Cutting into short logs reduces infestation by the bark beetle Polygraphus proximus

Polygraphus proximus Blandford (Coleoptera: Curculionidae: Scolytinae), a non‐aggressive bark beetle, has caused mortality of Abies spp. in Western Siberia, Central Asia, European Russia as well as Japan. Removing dead trees (e.g. salvage logging) and thinning stands are considered one of the best practices to reduce the risk of bark beetle outbreaks. However, salvaging or removing dead and thinned trees requires significant effort. Thus, to determine whether cutting trees into shorter logs would reduce P. proximus infestation, we exposed A. sachalinensis logs of various lengths on the ground in a forest stand. The attack density of P. proximus significantly decreased with shorter log lengths. It can be concluded that cutting into short logs and leaving them on the ground are effective practices to reduce the chance of P. proximus host location and outbreaks after natural disturbances and thinning.

thinned trees also demands considerable effort and these trees are often left in the stands, contributing to an increased probability of outbreaks.
Parent beetles must ensure that the mother galleries and larval galleries branching off from them have adequate space under the bark.If breeding materials are cut into shorter lengths, they may not provide sufficient space and parent beetles may avoid attacks.
Consequently, we hypothesized that cutting logs into shorter lengths can reduce bark beetle infestation, potentially decreasing the likelihood of future outbreaks.To determine whether cutting trees shorter would reduce P. proximus infestation and to what extent this reduction would occur, we exposed logs of various lengths in forest stands.Our aim in this study was to determine whether the attack density of P. proximus decreases with shorter log lengths.

| Study site
The present study was conducted at The University of Tokyo Hokkaido Forest (43°15′ N, 142°30′ E) in Furano City, Hokkaido, Japan (UTHF).Established in 1899, the UTHF covers 22,717 ha within a hemiboreal mixed conifer-broadleaf forest zone.The dominant tree species include evergreen conifers such as A. sachalin-

| Field experiments
To determine the attack density of P. proximus on different lengths of A. sachalinensis logs, we cut down five A. sachalinesis trees (diameter at breast height: 13-17.5 cm) at the study site, on April 14, 2023.We obtained one set of logs of different lengths (10 cm, 20 cm, 40 cm, 60 cm and 80 cm) from each tree.The cut ends of logs were covered with paraffin to prevent the logs from drying.Then, we exposed the logs on the ground at the study site, dividing them into two groups placed 20 m apart on April 17, 2023.Within each group, the logs of various lengths were randomly and closely placed.
We retrieved the logs on July 7, 2023, although we could not find one 20 cm log.Then, we measured circumferences at cut ends and the middle of each log and confirmed that the circumferences of the logs were not significantly different among the various lengths (12.9 ± 1.37 SD cm for 10 cm logs, 14.0 ± 1.92 SD cm for 20 cm logs, 13.9 ± 1.41 SD cm for 40 cm logs, 14.5 ± 1.78 SD cm for 60 cm logs and 13.9 ± 2.03 SD cm for 80 cm logs; ANOVA, F 4,19 = 0.559, n = 24, p = 0.695).Then, we peeled the bark from logs, recording the number of attempts by P. proximus (i.e.entrance holes).

| Statistical analyses
To determine the effect of length of logs on P. proximus attack, we used a generalized linear mixed model (GLMM) with a Poisson's distribution with log link, where the response variable was the number of entrance holes and explanatory variable was length of logs.Tree individuals considered as a random effect.The model included, as an offset, the bark surface area calculated by multiplying the length of logs by the length of the average circumference measured at cut ends and the middle of the logs (Dobson & Barnett, 2018).pvalues were calculated using Wald chi-square tests.All statistical analyses were performed using 'lme4' (Bates et al., 2015) and 'car' (Fox & Weisberg, 2019) packages in the R v.4.3.0 software (R Core Team, 2023).

| RE SULTS AND D ISCUSS I ON
The logs harboured a mean of 70.1 ± 20.2 SE entrance holes per m 2 .
No entrance holes were found in three logs each of 10 cm, 20 cm and 40 cm lengths and two logs of 60 cm length, while all logs measuring 80 cm had entrance holes.The attack density of P. proximus significantly increased with longer lengths (Wald χ 2 = 147.3,n = 24, p < 0.001) (Figure 1).This suggests that P. proximus more easily recognized or preferred these host longer logs.Therefore, it can be concluded that when the dead and thinned trees are left on the ground, cutting them into short logs might be an effective practice to reduce the chance of P. proximus outbreaks after natural disturbances and thinning.Köbayashi et al. (2022) showed that the mean (±SE) attack density on naturally fallen trees in the same study site was 153.8 (±31.5) per m 2 .In this study, the mean (±SE) attack density on the 80 cm logs was 188.8 (±47.3) per m 2 , while that on the 60 cm logs was 82.0 (±52.3) per m 2 .This suggests that 80 cm logs were sufficiently long to obtain normal density, but already 60 cm logs started to be too short for allowing a normal attack density of P.

proximus.
Bark beetles exhibit an attraction towards the volatile emissions from their hosts (Brattli et al., 1998;Lehmanski et al., 2023;Seybold et al., 2006).Conifer bark beetles also avoid the green leaf volatiles (Poland et al., 1998;Wilson et al., 1996;Zhang & Schlyter, 2004).In the present study, shorter logs may emit a smaller amount of host volatiles.being surrounded by understory vegetation such as bamboo grass in the study site, shorter logs could result in a significant decrease in attack density.
Bark beetles can also employ visual cues such as colour, shape and silhouettes in their host selection (Campbell & Borden, 2006;Goyer et al., 2004).In this study, shorter logs, more surrounded by understory vegetation, might be visually covered, further contributing to a significant decrease in attack density with shorter log lengths.Further studies are needed to fully demonstrate the effects of understory vegetation on bark beetle attacks.
Once the initial bark beetle has begun excavating an entrance hole in a host tree, it begins emitting a pheromone that serves as an attractant for other beetles.As time progresses, such chemical signals may increase, particularly in larger logs with space for more beetles.
Consequently, any small differences in attack density among the logs at the beginning might expand over time and likely contribute to the exponential increase of attack density of P. proximus with longer lengths observed in this study (Figure 1).While this study showed the difference in attack density, it is worth noting that reproductive success, such as number of eggs oviposited and larval survival rate, may also be influenced by log length.
Shorter logs resulted in lower parent density, but may provide more space for egg laying, potentially leading to better offspring survival rates.Conversely, shorter logs may dry out more quickly, which could possibly reduce the larval survival.Further studies are necessary to determine the effects of log length on the reproductive success of bark beetles.AUTH O R CO NTR I B UTI O N S Etsuro Takagi: Conceptualization; writing -review and editing; writing -original draft; visualization; funding acquisition; investigation; formal analysis.Mai Aochi: Conceptualization; writing -review and editing; investigation.

2008 and 2022, with a density of 440 trees/ha. For our field experiments, we selected another forest stand with A. sachalinensis trees planted in 1962, thinned in 2002 and 2015. This second forest stand also had naturally regenerated A. sachalinensis trees and various
deciduous broadleaved species, such as Populus siebolidii, Magnolia spp.and T. japonica.The tree density was 640 trees/ha.The bamboo grass Sasa veitchii covers understory of the forest (Appendix 1).