A flow network model of Norway rat dynamics in an urban landscape

Background/Question/Methods

Following a traumatic event such as a natural disaster, ecological and societal communities concurrently change and reassemble in response to one another. Loss of predator species, abandonment, and vegetation regrowth following such an event can provide opportunities for human commensal pests such as Norway rats (Rattus norvegicus) to recover and spread quickly. This species can carry and transmit several zoonotic pathogens, posing a potential health risk to humans and domestic animals. Despite the global ubiquity of this species, relatively little is known about how natural and human-related changes in urban landscapes affect its reproduction and movement. As part of a project investigating recovery of human and natural systems in New Orleans, LA after Hurricane Katrina, we are modeling movement and dynamics of Norway rat populations. To identify key factors relating landscape features to population spread, we use a flow network model, which represents distance-dependent movement between locations. Network nodes represent areas containing active colonies modeled using continuous-time stage-structured population dynamics. Varying demographic parameters and network structure, we use this model to predict the fraction of sites that will become occupied during an invasion and identify reservoir populations, movement corridors, and critical fractions of sites that would require control efforts for mitigating spread. 

Results/Conclusions

We find that lattice and lattice-like networks with high regularity of distance in node placements admit full network occupancy with significantly fewer edges than networks with randomly placed nodes. To test the effects of random and targeted population control strategies, we imposed sink-dynamics at randomly distributed nodes across the network.  The effect of sink node quantity and location on occupancy and invasion time also depends strongly on network structure. I will discuss the sensitivity of invasion time to demographic parameters and describe how population density correlates with network neighborhood properties.  I will also describe how we are using extensive local data on rat demographics and genetics, ground cover vegetation data, and GIS models to parameterize movement and life history parameters in the model for a New Orleans site-specific application. With this modeling approach we can make general predictions about invasion, and combined with current data, it can aid in testing hypotheses about gene flow of rodents as they reoccupied post-Katrina New Orleans. This will also provide a tool for designing targeted intervention strategies for preventing rodent-borne disease spread, and help identify human and natural landscape properties that shape the population distribution of this widespread species.