Main text

Carnivore diet composition is often inferred by morphological analysis of prey remains in scats, but metabarcoding as a genetic alternative is used with increasing frequency (Monterroso et al. 2019; Shi et al. 2021). For animals in general, the standard metabarcoding marker is a region of the COI gene while specialized vertebrate markers target mitochondrial 12S ribosomal RNA. Currently available markers were developed to meet the requirements of different DNA sources but their performance regarding mammalian prey species identification based on scats has usually not been assessed (Ando et al. 2020; Kocher et al. 2017).

To address this issue, we compared the performance of three different metabarcoding markers (Table 1) in wolf prey item identification from scat samples and compared them with the results obtained by morphological prey item analysis. The very short 12S-V5 (12S, ca. 100 bp, Riaz et al. 2011) was specifically developed for use with highly degraded fecal samples. In comparison, the longer Mam12S-340 (12S, ca. 340 bp, Kocher et al. 2017) and mlCOIint-F/jgHCO2198-R (CO1, ca. 310 bp, Leray et al. 2013; Geller et al. 2013) offer higher resolution to identify closely related vertebrate species (Mam12S-340) and a wider phylogenetic range, covering all animals as well as fungi and bacteria (mlCOIint-F/jgHCO2198-R) but have not been tested with fecal samples so far.

We started with 24 wolf fecal samples collected in the Glücksburger Heide, Saxonia-Anhalt, Germany. Detailed descriptions of study site, collection method, and morphological scat analysis are given in Lippitsch et al. (2024). Prior to morphological analysis, DNA samples were taken from both ends of each scat. DNA was extracted using the QIAamp Fast DNA Stool Mini Kit (QIAGEN, Hilden, Germany).

For all samples, we tested successful PCR amplification using agarose gel electrophoresis, eight samples that showed successful amplification were selected for metabarcoding. Wolf-specific blocking primers do not increase the number of prey items detected (Shi et al. 2021) and were therefore not used.

Sequencing was done on the Illumina MiSeq platform by LGC (Berlin, Germany). All samples and markers were sequenced individually. Read processing and amplicon filtering were done using Vsearch (Rognes et al. 2016). We used the following filtering commands and parameters: quality filtering (--fastq_maxee 1), denoising (--minsize 4 --unoise_alpha 2), indel filtering (--fastq_minlen [10% smaller than anticipated fragment length] --fastq_maxlen [10% larger than anticipated fragment length]), chimera filtering (--uchime3_denovo --nonchimeras). To account for possible contamination, sequences with a read count below 0.05% of the total read count of the respective sample were suppressed. Taxonomic assignment was done by using blastn (Johnson et al. 2008) against the NCBI GenBank database, applying a 0.99% of identity threshold.

Prey items identified within each sample are shown in Table 2. Despite successful amplification of wolf DNA, five samples failed to retrieve any prey items for Mam12S-340, and two for COI. For 12S-V5, prey item detection succeeded in all samples.

While the identity of detected prey species was comparable between markers, the number of identified prey items and reads per item varied strongly (Table 2). Morphological and genetic results were comparable and corresponded well with those from an exhaustive dietary study in the study area (Lippitsch et al. 2024). There were, however, some notable differences. In sample 5, the metabarcoding was able to identify the species of an item that could only be identified to family level morphologically. In samples 6 and 7, morphological and genetic results seem to differ. The morphological analysis identified only Capreolus capreolus while all metabarcodes unambiguously identified Cervus elaphus as the main prey item. In both cases, however, the genetic analysis picked up traces of C. capreolus. As material for the genetic analysis was taken from the ends of the scat, it is plausible that the metabarcoding detected the content of a different meal than the morphological analysis which studied the entire scat. The occurrence of fox (Vulpes vulpes) in two samples can likely be explained by the fact that foxes mark wolf scat, thereby contaminating the sample. While foxes occasionally form part of wolf diet (Nowak et al. 2011; Wagner et al. 2012), no macroscopic remains were identified in the respective samples. The identification of a vole in sample 6 reinforces the sensitivity of the marker.

Our study demonstrates differences in the performance of these three markers that would strongly affect the success of studies aimed at identifying prey items based on highly degraded DNA from scats. Very short metabarcodes like 12S-V5 seem to be the only viable option for this difficult material. The longer Mam12S-340 and COI metabarcodes were developed for the analysis of gut content which, unlike scats, was not subjected to environmental effects. If the broader phylogenetic range of the COI region is needed, Kocher et al. (2017) designed a short COI-metabarcode. To identify plant food items, which formed an important component in sample 6, a short metabarcode based on the TrnL-region developed by Taberlet et al. (2007) is promising (Boukhdoud et al. 2021). For studies focusing on vertebrates, 12S-V5 has proven to be a robust and sensitive marker. It offers an alternative to the morphological analysis of diet composition in carnivores and can expand the range of identifiable prey species in regard to smaller species.

Table 1 Details of the metabarcoding markers used in this study
Full size table
Table 2 Food items (and read counts) identified by morphological analysis and metabarcoding
Full size table