Datasets collected in multi-disciplinary soil science research present unique analytical challenges that arise due to a large number of variables measured across a small number of experimental treatments. This issue of dimensionality, along with confounding variation due to season and depth, plus the possible presence of nonlinear relationships, further limits the application of traditional data-guided methods in soil research. Here, we apply a hybrid data mining and visualization strategy to a dataset generated for soils sampled in a fertilizer and cultivar trial involving the herbaceous energy crop switchgrass (Panicum virgatum L.). The dataset included more than 45,000 records of soil carbon, nitrogen, plant and litter biomass, soil fractions, micro- and macro-aggregates, and various descriptors of microbial biomass and community composition. A visual interface provided a two-dimensional depiction of the information content of an independent variable versus its relationship with other variables of interest. For variables that varied by and hence were measured at multiple depths, an exponential decrease in variation was observed as depth increased for all variables. Residuals extracted after fitting multiple regression models accounting for depth and time were subsequently fed into regression models to extract information content and correlations, which in turn were used to develop correlation-based complex networks captured through visual interfaces. Insights from complex networks included dyads, triads, and subsets of variables were exclusively related to each other, and could be interpreted as simple associations. A majority of the variables appeared to be well integrated within a complex network with small-world properties where soil carbon and nitrogen form the link between variables related to soil properties and another related to biomass and biological activity. New insights and confirmation of understanding need to be interpreted based on current understanding of soil science and potentially examined for new hypothesis generation.