Logging Amazon forest increased the severity and spread of fires during the 2015–2016 El Niño

Forest fires degrade Amazon forest and its natural functions. Logging, deforestation and the increased frequency of prolonged droughts have contributed to the high recurrence of forest fires in the Amazon. Fires have impacted areas that, until recently, were considered immune to fire, such as the southern portion of the Brazilian state of Roraima, which is characterized by forest types that occur in environments with high natural humidity but that are now strongly impacted by selective logging (SL). The objective of this study was to determine the severity and spread of fire in the forests of southern Roraima, taking as a reference the great forest fire that occurred during the 2015–2016 El Niño. We mapped fire scars and forest biomass from remote sensing and data from forest inventories in a 6657.3 km2 study area, of which 6512.4 km2 (97.8%) had originally been forest and 5412.3 km2 (81.3%) was still forest in 2016. The 2015/2016 fires affected an estimated at 682.2 km2, or 12.6% of the area that was still forest in 2016. Vulnerability maps of the forest were made using the weights-of-evidence method. The biomass impacted by fire totaled 26.4 × 106 Mg, representing 9.5% of the total mapped for the study area (277.4 × 106 Mg). The biomass killed by the fire totaled 5.9 × 106 Mg, representing 22.3% of the biomass affected by the fires. The highest level of fire severity (very strong) proportionally affected 84.6% more forest biomass inside than outside SL areas. Forest vulnerability to fires increased by 265.5% in terms of area and by 400.7% in terms of biomass when exposed to SL. Logging also increased the severity of fires when they occurred: a hectare of burned forest was 85.9% more likely to have a “very strong” fire if it had been previously logged, and burned areas that had been logged lost, on average, 2.9% more of their pre-fire biomass to the fire than those that had not been logged (86.5 Mg ha−1 versus 84.0 Mg ha−1). Considering only the ombrophilous forest, the mean biomass of forest that was logged and burned was 310.7 Mg ha−1, or 30.8% lower than the mean biomass of 448.7 Mg ha−1 in logged but unburned areas, showing a substantial biomass loss to fire (average of 138.0 Mg ha−1). SL more than doubled the impact of fire on biomass loss as compared to the impact of the logging itself. In addition to its contribution to carbon emissions and other impacts, the amplifying effect of SL on forest fires indicates that the assumption that authorized forest management projects in Amazonia are sustainable is unwarranted. The future role of this practice should be rethought, existing projects should be subject to close inspection and control, and unauthorized logging should be identified and repressed. The policy of allowing sale of wood from clearcutting projects should be rethought because it provides a loophole for laundering wood from illegal logging.