Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Meeting the Global Ecosystem Collapse Challenge

Meeting the Global Ecosystem Collapse Challenge Assessing risks of ecosystem collapseGlobally, anthropogenic pressures are forcing ecosystems outside of safe operating spaces (sensu Rockström et al. ), increasing their likelihood of collapse. The IUCN aims to assess the risk of collapse of all of the world's ecosystems by 2025, using the Red List of Ecosystems (RLE) assessment framework (Keith et al. ). Ecosystem collapse (see Table for glossary) is a hot topic in ecology, receiving increasing treatment in the scientific and popular literature – with several recent articles reviewing potential early warning indicators of collapse (Dakos et al. ; Kéfi et al. ; Scheffer et al. ; see Table for glossary and examples of early warning indicators). We address the broad overarching question: Is ecosystem collapse predictable? The importance of answering this question is underscored by the fact that the concept is now formally codified under five criteria (A‐E) in the IUCN RLE (Box ), with Criterion E specifically seeking a quantification of collapse risk (Keith et al. ). This significantly raises the stakes in identifying and refining tractable methods to predict collapse, especially because collapse assessments inform listings in threat categories along a continuum from Least Concern to Critically Endangered according to the risks that they face (Keith et al. ; IUCN ). Here, we explored the challenges associated with predicting ecosystem collapse, providing commentary on the considerable implications those challenges have on numerically predicting collapse under Criterion E of the IUCN RLE. We conduct a formal systematic review to provide a detailed assessment of the ability to predict ecosystem collapse – and identify six key concerns and unresolved issues (stemming from the findings of our review) that must be addressed for researchers and practitioners to promote assessments of ecosystem collapse risk based on robust and objective predictions of ecosystem collapse. Robust predictions of ecosystem collapse are important if Criterion E of the IUCN RLE is to be successfully established and rigorously applied.Glossary of termsTerminologyDefinitionReferenceEarly warning indicatorA general term for dynamic patterns (“signals”) in ecosystem behavior that precede ecosystem collapse, e.g., increasing variance, increasing autocorrelation, or the presence of “flickering” in temporal or spatial data; see Kéfi et al. () and Dakos et al. () for a full summary of temporal and spatial indicatorsBoettiger et al. ()EcosystemA dynamic complex of plant, animal, and microorganism communities and their nonliving environment interacting as a functional unitUN ()Ecosystem collapseA transition beyond a bounded threshold in one or more variables that define the identity of the ecosystemKeith et al. ()Empirical studyStudies that collect and/or use observational and experimental data to answer ecological questionsHaller ()Mean maximum predictionAverage of the maximum number of years advance warning of ecosystem collapse reported in 21 articlesThis articleNarrative reviewA comprehensive narrative synthesis of previously published information, but does not employ rigorous and explicit literature search methodsLortie ()SurrogateA component of an ecosystem that can be more easily measured or managed than others and that is used as an indicator of the attribute/trait/characteristic/quality of that ecosystem.Lindenmayer et al. ()Systematic reviewA type of literature review that employs detailed, rigorous, and explicit methods to answer a specific questionLortie ()Theoretical studyStudies that use conceptual, mathematical, or simulation methods, often parameterized with real data, to answer ecological questionsHaller ()Within the growing literature on ecosystem collapse, we found that: (1) definitions of ecosystems, and what constitutes ecosystem collapse, are scant; (2) experimental tests of theory are rare – just four distinct experiments have been conducted; and (3) there is a mismatch between predicting ecosystem collapse, and predicting ecosystem collapse in time to implement management to avert collapse. We found warnings of ecosystem collapse ranged between 1 and 40 years prior to a shift (Contamin & Ellison ; Carpenter et al. ), but sometimes averting collapse was unlikely even if actions were taken more than 40 years prior to the onset of a shift in ecosystem state (Biggs et al. ). In fact, many early warnings of collapse may not be detected by managers as these warnings may precede ecosystem shifts by 2000 years (Spanbauer et al. ). While this means ecosystem shifts could be detected 2000 years into the future by managers, some imminent ecosystem shifts may have displayed early warning signals in the first century – and are undetectable for managers using early warning signals on contemporary (rather than paleoecological) data sets. Furthermore, a land manager's ability to predict collapse may also be confounded as supposed early warning signals may be detected after the collapse of an ecosystem (Carpenter et al. ).Experiences and challenges in predicting ecosystem collapseFor Criterion E under the IUCN RLE to work, we need to answer two key questions: Is predicting collapse possible? If so, how much advance warning is needed to avert collapse? Answers are essential to determine the level of threat an ecosystem is under. To answer these questions, we conducted a formal systematic review (see Table for glossary) of the scientific literature in four online databases on August 11, 2015, using a standardized search string (see Supplementary Methods and Table S1 for methodological details). Our search returned 10,696 articles, but only 64 focused on predicting ecosystem collapse (see Table S2 for details). Of these, six narrative reviews (Table S1) examined early warning indicators (Boettiger et al. ; Dakos et al. ), regime shifts (deYoung et al. ) and their application to management (Angeler et al. ). Of the 58 remaining articles, 29 were theoretical, and 29 were empirical. Nine empirical articles were experimental. In terms of ecosystems, aquatic systems were the focus of 42 studies, while just 13 articles focused on terrestrial systems (Table S3). For the empirical articles exploring these systems, only half provided a detailed description of the ecosystem under study. The number of publications investigating the predictability of ecosystem collapse increased considerably from 2000 to 2015 (no articles in 2000 to a peak of 12 articles in 2013, and 10 articles in 2015; Figure ).Number of papers predicting ecosystem collapse published each year since 2000.All but one study indicated that numerical prediction of ecosystem collapse is possible, but 21 identified significant practical challenges in predicting collapse – regardless of paper type (i.e., theoretical vs. empirical), ecosystem type (i.e., aquatic vs. terrestrial), surrogate type (i.e., abiotic vs. biotic), or early warning indicator (Figure a‐d). Predicting collapse requires intimate understanding of an ecosystem (50/58 studies), knowledge of the type of transition occurring (e.g., gradual, nonlinear, etc., 18/58 studies), and a suitable mathematical model of the ecosystem (25/58 studies) to guide the choice of appropriate early warning indicators. Even when these factors are known, reliable prediction is hindered by observation error or environmental variability (27/58 studies), choice of the surrogate that will rapidly display signs of system instability (18/58 studies), and insufficient temporal or spatial sampling of ecosystems (41/58 studies; Table S4).Practicality of predicting ecosystem collapse by (a) paper type (theoretical, empirical, or experimental), (b) ecosystem type (aquatic or terrestrial), (c) surrogate type (abiotic or biotic), and (d) early warning indicator type (only including early warning indicators [EWI] tested at least twice across all papers reviewed). Numbers may add up to more than the number of articles reviewed (58), as individual articles may have examined more than one ecosystem type, surrogate type, or early warning indicator type.Despite these limitations, 58 articles provided a numerical prediction of collapse: 17 based on disturbance intensity, 33 based on time (in time steps or years), and 8 based on risk/vulnerability. For temporal predictions, early warning indicators provided between 1and 40 years advance warning of collapse (mean maximum prediction ± SE = 7.6 ± 2.1 years; see Table for glossary). However, if management actions are delayed or acting on “slow” disturbances (e.g., shoreline incursion by rising sea levels), then averting collapse is unlikely – even if actions are taken more than 40 years prior to the onset of the shift (Biggs et al. ). Given that acting on early warnings may not always avert ecosystem collapse, it is more realistic to use early warning indicators as tools that can identify: (1) the need for intervention, and (2) a timeframe when it may still be possible to act to avert collapse, but not necessarily a time when intervention equates with easy or complete reversal of collapse (Donangelo et al. ).Toward better predictions of ecosystem collapseOur formal systematic review demonstrated the limited current ability to apply theoretical methods of predicting collapse to real‐world ecosystems, which has substantial implications for conducting robust, quantitative assessments of collapse risk under Criterion E of the IUCN RLE. We therefore identify six unresolved issues and areas of concern that must be addressed to more rapidly progress work on ecosystem collapse, and its application to ecosystem conservation. The unresolved issues and areas of concern we present below are to some extent hierarchical – the first two issues (defining ecosystems and ecosystem collapse) need to be resolved (i.e., consensus reached by researchers and practitioners) to then effectively address the subsequent issues.First, researchers must better define and conceptualize ecosystems when examining ecosystem collapse. Our review shows that only half of the empirical studies we examined contained a detailed definition of the ecosystem under study; but this is an essential part of the IUCN RLE assessment (Keith et al. ). Different definitions at different spatial scales can lead to researchers, managers, and policy makers talking about different levels of thematic complexity and – effectively – at cross‐purposes. For example, Higgins & Scheiter () found that vegetation shifts driven by changes in atmospheric CO2 occurred abruptly at local scales, but occurred smoothly when averaged at a continental scale. While it may be desirable to list ecosystems at multiple thematic levels (e.g., if different jurisdictions require different levels of information about an ecosystem), it is critical that researchers, managers and policy makers are clear about the scale at which they are defining an ecosystem and at which they are conducting assessments. Scale‐dependent differences in ecosystem response have substantial implications for robust estimates of collapse risk, and timely implementation of appropriate management interventions to avert collapse. Under a global assessment process like that proposed by the IUCN, consistency in definitions – and the scale of those definitions – is essential to minimize multiple listings of “equivalent” ecosystems, to encourage global information sharing between managers and researchers working in similar systems, and to facilitate optimal distribution of funds to conserve globally at‐risk ecosystems.Second, researchers must better define ecosystem collapse. While this issue has been raised by other researchers (e.g., Boitani et al. ), we underscore it here because our review highlights the extraordinary scarcity of descriptions or conceptualizations of collapsed ecosystems in the reviewed literature (7/58 studies). How ecosystem collapse manifests will be ecosystem specific (Boitani et al. ), which greatly increases the challenge of defining ecosystem collapse consistently across the globe. However, in understanding what desired ecosystems are not, we may begin to develop a general definition of ecosystem collapse (e.g., Tozer et al. ). Consistency in defining ecosystem collapse is essential to gauge the relative level of threat faced by different ecosystems (Boitani et al. ) under the IUCN RLE, and is critical to rigorous assessment under all criteria – including Criterion E (Rodríguez et al. ). To move toward a workable definition, we suggest that ecosystem collapse be defined relative to a benchmark or reference condition – as suggested by the IUCN RLE (Rodríguez et al. ). An appropriate reference condition should consider the existing and recent composition (species assemblages), structure (complexity and configuration), and function (processes and dynamics) of an ecosystem (McDonald et al. ). Moreover, we recommend that explicit definitions of reference conditions as well as explicit definitions of collapse for different ecosystems used by researchers be included in published research – particularly research concerning collapse (including literature on regime shifts, alternative stable states, tipping points, etc.). This will help to expedite consensus on what collapse looks like in different ecosystems and, indeed, what defines specific ecosystems.Third, there is an unprecedented demand for a far stronger meld of theory, experimentation, and practice to advance work on ecosystem collapse, and its application to initiatives like the IUCN RLE. Since 2009, the IUCN RLE has undergone extensive development, characterizing threat levels based on risk of ecosystem collapse (Keith et al. ; Box ). Yet, for Criterion E, published RLE documentation (e.g., Keith et al. , 2015; IUCN ) indicates that this process has evolved mostly independently from theoretical research on the same topic. In fact, there is almost no literature from the two areas in common – recent theoretical articles that explore early warning indicators of collapse are rarely cited within RLE documentation (except four articles published in the early 2000s including Scheffer et al. ), and RLE documentation has not been referenced in any theoretical ecosystem collapse articles. Yet, our review highlights the recognition in theoretical literature that in‐field testing of collapse predictors is required (11/29 studies), and empirical literature recognizes the need for further theoretical development of collapse predictors that take into consideration constraints on practical implementation (10/29 empirical studies identified significant challenges with implementing current collapse predictors). It is clear that greater integration of theory and practice is needed not only to advance the practical application of early warning indicators, but also to ensure theoretical advancements in early warning indicators are applied to ecosystem collapse concepts within Criterion E of the IUCN RLE assessment.Fourth, early warning indicators of collapse must be more practical as current indicators (such as increasingly variable or flickering time series; see Kéfi et al. and Dakos et al. for a full summary of temporal and spatial indicators) are often difficult for managers to apply (10/29 studies). This must be rectified given that conclusions drawn by managers can be strongly influenced by the early warning indicator used (Seekell & Dakos ), the variable monitored (Batt et al. ), and the presence of false positive or negative warning signals (Burthe et al. ). Given the complexity associated with interpreting current early warning indicators, simpler ecosystem‐specific indicators may be more practical for collapse assessments as they increase the chance that end‐users (i.e., land managers) can interpret them. Applying an adaptive surrogacy framework (Lindenmayer et al. ) to early warning indicators will help in identifying tractable early warning indicators for managers that are sensitive, cost‐effective, consistent, and that have realistic data requirements. This framework also could be used to evaluate the appropriateness of variables, such as remotely sensed data, for monitoring as surrogates for ecosystem collapse. The reviewed literature repeatedly recommended using remotely sensed data for the application of spatial early warning indicators (16/58 studies), but its surrogacy value has not been adequately evaluated. Remotely sensed data may meet the high‐frequency data demands of proposed early warning indicators (Carpenter et al. ; Burthe et al. ), but may also yield conservative estimates of collapse risk. For example, Burns et al. () used structural features (i.e., hollow‐bearing trees, an important resource for many unique fauna) to determine IUCN threat status of south‐eastern Australian forests. As remote sensing of tree hollows is not currently feasible, using remotely sensed data may overestimate the availability of these structural features, in turn underestimating collapse risk. Such erroneous estimates may have dire consequences for ecosystems (Biggs et al. ). However, we recognize that remote sensing encompasses many variables, and some remotely sensed applications may be useful as direct monitoring variables (Pereira et al. ), and therefore have potential for use in predicting ecosystem collapse.Fifth, if there are significant challenges in predicting collapse in practice, management needs to focus on improving understanding of ecosystems (as recognized by 50/58 studies in our review), particularly determining the relationship between variability and stability. Some researchers suggest that ecosystems become increasingly variable prior to collapse (Donangelo et al. ; Batt et al. ). Yet, environmental variability can enhance the stability of some ecosystems (Borgogno et al. ), as well as increase population growth rates and viability (Lawson et al. ). Actions designed to reduce variability may therefore have perverse outcomes (Lawson et al. ). Hence, manipulative experiments that push subsets of ecosystems beyond the bounds of natural variability are essential to enhance understanding of ecosystems and ecosystem thresholds – as already demonstrated with “extreme” manipulations of precipitation and temperature variability in tall grass prairies in North America (Hoover et al. ).Finally, we must ensure listing ecosystems under an IUCN framework and subsequent practical ecosystem management results in the conservation of biodiversity – as highlighted by Keith et al. (). This is because biodiversity plays critical roles in ecosystem function, dynamics, and stability (Reich et al. ). To know if this occurs, we recommend that IUCN ecosystem assessments be revisited regularly (e.g., bidecadally) to quantify: (1) the status of knowledge for the system: whether this has improved, declined, or stagnated – and how (e.g., through experimentation, lack of funding), (2) the effectiveness of monitoring for predicting collapse and improving understanding of ecosystem trajectories – and how this has come about (e.g., data collection or management, change of monitoring protocol, etc.), and (3) the efficacy of collapse surrogates (both monitored variables and early warning indicators) used.CaveatWe present a review of the global, peer‐reviewed literature available to date on predicting ecosystem collapse, providing commentary on the implications our findings have on numerically predicting collapse under the IUCN RLE. While this review provides new and important insight into the challenges associated with implementing the IUCN RLE – it has focused attention on Criterion E only. Further research – and review – of the ecological literature is required to assess each of the other IUCN RLE criteria, and to provide a comprehensive evaluation of the merits and potential challenges associated with conducting a global assessment of ecosystem threat levels using this new framework.Concluding remarksThere is strong empirical evidence for ecosystem collapse in recent history (Scheffer et al. ), but anticipating collapse is complex. We found that while there is evidence to suggest that numerically predicting collapse is possible, at present, early warning indicators cannot predict collapse reliably across all ecosystems. This means we are currently limited in our ability to provide reliable and robust quantitative predictions of ecosystem collapse using Criterion E of the RLE framework. Existing early warning indicators need refinement for general practical application. Improving the robustness of predictions demands an intimate, long‐term understanding of ecosystem dynamics and drivers. This, in turn, requires experimentation with, and long‐term monitoring of, ecosystems. Given the growing evidence for collapsed ecosystems and the formal codification of ecosystem collapse in the IUCN RLE, there is an immediate need for robust, generally applicable predictors of ecosystem collapse. Thus, the time for researchers, managers, and policy makers to collaborate is now. Never before has it been more important to bring theory, experimentation, and practice together to further the global conservation of ecosystems, and the biodiversity therein.1BOXSummary of the IUCN RLE criteriaIn May 2014, the IUCN ratified the criteria for a Red List of Ecosystems assessment framework at the 83rd session of the Council of the International Union for Conservation of Nature (Decision C/83/22; IUCN ). Criteria A to D are based on a decline in the spatial or functional attributes of ecosystems, while Criterion E focuses on a quantitative analysis of collapse risk. The summary of the criteria provided is based on Keith et al. () and IUCN ().Criterion A: Requires an assessment of the past, present, or future decline in spatial extent of a defined ecosystem.Criterion B: Requires an assessment of the extent or area of occupancy of a defined ecosystem.Criterion C: Requires an assessment of the past, present, or future degradation abiotic variable(s) critical to the functioning of a defined ecosystem.Criterion D: Requires an assessment of the past, present, or future disruption of biotic processes or interactions critical to the functioning of a defined ecosystem.Criterion E: Requires a quantitative analysis of the collapse risk of a defined ecosystem.AcknowledgmentWe thank E. Burns and two anonymous reviewers for comments that improved the work.ReferencesAngeler, D.G., Allen, C.R., Barichievy, C., et al. (2016). Management applications of discontinuity theory. J. Appl. Ecol., 53, 688‐698.Batt, R.D., Carpenter, S.R., Cole, J.J., Pace, M.L. & Johnson, R.A. (2013). Changes in ecosystem resilience detected in automated measures of ecosystem metabolism during a whole‐lake manipulation. Proc. Nat. Acad. Sci., 110, 17398‐17403.Biggs, R., Carpenter, S.R. & Brock, W.A. (2009). Turning back from the brink: detecting an impending regime shift in time to avert it. Proc. Nat. Acad. Sci., 106, 826‐831.Boettiger, C., Ross, N. & Hastings, A. (2013). Early warning signals: the charted and uncharted territories. Theor. Ecol., 6, 255‐264.Boitani, L., Mace, G.M. & Rondinini, C. (2015). Challenging the scientific foundations for an IUCN Red List of Ecosystems. Conserv. Lett., 8, 125‐131.Borgogno, F., D'Odorico, P., Laio, F. & Ridolfi, L. (2007). Effect of rainfall interannual variability on the stability and resilience of dryland plant ecosystems. Water Resour. Res., 43, W06411.Burns, E.L., Lindenmayer, D.B., Stein, J., et al. (2015). Ecosystem assessment of mountain ash forest in the Central Highlands of Victoria, south‐eastern Australia. Austral. Ecol., 40, 386‐399.Burthe, S.J., Henrys, P.A., Mackay, E.B., et al. (2016). Do early warning indicators consistently predict nonlinear change in long‐term ecological data? J. Appl. Ecol., 53, 666‐676.Carpenter, S.R., Brock, W.A., Cole, J.J., Kitchell, J.F. & Pace, M.L. (2008). Leading indicators of trophic cascades. Ecol. Lett., 11, 128‐138.Carpenter, S.R., Cole, J.J., Pace, M.L., et al. (2011). Early warnings of regime shifts: a whole‐ecosystem experiment. Science, 332, 1079‐1082.Contamin, R. & Ellison, A.M. (2009). Indicators of regime shifts in ecological systems: what do we need to know and when do we need to know it? Ecol. Appl., 19, 799‐816.Dakos, V., Carpenter, S.R., Brock, W.A., et al. (2012). Methods for detecting early warnings of critical transitions in time series illustrated using simulated ecological data. PLoS ONE, 7, e41010.Dakos, V., Carpenter, S.R., van Nes, E.H. & Scheffer, M. (2014). Resilience indicators: prospects and limitations for early warnings of regime shifts. Philos. Trans. R. Soc. Lond. B Bio. Sci., 370, 20130263.deYoung, B., Barange, M., Beaugrand, G., et al. (2008). Regime shifts in marine ecosystems: detection, prediction and management. Trends Ecol. Evol. 23, 402‐409.Donangelo, R., Fort, H., Dakos, V., Scheffer, M. & van Nes, E.H. (2010). Early warnings for catastrophic shifts in ecosystems: comparison between spatial and temporal indicators. Int. J. Bifurcat. Chaos, 20, 315‐321.Haller, B.C. (2014). Theoretical and empirical perspectives in ecology and evolution: a survey. Bioscience, 64, 907‐916.Higgins, S.I. & Scheiter, S. (2012). Atmospheric CO2 forces abrupt vegetation shifts locally, but not globally. Nature, 488, 209‐212.Hoover, D.L., Knapp, A.K. & Smith, M.D. (2014). Resistance and resilience of a grassland ecosystem to climate extremes. Ecology, 95, 2646‐2656.IUCN. (2014). Decisions of the 83rd Meeting of the IUCN Council (May 2014). http://www.iucn.org/sites/dev/files/import/downloads/decisions_of_the_83rd_meeting_of_the_iucn_council__may_2014_.pdf (visited Aug. 14, 2016).IUCN. (2016). Guidelines for the application of IUCN Red List of Ecosystems Categories and Criteria, Version 1.0 (eds. L. Bland, D.A. Keith, N.J. Murray, J.P. Rodriguez). https://www.iucn.org/sites/dev/files/content/documents/rle_guidelines_draft_dec_2015.pdf (visited Aug. 14, 2016).Kéfi, S., Guttal, V., Brock, W.A., et al. (2014). Early warning signals of ecological transitions: methods for spatial patterns. PLoS ONE, 9, e92097.Keith, D.A., Rodríguez, J.P., Brooks, T.M., et al. (2015). The IUCN Red List of Ecosystems: motivations, challenges, and applications. Conserv. Lett., 8, 214‐226.Keith, D.A., Rodríguez, J.P., Rodríguez‐Clark, K.M., et al. (2013). Scientific foundations for an IUCN Red List of Ecosystems. PLoS ONE, 8, e62111.Lawson, C.R., Vindenes, Y., Bailey, L. & Pol, M. (2015). Environmental variation and population responses to global change. Ecol. Lett., 18, 724‐736.Lindenmayer, D., Pierson, J., Barton, P., et al. (2015). A new framework for selecting environmental surrogates. Sci. Total Environ., 538, 1029‐1038.Lortie, C.J. (2014). Formalized synthesis opportunities for ecology: systematic reviews and meta‐analyses. Oikos, 123, 897‐902.McDonald, T., Jonson, J. & Dixon, K.W. (2016). National standards for the practice of ecological restoration in Australia. Restor. Ecol., 24, S4‐S32.Pereira, H.M., Ferrier, S., Walters, M. et al. (2013). Essential biodiversity variables. Science, 339, 277‐278.Reich, P.B., Tilman, D., Isbell, F., et al. (2012). Impacts of biodiversity loss escalate through time as redundancy fades. Science, 336, 589‐592.Rockström, J., Steffen, W.L., Noone, K., et al. (2009). Planetary boundaries: exploring the safe operating space for humanity. Ecol. Soc., 14, 32.Rodríguez, J.P., Keith, D.A., Rodríguez‐Clark, K.M., et al. (2015). A practical guide to the application of the IUCN Red List of Ecosystems criteria. Philos. Trans. R. Soc. Lond. B Bio. Sci. 370, 20140003.Scheffer, M., Carpenter, S., Foley, J.A., Folke, C. & Walker, B. (2001). Catastrophic shifts in ecosystems. Nature, 413, 591‐596.Scheffer, M., Carpenter, S.R., Dakos, V. & van Nes, E.H. (2015). Generic indicators of ecological resilience: inferring the chance of a critical transition. Ann. Rev. Ecol. Evol. Syst., 46, 145‐167.Seekell, D.A. & Dakos, V. (2015). Heteroskedasticity as a leading indicator of desertification in spatially explicit data. Ecol. Evol., 5, 2185‐2192.Spanbauer, T.L., Allen, C.R., Angeler, D.G., et al. (2014). Prolonged instability prior to a regime shift. PLoS ONE, 9, e108936.Tozer, M.G., Leishman, M.R. & Auld, T.D. (2015). Ecosystem risk assessment for Cumberland Plain Woodland, New South Wales, Australia. Austral. Ecol., 40, 400‐410.UN. (2014). System of environmental‐economic accounting 2012: experimental ecosystem accounting. United Nations, New York. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Conservation Letters Wiley

Meeting the Global Ecosystem Collapse Challenge

Sato, Chloe F.; Lindenmayer, David B.
Conservation Letters , Volume 11 (1) – Jan 1, 2018
Download PDF
7 pages

Loading next page...
 
/lp/wiley/meeting-the-global-ecosystem-collapse-challenge-VjQ1WbAKTt

References (37)

Publisher
Wiley
Copyright
Copyright © 2018 Wiley Periodicals, Inc.
ISSN
1755-263X
eISSN
1755-263X
DOI
10.1111/conl.12348
Publisher site
See Article on Publisher Site

Abstract

Assessing risks of ecosystem collapseGlobally, anthropogenic pressures are forcing ecosystems outside of safe operating spaces (sensu Rockström et al. ), increasing their likelihood of collapse. The IUCN aims to assess the risk of collapse of all of the world's ecosystems by 2025, using the Red List of Ecosystems (RLE) assessment framework (Keith et al. ). Ecosystem collapse (see Table for glossary) is a hot topic in ecology, receiving increasing treatment in the scientific and popular literature – with several recent articles reviewing potential early warning indicators of collapse (Dakos et al. ; Kéfi et al. ; Scheffer et al. ; see Table for glossary and examples of early warning indicators). We address the broad overarching question: Is ecosystem collapse predictable? The importance of answering this question is underscored by the fact that the concept is now formally codified under five criteria (A‐E) in the IUCN RLE (Box ), with Criterion E specifically seeking a quantification of collapse risk (Keith et al. ). This significantly raises the stakes in identifying and refining tractable methods to predict collapse, especially because collapse assessments inform listings in threat categories along a continuum from Least Concern to Critically Endangered according to the risks that they face (Keith et al. ; IUCN ). Here, we explored the challenges associated with predicting ecosystem collapse, providing commentary on the considerable implications those challenges have on numerically predicting collapse under Criterion E of the IUCN RLE. We conduct a formal systematic review to provide a detailed assessment of the ability to predict ecosystem collapse – and identify six key concerns and unresolved issues (stemming from the findings of our review) that must be addressed for researchers and practitioners to promote assessments of ecosystem collapse risk based on robust and objective predictions of ecosystem collapse. Robust predictions of ecosystem collapse are important if Criterion E of the IUCN RLE is to be successfully established and rigorously applied.Glossary of termsTerminologyDefinitionReferenceEarly warning indicatorA general term for dynamic patterns (“signals”) in ecosystem behavior that precede ecosystem collapse, e.g., increasing variance, increasing autocorrelation, or the presence of “flickering” in temporal or spatial data; see Kéfi et al. () and Dakos et al. () for a full summary of temporal and spatial indicatorsBoettiger et al. ()EcosystemA dynamic complex of plant, animal, and microorganism communities and their nonliving environment interacting as a functional unitUN ()Ecosystem collapseA transition beyond a bounded threshold in one or more variables that define the identity of the ecosystemKeith et al. ()Empirical studyStudies that collect and/or use observational and experimental data to answer ecological questionsHaller ()Mean maximum predictionAverage of the maximum number of years advance warning of ecosystem collapse reported in 21 articlesThis articleNarrative reviewA comprehensive narrative synthesis of previously published information, but does not employ rigorous and explicit literature search methodsLortie ()SurrogateA component of an ecosystem that can be more easily measured or managed than others and that is used as an indicator of the attribute/trait/characteristic/quality of that ecosystem.Lindenmayer et al. ()Systematic reviewA type of literature review that employs detailed, rigorous, and explicit methods to answer a specific questionLortie ()Theoretical studyStudies that use conceptual, mathematical, or simulation methods, often parameterized with real data, to answer ecological questionsHaller ()Within the growing literature on ecosystem collapse, we found that: (1) definitions of ecosystems, and what constitutes ecosystem collapse, are scant; (2) experimental tests of theory are rare – just four distinct experiments have been conducted; and (3) there is a mismatch between predicting ecosystem collapse, and predicting ecosystem collapse in time to implement management to avert collapse. We found warnings of ecosystem collapse ranged between 1 and 40 years prior to a shift (Contamin & Ellison ; Carpenter et al. ), but sometimes averting collapse was unlikely even if actions were taken more than 40 years prior to the onset of a shift in ecosystem state (Biggs et al. ). In fact, many early warnings of collapse may not be detected by managers as these warnings may precede ecosystem shifts by 2000 years (Spanbauer et al. ). While this means ecosystem shifts could be detected 2000 years into the future by managers, some imminent ecosystem shifts may have displayed early warning signals in the first century – and are undetectable for managers using early warning signals on contemporary (rather than paleoecological) data sets. Furthermore, a land manager's ability to predict collapse may also be confounded as supposed early warning signals may be detected after the collapse of an ecosystem (Carpenter et al. ).Experiences and challenges in predicting ecosystem collapseFor Criterion E under the IUCN RLE to work, we need to answer two key questions: Is predicting collapse possible? If so, how much advance warning is needed to avert collapse? Answers are essential to determine the level of threat an ecosystem is under. To answer these questions, we conducted a formal systematic review (see Table for glossary) of the scientific literature in four online databases on August 11, 2015, using a standardized search string (see Supplementary Methods and Table S1 for methodological details). Our search returned 10,696 articles, but only 64 focused on predicting ecosystem collapse (see Table S2 for details). Of these, six narrative reviews (Table S1) examined early warning indicators (Boettiger et al. ; Dakos et al. ), regime shifts (deYoung et al. ) and their application to management (Angeler et al. ). Of the 58 remaining articles, 29 were theoretical, and 29 were empirical. Nine empirical articles were experimental. In terms of ecosystems, aquatic systems were the focus of 42 studies, while just 13 articles focused on terrestrial systems (Table S3). For the empirical articles exploring these systems, only half provided a detailed description of the ecosystem under study. The number of publications investigating the predictability of ecosystem collapse increased considerably from 2000 to 2015 (no articles in 2000 to a peak of 12 articles in 2013, and 10 articles in 2015; Figure ).Number of papers predicting ecosystem collapse published each year since 2000.All but one study indicated that numerical prediction of ecosystem collapse is possible, but 21 identified significant practical challenges in predicting collapse – regardless of paper type (i.e., theoretical vs. empirical), ecosystem type (i.e., aquatic vs. terrestrial), surrogate type (i.e., abiotic vs. biotic), or early warning indicator (Figure a‐d). Predicting collapse requires intimate understanding of an ecosystem (50/58 studies), knowledge of the type of transition occurring (e.g., gradual, nonlinear, etc., 18/58 studies), and a suitable mathematical model of the ecosystem (25/58 studies) to guide the choice of appropriate early warning indicators. Even when these factors are known, reliable prediction is hindered by observation error or environmental variability (27/58 studies), choice of the surrogate that will rapidly display signs of system instability (18/58 studies), and insufficient temporal or spatial sampling of ecosystems (41/58 studies; Table S4).Practicality of predicting ecosystem collapse by (a) paper type (theoretical, empirical, or experimental), (b) ecosystem type (aquatic or terrestrial), (c) surrogate type (abiotic or biotic), and (d) early warning indicator type (only including early warning indicators [EWI] tested at least twice across all papers reviewed). Numbers may add up to more than the number of articles reviewed (58), as individual articles may have examined more than one ecosystem type, surrogate type, or early warning indicator type.Despite these limitations, 58 articles provided a numerical prediction of collapse: 17 based on disturbance intensity, 33 based on time (in time steps or years), and 8 based on risk/vulnerability. For temporal predictions, early warning indicators provided between 1and 40 years advance warning of collapse (mean maximum prediction ± SE = 7.6 ± 2.1 years; see Table for glossary). However, if management actions are delayed or acting on “slow” disturbances (e.g., shoreline incursion by rising sea levels), then averting collapse is unlikely – even if actions are taken more than 40 years prior to the onset of the shift (Biggs et al. ). Given that acting on early warnings may not always avert ecosystem collapse, it is more realistic to use early warning indicators as tools that can identify: (1) the need for intervention, and (2) a timeframe when it may still be possible to act to avert collapse, but not necessarily a time when intervention equates with easy or complete reversal of collapse (Donangelo et al. ).Toward better predictions of ecosystem collapseOur formal systematic review demonstrated the limited current ability to apply theoretical methods of predicting collapse to real‐world ecosystems, which has substantial implications for conducting robust, quantitative assessments of collapse risk under Criterion E of the IUCN RLE. We therefore identify six unresolved issues and areas of concern that must be addressed to more rapidly progress work on ecosystem collapse, and its application to ecosystem conservation. The unresolved issues and areas of concern we present below are to some extent hierarchical – the first two issues (defining ecosystems and ecosystem collapse) need to be resolved (i.e., consensus reached by researchers and practitioners) to then effectively address the subsequent issues.First, researchers must better define and conceptualize ecosystems when examining ecosystem collapse. Our review shows that only half of the empirical studies we examined contained a detailed definition of the ecosystem under study; but this is an essential part of the IUCN RLE assessment (Keith et al. ). Different definitions at different spatial scales can lead to researchers, managers, and policy makers talking about different levels of thematic complexity and – effectively – at cross‐purposes. For example, Higgins & Scheiter () found that vegetation shifts driven by changes in atmospheric CO2 occurred abruptly at local scales, but occurred smoothly when averaged at a continental scale. While it may be desirable to list ecosystems at multiple thematic levels (e.g., if different jurisdictions require different levels of information about an ecosystem), it is critical that researchers, managers and policy makers are clear about the scale at which they are defining an ecosystem and at which they are conducting assessments. Scale‐dependent differences in ecosystem response have substantial implications for robust estimates of collapse risk, and timely implementation of appropriate management interventions to avert collapse. Under a global assessment process like that proposed by the IUCN, consistency in definitions – and the scale of those definitions – is essential to minimize multiple listings of “equivalent” ecosystems, to encourage global information sharing between managers and researchers working in similar systems, and to facilitate optimal distribution of funds to conserve globally at‐risk ecosystems.Second, researchers must better define ecosystem collapse. While this issue has been raised by other researchers (e.g., Boitani et al. ), we underscore it here because our review highlights the extraordinary scarcity of descriptions or conceptualizations of collapsed ecosystems in the reviewed literature (7/58 studies). How ecosystem collapse manifests will be ecosystem specific (Boitani et al. ), which greatly increases the challenge of defining ecosystem collapse consistently across the globe. However, in understanding what desired ecosystems are not, we may begin to develop a general definition of ecosystem collapse (e.g., Tozer et al. ). Consistency in defining ecosystem collapse is essential to gauge the relative level of threat faced by different ecosystems (Boitani et al. ) under the IUCN RLE, and is critical to rigorous assessment under all criteria – including Criterion E (Rodríguez et al. ). To move toward a workable definition, we suggest that ecosystem collapse be defined relative to a benchmark or reference condition – as suggested by the IUCN RLE (Rodríguez et al. ). An appropriate reference condition should consider the existing and recent composition (species assemblages), structure (complexity and configuration), and function (processes and dynamics) of an ecosystem (McDonald et al. ). Moreover, we recommend that explicit definitions of reference conditions as well as explicit definitions of collapse for different ecosystems used by researchers be included in published research – particularly research concerning collapse (including literature on regime shifts, alternative stable states, tipping points, etc.). This will help to expedite consensus on what collapse looks like in different ecosystems and, indeed, what defines specific ecosystems.Third, there is an unprecedented demand for a far stronger meld of theory, experimentation, and practice to advance work on ecosystem collapse, and its application to initiatives like the IUCN RLE. Since 2009, the IUCN RLE has undergone extensive development, characterizing threat levels based on risk of ecosystem collapse (Keith et al. ; Box ). Yet, for Criterion E, published RLE documentation (e.g., Keith et al. , 2015; IUCN ) indicates that this process has evolved mostly independently from theoretical research on the same topic. In fact, there is almost no literature from the two areas in common – recent theoretical articles that explore early warning indicators of collapse are rarely cited within RLE documentation (except four articles published in the early 2000s including Scheffer et al. ), and RLE documentation has not been referenced in any theoretical ecosystem collapse articles. Yet, our review highlights the recognition in theoretical literature that in‐field testing of collapse predictors is required (11/29 studies), and empirical literature recognizes the need for further theoretical development of collapse predictors that take into consideration constraints on practical implementation (10/29 empirical studies identified significant challenges with implementing current collapse predictors). It is clear that greater integration of theory and practice is needed not only to advance the practical application of early warning indicators, but also to ensure theoretical advancements in early warning indicators are applied to ecosystem collapse concepts within Criterion E of the IUCN RLE assessment.Fourth, early warning indicators of collapse must be more practical as current indicators (such as increasingly variable or flickering time series; see Kéfi et al. and Dakos et al. for a full summary of temporal and spatial indicators) are often difficult for managers to apply (10/29 studies). This must be rectified given that conclusions drawn by managers can be strongly influenced by the early warning indicator used (Seekell & Dakos ), the variable monitored (Batt et al. ), and the presence of false positive or negative warning signals (Burthe et al. ). Given the complexity associated with interpreting current early warning indicators, simpler ecosystem‐specific indicators may be more practical for collapse assessments as they increase the chance that end‐users (i.e., land managers) can interpret them. Applying an adaptive surrogacy framework (Lindenmayer et al. ) to early warning indicators will help in identifying tractable early warning indicators for managers that are sensitive, cost‐effective, consistent, and that have realistic data requirements. This framework also could be used to evaluate the appropriateness of variables, such as remotely sensed data, for monitoring as surrogates for ecosystem collapse. The reviewed literature repeatedly recommended using remotely sensed data for the application of spatial early warning indicators (16/58 studies), but its surrogacy value has not been adequately evaluated. Remotely sensed data may meet the high‐frequency data demands of proposed early warning indicators (Carpenter et al. ; Burthe et al. ), but may also yield conservative estimates of collapse risk. For example, Burns et al. () used structural features (i.e., hollow‐bearing trees, an important resource for many unique fauna) to determine IUCN threat status of south‐eastern Australian forests. As remote sensing of tree hollows is not currently feasible, using remotely sensed data may overestimate the availability of these structural features, in turn underestimating collapse risk. Such erroneous estimates may have dire consequences for ecosystems (Biggs et al. ). However, we recognize that remote sensing encompasses many variables, and some remotely sensed applications may be useful as direct monitoring variables (Pereira et al. ), and therefore have potential for use in predicting ecosystem collapse.Fifth, if there are significant challenges in predicting collapse in practice, management needs to focus on improving understanding of ecosystems (as recognized by 50/58 studies in our review), particularly determining the relationship between variability and stability. Some researchers suggest that ecosystems become increasingly variable prior to collapse (Donangelo et al. ; Batt et al. ). Yet, environmental variability can enhance the stability of some ecosystems (Borgogno et al. ), as well as increase population growth rates and viability (Lawson et al. ). Actions designed to reduce variability may therefore have perverse outcomes (Lawson et al. ). Hence, manipulative experiments that push subsets of ecosystems beyond the bounds of natural variability are essential to enhance understanding of ecosystems and ecosystem thresholds – as already demonstrated with “extreme” manipulations of precipitation and temperature variability in tall grass prairies in North America (Hoover et al. ).Finally, we must ensure listing ecosystems under an IUCN framework and subsequent practical ecosystem management results in the conservation of biodiversity – as highlighted by Keith et al. (). This is because biodiversity plays critical roles in ecosystem function, dynamics, and stability (Reich et al. ). To know if this occurs, we recommend that IUCN ecosystem assessments be revisited regularly (e.g., bidecadally) to quantify: (1) the status of knowledge for the system: whether this has improved, declined, or stagnated – and how (e.g., through experimentation, lack of funding), (2) the effectiveness of monitoring for predicting collapse and improving understanding of ecosystem trajectories – and how this has come about (e.g., data collection or management, change of monitoring protocol, etc.), and (3) the efficacy of collapse surrogates (both monitored variables and early warning indicators) used.CaveatWe present a review of the global, peer‐reviewed literature available to date on predicting ecosystem collapse, providing commentary on the implications our findings have on numerically predicting collapse under the IUCN RLE. While this review provides new and important insight into the challenges associated with implementing the IUCN RLE – it has focused attention on Criterion E only. Further research – and review – of the ecological literature is required to assess each of the other IUCN RLE criteria, and to provide a comprehensive evaluation of the merits and potential challenges associated with conducting a global assessment of ecosystem threat levels using this new framework.Concluding remarksThere is strong empirical evidence for ecosystem collapse in recent history (Scheffer et al. ), but anticipating collapse is complex. We found that while there is evidence to suggest that numerically predicting collapse is possible, at present, early warning indicators cannot predict collapse reliably across all ecosystems. This means we are currently limited in our ability to provide reliable and robust quantitative predictions of ecosystem collapse using Criterion E of the RLE framework. Existing early warning indicators need refinement for general practical application. Improving the robustness of predictions demands an intimate, long‐term understanding of ecosystem dynamics and drivers. This, in turn, requires experimentation with, and long‐term monitoring of, ecosystems. Given the growing evidence for collapsed ecosystems and the formal codification of ecosystem collapse in the IUCN RLE, there is an immediate need for robust, generally applicable predictors of ecosystem collapse. Thus, the time for researchers, managers, and policy makers to collaborate is now. Never before has it been more important to bring theory, experimentation, and practice together to further the global conservation of ecosystems, and the biodiversity therein.1BOXSummary of the IUCN RLE criteriaIn May 2014, the IUCN ratified the criteria for a Red List of Ecosystems assessment framework at the 83rd session of the Council of the International Union for Conservation of Nature (Decision C/83/22; IUCN ). Criteria A to D are based on a decline in the spatial or functional attributes of ecosystems, while Criterion E focuses on a quantitative analysis of collapse risk. The summary of the criteria provided is based on Keith et al. () and IUCN ().Criterion A: Requires an assessment of the past, present, or future decline in spatial extent of a defined ecosystem.Criterion B: Requires an assessment of the extent or area of occupancy of a defined ecosystem.Criterion C: Requires an assessment of the past, present, or future degradation abiotic variable(s) critical to the functioning of a defined ecosystem.Criterion D: Requires an assessment of the past, present, or future disruption of biotic processes or interactions critical to the functioning of a defined ecosystem.Criterion E: Requires a quantitative analysis of the collapse risk of a defined ecosystem.AcknowledgmentWe thank E. Burns and two anonymous reviewers for comments that improved the work.ReferencesAngeler, D.G., Allen, C.R., Barichievy, C., et al. (2016). Management applications of discontinuity theory. J. Appl. Ecol., 53, 688‐698.Batt, R.D., Carpenter, S.R., Cole, J.J., Pace, M.L. & Johnson, R.A. (2013). Changes in ecosystem resilience detected in automated measures of ecosystem metabolism during a whole‐lake manipulation. Proc. Nat. Acad. Sci., 110, 17398‐17403.Biggs, R., Carpenter, S.R. & Brock, W.A. (2009). Turning back from the brink: detecting an impending regime shift in time to avert it. Proc. Nat. Acad. Sci., 106, 826‐831.Boettiger, C., Ross, N. & Hastings, A. (2013). Early warning signals: the charted and uncharted territories. Theor. Ecol., 6, 255‐264.Boitani, L., Mace, G.M. & Rondinini, C. (2015). Challenging the scientific foundations for an IUCN Red List of Ecosystems. Conserv. Lett., 8, 125‐131.Borgogno, F., D'Odorico, P., Laio, F. & Ridolfi, L. (2007). Effect of rainfall interannual variability on the stability and resilience of dryland plant ecosystems. Water Resour. Res., 43, W06411.Burns, E.L., Lindenmayer, D.B., Stein, J., et al. (2015). Ecosystem assessment of mountain ash forest in the Central Highlands of Victoria, south‐eastern Australia. Austral. Ecol., 40, 386‐399.Burthe, S.J., Henrys, P.A., Mackay, E.B., et al. (2016). Do early warning indicators consistently predict nonlinear change in long‐term ecological data? J. Appl. Ecol., 53, 666‐676.Carpenter, S.R., Brock, W.A., Cole, J.J., Kitchell, J.F. & Pace, M.L. (2008). Leading indicators of trophic cascades. Ecol. Lett., 11, 128‐138.Carpenter, S.R., Cole, J.J., Pace, M.L., et al. (2011). Early warnings of regime shifts: a whole‐ecosystem experiment. Science, 332, 1079‐1082.Contamin, R. & Ellison, A.M. (2009). Indicators of regime shifts in ecological systems: what do we need to know and when do we need to know it? Ecol. Appl., 19, 799‐816.Dakos, V., Carpenter, S.R., Brock, W.A., et al. (2012). Methods for detecting early warnings of critical transitions in time series illustrated using simulated ecological data. PLoS ONE, 7, e41010.Dakos, V., Carpenter, S.R., van Nes, E.H. & Scheffer, M. (2014). Resilience indicators: prospects and limitations for early warnings of regime shifts. Philos. Trans. R. Soc. Lond. B Bio. Sci., 370, 20130263.deYoung, B., Barange, M., Beaugrand, G., et al. (2008). Regime shifts in marine ecosystems: detection, prediction and management. Trends Ecol. Evol. 23, 402‐409.Donangelo, R., Fort, H., Dakos, V., Scheffer, M. & van Nes, E.H. (2010). Early warnings for catastrophic shifts in ecosystems: comparison between spatial and temporal indicators. Int. J. Bifurcat. Chaos, 20, 315‐321.Haller, B.C. (2014). Theoretical and empirical perspectives in ecology and evolution: a survey. Bioscience, 64, 907‐916.Higgins, S.I. & Scheiter, S. (2012). Atmospheric CO2 forces abrupt vegetation shifts locally, but not globally. Nature, 488, 209‐212.Hoover, D.L., Knapp, A.K. & Smith, M.D. (2014). Resistance and resilience of a grassland ecosystem to climate extremes. Ecology, 95, 2646‐2656.IUCN. (2014). Decisions of the 83rd Meeting of the IUCN Council (May 2014). http://www.iucn.org/sites/dev/files/import/downloads/decisions_of_the_83rd_meeting_of_the_iucn_council__may_2014_.pdf (visited Aug. 14, 2016).IUCN. (2016). Guidelines for the application of IUCN Red List of Ecosystems Categories and Criteria, Version 1.0 (eds. L. Bland, D.A. Keith, N.J. Murray, J.P. Rodriguez). https://www.iucn.org/sites/dev/files/content/documents/rle_guidelines_draft_dec_2015.pdf (visited Aug. 14, 2016).Kéfi, S., Guttal, V., Brock, W.A., et al. (2014). Early warning signals of ecological transitions: methods for spatial patterns. PLoS ONE, 9, e92097.Keith, D.A., Rodríguez, J.P., Brooks, T.M., et al. (2015). The IUCN Red List of Ecosystems: motivations, challenges, and applications. Conserv. Lett., 8, 214‐226.Keith, D.A., Rodríguez, J.P., Rodríguez‐Clark, K.M., et al. (2013). Scientific foundations for an IUCN Red List of Ecosystems. PLoS ONE, 8, e62111.Lawson, C.R., Vindenes, Y., Bailey, L. & Pol, M. (2015). Environmental variation and population responses to global change. Ecol. Lett., 18, 724‐736.Lindenmayer, D., Pierson, J., Barton, P., et al. (2015). A new framework for selecting environmental surrogates. Sci. Total Environ., 538, 1029‐1038.Lortie, C.J. (2014). Formalized synthesis opportunities for ecology: systematic reviews and meta‐analyses. Oikos, 123, 897‐902.McDonald, T., Jonson, J. & Dixon, K.W. (2016). National standards for the practice of ecological restoration in Australia. Restor. Ecol., 24, S4‐S32.Pereira, H.M., Ferrier, S., Walters, M. et al. (2013). Essential biodiversity variables. Science, 339, 277‐278.Reich, P.B., Tilman, D., Isbell, F., et al. (2012). Impacts of biodiversity loss escalate through time as redundancy fades. Science, 336, 589‐592.Rockström, J., Steffen, W.L., Noone, K., et al. (2009). Planetary boundaries: exploring the safe operating space for humanity. Ecol. Soc., 14, 32.Rodríguez, J.P., Keith, D.A., Rodríguez‐Clark, K.M., et al. (2015). A practical guide to the application of the IUCN Red List of Ecosystems criteria. Philos. Trans. R. Soc. Lond. B Bio. Sci. 370, 20140003.Scheffer, M., Carpenter, S., Foley, J.A., Folke, C. & Walker, B. (2001). Catastrophic shifts in ecosystems. Nature, 413, 591‐596.Scheffer, M., Carpenter, S.R., Dakos, V. & van Nes, E.H. (2015). Generic indicators of ecological resilience: inferring the chance of a critical transition. Ann. Rev. Ecol. Evol. Syst., 46, 145‐167.Seekell, D.A. & Dakos, V. (2015). Heteroskedasticity as a leading indicator of desertification in spatially explicit data. Ecol. Evol., 5, 2185‐2192.Spanbauer, T.L., Allen, C.R., Angeler, D.G., et al. (2014). Prolonged instability prior to a regime shift. PLoS ONE, 9, e108936.Tozer, M.G., Leishman, M.R. & Auld, T.D. (2015). Ecosystem risk assessment for Cumberland Plain Woodland, New South Wales, Australia. Austral. Ecol., 40, 400‐410.UN. (2014). System of environmental‐economic accounting 2012: experimental ecosystem accounting. United Nations, New York.

Journal

Conservation Letters – Wiley

Published: Jan 1, 2018

Keywords: ; ; ; ; ;

There are no references for this article.