Assessment of plant species distribution and diversity along a climatic gradient from Mediterranean woodlands to semi-arid shrublands

Climate and land-use change profoundly affect plant species distribution (SD) composition, the impact of these processes is expected to increase in coming years. As a proxy global changes, knowledge SD diversity along climatic gradients essential determine efforts needed for conservation. Plant spectral an emerging approach used as based on remote sensing. Thus, research aim was develop comprehensive methodology richness mapping study their relations with environmental human-derived factors, demonstrated Mediterranean semi-arid gradient. The addresses two main gaps regarding diversity: (1) improving accuracy woody classification by features extraction selection, using texture analysis ecosystem characterized high spatial variability relatively small-sized sparse vegetation; (2) developing better estimate local ‎richness response factors (i.e. climate, topography, substrate, land cover factors) across transition zone between woodlands dwarf shrublands. A hyperspectral image acquired 43-km strip area airborne flight AISA-FENIX (380–2500 nm, 420 bands) at end 2017 rainy season. dominant were surveyed, total number 247 trees shrubs, train machine learning support vector (SVM) mapping, which yielded overall 86.1%. feature selection developed, combining principal component neighborhood techniques, facilitating identification 33 diagnostic bands out 330 bands. decreased about 2% 84.2% only improved 7.1% seven large crown (93.3%) adding information. Later, calculated utilizing alpha index Shannon Index) 30-m grid cells tested topography) cover). highest sensitivity mean annual precipitation, slope, surface temperature. showed higher natural shrubland guarrigue located northern part climate We suggest that presented here significantly improves estimation areas heterogeneity steep gradients.