Biodiversity is unevenly distributed across the Earth, and empirical studies suggest that landscape features have a crucial role in shaping such patterns. Montane regions or riverine systems sustain high levels of species diversity, indicating that topography and habitat complexity are major drivers of biodiversity. Nonetheless, there exists a gap between theory and observations. The complexity of the eco-evolutionary processes that shape species’ evolution blur our understanding on how exactly diversification mechanisms are influenced by landscapes. Recently, insights from an eco-evolutionary model where landscapes are represented as a graphs - that capture landscape topographical complexity - revealed that simple graph properties have a major influence on the diversification processes. In particular, the theory demonstrates how heterogeneity in degree, characteristic length and habitat assortativity differently affect neutral and adaptive diversification. In this talk, we propose a simple methodology to project real landscapes on graphs, and illustrate how to characterize large geographical areas by using graph-based topography metrics. Because they connect differentiation patterns to generating eco-evolutionary mechanisms and landscape features, such metrics could improve our understanding of the origin of spatial biodiversity gradients.