How adaptive capacity shapes the Adapt, React, Cope Response to climate impacts: insights from small-scale fisheries

As the impacts of climate change on human society accelerate, coastal communities are vulnerable to changing environmental conditions. The capacity of communities and households to respond to these changes (i.e., their adaptive capacity) will determine the impacts of climate and co-occurring stressors. To date, empirical evidence linking theoretical measures of adaptive capacity to community and household responses remains limited. Here we conduct a global meta-analysis examining how metrics of adaptive capacity translate to human responses to change (Adapt, React, Cope Response) in 22 small-scale fishing case studies from 20 countries (n=191 responses). Using both thematic and Qualitative Comparative Analysis, we evaluate how responses to climate, environmental, and social change were influenced by domains of adaptive capacity. Our findings show that adaptive responses at the community level only occurred in situations where the community had Access to Assets, in combination with other domains including Diversity and Flexibility, Learning and Knowledge, and Natural Capital. In contrast, Access to Assets was nonessential for adaptive responses at the household level. Adaptive households demonstrated Diversity and Flexibility when supported by strong Governance or Institutions and were able often able to substitute Learning and Knowledge and Natural Capital with one another. Standardized metrics of adaptive capacity are essential to designing effective policies promoting resilience in natural resource-dependent communities and understanding how social and ecological aspects of communities interact to influence responses. Our framework describes how small-scale fishing communities and households respond to environmental changes and can inform policies that support vulnerable populations.Please see uploaded ReadMe file, as well as Table 1. We also submitted the DOIs of the publications (case studies) that we coded to retrieve the data used for the QCA.This data set was collected through a systematic review of the literature used to select case studies related to small-scale fisheries and adaptive capacity and used for a crisp set Qualitative Comparative Analysis. First, we conducted a Google Scholar search using the following strings ("small-scale fisheries" OR "artisanal fisheries" OR "subsistence fisheries") and ("social ecological systems" "community resilience environmental change response OR adapt") to identify instances of SSF responses to biological or physical stressors. The first set of terms allowed us to focus on the responses to stressors of SSF, defined by Teh and Sumaila (2013) as: (1) targeting species for household consumption or local markets; (2) occurring at a relatively low level of economic activity; (3) minimally mechanized; (4) operated nearshore; (5) performed by the fisher/fisher's family; (6) minimally managed; and/or (7) undertaken for cultural or ceremonial purposes. The second set was used to find studies where empirical responses to stressors (verified through the case study methods) were documented for social ecological systems, as we were interested specifically in the adaptive capacity of intertwined human-natural systems. Second, we reviewed abstracts for the initial 1,650 search results. We excluded non peer-reviewed primary research articles (i.e. government reports, unpublished theses, book chapters, and synthesis papers), that were unrelated to SSF, or that focused on hypothetical adaptive capacity rather than empirical assessments of responses. This yielded 85 potential case studies for further evaluation. Third, we limited our final analysis to case studies that met two criteria: (1) explicitly examined empirical adaptive response of SSF communities or households to a specific stressor (or set of stressors); and (2) considered situations where at least one primary stressor was biological or physical. The inclusion of biological or physical stressors ensured that our work was relevant to climatic adaptation. Responses were scaled by community and household to determine if domains influenced adaptive capacity response differently at these two levels. This selection process yielded examples of responses from 22 case studies. Once case studies were collected, they were coded according to a coding database. Specifically, we coded each paper for responses to single stressors (or multiple stressors) into three response categories coined the ARC Response: Adapt (proactive planning and/or collective action); React (unplanned response, but some action was taken); and Cope (no action, a passive acceptance) (Bennett et al. 2014). We considered the ARC Response at two levels: community and household/individual (hereafter "household"). We assigned these levels according to the case study authors' description. Responses at the level of an individual or household were assigned to the household level (e.g. individual or household choice to buy a new type of fishing gear). Responses at levels of organization broader than a household, were assigned to community (e.g. fishing associations that pursued aquaculture farming). Building on the existing adaptive capacity literature, we also coded for presence or absence of four broad categories of domains anticipated to influence adaptive capacity: Diversity and Flexibility, Access to Assets; Learning and Knowledge, Governance and Institutions (Adger et al. 2003b; Brooks et al. 2005; Allison et al. 2009; Bennett et al. 2014; Whitney et al. 2017; Table 1). Additionally, we added a fifth category—Natural Capital—stock(s) of natural resources generating ecosystem goods and services (Costanza et al. 1997). Each of the five domains was composed of specific determinants (Table 1). If any determinants were reported to be present, the parent domain was considered present. If none of the determinants were reported, we considered the parent domain absent. We recorded "not available" (N/A) for determinants/domains that were not described by the original authors to assure the coders did not infer associations that were not directly reported. Initially, we planned to code responses at the determinant level, but few case studies recorded determinant detail across all domains. Thus, we analyzed responses at an aggregated level