U.S. Department of Agriculture: Animal and Plant Health Inspection Service
Date of this Version
January 2002
Abstract
Accurate estimation of crop damage by wildlife often requires labor-intensive sampling procedures. Variable area transect (VAT) sampling has been identified as a potential labor-saving alternative to quadrat sampling. We fully enumerated all damaged corn plants in a field to assess accuracy of sampling by small quadrats 5m long by one or two row-widths, and by VAT sampling with searches from random starting points to the 3rd, 4th, 5th, and 6th damaged plant. Two widths for strip transects also were considered for VAT sampling: one and two crop rows. Two analytical formulae were considered for the VAT data. Thus, 18 estimates of damage were produced and compared for accuracy with the true damage level. Sampling and enumeration were carried out the same day, so that the damage level could not change between enumeration and sampling. Estimation from VAT sampling using the original formula showed promise as a low-labor assessment method by producing highly accurate estimates (o1.5% absolute relative bias) when strip transects of two row-widths were searched to the 4th or 5th damaged plant. The formula designed to improve estimation for VAT sampling in aggregated populations had high biases. Quadrat sampling is well-known to produce unbiased results when optimized, and we restricted our research to only the low-effort situation, which also produced biased results.
Comments
Published by Crop Protection 21 (2002) 101-105.