The form of fluvial landscapes is known to attain stationary network configurations that settle in dynamically accessible minima of total energy dissipation by landscape-forming discharges. Recent studies have highlighted the role of the dendritic structure of river networks in controlling population dynamics of the species they host and large-scale biodiversity patterns. Here, we systematically investigate the relation between energy dissipation, the physical driver for the evolution of river networks, and the ecological dynamics of their embedded biota. To that end, we use the concept of metapopulation capacity, a measure to link landscape structures with the population dynamics they host. Technically, metapopulation capacity is the leading eigenvalue (M) of an appropriate landscape matrix subsuming whether a given species is predicted to persist in the long run. (M) can conveniently be used to rank different landscapes in terms of their capacity to support viable metapopulations. We study how (M) changes in response to the evolving network configurations of spanning trees. Such sequence of configurations is theoretically known to relate network selection to general landscape evolution equations through imperfect searches for dynamically accessible states frustrated by the vagaries of Nature. Results show that the process shaping the metric and the topological properties of river networks, prescribed by physical constraints, leads to a progressive increase in the corresponding metapopulation capacity and therefore on the landscape capacity to support metapopulationswith implications on biodiversity in fluvial ecosystems.