Contents lists available at ScienceDirect Ecological Indicators journal homepage: www.elsevier.com/locate/ecolind A long-term assessment of fire regimes in a Brazilian ecotone between seasonally dry tropical forests and savannah Daihana S. Argibay ⁎ , Javier Sparacino, Giovana M. Espindola PRODEMA, Federal University of Piauí, 64049-538 Teresina, Brazil ARTICLE INFO Keywords: Seasonality Fire recurrence Caatinga Cerrado Precipitation ABSTRACT Fire is a recurrent phenomenon in semiarid seasonal ecosystems. The study of the relationship between fire and climate could provide new approaches for understanding fire dynamics in semiarid regions, which could be useful for strategically managing the loss and recovery of natural resources under threat. Here, we evaluated the patterns of fires occurrence in a semiarid Caatinga-Cerrado ecotone in the Northeast Region of Brazil by char- acterizing their spatiotemporal dynamics associated with climatic conditions. Using a temporal series of 306 Landsat images, we mapped a burned area database at a fine spatial resolution (30 m) from 1999 to 2017 in the Capivara-Confusões Mosaic of protected areas and their surroundings. Fire seasonality was analyzed through climatic seasons, distinguishing between the rainy season (RS), early dry season (EDS), middle dry season (MDS) and late dry season (LDS), by analysis of the daily precipitation of ground-based stations in the area. We created yearly, seasonal and recurrence maps of burned areas to assess fire regimes. The results showed that the 48% of the area was burned during the 19-year long period considered. Serra das Confusões National Park (SCoNP) had 302,644 ha burned, Serra da Capivara National Park (SCaNP) had 2056 ha burned and the ecological corridor had 215,718 ha burned. Most of the burned area was registered during the MDS (36% of the study area), there was almost no burned area in the EDS (< 0.01%), while the LDS (7%) and the RS (4%) had similar dimensions burned. The years that burned the most (2001, 2007, 2010, 2012, 2015) were followed by years with con- siderably less burned area. We identified a multiple-year process, consisting of a climatological year with above- average precipitation and consequent low burned area, followed by a climatological year with below-average precipitation and resulting in large extensions of burned area (periods: 1999–2001, 2008–2010, 2010–2012). Fire recurrence reached a maximum value of ten, with 47% of the area burned presenting some degree of recurrence. SCaNP presented no fire recurrence, while SCoNP had a maximum recurrence parameter of four. Areas with moderate fire recurrence were near roads, settlements or cities. Some recurrently burned areas presented similar shapes, which were associated with topography limiting the spread of the fires. The patterns that we described here constitute the first step toward understanding the fire regimes of the region to establish directions for improving management strategies and orienting policies in the area. 1. Introduction Fires alter the structure and composition of vegetation around the world, regulating the distribution of ecosystems (Bond et al., 2004). At the same time, vegetation is one of the key drivers in fire activity, re- garding its productivity (biomass as fuel), flammability (chemical composition and structure), availability to burn (moisture/dryness) and phenology (deciduous, semideciduous or evergreen) (Archibald et al., 2018). Together with vegetation and fuel variation, climate, weather during fire, and ignition rates determine fire activity (Bradstock, 2010). Fire regimes can be described as a particular combination of char- acteristics, such as frequency, intensity, severity, seasonality, size, type, extent and spatial pattern of fire occurrence (Bond and Keeley, 2005). Fire regimes have been modeling the Earth processes for hundreds of millions of years, long before anthropogenic burnings (Scott, 2000). In spite of that, fire regimes are altered by human activities that play a complex role, influencing the number and timing of ignitions, sup- pressing fires, affecting the fuel amount and vegetation connectivity, and indirectly, altering climate (Archibald et al., 2013; Bowman et al., 2009; 2011; Flannigan et al., 2009; Mayr et al., 2018). Worldwide, the spatial variability of fire regimes has been ex- plained, in part, by rainfall seasonality and interannual variability, ef- fective rainfall, the presence of long-term droughts and dry season length (Archibald et al., 2013; Argañaraz et al., 2015; Bradstock, 2010). https://doi.org/10.1016/j.ecolind.2020.106151 Received 18 February 2019; Received in revised form 18 January 2020; Accepted 28 January 2020 ⁎ Corresponding author. E-mail addresses: archibayds@gmail.com (D.S. Argibay), giovanamira@ufpi.edu.br (G.M. Espindola). Ecological Indicators 113 (2020) 106151 1470-160X/ © 2020 Elsevier Ltd. All rights reserved. T