Ferwerda, Carolin. Characterizing the relationship between Asian tiger mosquito abundance and habitat in urban New Jersey. Retrieved from https://doi.org/doi:10.7282/T3348KJN
DescriptionSince its introduction to North America in 1987, the Asian tiger mosquito (Aedes albopictus) has spread rapidly. Due to its unique ecology and preference for container breeding sites, Ae. albopictus commonly inhabits urban/suburban areas and is often in close contact with humans. An aggressive pest, this mosquito species is a vector of multiple arboviruses. In order for mosquito control efforts to remain effective, control of this important vector must be guided by spatially explicit habitat models that aid in predicting mosquito outbreaks.
Using linear regression, I determined the relationship between adult Ae. albopictus abundance and climate, census, and land use factors in nine urban/suburban study sites in central New Jersey. Systematically collected adult counts (females and males) from July to October 2008, served as estimates of abundance. Fine-scale land use/land cover data were obtained from object-oriented classifications of 2007 CIR orthophotos in Definiens eCognition. Mosquito abundance data were tested for spatial autocorrelation via Moran’s I, semivariograms, and hotspot analysis in order to reveal consistent patterns in abundance.
Spatial pattern analysis produced little evidence of consistent spatial autocorrelation, though several sites exhibited recurring hotspots, especially in areas near residential housing and vegetation. Stepwise multiple regression was able to explain 20-25 percent of variation in Ae. albopictus abundance at the ‘backyard’ or cell level and 72-78 percent of variation in abundance at the ‘neighborhood’ or study site level. Meteorological variables (temperature on the trap date and precipitation), census variables (vacant housing units and population density), and more detailed land use/land cover classes (deciduous woody vegetation, rights-of-way and vacant lots) were frequently selected in all eight models, though many other independent variables were included in the individual models. The results of the spatial statistics suggest that clustering may occur at a broader extent, while the superior predictive ability of the site level models over the finer grain cell level models supports this conclusion. Future work should focus on validating these models with 2009 field data and testing whether finer grain weather and census data enhance the models’ predictive ability. Given the major differences between individual county models, future studies should further explore variations in Ae. albopictus habitat preferences in different geographic locations.