Modeling the microbiome of Utah's Great Salt Lake: A regression analysis of key abiotic factors impacting growth of Dunaliella green algae in the GSL's South Arm
Since the mid-1800s, Utahs Great Salt Lake (GSL) has undergone dramatic changes. Due to the effects of climate change and an increase in agricultural, industrial, and residential water usage to support population growth, the present water level has fallen to about one-fourth of its highest recorded level in 1987 [1, 2]. As Earths global air and water temperatures continue to rise, evaporation rates from this closed basin will also rise, thus increasing the salinity of this already hypersaline lake. A shift in water chemistry from its current salinity of 15% to a halite saturation of 30% will negatively impact the populations of Dunaliella viridis - a halophilic species of green algae that form the basis of the simple but delicate food web in the South Arm of the GSL. Disruption of the D. viridis population through increased water temperature and salinity will spur a negative cascade throughout the food chain by reducing brine shrimp populations and thereby threaten local and migratory bird populations. Since increasing water temperature and salinity can have such deleterious ramifications on both D. viridis and the overall lake ecosystem, a predictive model that maps the impact of changing water temperature and salinity to specific growth values for D. viridis is needed for forecast-assisted management. In support of this goal, we developed a multiple linear regression model using twelve years of observational data consisting of chlorophyte (of which Dunaliella are the dominant species) population concentrations under co-varying water temperature and salinity. The resulting fitted data produced an R2 value of 0.17 with a RMSPE of 100.704, and additional diagnostics were conducted to verify the model. Overall, this model predicts that chlorophyte populations will decrease by 0.41 g/L for each 1% increase in salinity and decrease by 0.74 g/L for each 1{degrees}C increase in water temperature up to the extinction point of 30% salinity and 45{degrees}C. One limitation of the linear regression model is its inability to capture trace algal population concentrations at 0 g/L. To address this, we also developed a zero-inflated Poisson regression model, which predicts similar decreases in chlorophyte populations for increasing water temperature and salinity as the linear regression model. The fitted data for this model produced a pseudo-R2 value of 0.35 with a RMSPE of 90.026. This model predicts that chlorophyte populations will decrease by 0.16 g/L for each 1% increase in salinity and decrease by 0.13 g/L for each 1{degrees}C increase in water temperature up to the extinction point of 30% salinity and 45{degrees}C. Even for a limited climate change scenario of an increase in air/water temperature of 2.5{degrees}C and an associated increase in salinity by 7.5%, the linear regression model predicts a potential loss of [~]224, 000 kg total of chlorophytes from the South Arm of the GSL (based on the median chlorophyte concentration between 2001 and 2006), while the Poisson regression model predicts a potential loss of [~]173, 200 kg of chlorophytes. Continued research will include model selection and error quantification. More broadly, future work aims to constrain chlorophyta population predictions based on D. viridis growth limits under maximum water temperature and salinity thresholds obtained from controlled laboratory experiments, which can be used to identify a microbial tipping point of the GSL.