Ambrosia beetle response to ethanol concentration and host tree species

Ethanol emitted by stressed trees is an olfactory cue used by ambrosia beetles (Coleoptera; Curculionidae; Scolytinae) to locate susceptible hosts to colonize. In addition, ethanol enhances the growth of ambrosia beetle fungal symbionts, improving colonization. Whether host selection and colonization are affected also by the amount of ethanol produced by stressed trees and by tree species is unclear. To investigate these mechanisms, we induced attacks by ambrosia beetles in bolts of eight tree species by coring and filling them with either 5% or 90% ethanol solutions in water. For each ethanol concentration, bolts of the eight different tree species were replicated six times in a randomized complete block design. Entry holes were used as a proxy for host selection whereas gallery development stage was used as a proxy for colonization. Ethanol concentration differentially affected host selection of the three ambrosia beetles that were active during this study. Anisandrus dispar Fabricius preferentially attacked bolts with 90% ethanol concentration, Xylosandrus crassiusculus (Motschulsky) preferentially attacked bolts with 5% ethanol concentration, and Xyleborinus saxesenii (Ratzeburg) attacked bolts irrespective of ethanol concentration. Colonization of X. crassiusculus reflected the same pattern observed for entry holes. The effect of host tree species on host selection was most prominent for X. saxesenii, while X. crassiusculus established a higher number of developed galleries in Ostrya carpinifolia Scopoli bolts than on five of the other tested tree species. Our results suggest that ethanol concentration and host tree species may influence ecological niche partitioning among ambrosia beetle species.

Most ambrosia beetles colonize dying or recently dead hosts (Hulcr & Stelinski, 2017), but some species preferentially attack living but stressed trees Wang et al., 2021). To discern among stressed and healthy trees most ambrosia beetles exploit olfactory cues, in particular ethanol (Oliver & Mannion, 2001;Ranger et al., 2010Ranger et al., , 2012Ranger et al., , 2021Reding et al., 2011;Werle et al., 2019), which is induced and emitted by trees stressed by abiotic (Kelsey et al., 2014;La Spina et al., 2013;Ranger et al., 2013Ranger et al., , 2019 or biotic (Kelsey et al., 2013;McPherson et al., 2008) factors. Ethanol within host tissues can also influence ambrosia beetle colonization. Apparently-healthy trees baited with ethanol are attacked but not colonized by Xylosandrus germanus (Blandford) or other ambrosia beetles, while X. germanus foundresses establish fungal gardens and offspring within stems of trees irrigated with dilute ethanol solutions (Ranger et al., 2018). Ethanol incorporated into agar based media also promotes the growth of certain ambrosia beetle nutritional fungal symbionts and inhibits the growth of antagonistic fungi (Lehenberger et al., 2021;Ranger et al., 2018).
Ambrosia beetle host selection and colonization are also affected by the amount of ethanol associated with stressed trees (Kelsey & Joseph, 1997, 1999Klimetzek et al., 1986;Ranger et al., 2011). Kelsey et al. (2013) documented four times more ambrosia beetle attacks above ethanol-infused sapwood tissue than in the opposite side of the same log. Xylosandrus germanus and Xyleborinus saxesenii were differentially attracted to bolts soaked in varying concentrations of ethanol; the number of entry holes decreased with increasing ethanol concentration for X. germanus and increased for X. saxesenii . Only X. germanus established successful galleries, and the number of emerged X. germanus adults increased and then decreased with increasing ethanol concentrations , following the pattern observed for the growth of its fungal symbiont (Ranger et al., 2018). These results suggest that different ambrosia beetle species are more attracted by a certain ethanol concentration over others, which might correspond to the optimal concentration at which their colonization is maximized.
Although most ambrosia beetles attack a broad range of species (Hulcr et al., 2007), certain tree species may be preferentially attacked over others (Egonyu et al., 2017;Mayfield & Hanula, 2012;Reding & Ranger, 2020). For example, X. germanus preferentially attacks bolts of chestnut (Castanea sativa Mill.)  or red maple (Acer rubrum L.) (Reding & Ranger, 2020) over a number of other broadleaved species. Xylosandrus crassiusculus entrance holes were more than five times higher on California bay laurel (Umbellularia californica (Hook. & Arn.) Nutt) than on camphor tree (Cinnamomum camphora (L.) J. Presl), and only a few holes or none at all were found on seven other species (Mayfield et al., 2013).
In the present study, we examined the effect of ethanol concentration and host tree species on host selection and colonization of ambrosia beetles. We hypothesized that different host tree species and different ethanol concentrations within the host tissues would influence interspecific differences in the attraction and colonization of ambrosia beetles. These results are discussed in relation to a niche partitioning mechanism whereby host tree and ethanol concentration might assist in reducing interspecific competition for resources of ambrosia beetle species with overlapping flight activity period.

| Study site, bolt preparation and experimental scheme
The study was conducted in a broadleaf forest (45° 17' 14''N; 11° 41' 9''E) located on the northern slope of Monte Fasolo (260 m a.s.l.) in the Euganean Hills area, Veneto region, north-eastern Italy ( Figure S1). The forest is dominated by O. carpinifolia and Quercus pubescens Willd., with lesser amounts of C. sativa, Fraxinus ornus L., and Robinia pseudoacacia L., and was selected because it hosts a diversity of both native and exotic ambrosia beetle species. The climate is characterized by an annual mean temperature of approximately 14°C and precipitation of approximately 800 mm (Table S1).
In mid-May 2020, 96 bolts (diam. 6.0 ± 1.3 cm, length 30 ± 2.3 cm) were cut from branches of eight tree species (12 bolts per species), that is Acer platanoides L., C. sativa, Carpinus betulus L., Corylus avellana L., F. ornus, O. carpinifolia, Quercus robur L., and Picea abies (L.) H. Karst. ( Figure S1 and Figure S2a). All are widespread in temperate forests of Europe and are hosts for many ambrosia beetles. For each selected tree species, bolts were obtained from three different standing trees (four bolts per tree). For each bolt, a 10-cm deep hole was then made on one end using a 1.5-cm-diameter drill (hole vol. = 17.3 cm 3 ). Six bolts of each species were randomly selected to receive a 5% (v:v) aqueous ethanol solution and six to receive 90% aqueous ethanol solution. The 5% concentration was selected based on previous host selection and colonization studies on X. saxesenii and Xylosandrus spp. (Ranger et al., 2018;Rassati et al., 2020), and the 90% concentration was selected because it is highly attractive to many ambrosia beetle species (Klingeman et al., 2017). The aqueous ethanol solution was poured into the drilled hole, which was then plugged with a silicon cap ( Figure S1). Bolts were brought to the field on 22 May.
Using plastic cable ties attached to an eye bolt screwed into the wood near the silicon cap ( Figure S2a), bolts were hung approximately 2 m from the ground, a height at which ambrosia beetles are abundant (Marchioro et al., 2020;Miller et al., 2020). Bolts of the eight different tree species were hung 5 m apart in randomized complete blocks, with alternating blocks for each ethanol concentration 40 m apart ( Figure S1). We did not test the two ethanol concentrations within the same block because we wanted to avoid potential repellent effects of the highest ethanol concentration.
Bolts were checked every week until 10 July when bolts were brought back to the lab. At each check, the ethanol solutions were replenished, and entry holes bored during the previous week were circled with a permanent marker using a different colour for each week ( Figure S2b).

| Analysis of ambrosia beetle attacks and colonization
In the laboratory adult females blocking entry holes (Biedermann & Taborsky, 2011;Nuotclà et al., 2019) were carefully excised avoiding damage to the entry hole and then were identified to species.
Width of adult beetles was measured with a stage micrometer under a stereomicroscope. Pins of a thickness comparable to the width of beetles of each species were used to measure entry hole diameter. This methodology allowed us to reliably assign entry holes to X. saxesenii and X. crassiusculus. The other two colonizing species, A. dispar and the auger beetle, Scobicia chevrieri (Villa & Villa), have the same width and entry hole diameter. For the latter species, we analysed only galleries in which we found and identified an adult beetle. At the end of September, bolts were peeled using an electric planer ( Figure S2c,d) and galleries were classed as either developed or superficial ( Figure S3), indicating successful or unsuccessful attack, respectively.

| Statistical analysis
Generalized linear mixed-effects models (GLMMs) were used to analyse flight and attack activity for each species. The number of entry holes of a given species bored during each week (Poisson distribution) was the dependent variable, the week number since the starting day (and both quadratic and cubic terms) was the independent variable. The models were fitted using the 'glmer' function in the package 'lme4' (Bates et al., 2017) implemented in R (R Core Team, 2019). Overdispersion and residual distribution were checked via the 'DHARMa' package (Hartig, 2017). To determine the influence of tree species and ethanol concentration on host selection and colonization we used linear mixed-effects models (LMMs). For host selection, the number of entry holes of a given species visible on the bolt surface was used as dependent variable whereas tree species (categorical variable), ethanol concentration (categorical variable) and their interaction were independent variables. The interaction term was included in the model for X. saxesenii and X. crassiusculus but not for A. dispar and S. chevrieri due to the lack of data points for certain treatments. For colonization, the percentage of developed galleries of a given species per bolt was the dependent variable whereas tree species and ethanol concentration were independent variables. The effect of the interaction between the latter two variables was not tested due to the lack of data points for certain treatments. When necessary, the number of entry holes or the percentage of developed galleries was either log-or square root-transformed to satisfy the assumption of normality. The latter models were fitted using the 'lmer' function in the package 'lme4' (Bates et al., 2017) implemented in R (R Core Team, 2019). All post hoc pairwise comparisons among means were performed using the 'glht' function from the 'multcomp' package (Hothorn et al., 2016) with Tukey correction. In all three analyses, block was a random variable. Statistical significance was always set at the 0.05 level.

| RE SULTS
Bolts were attacked by two native ambrosia beetles (A. dispar and X. saxesenii) and one exotic ambrosia beetle (X. crassiusculus), and the auger beetle, S. chevrieri.
The temporal analysis of entry holes showed a significant nonlinear trend for A. dispar (Figure 1a), X. crassiusculus (Figure 1c), and S. chevrieri (Figure 1d) (Table S2), with a peak in mid-June for all three species. A significant non-linear trend was found also for X. saxesenii ( Figure 1b, Table S2), but the peak was less evident and was followed by a slight further increase at the end of the sampling period. For X. crassiusculus, the percentage of developed galleries per bolt was higher in June (33.4%) than in July (7.4%).

| Effect of ethanol concentration and host tree species on host selection
Host tree species and ethanol concentration significantly affected the number of A. dispar entry holes (LMMs, F = 2.33, p = 0.03 and F = 21.62, p < 0.001, respectively), with significantly more entry holes on O. carpinifolia than on P. abies, and no difference between O. carpinifolia and the other species (Figure 2a). In addition, the number of entry holes was significantly higher in bolts filled with 90% ethanol than with 5% ethanol (Figure 2b).
Host tree species and ethanol concentration also significantly affected the number of X. crassiusculus entry holes (LMMs, F = 2.38, p = 0.02 and F = 12.85, p < 0.01, respectively), with no interaction between host tree species and ethanol concentration. The number of entry holes was significantly higher on Q. robur than on P. abies, whereas no difference was observed between Q. robur and the other species ( Figure 2e). In addition, the number of entry holes was significantly higher in bolts filled with 5% ethanol than with 90% ethanol (Figure 2f).
The number of X. saxesenii entry holes was affected by host tree species (LMM, F = 11.23, p < 0.001, Figure 2c), but not ethanol concentration (LMM, F = 0.020, p = 0.88, Figure 2d) or the interaction between ethanol concentration and host tree species. The number of entry holes was significantly higher on Q. robur than on all the other species except F. ornus (Figure 2c).
Finally, the auger beetle S. chevrieri was affected by both host tree species (LMM, F = 5.32, p < 0.001, Figure 2g) and ethanol concentration (LMM, F = 49.09, p < 0.001, Figure 2h). The number of entry holes was significantly higher on A. platanoides than on C. betulus and F. ornus (Figure 2g), as well as on bolts filled with 90% ethanol than with 5% ethanol (Figure 2h).

| Effect of ethanol concentration and host tree species on colonization
Only X. crassiusculus constructed a sufficiently large number of developed galleries to justify analysing the effect of host tree species and ethanol concentration on colonization. The percentage of developed galleries per bolt was affected by host tree species (LMM, F = 3.41, p < 0.01, Figure 3a), being significantly higher on O. carpinifolia than A. platanoides, C. avellana, C. betulus, F. ornus and P. abies but similar to C. sativa and Q. robur (Figure 3a). The percentage of developed galleries was significantly higher in bolts filled with 5% ethanol than with 90% ethanol (LMM, F = 11.92, p < 0.01; Figure 3b). results demonstrate that these processes are also affected by ethanol concentration and host species, with major differences among ambrosia beetle species. Certain species seek high or low ethanol concentration emitted by their hosts, whereas others are less selective. Moreover, some species have more prominent host preference than others. Finally, for X. crassiusculus we show that tree species can affect colonization.

| D ISCUSS I ON
We found that bolts infused with two different concentrations of ethanol (5% and 90%) differentially affected host selection of the three ambrosia beetles and the auger beetle that were active under field conditions over the time span of this study. In particular, A. dispar and S. chevrieri preferentially attacked bolts infused with ethanol at the highest concentration (90%), X. crassiusculus preferentially attacked bolts with the lowest ethanol concentration (5%), whereas X. saxesenii was indifferent to ethanol concentration. All three species are known to be attracted to ethanol-emitting trees Reed et al., 2015;Tanasković et al., 2016), but our results suggest that the beetles use ethanol concentration as an indicator of host tree susceptibility, and possibly as a means to reduce interspecific competition.
The preference of A. dispar and S. chevrieri for bolts infused with 90% ethanol can be related to their habit of colonizing dying or severely stressed trees. This pattern is in line with previous studies showing that both species are strongly attracted to ethanol, with attraction increasing with ethanol concentration (Byers et al., 2020;Klimetzek et al., 1986;Schroeder & Lindelöw, 1989).
In contrast, the preference of X. crassiusculus for bolts infused with 5% ethanol suggests a preference for trees in the early stages of physiological stress. This differential host preference is in agreement with the higher ability of Ambrosiella hartigii Batra, the fungal symbiont of A. dispar, to tolerate ethanol in host tissue than Ambrosiella roeperi T.C. Harr. & McNew, the fungal symbiont of X. crassiusculus (Lehenberger et al., 2021;Ranger et al., 2018). This scenario would explain why exotic X. crassiusculus has become a major pest of stressed living trees in ornamental hardwood nurseries in the USA  and would also explain why A. dispar and X. crassiusculus do not simultaneously attack the same trees or logs despite their overlapping flight activity (Frank & F I G U R E 2 Effect of host tree species and ethanol concentration on the number of entry holes bored by A. dispar (a, b), X. saxesenii (c, d), X. crassiusculus (e, f) and the auger beetle S. chevrieri (g, h). Ap = Acer platanoides; Ca = Corylus avellana; Cb = Carpinus betulus; Cs = Castanea sativa; Fo = Fraxinus ornus; Oc = Ostrya carpinifolia; Pa = Picea abies; Qr = Quercus robur. Means with the same letter are not significantly different. Post hoc comparison was carried out using the 'glht' function with Tukey correction [Colour figure can be viewed at wileyonlinelibrary.com]  Klingeman et al., 2017;. Other ambrosia beetles such as Trypodendron spp. attacking apparently healthy trees in both urban and natural settings (Kühnholz et al., 2001) might show the same preference for trees producing and emitting low amounts of ethanol. The lack of a preference by X. saxesenii for bolts infused with 5% or 90% ethanol suggests an ability to attack trees under various stages of decline, which could explain its high abundance in ethanol-baited traps and on ethanol-infused trees or bolts irrespective of the ethanol dose (Galko et al., 2014;Kelsey et al., 2013;Oliver & Mannion, 2001;Reding et al., 2011). The ecological niche of X. saxesenii might thus differ from that of X. crassiusculus and A. dispar more in terms of seasonal flight pattern (Coyle et al., 2005) or in the portion of wood colonized (Iidzuka & Osawa, 2016) than on host tree chemistry and ethanol content.

Ranger
We also found a higher number of developed X. crassiusculus galleries in bolts filled with 5% than with 90% ethanol, reflecting the similar pattern observed for entry holes and again supporting the hypothesis that the symbiotic fungus A. roeperi is adapted to weakly stressed hosts with low ethanol concentration (Ranger et al., 2018). In fact, adult female ambrosia beetles neither oviposit nor construct developed galleries if the substrate is unsuitable for growth of their mutualistic fungi (Biedermann, 2020;Biedermann et al., 2009;Cruz et al., 2018). In further support of this hypothesis, Ranger et al. (2018) showed that growth of Ambrosiella grosmanniae McNew, C. Mayers, and T. C. Harr., the fungal symbiont of X. germanus, was enhanced when ethanol concentration in rearing media was between 0.5% and 3%. The main difference between the two fungi is that the activity of alcohol dehydrogenase is lower in A. roeperi than A. grosmanniae, and thus A. roeperi has a lower ethanol tolerance (Ranger et al., 2018). This pattern might explain why X. crassiusculus could establish developed galleries in logs infused with 5% ethanol but not 90% ethanol. The low number of developed galleries observed for A. dispar and X. saxesenii suggests that other factors, such as nutrient level, moisture level, or pH affect their colonization success (Adams et al., 2009;Maner et al., 2013;Menocal et al., 2017).  (Chen et al., 2021), the synergistic influence of benzaldehyde on X. saxesenii response to ethanol (Yang et al., 2018), and the general attraction of X. saxesenii to host produced compounds like quercivorol or α-copaene (Owens et al., 2019).
Xylosandrus crassiusculus established a higher number of developed galleries in O. carpinifolia bolts than in five other tree species.
Similarly, Mayfield et al. (2013) found that brood X. crassiusculus emerged from California bay laurel bolts was more than 20-fold higher than from camphor tree or silkbay, with no individuals emerging from bolts of five other tree species. Also, congeneric X. germanus produced more offspring in artificial media prepared with sawdust from European buckthorn compared to American beech, black walnut, and red oak (Castrillo et al., 2012). These results, along with ours, support (but do not confirm) the hypothesis that the symbiotic fungi of Xylosandrus spp. have variable capability to colonize and grow in different host tree species. In addition, developed galleries mainly originated from attacks during the first month of the sampling period (i.e., June), when bolts were still fresh. Although not important for other species (Sanguansub et al., 2012), this trend suggests that wood suitability for X. crassiusculus decreases with increasing wood oldness.
Our study provides support for the potential role of ethanol concentration and host tree species in determining the ecological niche be more prone to exchange fungal symbionts, which could lead to novel beetle-fungus associations with unpredictable impact (Carrillo et al., 2014;Rassati et al., 2019;Wingfield et al., 2016).

ACK N OWLED G EM ENTS
We thank the Regional Forest Service of the Veneto Region for pro-

CO N FLI C T O F I NTE R E S T
Declare that they have no conflict of interest.

AUTH O R CO NTR I B UTI O N S
DR, CMR, and GC conceived research and wrote the manuscript.
DR and GC conducted the experiments and statistical analyses. MF contributed material. DR and MF secured funding. All authors read, contributed to, and approved the manuscript.