State-dependent sampling bias in insects: implications for monitoring western tarnished plant bugs

Insect populations vary in the proportion of individuals exhibiting a particular ‘state’ (e.g., developmental stage, sex, egg load, or nutritional status). Because an insect’s developmental state often determines both its size and its behavior, it is likely that this will also affect the probability of being sampled. We propose that a comprehensive approach to pest management must consider the interaction between the structure of a pest population (i.e., the relative number of individuals in each state) and any statedependent sampling bias. To illustrate the usefulness of this method we sampled populations of the western tarnished pest bug, Lygus hesperus Knight (Heteroptera: Miridae), in cotton fields. Our sampling technique utilized large cages to measure the absolute densities of each L. hesperus stage and adult sex within a population. This technique allowed us to document a wide range of absolute stage structures and sex ratios across 10 L. hesperus populations in California. Using a combination of cage samples and sweep net samples, we quantified the state-dependent sampling bias by calculating the efficiency of sweep sampling as a function of L. hesperus developmental stage and sex. We found that the efficiency of sweep nets increased steadily with each successive developmental stage (i.e., nymphal instar) of L. hesperus. We also found that sweep nets are slightly more efficient in capturing male vs. female L. hesperus adults. Since other studies have documented that the stage and sex of L. hesperus can affect feeding impact on cotton flower buds, our results suggest that accurate predictions of Lygus damage will need to incorporate stage and sex-dependent sampling biases.