This study covered all the region of Extremadura (SW Spain), which extends over 41,634 km² and contains more than 1.2 million hectares of oak forests, most of them open woodlands. The most important oak species are, in order of importance, holm oak (Quercus ilex L.), cork oak (Quercus suber L.) and Pyrenean oak (Quercus pyrenaica Willd.). The climate is typically Mediterranean, with dry and warm summers (up to 40 °C) and mild-rainy winters. The mean annual temperature is close to 16 ºC (range 10-18 ºC) and the mean annual precipitation is about 700 mm (range 360-1500 mm). Altitude ranges from about 150 m (Guadiana and Tajo River valleys) to more than 2400 m (Central System Mountain range), although the oak species rarely populate areas above 1100-1200 m (holm and cork oaks) or 1500-1600 m (Pyrenean oak) (Torres-Vila et al., 2019, 2023a).

During the spring (April-June) of two consecutive years (2021-2022), a trapping network with synthetic sex pheromones was implemented throughout the Extremadura region (Fig. 2). This main sampling protocol was complemented with additional data obtained during the years 2020 and 2023. In total, information was collected from 38 sites, some of which were sampled for more than a year (42 site/year combinations) (Supplementary Table S1). Sampling sites were established mainly in reforestations and densifications of holm and cork oak, but also of Pyrenean oak. In each selected sampling site, standard delta traps were installed and baited with commercial sex pheromones (OpenNatur^®, manufacturer: Pherobank BV, The Netherlands) of each of the six oak-living sesid species reported in Iberia (De Freina & Witt, 1997; Laštůvka & Laštůvka, 2001), namely Paranthrene insolita Le Cerf, 1914 [= P. insolitus]; Synanthedon stomoxiformis (Hübner, 1790); Synanthedon conopiformis (Esper, [1782]); Synanthedon codeti (Oberthür, 1881); Synanthedon vespiformis (L., 1761); and Synanthedon spuleri (Fuchs, 1908). We included S. stomoxiformis among the target species because it was mentioned in oaks by De Freina & Witt (1997) based on early observations by Staudinger (1879: 303): “I suspect that Ses.[ia] stomoxyformis [sic] will live in oak trees, as I know it has been caught on oak trees more often”. However, the available evidence suggests that it is rather unlikely that this clear-wing moth uses oaks as host trees. The six studied species have a western Palearctic distribution (Laštůvka & Laštůvka, 2001).

In several sites and years (n = 7 combinations) moth flight was monitored with pheromone traps throughout the entire favourable season (April-November) to build the flight curves of the target species (site/year codes: 0, 22S, 31, 32, 33, 43S and 46F; Suppl. Table S1). Pheromone diffusers were replaced every 30-40 days following the manufacturer's instructions. Pheromone traps were inspected weekly, changing each sticky plate for a new one and taking the sticky plates with the catches to the laboratory for examination. The species membership of all captured adults was verified based on their external morphology (wing, thoracic and abdominal coloration patterns) under a stereomicroscope. In case of doubt or poor preservation, the adults were removed from the sticky plates to study their genitalia, diluting the glue with a drop of hexane. The taxonomy was performed according to the keys of Laštůvka & Laštůvka (2001). From the catch data (Suppl. Table S1), both abundance (mean catch number per trap per site per season in the subset of traps with at least one capture) and occupancy (percentage of traps with at least one capture) were estimated (Gaston et al., 2000; Torres-Vila et al., 2022).

The observation of numerous specimens of S. vespiformis on the sticky plates suggested that adults captured in spring were larger than those captured in other seasons of the year, so we studied the effect of season and year on moth size. To do this, we measured the imaged males (see phototraps below) in Montánchez during 2022-2023 (site/year codes: 44F and 46F; Suppl. Table S1) using the forewing length (base-apex) as an estimator of body size. Only imaged males with at least one wing in horizontal position were measured. The sticky plate grid was used as a scale. Sample sizes ranged 14-71 individuals per season per year (total sample size n = 192). Wing measurements were done with EclipseNet 1.20 software using a Nikon DS-U1 digital camera mounted on a Leica S6D stereomicroscope.

The diel activity of the most abundant sesid species (P. insolita, S. conopiformis and S. vespiformis) was investigated with automated traps baited with sex pheromones placed in two sites for two years (site/year codes: 22S, 43S, 44F and 46F; Suppl. Table S1). We used two types of automated traps that allowed us to know the time of adult capture by different methods.

First, we used a phototrap adapted from the PT2 trap model previously used with Cerambyx longhorn beetles (Torres-Vila et al., 2023b). The phototrap is formed by an image capture device (ICD) placed on the top of a standard delta trap sticky plate baited with the sex pheromone of the target species (Fig. 3A, 3B). The ICD consists of 1) a central processing unit (Raspberry Pi mod. A+) managed with Raspberry Pi OS v. 1.4 (formerly Raspbian); 2) a visible light camera (Raspberry Pi Camera Module v. 2.0); 3) a quartz clock installed in the Raspberry Pi GPIOs; and 4) a micro-SD port with a 32 Gb memory card for the operating system and image storage. ICD electronic components were arranged inside a watertight box (11 x 11 x 6 cm) and powered by portable solar panels. The device was equipped with four light-emitting diodes (LEDs) facing the image capture area (the entire sticky plate) to enable imaging at twilight and night. The LEDs turned on 5 seconds before each shot, enabling the digital camera to set optimal sensitivity and exposure time, and they were shut off once the shot was taken. The ICD script was set to take an image every 10 minutes. The phototrap setting was finished by placing some wooden skewers or a piece of wire mesh on the ICD top to keep birds from perching. A curved acetate sheet placed on top of the sticky plate greatly increased trap effectiveness since moths entered easily through the lateral apertures, but they crashed while in flight into the inner acetate wall to fall on the glue (Fig. 3B). Images were viewed using Irfan View^® free software, and image metadata (date, hour, site) were recorded for each male moth captured. Images allowed a correct specific identification in nearly all cases (> 99%), but in cases of doubt, the species membership of the imaged adults was verified by examining the adults glued to the sticky plates brought from the field.

Second, we used an infrared-sensor trap (Montero-Calvo, 2021) formed by an infrared light-emitting device enclosed in a standard funnel trap (covered with a small plastic roof to prevent rainwater from entering) baited with the sex pheromone of the target species (Fig. 3C). The device consists of a pass sensor with five infrared emitters and paired receivers sensitive to the 940 nm wavelength, a NSL-19m51 brightness sensor, a Heltec WiFi LoRa 32 v2.1 board with an ESP32 microprocessor and a LoRa transceiver (Fig. 3D) managed with a custom firmware. The components were powered by two lithium-ion batteries integrated into the mother board. The infrared barrier is always active, but when an insect passes through to enter the trap, the barrier deactivates and wakes up the microprocessor, which sends a capture event signal to a server via a LoRaWAN network. Once the server confirms the reception, the device reactivates the infrared barrier and goes back to sleep. The species membership of each capture event was verified weekly by examining the adults captured in the funnel trap container.

The automated traps were set, depending on trap type and site features, either on metal supports nailed to the ground, suspended from an oak branch, or fixed to the oak trunks at about 1.40-1.50 m height, as vertically as possible. Automated traps were usually inspected once a week to recover data and do maintenance operations. The number of catches in each trap was scored into 24 one-hour periods, and frequencies per hourly intervals were calculated prior to statistical analysis. Data from different sites, years and traps were pooled within species to build more robust curves of diel activity (site/year codes: 22S, 43S, 44F and 46F; Table S1).

The two-sample Kolmogorov-Smirnov (KS) test was used to compare frequency distributions of diel activity (hourly captures) between species (Sokal & Rohlf, 1995). Since S. vespiformis moth size was not normally distributed after a preliminary Shapiro-Wilk test, a generalised linear model (GLM) using the Poisson distribution with log link was computed to assess the effects on moth size of season and year (both with fixed effects) as explanatory variables. Season factor was scored in three classes (spring: April-June; summer: July-September; autumn: October-November) and year factor in two classes (2022 and 2023). Model error was tested for over- and under-dispersion, and the residual plot was checked to verify the assumptions of the fitted model. Analysis of deviance (Type II Wilks’ likelihood-ratio test [LR Chi²]) was used to assess the effects of season, year and their interaction on moth size. A post hoc Tukey contrast (|z| values) was completed for multiple comparisons of means to establish homogeneous groups among seasons. All statistics were computed with R 4.3.2 (R Core Team, 2023).

Four of the six target species, P. insolita, S. vespiformis, S. codeti and S. conopiformis were well distributed throughout the study region, occurring in the three sampled forest habitats (holm, cork, and Pyrenean oak) and occupying a considerable altitudinal range (Fig. 2, Suppl. Table S1). On the contrary, S. stomoxiformis and S. spuleri were never captured. The total number of catches per species, as well as the species-specific values of abundance and occupancy, differed noticeably among species (Fig. 4). S. conopiformis was the most abundant species, but the moth with the highest occupancy was P. insolita, despite its relatively low abundance (Fig. 4). Note, however, that S. vespiformis autumn flight (see below) is not included in these statistics since trapping protocol only covered the spring (April-June). It follows that S. vespiformis abundance values are underestimated if the entire year is considered rather than just the spring period.

The commercial sex pheromones used showed a good level of specificity. Even so, there was cross-attraction with three non-target sesids (not associated with oaks) that were captured in the traps (Suppl. Table S1, Fig. 4), in order of abundance: Bembecia uroceriformis (Treitschke, 1834), Pyropteron chrysidiformis (Esper, 1782) and Pyropteron (Synansphecia) leucomelaena (Zeller, 1847). The occupancy of these three non-target species did not exceed 20% in any case (Fig. 4). It is important to note that the S. conopiformis sex pheromone was extremely attractive to B. uroceriformis; in fact, all B. uroceriformis specimens were captured in traps baited with S. conopiformis pheromone (see Pühringer, 1996).

The flight curves showed that P. insolita (Fig. 5A) and S. conopiformis (Fig. 5B) exhibit spring activity (April-June), with flight peak dates that may vary up to almost a month depending on the site and year. The data available for S. codeti were more modest but clearly indicate that its flight period is longer and later (May-October) than in the two preceding species, with a marked summer peak (July) (Fig. 5C). In the case of S. vespiformis, adult activity covered the longest seasonal period (early April to early November), with two flight peaks in May and September-October (Fig. 5D).

Moth size in S. vespiformis was not normally distributed (Shapiro-Wilk test, W = 0.98, P < 0.01). Moth size was significantly affected by the season (Analysis of deviance, LR Chi² = 16.61, df = 2, P < 0.001) but not by the year (LR Chi² = 0.51, df = 1, P = 0.48), with no season × year interaction (LR Chi² = 0.06, df = 2, P = 0.97) (Fig. 6). So, we pooled both years and re-analysed the data with the season as the only explanatory variable (LR Chi² = 16.40, df = 2, P < 0.001). A post hoc of this model showed that S. vespiformis moths were significantly larger in spring than in either summer (|z| = 3.73, P < 0.001) or autumn (|z| = 3.01, P < 0.01), with no significant differences between the summer and autumn seasons (|z| = 0.05, P = 0.99) (Fig. 6).

Diel activity was notably different between the three target species for which data were available: P. insolita, S. vespiformis and S. conopiformis (Fig. 7). Although the three species were active from early afternoon until late evening (from about 14:00 to 22:30 h), the earliest flight peak was that of P. insolita (15:00-18:00 h), followed by S. vespiformis (17:00-19:00 h), and then by S. conopiformis (19:00-21:00 h) (Fig. 7). Diel activity was significantly earlier in S. vespiformis than in S. conopiformis (KS test, D-stat = 0.642 > D-crit = 0.110, P < 0.001). KS tests were not computed with P. insolita due to the low sample size recorded for this species in automated traps. Distinctly enough, the diel activity of B. uroceriformis (a non-target sesid species) took place in the morning, with a relatively early flight peak (9:00-11:00 h) (Fig. 7) (all local time, GMT+2).