Biodiversity in river networks: Theoretical and empirical perspectives .

摘要

Structured metapopulation models and dispersal kernels are developed to study biodiversity patterns in river networks, which are characterized by network structure and directional dispersal. Two types of interspecific dynamics are considered: the neutral and strictly hierarchical competition-colonization trade-off models. We find that dispersal directionality and network structure promote species that produce a large number of propagules at a species level. This translates into promoting competitively superior species in the neutral model, and competitively inferior species in the trade-off model. Our results suggest that the more stochastic the interspecific dynamics, the more sensitive the relative abundance patterns are on the dispersal directionality and network structure. The relative importance of the network structure and dispersal directionality is also investigated: alpha (local), beta (between-community), and gamma (global) diversities seem very sensitive to the dispersal directionality, even more so than the dispersal rate. The network structure, through its containment effect, leads to higher beta diversity in river networks and two-dimensional landscapes. The analysis shows that only in river networks with directional dispersal does the probability distribution of alpha diversity exhibit a power-law character, implying the high likelihood of biodiversity hotspots relative to the rest of the system. Interestingly, the analysis of spectral group diversity performed on remote sensing data of a river network in a semi-arid region finds the same signature of alpha diversity. The empirical spatial series of alpha diversity exhibits higher, more fluctuating values in the river network's downstream portions and long-range correlation. The remote sensing analysis also shows that riparian zones at stream links exhibit no strong trend with stream magnitude. Rather, it is exponentially distributed, which suggests that riparian vegetation communities may organize themselves into a configuration with maximum entropy, which may lead to optimal flexibility against their changing environment. Finally, a stochastic model of riparian vegetation biomass is developed to study the width of riparian zones. Applying the theory of Markovian dichotomous noise and empirical geomorphic relationships, we find that if the fluctuation of river flow increases too rapidly with stream magnitude, the riparian zones at streamlinks of high magnitudes collapse as the flooding-induced stress becomes too strong.