Model for Estimating Relative Abundance of Periplaneta fuliginosa (Dictyoptera: Blattidae) by Using House and Landscape Characteristics

Abstract

The relationship between smokybrown cockroach, Periplaneta fuliginosa (Serville), abundance and habitat was quantified using canonical correlation. Cockroach habitat was described, using 18 house and landscape characteristics based on a sample of 60 homes. Cockroach abundance, estimated by three monthly mean trap catches at each home, was correlated With habitat. After preliminary analysis, only 11 of the house and landscape characteristics were retained in the model. No characteristics dominated the description of habitat (an index) but some contributed positively and others negatively. The characteristics having the largest standardized weights in the habitat index were percentage of lot occupied by home, tree density, number of pets, number of residents, chemical pest control, age of house, and obvious cockroach harborages on tilt' property. Unbiased estimates of variables weights of the habitat index obtained by the Jackknife method demonstrated the fit of the model and revealed minimal effect of outliers; of 14 weights, 11 had standard deviations <20% of the mean. We produced that cumulative cockroach abundance would increase linearly with cockroach habitat index. Regressions of a control data set (n = 44) and a validation data set (n = 16) both produced positive slopes and, in addition, neither slopes nor intercepts were significantly different between these two data sets. These results validated the correlation model and indicated that it is applicable to all homes in Auburn, not just those in the control data set. Use of untransformed data produced similar results and showed that a cockroach habitat index may be used readily by pest control operators or homeowners.