Study sites
This study was conducted in 30 permanently monitored forest sites located in southeast Brazil (Figure 1). The largest distances among sites are of 900 km (latitude) and 177 km (longitude). Altitudes vary between 447 and 1490 meters above sea level, whereas mean annual precipitation (MAP) varies between 763 and 1831 mm, and mean annual temperature (MAT) between 17.1 and 25.4 °C. The data include three forest types distinguished by deciduousness and climate (details in Supplementary Material Table S1): (i) five evergreen forest sites, with little to insignificant deciduousness and under the coldest and wettest climate in the dataset (measured between 1995-2019); (ii) sixteen semideciduous forest sites, with 20-50% of canopy deciduousness during the dry season and under an intermediate climate in the dataset (measured between 1987-2019); (iii) and nine deciduous forest sites, with more than 50% canopy deciduousness during the dry season and under the driest and hottest climate of our dataset (measured between 2002-2019). All sites are closed-canopy and mixed-age forests with similar conservation statuses, with no indication of wood extraction or fire occurrence.
Vegetation data were collected from 400 m² plots, distributed across each site aiming to capture local heterogeneity reliably (total of 34 hectares sampled). Each site was measured at least twice, with an inclusion criterion equal to or higher than 5 cm of quadratic mean diameter at the reference height (1.30 m; dbh). All individuals that met the inclusion criterion were tagged and their point of measurement (POM) recorded. We used the POM as a reference for the subsequent measurements. When the POM of a given stem needed to change between measurements, we estimated stem diameter growth from the ratio between the current and previous POMs (Talbot, et al. 2014). Tree identification was performed by specialists in the field or by consulting herbaria. Species names followed APG IV (Angiosperm Phylogeny Group, 2016) and were standardised based on The Plant List (2020).
We extracted wood density values of all individuals from the global wood density database (Chave et al., 2009; Zanne et al. 2009). When wood density was not available at the species level, we used the average wood density value of other species within the same genus or family. We calculated each tree’s aboveground woody biomass (AGWB) with the pantropical allometric equation proposed by Chave et al. (2014), with package biomass (Réjou-Méchain, Tanguy, Piponiot, Chave, & Hérault, 2017) in the R environment. We used the modified version of this equation because information on tree height was unavailable. We corrected the values of AGWB productivity with the CIC1 equation by Talbot et al. (2014). We removed the productivity of recruited species because individual AGWB is non-numeric. We calculated the average AGWB (or average size) of each population based on their initial total biomass (sum of AGWB of all individuals) divided by the total number of individuals at the beginning of monitoring.
We considered as “locally extinct” those species which disappeared from their original occurrence area, in one or more sites; as “regionally extinct” those species which disappeared in all sites of our dataset; and as “recruited” those species which were absent from all areas in the first measurement, but met the inclusion criterion at a certain point and remained in the dataset until the last measurement.