PS 4-25 - Determinants of variation in and spatial patterning of vegetation cover between and within regenerating landslides in the Peruvian Andes

Monday, August 12, 2019
Exhibit Hall, Kentucky International Convention Center
Cathryn A. Freund, Stephen J. Bechtel and Miles R. Silman, Biology, Wake Forest University, Winston-Salem, NC
Background/Question/Methods

Understanding how ecosystems recover from natural disturbances is a fundamental issue in ecology. Tropical montane forests (TMF) are dynamic, disturbance-driven ecosystems, yet TMF disturbance regimes and their impacts on biodiversity and ecosystem functioning remain poorly understood. Landslides are the most common natural disturbance in the TMF in Peru’s Andes-to-Amazon region. Long-term studies here have laid a foundation for understanding forest composition and functioning across elevational gradients. However, the steep and remote slopes characteristic of TMF systems are challenging places to work and thus to date, most of this research draws on data from permanent forest inventory plots preferentially established in mature forest. Basic questions about the process of TMF regeneration and how successional trajectories vary in relation to landslide size and elevation remain unanswered.

Here we aim to elucidate drivers of vegetation patterning during forest regeneration after landslides using high-resolution drone imagery of 25 landslides across a 2.5-kilometer elevational gradient in the Peruvian Andes. We use vegetation classification methods, linear modeling, and a spatial autologistic model to ask,

  • How does vegetation cover on regenerating landslides vary with landslide age, size, and elevation?
  • How does vegetation cover vary within landslides with slope and distance from edge, after accounting for spatial autocorrelation?

Results/Conclusions

Landslides surveyed ranged in age from 5-18 years, in size from 945 – 78,038 m2, and occurred between 900 and 3,579 meters above sea level. The portion of each landslide covered with vegetation ranged from 45.6 – 99.6%. Preliminary results suggest that landslide size is a significant predictor of vegetation cover (p = 0.03), but landslide age and elevation are not.