For field work (collection of substrates with larvae and bioassays), two different study areas were used, one for each species (for details on the individual sites, see Online Resource 1, Tables S1 and S2). Anaglyptus mysticus was studied on the Swedish island of Öland in the Baltic Sea, Kalmar County. Several sites were used for collection of infested wood and insect trapping in the southern part of the Mittland forest (approximate center coordinates: DD 56.6800N, 16.6027E). The area consists of a diverse mixture of dense forest stands managed for production of timber and forest biofuel, interspersed with grazed woodlands, semi-natural grasslands, and agricultural fields. Hazel (C. avellana), pedunculate oak (Q. robur), European ash (Fraxinus excelsior), and elm (Ulmus spp.) dominate the tree cover. Anaglyptus mysticus occurs throughout the Mittland forest and is fairly abundant; no other area in Sweden has a higher number of recent observations of this species (see Lindhe et al. 2010; Swedish Species Observation System 2018a). The species’ comparatively high abundance in the Mittland forest may be explained by the large quantities of old hazel bushes present, because decaying hazel wood appears to be the preferred substrate of the larvae in Sweden (Ehnström and Holmer 2007). Therefore, we also focused on this tree species for collecting wood to rear out A. mysticus (see below).
Xylotrechus antilope was studied at Ecopark Hornsö (approx. center coordinates: DD 57.0120N, 16.0897E) on the Swedish mainland, Kalmar County. The Ecopark comprises a mixed coniferous–deciduous forest of ~ 9200 hectares that is one of the most important areas for rare and red-listed saproxylic insects in Northern Europe (Nilsson 2001; Anonymous 2008). The forest is predominantly managed for timber production, but several nature reserves have been established and management practices are tailored to support biodiversity throughout the area (Anonymous 2008). Small, open grasslands, rocky areas, and lakes and streams of different sizes increase the landscape’s heterogeneity. Young and middle-aged pedunculate oak trees are abundant and X. antilope is a common and widespread species within the Ecopark (Nilsson and Huggert 2001; MAM, pers. obs.), which may harbor the largest population of this species in Northern Europe (Nilsson and Huggert 2001).
Collection of beetles
Hazel wood with larvae of Anaglyptus mysticus was collected from three different locations in the Mittland forest in early December 2015. At each location, we sought to collect different types of dead hazel wood because we did not know which logs might contain developing larvae, or the preferred stage of decay. Thus, substrates in different decay stages were collected, from relatively fresh, hard wood to decaying wood that could easily be broken up with a knife. Diameters varied from ~ 3 to 12 cm. We collected standing and fallen hazel wood stems, from both sun-exposed and shady positions. In all, ~ 1 cubic meter of hazel wood was gathered. The collected wood samples represented only a small proportion of the entire amount of dead hazel wood present at each of the three surveyed sites.
The collected wood was cut into ~ 0.5-m sections and placed in transparent plastic containers (L × W × H: 56 × 39 × 42 cm). From late December, the containers were kept in a greenhouse with an average daily temperature of about 15 °C and examined twice daily for emerging beetles. Occasionally, a spray bottle was used to dampen the wood with water to reduce desiccation. The first individuals of A. mysticus emerged from the wood after about 4 weeks (in early February). Adults continued to emerge for about a week until a total of 29 individuals had emerged. The beetles were continuously removed from the boxes, separated by sex, and stored in two plastic containers with fresh pieces of hazel branches (C. avellana), in an effort to stimulate pheromone release. For nourishment, we added pieces of paper saturated with honey water, and exchanged them for new ones every second day. The beetles were sexed by comparing the length of the antennae to the length of the body (from the mouthparts to the tip of the abdomen); males of A. mysticus have antennae of a greater relative length than females (Hansen 1966). The containers were stored in a refrigerator (8 °C) before (and sometimes in between) collections of headspace odors.
Xylotrechus antilope was reared from wood substrates collected on two occasions in Ecopark Hornsö. In the early spring of 2011, freshly cut branches of pedunculate oak (Q. robur) collected from various sources in the southern Swedish counties of Skåne, Blekinge, and Kalmar were transported to Lindsudd near Barnebo Lake (Barnebo N, Online Resource 1, Table S2), held outdoors through the summer, and collected again in the autumn, after natural colonization by the insects. The wood was stored outdoors and brought into a greenhouse in early 2012. In 2015, substrates were collected from a large heap of dead twigs and branches (up to ~ 20 cm diameter) of pedunculate oak at Ecopark Hornsö (Barnebo NE, Online Resource 1, Table S2). The substrates were left-over wood residues from logging activities in an adjacent mature oak forest in the winter of 2013–2014. A large number of saproxylic insects had colonized the wood in the summer of 2014, when it had been left to dry along a sunny south-facing forest edge. Once dry, the material would eventually be shredded as biofuel, which occurred in late winter 2015. Thus, the heap of substrates formed a so-called ecological trap, as described by Hedin et al. (2008). In February 2015, before several species with short larval developmental times were about to emerge from the wood in the following summer season, we collected ~ 1.5 cubic meters of wood substrates of mixed dimensions from the top and sides of the heap, which were brought back to the SLU campus and held in a greenhouse. In both 2011 and 2015, the wood material was processed as described for A. mysticus (see above).
In both 2012 and 2015, the first individuals of X. antilope emerged after about 3 weeks in the greenhouse, and large numbers of adults continued to emerge over about 1 week. The numbers of emerging beetles were quantified only in 2015, when a total of 527 X. antilope emerged, of which a subset of about 25 individuals (selected at random from multiple containers with wood, and initially stored singly in tubes) was used for headspace sampling (see below). The roughly 25 individuals that were selected to be used for aerations (the technique used for collecting volatile compounds emitted by the insects, see below) were separated by sex and stored in two plastic containers with pieces of fresh oak branches and slips of paper soaked in honey water. Since females and males of X. antilope have nearly identical external morphology, to sex the beetles, we put two individuals (from the group of ~ 25 individuals) together in a small container and observed if the animals attempted to copulate. In most cases, when a male and female were put together, copulation was initiated immediately after the male first touched the female with his antennae. If the beetles did not make physical contact after ~ 0.5 min, or if they made physical contact but did not start to copulate, we excluded the animals and continued with a new pair of individuals. The containers with males and females (about ten individuals of each sex) were stored in a refrigerator before headspace collections (see below).
Collection of volatiles
We performed headspace sampling using the same approach and equipment for both study species. The beetles, in their containers with males and females separated, were transferred from the refrigerator to a climate-controlled chamber (25 °C constant temperature, light 8:30 AM–9 PM, 65% relative humidity) in the afternoon of the day before aerations were to commence. The next morning, 5–8 individuals of each sex were picked randomly and placed in two 1 L gas washing bottles (males and females in separate bottles) (Lenz Laborglas Gmbh, Wertheim, Germany). An identical, but empty bottle was used as a blank control. Strips of a fine metal mesh (previously rinsed with ethanol and acetone) were inserted into the bottles to increase the surface area over which the beetles were distributed to reduce the risk of aggression. The bottles were then transferred to a second climate chamber (but with the same settings), and headspace sampling was initiated.
Detailed specifications of the materials and techniques used for the headspace sampling can be found in Molander and Larsson (2018). Briefly, collectors consisted of Porapak™ Q adsorbent, enclosed in Teflon® tubing, and attached with Teflon® thread-sealing tape to one of the two top openings of each glass bottle. An identical collector was attached to the second opening of each bottle to remove volatiles from the incoming air. Polyvinyl chloride (PVC) tubing, plastic connectors, and Teflon® tape were used to connect the three collectors to a single air pump. A constant flow rate of 0.25 L min‒1 through each collector was established using a triple air flow meter and plastic valves on the PVC tubing. Two sets of glass bottles and several sets of collectors were alternated between different aerations, to avoid systematic errors. In between aerations, the bottles were rinsed with ethanol followed by acetone, and occasionally placed in an oven (400 °C) overnight.
Aerations lasted for about 5–6 h ( ~ 10 AM to 4 PM) because both species are diurnal with maximal activity in the late morning to early afternoon under field conditions (MAM, pers. obs.). In between aerations, the beetles were released back into the containers with fresh wood where they mixed with other same-sex individuals. Thus, most collections were carried out with mixtures of new and previously aerated individuals. After headspace collections had been initiated, the beetles were stored in the original climate chamber in between aerations (or in the refrigerator if aerations were not performed on consecutive days). A total of 4 headspace collection sessions were conducted with A. mysticus and 7 with X. antilope, with each session consisting of a collection from males, females, and a blank control sample.
Immediately after aeration, collectors were extracted with 300 µl of hexane, then washed with an additional 3 × 300 µl of hexane followed by 3 × 300 µl of acetone, before reuse. The extracts, in vials with closed butyl/PTFE seal screw caps (articles 11,090,210 and 08,151,653, Skandinaviska Genetec AB, Stockholm, Sweden), were kept in a freezer (− 18 °C) until analysis by gas chromatography–mass spectrometry (GC–MS; see below).
Identification of compounds
Extracts of volatiles from both species were analyzed with the same methods and GC–MS instruments. At the Alnarp campus, a 6890N model GC interfaced to a 5975 mass selective detector (Agilent Technologies, Palo Alto, CA, USA) was used, with the transfer line temperature set at 150 °C. The GC was fitted with an HP-5ms capillary column (95% polydimethylsiloxane/5% phenylmethylsiloxane, 60 m × 0.25 mm inner diam., 0.25 µm film thickness, Agilent Technologies). Injections of 2 µl of each aeration sample were made manually in splitless mode (split vent opened after 0.5 min, injector temperature 225 °C). Control injections of 2 µl of hexane were performed before and in between analyses of headspace samples. The carrier gas was helium with a constant flow rate of 1.8 ml min‒1 (inlet pressure 172 kPa). The oven program started at 30 °C, with a 3 min hold, thereafter rising by 8 °C min−1 to 260 °C, with a 10 min hold. We also analyzed the extracts on a second GC–MS (GC model 7890B and MS model 5977Agilent Technologies) equipped with a DB-WAX capillary column (polyethylene glycol, 60 m × 0.25 mm inner diam., 0.25 µm film thickness; J&W Scientific, Folsom, CA, USA). Injections, carrier gas, and settings were identical to those used with the HP-5ms column, except that the flow rate was 1.9 ml min‒1, and the final temperature was 230 °C. The mass spectrometers were set with a solvent delay of 7 min. Spectra were taken in electron impact ionization (EI) mode at 70 eV, with a scanning range of 29–400 m/z.
Calculations of the relative proportions of potential pheromone compounds (see “Results”) were based on the results from the DB-WAX column for A. mysticus because the trace components separated better on this column, and the HP-5ms column for X. antilope. Due to the temperatures used during analyses, we observed that one of the main pheromone components, 3-hydroxy-2-hexanone, was partially rearranged into 2-hydroxy-3-hexanone on both columns, with peaks partially separated on the DB-WAX column. We therefore pooled the peak areas of the two isomers from the DB-WAX chromatogram to achieve a combined estimate of the relative quantities of pheromone compounds produced by A. mysticus. When extracts were analyzed at lower temperatures on a chiral stationary phase GC column (see below), essentially a single peak corresponding to 3-hydroxy-2-hexanone was observed.
Extracts of volatiles were also analyzed at UC Riverside by GC–MS using an Agilent 7820A GC interfaced to a 5977E mass selective detector, and an autosampler. The GC was equipped with a DB-5 column (95% polydimethylsiloxane/5% phenylmethylsiloxane, 30 m × 0.25 mm inner diam., 0.25 μm film thickness; J&W Scientific). One µl injections were made in splitless mode (injector 250 °C, transfer line 280 °C, split vent opened after 0.5 min), with helium carrier gas (inlet pressure 89.6 kPa, linear velocity 34 cm/sec). The oven was programmed from 40 °C for 1 min, 10 °C min−1 to 280 °C, hold 20 min. Mass spectra were taken in EI mode at 70 eV, scanning a mass range from 40–450 m/z, with a 3 min solvent delay.
The chromatograms from extracts of males, females, and controls were compared visually with Agilent ChemStation software (version E.02.02.1431) using the overlay function. The observed sex-specific peaks were first tentatively identified by matching their mass spectra to reference spectra in the Wiley (10th edition) and NIST (NIST 14) commercial mass spectral databases where possible, and/or to reference spectra from previous work with cerambycid semiochemicals. The identities were then confirmed by comparing mass spectra and retention times to those of authentic standards.
The absolute configurations of the insect-produced compounds were determined by analyses of aliquots of the extracts at UC Riverside using a GC equipped with a chiral stationary phase Cyclodex B GC column (10.5% β-cyclodextrin in DB-1701, 30 m × 0.25 mm inner diam., 0.25 μm film thickness, J&W Scientific), with an injector temperature of 150 °C to minimize isomerization of the thermally labile hydroxyketones. Injections were made in split mode (split ratio ~ 20:1) at 172 kPa inlet pressure, with a temperature program of 50 °C/1 min, 3 °C/min to 220 °C, hold 20 min. The GC was equipped with a flame ionization detector (detector temperature 250 °C), and chromatograms were processed with Peak Simple software (SRI Instruments, Torrance, CA, USA). Authentic standards were analyzed using identical parameters, and the identification of the insect-produced enantiomer of each hydroxyketone was verified by coinjection of a blend of the insect extracts with the racemic standards, determining which of the two peaks in each racemate was enhanced.
Sources of chemicals
Racemic 3-hydroxy-2-hexanone (CAS number: 54123-75-0) was purchased from Bedoukian Research (Danbury, CT, USA). Standards of 2-nonanone (99% purity, CAS: 821-55-6), 2-heptanone (99%, CAS: 110-43-0), 2-undecanone (99%, CAS: 112-12-9) and 2-nonanol (99%, CAS: 628-99-9) were obtained from Sigma-Aldrich (Hamburg, Germany), and 2,3-hexanedione (90%, CAS: 3848-24-6) from Aldrich Chemical Co. (Milwaukee, WI, USA). Racemic 2-hydroxy-3-octanone was synthesized from racemic lactic acid methyl ester (TCI America, Portland, OR, USA), and (S)-2-hydroxy-3-octanone was synthesized from (S)-lactic acid methyl ester, using methods previously described by Hall et al. (2006). An authentic standard of 2,3-octanedione was produced by oxidation of a few mg of (S)-2-hydroxy-3-octanone with pyridinium dichromate in dichloromethane.
The biological activity of the identified compounds was examined in three separate field bioassays. For A. mysticus, we used racemic 3-hydroxy-2-hexanone and 2-nonanone because pure (R)-3-hydroxy-2-hexanone was not available when the bioassays were conducted. For X. antilope, racemic 2-hydroxy-3-octanone was used in 2015, whereas (S)-2-hydroxy-3-octanone was used in 2016. We choose to exclude 2,3-hexanedione and 2,3-octanedione from our bioassays because these compounds are likely to be artefacts from degradation of the corresponding ketols (Millar and Hanks 2017). In addition, we did not test a number of trace components (see “Results”) in the extracts of male A. mysticus.
Five treatments were used in the bioassay with A. mysticus in 2017, including racemic 3-hydroxy-2-hexanone and 2-nonanone as single components at 100 mg and 100 mg per ml in isopropanol, respectively, and three blends with variable quantities of 2-nonanone to that of the racemic hydroxyketone (20:100, 50:100, and 80:100 mg/ml respectively, in isopropanol). Thus, the resulting proportions of 2-nonanone to the insect-produced (R)-enantiomer of 3-hydroxy-2-hexanone were 40:100, 100:100, and 160:100. Polyethylene Grippie® zip-lock bags (6.5 cm × 5.5 × 40 µm, Grippie Light Nr-02, b.n.t. Scandinavia AB, Arlöv, Sweden) were used as dispensers. At the field sites, 0.5 ml of each treatment was loaded into each bag, and the bag was attached with wire to the south-facing side of the trap without puncturing the bag. A detailed description of the custom-built flight-intercept traps that were used in all bioassays can be found in Molander et al. (2019). The traps had two cross-vane panels mounted atop a funnel that emptied into a collecting jar, and a top cover for rainwater protection. The panels and the inside of the funnel were coated with Fluon® (polytetrafluoroethylene dispersion, 60 wt% in H2O, Sigma-Aldrich, St. Louis, Missouri, USA, further diluted 1:1 with water) to increase trap efficiency (see, e.g., Graham and Poland 2012). Each trap contained ~ 0.25 L propylene glycol as a preservative and killing agent.
Six replicates (sites) were used in the bioassay with A. mysticus, each consisting of one trap with each of the five treatments and a control trap (with a lure of 0.5 ml of isopropanol alone). The traps were deployed in linear transects at six sites in the Mittland forest on April 29 and positioned in glades or along sunny forest edges with large amounts of hazel in the immediate vicinity. Replicates were separated by a minimum of 300 m and posts of reinforcement steel were used to hang the traps in a row with 10 m between traps. The center of each trap, and the location of the pheromone lure, was ~ 1.5 m above ground. All replicates were checked three times (May 21, June 10, and July 2) resulting in three sampling periods which covered the main activity period of the species in southeastern Sweden (Lindhe et al. 2010).
For X. antilope, racemic 2-hydroxy-3-octanone was tested at 100 mg/ml of isopropanol (2015), whereas (S)-2-hydroxy-3-octanone was tested at 50 mg/ml (2016). Consistent with A. mysticus, Grippie® zip-lock bags were loaded with 0.5 ml of test solutions and control lures were loaded with 0.5 ml of isopropanol alone. In 2015, four replicates (sites) were deployed in Ecopark Hornsö, separated by 600 m or more. The traps were suspended from protruding oak branches at a height of ~ 1.5–2 m and the traps within each replicate were separated by ~ 10 to 20 m. Two sampling periods were used. For the first sampling period, three replicates were deployed on May 29 and the fourth replicate on May 30. Three replicates were checked on June 24 and the fourth replicate on June 25. The second sampling period ended on July 26 when the traps were emptied a second time and taken down. Late July marks the end of the species’ main activity period in Sweden (Lindhe et al. 2010). In 2016, when pure (S)-2-hydroxy-3-octanone was tested, five replicates were deployed in the Ecopark. Most sites were not the same as in 2015 and sites were separated by at least 1.1 km. In 2016, traps were hung from steel posts (see above) and spaced 10 m apart within each replicate. Only a single sampling period was used with all replicates deployed on June 5. Two replicates were checked on June 28 and three replicates on June 30.
In all three bioassays, the positions of the lures were randomized within the replicates at deployment, and each replicate for A. mysticus was re-randomized after the first and second sampling periods. For X. antilope, we switched the position of the treatment and the control after the first sampling period in 2015. Tea filters were used to separate the insects from the propylene glycol. The glycol was reused, while the filters with the insects were saved in plastic bags with details on site, date, lure, etc. recorded on each bag. After each sampling period, new pheromone lures were added to all traps by attaching a new Grippie® zip-lock bag.
Furthermore, in 2017 we undertook a large-scale landscape study (unpublished) of X. antilope and other longhorn beetles at 68 sites in southern Sweden that consisted of forests and woodlands dominated by mature oak trees (mainly Q. robur) (for locations see Fig. 5 and Online Resource 1, Table S3). At each site, we used three traps with (S)-2-hydroxy-3-octanone (50 mg/ml isopropanol) and three traps with a blend of heterospecific cerambycid pheromone compounds consisting of racemates of 3-hydroxy-2-hexanone (100 mg/ml isopropanol; from Bedoukian Research, see above), 3-hydroxy-2-decanone [20 mg/ml; synthesized as described in Allison et al. (2012)] and 3-hydroxy-2-octanone [10 mg/ml; made as described in Imrei et al. (2013)]. Traps (of the same type as used in the bioassays and hung from steel posts at all sites with polypropylene glycol as a preservative) were spaced ~ 10 m apart and lure treatments were assigned randomly. Lure solutions (0.5 ml/lure) were loaded into Grippie® zip-lock bags as above. All traps were deployed during the peak of the activity period of X. antilope (7–12 June), and emptied once (28 June–3 July) when trapping was terminated. The specimens of X. antilope that were captured in this study were used to compare the trapping efficiency of the species’ own pheromone to that of the blend of heterospecific pheromone compounds.
The numbers of A. mysticus and X. antilope per sample were counted and the sex ratio determined in a subset of samples (see “Results”) from the bioassays. The key by Ehnström and Holmer (2007) was used to identify specimens. For A. mysticus, the sex was mainly determined by the relative length of the antennae (see above), but ~ 15 specimens were also dissected to confirm that the character was reliable. For X. antilope, every beetle in the subset was dissected and the genitalia were studied to confirm the sex. The trapped beetles are preserved in 70% ethanol and voucher specimens will be deposited in the public Lund entomological collections (Biological Museum, Lund University, Sweden).
To analyze trap catches of A. mysticus, the effects of the five different lure treatments were compared to the control in a Generalized Linear Mixed Model (GLMM) using a Poisson distribution and a loge link function. The model compared how the number of captured A. mysticus (response variable) depended on lure treatment (fixed factor) and was set up with replicate (six sites) and sampling period (three trap counts) as random effects to control for dependencies. The model was checked for over- and underdispersion using Pearson residuals. The resulting dispersion statistic (φ = 1.24) was deemed acceptable (cf. Hilbe 2011), with no need for model corrections. The different treatments and the control were then compared by multiple pairwise post hoc tests (Tukey’s HSD test), using least squares means with the package emmeans (Lenth 2018). Flight activity of A. mysticus was low during the first sampling period (early May) and four replicates where no beetles were captured during this sampling period were omitted from the analyses (and Fig. 3).
GLMMs with Poisson distributions were also used to compare the treatment with the control for each of the two bioassays with X. antilope. A loge link function was used for the 2015 dataset, but exchanged for a square root link function in the 2016 dataset, because no beetles were caught in any control traps. For the 2015 dataset, replicate (four sites) and sampling period (two trap counts) were used as random effects, while only replicate (five sites) was used for the 2016 dataset. Dispersion statistics were calculated for both models using Pearson residuals and deemed negligible (φ = 1.05 for 2015, φ = 1.17 for 2016). Finally, a GLMM with Poisson distribution and loge link function was used to compare captures of X. antilope baited with the species’ own pheromone versus traps baited with the blend of heterospecific cerambycid compounds in the 2017 landscape study, with site as a random effect. Pearson residuals were used to calculate the models’ dispersion statistic, which was satisfactory (φ = 1.23).
Statistical significance was defined as probability values (P), or adjusted probability values (Pa) for post hoc tests, of less than 0.05. All statistical analyses were performed in R version 3.5.1 for Windows (R Core Team 2018) with the lme4 package for GLMMs (Bates et al. 2015).