Understanding the spatial distribution of specific environmental variables and the interdependencies of these variables is crucial for managing the environment in a sustainable way. Here we discuss two methods of mapping ? a Geographical Information System classification-based approach and a statistical model-based approach. If detailed, spatially comprehensive covariate datasets exist to complement the ecological-response data, then using a statistical model-based analysis provides the potential for greater understanding of underlying relationships, as well as the uncertainty in the spatial predictions. Further, the model-based approach facilitates scenario testing. Although similar methods are already adopted in species distribution modeling, the flexibility of the model framework used is rarely exploited to go beyond modeling occupancy or suitability for a single species, into modeling complex derived metrics such as community composition and indicators of natural capital. As an example, we assess the potential benefits of the statistical model-based approach to mapping natural capital through the use of two national survey datasets; The Centre for Ecology and Hydrology (CEH) Land Cover Map (LCM) and the British Geological Survey's (BGS) Parent Material Model (PMM), to predict national soil microbial community distributions based on data from a sample of > 1000 soils covering Great Britain. The results are mapped and compared against a more traditional, land classification-based approach. The comparison shows that, although the maps look broadly similar, the model-based approach provides better overall spatial prediction, and the contribution of individual model terms (along with their uncertainty) are far easier to understand and interpret, whilst also facilitating any scenario testing. We therefore both recommend the use of spatial statistical modelling techniques to map natural capital and anticipate that they will become more prominent over the forthcoming years.