Organized Oral Session 16: Linking data and theory in dendritic ecological networks: from ecological problems to rapid understanding, was organized by Evan H. Campbell Grant, with William F. Fagan and Winsor H. Lowe as co-organizers, and held during the 2009 ESA Annual Meeting in Albuquerque, New Mexico, on 4 August 2009.
Considerable work on complex systems has focused on “node-based” networks, such as food webs or metapopulations, where tools from network theory have been brought to bear on understanding linkages among system components. However, in many areas of ecology, spatial networks with explicit branching geometry are of particular interest (e.g., river networks, transportation infrastructure, individual plants). In order to maintain the persistence, functioning, and integrity of these systems, it is essential that we improve our understanding of when and how the details of their geometric configuration become important.
This session was structured to (1) investigate ecological problems made more interesting by dendritic network structure, (2) provide examples of how theory and data are combined to investigate these types of networks, and (3) describe recent developments in theory which is specifically applicable to dendritic networks. The objectives of the session were to bring together empirical and theoretical ecologists, to exchange ideas on conservation and management problems in dendritic ecological networks, and discuss how existing theoretical tools can be applied to these problems. The development of theory in other types of spatial networks (e.g., metapopulations, interaction networks, the World Wide Web) may be applied to dendritic networks, though special characteristics of these networks (e.g., unique architecture, lack of loops, dominance of advection, inherent connectedness lacking discrete patches, variability in continuous space) require an understanding of the ways in which dendritic networks are unique from their two-dimensional counterparts.
The complexities induced by the spatial structure of dendritic networks were a central theme of the session. The seven presentations were diverse, with two (Anderson, Muneepeerakul) developing theoretical tools useful for dendritic networks, two (Hitt, Gilliam) providing insight into the empirical systems and management problems made more complex by the dendritic architecture of stream networks, and three (Lynch, Grant, and Cuddington) linking theory and data to better understand ecosystem dynamics in dendritic systems. The session organization emphasized that the most rapid path to learning involves identifying problems arising from field observations of conservation problems, and developing or modifying existing theory to explicitly account for the dendritic architecture of these networks.
Identifying key challenges to habitat management is critical for planning responses at the appropriate temporal and spatial scales. On this topic, Than Hitt (USGS Leetown Science Center; “Management and conservation in stream networks”) and coauthors discussed how incorporating network architecture and spatial position could improve management and conservation in stream networks. Hitt identified three challenges to stream ecosystem management that require the explicit consideration of spatial configuration: (1) threshold responses to fragmentation, (2) pathogen and contaminant spread through a complex, directed network, and (3) the importance of matching observation and dispersal grains to make correct inferences.
Following the recommendations to understand how organisms utilize space, and how spatial complexity might underlie these patterns, Jim Gilliam (North Carolina State University; “Fish in the tips of dendritic drainages: a case study, with surprises”) and collaborators provided an empirical example. Gilliam identified two types of fragmentation in his Trinidad river system: spatial fragmentation (caused by the interaction between predator and prey fish species) and dynamical fragmentation (where the presence of a predator inhibits movement under normal conditions). Though the presence of a predator in the mainstem of the network influenced the distributional patterns of the prey species, periods of hydrologic disturbances (high-flow events) may have provided the opportunity for dispersal among distant tributaries.
To generalize patterns observed in networks with a dendritic architecture, it would be prudent to investigate the application of existing tools for understanding populations and communities in complex networks. Rachata Muneepeerakul (Princeton University; “On explicitly spatial theoretical tools for a dendritic world”) and coauthors did just this, applying theoretical tools for (1) understanding the distributional patterns of organisms in a large dendritic network, and (2) understanding how the restrictive architecture of a dendritic network influences population spread and the evolutionary stable state of dispersal. Muneepeerakul found that multiple distributional patterns of freshwater fish in the Mississippi–Missouri river basin can be understood under a modification to the unified neutral theory, while explicitly considering that the dendritic architecture improved the fit of the model to some empirical patterns. Using an application of game theory, Muneepeerakul finds that within the restrictive spatial structure of a dendritic ecological network (and without spatial correlation in patch quality), a population evolves simultaneously more local and widespread dispersal, which may represent a response to the network architecture.
There are several special properties of dendritic networks that warrant detailed investigation, including how the branching architecture may underlie patterns of movement and contribute to population stability. Evan Grant (USGS Patuxent Wildlife Research Center; “How dendritic ecological networks structure the movement and underlie the stability of stream salamanders”) tested this prediction in a streamassociated amphibian. Using individual mark–recapture of stream salamanders, Grant described the use of two movement pathways in headwater stream networks, empirically deriving the first estimates of both within-stream and overland dispersal in the same system. He then related these observations on movement behavior in stream salamanders to simulation model results, finding that the influence of the dendritic network architecture on the observed movement behavior can underlie population stability in stream salamanders.
Rapid learning in ecology comes from the intersection of theory and empirical observations. Kim Cuddington (University of Waterloo; “Diffusion-limitation in predator-prey systems”) took this approach in summarizing how dendritic networks may influence predator–prey dynamics differently than in their two-dimensional counterparts. Cuddington used a pea-aphid-ladybird system to test predictions from her earlier theoretical work on predator/prey dynamics. Her experimental work confirms that the branching structure of dendritic networks alters movement rates, predation rates, and thereby influences population dynamics.
While dendritic networks represent a wide range of terrestrial and aquatic ecosystems, the most numerous examples come from river systems, and indeed river networks featured prominently in the session. Kurt Anderson (University of California, Riverside; “Using simple models to predict population responses to spatial variability in dendritic stream networks”) pointed out that a dominant characteristic of stream networks is the strong advection. Anderson exploited this feature of stream networks to first develop a model of insect populations within a branching network, including variation in emigration induced by the connection points between two habitat branches. These results may explain the spatial structuring of populations in dendritic river networks, while more work needs to be done to link process and pattern in these types of networks.
Application of theoretical tools to management decisions should be a useful tool to resource managers and ecologists, especially where large-scale changes in ecosystems are proposed. Heather Lynch (University of Maryland; “India's Inter Basin Water Transfer project: the impact of network manipulation on freshwater fish communities”) and coauthors presented an interesting example of an applied problem in rivers management. Lynch reported on her work applying a neutral model to the Peninsular Indian river network, and investigated how proposed canal connections would be expected to alter patterns of species richness. She found that the addition of canals connecting hydrologically isolated river basins facilitates the spread of common species and increases average local species richness without changing the total species richness of the system. These impacts were sensitive to the parameters controlling the spatial scale of fish dispersal, and suggest that fish species capable of extensive dispersal will likely experience the greatest impacts from the IBWT project.
Overall, the session highlighted the ways in which dendritic ecological networks are an important class of spatially structured systems, and presenters in the session repeatedly pointed to complications arising from the network architecture. Many challenges for ecosystem and species management, conservation, and ecological services depend on our understanding of the spatial control of ecosystem patterns and processes. Rapid progress on understanding these systems will require simultaneously developing theory for the structure, dynamics, and function of dendritic ecological networks, and collecting empirical data explicitly designed to test this theory.