Binomial and numerical sampling for estimating density of Baris coerulescens (Coleoptera: Curculionidae) on Oilseed rape

• Published in: Journal of economic entomology 1999-08, Vol.92 (4), p.875-885
• Main Authors: Badenhausser, I, Lerin, J
• Format: Article
• Language: English
• Subjects: Baris; Baris coerulescens; binomial sampling; Biological and medical sciences; bootstrap simulation; Brassica napus oleifera; Brassica napus var. napus; Curculionidae; estimation; Fundamental and applied biological sciences. Psychology; Life Sciences; Phytopathology. Animal pests. Plant and forest protection; population density; Protozoa. Invertebrates; Records, symptoms, damages, economic importance, population surveys; sequential sampling; simulation; statistical analysis
• ISSN: 0022-0493; 1938-291X
• DOI: 10.1093/jee/92.4.875

Baris coerulescens Scopoli is a pest attacking the tap root of oilseed rape and other cruciferae. Sampling plans based on 6 yr of surveys were devised to estimate the mean number of immatures per plant. When the required precision (SE/mean) is around 0.10, sampling must be based on insect counts in the roots, and sequential sampling based on a variance-mean model provides a reasonable tool. When using the proportion of plants infested with at least 1 insect, a binomial sampling plan can be used to estimate the mean density with a level of precision > or = 0.40. A better precision can be reached (> or = 0.25) and a smaller sample size is required, when using the proportion of plants infested with at least 2 insects. Sampling characteristics of the binomial sampling plan developed with the proportion of plants with at least 2 insects and precision 0.30 showed that it can be used for intermediate means from 2 to 6 insects per plant. In these conditions, means are well estimated and the sampling time is reduced when compared with sampling based on insect counts. Minimizing sampling time led to developing a sampling strategy with precision 0.30, consisting of using binomial sampling when the proportion of infested plants with at least 2 insects is in the range 0.50-0.80 and in using numerical sequential sampling in other cases. Simulations of these sampling plans were carried out on 16 bootstrap samples that cover a wide range of means from 0 to 10 insects per plant.