Interrogating 1000 Insect Genomes for NUMTs: A Risk Assessment for Species Scans

Abstract

The nuclear genomes of most animal species include segments of the mitogenome, but the count of these NUMTs varies greatly. This study examines the incidence of NUMTs derived from a 658 bp region of the cytochrome c oxidase I (COI) gene as a proxy for other coding regions of the mitochondrial genome. Analysis focuses on the most diverse group of terrestrial organisms, insects, because COI-based identification systems play a key role in clarifying their diversity, an essential antecedent to genome sequencing. Nearly 10,000 COI NUMTs ≥ 100 bp were detected in the genomes of 1,002 insect species with a range from 0–443. NUMT counts were similar among congeners, but differences among genera in a family were often large with genome size explaining 56% of the mitogenome-wide variation in counts. While many of these NUMTs possessed an indel or premature stop codon allowing their exclusion, the others could complicate species diagnosis as they averaged 10.1% divergence from their mitochondrial homologue. The count of NUMTs varies widely among insect lineages, peaking in groups that employ direct development or incomplete metamorphosis. They can raise the apparent species count by up to 22% when the 658 bp barcode region is examined while shorter targets (300 bp, 150 bp) elevate exposure (58–111%) to “ghost” species. As a result, NUMTs represent a particular complication for protocols (e.g., eDNA, metabarcoding) which employ short amplicons for biodiversity assessments.

Metabarcoding reveals seasonal and spatial patterns of arthropod community assemblages in two contrasting habitats: Desert and oasis of the Baja California Peninsula, Mexico

“…NUMTs have been recognized as a critical issue in phylogenetics (Bensasson et al,, 2001), but they are particularly harmful to DNA barcoding and metabarcoding studies (Graham et al, 2021; Song et al, 2008; Sorenson & Quinn, 1998). Our results showed that even after strict data quality filtering (sequences with minimum coverage of less than 10, stop codons or InDels were removed), a high number of putative NUMTs may be detected (Hebert et al, 2022). Even considering that we obtained a comparatively large COI fragment (658 bp) has helped to decrease the number of NUMTs substantially (Hebert et al, 2022; Kaya & Çıplak, 2018).…”

“…Our results showed that even after strict data quality filtering (sequences with minimum coverage of less than 10, stop codons or InDels were removed), a high number of putative NUMTs may be detected (Hebert et al, 2022). Even considering that we obtained a comparatively large COI fragment (658 bp) has helped to decrease the number of NUMTs substantially (Hebert et al, 2022; Kaya & Çıplak, 2018). The species in order Orthoptera are well known to have large nuclear genomes (Hanrahan & Johnston, 2011), and they are also reported to have an extremely high number of nuclear mitochondrial DNA copies (Kaya & Çıplak, 2018; Song et al, 2008, 2014).…”

“…In contrast, in other groups, like Coleoptera or Araneae, the number of MOTUs never exceeded the number of pooled individuals (Figure S1a,b). Since it was impossible to distinguish authentic sequences from putative errors (Hebert et al, 2022; Song et al, 2014), we decided to conduct all further analyses, excluding Orthoptera and Zygentoma orders (only for the genetic data set). After the deletion of these taxa, the number of MOTUs dropped from 1034 to 667.…”

“…Several reasons exist for these results, such as sequence errors, chimeric sequences or poor‐quality reads (Piper et al, 2019). Fortunately, thorough data filtering can reduce their effect to a minimum (Hebert et al, 2022). Wolbachia DNA is also commonly co‐amplified from insect specimens, but removing all nonarthropod sequences can help to overcome parasite contamination.…”

“…Several reasons exist for these results, such as sequence errors, chimeric sequences or poor-quality reads (Piper et al, 2019). Fortunately, thorough data filtering can reduce their effect to a minimum (Hebert et al, 2022).…”