Shelterbelt systems establishment in Saskatchewan, Canada: a multi-criteria fuzzy logic approach to land suitability mapping

There is lack of guidelines helping land managers to locate suitable areas for planting new shelterbelt agroforestry systems on their landbases. The goal of this study was to create land suitability maps for deciduous, coniferous, and shrub shelterbelt agroforestry systems establishment across a wide range of climatic and soil zones of Saskatchewan, Canada. Spatial shelterbelt data and a suite of 50 predictor variables were analyzed using multivariate principal component analysis (PCA), principal component regression (PCR), fuzzy logic analysis, and GIS mapping techniques. Fifty spatial datasets were used as shelterbelt establishment predictor variables (4 groups): 21 climate (1980–2010 normals), 13 land management, 14 soils, and 2 topographic criteria. A shelterbelt carbon inventory spatial layer was used as the shelterbelt establishment indicator dataset. Using PCA and PCR analyses, the overall importance (cumulative loading: positive or negative) of all predictor variables was determined and used to create shelterbelt suitability maps by means of weighted-sum overlays in GIS. Statistically significant positive correlations between mapped shelterbelt suitability levels and observed mean shelterbelt carbon stocks were used to evaluate the resulting deciduous (4.86 million hectares (Mha) study area; p = 0.0033, R 2 = 0.79), coniferous (1.96 Mha; p = 0.0008, R 2 = 0.77), and shrub suitability maps (2.06 Mha; p = 0.0002, R 2 = 0.83). Additional 8.76, 7.90, and 9.77 Mha were identified as suitable for planting future deciduous, coniferous, and shrub shelterbelt systems, respectively, mapped as above-average or high suitability land. Shelterbelt suitability mapping is a means to delineating and ranking the land across large landscapes. The approach employed in this study can benefit other afforestation and agroforestry adoption studies across Canada and the world.