MODELLING THE EFFECT OF BIOCLIMATIC FACTORS ON THE PRESENT AND FUTURE DISTRIBUTION OF AFRICA LOCUST BEAN (PARKIA BIGLOBOSA) IN AFRICA

Abstract

With the advent of climate change, it is important to predict the effect it has on the potential distribution of species in order to identify those that are vulnerable for planning conservation strategies. This study was carried out to predict the effect of 2080 climate change on the potential distribution of Parkia biglobosa. A factorial combination of three resolutions (2.5, 5 and 10 arc-minutes; retrieved from WorldClim database) and eight sample sizes (5, 10, 25, 50, 75, 150, 200 and 305) were sampled from Parkia biglobosa location data retrieved from Global Biodiversity Information Facility (GBIF) and modelled using MaxEnt (version 3.4.4). All 19 bioclimatic variables in addition to four soil layers (retrieved fromHarmonized World Soil Databases (version 1.2)) were used for the modelling with 15 replications. A virtual species model was also ran based on the mean annual rainfall (mm) and mean annual temperature (℃) of the GBIF data on the software Virtualspecies”. Models were evaluated using Area Under Curve (AUC), True Skill Statistic (TSS) dKappa Statistic. Jaccard Similarity Index was also calculated between the predicted ranges and the true range for the virtual species.Results indicate that Parkia biglobosa has the potential to expand its present range by about 110% in the future 2080. Small sample sizes (<50) predicted imprecise ranges and also were less accurate in terms of evaluation statistics than larger sample sizes (≥50). The effect of resolution largely depended on sample size. Comparison between evaluation statistics and Jaccard Similarity Index suggests that evaluation statistics may not reflect accuracy of range prediction. It is concluded that, human assisted dispersal may be necessary to aid Parkia biglobosa to achieve its future potential.