Evaluation of the marker for metabarcoding of marine diatoms and inference of population structure of selected genera.

Diatoms are one of the most important phytoplankton groups in the world's oceans. There are responsible for up to 40% of the photosynthetic activity in the Ocean, and they play an important role in the silicon and carbon cycles by decoupling carbon from atmospheric interactions through sinking and export. These processes are strongly influenced by the taxonomic composition of diatom assemblages. Traditionally, these have been assessed using microscopy, which in some cases is not reliable or reproducible. Next-generation sequencing enabled us to study diversity in a high-throughput manner and uncover new distribution patterns and diversity. However, phylogenetic markers used for this purpose, such as various 18S rDNA regions, are often insufficient because they cannot distinguish between some taxa. In this work, we demonstrate the performance of the chloroplast-encoded marker for metabarcoding marine diatoms compared to microscopy and 18S-V9 metabarcoding using a series of monthly samples from the Gulf of Trieste (GoT), northern Adriatic Sea. We demonstrate that is able to detect more taxa compared to 18S-V9 metabarcoding or microscopy, while the overall structure of the diatom assemblage was comparable to the other two methods with some variations, that were taxon dependent. In total, 6 new genera and 22 new diatom species for the study region were identified. We were able to spot misidentification of genera obtained with microscopy such as , which was mistaken for , as well as genera that were completely overlooked, such as and several genera from the Cymatosiraceae family. Furthermore, on the example of two well-studied genera in the region, namely and particularly , we show how the method can be used to infer even deeper phylogenetic and ecologically significant differences at the species population level. Despite a very thorough community analysis obtained by the incompleteness of reference databases was still evident, and we shed light on possible improvements. Our work has further implications for studies dealing with taxa distribution and population structure, as well as carbon and silica flux models and networks.