The collapse of a mining dam with 62 million cubic meters of mud in the Rio Doce basin resulted in the destruction of whole communities and large areas of Atlantic Forest. As forest restoration activities are among the most costly conservation strategies, prioritization of restoration efforts is crucial. In the present article, we mapped priority areas for forest recovery in a portion of the Rio Doce Basin (DO1) using a GIS-based (geographic information system) multicriteria decision analysis (MCDA) employing the weighted linear combination (WLC) method. Five factors with different weights were taken into consideration according to their level of importance: distance from the drainage network, distance from the native vegetation patches, slope, soil class and precipitation. A map of priority areas was produced where 1.73% of the area was classified as very high priority for forest recovery, while 5.18% of the area was classified as high priority, 57.88% as medium priority, 1.34% as low priority and 0.00% as very low priority. The highest weights were both for the distance from the drainage network and the distance from native vegetation, revealing that areas of permanent preservation and those closer to forest fragments are priority areas for forest recovery. MCDA is a flexible and easy-to-implement method generating maps with suitable solutions for forest recovery. The approach taken can be replicated in regions that require support for decision making in environmental planning, such as the Pantanal biome, which is under considerable pressure from deforestation for the expansion of pastures.

A Floresta Atlântica é um dos ecossistemas mais fragmentado e explorado. Como atividades de restauração florestal são dispendiosas, a Análise de Decisão Multicritério (ADMC) integrada ao SIG (Sistema de Informações Geográficas) fornece um satisfatório suporte de decisão espacial para produção de mapas de forma eficiente. O colapso de uma barragem de mineração em áreas de floresta Atlântica, resultou na destruição de comunidades por rejeitos de mineração na bacia hidrográfica do Rio Doce. Assim, o objetivo deste estudo foi mapear áreas prioritárias para recuperação florestal na bacia do Rio Doce, Minas Gerais. Utilizou-se a ADMC baseada em SIG, e associada ao método do Processo Analítico Hierárquico (AHP) e Combinação Linear Ponderada (CLP). Cinco fatores foram utilizados com distintos pesos: distância da rede de drenagem, distância do fragmento de vegetação nativa, declividade, classe de solo e precipitação. De acordo com o mapa de áreas prioritárias produzido, 92,69% da área foi classificado como área de importância baixa ou muito baixa para recuperação florestal e, 7,31% como área de média, alta e muito alta prioridade. A ADMC é de fácil implementação, produzindo mapas que podem predizer as soluções adequadas para conduzir ações de recuperação, desde que a base de dados seja fidedigna para obter resultados satisfatórios.Palavras-chave: manejo de ecossistemas; combinação linear ponderada; processo analítico hierárquico. MULTRICRITERIA ANALYSIS TO DEFINE PRIORITY AREAS FOR FOREST RECOVERY IN THE RIO DOCE BASIN, MINAS GERAIS ABSTRACT: The Brazilian Atlantic forest is one of the most fragmented ecosystems and exploited Brazilian biome. As restoration activities are expensive, multicriteria decision analysis (MCDA) integrated with GIS (geographic information system) provide a satisfactory spatial decision support system to efficiently produce maps. The collapse of a mining dam in a region of Brazilian Atlantic forest, resulted in the destruction of communities by a river of mud and mining waste. Thus, the objective of this study was to map and identify priority areas for forest recover in the Rio Doce Basin, Minas Gerais. We used GIS-based multicriteria decision analysis associated with the analytic hierarchy process (AHP) and weighted linear combination (WLC) method in the aggregation of criteria. Five factors were used, receiving different weights: distance from the drainage network, distance from the native vegetation patches, slope, soil class and precipitation. According to the priority areas map, 92.69% of the area was classified as an area of low or very low importance for forest recovery and the remained (2.92%) of the Rio Doce basin was mapped as an area with high and very high priority for forest recovery. The ADMC is easy to implement, producing maps that can predict the right solutions to conduct recovery actions, provided the database is trusted for satisfactory results.Keywords: ecossystem management; linear weighted combination; analytical hierarchical process.