Record-breaking fires in the Brazilian Amazon associated with uncontrolled deforestation

Small forest patches and landscape-scale fragmentation exacerbate forest fire prevalence in Amazonia

Over recent decades, forest fire prevalence has increased throughout the tropics, necessitating improved understanding of the landscape-scale drivers of fire occurrence. Here, we use MapBiomas land-cover and fire scar data to evaluate relationships between forest fragmentation, land-use, and forest fire prevalence in a typically consolidated Amazonian agricultural frontier: Portal da Amazonia, Mato Grosso, Brazil. Using zero-/zero-one-inflated Beta regressions, we investigate effects of forest patch (area, shape, surrounding forest cover) and landscape-scale variables (forest edge length, land-cover composition) on forest fire occurrence and density between 1985 and 2021. We show that fire density was greatest in small, complex forest patches. Small patches (≤100 ha) were also the dominant contributors to annual, regional forest fire cover. At the landscape-scale (100 km2), forest edge length and urban land cover were positively associated with forest fire occurrence and density. Furthermore, forest fires were most likely to occur in landscapes consisting of ∼45% pasture cover, while fire density increased roughly linearly with pasture cover. Cropland cover was negatively associated with forest fire occurrence and density. Our findings indicate clear links between forest fragmentation and increased forest fire prevalence. This is cause for global concern, given that fragmentation rates throughout Amazonia are increasing, and fires are eroding the Amazon's capacity to act as a carbon sink. Efforts to minimise further fragmentation within Amazonia would likely help reduce forest fire prevalence. Within already fragmented regions, the conversion of pasture into crops, alongside targeted efforts to suppress fires within small forest patches and urbanized areas, may also limit fire prevalence.

Disentangling the seasonality effects of malaria transmission in the Brazilian Amazon basin

The evidence of seasonal patterns in malaria epidemiology in the Brazilian Amazon basin indicates the need for a thorough investigation of seasonality in this last and heterogeneous region. Additionally, since these patterns are linked to climate variables, malaria models should also incorporate them. This study applies wavelet analysis to incidence data from 2003 to 2020 in the Epidemiological Surveillance System for Malaria (SIVEP-Malaria) database. A mathematical model with climate-dependent parametrization is proposed to study counts of malaria cases over time based on notification data, temperature and rainfall. The wavelet analysis reveals marked seasonality in states Amazonas and Amapá throughout the study period, and from 2003 to 2012 in Pará. However, these patterns are not as marked in other states such as Acre and Pará in more recent years. The wavelet coherency analysis indicates a strong association between incidence and temperature, especially for the municipalities of Macapá and Manaus, and a similar association for rainfall. The mathematical model fits well with the observed temporal trends in both municipalities. Studies on climate-dependent mathematical models provide a good assessment of the baseline epidemiology of malaria. Additionally, the understanding of seasonality effects and the application of models have great potential as tools for studying interventions for malaria control.

Local and landscape factors influencing mercury distribution in water, bottom sediment, and biota from lakes of the Araguaia River floodplain, Central Brazil

Mercury (Hg) is a chemical element widely present in the Earth's crust. However, its high toxicity and ability to accumulate in organisms and biomagnify through food chains characterize it as a global pollutant of primary control. We assessed total mercury concentrations ([THg]) in abiotic and biotic compartments from 98 floodplain lakes associated with the Araguaia River and six tributaries (Midwest Brazil). [THg] quantification in water was performed by cold vapor atomic fluorescence spectroscopy. [THg] in bottom sediment was assessed using cold vapor generation atomic absorption spectrophotometry, while [THg] in macrophyte, periphyton, and plankton were quantified by thermal decomposition atomic absorption spectrometry. Hotspots of [THg] in water, bottom sediment, and macrophytes were determined in areas impacted by pasture and urban areas. In contrast, hotspots of [THg] in periphyton and forest fires were determined in preserved areas downstream. [THg] in plankton did not show a clear spatial distribution pattern. The mean bioaccumulation factor order was plankton (2.3 ± 1.8) > periphyton (1.3 ± 0.9) > macrophytes (0.7 ± 0.4) (KW = 55.09, p < 0.0001). Higher [THg] in water and bottom sediment were associated with high pH (R2adj = 0.118, p = 0.004) and organic matter (R2adj = 0.244, p < 0.0001). [THg] in macrophytes were positively influenced by [THg] in water (R2adj = 0.063, p = 0.024) and sediment (R2adj = 0.105, p = 0.007). [THg] in periphyton are positively related to forest fires (R2adj = 0.156, p = 0.009) and [THg] in macrophytes (R2adj = 0.061, p = 0.03) and negatively related to lake depth (R2adj = 0.045, p = 0.02). The transfer of Hg from water and sediment to the biota is limited. However, the progressive increase of the bioaccumulation factor between macrophyte, periphyton, and plankton may indicate Hg biomagnification along the food chain of the Araguaia River floodplain.

Reconfiguration of Amazon's connectivity in the climate system

With the recent increase in deforestation, forest fires, and regional temperatures, the concerns around the rapid and complete collapse of the Amazon rainforest ecosystem have heightened. The thresholds of deforestation and the temperature increase required for such a catastrophic event are still uncertain. However, our analysis presented here shows that signatures of changing Amazon are already apparent in historical climate data sets. Here, we extend the methods of climate network analysis and apply them to study the temporal evolution of the connectivity between the Amazon rainforest and the global climate system. We observe that the Amazon rainforest is losing short-range connectivity and gaining more long-range connections, indicating shifts in regional-scale processes. Using embeddings inspired by manifold learning, we show that the Amazon connectivity patterns have undergone a fundamental shift in the 21st century. By investigating edge-based network metrics on similar regions to the Amazon, we see the changing properties of the Amazon are noticeable in comparison. Furthermore, we simulate diffusion and random walks on these networks and observe a faster spread of perturbations from the Amazon in recent decades. Our methodology innovations can act as a template for examining the spatiotemporal patterns of regional climate change and its impact on global climate using the toolbox of climate network analysis.