This project focuses on the process of secondary forest regeneration following agricultural abandonment. Current independent studies conducted in Central Amazonia (Manaus, Brazil), northeastern Costa Rica, and Chiapas, Mexico reveal that observed vegetation changes within forests often deviate from predictions based on single-time studies of forests of different age since abandonment. This project unites four existing projects in wet tropical forests of Latin America. Coordinated annual vegetation censuses in long-term study plots will be conducted to test specific chronosequence predictions for a range of dependent variables including stem density, size distributions, basal area, biomass, species richness, species composition, and life-form composition. Observed rates of change in dependent variables will be compared with predicted rates from chronosequence studies within each region, based on the best-fit statistical models.
By understanding how stand-level dynamics conform to or deviate from overall chronosequence patterns within each of these regions, we will identify key ecological factors, processes, and mechanisms that influence successional pathways. The long-term data gathered will provide essential information to guide forest restoration, carbon sequestration, and biodiversity conservation in the Neotropics. Ultimately, our results will enable development of more predictive and more widely applied models of successional vegetation dynamics in today's complex tropical landscapes.
Federal Award ID: 0639114 Report Submission Period: 01/01/2012 to 09/30/2012 Chronosequence and long-term plot-based studies are complimentary research approaches to studying tropical forest succession in abandoned pastures and agriculture fields in Central Amazonia, Brazil, Northeastern Costa Rica, and Chiapas, México. Our studies indicate that despite high variability, successional pathways generally show predictable changes in structural characteristics, tree life history, functional traits, and species relatedness. Vegetation succession in the Central Amazon exhibits alternative pathways according to prior land use; Cecropia species dominate after clearcuts of mature forest, whereas Vismia species dominate after burning of abandoned pastures. Basal area in Cecropia stands increased quickly in the first decade, outpacing accumulation in Vismia stands, but both stand types converged in basal area after 25 years. Vismia stands were much more variable than Cecropia stands in stem density and basal area. Species density increased at a much faster rate in Cecropia than Vismia stands, creating a large divergence after 25 years. Cecropia sites became less similar in species composition during the first decade of succession, whereas Vismia sites maintained a high similarity. Population dynamics and species composition during succession are driven by random and non-random processes. Within each region, temporal variation in stem density and species richness showed higher variation than expected by purely random effects, even when controlling for land use and for age since abandonment. This result indicates that individual trajectories vary widely. In contrast, basal area exhibited more predictable successional trajectories. When initial conditions are taken into account, predictive models of age-related variation in structure and species richness performed well for sites in Costa Rica and Vismia stands in Brazil, whereas random factors were more important for predicting temporal patterns in Mexico and Cecropia stands in Brazil. In Costa Rican study sites, second-growth specialists showed higher mortality rates than old-growth specialists. Initial basal area (at the species level and above the family level) and initial population density (species level) were also significant predictors of mortality, emphasizing the importance of non-random, density-dependent effects. Our results show that variation in functional attributes among tree species is an important driver of successional change. We predicted a shift over time in abundance and basal area from species with traits that enable fast returns on tissue investment early in succession to species with traits that lead to slow economic returns later in succession. Based on wet secondary forest sites in Chiapas, Mexico and Costa Rica, we demonstrated a shift from ‘fast’ traits early in succession towards ‘slow’ traits later in succession. Functional characteristics of the plant community in Chiapas sites were described with the community-weighted mean calculated based on 12 functional traits. Consistent with predictions, leaf tissue density and dry matter content increased during succession, whereas specific leaf area decreased. Seed dispersal by animals also increased during succession. We also compared community-weighted mean functional traits of trees in six secondary forests and two old-growth forests in Costa Rica. Consistent with predictions and similar to the Chiapas sites, we found a decline in specific leaf area, and an increase in leaf dry matter content, leaf thickness, leaf toughness, leaf density, and wood specific gravity within and among successional plots over time. In general, smaller trees in each plot had traits associated with the slow end of the leaf economics spectrum. These results clearly show that functional attributes and species composition of trees in the secondary forest understory are converging with those of old-growth forests in the region. In all of our study areas, trees in early successional forests are more closely related than predicted by chance. We believe this phylogenetic clustering results from abiotic filtering for species traits that favor colonization and establishment in clearings. During succession, closely related species show higher mortality rates, whereas new recruits are more distantly related. These demographic processes increase the representation of different evolutionary groups in later stages of succession, reflecting biotic filtering processes. The shift from abiotic to biotic filtering contributes strongly to the predictive (deterministic) behavior of successional trajectories that we are observing in tropical wet forests of Costa Rica, Mexico, and Brazil. Training and education are a major focus of the neoSelvas LTREB. Since 2007, we mentored 14 undergraduate and masters students, 16 doctoral students,1 postdoctoral associate, and 1 high school teacher. Doctoral students come from the US, Costa Rica, Mexico, Peru, Brazil and The Netherlands. We trained 10 local field technicians to establish plots, identify tree species, map stems, measure tree DBH, and to enter data. Our project stimulated outreach activities with local communities in the Central Amazon sites and provided insights and new models for reforestation projects. Project investigators published over 55 peer-reviewed papers in scientific journals and edited books during the five-year duration.