1. Shifting Climate Disasters
Climate-related disasters in the Brazilian Amazon have shown a pronounced increase and shift in prevalence over the two-decade period in this study. Initially dominated by wet events, the region has seen a substantial increase in droughts, heatwaves, and fires since 2005 (Fig. 1a, b). These disasters, defined as significant disruptions caused by hazardous events interacting with population's exposure, vulnerability, and capacity, have led to extensive human, material, economic, and ecological losses. The frequency of wet events, including various types of floods and intense rainfall, increased by 124% from the mid-2000s (2006–2010) to 2018–2022. When comparing the mean annual values from 2000–2005 to 2018–2022, wet events have risen more than fivefold, and landslides have increased by 189% over the same mid-2000s period, showing an eightfold surge. Similarly, drought and heatwave occurrences have tripled when comparing the early 2000s (2000–2005) to 2018–2022, with a 15% increase noted when comparing the mid-2000s (2006–2010) to 2018–2022. Fire events, in particular, have surged by 409% compared to the mid-2000s, resulting in a more than nine fold increase from 2000–2005 to 2018–2022 (Fig. 1b). Other climatic disasters also exhibited significant increases over these periods.
Spatial analysis indicates that while wet events are uniformly distributed across the municipalities of the Brazilian Amazon, droughts predominantly affect the northern (RR), central (AM), and southern and southeastern municipalities (MT, TO, and MA) (Fig. 1a). Fire disasters are primarily concentrated in the municipalities of RR, MT, AC, and TO. Conversely, landslides and other disaster types do not exhibit a clear pattern of distribution across the region.
Figure 1 here
Figure 1: Number of climate disasters along municipalities in the Brazilian Amazon from 2000–2022. a) Spatial distribution of total number of disasters by municipality and disaster typology from 2000 to 2022; b) Temporal distribution of disasters over the years in all municipalities. c) Total number of disasters by climate typology in 2000–2022 (Data Source: Brazilian Atlas of Disasters).
2. Disaster Exposures on the Rise: A Growing Concern for People
Our analysis from 2018 to 2022 reveals a significant increase in the number of individuals affected by disasters, with 1.78 million people, or 6.44% of the region's population, now exposed annually. This marks a dramatic rise from the 2000–2004 period, when only 2,316 people, or 0.11% of the population, were exposed (Fig. 2a). Compared with the 2006–2010 period, 541,180 people, or 2% of the population, were exposed annually.
Wet events accounted for the majority of this exposure, affecting 62.08% of the population, followed by droughts and heatwaves (13.65%), fires (11.73%), landslides (8.7%), and other events (3.84%) (Fig. 2a, b). In years with extreme conditions, wet disasters impacted over 75% of the population in 408 municipalities, accounting for 53% of the region, and were evenly distributed throughout the Region (Fig. 2b). Severe droughts, heatwaves, and fire events affected more than 75% of the population in 130 municipalities, about 17% of the region, with these incidents primarily concentrated in the states of Acre (AC), Mato Grosso (MT), the southwest of Pará (PA), and Roraima (RR) (Fig. 2b).
Figure 2 here
Figure 2: Population exposed to disasters in municipalities of the Brazilian Amazon over the period 2000–2002. a) Temporal distribution of number of people affected by disaster typology (millions/year) and percentage of population exposed considering all disasters; b) Maximum annual percentage of affected people between 2000–2022 and its corresponding disaster typology by municipality.
3. Substantial Economic Impacts from Climate Disasters
Economic losses from climate-related disasters in the Brazilian Amazon reveal a striking increase over two decades. From 2000 to 2022, the total financial damages were substantial, totaling approximately $5.78 billion USD. The annual mean losses have surged from $32.8 million USD during 2006–2010 to $634.2 million USD in the period 2018–2022, marking an increase of 3.7 times (Fig. 3a). Breaking down the losses by disaster type, wet events were the most costly, accounting for 84.78% of total losses ($4.9 billion USD). Droughts and heatwaves followed, incurring 12.1% of the losses ($700 million USD), with landslides, fires, and other climate disasters contributing less significantly (Fig. 3b). Wet disasters dominated losses, particularly in farming (agriculture and livestock), which amounted to $2.77 billion USD (57% of losses from wet events). Infrastructure and housing were also heavily affected, with damages of $962 million USD and $447 million USD, respectively. Health-related impacts, including damage to healthcare facilities and services, totaled $53 million USD. Droughts and heatwave disasters primarily impacted farming, accounting for $650.5 million USD (97% of losses from these events). Health-related costs amounted to $5.87 million USD. Likewise, fire disasters resulted in $64.2 million USD in losses, predominantly affecting farming (92%), with health-related losses totaling $0.58 million USD. Landslide disasters caused significant damage to public infrastructure ($46.5 million USD). Across all disaster types, farming sustained 60.3% of total losses ($3.5 billion USD). Damages to public infrastructure and housing were also notable for wet disasters, totaling $1.1 billion USD and $470 USD, 45%) and housing ($18.9 million USD, 18%), with health-related costs of $1.77 million USD. Landslide disasters resulted in total losses of $101 million USD, with 45% ($46.5 million USD) attributed to infrastructure damages and 18% ($18.9 million USD) to housing (Fig. 3a).
Figure 3 here
Figure 3: Asset losses due to climate disasters in the Brazilian Legal Amazon. a) Total losses from 2000 to 2022, categorized by disaster type and affected sectors. b) Annual economic losses by disaster type. Data source: Brazilian Digital Atlas of Disasters.
4. Small Municipalities and Indigenous Peoples are disproportionately impacted
Our analysis reveals a striking disparity in the impact of climate-related disasters across different-sized municipalities and Indigenous populations within the Brazilian Amazon. The smallest municipalities, defined as those with populations under 50,000 suffered the most severe economic consequences of climate disasters, with losses averaging 9.58% of their economic growth from 2002 to 2020. In contrast, municipalities with populations exceeding 50,000 and 300,000 saw economic losses constituting 3.8% and 1.3% of their economic growth, respectively (Fig. 4d).
Municipal population size can be considered as a vulnerability, because in Brazil, municipalities with less than 20,000 people are not required to have a city master plan to guide the city's development. Often master plans indicate risk areas that should not be occupied. Smaller municipalities also show consistently higher numbers of climate disasters and lower Social Progress Index (SPI) scores compared to larger cities, while host a greater proportion of Indigenous residents (Fig. 4a, b, c).
Our statistical analysis indicates significant correlations between smaller municipalities and the number of Indigenous inhabitants, with a Kendall-tau of -0.063 and Spearman of 0.025 for SPI, and a Kendall-tau of 0.106 and Spearman of 0.16 for the proportion of Indigenous populations. These findings underscore the vulnerability of smaller, and indigenous-rich municipalities to climate disasters and highlight the urgent need for targeted adaptation and resilience strategies to reduce these risks for these populations.
Figure 4 here
Figure 4. Vulnerability to climate disasters based on municipalities' population size. Vulnerability is represented on the X-axis by population size and on the Y-axis as follows: a) number of climate disasters; b) Social Progress Index; c) Indigenous population; and d) Loss to Economic Growth ratio.
5. Compound Risks Across the Brazilian Amazon
Our study highlights significant variations in the multidimensional components of exposure and vulnerability to climate risks across municipalities in the Brazilian Amazon. We have identified distinct levels of compound risk based on the exposure of people and assets and the resultant economic impact. At the highest level of risk, defined by a "red rectangle," nine municipalities, each with a population under 20,000 people, experienced the most severe effects (Fig. 5a, b). In these areas, disasters affected over 50% of the population, and total economic losses exceeded 50% of their economic growth from 2002–2020 (Fig. 5a). This category includes municipalities located across various states: four in Amazonas (AM), two in Maranhão (MA), and one each in Mato Grosso (MT), Pará (PA), and Tocantins (TO) (Fig. 5a, b). The "orange rectangle" represents the second highest level of risk, comprising 187 municipalities (24.2% of the region) where disasters affected more than 50% of the population, yet the economic losses were less than 50% of the economic growth. This group is geographically diverse, with 119 municipalities (15% of the total) having populations under 20,000 (Fig. 5a, b). The "green rectangle" indicates a medium level of risk and includes just two municipalities located in eastern Amazonas, along the Amazon River, with populations under 50,000. Here, the climate disasters affected less than 50% of the population, but the economic losses totaled more than 50% of their growth between 2002 and 2020. The lowest level of risk, defined by a "blue rectangle," includes 575 municipalities (74.5% of the region). In these areas, climate disasters affected less than 50% of the population, and the economic losses were less than 50% of their economic growth. A majority, 355 municipalities (50%), have populations less than 20,000 and are spread across the Amazon.
Figure 5 here
Figure 5: Compound risk in each municipality, analyzing both population exposure in the year of the most extreme disaster and total economic loss in relation to economic growth. a) Scatterplot used to classify municipalities into 4 different levels of risk: red (high/high), grouping municipalities in which more than 50% of the population was affected in the most extreme disaster year and in which total economic losses surpassed 50% of municipality economic growth; yellow (high/low), grouping municipalities in which more than 50% of population was affected in the most extreme disaster year, but total economic losses were less than 50% of municipality economic growth; green (low/high), comprising municipalities with less than 50% people affected in the most extreme year and with total losses of more than 50% of economic growth; blue (low/low), comprising municipalities with less than 50% people affected in the most extreme year and with total losses of less than 50% of economic growth. b) map showing the spatial distribution of compound risk (classification obtained in a).