The Effect Of Habitat And Didymosphenia Geminata On Macroinvertebrate Community Structure In The South Holston River Tailwater

In the South Fork Holston River, near Bristol, Tennessee, we examined the effects of habitat and the diatom Didymosphenia geminata (only in riffle and run mesohabitats) on benthic macroinvertebrate communities in three river reaches along a tailwater river thermal gradient and in one control site located upstream of the reservoir between January 2015 and March 2016. Directly downstream of the dam, the South Fork Holston Tailwater has a very narrow annual thermal range (11.83°C) and few cumulative growing degree days (GDD=243); abundance of Ephemeroptera, Plecoptera and Trichoptera (EPT) was as much as 80% lower at this site than at the two downstream sites, which had higher annual thermal ranges (19.39°C, 21.54°C respectively) and more degree growing days (GDD=438, 522 respectively). Species richness at the control site (Mean S=20) was 2–3× higher than any of the tailwater sites, although there was little variation in richness between tailwater river reaches (Mean S=7–11). We also found that D. geminata ash-free dry mass (AFDM) and dry mass differed significantly between mesohabitats with significantly higher D. geminata AFDM and dry mass in run habitats. Partial least squares regression (PLS) revealed that surface flow, benthic flow, dissolved oxygen, conductivity, and turbidity were significant variables of importance for predicting the observed variation in D. geminata AFDM and accounted for 45.2% of the total variation in D. geminata AFDM between sites. A second PLS regression found that surface flow, benthic flow, temperature, conductivity, turbidity, and D. geminata AFDM were significant variables of importance for predicting Shannon diversity (H’) between sites and accounted for 51.4% of the variation in H’. However, D. geminata AFDM had a much lower VIP value relative to other significant environmental predictors (e.g. surface flow, benthic flow, temperature, and conductivity), and therefore is only a minimal contributor to the overall variation in Shannon diversity at study sites