We explore in-situ fluorescence spectroscopy as an instantaneous indicator of total bacterial abundance and faecal contamination in drinking water. Eighty-four samples were collected outside of the recharge season from groundwater-derived water sources in Dakar, Senegal. Samples were analysed for tryptophan-like (TLF) and humic-like (HLF) fluorescence in-situ, total bacterial cells by flow cytometry, and potential indicators of faecal contamination such as thermotolerant coliforms (TTCs), nitrate, and in a subset of 22 samples, dissolved organic carbon (DOC). Significant single-predictor linear regression models demonstrated that total bacterial cells were the most effective predictor of TLF, followed by on-site sanitation density; TTCs were not a significant predictor. An optimum multiple-predictor model of TLF incorporated total bacterial cells, nitrate, nitrite, on-site sanitation density, and sulphate (r2 0.68). HLF was similarly related to the same parameters as TLF, with total bacterial cells being the best correlated (ρs 0.64). In the subset of 22 sources, DOC clustered with TLF, HLF, and total bacterial cells, and a linear regression model demonstrated HLF was the best predictor of DOC (r2 0.84). The intergranular nature of the aquifer, timing of the study, and/or non-uniqueness of the signal to TTCs can explain the significant associations between TLF/HLF and indicators of faecal contamination such as on-site sanitation density and nutrients but not TTCs. The bacterial population that relates to TLF/HLF is likely to be a subsurface community that develops in-situ based on the availability of organic matter originating from faecal sources. In-situ fluorescence spectroscopy instantly indicates a drinking water source is impacted by faecal contamination but it remains unclear how that relates specifically to microbial risk in this setting.

Fluorescent natural organic matter at tryptophan-like (TLF) and humic-like fluorescence (HLF) peaks is associated with the presence and enumeration of faecal indicator bacteria in groundwater. We hypothesise, however, that it is predominantly extracellular material that fluoresces at these wavelengths, not bacterial cells. We quantified total (unfiltered) and extracellular (filtered at < 0.22 µm) TLF and HLF in 140 groundwater sources across a range of urban population densities in Kenya, Malawi, Senegal, and Uganda. Where changes in fluorescence occurred following filtration they were correlated with potential controlling variables. A significant reduction in TLF following filtration (ΔTLF) was observed across the entire dataset, although the majority of the signal remained and thus considered extracellular (median 96.9%). ΔTLF was only significant in more urbanised study areas where TLF was greatest. Beneath Dakar, Senegal, ΔTLF was significantly correlated to total bacterial cells (ρ_s 0.51). No significant change in HLF following filtration across all data indicates these fluorophores are extracellular. Our results suggest that TLF and HLF are more mobile than faecal indicator bacteria and larger pathogens in groundwater, as the predominantly extracellular fluorophores are less prone to straining. Consequently, TLF/HLF are more precautionary indicators of microbial risks than faecal indicator bacteria in groundwater-derived drinking water.

Soil hydraulic parameters (SHPs) required as inputs for numerical models are scarce in Sahelian regions. Instead, they are estimated using pedotransfer functions (PTFs), but their ability to simulate soil water dynamics has not been evaluated yet. Therefore, this study aims to parametrise SHPs with seven different PTFs and inverse modelling to gain insight into their ability to simulate water fluxes in Senegal’s Groundnut basin. We used four years of field measurements of soil moisture (SWC) and actual evapotranspiration (ET_a) under pearl-millet and groundnut crop rotation for model evaluation. Inverse modelling for SWC (RMSE ≤ 0.015 cm³ cm⁻³) and ET_a (RMSE ≤ 0.62 mm d⁻¹) yielded the best model performance values compared to PTFs (0.024-0.175 cm³ cm⁻³ and 0.68-0.96 mm d⁻¹). Where field measurements are lacking for inverse estimation, three of the seven tested PTFs yielded good modelling results and could be used as a parsimonious approach for cultivated Sahelian soils.