Landscape memory is mediated by climate in tropical mountains: Implications for the spatial distribution of landslides

Link To Share This Poster: https://cdmcd.co/Dwm498

Ana Kilgore, Organismal Biology and Ecology, Colorado College, Colorado Springs, CO, Laura O. Gómez, Biology, EAFIT University, Medellin, Colombia and Carla Restrepo, Department of Biology, University of Puerto Rico Rio Piedras, San Juan, PR

Background/Question/Methods
Landscape memory, the set of processes conferring resistance of disturbed areas to further disturbance, is critical for our understanding of ecosystem dynamics and community assembly at large scales. Although evidence from diverse disciplines suggests that the conditions at the time of disturbance as well as the history of past disturbances are central to landscape memory, there have been few attempts to explore their relative contribution in systems in which ecosystem and geomorphic processes strongly interact. Among these systems, mountains disturbed by landslides stand out due to their regional and global importance. In mountainous regions, topographic and morphological attributes contribute to landscape memory. Yet, vegetation attributes may also play a role in forming patterns of landscape memory. In this study, we ask how landslide co-occurrences are influenced by bioclimatic and topographic variables used to model species distributions (SDMs). Focusing on the Sierra de Las Minas of Guatemala, we created landslide inventories (years 1991, 1998 and 2006) based on remotely sensed data that were used to identify overlapping and non-overlapping landslide areas, model their relationship with bioclimatic and topographic variables, and ultimately, predict changes in their distribution based on future climate scenarios.
Results/Conclusions
Our landslide inventories yielded 972 (year 1991), 539 (year 1998), and 1,987 (year 2006) landslides. The three populations of landslides impacted 424 ha, 252 ha, and 129 ha in 1991, 1998, and 2006, respectively. Roughly 2.7% of the area impacted by landslides in 1991 was impacted in 1998; similarly, 8.1% of the area impacted by landslides in 1998 was impacted in 2006. Irrespective of year, landslide occurrence was explained by three bioclimatic (positive relationship with isothermality, temperature seasonality, and precipitation during the wettest month of the year) and three terrain (negative relationship with aspect and curvature, and positive relationship with slope) variables. The number and type of variables explaining landslide overlapping and non-overlapping areas varied by period of observation. Overall, we observed little overlap among landslide populations which suggests a great potential for recovery after disturbance. In addition, the retention of three bioclimatic variables in our models suggests that climate alone, or through its effect on vegetation, influences the occurrence of landslides as well as their recovery. Given these results, we expect areas with increased precipitation and more extreme temperature shifts to experience an increased landslide burden in the future.