Active fires during the COVID-19 lockdown period in the Llanos ecoregion, northern South America

Tropical savannas are the biome with the highest fire occurrences worldwide and play a key role in fire carbon emissions dynamics at regional to global scales. During the past decades, however, climate change and land use management have altered their fire regimes via fire suppression or ignition related to conservation and agricultural practices, and extreme weather conditions, among others. In particular, the ongoing COVID-19 pandemic has modified human activities in both urban and rural environments, and thus provides an opportunity to study the interactions between socio-economic and biophysical drivers of fires. Using satellite-based observations, we analyze the spatio-temporal patterns of active fires (AF, from MODIS-MCD14ML) in the Llanos ecoregion (northern South America between Colombia and Venezuela) during the COVID-19 lockdown period (mid-March to December 2020). We also examine fire carbon emissions (from GFED4s) as well as monthly precipitation (from CHIRPS), maximum temperature, and vapor pressure deficit (VPD, from TerraClimate). Our results show that 2020 was the year with the highest number of AF (>60%) and fire carbon emissions (>50%) compared to the 2001 to 2019 average. We found that these increases occur mainly during the peak of the fire season (March and April), which corresponds to the beginning of the lockdown period in Venezuela (March 17) and Colombia (March 20). Pixels (at 0.05° resolution) with significant positive AF anomalies (p<0.05) occur primarily in Venezuela and over grassland and agricultural land covers. A large proportion of these pixels interact with significant positive anomalies (p<0.05) in VPD (>70% of pixels) and maximum temperature (>50%) in March and April. Furthermore, our results highlight that the increase of AF could be associated not only with potential changes in land use management but also with weather patterns anomalies during the lockdown period in the Llanos ecoregion.