Andean montane forests are a biodiversity hotspot, store large amounts of carbon and provide numerous ecosystem services to inhabitants. After abandonment of centuries of agriculture and human use, many areas undergo forest recovery. Fieldwork studies have shown that montane forest recovery is often arrested or unsuccessful. However, little is known on large-scale and long-term recovery dynamics and on how forest recovery manifests through space and time across the mountain range.

The goal of this paper is to monitor 20 years of tropical montane forest recovery using satellite data on three case study landscapes and upscaled across the Andes. First, we map ‘potential recovery areas’ between 2000-2005 using a combination of Hansen Forest Loss, GFCC Tree Cover Change and MODIS Burned area datasets and use these areas to assess forest recovery trajectories for the following 15 years. For this we use Landsat time series analysis of Normalized Difference Water Index (NDWI), tree cover change analysis and a novel classification of ‘realized recovery’ trajectories based on sequential pattern of z-score anomalies to elucidate on recovery dynamics.

At a landscape scale our method captures forest recovery dynamics well, tracking trajectories of known successful natural regeneration in a Colombian National Park context, or detecting instances of arrested succession in a Peruvian high-altiplano landscape. Across the Andes land abandonment represents the major potential recovery class for the 5-year time frame (274km2). While tree cover increases across the Andes, most of the areas stay in early succession states (10-25% tree cover) and recovery of NDWI is slow. Of all potential forest recovery areas in the Andes, more than 37% show ‘ongoing recovery’, 28% show either ‘disrupted’ or ‘arrested recovery’, and 35% show ‘no recovery. These recovery trajectories vary greatly between countries, highlighting 'hotspots' and 'notspots' of forest recovery.

Our findings show that forest recovery through natural regeneration is largely disrupted, arrested or unsuccessful across the Andes, which has consequences for biodiversity recovery and provision of ecosystem services. Areas of low recovery may be targeted for active restoration interventions in this UN Decade on Restoration and future studies could aim at elucidating restoration priorities and suggested management strategies across the Andean montane forests.