Abstract
Climate change is modifying global biogeochemical cycles, and is expected to exert increasingly large effects in the future. How these changes will affect and interact with the structure and function of particular ecosystems is unclear, both because of scientific uncertainties and the very diversity of global vegetation models in use. Writing in ERL, Warszawski et al (2013 Environ. Res. Lett. 8 044018) aggregate results from a group of models, across a range of emissions scenarios and climate data, to investigate these risks. Although the models frequently disagree about which specific regions are at risk, they consistently predict a greater chance of ecosystem restructuring with more warming; this risk roughly doubles between a 2 and 3 ° C increase in global mean temperature. The innovative work of Warszawski et al represents an important first step towards fully consistent multi-model, multi-scenario assessments of the future risks to global ecosystems.
Export citation and abstract BibTeX RIS
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
We know that climate change is altering global biogeochemical cycles, and will likely continue to do so throughout the 21st century [1]. But how will these changes affect the structure and functioning of global ecosystems and the services they provide? Many ecosystem- to regional-scale studies have examined this question, looking for example at the effects of changing fire regime on forest succession [2] or what local species might 'win' in a changing climate [3].
For larger-scale assessments, researchers often turn to global vegetation models (GVMs). GVMs typically abstract the immense diversity of global vegetation, in both form and function, into a small number of functional types or growth strategies (e.g. [4]). Such groupings allow researchers to simulate the effects of climate change on ecosystems and the services they provide, and simulate how biogeochemical fluxes from the terrestrial biosphere might feed back to the climate system. A recurring challenge, however, is to understand which results arise from a model's particular assumptions, structure, or omissions, a problem compounded by the wide diversity among GVMs [5].
This is where Warszawski et al [7] break new ground. Aggregating GVM results across a range of models, scenarios, and driver data, they examine how the models, on average, predict biogeochemical shifts and thus ecological transformations. Such multi-model evaluations have been used extensively in the climate modeling community and typically provide more robust predictions than do individual models [6]. The 'biogeochemical shifts' examined here include carbon stocks and fluxes, water fluxes, and species composition, all normalized and combined into a single metric Γ (gamma). This Γ is spatially and temporally explicit and, the authors argue, can be regarded as correlated with the risk of ecosystem restructuring.
Warszawski et al find that the risk of ecosystem disruption rises with warming, roughly doubling as global temperature rises from 2 to 3 ° C above 1980–2010 levels. This increased risk is consistent across GVMs and driven largely by carbon fluxes, except in the Amazon, where water flux changes are prominent as well. Not all the GVMs were dynamic—i.e., allowed for changes in vegetation composition during a model run—but among those that were, northern boreal forests and high-altitude ecosystems were found to be at particular risk of vegetation change. This finding is consistent with other recent model results projecting widespread vegetation change across the Arctic by the 2050s under a range of climate change and plant dispersal scenarios [8]. A significant fraction of terrestrial ecosystems was found to be at risk by the end of the 21st century. Finally, the uncertainty in Γ arising from the GVMs themselves was about twice as large as the uncertainty in Γ from the underlying climate models.
How robust is the fundamental assumption made by Warszawski et al, that biogeochemical shifts imply inevitable changes in species composition, trophic structures, etc? The authors are careful and conservative here, employing a space-for-time substitution to test Γ, and arguing that while it may not be a perfect indicator of ecosystem state shifts, should certainty signal risk of change based on what we understand about how plant productivity and water fluxes structure ecosystems. In any event, one of their most important results—that GVMs have much more spatial and temporal uncertainty than the climate models driving them—is independent of this assumption, and consistent with the AgMIP intercomparison of agricultural models [9]. This implies that ecosystem risk projections will benefit from greater consistency across models in the treatment of biogeochemical processes and how they respond to rising temperature and CO2.
In summary, the analysis by Warszawski et al illuminates the risks projected by current vegetation models; where these models are consistent; and, critically, where they differ. It also complements other large-scale efforts to quantify climate change effects on biodiversity and conservation priorities [10, 11]. This study provides a powerful argument about the importance of identifying uncertainty sources, and the need for greater consistency across GVM biogeochemical processes. This in turn will enable future, more powerful, multi-model studies of risks in the future interactions between climate and global ecosystems.
Acknowledgment
BB-L was supported by the Terrestrial Ecosystem Science program at the US Department of Energy.