Bayesian inversion of bacterial physiology and dissolved organic carbon biodegradability on water incubation data

Abstract

In aquatic ecosystems, dissolved organic carbon (DOC) plays a significant role in the global carbon cycle. Microorganisms mineralize biodegradable DOC, releasing greenhouse gases (carbon dioxide, methane) into the atmosphere. Extensive research has focused on the concentrations and biodegradability of DOC in aquatic systems worldwide. However, little attention has been given to uncertainties regarding the physiological characteristics of heterotrophic bacteria, which are crucial for biogeochemical modeling. In this study, the physiological properties of heterotrophic bacteria and the properties of DOC biodegradability in water are inferred through a Bayesian inversion approach. To achieve this, treated and natural water samples collected from the Seine River basin, were inoculated and incubated in laboratory. During incubation, the concentrations of DOC and heterotrophic bacteria biomass were measured. Then, a multiple Monte Carlo Markov Chains method and the HSB model (High-weight polymers, Substrate, heterotrophic Bacteria) are applied on the water incubation data. The results indicate a higher biodegradable fraction of DOC in natural water compared to treated water and significant variability in the fraction of fast biodegradable DOC within 5 days in both water samples. The significant variability highlights the uncertainties/challenges in the HSB model parameterization. The seven water samples used in the paper serve as a proof of concept. They are from various origins and display the potential of the method to identify parameter values in a large range of values. Because mortality rate of heterotrophic bacteria at 20 ^∘C (k_d20) showed a remarkable stability at 0.013 h^-1, we considered that this parameter can be fixed at this value. The maximum growth rates at 20 ^∘C (μ_max20) was 0.061 h^-1 while optimal growth yield (Y) estimated at 0.34 for treated water and at 0.25 for natural water. All these parameter values are well in accordance with previous determinations.

Keywords: Aquatic biogeochemical modeling; Bacterial physiology; Bayesian inversion; DOC biodegradability; Parameter uncertainties.