Predicting forest recovery at landscape scales can help to integrate natural regeneration as a forest restoration strategy. The first step for a successful forest recovery is recruitment, so predicting tree recruit abundance could assist efforts to identify sites with high potential for natural regeneration. One prerequisite for recruitment is seed sources. However, previous work has revealed wide variation in the effect of landscape seed sources on seedling abundance, from positive to no effect.
The objectives of this study are (1) to quantify the relationship between adult tree seed sources and tree recruits and (2) to predict where natural recruitment would occur in a fragmented, tropical, agricultural landscape.
We used a hierarchical Bayesian zero-inflated model to predict landscape-scale recruit abundance. We then used a map of species-specific tree crowns derived from hyperspectral and lidar imagery to characterize landscape seed sources. We also used property ownership data to represent different management strategies and collected recruitment field data from five species to describe the recruitment abundance in the study area.
Our models revealed that species-specific maps of tree crowns improved recruit abundance predictions compared with a model without the species-specific maps of tree crowns. The conspecific crown area had a stronger impact on recruitment abundance (8.00% increase in recruit abundance when conspecific tree density increases from zero to one tree; 95% CI: 0.80 to 11.57%) compared with the heterospecific crown area (0.03% increase with the addition of a single heterospecific tree, 95% CI: -0.60 to 0.68%). The best performing model had varying effects of the conspecific and the heterospecific crown area on recruit abundance depending on individual property ownership, which indicates that individual property ownership was also an important predictor of recruit abundance.
We showed how novel high-resolution remote sensing could be combined with field data and cadastral data to generate landscape-level maps of tree recruit abundance. Spatial models parameterized with the field, cadastral, and remote sensing data are poised to assist decision support for forest landscape restoration.
agricultural landscape, forest landscape restoration, hyperspectral imagery, natural regeneration