Rishi Aree
Pronouns: he/him
Research Mentor(s): Erik Kiledal
Research Mentor School/College/Department: Earth and Environmental Sciences / LSA
Program:
Authors: Rishi Aree, Greg Dick, Anders Kiledal
Session: Session 5: 2:40 pm – 3:30 pm
Poster: 30
Abstract
Harmful algal blooms dominated by the cyanobacteria Microcystis have plagued Lake Erie for many years. The blooms are largely driven by nutrient runoff, specifically phosphorus runoff carried by the Maumee river from farmland in northeast Ohio to Lake Erie. The production of toxins, particularly microcystin, has an effect on surrounding wildlife and impacts the lake as a drinking water source. We have collected microbiome data during routine monitoring of the algal blooms in Lake Erie from 2015-2021 in the form of the 16S amplicon sequences, a specific gene sequence used to identify the microbial community composition. We employed machine learning methods to predict potential interactions and associations between organisms and toxin production, yielding models that can be used to inform potential for toxin production from the microbial community present. In particular, we evaluated multiple different machine learning algorithms to determine which was best able to predict toxin production, and which features, including microbes and environmental factors most informed the model. With this information, we identify additional targets for future laboratory work on interactions between bloom forming cyanobacteria such as Microcystis and associated heterotrophic bacteria. These findings could allow toxin production to be added to existing bloom forecasts, providing valuable information to Lake Erie stakeholders.
