Comparison of morphological identification and DNA metabarcoding for dietary analysis of faeces from a subtropical lizard

Context The common methods for detecting prey in faeces are traditional morphological examination of digested prey remains and faecal DNA (fDNA) metabarcoding. Although recent studies have shown that fDNA metabarcoding can identify prey that is excreted without morphologically identifiable remains, it is unclear whether fDNA metabarcoding provides a more comprehensive assessment of diet than morphological analyses. Aims We aimed to compare prey detection abilities of morphological and fDNA-metabarcoding assays by evaluating similarities and differences of estimated prey communities. We also aimed to provide the first comprehensive prey composition of an endangered ground-dwelling gecko, Goniurosaurus kuroiwae, which will contribute to the conservation management of this species. Methods We identified the prey consumed by G. kuroiwae by morphological examination and fDNA metabarcoding using the same series of faecal samples. For these two methods, we compared the overall community composition of prey items and the detection rate for each prey type. We also tested for environmental and ontogenetic variation of the composition of prey items consumed. Key results When comparing the two methods there were significant differences in the composition of prey communities that were identified from the same series of faecal samples. The fDNA metabarcoding analyses showed the higher rate of detection of earthworms, amphipods and coleopteran and lepidopteran larvae, which lack hard exoskeleton or have small body sizes. The morphological assay allowed identification of a few prey types, including spiders and earwigs, more frequently than fDNA metabarcoding. Combined use of both methods revealed that G. kuroiwae consumed large-sized invertebrates with a slight ontogenetic shift from small-sized invertebrates to large-sized ones. Conclusions Although fDNA metabarcoding showed the higher overall prey detection rate, it did not include all the prey items detected by morphological analysis. The complementary use of both methods can yield better understanding of the prey composition of invertebrate-consuming animals represented by reptiles and amphibians. Implications We have shown that the most comprehensive reconstruction of prey composition from faeces will be achieved by combining results from both fDNA metabarcoding and morphological analyses.