Structural complexity in riverine and terrestrial habitat networks affects population abundance and diversity

Abstract

Landscape structure is an important determinant of spatial variation in population and community processes. Theoretical research commonly assumes that extinction and recolonisation can be modelled without considering population dynamics, and ignores the latter in metapopulation models. However, the scales at which landscapes must be actively managed are generally commensurate with scales of population dynamics, suggesting the importance of developing models of (meta-)population response that explicitly consider population dynamics. It has been previously demonstrated that habitat network structure affects the responses of metapopulations in two-dimensional habitat networks typical of terrestrial systems. However, other types of habitat networks have largely been ignored. In particular, river systems follow a rigid hierarchical branching structure to form a Dendritic Ecological Network (DEN), which differs fundamentally from terrestrial systems because habitat patches are only ever linked by a single path. It has been hypothesized that this unique structure must have important implications for resident populations and communities. We developed a theoretical method to relate both population abundance and two different measures of population diversity to landscape structure. We simulated lattice and dendritic networks of differing complexity, representing open terrestrial landscapes with no particular constraints on patterns of connectivity, and river networks constrained to bifurcating connectivity patterns, respectively. We used two different measures to quantify structural complexity. Landscapes were simulated as directed graphs, with nodes representing individual habitat patches defined by a habitat quality or carrying capacity. Patches were connected by directed graph edges, with asymmetrical edge weights quantifying the probability of or-ganisms being able to move in each direction between pairs of nodes. Density dependent population processes were modelled stochastically within individual patch-level populations. Results were affected both by the way in which structural complexity was represented, and also by the way in which diversity was quantified. Nevertheless, both the type and the complexity of the habitat network affected population abundance and diversity. Complex lattice and dendritic networks supported the highest and lowest abundances, respectively. As complexity was reduced, network-scale carrying capacities of the two network types converged. The results for diversity were quite different. Diversity was almost unaffected by structural complexity in dendritic networks, and was higher than for any lattice network. Within lattice networks, simple structures had the highest diversity. Simple lattice networks are structurally similar to complex dendritic networks, except for the occasional presence of loops and multiple paths. These features appear to have profound effects on network-scale abundance and diversity. These results show that the unique structure of dendritic ecological networks do indeed have unique implications for resident organisms. They suggest that highly branched riverine systems are more able to foster diverse ecosystems that less branched equivalents and terrestrial landscapes. They also call into question the predominate use of metapopulation models that ignore population processes when studying the effects of habitat network structure.