The tropical forest carbon sink is known to be drought sensitive, but it is unclear which forests are the most vulnerable to extreme events. Forests with hotter and drier baseline conditions may be protected by prior adaptation, or more vulnerable because they operate closer to physiological limits. Here we report that forests in drier South American climates experienced the greatest impacts of the 2015–2016 El Niño, indicating greater vulnerability to extreme temperatures and drought. The long-term, ground-measured tree-by-tree responses of 123 forest plots across tropical South America show that the biomass carbon sink ceased during the event with carbon balance becoming indistinguishable from zero (−0.02 ± 0.37 Mg C ha⁻¹ per year). However, intact tropical South American forests overall were no more sensitive to the extreme 2015–2016 El Niño than to previous less intense events, remaining a key defence against climate change as long as they are protected.

Tropical forests have played an important role as a carbon sink over time. However, the carbon dynamics of Brazilian non-Amazon tropical forests are still not well understood. Here, we used data from 32 tropical seasonal forest sites, monitored from 1987 to 2020 (mean site monitoring length, ~15 years) to investigate their long-term trends in carbon stocks and sinks. Our results highlight a long-term decline in the net carbon sink (0.13 Mg C ha⁻¹ year⁻¹) caused by decreasing carbon gains (2.6% by year) and increasing carbon losses (3.4% by year). The driest and warmest sites are experiencing the most severe carbon sink decline and have already moved from carbon sinks to carbon sources. Because of the importance of the terrestrial carbon sink for the global climate, policies are needed to mitigate the emission of greenhouse gases and to restore and protect tropical seasonal forests.

Large‐scale data compilation is increasing steadily in tropical forest research, but the lack of standardized methods for data collection limits drawing inference from large datasets and cross‐biome analyses. Different inclusion methods and minimum tree diameter threshold are among these varying factors. To tackle this issue, we evaluated how different approaches for tree sampling affects our understanding of diversity and functioning in different tropical vegetation types. We used a unique dataset of 44 inventory plots (43.54 ha) encompassing an aridity gradient: evergreen moist forests, semideciduous and deciduous tropical forests. Data were collected using the by‐tree inclusion method, in which, all stems are measured if the equivalent diameter of the tree reaches the minimum threshold. We simulated the impact of adopting different inclusion methods (by‐stem and by‐tree) and different minimum diameter thresholds on the estimation of number of trees and stems, biomass and species richness. We used linear and nonlinear mixed models to investigate the effect of minimum diameter threshold and inclusion method on our different response variables. We also evaluated species chance to be sampled under different minimum inclusion criteria. Inclusion method and minimum diameter threshold mainly affect the estimation of number of trees and stems and species richness, especially in deciduous and semideciduous forests, where resprouting is a prevalent strategy. In these forests, many trees that have several stems do not reach the minimum size individually when adopting the by‐stem method, yet they do reach the minimum size threshold when all stems are considered together. For these environments under water stress, our analysis showed that using large minimum sizes, such as the 10 cm typically used in rainforests, implies large sampling losses, especially when used jointly with the by‐stem inclusion method. The by‐tree inclusion method represents an alternative approach that offers a more reliable sampling in different vegetation types, particularly in those habitats where resprouting is a widely encountered strategy along all age classes. We demonstrate the infeasibility of adopting broad and standard minimum thresholds for different tropical vegetation types, particularly considering their widely different ecological strategies.

The aim of this study was to evaluate the vegetation identity of local-scale ecotones and its importance to landscape biodiversity in a transition between savanna and forest vegetation types in Brazil. We surveyed the tree community (diameter at breast height ≥5 cm) within 25 plots of 400 m2 across three core vegetation types and two ecotones among them (totalling five vegetation types). We then evaluated similarities in species composition, community structure and phylogenetic diversity across the transitions in order to assess the relationship between the ecotones and the core areas. Ecotones were distinct floristic units with a high number of unique species and floristic and phylogenetic clustering, and hence these environments are additional vegetation types in relation to the core areas. Some species showed maximum abundance in ecotones, which harboured distinct ecological patterns, demonstrating the importance of the ecotones in the overall ecosystem. Results are related to the distribution of the species present in the regional pool across local-scale microhabitats, with ecotones being a product of a distinct environmental conditions resulting from the distinct adjacent biomes (savanna and forest), which enable the maintenance of biological diversity.