Do inferences about freshwater phytoplankton communities change when based on microscopy or high-throughput sequencing data?

1. Microscopy and high-throughput sequencing (HTS) detect and quantify algae differently. Little is known if microscopy-based abundance or biomass better compare to HTS data and how methodological differences affect ecological inferences about the phytoplankton communities studied. 2. We investigated methodological (abundancemicroscopy versus abundanceHTS, biomassmicroscopy versus abundanceHTS), habitat (littoral, pelagic, deep hypolimnion), and year differences (2014 versus 2017) for phytoplankton communities of Lake Tovel (Italy) using ANOVA. Specifically, we tested the hypothesis that depending on comparing abundancemicroscopy or biomassmicroscopy, respectively, to abundanceHTS different effects would be indicated; we called this the metric effect. Furthermore using samples from 2014 to 2017, we investigated environment-community relationships by a redundancy analysis (RDA) based on abundancemicroscopy, biomassmicroscopy, and abundanceHTS, and compared the results. 3. Approximately nine times more operational taxonomic units (OTUs) were reported with HTS (n2014 = 819, n2017 = 891) than algal taxa with microscopy (n2014 = 90, n2017 = 109) in 2014 and 2017. While microscopically assessed algal taxa were evenly distributed among phyla, the vast majority of OTUs were attributed to Chrysophyta (2014 = 54%, 2017 = 62%) and Bacillariophyta (2014 = 19%, 2017 = 17%). A metric effect for method differences was generally observed comparing abundancemicroscopy to abundanceHTS with Chlorophyta, Cryptophyta, and Dinophyta showing higher % abundance with microscopy while richness and Chrysophyta showed higher values with HTS. Almost no metric effects were found in 2014, but they were common across phyla in 2017. Bacillariophyta and Eustigmatophyta showed the same habitat differences when comparing biomassmicroscopy to abundanceHTS. Dinophyta showed habitat differences only with microscopy, while Chyrsophyta showed habitat differences only with HTS; these results were probably related to technical bias and strengths of HTS, respectively. 4. Habitat differences of phyla were reasonably related to their ecological niche and linked to factors such as temperature and feeding preferences; furthermore, phyla often showed a significant 2014-versus-2017 year effect. The year 2014 was very wet while 2017 had a dry winter, and we attributed the patterns found to allochthonous nutrient input by rain and decreased turbulence. RDAs based on phytoplankton communities assessed with microscopy and HTS, respectively, equally indicated the importance of hydrology, nutrients, and temperature for phytoplankton communities and discriminated the littoral from the deep hypolimnion. However, variance explained was higher with HTS, and the pelagic was similar to the deep hypolimnion with microscopy but to the littoral with HTS. 5. Despite the different strengths of microscopy and HTS for biodiversity assessment, both datasets outlined similar large-scale patterns linked to strong environmental control of phytoplankton communities as they related to habitat and year differences. According to our hypothesis, metric effects were common; however, no general rule was found if either abundancemicroscopy or biomassmicroscopy, respectively, should be compared to abundanceHTS. Notwithstanding metric effects, HTS-based data provided similar and more detailed information than microscopy, supporting the promise of HTS becoming the tool of the future for biodiversity research