Two sites in southern Moravia, Czech Republic were chosen for field observation. Massive outbreaks of geometrids Operophtera brumata and Erannis defoliaria have been repeatedly observed in the past in forest Zeletice (ZL; 350 m a. s. l.; 49.02°N, 17.02°E). Second observed site is situated in Zdanicky forest, close to established point called Cerveny Kriz (CL; 350 m a. s. l.; 49.09°N, 17.02°E), where no records of massive outbreaks in the past was observed, but it was known that geometrids are present. Both sites are forests composed mainly of Quercus robur, Q. petraea, and Carpinus betulus, with sporadic occurrence of Fagus sylvatica.
Moth availability
Fifty trees per site were selected randomly, and grease bands were installed on tree trunks approximately one meter above the ground. This sampling took place from the 25th of October to the 1st of December 2017. The grease band surface was cleaned, and the number, species and sex of moths were observed once a week. Results from this sampling was used to establish period “before peak” and “peak of abundance”. During the abundance peak, one independent sampling took place on both sites, where a UV lamp lure was used.
Moths sampled from grease bands were difficult to identify to the species level when their morphology was too similar. Therefore, we separated two complexes of moth species. Erannis defoliaria complex refers to E. defoliaria, Alsophila aceraria and Agriopis marginaria (abbreviated as EDC). Operophtera brumata complex refers to O. brumata and O. fagata (abbreviated as OBC). All the other moths were categorised as OTHER, and it consisted mostly of autumn noctuoids. All moths were determined according to the key of Macek et al. (2012).
Flight activity of bats
Acoustic activity of bat species foraging moths (Andreas et al. 2012) was monitored on both sites between 25th October and 25th November 2017, using SM4BAT detectors (Wildlife Acoustics, Inc., Maynard, USA) equipped with an SMX-II microphone was set to record from sunset to sunrise. The detectors were attached to a tree trunk 6 m above the surrounding vegetation so that the branches did not block the microphones. The detectors were options set same as in Blažek et al. (2021). Recordings were analysed using the Sonochiro software package (version 4, Biotope, Méze, France), we applied the northern temperate recordings library (classifier) edited by recordings from the Czech Republic. Software settings, post-processing and data analysis followed Bartonička et al. 2019. Bat calls were grouped as aerial-hawkers (Barbastella barbastellus and Rhinolophus hipposideros), gleaners (bats of the genus Myotis, mainly M. nattereri and M. bechsteinii) and opportunists in foraging strategy (Plecotus auritus and P. austriacus). Acoustic activity of other bat species was not considered.
Bat faecal sampling
Bats were mist-netted during November of 2017. In total we sampled 40 bats grouped according to the foraging strategies, i.e., 12 Barbastella barbastellus, 12 Rhinolophus hipposideros (grouped as aerial-hawkers; n=24), 8 Myotis nattereri, 1 Myotis bechsteinii (grouped as gleaners; n=9) and 4 Plecotus auritus, 3 Plecotus austriacus (as opportunists; n=7). Bats were sampled according to Blažek et al. (2021) as species that regularly forage geometrids. Each dropping sample then underwent morphological analysis (magnification ranging from 1.6x to 5.6x) where the percentage occurrence of identified taxa (%oc) was estimated following McAney et al. 1991. For every twenty samples, we added one negative control sample (two in total) to test possible contamination. After morphological analysis, all 40 samples (including negative controls) underwent DNA isolation using the same workflow as in Blažek et al. 2021. Isolation kit NucleoSpin DNA Stool kit (Macherey-Nagel GmbH & Co. KG, Düren, Germany), PCR amplification using Zeale primers (Zeale et al. 2011) that were MID-tagged for pooling, visualisation of PCR products on 1% agarose gel, purification using standard EXOCIP purification, and DNA pooling (concentration about 11 ng/ul). Sequencing was processed using the MiSeq Reagent Kit version 3 system (Illumina) by the SEQME company (Dobříš, Czech Republic; www.seqme.eu). Negative control samples showed no signs of amplification and thus were not added to the sequenced pool.
Bioinformatics analysis
Bioinformatic pipeline followed Blažek et al. (2021). Merging of paired-end 250bp reads and the quality check was done using USEARCH v10 (https://www.drive5.com/) software package, using the algorithms -fastq_mergepairs, -fastq_trunctail 30 and -fastq_minovlen 50. Python program Cutadapt v1.15 (https://cutadapt.readthedocs.io/) was used to remove tagged primers, to demultiplex samples according to the tags and to discard all reads shorter than 150 bp and longer than 165 bp (Martin 2011). We dereplicated reads using the USEARCH fastx_uniques algorithm, using the option minuniquesize 2 and applied the USEARCH UNOISE3 algorithm to cluster these unique reads into zOTUs and automatically filter chimaeras. Further, the reads were mapped back to the original samples using the USEARCH out tab algorithm.
We compared our data with the GENBANK database to match zOTUs with appropriate taxa at the highest level possible, using a minimum sequence similarity of 98% (Benson etal. 2007). When several prey species provided an equal best match, selection was based on the species known to be present in the study area.