Our main objective was to compare the relative importance of continuous spectral information (i.e., NDVI and spatio-temporal properties of NDVI) versus landscape metrics derived from a discrete land cover classification as predictors for describing the variability in species richness of birds, butterflies, and plants within agricultural environments. Our results show that models based on predictors derived from continuous information consistently outranked models based on predictors derived from discrete information. This finding is clearly evident when examining individual model weights (Tables 5 and 6), calculated model evidence ratios, and when inspecting model-averaged coefficients and associated 95% CIs (Fig. 3).