Primer biases in the molecular assessment of diet in multiple insectivorous mammals

Our understanding of trophic interactions of small insectivorous mammals has been drastically improved with the advent of DNA metabarcoding. The technique has continued to be optimised over the years, with primer choice repeatedly being a vital factor for dietary inferences. However, the majority of dietary studies examining the effect of primer choice often rely on in silico analyses or comparing between species that occupy an identical niche type. Here we apply DNA metabarcoding to empirically compare the prey detection capabilities of two widely used primer sets when assessing the diets of a flying (lesser horseshoe bat; Rhinolophus hipposideros) and two ground dwelling insectivores (greater white-toothed shrew; Crocidura russula and pygmy shrew; Sorex minutus). Although R. hipposideros primarily rely on two prey orders (Lepidoptera and Diptera), the unique taxa detected by each primer shows that a combination of primers may be the best approach to fully describe bat trophic ecology. However, random forest classifier analysis suggest that one highly degenerate primer set detected the majority of both shrews' diet despite higher levels of host amplification. The wide range of prey consumed by ground-dwelling insectivores can therefore be accurately documented from using a single broad-range primer set, which can decrease cost and labour. The results presented here show that dietary inferences will differ depending on the primer or primer combination used for insectivores occupying different niches (i.e. hunting in the air or ground) and demonstrate the importance of performing empirical pilot studies for novel study systems.