Fruit and Seed fall in southeast Amazonia following experimental fires

This dataset was used to investigate how wildfires and edge effects affected fruit and seed (FS) production and diversity of a tropical forest. The research was conducted in southeast Amazonia from 2005-2018, using three 50-hectare plots with varying fire regimes: a control plot, a plot burned annually, and a plot burned every three years (with exception of 2008). These plots also experienced edge effects, two droughts, and a blowdown event. The findings indicate that while low-intensity fires alone had a minor impact on FS production, the combined effects of fires, droughts, and windthrow led to significant declines in FS diversity and production. Additionally, the species composition shifted towards trees with acquisitive strategies, characterized by faster growth, thinner leaves and bark, and shorter stature, particularly along the edges of the burned plots. This research highlights the complex interactions between different types of disturbances and their cumulative impact on tropical forest ecosystems

Methods
Litterfall was collected biweekly from August 2004 to August 2018 using 0.5-m2 screen litter traps (N: 90 per treatment plot) suspended 1 m above the forest floor and distributed systematically in grids throughout the plots to capture spatial variability random variability and potential edge effects (details in Balch et al., 2008). Litter was oven-dried (65oC for 48 hrs) and weighed to calculate dry mass. Fruits and seeds were separated from the litter and identified to species. Of the total fruit/seed fall, we could not identify 11 % of the sampled species. Nine of these were treated as different morpho species. The same technician sorted and identified all species in order to reduce identification errors

Annual or biannual inventories in the unburned Control (details in Balch et al., 2008) were used to calculate maximum tree size (based on dbh, 97.5% percentile) and relative tree growth per species. In addition, we sampled 413 individuals in the unburned Control and 451 in the burned plots for stem-specific density (SSD), maximum tree height (MTH), relative growth rate (RGR), specific leaf area (SLA) and bark thickness (BT), following Pérez-Harguindeguy et al. (2013). To calculate the community-weighted mean, we weighted trait values by the relative abundance of each species, which was estimated as the number of traps containing the FS of a given species in each year divided by the total number of FS traps. We then evaluated how community-weighted traits changed over time across the three treatments.