Figure 3.

Variation predicted from GAMMs in (a) observed species richness and (b) Chao1 richness estimates assessed through the metabarcoding of DNA extracted from 96% ethanol used to preserve five unprocessed freshwater macroinvertebrate bulk samples, in relation to ethanol subsampling day and DNA extraction method, while controlling for variation in the number of reads across samples. Subsampling was conducted at 1, 2, 3, 5, 7 and 14 days after field sampling, and DNA extractions were performed using three methods: BEAD (purple, solid line), TISSUE (blue, dashed line) and SOIL (green, dotted line). Temporal variation trend lines are provided with the corresponding standard errors. Summary statistics of the GAMMs are provided in Table 2 [Colour figure can be viewed at http://www.wileyonlinelibrary.com]