Abstract
Cyanobacterial communities of three co-located eutrophic sandpit lakes were surveyed during 2016 and 2017 over season and depth using high-throughput DNA sequencing of the 16S rRNA gene. All three lakes were stratified except during April 2017 when the lakes were recovering from a strong mixing event. 16S rRNA gene V4 sequences were parsed into operational taxonomic units (OTUs) at 99% sequence identity. After rarefaction of 139 samples to 25,000 sequences per sample, a combined total of 921,529 partial 16S rRNA gene sequences were identified as cyanobacteria. They were binned into 19,588 unique cyanobacterial OTUs. Of these OTUs, 11,303 were Cyanobium. Filamentous Planktothrix contributed 1537 and colonial Microcystis contributed 265. The remaining 6482 OTUs were considered unclassified. For Planktothrix and Microcystis one OTU accounted for greater than 95% of the total sequences for each genus. However, in both cases the non-dominant OTUs clustered with the dominant OTUs by date, lake, and depth. All Planktothrix OTUs and a single Cyanobium OTU were detected below the oxycline. All other Cyanobium and Microcystis OTUs were detected above the oxycline. The distribution of Cyanobium OTUs between lakes and seasons can be explained by an epidemic-like response where individual OTUs clonally rise from a diverse hypolimnion population when conditions are appropriate. The importance of using 99% identity over the more commonly used 97% is discussed with respect to cyanobacterial community structure. The approach described here can provide another valuable tool for assessing cyanobacterial populations and provide greater insight into the controls of cyanobacterial blooms.