Patterns of dispersal through stream networks respond simply to multiple structural modifications

Planning of catchment-scale programs of stream restoration requires an ability to predict the combination of small-scale projects that will provide the greatest ecological benefit. However, we lack the tools to predict where within stream systems restoration should be undertaken to achieve maximum ecological benefit, or to quantify the cumulative effect of multiple small-scale projects. Stream systems form Dendritic Ecological Networks (DENs), which differ from other ecological networks in terms of their complexity, and possible routes of dispersal for organisms. The restricted topography of a DEN means that a single intervention in a network may have greater consequences than would such an intervention in a network of less restricted topography. Moreover, the strong directional connectivity inherent in stream systems suggests that multiple interventions in a network will interact, leading to a network-scale effect that may be more than just the sum of the parts. These properties appear to make planning restoration actions in stream networks a complex problem, best suited to approaches such as numeric optimization. We developed a framework to describe the movement of fish throughout stream networks, and how those patterns of movement are affected by changes to habitat quality and reduced connectivity caused by barriers to passage. We used this framework to investigate the nature of interactions between multiple restoration projects designed to improve either habitat quality or connectivity. Contrary to the expectations outlined above, our mathematical analysis showed that the interaction amongst multiple projects is small, and that under almost all circumstances, the total network-scale effect of multiple restoration projects is well-approximated by the sum of individually predicted effects. Interactions among multiple projects to improve connectivity will only be large when two projects cause very large increases in connectivity, and when those two projects are also close together. The predictions of the mathematical analysis were confirmed by simulation analyses. These conclusions apply to equilibrium conditions within dendritic networks, and the transferability of such conclusions to dynamically evolving systems is not yet known. However, even if non-additivity emerges in the dynamic evolution of a system, near-additivity will always be inherent in the underlying equilibrium framework. These findings are immediately important for planning of stream restoration programs. Rather than having to consider many combinations of projects using advanced approaches such as optimization, managers can rank individual projects by their expected individual benefit, and implement the top ranked projects. The framework can be implemented with the type of data that exist for many stream systems, and can readily consider multiple species and ontogenetic behavioural changes. In circumstances where additivity of effects cannot be assumed, the framework can still be used to assess the effects of combinations of projects. More generally, these findings are important for the field of riverine landscape ecology. Whilst rivers have become increasingly viewed as landscapes in their own right, it has been suspected that the topological restrictions of DENs may affect population processes in these landscapes. Accordingly, there have been recent calls for theoretical research to better understand population processes within DENs. We have shown here both analytically and through simulations, that dispersal and movement through DENs responds more simply to interventions in the landscape than has previously been assumed. Accordingly, our work constitutes an immediately useful contribution to the research required to better understand these unique systems.