Sample size
Although the DNA extraction method may have a strong impact on taxa recovery in metabarcoding studies (Schiebelhut, Abboud, Gómez Daglio, Swift & Dawson 2017; Sinha et al. 2017), previous comparisons of different DNA isolation kits for diatoms have demonstrated compatible patterns for diversity and community assembly (Vasselon, Domaizon, Rimet, Kahlert & Bouchez 2017). The present study also shows that richness and community structure of diatoms are highly correlated between metabarcoding data of 10 g vs . 0.5 g of sediment samples. Interestingly, the correlations of relative abundances of matching diatom genera between metabarcoding and microscopy data sets, resulted in higher correlation values for the HTS 0.5 data (Fig. S4). However, other studies comparing metabarcoding results from DNA extracts of various amounts of substrate have reported contrasting results. For example, Penton, Gupta, Yu and Tiedje (2016) reported significant effects of sample size (in terms of input quantity) on the community structure of fungi and bacteria from soil. Higher diversity estimates were associated with 10 g of soil DNA extracts compared with 5 g, 1 g and 0.25 g. Studying meiofaunal communities from marine sediment samples, Brannock and Halanych (2015) found that different extraction quantities did not result in significantly different diversity estimates, however, the OTU community compositions were different. Exploring various eukaryotes from sediment samples, Nascimento, Lallias, Bik and Creer (2018) also found significantly different diversity metrics and community compositions for various sample sizes. They suggested that larger volumes of sediment are necessary to capture the representative metazoan communities compared to the non-metazoan eukaryotes. Therefore, the choice of a quantity of sediment for DNA extraction may depend on the expected distribution of the target groups in the substrate, where the detection of more patchily distributed metazoan communities requires larger quantities of sediment for analyses. Although it could be hypothesized that sample size may affect also the recovery of some microbial groups, here we demonstrate that this was not the case for diatoms from lake sediment samples (when comparing sample size on 10 g vs . 0.5 g). The benefits of using only up to 0.5 g of the sample include a more time- and cost-effective DNA extraction procedure, and the possibility to conduct meaningful analyses when only a limited amount of sediment is available.