Documenting the state of adaptation for the global stocktake of the Paris Agreement

Article 7, paragraph 14 of the Paris Agreement to the United Nations Framework Convention on Climate Change commits Parties to create a five yearly assessment of observed adaptation to track progress and enable appropriate future commitments through the Nationally Determined Contributions and National Adaptation Plans. No large‐scale study exists that shows the types of adaptation, the spatial distribution of types of adaptation, and the numbers of people engaging in that adaptation. To address this gap, and to feed into debates about the modalities for the global stocktake, in this paper we propose a new “stocktaking” approach to document the spectrum and prevalence of observed adaptation over large scales. The four‐step stocktaking approach focuses on: (a) obtaining consensus on the objectives of adaptation; (b) agreeing the sources of evidence; (c) agreeing the search method; and (d) categorizing the adaptations. By focusing on documenting rather than evaluating adaptation, the simple approach avoids some of the adaptation heuristic traps. With guidance to countries on how to operationalize, this approach could improve the transparency of adaptation data collection and analysis, ensure comparability of findings across space and time, and inform the Adaptation Communications (Article 7.10)—a prerequisite to strengthening future ambition commitments within the Paris Agreement.

adaptation, through the Paris Agreement (Christiansen, Martinez, & Naswa, 2018). Lack of consensus on the nature of adaptation is highlighted by the focus on planning through instruments such as the National Adaptation Programmes of Action and, more recently, the NAPs. NAPs have focused on reducing vulnerability to climate change and facilitating the integration of adaptation into policies and programs (LDC Expert Group, 2012). However, they have done so without first generating a baseline assessment of the types and distribution of adaptation practice. Attention has been given to financing adaptation as outlined in the NAPs, without due consideration of the effectiveness (or equity) of the plans in enabling adaptation . The Paris Agreement advances a broader framing of adaptation, that is within the content of multiple stressors, but provides little detail on what this means in practice (Lesnikowski et al., 2017).
Conceptual challenges are exacerbated by unstated normative assumptions about the nature of adaptation and its effectiveness. Preston and colleagues argue that assumptions about adaptation such as "adaptation should generate positive benefits," "adaptation is different to coping," or "reactive adaptation is inefficient" create confusion over the interpretation of the term, and constrain research and practice (Preston, Mustelin, & Maloney, 2015). Undefined relationships between concepts such as development, coping, and maladaptation add to the confusion, leading to questions such as: should development initiatives that address the current adaptation deficit be considered as adaptation, and should coping, maladaptations, and serendipitous adaptations be included as adaptations (Agrawal & Lemos, 2015;Berman, Quinn, & Paavola, 2012;Dilling, Daly, Travis, Wilhelmi, & Klein, 2015)? Answers to these questions have, in the past, influenced how wide the net is cast and what is included as an adaptation. We argue that lack of clarity on these questions should not impede progress on documenting a baseline of adaptation.
Yet, to date, without conceptual consistency on adaptation, it has proven difficult to develop a methodology for empirical analysis. The vast majority of documented adaptations come from individual research projects and programs, for example, as documented in IPCC Fifth Assessment Report by Working Group II on Impacts, Adaptation and Vulnerability (Field et al., 2014). The IPCC reviews some adaptations, but it neither describes the spectrum nor the prevalence of actions. Attempts have been made to document the nature and distribution of adaptation, notably by the UNFCCC, which has created databases of examples of adaptation. For example, the Adaptation Knowledge Portal, through the Nairobi Work Programme, documents examples of adaptation by theme and region; the Private Sector Initiative database compiles examples of private action on adaptation; and the local coping strategies database provides examples of household and community adaptations. Yet all of these databases rely on voluntary submissions of evidence. They thus suffer from being self-selecting and are not comprehensive, hence preventing any assessment of the frequency or the spatial spread of adaptation activity.
Methods used for adaptation reporting have been criticized for focusing on vulnerability assessments and intention to act, rather than observed adaptation (Berrang-Ford, Ford, & Paterson, 2011). For example, researchers document assets deemed to create adaptive capacity, without consideration of whether and how such latent capacity translates into adaptation outcomes (Mortreux & Barnett, 2017). In comparison researchers focusing on tracking, monitoring, and evaluating adaptation tend to prioritize measuring progress toward delivery of adaptation policy or the effectiveness of the process, and not on documenting evidence of the amount and distribution of adaptation by individuals, communities, or sectors (Brooks et al., 2011;GEF, 2014;Leiter, 2017;Lesnikowski, Ford, Biesbroek, Berrang-Ford, & Heymann, 2016).
The conceptual and methodological ambiguity has constrained discussion on the collation of evidence of adaptation at a large scale (e.g., national or ecosystem scale). No formal guidance exists on how to search for, collate and document adaptations. This has consequences for the international policy process and, particularly, how the global stocktake will be undertaken. Although the debate on the meaning of adaptation may continue, it should not hinder development of a replicable and usable method to document who is (and how many are) adapting to what and why, and where this adaptation is occurring.

| CHARTING A WAY FORWARD
Given that the Paris Agreement mandates a global stocktake, this leads us to ask: what opportunities are there for improving documentation of adaptation at large scales, for example, across nations, regions and social-ecological zones? Emerging research into adaptation heuristics attempts to create conceptual consistency by asking researchers to clarify the objectives, purpose, and expectations of adaptation Preston et al., 2015). Instead of focusing on the usual heuristics (e.g., "adaptation is local," "predict and respond"), and being constrained by the notion that adaptation has to be "incremental," we propose a way forward by seeking consensus on the broadly agreed objectives of adaptation to multiple stressors: (a) to reduce socioeconomic vulnerability, (b) to address disaster risk, and (c) to support social-ecological resilience. Starting with these accepted objectives avoids becoming enmeshed in debates about defining adaptation that have remained unresolved since the 1990s, and that continue to stall progress on documenting and understanding adaptation. These three objectives are also consistent with principles of sound development, as espoused in the Paris Declaration on Aid Effectiveness (2005), and support core elements of the 2030 Sustainable Development Goals (such as 1: No Poverty; 11: Sustainable Communities, and 13: Climate Action).

| THE " STOCKTAKING" APPROACH
To provide a starting point to create baselines of adaptation that can say something about the range and prevalence of adaptation, we propose a "stocktaking" approach. This approach comprises four steps to track and monitor observed adaptation over large scales: (a) Step 1: obtain consensus on the objectives of adaptation; (b) Step 2: agree sources of evidence; (c) Step 3: agree search methods; and (d) Step 4: categorize the adaptations (Figure 1). By focusing on documenting rather than evaluating adaptation, the simple framework avoids some of the challenges documented above that hinder progress. The stocktaking approach acknowledges two key aspects of adaptation data collection: (a) the numbers of people who adapt and (b) where and by whom the adaptation occurs. Combined, this information allows some weighting as to the distribution of the adaptation. The second aspect further allows the adaptations to be linked to spatial databases allowing both visualization and potentially other forms of analysis. This shows how starting the stocktaking approach could develop and incrementally lead to growing insight and linkages with other adaptation and development activities.

| Step 1: Obtaining consensus on the objectives of adaptation
Debates about what constitutes adaptation have continued for decades. Instead of requiring a specific definition of adaptation, there are merits in focusing on the objectives of adaptation, where more consensus exists as to what are worthwhile goals (Eakin, Tompkins, Nelson, & Anderies, 2009;Ensor & Berger, 2009;McGray, Hammill, & Bradley, 2007). Adaptation objectives broadly include: (a) reducing socioeconomic vulnerability, (b) addressing disaster risk, and/or (c) supporting socialecological resilience. The first two elements have been widely reported in IPCC reports since 2001 (Adger et al., 2007;Noble et al., 2014;Smit et al., 2001), although evidence of measures to support or enhance social-ecological resilience is limited (Ranger, Reeder, & Lowe, 2013). Collectively these three elements describe the broad scope of adaptation and can be linked as shown in Figure 2.
Reduction of present vulnerabilities is the foundation of adaptation as it focuses on reducing poverty by addressing structural issues which, evidence shows, inhibit adaptation action (Duncan, Tompkins, Dash, & Tripathy, 2017). Despite this empirical evidence, it appears that much adaptation action is having little impact on vulnerability (Atteridge & Remling, 2018 The three building blocks for adaptation context. For example, reducing vulnerability to drought in Iran or Niger (Asfaw, Pallante, & Palma, 2018;Keshavarz, Maleksaeidi, & Karami, 2017) looks very different to addressing physical vulnerability in Grenada (Canevari-Luzardo, Bastide, Choutet, & Liverman, 2017). Following Roberts and Pelling (2018), we argue that reducing vulnerability by addressing the structural issues that entrench poverty can create the space for individuals to build adaptive capacity . Indeed, without this, Atteridge and Remling (2018) argue that vulnerable communities could become more marginalized due to risks or vulnerabilities being transferred to them through others' adaptation actions. Once vulnerability reduction is occurring, it becomes more effective to engage in present and future disaster risk reduction (DRR). Together, these actions collectively form the foundation to build and enhance long-term social-ecological system resilience. All of these elements, we argue, are worth documenting as adaptations.
Obtaining consensus on these concepts is a vital first step. Clarity over objectives also helps generate the key words and synonyms, for the search for evidence on adaptation. To capture the range of adaptations, a comprehensive and replicable search should use search terms that represent the objectives of adaptation. Tiered searches, where terms from each tier are combined, may be the most useful to ensure that the population of literature is identified. A first tier could comprise synonyms for the area/social-ecological system under review (e.g., nation, regional, or ecosystem). A second tier might consider variations of the terms for adaptation, for example, adapt, risk, vulnerability, coping. A third tier might consider the nature of the hazards, for example, weather, hazard, climate, variability, change.

| Step 2: Agree the sources of evidence
Adaptation is documented in a variety of media, online, offline, and printed, including technical reports from the private, third, and public sectors; journal articles; academic and multilateral agency working papers; and various submissions to the UNFCCC, to name a few. We propose that all secondary sources of information are searched to deliver a complete baseline assessment of adaptation. The success of the search comes from triangulating sources to reduce the risks of misrepresentation and double-counting. Adaptation can occur in any sector, at any scale, and be undertaken by any actor or system prior to, or in response to, a variety of climate-related and nonclimate drivers. This messiness of adaptation adds to the complexity of documentation. For the initial stocktake to reduce the complexity of the task, the search could be narrowed to focus on weathersensitive and natural resource-dependent sectors and locations. Example sectors include agriculture; forestry; fisheries; water resources and flood management; DRR/management; coastal zone management; public health, waste, and sanitation; urban infrastructure and utilities; and rural development (Bierbaum et al., 2014;H.M. Government, 2012).

| Step 3: Agree search methods
We identify three main methods to search for evidence of observed adaptation at a large scale: the "systematic review" approach, the "call for evidence" approach, and the "inventory" approach.
Using established methods (e.g., Moher, Liberati, Tetzlaff, & Altman, 2009), the "systematic review approach" can synthesize findings from a wide range of empirical adaptation research (Berrang-Ford et al., 2011;. The approach uses predefined search terms and search methods, and applies formal inclusion and exclusion criteria to the search (see, e.g., Biesbroek, Klostermann, Termeer, &Kabat, 2013, andPorter, Dessai, &Tompkins, 2014). Focusing only on high-quality published literature means that the systematic review has the narrowest scope of the three approaches; it is least likely to capture unpublished private sector, individual, household, or community level adaptation activity. This approach may be able to offer some insights into spatial distribution of types of adaptation, but is unlikely to offer detail on numbers of people adapting.
The "call for evidence" approach requires adaptation researchers and practitioners to respond to a call to submit evidence of adaptation, such as within IPCC assessments. IPCC chapter authors review submissions in conjunction with their own literature search to assess evidence of adaptation (IPCC Secretariat, n.d.). There is no requirement to document the search method, and outputs of this approach summarize types of adaptation, but do not provide insight into prevalence or distribution of adaptation.
The "inventory" approach, developed by Tompkins et al. (2010), searches multiple sources, including private and public sector gray literature and documents the adaptations using a predefined template shaped around the core questions asked by Smit et al. (2001). For example, questions could be: how are people and systems adapting; how many people are adapting; where are people adapting; what is the purpose and form of the adaptation; what is the temporal and spatial prevalence and distribution of adaptation; who is delivering the adaptation and who is adapting (provider and beneficiary); to what are people adapting/what is driving the adaptation; are there side effects/co-benefits from the adaptation; and what is the cost of delivering the adaptation? By using the broadest interpretation of "literature" and by focusing on collecting spatially and temporally explicit evidence from multiple data sources, this method has the potential to document spatial spread and distribution of adaptation, although it requires careful triangulation to avoid overreporting of adaptation (Tompkins et al., 2010).
All three approaches have been used to document adaptation, yet all three approaches suffer from weaknesses. These include the potential for: double-counting the same adaptation that occurs in multiple sources; sectoral bias as some sectors may document adaptation more thoroughly than others, such as water supply and coastal hazard responses; and publication bias, whereby only adaptations that are published in the sources selected are included. To reduce the risk of publication bias, we propose the use of the systematic review combined with the inventory approach, as together they would include a wide variety of literature, can be applied by non-experts, and can be cross-checked to avoid double-counting. The first search using this approach will inevitably be time consuming; however, if the search is documented carefully (the search engine, the key words and synonyms, the combinations of search terms, and explicit exclusion criteria), then it can be repeated and updated more easily in later years.

| Step 4: Categorize the adaptations
We suggest that the spectrum and prevalence of adaptations are best identified by categorizing adaptation by its objectives (i.e., reducing socioeconomic vulnerability, addressing disaster risk, or building social-ecological resilience-see Figure 2). Drawing on several decades of work on sustainable livelihoods, DRR, and ecosystem services, we argue that: (a) following Department for International Development [DFID] (1999) vulnerability reduction can be measured by changes in people's stocks of assets; (b) following UNISDR (1994) DRR can be measured by investment in disaster mitigation, preparedness, response, and recovery; and (c) following Millennium Ecosystem Assessment (2005) social-ecological resilience can be measured by changes in ecosystem services, that is, provisioning, regulating, habitat, and cultural services. Collectively these 13 components can describe the broad range of adaptations (see Figure 3). In addition to being well-known and extensively trialed, a further merit to categorizing adaptation in this way is the ability to capture changes in both stocks and flows of assets and services that occur as part of adaptation. Careful definition of the three main objectives of adaptation is needed (in step 1) to clarify where to place different adaptations. For example, unless it is clearly explained that natural capital represents stocks of assets that people use, and provisioning services relate to the flows of services that people obtain from ecosystems, there could be confusion between the two categories.
Over the past two decades significant research has been undertaken in each of these three areas, with advances made in categorizing types of activity. The five capital assets within the sustainable livelihoods approach (SLA) reveal the stocks of assets that people draw on to sustain livelihoods, that is, access to: the environment (natural capital); health, education, and well-being (human capital); networks and connections (social capital); money and finance (financial capital); and technology and built resources (physical capital). The four components of the DRR framework within the Hyogo and Sendai Frameworks for DRR (UNISDR, 2005(UNISDR, , 2015 identify changes in both the stocks and flows of DRR: mitigation of long term hazard risk; regular preparatory activity for known hazards; response planning for hazards; and planned post disaster recovery. The four ecosystem services within the Millennium Ecosystem Assessment (2005) recognize the flows of services that ecosystems provide.
Intellectually, each of the three approaches builds on significant conceptual and empirical work dating back at least two decades: SLA (DFID, 1999;Scoones, 1998), DRR (UNISDR, 1994, and ecosystem services (Costanza et al., 1997). In that Components of adaptation time, each approach has not only been critiqued but also counter-critiqued (e.g., Scoones, 2009, on the SLA, andSchröter et al., 2014, on ecosystem services), leading to improved versions of each. Further, all have been applied to document humanenvironment interactions in a multitude of spatial and temporal settings, in relation to a variety of environmental shocks and stresses. To create an initial baseline of adaptation, these extensively tested frameworks potentially provide a useful 13 component categorization of types of adaptation to support the global stocktake ( Figure 3).

| CONCLUSION
With the first global stocktake under the Paris Agreement due to be completed by 2023, and agreement on modalities required by December 2018, continued debates about the definition of adaptation can no longer hamper progress on documenting adaptation. Moving beyond definitional arguments, and using widely agreed objectives of adaptation, we propose a replicable, transparent, and comparable four-step process for tracking and monitoring the state of adaptation. It requires a broad understanding of adaptation before data collection can begin, but it can integrate multiple adaptation types if these are found, leading to a comprehensive assessment of the type of adaptation, who is adapting, where adaptation is occurring and what is driving the adaptation. With careful guidance to countries on how to operationalize this stocktaking approach, this method would improve transparency of process, and enable comparability of findings across space and time. It would also create an analyzable baseline data set on the state of adaptation, and enable tracking and monitoring of progress with adaptation-a prerequisite to strengthening future ambition commitments within the Paris Agreement.