F. Fages and C. Piazza (Eds.): FMMB 2014, LNBI 8738, pp. 159–161, 2014.
© Springer International Publishing Switzerland 2014
The Challenges of Developing Spatially Explicit Network
Models for the Management of Disease Vectors
in Ecological Systems
Brendan Trewin
1,2
, Hazel Parry
1
, Myron Zalucki
2
,
David Westcott
1
, and Nancy Schellhorn
1
1
CSIRO, Brisbane, Australia
2
University of Queensland, Brisbane, Australia
Challenges of modelling vector-borne disease systems result from complexities and
uncertainities inherent in the vector’s behavioural ecology and its interactions in a
landscape context. Network models provide a number of approaches and measures to
quantify spatially-explicit systems that are consistent with the ecological process of
vector dispersal, with implications for disease transmission and spread [1,2]. Here we
discuss two spatially explicit vector systems as network models; (1) the movement of
the invasive mosquito Aedes aegypti, which vectors a number of diseases including
dengue fever, through rainwater tanks in a major urban area, (2) the movement of bats
(flying-foxes), which vector Hendra virus, through urban and rural landscapes [3]. We
contrast the design and applicability of these networks, comparing features and chal-
lenges inherent in modelling these systems, and discuss the use of network models as
disease vector management tools with implications for disease spread.
In an ecological context, nodes often represent metapopulations and compartmental-
ize important demographic characteristics such as growth rate, disease transmission rate
and spatial location within landscapes. In our mosquito model, rainwater tanks are
nodes that are fixed in both space and time, with accurate location data available from
government rebate schemes. Depending on whether nodes are exposed to the environ-
ment (non-compliant) or not, tanks are nodes that may act as sources or sinks for mos-
quito vectors respectively. Characteristics that govern population growth within each
source node are simple to collect and model as there is a vast literature on simulating
population growth within containers [4]. Within the bat-Hendra model nodes are likely
to be bat camps (roosts), containing populations of vectors. The highly seasonal nature
of camps and their susceptibility to variations in environment and climate result in un-
certainties in spatial location of the camps. Bats have high dispersal abilities with com-
plex movement and social behaviours. This leads to large fluctuations in the formation
and removal of nodes through space and time. Foraging sites could be additional nodes
within this system, but are difficult to model explicitly due to their inherent stochasticity
and have so far been ignored. Important simplifying assumptions are made in characte-
rizing bat camps as nodes in a network model compared to rainwater tanks, as the tanks
better reflect our compartmentalized concept of ‘nodes’ in a network. These assump-
tions introduce uncertainty into any conclusions that are drawn about the bat disease
vector system, but this uncertainty is not made explicit.