Mapping boron deficiency risk in soils of south-west Western Australia using a weight of evidence model

Abstract

The aim of this work was to develop a risk map for boron (B) deficiency in the grain cropping regions of Western Australia (WA), whilst avoiding the high costs associated with direct B measurements for an area as vast as the south-west of WA. The study firstly determined relationships between 0.01 m CaCl2-extractable soil B levels and readily available data on soil properties and parent materials for Reference Soils of south-west Australia and secondly assembled direct evidence of B deficiency risk from surveys of farmers’ crops and soils and from glasshouse experiments. Across 73 Reference Soils, there was a positive relationship between 0.01 m CaCl2-extractable soil B levels and clay (r 2 = 0.50) and pH (r 2 = 0.43) in the surface horizon. Soils containing <0.5 mg B/kg generally had <5% clay and pH CaCl2 <5.5. Plant and soil analysis surveys in farmers’ fields revealed 10–20% of fields had B levels below tentative critical levels. In a glasshouse experiment, B response in oilseed rape was obtained in 4 sandy acid soils, all developed on sandstone parent materials. From this prior evidence of B deficiency, spatial data layers for surface soil pH, subsurface pH, surface clay level, and geology in south-western Australia were weighted and combined using the Dempster-Shafer weight of evidence model to map B-deficiency risk. The weightings of evidence layers were revised to increase the correspondence between predicted areas of high risk and field areas with measured low B or B deficiency from a validation dataset. The model helps overcome the high cost associated with direct B measurements for risk mapping. A similar approach may have value for mapping risk of other deficiencies of relevance to agriculture.