Evaluation of different sampling plans for estimating Coraebus florentinus (Herbst) (Coleoptera: Buprestidae) population density in Mediterranean cork oak forests

Abstract: The black-banded oak borer, Coraebus florentinus, is one of the main emerging pests of cork oak (Quercus suber L.) trees in the Mediterranean area. Larvae of this insect are
xylophagous and develop in tree branches, where they feed on conducting tissues and cause
dieback of infested branches. Despite the increasing spread of this pest throughout cork oak
forests in the Mediterranean area, a well-defined sampling protocol to estimate its population
density has not been defined yet. For this reason, different standard sampling protocols (i. e.,
linear transect, random sampling, and fixed-time sampling) were tested in ten cork oak forest
stands in Sardinia (Italy) in 2021 and the correlation between infestation density estimates and relative abundance of C. florentinus was determined. Moreover, binomial sequential sampling plans based on the functional relationship between the proportion of occupied sampling units (i.e., trees) and the average density (i.e., infested branches/tree) developed by Kono and Sugino (KS) and Wilson and Room (NB) empirical models at two different levels of precision (D) (0.25 and 0.15). In the study area, the average density of C. florentinus ranged from 0.42 ± 0.04 (mean ± SE) to 1.20 ± 0.14 infested branches/tree. Among all tested sampling protocols, the 100-m linear transect and the fixed-time sampling of 15 and 20 minutes were the most valuable methods to estimate C. florentinus infestation density, whereas random samplings tended to underestimate it. A strong and positive relationship between the average density and the proportion of occupied sampling units was found for both empirical models tested (KS: R2 = 0.85; NB: R2 = 0.85;). The sample size required to estimate the observed minimum density (0.42 infested branches/tree) at D = 0.25 was approximately 25 and 20 trees for KS and NB empirical models, respectively. Compared to standard sampling methods, the application of the binomial plan might reduce the time required for sampling, thus allowing to increase the number of sites that can be sampled throughout an infested area.