Under predicted climate change: Distribution and ecological niche modelling of six native tree species in Gilgit-Baltistan, Pakistan

This study presents the tree species distribution and habitat suitability maps in Gilgit-Baltistan, Pakistan at 1 km spatial resolution. This study is based on bioclimatic and topographical variables and 440 samples of six native trees species: Abies pindrow, Betula utilis, Cedrus deodara, Picea smithiana, Pinus wallichiana, and Quercus ilex. Data is collected through field survey. Exclusively for each tree species, a multicollinearity test was performed among 24 independent or environment variables (21 bioclimatic and 3 topographic). The highly correlated independent variables (r ≥ 0.9, Pearson correlation coefficient) were eliminated from the independent variables list. In this study, we employed the Maximum Entropy (MaxEnt) model to produce current (2015–2016) as well as RCP4.5 and RCP8.5 climate-change scenarios by 2050 for tree species spatial distribution results. The jackknife test was carried out to depict the importance of variables with the highest gain and it was observed that overall elevation, precipitation, and temperature are the factors with the highest gain. The results of the MaxEnt model for each tree species were satisfactory with ROC (receiver operating characteristic) AUC (area under the curve) curve training and testing values greater than 0.9 and 0.84 respectively. Based on 10-percentile training presence threshold-dependent values, the overall accuracy of True Skill Statistics (TSS) was more than 80%. The maximum area coverage of all tree species existed under “inadmissible natural surroundings (0–0.2 probability)” and least area fell under “exceptionally appropriate environment (0.6–0.7 probability)” to “profoundly reasonable living space (0.7–1.0 probability)”. A tree species diversity map prepared through equal weighted average overlay analysis, using all six developed tree species probability outputs. The field observation might possess certain limitations because it was difficult for the field crew to access the areas with rough terrain, long distances, harsh weather conditions, and locations of forest in steep, narrow valleys. Overall, this study contributes to enlarge tree species distribution research datasets applicability in Pakistan and over the Hindu Kush Himalayan (HKH) mountains region. It may also provide interesting insight, which could be used for the habitat corridor suitability modelling of endangered species, and ground intervention to protect and expand tree species distributions.