Do biofilm communities respond to the chemical signatures of fracking? A test involving streams in North-central Arkansas

Wilson H. Johnson, Marlis R. Douglas, Jeffrey A. Lewis, Tara N. Stuecker, Franck G. Carbonero, Bradley J. Austin, Michelle A. Evans-White, Sally A. Entrekin, Michael E. Douglas

Unconventional natural gas (UNG) extraction (fracking) is ongoing in 29 North American shale basins (20 states), with ~6000 wells found within the Fayetteville shale (north-central Arkansas). If the chemical signature of fracking is detectable in streams, it can be employed to bookmark potential impacts. We evaluated benthic biofilm community composition as a proxy for stream chemistry so as to segregate anthropogenic signatures in eight Arkansas River catchments. In doing so, we tested the hypothesis that fracking characteristics in study streams are statistically distinguishable from those produced by agriculture or urbanization. Four tributary catchments had UNG-wells significantly more dense and near to our sampling sites and were grouped as 'potentially-impacted catchment zones' (PICZ). Four others were characterized by significantly larger forested area with greater slope and elevation but reduced pasture, and were classified as 'minimally-impacted' (MICZ). Overall, 46 bacterial phyla/141 classes were identified, with 24 phyla (52%) and 54 classes (38%) across all samples. PICZ-sites were ecologically more variable than MICZ-sites, with significantly greater nutrient levels (total nitrogen, total phosphorous), and elevated Cyanobacteria as bioindicators that tracked these conditions. PICZ-sites also exhibited elevated conductance (a correlate of increased ion concentration) and depressed salt-intolerant Spartobacteria, suggesting the presence of brine as a fracking effect. Biofilm communities at PICZ-sites were significantly less variable than those at MICZ-sites. Study streams differed by Group according to morphology, land use, and water chemistry but not in biofilm community structure. Those at PICZ-sites covaried according to anthropogenic impact, and were qualitatively similar to communities found at sites disturbed by fracking. The hypothesis that fracking signatures in study streams are distinguishable from those produced by other anthropogenic effects was statistically rejected. Instead, alterations in biofilm community composition, as induced by fracking, may be less specific than initially predicted, and thus more easily confounded by agriculture and urbanization effects (among others). Study streams must be carefully categorized with regard to the magnitude and extent of anthropogenic impacts. They must also be segregated with statistical confidence (as herein) before fracking impacts are monitored.