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

Learn More →

The IUCN Red List of Ecosystems: Motivations, Challenges, and Applications

The IUCN Red List of Ecosystems: Motivations, Challenges, and Applications Introduction Policy imperatives to assess risks to biodiversity above the species level are underscored by IUCN's recent endorsement of new criteria and categories for Red Lists of Ecosystems (RLE) as a global standard for ecosystem risk assessment, and the inclusion of the RLE in the 2014 Horizon Scan of environmental issues (Sutherland et al . ). The RLE method is based on five quantitative criteria (Appendix 1) designed to evaluate symptoms of risk in terrestrial, subterranean, freshwater, and marine ecosystems (Keith et al . ). What are the practical challenges in applying such a tool in environmental policy and management for biodiversity conservation, and how can they be met? In this perspective, our central purpose is to review the major challenges in developing, interpreting and applying the RLE method and consider trade‐offs inherent in the design of solutions. To support our discussion, we first elucidate the motivations and goals for Red Listing ecosystems and later identify current and potential applications of the risk assessment products. Overall, we suggest that the IUCN Red List of Ecosystems should be judged by whether: it achieves conservation ends and improves environmental management; its limitations are acceptable given its benefits; and it outperforms alternative methods for similar tasks. We draw affirmative evidence from an extensive global consultation process including scientific workshops, stakeholder meetings, reviews, and published case studies (Nicholson et al . ; Keith ; Rodriguez et al . , ; Keith et al . ; IUCN ; Nicholson et al . ). We also recommend ongoing performance evaluations to support users who need to make judgements about RLE applications. Motivations, goals, and trade‐offs in ecosystem risk assessment An IUCN tool to assess ecosystems has a sound conservation justification (Boitani et al . ). There are several well established imperatives to address multiple levels of biodiversity in conservation planning (Noss ). First, broad‐scale features such as ecosystems identify major assemblages of biota requiring protection across landscapes and seascapes, and serve as surrogates for poorly known fine‐scale features (Margules & Pressey ). Simultaneous evaluation of broad‐ and fine‐scale features helps to ensure that key elements such as threatened species are not overlooked in planning decisions. Evaluation at both scales is more likely to achieve “comprehensive,” “adequate,” and “representative” (CAR) conservation outcomes (Margules & Pressey ). Second, ecosystem‐level assessment can address ecological processes, such as interactions among populations and with abiotic components of landscapes and seascapes, which are generally excluded by taxon‐level assessments (Sabo ). Ecological processes are crucial in diagnosing threats to individual species and resolving potential management conflicts for coexisting species, and hence underpin the “adequacy” of conservation decisions. Third, most ecosystem functions rely on common species (Gaston & Fuller ). These are central to description and risk assessment of ecosystems (Keith et al . ), but important changes in common species rarely feature in threatened species analysis. Finally, people value and relate to ecosystems and places, even though they may not recognize many component species (Dallimer et al . 2012; Wunder et al . ). Human well‐being depends on ecosystems for many ecological services (Costanza et al . ). These qualities make ecosystem‐level assessments useful communication and education tools to support biodiversity conservation, macroeconomic planning and sustainable management of land and water. These imperatives for ecosystem‐level approaches to conservation underpin the central goal of the RLE criteria, which is to support conservation in resource use and management decisions by identifying ecosystems most at risk of biodiversity loss (Keith et al . ). To meet this goal, a balance of four qualities was sought in the design of the RLE method: generality, precision, realism, and simplicity (building on Levins ). Generality was seen as particularly important to ensure that the RLE criteria could be applied to terrestrial, subterranean, freshwater and marine ecosystem types, and sufficiently flexible to handle data of varying quality and detail (Keith et al . ). Precision promotes consistency, transparency, and repeatability in applications across assessors, regions and ecosystem types, and was a strong motivation for adopting quantitative listing criteria (Keith et al . ). Realism underpins reliable and accurate scientific assessments, and is enhanced in the RLE listing criteria by the ability to use direct ecosystem‐specific diagnostic variables to assess ecological processes (Keith et al . ). We added simplicity to Levins' framework (after Einstein ) because it enhances clarity of purpose, accessibility to users and communication of resulting conservation messages. The direct focus of RLE criteria on symptoms of biodiversity loss contributes to simplicity by decoupling the risk assessment from priority setting and valuation of ecosystem services (Keith ). We further discuss relationships between biodiversity, ecosystem functions, and services below. The four qualities sought in designing the RLE involve unavoidable trade‐offs that are well known in model development across diverse disciplines (Levins ). For example, Red List criteria for both species and ecosystems sacrifice some precision (hence consistency) to achieve generality across species with different life histories, and ecosystems with different governing processes. The trade‐offs and appropriate balance will depend on the goals and scope of risk assessment. When the scope is broad, as is the case with Red Lists, emphasis inevitably falls on generality with some trade‐offs on other desired qualities. The issue is not whether imprecision and inconsistency exist (Boitani et al . ), but whether improving precision is worth the trade‐offs in generality, realism, and simplicity. Hobday et al . ( ), in their risk assessment protocol for managed fisheries, use a hierarchical approach to manage such trade‐offs; higher levels of assessment achieve more precise estimates of risk than lower levels, but at the cost of additional complexity and reduced generality. This approach may prove useful for Red Listing in cases where available data limit the certainty of assessment outcomes. Investments in more detailed risk assessments may be warranted when low‐level assessments suggest risks may be high (Hobday et al . ). Challenges and solutions in RLE design and application Ecosystem types as assessment units The RLE protocol defines ecosystems as units of assessment that represent complexes of organisms and their associated physical environment within an area (Keith et al . ). This definition incorporates four elements—biotic and abiotic complexes, interactions, and spatial location (after Tansley 1935, pp. 299‐300)—to establish a framework for describing assessment units. These defining features set a framework to distinguish among ecosystem types, and also identify a suite of characteristics potentially sensitive to identified hazards. In RLE assessments, they set the requirements for description of ecosystem types, are used to establish the levels of change that signal ecosystem collapse, and thus underpin the generality of the RLE protocol (Keith et al . ). While any unit that is distinguishable on the basis of these defining features meets the conceptual definition of ecosystems , this should not be confused with the need for appropriately scaled descriptions and classifications of ecosystem types for application of the RLE criteria (Boitani et al . ). Systematic classifications based on the defining features of ecosystem types place each unit in context and promote consistent identification and use. This is very important for comprehensive global and regional risk assessments aimed to support systematic conservation planning (e.g., Lindgaard & Henrisksen ), but less important where risk assessments are used individually to support strategic ecosystem management (Keith ). Two aspects of ecosystem typologies need to be addressed: (1) the classification method and resulting units; and (2) the extent to which these ecosystem types can be mapped. The latter is critical for assessment of RLE criteria A and B (Appendix 1), which address rates of change in distribution, and whether the extent is restricted and declining. A lack of spatial data for variables used in ecosystem classification can limit or preclude mapping of some ecosystem types (Mackey et al . ). The development of suitable ecosystem typologies (or adaptation of existing ones) to support RLE assessments is challenging partly because biotic and abiotic features vary continuously (not discretely), and partly because the patterns of variation are scale‐dependent (Keith ). A suitable typology would also need to be hierarchical or nested, allowing ecosystems assessed at national or finer scales to be related transparently to those applied in global‐scale assessments. Two main lines of evidence suggest that these challenges can be met in workable applications of the RLE method. First, risk assessments of ecosystem‐type units have already been applied successfully at national and regional levels over several decades in Europe, the Americas, Australia, and Africa (Table ). Systematic typologies support most assessments, but in Australia listings of ecosystem types (locally termed “ecological communities”) derive from public nominations and each nominated unit is critically evaluated and adapted (where necessary to ensure robustness) by statutory committees of scientists as part of the assessment process (Keith ). Experience from this process shows that, even without a formal typology, red‐listed ecosystem types can be reliably interpreted by consultants and agency staff and can withstand adversarial legal challenge, so long as their defining features are adequately described (Preston & Adam ). National applications involve a variety of listing criteria, typologies, and nomenclature for the assessment units (Table and references therein), but they demonstrate that inherent uncertainties in unit delimitation can be managed and pose no barrier to productive risk assessment or application of its outcomes to real‐world conservation problems. Current legislative, policy, and reporting applications of ecosystem types and similar units Jurisdiction Application Assessment unit Goal and application Reference European Union Habitats Directive 92/43/EEC (European Commission) Habitat type. “Plant and animal communities as the characterizing elements of the biotic environment, together with abiotic factors operating together at a particular scale.” To maintain and restore, at favorable conservation status, natural habitats. Listed habitat types protected and the status of protected areas reported under Habitats Directive 1992. Council of the European Communities ( ) Germany Red List of biotopes (Federal Environment Agency) Biotope. “Habitat of a community of fauna and flora living in the wild.” To permanently safeguard biological diversity by countering threats to naturally occurring ecosystems, biotopes and species under the Federal Nature Conservation Act 2009 Riecken et al . ( ) Finland Red List of habitat types (Finnish Environment Institute) Habitat type. “Spatially definable land or aquatic areas with characteristic environmental conditions and biota which are similar between these areas but differ from areas of other habitat types.” To provide a complete description of the current state of the habitat types found in Finland, their development during recent decades, and the threats they are likely to face in the near future. Kontula & Raunio ( ) Norway Red List of ecosystems and habitat types (Norwegian Biodiversity Information Centre) Habitat type. “A homogeneous environment, including all plant and animal life and environmental factors that operate there.” To preserve biodiversity and assess performance against national targets and international obligations. Listed “Nature types” protected under Nature Diversity Act 2009. Lindgaard & Henrisksen ( ) Venezuela National Red List of ecosystems (Provita) Major vegetation types for national assessment; satellite‐derived land types for subnational assessments. To support conservation policies and decision making. Rodriguez et al . (2010) Canada State threatened species and ecosystems legislation (Manitoba Conservation and Water Stewardship Department) Ecosystem. “A dynamic complex of plant, animal, and micro‐organism communities and their nonliving environment interacting as a functional unit.” To conserve and protect endangered and threatened ecosystems in the province and promote the recovery of those ecosystems under the Endangered Species and Ecosystems Act 2013. Government of Manitoba ( ) Australia Lists of threatened ecological communities at national and state levels (Federal Department of Environment, state environment agencies) Ecological community. “An assemblage of native species that inhabits a particular area in nature.” To conserve biodiversity and support state of environment reporting. Listed “ecological communities” protected under Australian Environment Protection and Biodiversity Conservation Act 1999, and respective state legislation (see Keith for details) Commonwealth of Australia ( ), Keith ( ), Nicholson et al . ( ) South Africa National biodiversity legislation (South African National Biodiversity Institute) Ecosystem. “A dynamic complex of animal, plant, and micro‐organism communities and their nonliving environment interacting as a functional unit.” To provide for the protection of ecosystems that are threatened or in need of protection to ensure the maintenance of their ecological integrity under the South African Biodiversity Act 2004 and give effect to the Republic's obligations under international agreements. Republic of South Africa ( ), Driver et al . ( ) Second, the RLE protocol can learn from the Red List of Threatened Species (RLTS) when dealing with uncertainties in classification of assessment units. Mace ( ), on species conservation, suggests that “Units for listing need to be consistent, change rarely and can be somewhat arbitrary.” Yet even for species and extinction, the boundaries of taxa and the understanding of decline depend on context and on indeterminate, dynamic and inconsistently applied theoretical concepts (Regan et al . ). The same is true of ecosystems and their collapse (Table ). Although species and ecosystem concepts often impose essentially arbitrary boundaries on continua (Regan et al . ), they serve to identify sets of relatively unique, ecologically and evolutionarily important components of biodiversity with which scientists, managers, and the public identify. Methods are developing to strengthen the consistency of species delimitation (Brooks & Helgen ). The RLTS is equipped with tools to deal with classification uncertainties (Akçakaya et al . ; Butchart et al . ), and similar tools are appropriate for the RLE. Species and ecosystems are imprecise entities, but their applications to Red Lists of species (for some 50 years) and of ecosystem types (for more than 20 years at national scales) have shown them to be useful frameworks for biodiversity assessment, conservation, macroeconomic planning, and resource use decisions (Margules & Pressey ; Rodrigues et al . ). Sources of uncertainty in species and ecosystem classifications. Tools such as interval arithmetic, fuzzy arithmetic (Akçakaya et al . ) and indices of genuine change (Butchart et al . ) help to deal with these in Red List assessments of both species and ecosystems Source of uncertainty Species Ecosystems Legacies of species and ecosystem concepts that have evolved over time Species classifications in current use include taxa whose names and circumscriptions are the cumulative legacy of several hundred years of classification and description. Hence they are based on a mix of contrasting species concepts including pre‐Darwinian morphological, biological (gene flow barriers), evolutionary, phenetic morphometric, and phylomolecular concepts. Although numerous variations on the ecosystem concept have been proposed, many retain the four elements in Tansley's original concept, albeit with different relative weightings depending on the application. The exposure of a global RLE to legacy effects will depend on how its adopted concept of an ecosystem evolves over time. Alphataxonomic boundary delineations are subjective and vary between experts and lineages There is no widely applied means of standardizing or calibrating levels of genetic variability within and between species. Degrees of splitting and lumping vary subjectively between lineages, despite nomenclatural rules and peer review. This directly determines which elements are eligible or ineligible for Red Listing. Similar difficulties apply in constructing a classification to represent consistent levels of dissimilarity between ecosystem types. This would be analytically possible if a globally representative set of samples existed. In the absence of such data, calibration will require subjective judgements. Numbers of described units are rapidly increasing, while large numbers of units remain undescribed Reinterpretation, new analyses, and discoveries are rapidly increasing the number of described taxa at the rate of approximately 10,000 per year (May ). The deficit of undescribed taxa is biased across major taxa and will be prolonged by current declines in the taxonomic workforce and alphataxonomic research activity. Numbers of ecosystem types may also increase over time as new variants are discovered, although the proportional rate of increase is likely to be slower than for species. The deficit of undescribed ecosystems is biased spatially and by environments (e.g., deep sea cf. terrestrial ecosystems). Boundaries delimiting related units are routinely revised Most taxonomic revisions involve new interpretations of species boundaries. Almost none find an existing classification to be an adequate representation of variation. Improved inventory data is likely to lead to revised circumscriptions and spatial boundaries between ecosystem types. Unrecognized cryptic units An unknown number of biologically isolated and genetically discrete taxa are morphologically cryptic and hence unrecognized (e.g., Mant et al . ). An unknown number of distinctive ecosystems may exist in cryptic environments such as subterranean and deep sea realms (e.g., Ramirez‐Llodra et al . ). New technology The molecular revolution is resulting in radical reorganization of classifications at generic level and above, and has only just begun to have an impact on species delineation. Below the species level, molecular analyses have revealed signatures of deep coalescence, sporadic introgression, unidentified sympatry, and evidence of polyphyly that require taxonomic correction (Weston ). New remote sensing and environmental DNA assay technologies (e.g., Jerde et al . ) may contribute to dynamism in classification of ecosystem types referred to above. While approaches to ecological classification are currently less unified than for species taxonomy, classification and mapping of broad landscape‐ and seascape‐scale ecosystem types is likely to be burdened by fewer historical legacies than species classification (Table ). Recent progress towards a global classification of terrestrial vegetation (Faber‐Langendoen et al . 2014), and other frameworks of continental and national scope, indicate that development of hierarchical ecosystem typologies is feasible, although establishing clear relationships (cross‐walks) with classifications in local use will be important for RLE applications. Vegetation types are often workable proxies for ecosystem types at terrestrial landscape scales; the macrogroup or group levels of classification could be appropriate for the global RLE (Faber‐Langendoen et al . 2014; IUCN ). However, restricting the RLE to vegetation types (as suggested by Boitani et al . ) would not resolve continuum issues which have long been recognized in plant community ecology (Austin ). This restriction would also reduce the generality of the RLE by precluding its application to many lentic, lotic, marine, subterranean, and some terrestrial ecosystems that cannot be characterized by vegetation alone. Conceptualizing ecosystem dynamics and collapse Ecosystems may decline in quality as well as quantity (Noss ). Erosion of the functional diversity and complementarity of biota is a particular concern because these features apparently support ecosystem function and resilience (Cardinale et al . ). Interpreting which aspects of ecosystem dynamics are relevant to conservation and defining collapse in an ecosystem‐specific context are challenging prerequisites for assessing risks of ecosystem collapse. Contemporary theory on ecosystem resilience (Folke et al . ) and state‐and‐transition frameworks for ecosystem dynamics (Hobbs & Suding ) support an effective trade‐off between realism and simplicity for this task in risk assessment, and have also proved useful in ecosystem management, restoration and other applications (e.g., Knapp et al . ). Tozer et al . ( ), for example, use a state‐and‐transition framework to identify processes that drive transitions between different states of a woodland ecosystem, and identify the states that represent ecosystem collapse (Figure ). This and other examples may be generalized (Figure ) to show how ecosystem dynamics can be interpreted for RLE assessment. A key task is to identify transitions between states either as part of natural variability within an ecosystem type, or as a process of collapse and replacement by a different or novel ecosystem type (Figure ). The outcomes of this interpretive process may be uncertain (Figure ), and depends on whether transitions to particular states involve loss of the defining features (characteristic biota and processes) that explicitly describe the ecosystem type (Table in Keith et al . ). Once explicitly identified (Figure ), model uncertainty can be incorporated into Red List assessments (Regan et al . ). This approach makes no assumptions about Clementsian climaxes, absolute stability, linearity between risk of collapse and diagnostic variables, or individualistic dynamics (cf. Boitani et al . ). Example of a diagrammatic process model for a woodland ecosystem illustrating transitions between alternative states driven by different threatening processes. Collapsed states may be reached via multiple trajectories. Suitable diagnostic variables for assessing Red List criteria C and D are those that measure progression along these trajectories (from Tozer et al . ). Generalized schematic representing a state and transition model of ecosystem dynamics (adapted from Folke et al . ). States A–G are defined by two state variables represented on the X and Y axes. The vertical axis (Z) represents potential for change. The RLE framework requires explicit description of state variables that represent defining features of the system (characteristic native biota and ecological processes) (Table in Keith et al . ). The two broken lines represent alternative interpretations of ecosystem collapse. For the inner line, transitions between states A, B, and C (e.g., white arrow) represent natural variability without loss of key defining features, while transitions across broken lines (e.g., grey arrow) to states D, E, F, and G represent collapse and replacement by novel ecosystems that lack these features. Progression along different pathways of ecosystem collapse are assessed using variables X and Y or other ecosystem‐specific diagnostic variables that reflect the loss of characteristic native biota and functionality. The outer broken line represents an alternative interpretation of ecosystem collapse in which state E is included within natural variation of the ecosystem type. Model uncertainty may be incorporated into Red List assessments by evidence‐weighted model averaging in calculations of diagnostic variables based on the two alternative models. Ecosystem stability depends on scale and context (Wiens ). The RLE method requires assessors to explicitly define the scale, context and characteristics of alternative states, as well as appropriate scaling functions for standardization (Rodriguez et al . ). These requirements promote transparency and robustness, allowing assessments to be critically reviewed, updated and repeated, with a level of consistency not attained by previous protocols (Nicholson et al . ). Self‐evidently, “states” are not useful for risk assessment when defined at such fine scales that they are always unstable or so broadly that they always appear stable. Ecosystem responses to climate change should be interpreted as transitions to collapsed states only if defining features and functionality are lost (Figure. ). If many ecosystem types qualify for threatened RLE status as a consequence of identified climate‐related mechanisms, it suggests that the method is sensitive to this serious threat to persistence and functionality of ecosystems and their component biota. Existing case studies show that the RLE status of several ecosystem types is influenced by identified threats attributed to global climate change, including quantified trends in cloud cover, flood regimes, fire regimes, and extreme temperatures (Keith ). Indeed, one important role of the RLE is to identify which ecosystems are at greatest risk of collapse from climate change. How this translates to priorities for action will depend on socioeconomic values, costs, and likelihood of management success (Carwardine et al . ), as well as the values of the novel systems that may replace those that collapse as the climate changes. Replacement by novel ecosystems may sometimes be interpreted as desirable adaptive responses. The role of Red Listing is to report on the risk of such transitions occurring, and the separate task of developing management strategies and priorities may or may not generate remedial management actions. The interpretation of novel ecosystems driven by no‐analog climates nonetheless remains a major challenge for ecosystem risk assessment. The main contribution of the RLE approach is to provide a framework for transparent reasoning and treatment of uncertainty where climate change generates competing models of ecosystem dynamics (Figure ). Quantifying ecosystem degradation and collapse Stochastic process‐based ecological models provide the most powerful means of quantifying the risks of ecosystem collapse because they can incorporate multiple interacting mechanisms of ecosystem decline (e.g., Lester & Fairweather ; Plagányi et al . ). The RLE method provides for the application of such models through criterion E (Appendix 1; Keith et al . ; Burns et al . ). For many ecosystem types, however, there may be insufficient data on which to build models with adequate realism. Criteria C and D (Appendix 1) enable use of diagnostic variables that can be related directly to ecosystem degradation. Risk assessment protocols that incorporate both quantitative models and diagnostic variables provide the generality (hence flexibility) and simplicity needed to handle varying data quality and diverse symptoms of risk (IUCN ; Hobday et al . ; Keith et al . ). The trade‐offs for realism and precision will depend on how well the diagnostic variables quantify degradation. Four options for assessing the severity of functional decline were evaluated during RLE consultation workshops: unstructured qualitative ranking; aggregated indices of health/condition; one or a few prescribed generic ecosystem variables; and assessor‐defined ecosystem‐specific variables. Many existing protocols rank the severity of functional decline using unstructured qualitative methods (Nicholson et al . ). This is the least consistent, transparent and repeatable option because assessments cannot be effectively calibrated. Aggregated indices or multivariate measures of ecosystem health or condition may appear more consistent, but missing data, aggregation artefacts, threshold behavior and averaging effects reduce their sensitivity (e.g., Jennings ; Gorrod et al . ), and different types of ecosystems are likely to require structurally different indices (e.g., marine vs. terrestrial). The other two options focus on particular variables correlated with functional decline. They have complementary strengths and weaknesses. The prescriptive option (as suggested by Boitani et al . ) requires estimates of trends in one or a few preset, generic variables such as productivity, species richness, or diversity. This appears to promote consistency, as all assessments use a common ‘currency’ to measure ecosystem degradation. The more flexible option, which permits assessor‐defined ecosystem‐specific variables, trades some of this apparent consistency for increased generality and realism, measuring more direct proxies of the ecological processes that threaten characteristic native biota. Thus, functionally contrasting systems may be assessed using different variables, which are scaled using an inferred threshold of collapse for ranking against a common set of ordinal threat categories (e.g., Figure 6 in Keith et al . ). Examples of suitable assessment variables include: measures of structural complexity or abundance of foundation species in forest or reef systems threatened by disturbance (Alvarez‐Filip et al . ; Burns et al . ); trends in hydrological or water quality variables where there is evidence of biotic dependency in wetlands threatened by water extraction or pollution (Lester & Fairweather in Keith et al . ; Pisanu et al . ); and trends in the abundance of top predators or other species that provide crucial links in food webs in systems sustained by top–down regulatory processes (Schmitz et al . ; Estes et al . ). The consultation process identified four reasons why gains in generality and realism of an ecosystem‐specific approach should be worth an apparent reduction in consistency. First, scaled ecosystem‐specific response variables should be more sensitive and consistent indicators of ecosystem degradation than generic variables when the relationship between the latter and the persistence of characteristic native biota is noisy or inconsistent across different ecosystems (e.g., Adler et al . ; Hooper et al . ). Prescribed response variables are workable for a narrow range of processes or ecosystem types (e.g., exploited fisheries; Hobday et al . ), but impose undue constraints for risk assessments across a wide range of ecosystems and processes. Second, the scaling of ecosystem‐specific variables promotes consistency between assessments, so long as inferred thresholds of collapse are evidence‐based and calculations of the relative severity of declines incorporate the uncertainties. Third, an ecosystem‐specific approach is less limited by data availability than a prescriptive one‐size‐fits‐all approach which requires the same variables to be quantified across all ecosystem types. Finally, ecosystem‐specific approaches promote critical examination of the evidence in diagnosing causes of decline, the pathways of collapse, the most sensitive means of measuring decline along those pathways and setting an explicit threshold for unacceptable loss of characteristic native biota (Figures and ). Further improvements in consistency should be possible by narrowing the range of ecological variables deemed appropriate for assessing particular types of ecosystem degradation, based on accumulating empirical experience. This will be increasingly possible as the RLE criteria are applied to many contrasting ecosystems by a diverse community of expert assessors. Imposing a prescriptive approach too early would stifle the exploration and evidence gathering essential to further development. In the meantime, published assessments provide guidance, and an IUCN peer‐review process for global assessments will encourage consistency and promote rigor across assessments (IUCN ; Rodriguez et al . ). Delimiting ordinal categories of risk The RLE criteria use numerical decision thresholds to assign ecosystem types to ordinal categories of risk (Appendix 1; Keith et al . ), a feature contributing to the simplicity of risk assessment outputs. A distinction exists between decision thresholds , which specify the conditions required to invoke a decision (e.g., categorization), and ecological thresholds , which are theoretical devices to explain nonlinear ecosystem behavior (Martin et al . ). The rationale for each RLE criterion derives from theory, while utility is the primary consideration for setting particular threshold values that delimit threat categories within each criterion, i.e., to make the categories informative about relative levels of risk (Keith et al . , p. 10). Calls for a theoretical basis for decision threshold values (Boitani et al . ) thus confuse them with ecological thresholds. Utility was also important in setting thresholds in the RLTS. For example, the threshold delimiting Vulnerable from least concern for criterion A (population reduction) in the RLTS was increased from 20% (in v2.3) to 30% (v3.1, IUCN ) because of practical concerns that high proportions of some taxa would exceed the lower threshold, not because of new developments in population theory. Category thresholds may similarly require adjustments in the RLE as case studies accumulate. In addition to considering utility in the RLE, the thresholds of spatial criteria (B1 and B2) are scale dependent (Nicholson et al . ). The scaling of these thresholds in different subglobal domains requires research to support subglobal‐scale RLE assessments. Current and potential RLE applications The RLE criteria may be applied systematically to a set of ecosystem types within a specified domain (global or subglobal) or to single ecosystem types. The applications described below are for systematic assessments, but some may also be relevant to assessments of single ecosystems. The IUCN will lead the development of a global RLE, but will also support stakeholders to develop national and regional RLEs (IUCN ), as these are the scales where many resource management and development decisions are made. Red Lists of Ecosystems (sensu lato) are already part of the legislative or policy infrastructure for biodiversity conservation in some countries (Table ). IUCN's new global standard is influencing federal jurisdictions such as Australia and the European Union, where member states are aiming to harmonize their listing methods (Nicholson et al . ). The IUCN RLE criteria also provide a benchmark for other countries wanting to develop their policy infrastructure (Government of Rwanda ). The RLE can contribute significantly to monitoring and reporting the status of biodiversity. Some national reporting frameworks on threatened ecosystems are already in place (e.g., Riecken et al . ; Australian State of Environment Committee ; Driver et al . ). Tools such as the Red List Index (Butchart et al . ) promote temporal consistency. Internationally, global and subglobal RLEs could contribute to reporting instruments such as the Aichi Targets under the Convention on Biodiversity (especially target 5), the International Platform on Biodiversity and Ecosystem Services, and the evolving Sustainable Development Goals (IUCN ). A third application of RLEs is in the design of protected area networks, where they can identify locations of poorly reserved ecosystems that would benefit most (in terms of risk reduction) from inclusion in protected areas (Margules & Pressey ). The presence of threshold proportions of threatened and restricted‐range ecosystems are proposed criteria for the identification of key biodiversity areas as sites contributing significantly to the global persistence of biodiversity (IUCN ). Fourth, RLEs can inform prioritization of investment decisions for conservation, restoration and development, for example, as inputs to cost efficiency analyses in allocating national or corporate budgets (Carwardine et al . ), and by contributing data to financial institutions for assessing investment risks for development projects (e.g., see Equator Principles, http://www.equator‐principles.com/ ). RLEs can support adaptive management strategies (Williams ), for example by informing sustainable allocations of water resources to environmental flows and agricultural production, or the zonation and quotas in fisheries management areas (Green et al . ). While the status of ecosystem types is the primary output of assessments, this is founded on an evidence‐based diagnosis of threats and derivation of ecosystem‐specific diagnostic variables that should be suitable for monitoring responses to alternative management actions. RLEs may also inform the sustainable management and delivery of ecosystem services. The relationship between the RLE (for which the primary goal is to assess risks to ecosystem‐level biodiversity) and ecosystem services is complex because the latter depend on both ecosystems and social environments (Costanza et al . ). To enhance clarity of purpose and simplicity of the RLE protocol, the RLE criteria focus on the loss of characteristic native biota (genes, populations, species, assemblages, and functional groups) and disruption to ecological processes that sustain it. Additional tools may thus be required to evaluate risks to ecosystem stocks, functions and services, especially if novel ecosystems deliver new or improved services relative to their collapsed predecessors. Nonetheless, there are cases where the RLE assessments can provide important information about ecological changes that have major consequences for ecosystem services (Micklin ). The same causal factors that drive loss of biodiversity may also result in decline of ecosystem stocks, functions, and services (Cardinale et al . ). For example, Burns et al . ( ) assessed the Red List status of a forest ecosystem that had undergone major transformations in structure due to timber harvesting and successive canopy fires. At the crux of its Red List status was a loss of characteristic native biota associated with the declining abundance of large old trees, including mammals and birds that shelter and breed in tree cavities, insectivorous birds that forage in large canopies and under loose bark on large tree trunks, and poorly documented microbial and fungal diversity associated with deep forest litter. The same widespread transformation from old growth to regrowth that underpins the Red List status of this forest type has major implications for hydrological productivity (Kuczera ), carbon stocks and sequestration (Keith et al . ), timber production, as well as overall resilience of the system (Lindenmayer et al . 2011). Although the RLE criteria are focused on biodiversity, the outcomes of RLE assessments clearly have broader implications if interpreted with appropriate rigor and caution. Finally, the RLE offers great potential for environmental education through extension programs, media articles, and resource materials for schools and community groups ( http://www.iucnredlistofecosystems.org/ ). The RLE is at an early stage of development but has already generated considerable public interest (Marshall ). Conclusions and future directions In risk assessments generally, few hazards are defined precisely (Burgman ). Greater consistency could be achieved in the RLE through more prescriptive approaches to assessment (Boitani et al . ), but will these gains be worth the necessary trade‐offs in generality and realism? The issue is not whether the concepts of ecosystems and collapse are imprecise or inconsistent (they are, undoubtedly) but whether the working definitions are useful and produce valuable outcomes. That is, will the RLE criteria (Keith et al . ) lead to better conservation decisions than would be achieved without them? In sum, published assessments to date and our global consultation process suggest that the RLE framework achieves a workable balance among inevitable trade‐offs (Keith et al . ), though we welcome empirical evidence indicating where a better balance is possible. We suggest a way forward by developing the RLE based on: (1) learning from systematic applications of the criteria across different ecosystem types and scales; (2) updating application guidelines regularly; (3) classifying and describing ecosystem types systematically as an integrated part of RLE assessments; (4) implementing an active research and development agenda; and (5) refining the conceptual framework and working definitions. In this, adaptive development of the RLE can benefit much from the history of the RLTS which, as one of IUCN's flagship products, was trialed and adjusted iteratively over 50 years from rudimentary beginnings (Scott et al . ). Acknowledgments We are grateful to: the MAVA Foundation, Gordon and Betty Moore Foundation, the Australian Research Council (LP 130100435) and the ARC Centre of Excellence for Environmental Decisions for their support, and to all participants in the RLE consultation process for their willingness to engage in challenging and fruitful discussions. Peter Letten and Julian Keith drafted Figure . Thanks also to June Keith. We thank two anonymous reviewers for their constructive comments. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Conservation Letters Wiley

Loading next page...
 
/lp/wiley/the-iucn-red-list-of-ecosystems-motivations-challenges-and-SZzvmHvp43

References (84)

Publisher
Wiley
Copyright
Copyright © 2015 Wiley Periodicals, Inc.
ISSN
1755-263X
eISSN
1755-263X
DOI
10.1111/conl.12167
Publisher site
See Article on Publisher Site

Abstract

Introduction Policy imperatives to assess risks to biodiversity above the species level are underscored by IUCN's recent endorsement of new criteria and categories for Red Lists of Ecosystems (RLE) as a global standard for ecosystem risk assessment, and the inclusion of the RLE in the 2014 Horizon Scan of environmental issues (Sutherland et al . ). The RLE method is based on five quantitative criteria (Appendix 1) designed to evaluate symptoms of risk in terrestrial, subterranean, freshwater, and marine ecosystems (Keith et al . ). What are the practical challenges in applying such a tool in environmental policy and management for biodiversity conservation, and how can they be met? In this perspective, our central purpose is to review the major challenges in developing, interpreting and applying the RLE method and consider trade‐offs inherent in the design of solutions. To support our discussion, we first elucidate the motivations and goals for Red Listing ecosystems and later identify current and potential applications of the risk assessment products. Overall, we suggest that the IUCN Red List of Ecosystems should be judged by whether: it achieves conservation ends and improves environmental management; its limitations are acceptable given its benefits; and it outperforms alternative methods for similar tasks. We draw affirmative evidence from an extensive global consultation process including scientific workshops, stakeholder meetings, reviews, and published case studies (Nicholson et al . ; Keith ; Rodriguez et al . , ; Keith et al . ; IUCN ; Nicholson et al . ). We also recommend ongoing performance evaluations to support users who need to make judgements about RLE applications. Motivations, goals, and trade‐offs in ecosystem risk assessment An IUCN tool to assess ecosystems has a sound conservation justification (Boitani et al . ). There are several well established imperatives to address multiple levels of biodiversity in conservation planning (Noss ). First, broad‐scale features such as ecosystems identify major assemblages of biota requiring protection across landscapes and seascapes, and serve as surrogates for poorly known fine‐scale features (Margules & Pressey ). Simultaneous evaluation of broad‐ and fine‐scale features helps to ensure that key elements such as threatened species are not overlooked in planning decisions. Evaluation at both scales is more likely to achieve “comprehensive,” “adequate,” and “representative” (CAR) conservation outcomes (Margules & Pressey ). Second, ecosystem‐level assessment can address ecological processes, such as interactions among populations and with abiotic components of landscapes and seascapes, which are generally excluded by taxon‐level assessments (Sabo ). Ecological processes are crucial in diagnosing threats to individual species and resolving potential management conflicts for coexisting species, and hence underpin the “adequacy” of conservation decisions. Third, most ecosystem functions rely on common species (Gaston & Fuller ). These are central to description and risk assessment of ecosystems (Keith et al . ), but important changes in common species rarely feature in threatened species analysis. Finally, people value and relate to ecosystems and places, even though they may not recognize many component species (Dallimer et al . 2012; Wunder et al . ). Human well‐being depends on ecosystems for many ecological services (Costanza et al . ). These qualities make ecosystem‐level assessments useful communication and education tools to support biodiversity conservation, macroeconomic planning and sustainable management of land and water. These imperatives for ecosystem‐level approaches to conservation underpin the central goal of the RLE criteria, which is to support conservation in resource use and management decisions by identifying ecosystems most at risk of biodiversity loss (Keith et al . ). To meet this goal, a balance of four qualities was sought in the design of the RLE method: generality, precision, realism, and simplicity (building on Levins ). Generality was seen as particularly important to ensure that the RLE criteria could be applied to terrestrial, subterranean, freshwater and marine ecosystem types, and sufficiently flexible to handle data of varying quality and detail (Keith et al . ). Precision promotes consistency, transparency, and repeatability in applications across assessors, regions and ecosystem types, and was a strong motivation for adopting quantitative listing criteria (Keith et al . ). Realism underpins reliable and accurate scientific assessments, and is enhanced in the RLE listing criteria by the ability to use direct ecosystem‐specific diagnostic variables to assess ecological processes (Keith et al . ). We added simplicity to Levins' framework (after Einstein ) because it enhances clarity of purpose, accessibility to users and communication of resulting conservation messages. The direct focus of RLE criteria on symptoms of biodiversity loss contributes to simplicity by decoupling the risk assessment from priority setting and valuation of ecosystem services (Keith ). We further discuss relationships between biodiversity, ecosystem functions, and services below. The four qualities sought in designing the RLE involve unavoidable trade‐offs that are well known in model development across diverse disciplines (Levins ). For example, Red List criteria for both species and ecosystems sacrifice some precision (hence consistency) to achieve generality across species with different life histories, and ecosystems with different governing processes. The trade‐offs and appropriate balance will depend on the goals and scope of risk assessment. When the scope is broad, as is the case with Red Lists, emphasis inevitably falls on generality with some trade‐offs on other desired qualities. The issue is not whether imprecision and inconsistency exist (Boitani et al . ), but whether improving precision is worth the trade‐offs in generality, realism, and simplicity. Hobday et al . ( ), in their risk assessment protocol for managed fisheries, use a hierarchical approach to manage such trade‐offs; higher levels of assessment achieve more precise estimates of risk than lower levels, but at the cost of additional complexity and reduced generality. This approach may prove useful for Red Listing in cases where available data limit the certainty of assessment outcomes. Investments in more detailed risk assessments may be warranted when low‐level assessments suggest risks may be high (Hobday et al . ). Challenges and solutions in RLE design and application Ecosystem types as assessment units The RLE protocol defines ecosystems as units of assessment that represent complexes of organisms and their associated physical environment within an area (Keith et al . ). This definition incorporates four elements—biotic and abiotic complexes, interactions, and spatial location (after Tansley 1935, pp. 299‐300)—to establish a framework for describing assessment units. These defining features set a framework to distinguish among ecosystem types, and also identify a suite of characteristics potentially sensitive to identified hazards. In RLE assessments, they set the requirements for description of ecosystem types, are used to establish the levels of change that signal ecosystem collapse, and thus underpin the generality of the RLE protocol (Keith et al . ). While any unit that is distinguishable on the basis of these defining features meets the conceptual definition of ecosystems , this should not be confused with the need for appropriately scaled descriptions and classifications of ecosystem types for application of the RLE criteria (Boitani et al . ). Systematic classifications based on the defining features of ecosystem types place each unit in context and promote consistent identification and use. This is very important for comprehensive global and regional risk assessments aimed to support systematic conservation planning (e.g., Lindgaard & Henrisksen ), but less important where risk assessments are used individually to support strategic ecosystem management (Keith ). Two aspects of ecosystem typologies need to be addressed: (1) the classification method and resulting units; and (2) the extent to which these ecosystem types can be mapped. The latter is critical for assessment of RLE criteria A and B (Appendix 1), which address rates of change in distribution, and whether the extent is restricted and declining. A lack of spatial data for variables used in ecosystem classification can limit or preclude mapping of some ecosystem types (Mackey et al . ). The development of suitable ecosystem typologies (or adaptation of existing ones) to support RLE assessments is challenging partly because biotic and abiotic features vary continuously (not discretely), and partly because the patterns of variation are scale‐dependent (Keith ). A suitable typology would also need to be hierarchical or nested, allowing ecosystems assessed at national or finer scales to be related transparently to those applied in global‐scale assessments. Two main lines of evidence suggest that these challenges can be met in workable applications of the RLE method. First, risk assessments of ecosystem‐type units have already been applied successfully at national and regional levels over several decades in Europe, the Americas, Australia, and Africa (Table ). Systematic typologies support most assessments, but in Australia listings of ecosystem types (locally termed “ecological communities”) derive from public nominations and each nominated unit is critically evaluated and adapted (where necessary to ensure robustness) by statutory committees of scientists as part of the assessment process (Keith ). Experience from this process shows that, even without a formal typology, red‐listed ecosystem types can be reliably interpreted by consultants and agency staff and can withstand adversarial legal challenge, so long as their defining features are adequately described (Preston & Adam ). National applications involve a variety of listing criteria, typologies, and nomenclature for the assessment units (Table and references therein), but they demonstrate that inherent uncertainties in unit delimitation can be managed and pose no barrier to productive risk assessment or application of its outcomes to real‐world conservation problems. Current legislative, policy, and reporting applications of ecosystem types and similar units Jurisdiction Application Assessment unit Goal and application Reference European Union Habitats Directive 92/43/EEC (European Commission) Habitat type. “Plant and animal communities as the characterizing elements of the biotic environment, together with abiotic factors operating together at a particular scale.” To maintain and restore, at favorable conservation status, natural habitats. Listed habitat types protected and the status of protected areas reported under Habitats Directive 1992. Council of the European Communities ( ) Germany Red List of biotopes (Federal Environment Agency) Biotope. “Habitat of a community of fauna and flora living in the wild.” To permanently safeguard biological diversity by countering threats to naturally occurring ecosystems, biotopes and species under the Federal Nature Conservation Act 2009 Riecken et al . ( ) Finland Red List of habitat types (Finnish Environment Institute) Habitat type. “Spatially definable land or aquatic areas with characteristic environmental conditions and biota which are similar between these areas but differ from areas of other habitat types.” To provide a complete description of the current state of the habitat types found in Finland, their development during recent decades, and the threats they are likely to face in the near future. Kontula & Raunio ( ) Norway Red List of ecosystems and habitat types (Norwegian Biodiversity Information Centre) Habitat type. “A homogeneous environment, including all plant and animal life and environmental factors that operate there.” To preserve biodiversity and assess performance against national targets and international obligations. Listed “Nature types” protected under Nature Diversity Act 2009. Lindgaard & Henrisksen ( ) Venezuela National Red List of ecosystems (Provita) Major vegetation types for national assessment; satellite‐derived land types for subnational assessments. To support conservation policies and decision making. Rodriguez et al . (2010) Canada State threatened species and ecosystems legislation (Manitoba Conservation and Water Stewardship Department) Ecosystem. “A dynamic complex of plant, animal, and micro‐organism communities and their nonliving environment interacting as a functional unit.” To conserve and protect endangered and threatened ecosystems in the province and promote the recovery of those ecosystems under the Endangered Species and Ecosystems Act 2013. Government of Manitoba ( ) Australia Lists of threatened ecological communities at national and state levels (Federal Department of Environment, state environment agencies) Ecological community. “An assemblage of native species that inhabits a particular area in nature.” To conserve biodiversity and support state of environment reporting. Listed “ecological communities” protected under Australian Environment Protection and Biodiversity Conservation Act 1999, and respective state legislation (see Keith for details) Commonwealth of Australia ( ), Keith ( ), Nicholson et al . ( ) South Africa National biodiversity legislation (South African National Biodiversity Institute) Ecosystem. “A dynamic complex of animal, plant, and micro‐organism communities and their nonliving environment interacting as a functional unit.” To provide for the protection of ecosystems that are threatened or in need of protection to ensure the maintenance of their ecological integrity under the South African Biodiversity Act 2004 and give effect to the Republic's obligations under international agreements. Republic of South Africa ( ), Driver et al . ( ) Second, the RLE protocol can learn from the Red List of Threatened Species (RLTS) when dealing with uncertainties in classification of assessment units. Mace ( ), on species conservation, suggests that “Units for listing need to be consistent, change rarely and can be somewhat arbitrary.” Yet even for species and extinction, the boundaries of taxa and the understanding of decline depend on context and on indeterminate, dynamic and inconsistently applied theoretical concepts (Regan et al . ). The same is true of ecosystems and their collapse (Table ). Although species and ecosystem concepts often impose essentially arbitrary boundaries on continua (Regan et al . ), they serve to identify sets of relatively unique, ecologically and evolutionarily important components of biodiversity with which scientists, managers, and the public identify. Methods are developing to strengthen the consistency of species delimitation (Brooks & Helgen ). The RLTS is equipped with tools to deal with classification uncertainties (Akçakaya et al . ; Butchart et al . ), and similar tools are appropriate for the RLE. Species and ecosystems are imprecise entities, but their applications to Red Lists of species (for some 50 years) and of ecosystem types (for more than 20 years at national scales) have shown them to be useful frameworks for biodiversity assessment, conservation, macroeconomic planning, and resource use decisions (Margules & Pressey ; Rodrigues et al . ). Sources of uncertainty in species and ecosystem classifications. Tools such as interval arithmetic, fuzzy arithmetic (Akçakaya et al . ) and indices of genuine change (Butchart et al . ) help to deal with these in Red List assessments of both species and ecosystems Source of uncertainty Species Ecosystems Legacies of species and ecosystem concepts that have evolved over time Species classifications in current use include taxa whose names and circumscriptions are the cumulative legacy of several hundred years of classification and description. Hence they are based on a mix of contrasting species concepts including pre‐Darwinian morphological, biological (gene flow barriers), evolutionary, phenetic morphometric, and phylomolecular concepts. Although numerous variations on the ecosystem concept have been proposed, many retain the four elements in Tansley's original concept, albeit with different relative weightings depending on the application. The exposure of a global RLE to legacy effects will depend on how its adopted concept of an ecosystem evolves over time. Alphataxonomic boundary delineations are subjective and vary between experts and lineages There is no widely applied means of standardizing or calibrating levels of genetic variability within and between species. Degrees of splitting and lumping vary subjectively between lineages, despite nomenclatural rules and peer review. This directly determines which elements are eligible or ineligible for Red Listing. Similar difficulties apply in constructing a classification to represent consistent levels of dissimilarity between ecosystem types. This would be analytically possible if a globally representative set of samples existed. In the absence of such data, calibration will require subjective judgements. Numbers of described units are rapidly increasing, while large numbers of units remain undescribed Reinterpretation, new analyses, and discoveries are rapidly increasing the number of described taxa at the rate of approximately 10,000 per year (May ). The deficit of undescribed taxa is biased across major taxa and will be prolonged by current declines in the taxonomic workforce and alphataxonomic research activity. Numbers of ecosystem types may also increase over time as new variants are discovered, although the proportional rate of increase is likely to be slower than for species. The deficit of undescribed ecosystems is biased spatially and by environments (e.g., deep sea cf. terrestrial ecosystems). Boundaries delimiting related units are routinely revised Most taxonomic revisions involve new interpretations of species boundaries. Almost none find an existing classification to be an adequate representation of variation. Improved inventory data is likely to lead to revised circumscriptions and spatial boundaries between ecosystem types. Unrecognized cryptic units An unknown number of biologically isolated and genetically discrete taxa are morphologically cryptic and hence unrecognized (e.g., Mant et al . ). An unknown number of distinctive ecosystems may exist in cryptic environments such as subterranean and deep sea realms (e.g., Ramirez‐Llodra et al . ). New technology The molecular revolution is resulting in radical reorganization of classifications at generic level and above, and has only just begun to have an impact on species delineation. Below the species level, molecular analyses have revealed signatures of deep coalescence, sporadic introgression, unidentified sympatry, and evidence of polyphyly that require taxonomic correction (Weston ). New remote sensing and environmental DNA assay technologies (e.g., Jerde et al . ) may contribute to dynamism in classification of ecosystem types referred to above. While approaches to ecological classification are currently less unified than for species taxonomy, classification and mapping of broad landscape‐ and seascape‐scale ecosystem types is likely to be burdened by fewer historical legacies than species classification (Table ). Recent progress towards a global classification of terrestrial vegetation (Faber‐Langendoen et al . 2014), and other frameworks of continental and national scope, indicate that development of hierarchical ecosystem typologies is feasible, although establishing clear relationships (cross‐walks) with classifications in local use will be important for RLE applications. Vegetation types are often workable proxies for ecosystem types at terrestrial landscape scales; the macrogroup or group levels of classification could be appropriate for the global RLE (Faber‐Langendoen et al . 2014; IUCN ). However, restricting the RLE to vegetation types (as suggested by Boitani et al . ) would not resolve continuum issues which have long been recognized in plant community ecology (Austin ). This restriction would also reduce the generality of the RLE by precluding its application to many lentic, lotic, marine, subterranean, and some terrestrial ecosystems that cannot be characterized by vegetation alone. Conceptualizing ecosystem dynamics and collapse Ecosystems may decline in quality as well as quantity (Noss ). Erosion of the functional diversity and complementarity of biota is a particular concern because these features apparently support ecosystem function and resilience (Cardinale et al . ). Interpreting which aspects of ecosystem dynamics are relevant to conservation and defining collapse in an ecosystem‐specific context are challenging prerequisites for assessing risks of ecosystem collapse. Contemporary theory on ecosystem resilience (Folke et al . ) and state‐and‐transition frameworks for ecosystem dynamics (Hobbs & Suding ) support an effective trade‐off between realism and simplicity for this task in risk assessment, and have also proved useful in ecosystem management, restoration and other applications (e.g., Knapp et al . ). Tozer et al . ( ), for example, use a state‐and‐transition framework to identify processes that drive transitions between different states of a woodland ecosystem, and identify the states that represent ecosystem collapse (Figure ). This and other examples may be generalized (Figure ) to show how ecosystem dynamics can be interpreted for RLE assessment. A key task is to identify transitions between states either as part of natural variability within an ecosystem type, or as a process of collapse and replacement by a different or novel ecosystem type (Figure ). The outcomes of this interpretive process may be uncertain (Figure ), and depends on whether transitions to particular states involve loss of the defining features (characteristic biota and processes) that explicitly describe the ecosystem type (Table in Keith et al . ). Once explicitly identified (Figure ), model uncertainty can be incorporated into Red List assessments (Regan et al . ). This approach makes no assumptions about Clementsian climaxes, absolute stability, linearity between risk of collapse and diagnostic variables, or individualistic dynamics (cf. Boitani et al . ). Example of a diagrammatic process model for a woodland ecosystem illustrating transitions between alternative states driven by different threatening processes. Collapsed states may be reached via multiple trajectories. Suitable diagnostic variables for assessing Red List criteria C and D are those that measure progression along these trajectories (from Tozer et al . ). Generalized schematic representing a state and transition model of ecosystem dynamics (adapted from Folke et al . ). States A–G are defined by two state variables represented on the X and Y axes. The vertical axis (Z) represents potential for change. The RLE framework requires explicit description of state variables that represent defining features of the system (characteristic native biota and ecological processes) (Table in Keith et al . ). The two broken lines represent alternative interpretations of ecosystem collapse. For the inner line, transitions between states A, B, and C (e.g., white arrow) represent natural variability without loss of key defining features, while transitions across broken lines (e.g., grey arrow) to states D, E, F, and G represent collapse and replacement by novel ecosystems that lack these features. Progression along different pathways of ecosystem collapse are assessed using variables X and Y or other ecosystem‐specific diagnostic variables that reflect the loss of characteristic native biota and functionality. The outer broken line represents an alternative interpretation of ecosystem collapse in which state E is included within natural variation of the ecosystem type. Model uncertainty may be incorporated into Red List assessments by evidence‐weighted model averaging in calculations of diagnostic variables based on the two alternative models. Ecosystem stability depends on scale and context (Wiens ). The RLE method requires assessors to explicitly define the scale, context and characteristics of alternative states, as well as appropriate scaling functions for standardization (Rodriguez et al . ). These requirements promote transparency and robustness, allowing assessments to be critically reviewed, updated and repeated, with a level of consistency not attained by previous protocols (Nicholson et al . ). Self‐evidently, “states” are not useful for risk assessment when defined at such fine scales that they are always unstable or so broadly that they always appear stable. Ecosystem responses to climate change should be interpreted as transitions to collapsed states only if defining features and functionality are lost (Figure. ). If many ecosystem types qualify for threatened RLE status as a consequence of identified climate‐related mechanisms, it suggests that the method is sensitive to this serious threat to persistence and functionality of ecosystems and their component biota. Existing case studies show that the RLE status of several ecosystem types is influenced by identified threats attributed to global climate change, including quantified trends in cloud cover, flood regimes, fire regimes, and extreme temperatures (Keith ). Indeed, one important role of the RLE is to identify which ecosystems are at greatest risk of collapse from climate change. How this translates to priorities for action will depend on socioeconomic values, costs, and likelihood of management success (Carwardine et al . ), as well as the values of the novel systems that may replace those that collapse as the climate changes. Replacement by novel ecosystems may sometimes be interpreted as desirable adaptive responses. The role of Red Listing is to report on the risk of such transitions occurring, and the separate task of developing management strategies and priorities may or may not generate remedial management actions. The interpretation of novel ecosystems driven by no‐analog climates nonetheless remains a major challenge for ecosystem risk assessment. The main contribution of the RLE approach is to provide a framework for transparent reasoning and treatment of uncertainty where climate change generates competing models of ecosystem dynamics (Figure ). Quantifying ecosystem degradation and collapse Stochastic process‐based ecological models provide the most powerful means of quantifying the risks of ecosystem collapse because they can incorporate multiple interacting mechanisms of ecosystem decline (e.g., Lester & Fairweather ; Plagányi et al . ). The RLE method provides for the application of such models through criterion E (Appendix 1; Keith et al . ; Burns et al . ). For many ecosystem types, however, there may be insufficient data on which to build models with adequate realism. Criteria C and D (Appendix 1) enable use of diagnostic variables that can be related directly to ecosystem degradation. Risk assessment protocols that incorporate both quantitative models and diagnostic variables provide the generality (hence flexibility) and simplicity needed to handle varying data quality and diverse symptoms of risk (IUCN ; Hobday et al . ; Keith et al . ). The trade‐offs for realism and precision will depend on how well the diagnostic variables quantify degradation. Four options for assessing the severity of functional decline were evaluated during RLE consultation workshops: unstructured qualitative ranking; aggregated indices of health/condition; one or a few prescribed generic ecosystem variables; and assessor‐defined ecosystem‐specific variables. Many existing protocols rank the severity of functional decline using unstructured qualitative methods (Nicholson et al . ). This is the least consistent, transparent and repeatable option because assessments cannot be effectively calibrated. Aggregated indices or multivariate measures of ecosystem health or condition may appear more consistent, but missing data, aggregation artefacts, threshold behavior and averaging effects reduce their sensitivity (e.g., Jennings ; Gorrod et al . ), and different types of ecosystems are likely to require structurally different indices (e.g., marine vs. terrestrial). The other two options focus on particular variables correlated with functional decline. They have complementary strengths and weaknesses. The prescriptive option (as suggested by Boitani et al . ) requires estimates of trends in one or a few preset, generic variables such as productivity, species richness, or diversity. This appears to promote consistency, as all assessments use a common ‘currency’ to measure ecosystem degradation. The more flexible option, which permits assessor‐defined ecosystem‐specific variables, trades some of this apparent consistency for increased generality and realism, measuring more direct proxies of the ecological processes that threaten characteristic native biota. Thus, functionally contrasting systems may be assessed using different variables, which are scaled using an inferred threshold of collapse for ranking against a common set of ordinal threat categories (e.g., Figure 6 in Keith et al . ). Examples of suitable assessment variables include: measures of structural complexity or abundance of foundation species in forest or reef systems threatened by disturbance (Alvarez‐Filip et al . ; Burns et al . ); trends in hydrological or water quality variables where there is evidence of biotic dependency in wetlands threatened by water extraction or pollution (Lester & Fairweather in Keith et al . ; Pisanu et al . ); and trends in the abundance of top predators or other species that provide crucial links in food webs in systems sustained by top–down regulatory processes (Schmitz et al . ; Estes et al . ). The consultation process identified four reasons why gains in generality and realism of an ecosystem‐specific approach should be worth an apparent reduction in consistency. First, scaled ecosystem‐specific response variables should be more sensitive and consistent indicators of ecosystem degradation than generic variables when the relationship between the latter and the persistence of characteristic native biota is noisy or inconsistent across different ecosystems (e.g., Adler et al . ; Hooper et al . ). Prescribed response variables are workable for a narrow range of processes or ecosystem types (e.g., exploited fisheries; Hobday et al . ), but impose undue constraints for risk assessments across a wide range of ecosystems and processes. Second, the scaling of ecosystem‐specific variables promotes consistency between assessments, so long as inferred thresholds of collapse are evidence‐based and calculations of the relative severity of declines incorporate the uncertainties. Third, an ecosystem‐specific approach is less limited by data availability than a prescriptive one‐size‐fits‐all approach which requires the same variables to be quantified across all ecosystem types. Finally, ecosystem‐specific approaches promote critical examination of the evidence in diagnosing causes of decline, the pathways of collapse, the most sensitive means of measuring decline along those pathways and setting an explicit threshold for unacceptable loss of characteristic native biota (Figures and ). Further improvements in consistency should be possible by narrowing the range of ecological variables deemed appropriate for assessing particular types of ecosystem degradation, based on accumulating empirical experience. This will be increasingly possible as the RLE criteria are applied to many contrasting ecosystems by a diverse community of expert assessors. Imposing a prescriptive approach too early would stifle the exploration and evidence gathering essential to further development. In the meantime, published assessments provide guidance, and an IUCN peer‐review process for global assessments will encourage consistency and promote rigor across assessments (IUCN ; Rodriguez et al . ). Delimiting ordinal categories of risk The RLE criteria use numerical decision thresholds to assign ecosystem types to ordinal categories of risk (Appendix 1; Keith et al . ), a feature contributing to the simplicity of risk assessment outputs. A distinction exists between decision thresholds , which specify the conditions required to invoke a decision (e.g., categorization), and ecological thresholds , which are theoretical devices to explain nonlinear ecosystem behavior (Martin et al . ). The rationale for each RLE criterion derives from theory, while utility is the primary consideration for setting particular threshold values that delimit threat categories within each criterion, i.e., to make the categories informative about relative levels of risk (Keith et al . , p. 10). Calls for a theoretical basis for decision threshold values (Boitani et al . ) thus confuse them with ecological thresholds. Utility was also important in setting thresholds in the RLTS. For example, the threshold delimiting Vulnerable from least concern for criterion A (population reduction) in the RLTS was increased from 20% (in v2.3) to 30% (v3.1, IUCN ) because of practical concerns that high proportions of some taxa would exceed the lower threshold, not because of new developments in population theory. Category thresholds may similarly require adjustments in the RLE as case studies accumulate. In addition to considering utility in the RLE, the thresholds of spatial criteria (B1 and B2) are scale dependent (Nicholson et al . ). The scaling of these thresholds in different subglobal domains requires research to support subglobal‐scale RLE assessments. Current and potential RLE applications The RLE criteria may be applied systematically to a set of ecosystem types within a specified domain (global or subglobal) or to single ecosystem types. The applications described below are for systematic assessments, but some may also be relevant to assessments of single ecosystems. The IUCN will lead the development of a global RLE, but will also support stakeholders to develop national and regional RLEs (IUCN ), as these are the scales where many resource management and development decisions are made. Red Lists of Ecosystems (sensu lato) are already part of the legislative or policy infrastructure for biodiversity conservation in some countries (Table ). IUCN's new global standard is influencing federal jurisdictions such as Australia and the European Union, where member states are aiming to harmonize their listing methods (Nicholson et al . ). The IUCN RLE criteria also provide a benchmark for other countries wanting to develop their policy infrastructure (Government of Rwanda ). The RLE can contribute significantly to monitoring and reporting the status of biodiversity. Some national reporting frameworks on threatened ecosystems are already in place (e.g., Riecken et al . ; Australian State of Environment Committee ; Driver et al . ). Tools such as the Red List Index (Butchart et al . ) promote temporal consistency. Internationally, global and subglobal RLEs could contribute to reporting instruments such as the Aichi Targets under the Convention on Biodiversity (especially target 5), the International Platform on Biodiversity and Ecosystem Services, and the evolving Sustainable Development Goals (IUCN ). A third application of RLEs is in the design of protected area networks, where they can identify locations of poorly reserved ecosystems that would benefit most (in terms of risk reduction) from inclusion in protected areas (Margules & Pressey ). The presence of threshold proportions of threatened and restricted‐range ecosystems are proposed criteria for the identification of key biodiversity areas as sites contributing significantly to the global persistence of biodiversity (IUCN ). Fourth, RLEs can inform prioritization of investment decisions for conservation, restoration and development, for example, as inputs to cost efficiency analyses in allocating national or corporate budgets (Carwardine et al . ), and by contributing data to financial institutions for assessing investment risks for development projects (e.g., see Equator Principles, http://www.equator‐principles.com/ ). RLEs can support adaptive management strategies (Williams ), for example by informing sustainable allocations of water resources to environmental flows and agricultural production, or the zonation and quotas in fisheries management areas (Green et al . ). While the status of ecosystem types is the primary output of assessments, this is founded on an evidence‐based diagnosis of threats and derivation of ecosystem‐specific diagnostic variables that should be suitable for monitoring responses to alternative management actions. RLEs may also inform the sustainable management and delivery of ecosystem services. The relationship between the RLE (for which the primary goal is to assess risks to ecosystem‐level biodiversity) and ecosystem services is complex because the latter depend on both ecosystems and social environments (Costanza et al . ). To enhance clarity of purpose and simplicity of the RLE protocol, the RLE criteria focus on the loss of characteristic native biota (genes, populations, species, assemblages, and functional groups) and disruption to ecological processes that sustain it. Additional tools may thus be required to evaluate risks to ecosystem stocks, functions and services, especially if novel ecosystems deliver new or improved services relative to their collapsed predecessors. Nonetheless, there are cases where the RLE assessments can provide important information about ecological changes that have major consequences for ecosystem services (Micklin ). The same causal factors that drive loss of biodiversity may also result in decline of ecosystem stocks, functions, and services (Cardinale et al . ). For example, Burns et al . ( ) assessed the Red List status of a forest ecosystem that had undergone major transformations in structure due to timber harvesting and successive canopy fires. At the crux of its Red List status was a loss of characteristic native biota associated with the declining abundance of large old trees, including mammals and birds that shelter and breed in tree cavities, insectivorous birds that forage in large canopies and under loose bark on large tree trunks, and poorly documented microbial and fungal diversity associated with deep forest litter. The same widespread transformation from old growth to regrowth that underpins the Red List status of this forest type has major implications for hydrological productivity (Kuczera ), carbon stocks and sequestration (Keith et al . ), timber production, as well as overall resilience of the system (Lindenmayer et al . 2011). Although the RLE criteria are focused on biodiversity, the outcomes of RLE assessments clearly have broader implications if interpreted with appropriate rigor and caution. Finally, the RLE offers great potential for environmental education through extension programs, media articles, and resource materials for schools and community groups ( http://www.iucnredlistofecosystems.org/ ). The RLE is at an early stage of development but has already generated considerable public interest (Marshall ). Conclusions and future directions In risk assessments generally, few hazards are defined precisely (Burgman ). Greater consistency could be achieved in the RLE through more prescriptive approaches to assessment (Boitani et al . ), but will these gains be worth the necessary trade‐offs in generality and realism? The issue is not whether the concepts of ecosystems and collapse are imprecise or inconsistent (they are, undoubtedly) but whether the working definitions are useful and produce valuable outcomes. That is, will the RLE criteria (Keith et al . ) lead to better conservation decisions than would be achieved without them? In sum, published assessments to date and our global consultation process suggest that the RLE framework achieves a workable balance among inevitable trade‐offs (Keith et al . ), though we welcome empirical evidence indicating where a better balance is possible. We suggest a way forward by developing the RLE based on: (1) learning from systematic applications of the criteria across different ecosystem types and scales; (2) updating application guidelines regularly; (3) classifying and describing ecosystem types systematically as an integrated part of RLE assessments; (4) implementing an active research and development agenda; and (5) refining the conceptual framework and working definitions. In this, adaptive development of the RLE can benefit much from the history of the RLTS which, as one of IUCN's flagship products, was trialed and adjusted iteratively over 50 years from rudimentary beginnings (Scott et al . ). Acknowledgments We are grateful to: the MAVA Foundation, Gordon and Betty Moore Foundation, the Australian Research Council (LP 130100435) and the ARC Centre of Excellence for Environmental Decisions for their support, and to all participants in the RLE consultation process for their willingness to engage in challenging and fruitful discussions. Peter Letten and Julian Keith drafted Figure . Thanks also to June Keith. We thank two anonymous reviewers for their constructive comments.

Journal

Conservation LettersWiley

Published: May 1, 2015

There are no references for this article.