*3.5. ITS-Based BLAST Performance*

BLAST performance was similar overall among the four scenarios (ITS1 region only, ITS1 including ASVs, ITS2 region only, and short subterminal ITS2 region). In all four cases, BLAST hits with other terminals of the same ad hoc-delimited species yielded E scores comparable to those obtained with self hits, although slightly lower on average, reflecting infraspecific variation (Figure 8). E scores also discriminated well between hits and misses. Upon closer examination, however, the four scenarios showed subtle differences. The best level of discrimination between hits and misses, and hence the highest level of accuracy was found with the ITS2 region: 88% of the queries accurately discriminated between hits and misses (min hit score > max miss score); however, in 53% the min hit score was 10% larger than the max miss score and in 23% of the queries it was 25% larger. The second best-performing subset was the ITS1 region, with 74%, 32%, and 13%, respectively, still spanning a large number of accurate queries but overall with a lower BLAST gap. When incorporating all available ASVs (including smaller fragments from Sanger and HTS) the ITS1 region performed slightly worse (66%, 25%, 11%), due to a number of incomplete query sequences missing parts of diagnostic regions. The short subterminal ITS2 string, making up about one fourth of the entire ITS2 region, had a larger BLAST gap than the other three scenarios, with 27% and 41% of the queries having the min hit score more than 25% and 10% above the max miss score, respectively, but, on the other hand, only 54% of the queries had a min hit score larger than the max miss score, resulting in potential inaccuracy.

**Figure 8.** BLAST results for the four different scenarios (ITS1 only, ITS1 including ASVs, ITS2, and short subterminal ITS2 region). The left column of graphs shows the terminals for each subset ranked by the minimum E score for hits (blue line), as opposed by the maximum E score for misses (orange line). The right row of graphs shows the corresponding overall distribution of E scores for self-hits, hits, and misses, for each subset.
