Morphology and metabarcoding: a test with stream diatoms from Mexico highlights the complementarity of identification methods

Diatoms are among the most commonly used bioindicators. Correct taxonomic identifications are critical to their use as bioindicators because closely related diatom species can respond differently to water physicochemical characteristics and pollutants. However, diatom identification based on morphology can be time consuming, and requires highly specialized taxonomic skills. To optimize diatom identification, DNA metabarcoding is increasingly used because it is generally less time consuming and may be more accurate than morphological identification. To date, however, neither DNA metabarcoding nor DNA barcoding diatom studies have been conducted in Mexico. Thus, we studied epilithic diatoms from streams in Central Mexico with a combination of morphological and metabarcoding techniques, and compared the diatoms identified and quantified by each method. We also assembled a barcode reference library based on clonal culturing. This library is composed of 190 strains that belong to 72 species in 24 genera. The morphological analysis of environmental samples resulted in the identification of 204 infrageneric taxa in 42 genera, but clonal culturing from the same samples retrieved 12 additional infrageneric taxa and 1 additional genus, thereby revealing concealed diversity. The metabarcoding approach resulted in the identification of 266 infrageneric taxa that belonged to 35 genera. Together, these methods detected 49 genera. Of these genera, 14 were identified only by morphology, 29 were identified by both methods, and 6 were only identified by metabarcoding. Of the 266 taxa we retrieved with metabarcoding, we confidently assigned 94 infrageneric taxa because a direct morphological or barcode sequence correlation was possible. Thirty-four of these 94 taxa were only detected with the metabarcoding method. One-fourth (23) of the assignments were only possible because of the barcode reference library we developed during this study, because there were no existing barcode sequences that matched these barcodes in the International Nucleotide Sequence Database Collaboration databases. Large disparities existed between relative abundances based on valve counts and sequence reads of the most abundant taxa, even after we corrected for cell biovolume. Overall, we conclude that the combination of morphological and molecular methods increases the detection and identification of diatoms.