Predicting the potential impact of climate change on the declining agroforestry species Borassus aethiopum Mart. in Benin: a mixture of geostatistical and SDM approach

摘要

Predicted effects of climate change (CC) on plant species distribution have raised concerns on their conservation and domestication. Appropriate stand density may enhance species ability to adapt to CC. Therefore, combining species distribution modeling (SDM) and spatial pattern of density should provide insightful information for setting conservation actions. We combined geostatistical and SDM techniques to assess (1) current tree density spatial pattern and its relationship with bioclimatic zone (humid, sub-humid, and semi-arid), land-use type (protected areas vs. agrosystems), and soil type (eight types), and (2) present-day and future distributions of suitable habitats under low-RCP4.5 and high-RCP8.5 emissions scenarios for Borassus aethiopum, a declining agroforestry palm in Benin. Data were obtained from 2880 one-ha plots. Semivariogram and kriging were used to model spatial patterns of density while Maximum Entropy was used for SDM. Tree density followed an isotropic spatial model with a range of 2.15km, indicating extremely fragmented density pattern. Tree density was 8-times higher in protected areas (PAs, 68.6 +/- 5.09 treesha(-1)) than in agrosystems (8.4 +/- 0.31 treesha(-1)) and greater on ferruginous soils. Though 80% of the country was currently highly suitable with similar trend for PAs and agrosystems, future predictions showed major habitat loss (20-61%), particularly under RCP8.5. While changes were similar between PAs and agrosystems, the decrease in habitat suitability was pronounced in the semi-arid zone where the species is currently widely-distributed with higher abundance. Very weak link was found between present-day abundance and present-day and future distribution. It is concluded that B. aethiopum has a fragmented density pattern and will be sensitive to CC. In-situ and circa-situ conservations or orchards establishment were suggested depending on the projected changes and the bioclimatic zone. The approach used here is exemplary for other agroforestry tree species.