Understanding the full diet of natural enemies is necessary for evaluating their role as biocontrol agents, because many enemy species do not only feed on pests but also on other natural enemies. Such intraguild predation can compromise pest control if the consumed enemies are actually better for pest control than their predators. In this study, we used gut metabarcoding to quantify diets of all common arachnid species in Swedish and Spanish apple orchards. For this purpose, we designed new primers that reduce amplification of arachnid predators while retaining high amplification of all prey group. Results suggest that most arachnids consume a large range of putative pest species on apple but also a high proportion of other natural enemies, where the latter constitute almost a third of all prey sequences. Intraguild predation also varied between regions, with a larger content of heteropteran bugs in arachnid guts from Spanish orchards. There was also a tendency for cursorial spiders to have more intraguild prey in the gut than web spiders. Two groups that may be overlooked as important biocontrol agents in apple orchards seem to be theridiid web spiders and opilionids, where the latter had several small-bodied pest species in the gut. These results thus provide important guidance for what arachnid groups that should be targets of management actions, even though additional information is needed to quantify all direct and indirect interactions occurring in the complex arthropod food webs in fruit orchards.

The data were collected from 10 orchards in Sweden and Spain, respectively, during a short period one month after peak flowering. Arachnids used for gut content analyses were collected individually by hand-picking from apple leaves (Sweden) or by beat sampling (Spain), and directly transferred to individual tubes with 95% ethanol to reduce contamination risk. Collected individuals were identified to the lowest taxonomic level possible, to species for adults and to genus or family for most juveniles. Fro metabarcoding of gut contents, we sequenced three sections, depending on arachnid taxon, within the Folmer region of COI  using parallel high throughput sequencing on the Illumina MiSeq3 platform at SciLifeLab in Stockholm. We used two new primers designed to reduce amplification of spiders (NoAranR) and opilionids (NoOpiF) respectively, and one published primer (NoSpi2), combined with general forward or reverse primers. We cleaned and dereplicated sequence data using OBITOOLS 1.2.9 using default settings. Following identification, we removed contaminations, plant and fungal sequences, and sequences that resulted in no identification (threshold = 97%). We also removed all sequences belonging to the same genus as the consumer.