Spatial Analysis of <I>Aedes albopictus</I> (Diptera: Culicidae) Oviposition in Suburban Neighborhoods of a Piedmont Community in North Carolina

Abstract

Temporal and spatial distribution of egg-laying by Aedes albopictus (Skuse) (Diptera: Culicidae) was investigated in suburban neighborhoods in Raleigh, NC, by using oviposition traps (ovitraps) at fixed sampling stations during the 2002 and 2003 mosquito seasons. Variations in the phenology of oviposition between the two mosquito seasons resulted from differences in the patterns and amounts of rainfall early in the season. Aerial images of each study neighborhood were digitized, and the proportions of specific types of land cover within buffer zones encompassing ovitraps were estimated. Retrospective analyses showed that in some neighborhoods, oviposition intensity was significantly associated with specific types of land cover. However, in general, it seemed that gravid Ae. albopictus searched throughout the landscape for water-filled containers in which to lay eggs. Peridomestic surveys were carried out concurrently with ovitrap collections to estimate production of Ae. albopictus pupae in discarded water-filled containers and the abundance of females in vegetation that made up the resting habitat. Results of linear regression analyses indicated that the mean standing crop of pupae (total and per container) per residence was not a significant predictor of mean egg densities in ovitraps. However, the mean standing crop of adult females was a significant but weak predictor variable, because the magnitude and sign of regression coefficients varied between neighborhoods. Linear spatial regression analyses revealed that oviposition intensity was not spatially dependent on pupal standing crop or the numbers of pupae-positive containers distributed peridomestically. However, a weak spatial dependence on the standing crop of adult females was found in some neighborhoods. Based on spherical variogram models, kriging was carried out to predict the spatial patterns of oviposition in suburban neighborhoods. Focal areas of high and low oviposition intensity were evident in most neighborhoods; however, the spatial patterns of oviposition changed between mosquito seasons. Kriging predictions were evaluated, using cross-validation, by systematically removing each data point from our data set and predicting the removed point by using the remaining points. The root mean square (standardized) error values of best fitting variogram models approximated 1, and plots of standardized PRESS residuals showed no distinct pattern for most neighborhoods, indicating that predictions of the spatial distribution of oviposition intensity were valid. Spherical variogram models are a satisfactory method for describing the spatial distribution of Ae. albopictus oviposition, and kriging can be a useful technique for predicting oviposition intensity at locations that have not been sampled.