The research was carried out in central Bohemia, Czechia, close to the town of Chlumec nad Cidlinou. The forest area was nearly 1,000 hectares. The altitude ranged from 200 to 260 m a.s.l. The majority of the stands were dominated by oaks, among which the most widely represented species were Quercus robur L. and Quercus petraea (Matt.) Liebl. The research site was a part of a game reserve where the stands were managed by selective logging.

We selected five stands per year during our research from 2015 to 2017 (Table 1). The stands and their surroundings were composed of vital trees. The selected oak stands were harvested during the spring of a particular studied year. All stands consisted of mature oaks (i.e., more than 100 years old), and the proportion of oaks was at least 60%, i.e., oak was the dominant tree species in the stand species composition.

We conducted research on the stumps that, after tree felling, can contribute to P. cylindrus population build-up in the next season. A postharvesting survey of stumps was performed in each of the study stands after the peak flight activity of P. cylindrus adults in the second half of August.

During the assessment, the number of entry holes (variable intensity = number of entry holes per 1 dm²) was counted within two linear transects with a length of 200 meters that included 14 to 24 stumps (Table 1). All stumps in the transect were evaluated, regardless of whether P. cylindrus infestation was detected or not. Each analyzed tree stump was completely debarked with an axe (Figure 1) and the bark area was measured. The determined number of entry holes was converted to a number per reference area of 1 dm². The diameter (variable diameter) of a particular stump with bark was measured in centimeters.

Our experiments began at the beginning of June and lasted until the end of August. At the five study sites, commercial flight barrier traps (type Theysohn - Theyson Kunstststoff. GmbH, Germany; black color) were placed 1.5 m above the ground approximately 10 m from the nearest standing oak tree (Figure 1). The sampling of trapped beetles and other arthropods took place regularly every 7-10 days (variable day) during all the years (variable year) of the research.

In 2015, one trap per study site with the pheromone lure Cylindriwit (Witasek Pflanzenschutz GmbH, Austria) was used. Cylindriwit^® contained ethanol, (±) 6-methyl-5-hepten-2-ol, 6-methyl-5-hepten-2-one and 1-hexanol. The Cylindriwit^® lures were changed after 10 weeks (as recommended by the manufacturer), and ethanol evaporators were added continuously during monitoring. The insects were stored in a 70% ethanol solution.

In 2016, there were two traps used in the study site. One trap was baited with 96% ethanol in a polyethylene bottle with holes at the top for evaporation, and the second was baited with Cylindriwit^®. The traps were approximately 20 m apart. In each sample, nontarget coleopteran species were also counted, but their identity was not determined in more detail.

In 2017, there was installed just one trap for study site again with Cylindriwit^® pheromone lure.

In each sample, the number of P. cylindrus specimens was counted, and the sex was determined (based on the outer marks on the abdomen) (Figure 1).

A detailed analysis of the nontarget organisms found in the pheromone traps baited by Cylindriwit^® was carried out in the 2017 samples. The nontarget species were classified into taxonomic groups (orders and families).

We used the glmmADMB R package for the computation of the GLMMs (generalized linear mixed-effect models). The models were computed as zero-inflated with a negative binomial distribution of the dependent variables. All the independent variables were first controlled for multicollinearity using the VIF (package HH).

For the first research aim, intensity was the dependent variable, site was used as a random factor, and diameter and year were the fixed factors. For the second aim, the adult males, females, and nontarget taxa trapped were used as dependent variables, day and compound (ethanol or Cylindriwit) were the independent variables, and rank was a random factor. In the third aim, we used all the individuals (variable adults) trapped and then the males and females separately as the dependent variables. The year was used as an independent factor, and the site was a random factor.

A zero-inflated model was not used in the case of the nontarget arthropods. We computed the GLMMs using the abundance of nontarget species as the dependent variable and the number of adults of P. cylindrus as an independent variable with the site as a fixed factor. Only the nontarget taxa with abundances higher than or equal to five individuals were analyzed.

We also used the package party for the threshold computation using the conditional inference tree method (ctree) for the analysis of the first aim (Hothorn et al., 2006), with intensity as the dependent variable and the diameter and year as the independent predictors.

All analyses were performed in R 3.0.2.

We examined 1,684 oak pinhole borer entry holes in 314 studied stumps. The mean attack intensity was 1.10 entry hole per dm² from 2015 to 2017 (Table 2). The attack intensity by the oak pinhole borer significantly increased with increasing diameter of the oak stumps (mean = 61.42 ± 13.43 cm, minimum = 33 cm, maximum = 103 cm; Table 2). The year did not influence the attack intensity (Table 3).

We found a significant threshold (Statistic = 8.29; P = 0.008) for the diameter of the oak stumps regarding the attack intensity by the oak pinhole borer. The threshold was found at a stump diameter of 61 cm (Figure 2). The number of stumps under or equal to this value was 169; the number of stumps above this threshold was 145. The mean value under or equal to this threshold was 0.66 entry holes, and 1.61 holes per dm² was the mean above this value.

In 2016, we trapped 5,747 oak pinhole borer adults, of which 4,646 were females. We also observed 348 trapped individuals of nontarget beetles, which was 5.7% of the total number of individuals. In total, 160 nontarget beetles in Cylindriwit and 188 in ethanol were observed. The number of trapped adults and nontarget beetles decreased from the beginning of the experiment to the peak flight activity of P. cylindrus. The type of compound used, had a positive impact on the trapping efficacy of the oak pinhole borer males, females, and total individuals. Pheromone traps proved to be more effective than ethanol-baited traps (Figure 3). The type of compound had no significant effect on the nontarget beetles (Table 4). We trapped 12,504 adults, of which 8,999 were females, during the 3 years of the study. We observed a significant difference in trapping success among the three study years (Figure 3). The number of males, females, and adults in total was significantly lower in the last year than in the previous 2 years (Table 5).

We trapped 181 nontarget arthropods in the pheromone-baited oak pinhole borer traps (Table 6). Jewel (Buprestidae) and click (Elateridae) beetles were significantly more abundant in the traps with an increasing number of trapped oak pinhole borers, while other taxa did not demonstrate a significant relationship (Table 6).

The key factors involved in oak diebacks are the actual site conditions and management strategies (Führer, 1998; McDowell and Allen, 2015). These factors affect both forest stands and individual trees (Führer, 1998; Thomas et al., 2002). P. cylindrus belongs to species whose abundance coincides with local oak densities. It has been shown that their population densities are highest in mid-size gaps (Bouget and Noblecourt, 2005). The mean intensity of oak pinhole borer attack was 1.1 entry holes per dm² on the stumps in our study, and the average density of individuals was approximately 55 beetles per dm³ (Belhoucine et al., 2011a).

Host selection by the oak pinhole borer depended primarily on the diameter of the oak stumps. We found a significant increase in ambrosia beetle density with the diameter of the oak stumps. This partly coincides with the spatiotemporal distribution of P. cylindrus, which was not random, indicated by cork oak stands. Nevertheless, there were more factors related to the attacks beside stump dimensions, such as weakness and exploitation (Sousa and Debouzie, 1999). In particular, oak pinhole borer males prefer trees that are already weakened, show discoloration symptoms and have higher diameters (Sousa and Debouzie, 1999). In other research, oak pinhole borer abundance was associated not only with stand and tree characteristics such as dimension, genes and localization (Mattson et al., 1991) but also with debarking and cutting methods in the case of cork oak (Sousa and Inacio, 2005).

Important information regarding commercial losses is that timber with holes caused by P. cylindrus is not destroyed significantly (it is mainly in sapwood), but the appearance of the final products will be spoiled (Tilbury, 2010). To ensure the commercial potential of oak timber, effective management at sites with high abundances of P. cylindrus appears to be shelterwood cutting. This allows for medium sunlight compared to standards (better fructification and growth). At the same time, there will be no intense sunlight on the remaining oak trees and forest floor in the stand, and possible drought stress will not be significant (Schlesinger et al., 1993). Oak stands could also be underplanted with seedlings, including those of other tree species, to make the microclimate of the stand less favorable to wood-boring insects (Thomas et al., 2002; Tilbury, 2010). To prevent damage by the oak pinhole borer, the cutting of oaks with a target diameter of more than 61 cm could be suggested.

Regarding interannual variability, a good management strategy would be to avoid harvesting from June to September, i.e., during the time when the adult beetles are dispersing and colonizing logs. All harvested oak timber should be removed or debarked (including stumps) from forest stands as quickly as possible, mostly before the flight period (Tilbury, 2010).

Even if this strategy is not suitable for mixed stands (Jonsell and Schroeder, 2014; Miklín and Čížek, 2014), in some cases, stump harvesting as a renewable energy source might also be used, especially in places with a strong threat of P. cylindrus outbreaks (dieback occurrence or sanitary or salvage cuttings after wind or fire damage). Nevertheless, this strategy is mainly used in boreal forests and for softwood tree species. One solution can be extracting stumps with the use of wood-chipping, which is sometimes applied in central Europe. This approach limits the mass of deadwood and thus affects the total diversity of the stand. However, it should be avoided in conservation forests and stands with protection status (Miklín and Čížek, 2014).

The flight activity of adults typically begins in May and ends in September in Europe (Sousa and Inacio, 2005). Nevertheless, sporadic emergence of P. cylindrus occurs throughout the whole vegetation season (Baker, 1963; Sousa and Inacio, 2005). In our study; most of the individuals were trapped in June and July. In the Mediterranean region, flying lasts until November (Catry et al., 2017). The main reasons are the life cycle of this cryptic species inside the tree for a long period of time and the high variability in egg laying, which means variability in life stage occurrence. This life strategy leads to a long emergence period (from spring to autumn) and may continue for a second generation during the spring of the following year. Imago activity stops only in winter, before which time males block entry holes of the gallery systems with a mixture of sawdust and secretions (Sousa and Inacio, 2005).

The abundance in the traps was significantly higher with the pheromone lure. This finding may be important for pest management and monitoring. Ethanol also attracted adults, but, consistent with other works, we found the abundance of adults attracted to be low (Markalas and Kalapanida, 2005). Monitoring using pheromone traps is a suitable detection method for estimating the size of a P. cylindrus population, focusing on installation at peak population densities in June and July in Central Europe. We do not recommend monitoring during other time periods outside the peak of flight activity because this method will eliminate nontarget insect species, as in the case of other saproxylic insect species (Lubojacký and Holuša, 2013, 2014).

More than three times more females than males were trapped in the pheromone traps at our study localities. The detected sex ratios were much higher than those previously reported in southern Europe (Catry et al., 2017). These sex ratio differences were most likely due to the different attractiveness to the pheromone compounds for the sexes. The building of galleries is initiated by males, which produce pheromones to attract a female. By producing pheromones, other beetles are attracted, which leads to frequent concurrent attacks (Atkinson, 2004).

P. cylindrus abundance was only related to the occurrence of the beetle families Buprestidae and Elateridae in 2017. One of the most plausible reasons is that jewel beetles (mainly from the genus Agrilus) frequently occur during oak declines and sometimes strongly affect forest stands (Evans et al., 2004; Sallé et al., 2014). Increasing population densities of Agrilus biguttatus (Fabricius, 1776) are often connected with oak declines after defoliations or severe drought events (Moraal and Hilszczanski, 2000; Hilszczański and Sierpinski, 2006). This might explain the correlation in abundance with P. cylindrus, whose abundance also increases due to oak stress (Henriques et al., 2006; Akbulut et al., 2008), and both groups of pests are apparently attracted to similar volatile substances associated with oak declines. Nevertheless, jewel beetles are not frequently found in passive traps; thus, it is hypothesized that they might use the pheromones of the study species as a kairomone. The same situation might be the case for click beetles. However, the most important reason is probably the fact that adult click beetles are very active, frequently collected using window traps (Horák and Rébl, 2013) and sometimes found in conspicuous numbers in oak forest stands (Loskotová and Horák, 2016).

The population density of the oak pinhole borer still may not reach sufficient level to contribute to oak diebacks in central Europe. However, our study indicates that the population density could be very high in oak stumps. A coarse stump diameter is likely the most important attractant for borer attack; the observed presence of the borer in European beech coarse stumps in mountains (i.e., above 600 m a.s.l.) coincides with the above-mentioned results but also indicates two other risks. The first is the possibility of attacks on other commercially important tree species. The second is a shift to high elevations, which is highlighted by its observed long flying period and the habitats it utilizes (wood). Comparable catches of nontarget beetles in traps baited with Cylindriwit and ethanol support the use of pheromone lures for oak pinhole borer monitoring. This result explains the composition of the lure, the main ingredient of which is also ethanol. An insignificant effect on nontarget invertebrates indicates that additional pheromone lure composition attracts target pests but does not affect other groups. This information appears to be important regarding the use of biodiversity-friendly monitoring methods that do not affect other species.