The debate about the relative merits of the 'land-sparing' and 'land-sharing' approaches to biodiversity conservation is usually addressed at local scale. Here, however, we undertake a regional-scale approach to this issue by exploring the association between the Human Appropriation of Net Primary Production (HANPP) and biodiversity components (plants, amphibians, reptiles, birds and mammals) across a gradient of human-transformed landscapes in Catalonia, Spain. We propose an Intermediate Disturbance Complexity (IDC) model to assess how human disturbance of the photosynthetic capacity affects the landscape patterns and processes that host biodiversity. This model enables us to explore the association between social metabolism (HANPP), landscape structure (composition and spatial configuration) and biodiversity (species richness) by using Negative Binomial Regression (NBR), Exploratory Factor Analysis (EFA) and Structural Equation Modelling (SEM). The empirical association between IDC and landscape complexity and HANPP in Catalonia confirms the expected values of the intermediate disturbance hypothesis. There is some increase in biodiversity when high IDC values correspond to landscape mosaics. NBR and EFA show positive associations between species richness and increasing values of IDC and forest cover for all biodiversity groups except birds. SEMshows that total biodiversity is positively determined by forest cover and, to a lesser extent, by HANPP, and that both factors are negatively associatedwith each other. The results suggest that 'natural' landscapes (i.e. those dominated by forests) and agroforestrymosaics (i.e. heterogeneous landscapes characterized by a set of land uses possessing contrasting disturbances) provide a synergetic contribution to biodiversity conservation. This 'virtuous triangle' consisting of forest cover, HANPP and biodiversity illustrates the complex human-nature relationships that exist across landscape gradients of human transformation. This energy-landscape integrated analysis provides a robust assessment of the ecological impact of land-use policies at regional scale.