Tree mortality and recruitment in secondary Andean tropical mountain forests along a 3000 m elevation gradient

Abstract

This study addresses the understudied dynamics of mortality and recruitment in Tropical Mountain forests, critical determinants of forest structural processes and biomass turnover. We examine how these demographic processes change with elevation and varying degrees of forest recovery by utilizing two forest censuses (2015 and 2019) from 16 plots (0.36 ha) across a 600-3500 m asl elevation gradient in the Ecuadorian Andes. Employing multivariate PCA analyses, we characterize successional forest dynamics and explore relationships between demographic rates, elevation, and indicators of forest recovery using standard linear regression and generalized additive models (GAMs). Contrary to our hypothesis, mortality exhibits a unimodal response, peaking at mid-elevations, with no significant relationship to above-ground biomass productivity (AGBp). In our successional forests, dominance by fast-growing species alters expected patterns, leading to increased mortality rates and AGBp, particularly at low-mid elevations. Forest recovery emerges as a significant driver of mortality and the sole predictor of recruitment, especially across different recovery statuses. Although forest recovery doesn't impact mortality rates, it elucidates the identity of declining species in forests with varying recovery degrees. Our findings underscore that while forest recovery does not alter mortality rates, it provides critical insights into understanding which species are affected under varying recovery conditions. Recruitment, primarily driven by successional dynamics, exhibits higher rates in sites with less recovery. Furthermore, we demonstrate the utility of forest structure indicators, such as above-ground biomass, in inferring successional dynamics when the time since the last disturbance is unknown. The study emphasizes the importance of considering disturbances in comprehending the intricate interplay between the environment and forest dynamics in secondary forests.

Fig 1. Principal component analysis (PCA) plot of sample scores (forest plots) across the first 2 PCA axes (variance explained in each axis between parenthesis).

Colors represent two forest types identified by K-means clustering. CT = competitive thinning forests (green) and MT = mature thinning forests (blue).

Fig 2. Boxplots of variables used as indicators of forest recovery in the best PCA model and elevation by forest recovery type.

Colors represent two forest types identified by K-means clustering. CT = competitive thinning forests (green) and MT = mature thinning forests (blue). Significant differences between groups were tested with a 2-sided t-test.

Fig 3. Best bivariate model: Tree mortality rates vs. elevation (GAMs parameters of the smooth term: EDF = 2.435, P = 0.038).

Communities are colored by forest recovery type: CT = competitive thinning forests (green) and MT = mature thinning forests (blue), but GAM model has been fit for the whole dataset. DE = deviance explained.

Fig 4

Multivariate models of mortality vs. A) PCA1 forest recovery and B) PCA2 endemism richness recovery. Models include interactions: varying intercepts and slopes by forest type: CT = competitive thinning forests; MT = mature thinning forests. Color gradient represents plot elevation in m asl.

Fig 5

Relationship between Above-ground biomass productivity (AGBp) vs. elevation (A), and AGBp vs. tree mortality rates (B). Plots that deviate from the expected pattern are marked with a black border and plot name. DE = deviance explained.