A Custom Regional DNA Barcode Reference Library for Lichen-Forming Fungi of the Intermountain West, USA, Increases Successful Specimen Identification

Abstract

DNA barcoding approaches provide powerful tools for characterizing fungal diversity. However, DNA barcoding is limited by poor representation of species-level diversity in fungal sequence databases. Can the development of custom, regionally focused DNA reference libraries improve species-level identification rates for lichen-forming fungi? To explore this question, we created a regional ITS database for lichen-forming fungi (LFF) in the Intermountain West of the United States. The custom database comprised over 4800 sequences and represented over 600 formally described and provisional species. Lichen communities were sampled at 11 sites throughout the Intermountain West, and LFF diversity was characterized using high-throughput ITS2 amplicon sequencing. We compared the species-level identification success rates from our bulk community samples using our regional ITS database and the widely used UNITE database. The custom regional database resulted in significantly higher species-level assignments (72.3%) of candidate species than the UNITE database (28.3-34.2%). Within each site, identification of candidate species ranged from 72.3-82.1% using the custom database; and 31.5-55.4% using the UNITE database. These results highlight that developing regional databases may accelerate a wide range of LFF research by improving our ability to characterize species-level diversity using DNA barcoding.

Keywords: Illumina; OTUs; UNITE; internal transcribed spacer (ITS); metabarcoding; species hypothesis; taxonomic assignment.

Figure 1

Species/species hypotheses (SH) identification success rates using the custom regional database (“LIMW database” on BOLD; https://boldsystems.org (accessed on 8 May 2023)) and the UNITE databases at two different sequence similarity thresholds. Each bar shows the proportion of identified versus unidentified species hypotheses (SHs) out of the total SHs. For all figures, “Regional” refers to the database made for this study; “UNITE __%” refers to UNITE as assessed through the BLAST Percentage ID statistic through the NCBI BLAST interface. The x-axis shows the database in question, while the y-axis shows the proportion of SHs identified to species-level.

Figure 2

Species/species hypotheses identification success rate by family using the custom regional database (“LIMW database” on BOLD; https://boldsystems.org (accessed on 8 May 2023)) and the UNITE database at two sequence similarity thresholds. Each bar shows the proportion of identified versus unidentified species hypotheses when grouped in taxonomic families. The x-axis shows the database in question, while the y-axis shows the proportion (out of 1.00). Each facet graph is one taxonomic family.

Figure 3

Species/species hypotheses identification success rate by site using the custom regional database (“LIMW database” on BOLD; https://boldsystems.org (accessed on 8 May 2023)) and the UNITE database at two sequence similarity thresholds. Each bar shows the proportion of identified versus unidentified species hypotheses when grouped by geographic sampling site. The x-axis shows the database in question, while the y-axis shows the proportion (out of 1.00). Each facet graph is one geographic sampling site.

Figure 4

Venn diagrams comparing the overlap of species/species hypotheses among five regions: the “Central Basin and Range” ecoregion; an urban site (Provo, UT, USA) in the Central Basin and Range; the Pine Valley Mountains, a transition zone in southwestern Utah between the Colorado Plateau and Great Basin; the Escalante River Watershed in southern Utah (a major drainage into the Colorado River); and an alpine steppe habitat on a desert sky island in southern Utah. The total number of species/species hypotheses for each region is given in parentheses in bold text, and the number of shared species/species hypotheses is given for each comparison.