DNA metabarcoding has developed into a routine tool for ecosystem monitoring in the past years. Recent work suggests that this powerful approach even enables community level analysis of intraspecific genetic diversity, revolutionizing our understanding of biodiversity assembly (Andújar et al., 2021). However, despite its promises, metabarcoding is subject to biases that can substantially distort estimates of diversity. Examples include external contamination of specimens, secondary predation, priming and polymerase biases, copy number variation, PCR and sequencing errors and the formation of chimeric sequences. Many of these biases can be experimentally (Kennedy et al., 2020;Krehenwinkel et al., 2017) or bioinformatically controlled (Edgar, 2016). Yet another problem that is particularly hard to control for are numts. Numts are ubiquitous components of eukaryote genomes and a long-recognized problem in phylogenetics and population genetics (Bensasson et al., 2001), but pose a particular challenge to DNA barcoding (Song et al., 2008).When a pseudogene copy is amplified it can result in incorrect inference of the evolutionary history of the lineage in question. Numts have accumulated in some lineages more than others and comparative genomic analyses are just beginning to highlight the breadth of the problem (Hazkani-Covo et al., 2010). Numt content is often evolutionarily labile (Figure 1) and can evolve quickly even within genera or species, making it particularly hard to identify them. Most evolutionary old numts can be identified by accumulated frameshift mutations and stop codons. However, the high similarity of recently evolved numts or old numts located in highly conserved DNA sequences to true mitochondrial variants makes it difficult to score intraspecific diversity. Numts usually show lower copy numbers than true mitochondrial sequences. But sequence abundance thresholds