Wildfires are a major hazard in Portugal. The current policy approach has prioritized firefighting capacity rather than preventing fires or reducing human exposure and vulnerability. As regards fuel management for fire prevention, a top-down commandand-control approach has been adopted with weak implementation and few results. The expected impacts of climate change and rural population decline lead to anticipate an increased severity of this problem if the current policy approach persists. The wildfire problem and its possible solutions have mainly been framed in technical terms, which tends not to take seriously into account the fact that implementing these solutions means changing thousands of people’s current practices.

This project aims at refocusing the wildfire problem by assuming that many of those practices are rational within the current policy and socioeconomic setting. Understanding people’s current choices and how policy can actually change them is key to design effective risk-reduction policies. The proposed research is aimed at developing and testing a new analytical framework to support the development and evaluation of new, people-centred policies for wildfire risk-reduction. The development of this analytical framework will be carried out by linking, in an interdisciplinary way, 3 separate areas of research. The first is risk analysis models, which have been developed to factor in the main drivers of the overall level of damage caused by wildfires. Many drivers of fire hazard or exposure are the result of human choices, which are nevertheless not explicitly addressed in risk models. Choice models, which predict how people would choose under specific policy circumstances, is thus the second research area. Linking choice models to risk models will allow the team to directly link policy to people’s choices on e.g. fuel management, which will then lead to specific land cover to be factored in as a driver of fire hazard. This is one of the most innovative aspects of the proposed research. Choice models will be estimated from data collected from survey-based choice experiments in which forest owners and other land managers will be invited to select their preferred management options under a specific policy environment. Linking choice models to hazard levels and risk models will allow us to simulate the costs and risk-reduction potential of different policy options, and thus to compare them on cost-effectiveness grounds. Choice models are not good at exploring new policy options or including the knowledge of relevant stakeholders. This is why the proposed research includes a significant role for stakeholder involvement in understanding wildfire-relevant practices, coproducing effective policy solutions and deliberating on alternative policy scenarios. Stakeholder involvement is our third research area. A relevant innovation here is developing stakeholder deliberations on model-based policy scenarios (linked riskchoice models) and thus combining analytical and participatory work in cost-effectiveness evaluation of policy options. The study area used to develop and test the analytical framework is the region of Pinhal Interior, a hilly region where forest and scrubland dominate the land cover, forest ownership is very-small-scale, population density is low and people live in dispersed settlements, as ‘islands in a forest sea’. This was one of the most seriously hit areas in the wildfires of 2017. This project will advance the state of the art mostly through expanding the interdisciplinary boundary of the scientific knowledge about wildfires and wildfire policy, by linking together those 3 key scientific areas, in addition to the abovementioned, more specialized innovations.

The main result of the proposed research is a tested new analytical framework to support the development and evaluation of new, integrated and people-centred policy approaches to wildfires. A policy brief, a decision-support platform and several guidelines for best practices on adaptation and emergency self-defensive strategies will be produced.