis SUPPLEMENTARY MATERIAL and Intercomparison provides coordinating support and led The Supplementary Material for this article can be found development of software infrastructure in partnership with the online at: http://journal.frontiersin.org/article/10.3389/fmars. Global Organization for Earth System Science Portals. All images 2017.00308/full#supplementary-material belong to the British Antarctic Survey. REFERENCES Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, and P. M. Midgley (Cambridge; New York, NY: Cambridge University Press), 1029–1136. Atkinson, A., Siegel, V., Pakhomov, E. A., Rothery, P., Loeb, V., Ross, R. M., et al. Comiso, J. C. (2000). Bootstrap Sea Ice Concentrations from Nimbus-7 SMMR and (2008). Oceanic circumpolar habitats of Antarctic krill. Mar. Ecol. Prog. Ser. DMSP SSM/I-SSMIS. Version 2. Updated 2015. Boulder, CO: NASA DAAC at 362, 1–23. doi: 10.3354/meps07498 the National Snow and Ice Data Center. Barange, M., Merino, G., Blanchard, J. L., Scholtens, J., Harle, J., Allison, Constable, A. J., Costa, D. P., Schofield, O., Newman, L., Urban, R. D., Fulton, E. H., et al. (2014). Impacts of climate change on marine ecosystem E. A., et al. (2016). Developing priority variables (“ecosystem Essential Ocean production in societies dependent on fisheries. Nat. Clim. Chang. 4, 211–216. Variables” — eEOVs). for observing dynamics and change in Southern doi: 10.1038/nclimate2119 Ocean ecosystems. J. Mar. Syst. 161, 26–41. doi: 10.1016/j.jmarsys.2016. Bednarek, A. T., Cooper, A. B., Cresswell, K. A., Mangel, M., Satterthwaite, 05.003 W. H., Simpfendorfer, C. A., et al. (2011). The certainty of uncertainty in Constable, A. J., Melbourne-Thomas, J., Corney, S. P., Arrigo, K. R., Barbraud, C., marine conservation and what to do about it. Bull. Mar. Sci. 87, 177–195. Barnes, D. K., et al. (2014). Climate change and Southern Ocean ecosystems I: doi: 10.5343/bms.2010.1060 how changes in physical habitats directly affect marine biota. Glob. Chang. Biol. Blois, J. L., Zarnetske, P. L., Fitzpatrick, M. C., and Finnegan, S. (2013). Climate 20, 3004–3025. doi: 10.1111/gcb.12623 change and the past, present, and future of biotic interactions. Science 341, Corney, S., Grose, M., Bennett, J. C., White, C., Katzfey, J., McGregor, 499–504. doi: 10.1126/science.1237184 J., et al. (2013). Performance of downscaled regional climate simulations Böning, C. W., Dispert, A., Visbeck, M., Rintoul, S. R., and Schwarzkopf, F. U. using a variable-resolution regional climate model: Tasmania as a test (2008). The response of the Antarctic Circumpolar Current to recent climate case. J. Geophys. Res. Atmos. 118, 11936–11950. doi: 10.1002/2013JD0 change. Nat. Geosci. 1, 864–869. doi: 10.1038/ngeo362 20087 Bopp, L., Resplandy, L., Orr, J. C., Doney, S. C., Dunne, J. P., Gehlen, Doney, S. C., Ruckelshaus, M., Duffy, J. E., Barry, J. P., Chan, F., English, C. A., M., et al. (2013). Multiple stressors of ocean ecosystems in the 21st et al. (2012). Climate change impacts on marine ecosystems. Ann. Rev. Mar. century: projections with CMIP5 models. Biogeosciences 10, 6225–6245. Sci. 4, 11–37. doi: 10.1146/annurev-marine-041911-111611 doi: 10.5194/bg-10-6225-2013 Dosio, A., Panitz, H.-J., Schubert-Frisius, M., and Lüthi, D. (2014). Dynamical Boyd, P. W., Lennartz, S. T., Glover, D. M., and Doney, S. C. (2014). Biological downscaling of CMIP5 global circulation models over CORDEX-Africa ramifications of climate-change-mediated oceanic multi-stressors. Nat. Clim. with COSMO-CLM: evaluation over the present climate and analysis of Chang. 5, 71–79. doi: 10.1038/nclimate2441 the added value. Clim. Dyn. 44, 2637–2661. doi: 10.1007/s00382-014- Bracegirdle, T. J., Connolley, W. M., and Turner, J. (2008). Antarctic 2262-x climate change over the twenty first century. J. Geophys. Res. 113:D03103. Ducklow, H. W., Baker, K., Martinson, D. G., Quetin, L. B., Ross, R. M., Smith, doi: 10.1029/2007JD008933 R. C., et al. (2007). Marine pelagic ecosystems: the West Antarctic Peninsula. Bracegirdle, T. J., Stephenson, D. B., Turner, J., and Phillips, T. (2015). The Philos. Trans. R. Soc. Lond. B. Biol. Sci. 362, 67–94. doi: 10.1098/rstb.2006.1955 importance of sea ice area biases in 21st century multimodel projections of Ferreira, D., Marshall, J., Bitz, C. M., Solomon, S., and Plumb, A. (2015). Antarctic Antarctic temperature and precipitation. Geophys. Res. Lett. 42, 10832–10839. ocean and sea ice response to ozone depletion: a two-time-scale problem. J. doi: 10.1002/2015GL067055 Clim. 28, 1206–1226. doi: 10.1175/JCLI-D-14-00313.1 Branch, T. A., Matsuoka, K., and Miyashita, T. (2004). Evidence for increases Gagné, M. È., Gillett, N. P., and Fyfe, J. C. (2015). Observed and simulated changes in Antarctic Blue Whales based on Bayesian modelling. Mar. Mamm. Sci. 20, in Antarctic sea ice extent over the past 50 years. Geophys. Res. Lett. 42, 90–95. 726–754. doi: 10.1111/j.1748-7692.2004.tb01190.x doi: 10.1002/2014GL062231 CCAMLR (2009). CCAMLR Resolution 30/XXVII, 2009. Climate Change. Hobart, García Molinos, J., Halpern, B., Schoeman, D., Brown, C. J., Kiessling, W., Moore, TAS: CCAMLR. P. J., et al. (2015). Climate velocity and the future global redistribution of CCAMLR (2015). CCAMLR, Statistical Bulletin. Hobart, TAS: CCAMLR marine biodiversity. Nat. Clim. Chang. 6, 83–88. doi: 10.1038/nclimate2769 CCAMLR (2016). CCAMLR, 2016. Report of the Thirty-third meeting of the Giorgi, F., and Gutowski, W. J. (2015). Regional dynamical downscaling Scientific Committee (SC-CAMLR-XXXIII) CCAMLR, Hobart, Australia. and the CORDEX initiative. Annu. Rev. Environ. Resour. 40, 467–490. Chave, J. (2013). The problem of pattern and scale in ecology: what have we learned doi: 10.1146/annurev-environ-102014-021217 in 20 years? Ecol. Lett. 16 (Suppl 1), 4–16. doi: 10.1111/ele.12048 Graham, J. A., Dinniman, M. S., and Klinck, J. M. (2016). Impact of model Cheung, W. W., Watson, R., and Pauly, D. (2013). Signature of ocean warming in resolution for on-shelf heat transport along the West Antarctic Peninsula. J. global fisheries catch. Nature 497, 365–368. doi: 10.1038/nature12156 Geophys. Res. Oceans 121, 7880–7897. doi: 10.1002/2016JC011875 Clarke, A., Griffiths, H. J., Barnes, D. K. A., Meredith, M. P., and Grant, Grose, M. R., Corney, S. P., Katzfey, J. J., Bennett, J. C., Holz, G. K., White, C. S. M. (2009). Spatial variation in seabed temperatures in the Southern J., et al. (2012). A regional response in mean westerly circulation and rainfall to Ocean: implications for benthic ecology and biogeography. J. Geophys. Res. projected climate warming over Tasmania, Australia. Clim. Dyn. 40, 2035–2048. 114:G03003. doi: 10.1029/2008JG000886 doi: 10.1007/s00382-012-1405-1 Clarke, A., Johnston, N. M., Murphy, E. J., and Rogers, A. D. (2007). Introduction. Gruber, N. (2011). Warming up, turning sour, losing breath: ocean Antarctic ecology from genes to ecosystems: the impact of climate change and biogeochemistry under global change. Philos. Trans. A. Math. Phys. Eng. the importance of scale. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 362, 5–9. Sci. 369, 1980–1996. doi: 10.1098/rsta.2011.0003 doi: 10.1098/rstb.2006.1943 Gutt, J., Bertler, N., Bracegirdle, T. J., Buschmann, A., Comiso, J., Hosie, G., et al. Collins, M., Knutti, R., Arblaster, J., Dufresne, J. L., Fichefet, T., Friedlingstein, (2015). The Southern Ocean ecosystem under multiple climate change stresses– P., et al. (2013). “Long-term climate change: projections, commitments and an integrated circumpolar assessment. Glob. Chang. Biol. 21, 1434–1453. irreversibility,” in Climate Change 2013: The Physical Science Basis. Contribution doi: 10.1111/gcb.12794 of Working Group I to the Fifth Assessment Report of the Intergovernmental. Gutt, J., Zurell, D., Bracegridle, T., Cheung, W., Clark, M. S., Convey, P., et al. Panel on Climate, eds T. F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S. K. (2012). Correlative and dynamic species distribution modelling for ecological Frontiers in Marine Science | www.frontiersin.org 10 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology predictions in the Antarctic: a cross-disciplinary concept. Polar Res. 31:11091. Lima, F. P., Ribeiro, P. A., Queiroz, N., Hawkins, S. J., and Santos, A. doi: 10.3402/polar.v31i0.11091 M. (2007). Do distributional shifts of northern and southern species of Halpern, B. S., Frazier, M., Potapenko, J., Casey, K. S., Koenig, K., Longo, C., et al. algae match the warming pattern? Glob. Chang. Biol. 13, 2592–2604. (2015). Spatial and temporal changes in cumulative human impacts on the doi: 10.1111/j.1365-2486.2007.01451.x world’s ocean. Nat. Commun. 6:7615. doi: 10.1038/ncomms8615 Loeb, V. J., Hofmann, E. E., Klinck, J. M., Holm-Hansen, O., and White, W. Harris, R. M. B., Grose, M. R., Lee, G., Bindoff, N. L., Porfirio, L. L., and Fox- B. (2008). ENSO and variability of the Antarctic Peninsula pelagic marine Hughes, P. (2014). Climate projections for ecologists. Wiley Interdiscipl. Rev. ecosystem. Antarct. Sci. 21, 135. doi: 10.1017/S0954102008001636 Clim. Change 5, 621–637. doi: 10.1002/wcc.291 Macias, D., Garcia-Gorriz, E., and Stips, A. (2013). Understanding the Hauck, J., and Völker, C. (2015). Rising atmospheric CO leads to large impact causes of recent warming of mediterranean waters. How much could be of biology on Southern Ocean CO uptake via changes of the Revelle factor. attributed to climate change? PLoS ONE 8:e81591. doi: 10.1371/journal.pone. Geophys. Res. Lett. 42, 1459–1464. doi: 10.1002/2015GL063070 0081591 Hauck, J., Völker, C., Wang, T., Hoppema, M., Losch, M., and Wolf-Gladrow, Massom, R. A., and Stammerjohn, S. E. (2010). Antarctic sea ice change and D. A. (2013). Seasonally different carbon flux changes in the Southern Ocean variability – Physical and ecological implications. Polar Sci. 4, 149–186. in response to the southern annular mode. Glob. Biogeochem. Cycles 27, doi: 10.1016/j.polar.2010.05.001 1236–1245. doi: 10.1002/2013GB004600 Meehl, G. A., Goddard, L., Murphy, J., Stouffer, R. J., Boer, G., Danabasoglu, G., Helmuth, B., Mieszkowska, N., Moore, P., and Hawkins, S. J. (2006). Living et al. (2009). Decadal prediction. Bull. Ame. Meteorol. Soc. 90, 1467–1485. on the edge of two changing worlds: forecasting the responses of rocky doi: 10.1175/2009BAMS2778.1 intertidal ecosystems to climate change. Annu. Rev. Ecol. Evol. Syst. 37, 373–404. Meijers, A. J. (2014). The Southern ocean in the coupled model intercomparison doi: 10.1146/annurev.ecolsys.37.091305.110149 project phase 5. Philos. Trans. A Math. Phys. Eng. Sci. 372:20130296. Hickman, J. E. (2015). Carbon sequestration: biology’s growing role. Nat. Geosci. 8, doi: 10.1098/rsta.2013.0296 173–173. doi: 10.1038/ngeo2386 Meinshausen, M., Smith, S. J., Calvin, K., Daniel, J. S., Kainuma, M. L. T., Hill, S. L., Phillips, T., and Atkinson, A. (2013). Potential climate change effects Lamarque, J. F., et al. (2011). The RCP greenhouse gas concentrations on the habitat of antarctic krill in the weddell quadrant of the southern ocean. and their extensions from 1765 to 2300. Clim. Change 109, 213–241. PLoS ONE 8:e72246. doi: 10.1371/journal.pone.0072246 doi: 10.1007/s10584-011-0156-z Hoegh-Goldberg, O., and Bruno, J. F. (2010). The impact of climate Melbourne-Thomas, J., Constable, A., Wotherspoon, S., and Raymond, B. (2013). change on the world’s marine ecosystems. Science 328, 1523–1528. Testing paradigms of ecosystem change under climate warming in Antarctica. doi: 10.1126/science.1189930 PLoS ONE 8:e55093. doi: 10.1371/journal.pone.0055093 Hollowed, A. B., Barange, M., Beamish, R. J., Brander, K., Cochrane, K., Melbourne-Thomas, J., Corney, S. P., Trebilco, R., Meiners, K. M., Stevens, R. P., Drinkwater, K., et al. (2013). Projected impacts of climate change on marine Kawaguchi, S., et al. (2016). Under ice habitats for Antarctic krill larvae: could fish and fisheries. ICES J. Mar. Sci. 70, 1023–1037. doi: 10.1093/icesjms/fst081 less mean more under climate warming? Geophys. Res. Lett. 43, 10322–10327. Hunt, G. L., Drinkwater, K. F., Arrigo, K., Berge, J., Daly, K. L., Danielson, doi: 10.1002/2016GL070846 S., et al. (2016). Advection in polar and sub-polar environments: Mieszkowska, N., Kendall, M. A., Hawkins, S. J., Leaper, R., Williamson, P., impacts on high latitude marine ecosystems. Prog. Oceanogr. 149, 40–81. Hardman-Mountford, N. J., et al. (2006). Changes in the range of some doi: 10.1016/j.pocean.2016.10.004 common rocky shore species in Britain – a response to climate change? Hunter, C. M., Caswell, H., Runge, M. C., Regehr, E. V., Amstrup, S. C., and Hydrobiologia 555, 241–251. doi: 10.1007/s10750-005-1120-6 Stirling, I. (2010). Climate change threatens polar bear populations: a stochastic Murphy, E. J., Cavanagh, R. D., Hofmann, E. E., Hill, S. L., Constable, demographic analysis. Ecology 91, 2883–2897. doi: 10.1890/09-1641.1 A. J., Costa, D. P., et al. (2012). Developing integrated models of IPCC (2013). “Summary for policymakers,” in Climate Change 2013: The Physical Southern Ocean food webs: including ecological complexity, accounting Science Basis. Contribution of Working Group I to the Fifth Assessment Report of for uncertainty and the importance of scale. Prog. Oceanogr. 102, 74–92. the Intergovernmental Panel on Climate Change, eds T. F. Stocker, D. Qin, G.-K. doi: 10.1016/j.pocean.2012.03.006 Plattner, M. Tignor, S. K. Allen, J. Boschung, et al. (Cambridge, UK; New York, Murphy, E. J., Cavanagh, R. D., Johnston, N. M., Reid, K., and Hofmann, E. E. NY: Cambridge University Press), 3–29. (2008). Integrating Climate and Ecosystem Dynamics in the Southern Ocean Jenouvrier, S., Holland, M., Stroeve, J., Serreze, M., Barbraud, C., Weimerskirch, (ICED) Science Plan and Implementation Strategy. Plymouth; Bergen: GLOBEC; H., et al. (2014). Projected continent-wide declines of the emperor penguin IMBER. under climate change. Nat. Clim. Chang. 4, 715–718. doi: 10.1038/nclimate2280 Murphy, E. J., Clarke, A., Abram, N. J., and Turner, J. (2014). Variability of sea-ice Katragkou, E., García-Díez, M., Vautard, R., Sobolowski, S., Zanis, P., Alexandri, in the northern Weddell Sea during the 20th century. J. Geophys. Res. Oceans G., et al. (2015). Regional climate hindcast simulations within EURO- 119, 4549–4572. doi: 10.1002/2013JC009511 CORDEX: evaluation of a WRF multi-physics ensemble. Geoscientific Model Murphy, E. J., Hofmann, E. E., Watkins, J. L., Johnston, N. M., Pinones, Dev. 8, 603–618. doi: 10.5194/gmd-8-603-2015 A., Ballerini, T., et al. (2013). Comparison of the structure and function Kawaguchi, S., Ishida, A., King, R., Raymond, B., Waller, N., Constable, A., of Southern Ocean regional ecosystems: the Antarctic Peninsula and et al. (2013). Risk maps for Antarctic krill under projected Southern Ocean South Georgia. J. Mar. Syst. 109–110, 22–42. doi: 10.1016/j.jmarsys.2012. acidification. Nat. Clim. Chang. 3, 843–847. doi: 10.1038/nclimate1937 03.011 Kennicutt, M. C., and Chown, S. L. (2014). Six priorities for Antarctic science. Murphy, E. J., Watkins, J. L., Trathan, P. N., Reid, K., Meredith, M. P., Thorpe, S. Nature 512, 23–25. doi: 10.1038/512023a E., et al. (2007). Spatial and temporal operation of the Scotia Sea ecosystem: a Landschützer, P. N., Gruber, N., Haumann, F. A., Rodenbeck, C., Bakker, D. C. E., review of large-scale links in a krill centred food web. Philos. Trans. R. Soc. B. van Heuven, S., et al. (2015). The reinvigoration of the Southern Ocean carbon Lond. Biol. Sci. 362, 113–148. doi: 10.1098/rstb.2006.1957 sink. Science 349, 1221–1224. doi: 10.1126/science.aab2620 Nicol, S., Pauly, T., Bindoff, N., Wright, S., Thiele, D., Hosie, G. W., et al. (2000). Larsen, J. N., Anisimov, O. A., Constable, A., Hollowed, A. B., Maynard, N., Ocean circulation off east Antarctica affects ecosystem structure and sea-ice Prestrud, P., et al. (2014). “Part B: Regional Aspects. Contribution of Working extent. Nature 406, 504–507. doi: 10.1038/35020053 Group II to the Fifth Assessment Report of the Intergovernmental Panel Noad, M., Dunlop, R., Paton, D., and Kniest, H. (2011). “Abundance estimates of on Climate Change,” in Climate Change 2014: Impacts, Adaptation, and the east Australian humpback whale population: 2010 survey,” in Final Report Vulnerability, eds V. R. Barros, C. B. Field, D. J. Dokken, M. D. Mastrandrea, for the Australian Marine Mammal Centre (Australian Antarctic Division). K. J. Mach, T. E. Bilir, M. Chatterjee, K. L. Ebi, Y. O. Estrada, R. C. Genova, O’Kane, T. J., Matear, R. J., Chamberlain, M. A., Risbey, J. S., Sloyan, B. M., B. Girma, E. S. Kissel, A. N. Levy, S. MacCracken, P. R. Mastrandrea, and L. L. and Horenko, I. (2013). Decadal variability in an OGCM Southern Ocean: White (Cambridge; New York, NY: Cambridge University Press), 1567–1612. intrinsic modes, forced modes and metastable states. Ocean Model. 69, 1–21. Le Quere, C., Rodenbeck, C., Buitenhuis, E. T., Conway, T. J., Langenfelds, R., doi: 10.1016/j.ocemod.2013.04.009 Gomez, A., et al. (2007). Saturation of the southern ocean CO sink due to Olsen, A., Key, R. M., Van Heuven, S., Lauvset, S. K., Velo, A., Lin, X., et al. (2016). recent climate change. Science 316, 1735–1738. doi: 10.1126/science.1136188 The Global Ocean Data Analysis Project version 2 (GLODAPv2) – an internally Frontiers in Marine Science | www.frontiersin.org 11 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology consistent data product for the world ocean. Earth Syst. Sci. Data 8, 297–323. Thompson, D. W. J., Solomon, S., Kushner, P. J., England, M. H., Grise, doi: 10.5194/essd-8-297-2016 K. M., and Karoly, D. J. (2011). Signatures of the Antarctic ozone hole Passow, U., and Carlson, C. A. (2012). The biological pump in a high CO world. in Southern Hemisphere surface climate change. Nat. Geosci. 4, 741–749. Mar. Ecol. Prog. Ser. 470, 249–271. doi: 10.3354/meps09985 doi: 10.1038/ngeo1296 Peck, L. S. (2011). Organisms and responses to environmental change. Mar. Thorpe, S. E., Murphy, E. J., and Watkins, J. L. (2007). Circumpolar connections Genomics 4, 237–243. doi: 10.1016/j.margen.2011.07.001 between Antarctic krill (Euphausia superba Dana) populations: investigating Pinkerton, M. H., Bradford-Grieve, J. M., and Hanchet, S. M. (2010). A balanced the roles of ocean and sea ice transport. Deep Sea Res. Part I Oceanogr. Res. Pap. model of the food web of the Ross Sea, Antarctica. CCAMLR Sci. 17, 1–31. 54, 792–810. doi: 10.1016/j.dsr.2007.01.008 Piñones, A., and Fedorov, A. V. (2016). Projected changes of Antarctic krill Trathan, P. N., and Agnew, D. (2010). Climate change and the Antarctic marine habitat by the end of the 21st century. Geophys. Res. Lett. 43, 8580–8589. ecosystem: an essay on management implications. Antarct. Sci. 22, 387–398. doi: 10.1002/2016GL069656 doi: 10.1017/S0954102010000222 Rintoul, S. R., Sparrow, M., Meredith, M. P., Wadley, V., Speer, K., Hofmann, Tulloch, V. J. D., Plagányi, É. E., Matear, R., Brown, C. J., and Richardson, A. J. E., et al. (2012). The Southern Ocean Observing System: Initial Science and (2017). Ecosystem modelling to quantify the impact of historical whaling on Implementation Strategy. Cambridge, UK: SCAR. Southern Hemisphere baleen whales. Fish Fish. doi: 10.1111/faf.12241. [Epub Risbey, J. S., Lewandowsky, S., Langlais, C., Monselesan, D. P., O’Kane, ahead of print]. T. J., and Oreskes, N. (2014). Well-estimated global surface warming in Turner, J., Bracegirdle, T. J., Phillips, T., Marshall, G. J., and Hosking, J. S. (2013). climate projections selected for ENSO phase. Nat. Clim. Chang 4, 835–840. An initial assessment of Antarctic Sea ice extent in the CMIP5 models. J. Clim. doi: 10.1038/nclimate2310 26, 1473–1484. doi: 10.1175/JCLI-D-12-00068.1 Simmonds, I. (2015). Comparing and contrasting the behaviour of Arctic and Turner, J., Comiso, J. C., Marshall, G. J., Lachlan-Cope, T. A., Bracegirdle, T., Antarctic sea ice over the 35 year period 1979–2013. Ann. Glaciol. 56, 18–28. Maksym, T., et al. (2009). Non-annular atmospheric circulation change induced doi: 10.3189/2015AoG69A909 by stratospheric ozone depletion and its role in the recent increase of Antarctic Smith, W. O., Ainley, D. G., and Cattaneo-Vietti, R. (2007). Trophic interactions sea ice extent. Geophys. Res. Lett. 36:L08502. doi: 10.1029/2009GL037524 within the Ross Sea continental shelf ecosystem. Philos. Trans. R. Soc. B. Lond. Turner, J., Lu, H., White, I., King, J. C., Phillips, T., Hosking, J. S., et al. (2016). Biolol. Sci. 362, 95–111. doi: 10.1098/rstb.2006.1956 Absence of 21st century warming on Antarctic Peninsula consistent with Smith, W. O., Dinniman, M. S., Hofmann, E. E., and Klinck, J. M. (2014). natural variability. Nature 535, 411–415. doi: 10.1038/nature18645 The effects of changing winds and temperatures on the oceanography Wang, M., and Overland, J. E. (2009). A sea ice free summer Arctic within 30 years? of the Ross Sea in the 21st century. Geophys. Res. Lett. 41, 1624–1631. Geophys. Res. Lett. 36:L07502. doi: 10.1029/2009GL037820 doi: 10.1002/2014GL059311 Xavier, J. C., Brandt, A., Ropert-Coudert, Y., Badhe, R., Gutt, J., Havermans, C., Snover, A. K., Mantua, N. J., Littell, J. S., Alexander, M. A., McClure, M. M., and et al. (2016). Future challenges in Southern ocean ecology research. Front. Mar. Nye, J. (2013). Choosing and using climate-change scenarios for ecological- Sci. 3:94. doi: 10.3389/fmars.2016.00094 impact assessments and conservation decisions. Conserv. Biol. 27, 1147–1157. doi: 10.1111/cobi.12163 Conflict of Interest Statement: The authors declare that the research was Southward, A. J., Hawkins, S. J., and Burrows, M. T. (1995). Seventy years’ conducted in the absence of any commercial or financial relationships that could observations of changes in disrtibution and abundance of zooplankton and be construed as a potential conflict of interest. intertidal organisms in the western English Channel in relation to rising sea temperatures. J. Therm. Biol. 20, 127–155. doi: 10.1016/0306-4565(94)00043-I The handling Editor declared a shared affiliation, though no other collaboration, Stock, C. A., Alexander, M. A., Bond, N. A., Brander, K. M., Cheung, W., with one of the authors DW. Curchitser, E. N., et al. (2011). On the use of IPCC-class models to assess the impact of climate on Living Marine Resources. Prog. Oceanogr. 88, 1–27. Copyright © 2017 Cavanagh, Murphy, Bracegirdle, Turner, Knowland, Corney, doi: 10.1016/j.pocean.2010.09.001 Smith, Waluda, Johnston, Bellerby, Constable, Costa, Hofmann, Jackson, Staniland, Sydeman, W. J., Poloczanska, E., Reed, T. E., and Thompson, S. A. Wolf-Gladrow, Xavier. This is an open-access article distributed under the terms (2015). Climate change and marine vertebrates. Science 350, 772–777. of the Creative Commons Attribution License (CC BY). The use, distribution or doi: 10.1126/science.aac9874 reproduction in other forums is permitted, provided the original author(s) or licensor Taylor, K. E., Stouffer, R. J., and Meehl, G. A. (2012). An overview of are credited and that the original publication in this journal is cited, in accordance CMIP5 and the experiment design. Bull. Ame. Meteorol. Soc. 93, 485–498. with accepted academic practice. No use, distribution or reproduction is permitted doi: 10.1175/BAMS-D-11-00094.1 which does not comply with these terms. Frontiers in Marine Science | www.frontiersin.org 12 September 2017 | Volume 4 | Article 308 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Frontiers in Marine Science Unpaywall

Loading next page...
 
/lp/unpaywall/a-synergistic-approach-for-evaluating-climate-model-output-for-H90M1hLhnP

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Unpaywall
ISSN
2296-7745
DOI
10.3389/fmars.2017.00308
Publisher site
See Article on Publisher Site

Abstract

ORIGINAL RESEARCH published: 26 September 2017 doi: 10.3389/fmars.2017.00308 A Synergistic Approach for Evaluating Climate Model Output for Ecological Applications 1 1 1 1 Rachel D. Cavanagh *, Eugene J. Murphy , Thomas J. Bracegirdle , John Turner , 1† 2 3 1 Cheryl A. Knowland , Stuart P. Corney , Walker O. Smith, Jr. , Claire M. Waluda , 1 4, 5 2, 6 7 Nadine M. Johnston , Richard G. J. Bellerby , Andrew J. Constable , Daniel P. Costa , Edited by: 8 1 1 9 Eileen E. Hofmann , Jennifer A. Jackson , Iain J. Staniland , Dieter Wolf-Gladrow and Elvira S. Poloczanska, 1, 10 José C. Xavier Alfred-Wegener-Institut für Polar- und Meeresforschung, Germany 1 2 British Antarctic Survey, Cambridge, United Kingdom, Antarctic Climate and Ecosystems Cooperative Research Centre, Reviewed by: University of Tasmania, Hobart, TAS, Australia, Virginia Institute of Marine Science, College of William and Mary, Gloucester Diego M. Macias, Point, VA, United States, SKLEC-NIVA Centre for Marine and Coastal Research, East China Normal University, Shanghai, 5 6 European Commission. Joint China, Norwegian Institute for Water Research, Bergen, Norway, Australian Antarctic Division, Australian Commonwealth Research Center, Italy Department of Environment and Energy, Kingston, TAS, Australia, Department of Ecology and Evolutionary Biology, Nova Mieszkowska, University of California, Santa Cruz, Santa Cruz, CA, United States, Center for Coastal Physical Oceanography, Old Marine Biological Association of the Dominion University, Norfolk, VA, United States, Alfred-Wegener-Institut Helmholtz Zentrum für Polar- und United Kingdom, United Kingdom Meeresforschung, Bremerhaven, Germany, Departamento das Ciências da Vida, Marine and Environmental Sciences Centre, Universidade de Coimbra, Coimbra, Portugal *Correspondence: Rachel D. Cavanagh rcav@bas.ac.uk Increasing concern about the impacts of climate change on ecosystems is prompting Present Address: ecologists and ecosystem managers to seek reliable projections of physical drivers of Cheryl A. Knowland, change. The use of global climate models in ecology is growing, although drawing King’s College London, Graduate School, London, United Kingdom ecologically meaningful conclusions can be problematic. The expertise required to access and interpret output from climate and earth system models is hampering progress Specialty section: in utilizing them most effectively to determine the wider implications of climate change. To This article was submitted to Global Change and the Future Ocean, address this issue, we present a joint approach between climate scientists and ecologists a section of the journal that explores key challenges and opportunities for progress. As an exemplar, our focus Frontiers in Marine Science is the Southern Ocean, notable for significant change with global implications, and on Received: 13 June 2017 sea ice, given its crucial role in this dynamic ecosystem. We combined perspectives Accepted: 08 September 2017 Published: 26 September 2017 to evaluate the representation of sea ice in global climate models. With an emphasis Citation: on ecologically-relevant criteria (sea ice extent and seasonality) we selected a subset of Cavanagh RD, Murphy EJ, eight models that reliably reproduce extant sea ice distributions. While the model subset Bracegirdle TJ, Turner J, Knowland CA, Corney SP, shows a similar mean change to the full ensemble in sea ice extent (approximately 50% Smith WO Jr, Waluda CM, Johnston decline in winter and 30% decline in summer), there is a marked reduction in the range. NM, Bellerby RGJ, Constable AJ, Costa DP, Hofmann EE, Jackson JA, This improved the precision of projected future sea ice distributions by approximately Staniland IJ, Wolf-Gladrow D and one third, and means they are more amenable to ecological interpretation. We conclude Xavier JC (2017) A Synergistic that careful multidisciplinary evaluation of climate models, in conjunction with ongoing Approach for Evaluating Climate Model Output for Ecological modeling advances, should form an integral part of utilizing model output. Applications. Front. Mar. Sci. 4:308. doi: 10.3389/fmars.2017.00308 Keywords: IPCC, CMIP5, climate models, Southern Ocean, marine ecosystems, climate change, sea ice Frontiers in Marine Science | www.frontiersin.org 1 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology INTRODUCTION (Stock et al., 2011), and the time frames of human activities (such as fishing) and political decision-making. At these shorter Natural variability in the climate system and anthropogenic temporal scales it is difficult or impossible to distinguish between climate change result in a complex array of physical and natural variability and the background climate change signal biological responses. Marine ecosystems are inextricably (Macias et al., 2013). Furthermore, many key ecological processes connected to the climate system and significant changes to their occur at regional (i.e., tens to hundreds of kilometers in structure and function are both observed and expected (Doney extent) or smaller scales, hence biological responses to change et al., 2012; Blois et al., 2013; Sydeman et al., 2015). Effects also vary at these scales (Helmuth et al., 2006; Clarke et al., may be direct (e.g., temperature changes affecting physiological 2009; Peck, 2011; Chave, 2013). Similarly, resource conservation processes such as growth, reproduction, consumption and and management is also often concerned with regional or respiration), or indirect (e.g., those resulting from changes smaller scales (e.g., subareas, divisions or subdivisions in the to primary productivity, which in turn can influence species case of fishing areas) that may contain relatively discrete abundance, distributions and interactions; Constable et al., populations of certain species (Stock et al., 2011; Sydeman et al., 2014). Biological feedbacks to the climate system such as the 2015). role of biology in carbon sequestration are recognized (Hauck Here we focus on the Southern Ocean as an exemplar region and Völker, 2015; Hickman, 2015) although difficult to quantify for exploring the use of IPCC-class models in studies of ecological (Passow and Carlson, 2012). change. The region was the subject of much attention during Concern about the effects of climate change on marine IPCC Fifth Assessment Report (AR5), owing to its important ecosystems is growing (Hoegh-Goldberg and Bruno, 2010; role in the global climate system and the significant changes Cheung et al., 2013; Halpern et al., 2015) and is a major focus already observed (Le Quere et al., 2007; Böning et al., 2008; of global environmental research programmes and targets such Hauck et al., 2013; Kennicutt and Chown, 2014; Larsen et al., as Future Earth, the Scientific Committee on Oceanic Research 2014; Landschützer et al., 2015). One of the most notable changes (SCOR) and the United Nations Sustainable Development Goals. in this region has been the increase in westerly wind speeds Ecologists are increasingly being asked to provide advice in associated with stratospheric ozone depletion (Turner et al., 2009; this regard to support conservation and management decisions Thompson et al., 2011; Meijers, 2014). There is strong evidence (Hollowed et al., 2013; Barange et al., 2014; García Molinos that this has had important effects on the Southern Ocean, in et al., 2015). Understanding and predicting these effects requires particular sea ice (Ferreira et al., 2015). However, major issues knowledge of the processes that determine the distribution remain with regard to reproducing the observed physical state and abundance of species, the structure and functioning of of the Southern Ocean in global climate models, and these have ecosystems within which they occur, and their past and present implications for the reliability of predicted responses to future dynamics (Southward et al., 1995; Mieszkowska et al., 2006; Lima climate forcing (Turner et al., 2013; Meijers, 2014). et al., 2007; Doney et al., 2012; Murphy et al., 2012). This is Considerable progress has been made with qualitative coupled with the need to understand the key physical drivers assessments of the effects of physical changes on Southern of change and for reliable projections of them. As such, the Ocean ecosystems, revealing key drivers (including sea ice, use of Intergovernmental Panel on Climate Change (IPCC)- wind, various water mass properties and mixed layer depth) class climate models (Box 1) is rapidly gaining momentum in and complex responses throughout the food web (e.g., Murphy ecological studies of change (Hunter et al., 2010; Bopp et al., 2013; et al., 2007; Massom and Stammerjohn, 2010; Gutt et al., 2012; Jenouvrier et al., 2014; Piñones and Fedorov, 2016). Constable et al., 2014; Hunt et al., 2016; Xavier et al., 2016). Although these models are designed to provide realistic Broadly, the effects are likely to include impacts on habitat, representations of the climate system and projections of climate physiology, distribution, population densities, phenology, variables (e.g., atmospheric and ocean temperatures, sea ice and community interactions (Trathan and Agnew, 2010). extent and winds) that are known to influence ecological Quantitative assessments are more difficult. Relevant biological processes, applying them to ecological problems is challenging data are often patchy or imprecisely known, while effects are (Stock et al., 2011; Snover et al., 2013; Harris et al., 2014). often indirect and multifaceted (Bednarek et al., 2011; Murphy No model is perfect and none reproduce all current and past et al., 2012; Melbourne-Thomas et al., 2013; Gutt et al., 2015). climates, so projections need to be used with care. There are This makes the challenges of assessing future change even greater important caveats that users must consider and these issues (Hill et al., 2013; Kawaguchi et al., 2013; Melbourne-Thomas are likely to be compounded in the ecosystem context where et al., 2016). additional considerations, such as differences in spatial and Despite the challenges, quantitative assessments of the temporal scale, may be significant. Most experiments using effects of change are urgently required, not least because the IPCC-class models are designed to provide projections on time conservation and management of the Southern Ocean must scales of the order multiple decades (>30 years) and tend to account for the dual realities of climate-driven ecosystem change be run with spatial grids of around 200 km. On such temporal and growing demand for fishery resources (Murphy et al., 2008; scales, climate change signals should be clearly distinguishable CCAMLR, 2009; Trathan and Agnew, 2010; Hill et al., 2013). from natural variability. However, from an ecological perspective, Input from the scientific community is critical in providing shorter (<30 years) time frames are key due to biological policy-relevant information on climate change impacts for use in processes, including life cycles, generation lengths and phenology decision making. Frontiers in Marine Science | www.frontiersin.org 2 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology BOX 1 | IPCC-class climate models. By “IPCC-class models” we refer to full complexity coupled atmosphere-ocean-sea ice- climate and earth system models. This type of model is a fundamental tool for quantifying how the environment may change in the future. They are increasingly able to take account of complex processes, including the “ozone hole,” carbon uptake in the ocean, and atmospheric aerosols. Because of the complexity, non-linearity and small horizontal scale of many meteorological and oceanographic processes, these models require intensive computer processing power and are therefore run at only a few dozen climate research centres. They can be run with “pre-industrial” concentrations of greenhouse gases (GHG) before anthropogenic forcing is introduced (from mid-nineteenth century) to quantify the impacts of the known increase in greenhouse gas concentrations and the development of the ozone hole. In addition, possible trajectories of twenty- first century climate change can be derived from climate model simulations run under different (GHG) emission scenarios and recovery of stratospheric ozone amounts. Future GHG emissions will be the product of complex dynamic systems, determined by a combination of political decisions, demography, socio-economic development and technological change. Their evolution is highly uncertain and a range of plausible scenarios have been defined to provide a common set of future emission outcomes that can be used to help compare results across different climate models. The set of scenarios of future change in climate forcings that were used in the IPCC Fifth Assessment Report (AR5) are termed Representative Concentration Pathways (RCPs) (Meinshausen et al., 2011). There are four RCPs (RCP8.5 which corresponds to the pathway with the highest GHG emissions, followed by RCP6, RCP4.5, and RCP2.6), each one represents a different trajectory and cumulative emission concentration to 2100. A major initiative that provided much of the climate model data for the most recent IPCC report is the Coupled Model Intercomparison Project Phase 5 (CMIP5, managed by the World Climate Research Programme) (Taylor et al., 2012). The CMIP5 dataset is a comprehensive set of outputs from approximately 60 of the world’s most sophisticated climate models from major climate modeling centres (such collections are often referred to as climate model ensembles). The CMIP model intercomparisons have historically been synchronized with the preparation phases of IPCC reports, hence the models (and model setups) are often referred to as “IPCC-class” models. The most up-to-date climate models are currently being prepared for CMIP6 in advance of the IPCC Sixth Assessment Report (AR6), which is scheduled to be released around 2021-2022. IPCC-class models are essential tools for quantitatively of the Southern Ocean sea ice extent (SIE) cycle (e.g., late- integrating knowledge of the climate system and making summer minimum and amplitude of the annual cycle), we used projections. There are general recommendations for ecologists selection criteria following the methodology of the Arctic sea on their use (e.g., Snover et al., 2013; Harris et al., 2014), ice analysis of Wang and Overland (2009), adjusted to reflect but information specifically for climate scientists about the the timing of austral seasons. This specifies that simulated requirements of ecologists is lacking. Furthermore, while sea ice must fall within ±20% of satellite data for mean generic guidelines are useful, detail specific to particular minimum SIE and mean seasonality (where seasonality is regions is also important, and expert-agreed benchmarks the annual difference between the maximum and minimum would be valuable. Here, with our focus on the Southern SIE). Ocean, and on sea ice, given its crucial role in this dynamic system (Box 2), we ask how these models can be used most effectively to understand, and manage, the responses Data of species, communities and ecosystems to change. Our For the analysis presented here monthly-mean sea ice study represents a first step in bringing members of the concentration data was used (CMIP5 variable name “sic”), Southern Ocean climate and ecological community together to which is the proportion of each model grid cell that is covered by jointly explore key challenges and consider opportunities for sea ice. Output from two different types of model experiments progress. were retrieved. The CMIP5 data can be downloaded from the Earth System Grid https://pcmdi.llnl.gov/projects/esgf- MATERIALS AND METHODS llnl/. Firstly “historical” simulations, which are run with observed concentrations of greenhouse gases and other Capabilities and Limitations of IPCC-class known important climate drivers between the mid-nineteenth Climate Models century to the present. Second, global warming simulations The first step in this study was the compilation of a set of following a scenario of high emissions of greenhouse gases key questions that Southern Ocean ecologists would ideally like (RCP8.5). Future climate conditions are taken from the late to address through the use of climate models. To begin to twenty-first century in the RCP8.5 simulations (2079-2099) consider the challenges of using IPCC-class models to address and these are compared against baseline (i.e., as a reference such questions, we focused in detail on one important driver, sea period from which to define twenty-first century change and ice, owing to its crucial role in the Southern Ocean (Box 2) and over which to compare observations against climate model identified properties of particular ecological significance. output) conditions taken from the period 1979-1999 in the historical simulations. After first eliminating the models for Sea Ice in CMIP5 Models which both monthly sea ice concentration data and data for the We evaluated the representation of Southern Hemisphere sea ice RCP8.5 high emissions scenario were not available, a total of in CMIP5 models with an emphasis on extent and seasonality 35 different CMIP5 models were identified (see below). These due to the crucial ecological roles of these properties (see baseline conditions were also used for comparing models and below). To assess which models most closely reproduced aspects observations in the sub-setting procedure. The dataset used was Frontiers in Marine Science | www.frontiersin.org 3 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology BOX 2 | Southern Hemisphere sea ice. Southern Hemisphere sea ice has a range of influences in the Earth System including on global heat and carbon cycles, and sea-level rise, and is a sensitive indicator of change in the Polar Regions; therefore, understanding its variability in the climate system is critical. It forms one of the largest, most dynamic marine ecosystems on Earth, with marine biota being highly adapted to its presence, seasonality and properties over evolutionary timescales (Clarke et al., 2007; Massom and Stammerjohn, 2 2 2010; Constable et al., 2014). Overall, sea ice extent around the Antarctic ranges from a late winter peak of ∼19 million km to a minimum of ∼3–4 million km in summer (Massom and Stammerjohn, 2010). Since the late 1970s the annual mean total Southern Hemisphere sea ice extent has increased by ∼3%, albeit with considerable regional contrast and variability (Turner et al., 2013; Gagné et al., 2015; Simmonds, 2015). However, as the impacts of shorter term variability (associated with natural variability and ozone depletion) recede over longer time frames, this overall increasing trend is projected to reverse in direction (Bracegirdle et al., 2008; Turner et al., 2013). the Bootstrap version 2 product described in Comiso (2000; Supplementary Table 3 also includes information for other key updated 2015). drivers, detailing aspects of these that are particularly important ecologically, providing a baseline for future work. Sea Ice Diagnostics and CMIP5 Model Selection Sea Ice in CMIP5 Models The basic parameter used for evaluation was total Southern The criteria we applied reduced the available 35 models Hemisphere sea ice extent (SIE), which was calculated from (Supplementary Table 4) (referred to hereafter as the full monthly mean sea ice concentration data (both the CMIP5 and ensemble) to a subset of eight models (referred to hereafter as the satellite Bootstrap 2 datasets). It was defined as the area enclosed subset). The importance of this sub-setting is presented below. by the 15% concentration contour of the sea ice concentration fields over the Southern Hemisphere. Effect of Sub-Setting on Historical and Future Further details are provided in Supplementary Material CMIP5-Derived Sea Ice Distributions Methods. Late twentieth century sea ice distributions from the full ensemble are shown both for austral summer (December- February) (Figure 1A) and winter (June-August) (Figure 2A). RESULTS These highlight the large ranges in the summer and winter Capabilities and Limitations of IPCC-class distributions in the full ensemble for the 1979-99 period. Large ranges are also seen in projected sea ice distributions in the late Climate Models Considering the key questions identified by Southern Ocean twenty-first century (Figures 1B, 2B). The inter-model range of sea ice in the historical runs is ecologists alongside current capabilities and limitations of the models, revealed fundamental differences in perspectives and much smaller for the subset (Figures 3A, 4A). This is expected since those with large deviations have been removed. However, approaches between the two disciplines. These include methods; research interests and priorities; requirements of the models Figures 3B, 4B illustrate that this narrower range is maintained in the projected sea ice spatial distribution in the late twenty- (Supplementary Table 1); and terminology (Supplementary Table 2). Issues around temporal and spatial scales were noted in first century. This emphasizes that model-projected future sea ice distributions (and associated variables) are highly dependent particular, as was the need to resolve detailed features such as the Marginal Ice Zone (MIZ) (i.e., the transition area between open on the match between simulated and observed climatological SIE conditions (Risbey et al., 2014; Bracegirdle et al., 2015). ocean and sea ice). In terms of ensemble mean change in sea ice concentration, Sea Ice As an Exemplar Variable it is evident that the subset gives more clearly defined regions Focusing on sea ice as an exemplar variable, information was of ice reduction with larger changes in many regions. This is compiled on properties of particular ecological relevance. Sea particularly apparent in winter (Figures 2C, 4C). These regional ice forms an essential surface habitat for resting, breeding and differences between the full ensemble and subset would likely feeding, and sub-surface habitat for food and refugia, it also be more distinct in evaluations of sea ice at specific locations or influences water column properties, and affects the reproductive within specific sectors. cycles, recruitment and foraging behavior of a wide range of The mean twenty-first century change in SIE is similar both species. It plays a pivotal role in Southern Ocean biogeochemical in the full ensemble and the subset, with respective changes of 49 cycles, and influences fisheries, not only through its impact on and 52% in summer and 31% for both in the winter. However, the the distribution and potentially the abundance of target species, range in projected CMIP5 SIE change is smaller across models but by affecting the access of vessels to fishing grounds. It is that more closely reproduce the observed SIE climatology, with a clear that the seasonal advance and retreat of sea ice (timing standard deviation from 12 to 8% in winter and a reduction for and extent) is a major driver of the structure and functioning of the summer from 27 to 19% (Supplementary Table 5). It should the Southern Ocean pelagic ecosystem, and this formed the basis be noted that proportional changes are not as strongly influenced for our evaluation of the models (see Materials and Methods). by weighting, since, for example, models with too little sea ice Supplementary Table 3 summarizes the information collated on produce smaller absolute changes that are a similar proportion of sea ice. Recognizing the range of multiple stressors involved, their own too-small SIE. Frontiers in Marine Science | www.frontiersin.org 4 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology FIGURE 1 | Austral summer (DJF) sea ice distribution for all available 35 models from historical simulations over the period 1979-1999 (A) and RCP8.5 (2079-2099) (B) and the RCP8.5-historical difference (C). On each plot the filled contours show the all-model ensemble mean sea ice concentration, the red lines show the ice edge for the model with most ice and the green lines show the ice edge for the model with least ice. The black lines on (A) show the observed extent (for 1979-1999 period) from the Comiso Bootstrap 2 dataset (see Materials and Methods). FIGURE 2 | As Figure 1 above, but for winter (JJA). DISCUSSION biased projection) but, not sufficient (i.e., an accurate baseline climatology will not necessarily produce an accurate projection Sea Ice in IPCC-class Models of future change) condition for producing reliable projections The first step in the study highlighted the importance of of future change. Despite the large differences between observed understanding the nature and capabilities of models and the and simulated climatological sea ice extent in many climate need for careful communication and interpretation of their models (Turner et al., 2013), previous projections for Southern outputs. Crucially, we have demonstrated the importance of Hemisphere sea ice have either treated all CMIP models equally model evaluation as a means to improve projections of change. (Collins et al., 2013) or used weightings based on other variables For both communities, it is imperative that models reproduce (Bracegirdle et al., 2008). the historical climate satisfactorily. In the case of sea ice this is Our analysis shows that applying criteria, motivated by a necessary (i.e., a biased baseline climatology will produce a ecological considerations, to select a subset of the available Frontiers in Marine Science | www.frontiersin.org 5 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology FIGURE 3 | As Figure 1 above, but for the subset of 8 models. FIGURE 4 | As Figure 2 above, but for the subset of 8 models. CMIP5 models dramatically reduces the range in projected subset and associated projected sea ice changes presented above. late-twenty-first century sea ice distribution and twenty-first However, interpretation of these must be accompanied by an minus twentieth century absolute sea ice extent change. By understanding of the associated uncertainty and caveats. Key better representing Southern Hemisphere sea ice extent and issues include the subjectivity in defining thresholds for sub- seasonality, the subset provides more ecologically meaningful setting (e.g., the 20% threshold used in this study), and the results (Figures 1–4). Box 3 captures some of the potential possibility that the “good” models share common biases and consequences for projecting ecosystem change based on poorly “get the right answer for the wrong reason.” Nevertheless, it is represented sea ice extent and seasonality. clear that a necessary condition for capturing realistic changes in ice-edge environments would be a satisfactory representation of Key Challenges and Recommendations past and current conditions (Bracegirdle et al., 2015). The sub- Therefore, as a first order assessment of the utility of IPCC-class setting technique used here improved projections by taking this climate models in Southern Ocean ecosystem research, and in into account. The difficulty comes in determining the reliability particular sea ice, we propose that ecologists use the eight-model of changes exhibited in sub-sets of “better” models. As the Frontiers in Marine Science | www.frontiersin.org 6 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology BOX 3 | Implications of using models that poorly represent Southern Hemisphere sea ice to project ecological responses to change. Efforts to incorporate biology into physical models tend to include only components of the lower levels (e.g., phytoplankton) (Murphy et al., 2012). However, physical changes can affect all trophic levels. Sea ice has complex multi-directional effects throughout the food web that influence the structure, function and dynamics of Southern Ocean ecosystems (Figure 5, Supplementary Table 3). These can be direct effects (D) as with the provision of crucial habitat and its impact as a physical barrier. They can also be indirect (I) through effects on physical conditions such as upper ocean temperature, irradiance and vertical mixing that in turn influence key ecological processes, food type and availability, and species composition, distributions and abundance. To capture some of the consequences of using models that poorly represent sea ice here we present a simplified view of some of its key effects in a Southern Ocean pelagic food web (for examples of detailed Southern Ocean food webs see, e.g., Ducklow et al., 2007; Murphy et al., 2007, 2012, 2013; Smith et al., 2007). Consider a model(s) that has sea ice melting a month early or a month late. The details of confounding factors and interactions aside, poorly-representing sea ice (in this case, melting too early or too late) will significantly impact phytoplankton and hence the overall productivity of the system: Physical factors—One possible (greatly simplified) outcome is that ice melting too early exposes the water column to increased light, as well as increased air-sea interaction. The opposite would be the case for late melting. Phytoplankton (a)—The early melting would provide increased irradiance for phytoplankton growth, altering bloom timing, composition and extent (D). Productivity would increase under these conditions. Conversely, productivity is likely to decrease if melting is late. There will also be indirect effects, e.g., on vertical flux (I). Zooplankton (b and c)—Phytoplankton dynamics markedly influence zooplankton dynamics. If the ice melt is too early or too late, this will affect food availability and phytoplankton-zooplankton coupling (I). The effects of melt timing on sea ice as a physical habitat are also important, for example this may affect essential overwintering habitat for zooplankton larvae and disrupt life history patterns (D). Fish and air-breathing predators (d and e)—Given the above effects, the timing of ice melt also influences the availability of prey (e.g., krill) for fish and higher predators (whales, seals and penguins) (I). Many Southern Ocean fish and higher predators are directly dependent on sea ice as a habitat for feeding, breeding and haul-out for which timing is crucial (D). Its presence or absence can also affect access to breeding and feeding grounds (D). While there will always be associated uncertainty, by reducing the range and thereby better representing sea ice (Figures 1–4), the more confident we can be in the model output, and in this case, in generating ecologically-relevant results (Figure 6). understanding of how to deal with these and other caveats for Antarctic krill (CCAMLR, 2015). The timing of regional (Supplementary Table 1) progresses within the climate modeling ice arrival, duration and retreat significantly influences habitat community, it will be important for those interested in climate availability, food type and availability, species distributions and change impacts on ecosystems to remain actively involved. vessel access (Supplementary Table 3). Furthermore, the MIZ is Challenges remain in Southern Ocean modeling, and many an area of great ecological importance, yet this and other detailed of these are common to areas other than the Antarctic (Murphy features such as eddies are very difficult to capture in climate et al., 2012; IPCC, 2013; Giorgi and Gutowski, 2015). Some of models due to their small spatial scale (Supplementary Table 1). these are particularly pressing for ecologists and those involved By beginning to collectively understand and address these in conservation and management decision-making. Due to the needs (and others documented in Supplementary Table 1), timescale of many ecological processes, projections of change for information from climate models can be more usefully applied, the next two to three decades are required (Trathan and Agnew, and future research priorities can be jointly determined and 2010). Such timescales are also important for those studying addressed. In the meantime, if the sea ice projections are to sea ice and other aspects of the climatology. However, on such be used, for example, to understand the changing dynamics of decadal time scales future change will be dominated by natural individual species at regional or even local spatial scales, then variability of the climate system which is challenging to predict reconciling the information from global climate models requires with the current climate modeling tools (Meehl et al., 2009; careful interpretation. However, in some situations, particularly O’Kane et al., 2013; Risbey et al., 2014). This results in a mismatch for more immediate (urgent) and high-level decision-making, in temporal scale between what the models can deliver and a high degree of complexity may not always be required. the relevant time-window for ecological considerations (Massom For example, the projections could be used to help identify and Stammerjohn, 2010; Macias et al., 2013; Supplementary Table where particular research efforts should be concentrated, or in 1). highlighting the main trends, at least in terms of direction and Given the variability in sea ice around the continent, and range of change. Large-scale projections can also help to focus the fact that many ecological processes, and conservation and more detailed studies. Smaller-scale regional studies and models management decisions, take place at relatively small spatial are invaluable and continue to be developed for the Southern scales, improving the regional performance of climate models Ocean (e.g., Pinkerton et al., 2010; Smith et al., 2014; Graham is another clear need (Supplementary Table 1). As a case in et al., 2016), and IPCC model projections may be considered point, sea ice is highly variable in the Antarctic Peninsula and as boundary forcing for downscaled, regional or local fields. Scotia Sea region (Murphy et al., 2014; Turner et al., 2016) Regional climate downscaling is a growing field of research, with which is also the location of the main commercial fishery CORDEX providing an example (Dosio et al., 2014; Katragkou Frontiers in Marine Science | www.frontiersin.org 7 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology FIGURE 5 | A simplified representation of a Southern Ocean pelagic food web, where the trophic levels are represented by phytoplankton (a), small zooplankton (copepods) (b), large zooplankton (krill) (c), fish (d), and air-breathing predators (seals, whales, penguins) (e). Sea ice exerts direct (D) (black arrow) and indirect (I) (dotted lines) effects at each trophic level. et al., 2015), although this needs to be carefully applied with an also Nicol et al., 2000; Thorpe et al., 2007; Atkinson et al., 2008; appreciation of its strengths and weaknesses (Grose et al., 2012; Loeb et al., 2008; Murphy et al., 2013; Melbourne-Thomas et al., Corney et al., 2013). 2016), large-scale information is also valuable. Similarly, regional In essence, a combination of large-scale and regional information is important in the design and implementation of information is key for the conservation and management of marine protected areas (MPAs)—this might include ensuring Antarctic marine resources. The circumpolar nature and high appropriate protection for vulnerable areas or those identified as connectivity of the Southern Ocean, including to the global likely to be most resilient to change - with large-scale information ocean, means that a large-scale view is crucial in understanding needed to ensure that regions are not considered in isolation. change (Murphy et al., 2008, 2012). Both regional and large-scale information is required for ecosystem-based management such Future Work as that of the Antarctic krill fishery in the Scotia Sea. Regional This work initiates an iterative process, alongside climate information is needed to provide an understanding of how model development for IPCC AR6, to provide model outputs harvested populations, dependent species, and access to fishing (projections) that can serve as a common resource for use in areas may change over time (CCAMLR, 2016). However, because ecological studies. This evaluation was concerned with processes conditions further afield are known to influence the population at the ocean surface whereas different life stages of species dynamics of krill and their predators (see Supplementary Table 3 can often be associated with deeper waters, thus our approach Frontiers in Marine Science | www.frontiersin.org 8 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology FIGURE 6 | Schematic illustrating that the deviation between observed (actual) and predicted state (in this case sea ice extent) has the potential to amplify as ecological complexity increases. The pale central area and white arrows represent reductions in uncertainty by using the best available models (in this case the eight-model subset). may not be optimal for whole system studies. In building a be predicted and appropriate conservation and management more comprehensive understanding of Southern Ocean change decisions can be made. it must also be recognized that physical and biogeochemical changes do not act in isolation (Gruber, 2011; Bopp et al., AUTHOR CONTRIBUTIONS 2013), although understanding and modeling the interactive and cumulative effects of multiple stressors presents a significant RC, EM, JT, TB, and NJ conceived the study; CK and TB analysed challenge (Murphy et al., 2012; Boyd et al., 2014; Gutt et al., 2015). the data; TB, RC, EM, and JT interpreted the results; RC led the Progress can now be made in extending this work to evaluate writing of the paper together with EM, JT, and TB; CK, SC, WS, model outputs for other key physical variables (e.g., winds, and CW contributed to the writing of the paper; all other authors various water mass properties, mixed layer depth, temperature, reviewed the manuscript; all authors approved the final version. and biogeochemical changes, see Supplementary Table 3) from an ecological perspective. Comparison of model results and performance with existing and emerging observations, ACKNOWLEDGMENTS data products and technology (e.g., new GLODAPv2 (Global Ocean Data Analysis Project Version 2), satellite products This paper builds on discussions that took place at a for chlorophyll, primary production, data from the Southern multidisciplinary workshop convened by the Integrating Climate Ocean Continuous Plankton Recorder and Southern Ocean and Ecosystem Dynamics in the Southern Ocean programme Observing System, FISH-MIP (Fisheries and Marine Ecosystem (ICED) and hosted by the British Antarctic Survey (BAS). Model Intercomparison Project) and others (Rintoul et al., We thank all the workshop participants. The study (and 2012; Constable et al., 2016; Olsen et al., 2016) will be an specifically RC, EM, NJ, JT, CK) was supported by ICED under essential aspect of future work. In addition, more research is a Natural Environment Research Council (NERC) International needed on the mechanisms that link the physical variables to Opportunities Fund Grant NE/I029943/1, together with NERC ecological processes. Beyond that there is a need to consider core funding to BAS. Additional workshop funding was provided ecological scenarios, including the effects of the recovery of by Integrated Marine Biosphere Research (IMBeR). SC and AC over-exploited baleen whale populations (Branch et al., 2004; were supported by the Australian Government’s Cooperative Noad et al., 2011; Tulloch et al., 2017) toward fully integrated Research Centres Programme through the Antarctic Climate and system understanding. Future fishery scenarios are also required Ecosystems Cooperative Research Centre. JX was supported by to explore how demand may change over time. the Investigator FCT program (IF/00616/2013) and this study We advocate the need for synergy between the climate benefited from the strategic program of MARE, financed by modeling community and a range of disciplines, including FCT (MARE- UID/MAR/04292/2013). We acknowledge Tony but certainly not limited to, marine ecologists, as state-of- Philips (BAS) for downloading and managing local copies the-art models continue to advance. We stress there is no of the required CMIP5 data. We acknowledge the World prescriptive method, rather a combination of approaches is Climate Research Programme’s Working Group on Coupled required. By refining large scale climate model output and Modelling (responsible for CMIP), and we thank the climate carefully using large-scale projections, together with specific modelling groups (Supplementary Table 4) for producing and regional downscaling, the likely impacts of change on biota can making their model output available. For CMIP the U.S. Frontiers in Marine Science | www.frontiersin.org 9 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology Department of Energy’s Program for Climate Model Diagnosis SUPPLEMENTARY MATERIAL and Intercomparison provides coordinating support and led The Supplementary Material for this article can be found development of software infrastructure in partnership with the online at: http://journal.frontiersin.org/article/10.3389/fmars. Global Organization for Earth System Science Portals. All images 2017.00308/full#supplementary-material belong to the British Antarctic Survey. REFERENCES Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, and P. M. Midgley (Cambridge; New York, NY: Cambridge University Press), 1029–1136. Atkinson, A., Siegel, V., Pakhomov, E. A., Rothery, P., Loeb, V., Ross, R. M., et al. Comiso, J. C. (2000). Bootstrap Sea Ice Concentrations from Nimbus-7 SMMR and (2008). Oceanic circumpolar habitats of Antarctic krill. Mar. Ecol. Prog. Ser. DMSP SSM/I-SSMIS. Version 2. Updated 2015. Boulder, CO: NASA DAAC at 362, 1–23. doi: 10.3354/meps07498 the National Snow and Ice Data Center. Barange, M., Merino, G., Blanchard, J. L., Scholtens, J., Harle, J., Allison, Constable, A. J., Costa, D. P., Schofield, O., Newman, L., Urban, R. D., Fulton, E. H., et al. (2014). Impacts of climate change on marine ecosystem E. A., et al. (2016). Developing priority variables (“ecosystem Essential Ocean production in societies dependent on fisheries. Nat. Clim. Chang. 4, 211–216. Variables” — eEOVs). for observing dynamics and change in Southern doi: 10.1038/nclimate2119 Ocean ecosystems. J. Mar. Syst. 161, 26–41. doi: 10.1016/j.jmarsys.2016. Bednarek, A. T., Cooper, A. B., Cresswell, K. A., Mangel, M., Satterthwaite, 05.003 W. H., Simpfendorfer, C. A., et al. (2011). The certainty of uncertainty in Constable, A. J., Melbourne-Thomas, J., Corney, S. P., Arrigo, K. R., Barbraud, C., marine conservation and what to do about it. Bull. Mar. Sci. 87, 177–195. Barnes, D. K., et al. (2014). Climate change and Southern Ocean ecosystems I: doi: 10.5343/bms.2010.1060 how changes in physical habitats directly affect marine biota. Glob. Chang. Biol. Blois, J. L., Zarnetske, P. L., Fitzpatrick, M. C., and Finnegan, S. (2013). Climate 20, 3004–3025. doi: 10.1111/gcb.12623 change and the past, present, and future of biotic interactions. Science 341, Corney, S., Grose, M., Bennett, J. C., White, C., Katzfey, J., McGregor, 499–504. doi: 10.1126/science.1237184 J., et al. (2013). Performance of downscaled regional climate simulations Böning, C. W., Dispert, A., Visbeck, M., Rintoul, S. R., and Schwarzkopf, F. U. using a variable-resolution regional climate model: Tasmania as a test (2008). The response of the Antarctic Circumpolar Current to recent climate case. J. Geophys. Res. Atmos. 118, 11936–11950. doi: 10.1002/2013JD0 change. Nat. Geosci. 1, 864–869. doi: 10.1038/ngeo362 20087 Bopp, L., Resplandy, L., Orr, J. C., Doney, S. C., Dunne, J. P., Gehlen, Doney, S. C., Ruckelshaus, M., Duffy, J. E., Barry, J. P., Chan, F., English, C. A., M., et al. (2013). Multiple stressors of ocean ecosystems in the 21st et al. (2012). Climate change impacts on marine ecosystems. Ann. Rev. Mar. century: projections with CMIP5 models. Biogeosciences 10, 6225–6245. Sci. 4, 11–37. doi: 10.1146/annurev-marine-041911-111611 doi: 10.5194/bg-10-6225-2013 Dosio, A., Panitz, H.-J., Schubert-Frisius, M., and Lüthi, D. (2014). Dynamical Boyd, P. W., Lennartz, S. T., Glover, D. M., and Doney, S. C. (2014). Biological downscaling of CMIP5 global circulation models over CORDEX-Africa ramifications of climate-change-mediated oceanic multi-stressors. Nat. Clim. with COSMO-CLM: evaluation over the present climate and analysis of Chang. 5, 71–79. doi: 10.1038/nclimate2441 the added value. Clim. Dyn. 44, 2637–2661. doi: 10.1007/s00382-014- Bracegirdle, T. J., Connolley, W. M., and Turner, J. (2008). Antarctic 2262-x climate change over the twenty first century. J. Geophys. Res. 113:D03103. Ducklow, H. W., Baker, K., Martinson, D. G., Quetin, L. B., Ross, R. M., Smith, doi: 10.1029/2007JD008933 R. C., et al. (2007). Marine pelagic ecosystems: the West Antarctic Peninsula. Bracegirdle, T. J., Stephenson, D. B., Turner, J., and Phillips, T. (2015). The Philos. Trans. R. Soc. Lond. B. Biol. Sci. 362, 67–94. doi: 10.1098/rstb.2006.1955 importance of sea ice area biases in 21st century multimodel projections of Ferreira, D., Marshall, J., Bitz, C. M., Solomon, S., and Plumb, A. (2015). Antarctic Antarctic temperature and precipitation. Geophys. Res. Lett. 42, 10832–10839. ocean and sea ice response to ozone depletion: a two-time-scale problem. J. doi: 10.1002/2015GL067055 Clim. 28, 1206–1226. doi: 10.1175/JCLI-D-14-00313.1 Branch, T. A., Matsuoka, K., and Miyashita, T. (2004). Evidence for increases Gagné, M. È., Gillett, N. P., and Fyfe, J. C. (2015). Observed and simulated changes in Antarctic Blue Whales based on Bayesian modelling. Mar. Mamm. Sci. 20, in Antarctic sea ice extent over the past 50 years. Geophys. Res. Lett. 42, 90–95. 726–754. doi: 10.1111/j.1748-7692.2004.tb01190.x doi: 10.1002/2014GL062231 CCAMLR (2009). CCAMLR Resolution 30/XXVII, 2009. Climate Change. Hobart, García Molinos, J., Halpern, B., Schoeman, D., Brown, C. J., Kiessling, W., Moore, TAS: CCAMLR. P. J., et al. (2015). Climate velocity and the future global redistribution of CCAMLR (2015). CCAMLR, Statistical Bulletin. Hobart, TAS: CCAMLR marine biodiversity. Nat. Clim. Chang. 6, 83–88. doi: 10.1038/nclimate2769 CCAMLR (2016). CCAMLR, 2016. Report of the Thirty-third meeting of the Giorgi, F., and Gutowski, W. J. (2015). Regional dynamical downscaling Scientific Committee (SC-CAMLR-XXXIII) CCAMLR, Hobart, Australia. and the CORDEX initiative. Annu. Rev. Environ. Resour. 40, 467–490. Chave, J. (2013). The problem of pattern and scale in ecology: what have we learned doi: 10.1146/annurev-environ-102014-021217 in 20 years? Ecol. Lett. 16 (Suppl 1), 4–16. doi: 10.1111/ele.12048 Graham, J. A., Dinniman, M. S., and Klinck, J. M. (2016). Impact of model Cheung, W. W., Watson, R., and Pauly, D. (2013). Signature of ocean warming in resolution for on-shelf heat transport along the West Antarctic Peninsula. J. global fisheries catch. Nature 497, 365–368. doi: 10.1038/nature12156 Geophys. Res. Oceans 121, 7880–7897. doi: 10.1002/2016JC011875 Clarke, A., Griffiths, H. J., Barnes, D. K. A., Meredith, M. P., and Grant, Grose, M. R., Corney, S. P., Katzfey, J. J., Bennett, J. C., Holz, G. K., White, C. S. M. (2009). Spatial variation in seabed temperatures in the Southern J., et al. (2012). A regional response in mean westerly circulation and rainfall to Ocean: implications for benthic ecology and biogeography. J. Geophys. Res. projected climate warming over Tasmania, Australia. Clim. Dyn. 40, 2035–2048. 114:G03003. doi: 10.1029/2008JG000886 doi: 10.1007/s00382-012-1405-1 Clarke, A., Johnston, N. M., Murphy, E. J., and Rogers, A. D. (2007). Introduction. Gruber, N. (2011). Warming up, turning sour, losing breath: ocean Antarctic ecology from genes to ecosystems: the impact of climate change and biogeochemistry under global change. Philos. Trans. A. Math. Phys. Eng. the importance of scale. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 362, 5–9. Sci. 369, 1980–1996. doi: 10.1098/rsta.2011.0003 doi: 10.1098/rstb.2006.1943 Gutt, J., Bertler, N., Bracegirdle, T. J., Buschmann, A., Comiso, J., Hosie, G., et al. Collins, M., Knutti, R., Arblaster, J., Dufresne, J. L., Fichefet, T., Friedlingstein, (2015). The Southern Ocean ecosystem under multiple climate change stresses– P., et al. (2013). “Long-term climate change: projections, commitments and an integrated circumpolar assessment. Glob. Chang. Biol. 21, 1434–1453. irreversibility,” in Climate Change 2013: The Physical Science Basis. Contribution doi: 10.1111/gcb.12794 of Working Group I to the Fifth Assessment Report of the Intergovernmental. Gutt, J., Zurell, D., Bracegridle, T., Cheung, W., Clark, M. S., Convey, P., et al. Panel on Climate, eds T. F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S. K. (2012). Correlative and dynamic species distribution modelling for ecological Frontiers in Marine Science | www.frontiersin.org 10 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology predictions in the Antarctic: a cross-disciplinary concept. Polar Res. 31:11091. Lima, F. P., Ribeiro, P. A., Queiroz, N., Hawkins, S. J., and Santos, A. doi: 10.3402/polar.v31i0.11091 M. (2007). Do distributional shifts of northern and southern species of Halpern, B. S., Frazier, M., Potapenko, J., Casey, K. S., Koenig, K., Longo, C., et al. algae match the warming pattern? Glob. Chang. Biol. 13, 2592–2604. (2015). Spatial and temporal changes in cumulative human impacts on the doi: 10.1111/j.1365-2486.2007.01451.x world’s ocean. Nat. Commun. 6:7615. doi: 10.1038/ncomms8615 Loeb, V. J., Hofmann, E. E., Klinck, J. M., Holm-Hansen, O., and White, W. Harris, R. M. B., Grose, M. R., Lee, G., Bindoff, N. L., Porfirio, L. L., and Fox- B. (2008). ENSO and variability of the Antarctic Peninsula pelagic marine Hughes, P. (2014). Climate projections for ecologists. Wiley Interdiscipl. Rev. ecosystem. Antarct. Sci. 21, 135. doi: 10.1017/S0954102008001636 Clim. Change 5, 621–637. doi: 10.1002/wcc.291 Macias, D., Garcia-Gorriz, E., and Stips, A. (2013). Understanding the Hauck, J., and Völker, C. (2015). Rising atmospheric CO leads to large impact causes of recent warming of mediterranean waters. How much could be of biology on Southern Ocean CO uptake via changes of the Revelle factor. attributed to climate change? PLoS ONE 8:e81591. doi: 10.1371/journal.pone. Geophys. Res. Lett. 42, 1459–1464. doi: 10.1002/2015GL063070 0081591 Hauck, J., Völker, C., Wang, T., Hoppema, M., Losch, M., and Wolf-Gladrow, Massom, R. A., and Stammerjohn, S. E. (2010). Antarctic sea ice change and D. A. (2013). Seasonally different carbon flux changes in the Southern Ocean variability – Physical and ecological implications. Polar Sci. 4, 149–186. in response to the southern annular mode. Glob. Biogeochem. Cycles 27, doi: 10.1016/j.polar.2010.05.001 1236–1245. doi: 10.1002/2013GB004600 Meehl, G. A., Goddard, L., Murphy, J., Stouffer, R. J., Boer, G., Danabasoglu, G., Helmuth, B., Mieszkowska, N., Moore, P., and Hawkins, S. J. (2006). Living et al. (2009). Decadal prediction. Bull. Ame. Meteorol. Soc. 90, 1467–1485. on the edge of two changing worlds: forecasting the responses of rocky doi: 10.1175/2009BAMS2778.1 intertidal ecosystems to climate change. Annu. Rev. Ecol. Evol. Syst. 37, 373–404. Meijers, A. J. (2014). The Southern ocean in the coupled model intercomparison doi: 10.1146/annurev.ecolsys.37.091305.110149 project phase 5. Philos. Trans. A Math. Phys. Eng. Sci. 372:20130296. Hickman, J. E. (2015). Carbon sequestration: biology’s growing role. Nat. Geosci. 8, doi: 10.1098/rsta.2013.0296 173–173. doi: 10.1038/ngeo2386 Meinshausen, M., Smith, S. J., Calvin, K., Daniel, J. S., Kainuma, M. L. T., Hill, S. L., Phillips, T., and Atkinson, A. (2013). Potential climate change effects Lamarque, J. F., et al. (2011). The RCP greenhouse gas concentrations on the habitat of antarctic krill in the weddell quadrant of the southern ocean. and their extensions from 1765 to 2300. Clim. Change 109, 213–241. PLoS ONE 8:e72246. doi: 10.1371/journal.pone.0072246 doi: 10.1007/s10584-011-0156-z Hoegh-Goldberg, O., and Bruno, J. F. (2010). The impact of climate Melbourne-Thomas, J., Constable, A., Wotherspoon, S., and Raymond, B. (2013). change on the world’s marine ecosystems. Science 328, 1523–1528. Testing paradigms of ecosystem change under climate warming in Antarctica. doi: 10.1126/science.1189930 PLoS ONE 8:e55093. doi: 10.1371/journal.pone.0055093 Hollowed, A. B., Barange, M., Beamish, R. J., Brander, K., Cochrane, K., Melbourne-Thomas, J., Corney, S. P., Trebilco, R., Meiners, K. M., Stevens, R. P., Drinkwater, K., et al. (2013). Projected impacts of climate change on marine Kawaguchi, S., et al. (2016). Under ice habitats for Antarctic krill larvae: could fish and fisheries. ICES J. Mar. Sci. 70, 1023–1037. doi: 10.1093/icesjms/fst081 less mean more under climate warming? Geophys. Res. Lett. 43, 10322–10327. Hunt, G. L., Drinkwater, K. F., Arrigo, K., Berge, J., Daly, K. L., Danielson, doi: 10.1002/2016GL070846 S., et al. (2016). Advection in polar and sub-polar environments: Mieszkowska, N., Kendall, M. A., Hawkins, S. J., Leaper, R., Williamson, P., impacts on high latitude marine ecosystems. Prog. Oceanogr. 149, 40–81. Hardman-Mountford, N. J., et al. (2006). Changes in the range of some doi: 10.1016/j.pocean.2016.10.004 common rocky shore species in Britain – a response to climate change? Hunter, C. M., Caswell, H., Runge, M. C., Regehr, E. V., Amstrup, S. C., and Hydrobiologia 555, 241–251. doi: 10.1007/s10750-005-1120-6 Stirling, I. (2010). Climate change threatens polar bear populations: a stochastic Murphy, E. J., Cavanagh, R. D., Hofmann, E. E., Hill, S. L., Constable, demographic analysis. Ecology 91, 2883–2897. doi: 10.1890/09-1641.1 A. J., Costa, D. P., et al. (2012). Developing integrated models of IPCC (2013). “Summary for policymakers,” in Climate Change 2013: The Physical Southern Ocean food webs: including ecological complexity, accounting Science Basis. Contribution of Working Group I to the Fifth Assessment Report of for uncertainty and the importance of scale. Prog. Oceanogr. 102, 74–92. the Intergovernmental Panel on Climate Change, eds T. F. Stocker, D. Qin, G.-K. doi: 10.1016/j.pocean.2012.03.006 Plattner, M. Tignor, S. K. Allen, J. Boschung, et al. (Cambridge, UK; New York, Murphy, E. J., Cavanagh, R. D., Johnston, N. M., Reid, K., and Hofmann, E. E. NY: Cambridge University Press), 3–29. (2008). Integrating Climate and Ecosystem Dynamics in the Southern Ocean Jenouvrier, S., Holland, M., Stroeve, J., Serreze, M., Barbraud, C., Weimerskirch, (ICED) Science Plan and Implementation Strategy. Plymouth; Bergen: GLOBEC; H., et al. (2014). Projected continent-wide declines of the emperor penguin IMBER. under climate change. Nat. Clim. Chang. 4, 715–718. doi: 10.1038/nclimate2280 Murphy, E. J., Clarke, A., Abram, N. J., and Turner, J. (2014). Variability of sea-ice Katragkou, E., García-Díez, M., Vautard, R., Sobolowski, S., Zanis, P., Alexandri, in the northern Weddell Sea during the 20th century. J. Geophys. Res. Oceans G., et al. (2015). Regional climate hindcast simulations within EURO- 119, 4549–4572. doi: 10.1002/2013JC009511 CORDEX: evaluation of a WRF multi-physics ensemble. Geoscientific Model Murphy, E. J., Hofmann, E. E., Watkins, J. L., Johnston, N. M., Pinones, Dev. 8, 603–618. doi: 10.5194/gmd-8-603-2015 A., Ballerini, T., et al. (2013). Comparison of the structure and function Kawaguchi, S., Ishida, A., King, R., Raymond, B., Waller, N., Constable, A., of Southern Ocean regional ecosystems: the Antarctic Peninsula and et al. (2013). Risk maps for Antarctic krill under projected Southern Ocean South Georgia. J. Mar. Syst. 109–110, 22–42. doi: 10.1016/j.jmarsys.2012. acidification. Nat. Clim. Chang. 3, 843–847. doi: 10.1038/nclimate1937 03.011 Kennicutt, M. C., and Chown, S. L. (2014). Six priorities for Antarctic science. Murphy, E. J., Watkins, J. L., Trathan, P. N., Reid, K., Meredith, M. P., Thorpe, S. Nature 512, 23–25. doi: 10.1038/512023a E., et al. (2007). Spatial and temporal operation of the Scotia Sea ecosystem: a Landschützer, P. N., Gruber, N., Haumann, F. A., Rodenbeck, C., Bakker, D. C. E., review of large-scale links in a krill centred food web. Philos. Trans. R. Soc. B. van Heuven, S., et al. (2015). The reinvigoration of the Southern Ocean carbon Lond. Biol. Sci. 362, 113–148. doi: 10.1098/rstb.2006.1957 sink. Science 349, 1221–1224. doi: 10.1126/science.aab2620 Nicol, S., Pauly, T., Bindoff, N., Wright, S., Thiele, D., Hosie, G. W., et al. (2000). Larsen, J. N., Anisimov, O. A., Constable, A., Hollowed, A. B., Maynard, N., Ocean circulation off east Antarctica affects ecosystem structure and sea-ice Prestrud, P., et al. (2014). “Part B: Regional Aspects. Contribution of Working extent. Nature 406, 504–507. doi: 10.1038/35020053 Group II to the Fifth Assessment Report of the Intergovernmental Panel Noad, M., Dunlop, R., Paton, D., and Kniest, H. (2011). “Abundance estimates of on Climate Change,” in Climate Change 2014: Impacts, Adaptation, and the east Australian humpback whale population: 2010 survey,” in Final Report Vulnerability, eds V. R. Barros, C. B. Field, D. J. Dokken, M. D. Mastrandrea, for the Australian Marine Mammal Centre (Australian Antarctic Division). K. J. Mach, T. E. Bilir, M. Chatterjee, K. L. Ebi, Y. O. Estrada, R. C. Genova, O’Kane, T. J., Matear, R. J., Chamberlain, M. A., Risbey, J. S., Sloyan, B. M., B. Girma, E. S. Kissel, A. N. Levy, S. MacCracken, P. R. Mastrandrea, and L. L. and Horenko, I. (2013). Decadal variability in an OGCM Southern Ocean: White (Cambridge; New York, NY: Cambridge University Press), 1567–1612. intrinsic modes, forced modes and metastable states. Ocean Model. 69, 1–21. Le Quere, C., Rodenbeck, C., Buitenhuis, E. T., Conway, T. J., Langenfelds, R., doi: 10.1016/j.ocemod.2013.04.009 Gomez, A., et al. (2007). Saturation of the southern ocean CO sink due to Olsen, A., Key, R. M., Van Heuven, S., Lauvset, S. K., Velo, A., Lin, X., et al. (2016). recent climate change. Science 316, 1735–1738. doi: 10.1126/science.1136188 The Global Ocean Data Analysis Project version 2 (GLODAPv2) – an internally Frontiers in Marine Science | www.frontiersin.org 11 September 2017 | Volume 4 | Article 308 Cavanagh et al. Evaluating Climate Models for Ecology consistent data product for the world ocean. Earth Syst. Sci. Data 8, 297–323. Thompson, D. W. J., Solomon, S., Kushner, P. J., England, M. H., Grise, doi: 10.5194/essd-8-297-2016 K. M., and Karoly, D. J. (2011). Signatures of the Antarctic ozone hole Passow, U., and Carlson, C. A. (2012). The biological pump in a high CO world. in Southern Hemisphere surface climate change. Nat. Geosci. 4, 741–749. Mar. Ecol. Prog. Ser. 470, 249–271. doi: 10.3354/meps09985 doi: 10.1038/ngeo1296 Peck, L. S. (2011). Organisms and responses to environmental change. Mar. Thorpe, S. E., Murphy, E. J., and Watkins, J. L. (2007). Circumpolar connections Genomics 4, 237–243. doi: 10.1016/j.margen.2011.07.001 between Antarctic krill (Euphausia superba Dana) populations: investigating Pinkerton, M. H., Bradford-Grieve, J. M., and Hanchet, S. M. (2010). A balanced the roles of ocean and sea ice transport. Deep Sea Res. Part I Oceanogr. Res. Pap. model of the food web of the Ross Sea, Antarctica. CCAMLR Sci. 17, 1–31. 54, 792–810. doi: 10.1016/j.dsr.2007.01.008 Piñones, A., and Fedorov, A. V. (2016). Projected changes of Antarctic krill Trathan, P. N., and Agnew, D. (2010). Climate change and the Antarctic marine habitat by the end of the 21st century. Geophys. Res. Lett. 43, 8580–8589. ecosystem: an essay on management implications. Antarct. Sci. 22, 387–398. doi: 10.1002/2016GL069656 doi: 10.1017/S0954102010000222 Rintoul, S. R., Sparrow, M., Meredith, M. P., Wadley, V., Speer, K., Hofmann, Tulloch, V. J. D., Plagányi, É. E., Matear, R., Brown, C. J., and Richardson, A. J. E., et al. (2012). The Southern Ocean Observing System: Initial Science and (2017). Ecosystem modelling to quantify the impact of historical whaling on Implementation Strategy. Cambridge, UK: SCAR. Southern Hemisphere baleen whales. Fish Fish. doi: 10.1111/faf.12241. [Epub Risbey, J. S., Lewandowsky, S., Langlais, C., Monselesan, D. P., O’Kane, ahead of print]. T. J., and Oreskes, N. (2014). Well-estimated global surface warming in Turner, J., Bracegirdle, T. J., Phillips, T., Marshall, G. J., and Hosking, J. S. (2013). climate projections selected for ENSO phase. Nat. Clim. Chang 4, 835–840. An initial assessment of Antarctic Sea ice extent in the CMIP5 models. J. Clim. doi: 10.1038/nclimate2310 26, 1473–1484. doi: 10.1175/JCLI-D-12-00068.1 Simmonds, I. (2015). Comparing and contrasting the behaviour of Arctic and Turner, J., Comiso, J. C., Marshall, G. J., Lachlan-Cope, T. A., Bracegirdle, T., Antarctic sea ice over the 35 year period 1979–2013. Ann. Glaciol. 56, 18–28. Maksym, T., et al. (2009). Non-annular atmospheric circulation change induced doi: 10.3189/2015AoG69A909 by stratospheric ozone depletion and its role in the recent increase of Antarctic Smith, W. O., Ainley, D. G., and Cattaneo-Vietti, R. (2007). Trophic interactions sea ice extent. Geophys. Res. Lett. 36:L08502. doi: 10.1029/2009GL037524 within the Ross Sea continental shelf ecosystem. Philos. Trans. R. Soc. B. Lond. Turner, J., Lu, H., White, I., King, J. C., Phillips, T., Hosking, J. S., et al. (2016). Biolol. Sci. 362, 95–111. doi: 10.1098/rstb.2006.1956 Absence of 21st century warming on Antarctic Peninsula consistent with Smith, W. O., Dinniman, M. S., Hofmann, E. E., and Klinck, J. M. (2014). natural variability. Nature 535, 411–415. doi: 10.1038/nature18645 The effects of changing winds and temperatures on the oceanography Wang, M., and Overland, J. E. (2009). A sea ice free summer Arctic within 30 years? of the Ross Sea in the 21st century. Geophys. Res. Lett. 41, 1624–1631. Geophys. Res. Lett. 36:L07502. doi: 10.1029/2009GL037820 doi: 10.1002/2014GL059311 Xavier, J. C., Brandt, A., Ropert-Coudert, Y., Badhe, R., Gutt, J., Havermans, C., Snover, A. K., Mantua, N. J., Littell, J. S., Alexander, M. A., McClure, M. M., and et al. (2016). Future challenges in Southern ocean ecology research. Front. Mar. Nye, J. (2013). Choosing and using climate-change scenarios for ecological- Sci. 3:94. doi: 10.3389/fmars.2016.00094 impact assessments and conservation decisions. Conserv. Biol. 27, 1147–1157. doi: 10.1111/cobi.12163 Conflict of Interest Statement: The authors declare that the research was Southward, A. J., Hawkins, S. J., and Burrows, M. T. (1995). Seventy years’ conducted in the absence of any commercial or financial relationships that could observations of changes in disrtibution and abundance of zooplankton and be construed as a potential conflict of interest. intertidal organisms in the western English Channel in relation to rising sea temperatures. J. Therm. Biol. 20, 127–155. doi: 10.1016/0306-4565(94)00043-I The handling Editor declared a shared affiliation, though no other collaboration, Stock, C. A., Alexander, M. A., Bond, N. A., Brander, K. M., Cheung, W., with one of the authors DW. Curchitser, E. N., et al. (2011). On the use of IPCC-class models to assess the impact of climate on Living Marine Resources. Prog. Oceanogr. 88, 1–27. Copyright © 2017 Cavanagh, Murphy, Bracegirdle, Turner, Knowland, Corney, doi: 10.1016/j.pocean.2010.09.001 Smith, Waluda, Johnston, Bellerby, Constable, Costa, Hofmann, Jackson, Staniland, Sydeman, W. J., Poloczanska, E., Reed, T. E., and Thompson, S. A. Wolf-Gladrow, Xavier. This is an open-access article distributed under the terms (2015). Climate change and marine vertebrates. Science 350, 772–777. of the Creative Commons Attribution License (CC BY). The use, distribution or doi: 10.1126/science.aac9874 reproduction in other forums is permitted, provided the original author(s) or licensor Taylor, K. E., Stouffer, R. J., and Meehl, G. A. (2012). An overview of are credited and that the original publication in this journal is cited, in accordance CMIP5 and the experiment design. Bull. Ame. Meteorol. Soc. 93, 485–498. with accepted academic practice. No use, distribution or reproduction is permitted doi: 10.1175/BAMS-D-11-00094.1 which does not comply with these terms. Frontiers in Marine Science | www.frontiersin.org 12 September 2017 | Volume 4 | Article 308

Journal

Frontiers in Marine ScienceUnpaywall

Published: Sep 26, 2017

There are no references for this article.