Get Complete Project Material File(s) Now! »
When present in the environment, FIB is exposed to various environmental stressors, considerably different than their primary habitats. A significant number of studies have been exploring the relationships between the FIB concentrations in surface water and environmental factors such as temperature, pH, salinity, nutrients and organic matter, sunlight, microbiota, association to suspended sediment, etc. (Van Elsas et al., 2011). While most of the published studies were conducted in temperate regions have provided an overview of the major factors involved in FIB survival (Ferguson and Signoretto, 2011; Pachepsky and Shelton, 2011; Philipsborn et al., 2016), major knowledge gaps remain in tropical ecosystems (Rochelle-Newall et al., 2015).
During rainfall events, secondary and tertiary FIB sources are mobilized, and disseminated in the environment in both temperate and tropical regions. However, in temperate regions, higher rainfall occur during the cold winter season thereby tending to reduce FIB number (Mitch et al., 2010), whereas in humid tropical regions, heavy and erosive rainfall occur during the warm wet season which can affect both survival and the transport of FIB (Pandey et al., 2012). FIB number is expected to show a seasonal variability, yet its behavior may vary considerably across different climates.
Researchers in temperate regions investigated FIB seasonality in a small catchment in Canada, where FIB loading were higher in stormflow when compared to baseflow (Sinclair et al., 2009). This result is further confirmed in another study conducted in North Carolina by Stumpf et al., 2010. The latter reported an average total stormflow loading of E. coli and Enterococcus sp. higher by 30 and 37 times respectively than baseflow loading. Similar trends were noted in a highly urbanized watershed in Republic of Korea where greater FIB concentrations occurred at higher rainfall intensity (Cho et al., 2010a, 2010b). Epidemiological study in the UK showed significant relationship between diarrhea and climate variables (Nichols et al., 2009). Several authors linked the deterioration of surface water quality by increasing FIB concentrations, to streambed sediments resuspension and to an increased runoff washing-off contaminated soils (Henry et al., 2016; Hrdinka et al., 2012; L et al., 2019; Stumpf et al., 2010; Vermeulen and Hofstra, 2014).
These processes are further exacerbated in tropical ecosystems due to erosive frequent rainfall events (Lal, 1983) that generate strong overland flow and soil erosion as well as sediment and bacteria transport to downstream environment (Gourdin et al., 2014; Olivier; Ribolzi et al., 2011). Therefore, it is not surprising to find an increasing number of studies identifying the significant relationship between FIB concentrations and rainfall (Staley et al., 2012; Walters et al., 2011). A recent study done in a subtropical catchment in Australia, have reported a tenfold increase of E. coli and Enterococcus sp. in wet compared to dry weather samples (Ahmed et al., 2018). A one year monitoring of three small catchments in Lao PDR, Vietnam and Thailand reported continuous FIB in-stream concentrations, with the highest occurring in wet season (Rochelle-Newall et al., 2016). The latter attributed the smaller E. coli peaks during dry season to the intense farming and the small episodic rainfall events. Thereby the latter highlighted the importance of an adequate sampling protocol during stormflow, in order to capture the FIB variability and peaks.
In addition to the increased overland flow during tropical rainfall events, the importance of groundwater during storm events (Kirchner, 2003) contributions to FIB dissemination was highlighted in few studies (Ribolzi et al., 2016; Rochelle-Newall et al., 2016). Based on a method using fallout radionuclides (7Be and 210Pbxs) and hydrograph separations, applied in a tropical mountain catchment, authors found that overland flow contributed just over one tenth of total flood volume, but was responsible for more than two-third of the E. coli transferred downstream. On the other hand, groundwater flow typically comprising low E. coli concentrations, contributed up to 89 % of the flood volume. While groundwater can control the FIB dilution magnitude in the stream flow, it can also increase the streambed sediments and FIB resuspension. In a recent study on a small tropical mountain headwater catchment, showed that overland flow contributed of about 41% to overland flow and an average of 89% to the in-stream E. coli concentrations (Boithias et al., 2021b). The latter reported that while sub-surface flow was dominating, mean E. coli was lower yet still exceeding 1,000 MPN 100 mL−1, which suggests that streambeds can be a considerable source and sink of E. coli (Pachepsky et al., 2017; Smith et al., 2008).
Contrary to the widely reported positive association between rainfall events and FIB concentrations in surface water, contradictory findings of epidemiological studies highlight the complex relationship between hydro-meteorological factors and the seasonality of waterborne diseases outbreaks (Guzman et al., 2015; Jagai et al., 2009). As stated by Boithias et al., 2016, diarrhea epidemics in the Luang Prabang area started during dry season triggered by water shortage and ended during the wet season due to aquifer refill. These findings suggest that anthropogenic drivers, like type of water supply, and human behavior, were at least as important as environmental factors in predicting diarrheal risks.
The risk of waterborne disease transmission is likely to increase with the increase with future shifts in climate, and extreme events related to climate change, such as floods and drought, especially in vulnerable regions of the tropical belt (Hofstra, 2011; UNESCO and UN-Water, 2020). Therefore, in addition to hydrology, other important interactive factors are to be taken into consideration when investigating major FIB controlling factors in surface water, like attachment to suspended sediments, rapid land use and climate changes, hygiene practices, etc. This stresses the need to further investigate FIB fate, which up to now, still has not been sufficiently explored in developing tropical countries.
Association to suspended particles, and aquatic organisms
The ability of bacteria to attach to surfaces and particles has been of considerable interest to many researchers in various fields. In fact, bacterial cells generally have a net negative charge on their cell walls that varies depending on the species, ionic strength, pH, etc. The attachment of bacteria including FIB to particles, is mediated by a complex range of physical, chemical and electrostatic bindings (Olsen et al., 1982; Palmer et al., 2007).
In soil and sediments, bacteria is likely to be found attached to particles rather than free-living (Oliver et al., 2007). However, in aquatic ecosystems, the particle-attached bacterial proportion is highly variable across study areas, ranging from 10% in clear water to over 70% in highly turbid estuaries (Crump et al., 1998; Lemée et al., 2002). Turbidity tend to be higher in tropical systems during the wet season following heavy erosive rainfall events (Milliman and Syvitski, 1991; Milliman et al., 1983). A study conducted in a high altitude tropical catchment in Uganda, showed strong positive correlations between the presence of FIB in aquatic systems and suspended sediments concentrations (Byamukama et al., 2005). Another study in the Hudson River in USA, stated a significant positive correlation between turbidity and FIB. The same study showed that over half of the Enterococci was attached to particles in the water column (Suter et al., 2011). Furthermore, previous investigations identified significant negative correlations between particle size and bacterial association, which raises concerns on the bacterial transport further downstream and its consequences on water quality (Oliver et al., 2007; Petersen and Hubbart, 2020a).
In addition to the attachment to suspended particles, many bacteria are known capable of producing and proliferating in biofilms, enhancing their access to nutrients and organic matter (September et al., 2007; Wingender and Flemming, 2011). Moreover, few studies in temperate and sub-tropical systems, found FIB associated with macrophytes and filamentous cyanobacteria (Byappanahalli et al., 2003; Vijayavel et al., 2013), which increased the FIB persistence in the water column of a constructed wetland in Arizona (Karim et al., 2004). However, limited information is available on bacterial associations to macrophytes, biofilms or cyanobacteria in tropical humid climate.
Given current understanding, all these bacterial associations may reflect favorable conditions for bacterial survival, such as providing access to nutrients, and protection from stressors like predation and sunlight (Amalfitano et al., 2017; Walters et al., 2014). Much is still unknown on the partition of particle-attached and free-living bacteria in tropical systems and on the attachment impact on FIB and pathogens survival, decay and transport in a wide range of tropical systems. Overall, the impact of land use changes subject to more erosive processes, on the FIB transport and persistence has important health risk implications and remains to be evaluated.
Modelling studies for FIB fate and transport
There is a need for studies providing new insights on mechanistic understanding of diverse global changes like climate change impacting surface water quality. More particularly, tropical regions are expected to face significant climate change consequences (frequency and intensity of extreme rainfall and/or drought events) (Vörösmarty et al., 2010; Vorosmarty and Sahagian, 2000). The latter, combined with the population growth, and with rapid changes in land and water management practices, is likely to result in significant changes in overland flow, soil erosion, sediment and bacterial transport across the watershed. In order to address these challenges, modeling tools can be useful to (i) understand FIB sources, fate and transport, (ii) help watershed stakeholders to establish adequate strategies to reduce the exposure risk on public health, (iii) analyze various future global changes scenarios, and (iv) develop early warning predictive tools for vulnerable communities affected by water contaminations. Three main types of mathematical approach are widely used to investigate and model the complex hydrological sedimentary and bacterial responses to changes in land use and climate (Miller et al., 2013).
The first approach is based on statistical models like empirical regression such as multiple regressions (Chu et al., 2011; Mahloch, 1974) and double mass curves (Tang et al., 2013; Wang et al., 2012) as well as artificial neural networks (Thoe et al., 2012). These approaches usually use environmental variables as input data, and provide FIB concentrations as an output.
The second approach uses mechanistic or physical process-based models. These models can simulate FIB transport through processes like advection dispersion equations (Wilkinson et al., 1995), streambed resuspension (Cho et al., 2010b). Unlike the first approach, this one allows considering and ranking underlying mechanisms responsible for the mobilization, transfer and concentration of FIB in the system.
The third approach is based on spatially distributed and semi-distributed models at watershed scale, e.g. ECOMSED (Blumberg and Mellor, 2012), SWAT (Arnold et al., 1998), SENEQUE/Riverstrahler (Billen et al., 1994), MIKE (DHI, 2011), and HSPF (Abbas et al., 2021). These models consider watershed morphology, soil properties, hydrology, land use, and pollution point sources. They can also take into account processes like streambed bacteria resuspension (Kim et al., 2010), FIB die-off rates, and, FIB transport model (Chin et al., 2009; Dorner et al., 2006). This approach allows to test complex scenarios and simulate the impact of future changes in the system and their potential impacts on public health on the catchment (Cho et al., 2012; Kashefipour et al., 2002).
However, these models require comprehensive knowledge of (i) the drainage basin morphology, and hydro-meteorology data for model parametrization (ii) long-term water quality data for calibration (iii) FIB die-off, deposition, runoff and resuspension rates, and (iv) FIB sources. Access to these data can be challenging and limited in some tropical and developing countries with sparse water quality monitoring networks. Nonetheless, this research area is expanding and some of the existing models have been adapted to tropical systems (Le et al., 2005; Luu et al., 2010). Few, to our knowledge, were successfully capable of simulating the transport and fate of FIB in these systems (Causse et al., 2015; Coffey et al., 2013; Kim et al., 2017; Thoe et al., 2012), which provides new opportunities to explore this research area.
Each model presents some advantages and limitations. For instance, MIKE model is good for small river basins or water bodies, and can generate hourly output data. On the other hand, the simulation time is longer which makes it difficult to apply it for climate change scenario analysis (Islam et al., 2021). SWAT (i) has a large user community, (ii) covers a large range of processes at watershed-scale, (iii) can be used for various applications, and (iv) simulates from sub-daily time step to inter-annual time step. Another advantage of SWAT over other models like hydrological simulation program fortran (HSPF) is that it allows the simulation of persistent and less-persistent bacteria populations in the same model run (Niazi et al., 2015; Qiu et al., 2018).
Progress has been made towards improving SWAT performance in simulating FIB fate and transport at watershed-scale. SWAT is able to estimate FIB sources and loading in watershed (Coffey et al., 2010), and to assess the magnitude of FIB sources within the watershed (Coffey et al., 2013). SWAT was further improved by including in-stream processes to estimate the impact of the streambed sediment re-suspension or deposition on FIB number (Kim et al., 2010; Pandey et al., 2012). Furthermore, SWAT was able to better reproduce the seasonal variability of bacteria after including bacterial growth/die-off adjustable by changing temperature (Cho et al., 2016). Recent advances showed better simulation of low concentrations of bacteria during the dry season with associated base flow, after taking into account additional bacteria in-stream processes like hyporheic exchange process (Kim et al., 2017).
Specific objectives and scientific approach
Surface water contamination by fecal pathogens remains a major threat to public health in developing countries of the tropical region. Despite the significant advances made towards a better understanding of the FIB dynamics in temperate regions, many knowledge gaps exist on the underlying mechanisms of fecal contamination in tropical conditions. There is a need for studies to provide new insights regarding FIB transport and survival in tropical conditions at watershed-scale, in order to mitigate health risks associated to the use of contaminated water.
In the Mekong basin, over 70 million people rely on unimproved surface water for their domestic requirements. Communities living in these areas are not only exposed to continuous fecal contamination from point and diffuse sources, but are also facing rapid global changes (land use, climate change, hydropower dams) with various consequences on FIB fate and public health. These anthropic activities are likely to have consequences on water contamination at various scales.
Therefore, the main objective of this research was to characterize the dissemination of FIB and its dynamics at different spatio-temporal scales in the lower Mekong basin. This thesis work aims to answer the following questions:
(i) What are the different factors that control E. coli at the scale of large Mekong River tributaries in Lao PDR?
(ii) How do two key environmental factors (solar radiation exposition and suspended particles deposition) affect the decay or survival of E. coli in a tropical mountainous wetland?
In the current version of the SWAT model, bacteria fate processes are described for both the land and the routing phase. Bacteria is introduced into the environment through manure application, where they can be applied on soil surface or intercepted by foliage. SWAT model partitions bacteria on the soil surface into free-living bacteria in soil solution and particle-attached bacteria. Depending on the hydrology, and on the free-living/particle-attached bacteria partition, bacteria can be transferred to the river network with surface runoff during rain events: × ℎ × , = × (2.14).
where bactsurf is the bacteria number transported in surface runoff (MPN m-2), kbact,surf is the parameter of bacteria soil partitioning (m3 Mg-1), and Qsurf is the amount of surface runoff (mm). ℎ= 0.0001 ×× ( ) × , (2.15).
where bactsed is the bacteria number transported with suspended sediments in surface runoff (MPN m -2), concsed,bact is the amount of bacteria attached to soil particles in the top soil (MPN ton-1), sed is the yield of suspended solids (ton), areahru is the HRU area (ha), and εbact,sed is the ratio of bacteria enrichment.
Bacteria can be leached along the soil profile and assumed to die in deeper soil layers. The decay/regrowth of both free-living and particle-attached bacteria is modeled on foliage, and on surface soil. Chick’s Law first order equation determines the quantity of removed bacteria by decay or the quantity of added bacteria by regrowth: = 0 exp(−µ ) (2.16).
where N is the number of bacteria in a given time t (CFU 100 mL-1), N0 is the original number of bacteria indicator (CFU), μ is the bacterial decay rate constant (h−1), and t represents time (h). In SWAT model, temperature is one of the variables that determines the decay rate which can be obtained as: µ = µ20 −20 (2.17).
where μ20 refers to the decay rate at 20°(h-1), θ is the temperature correction parameter for the first-order decay, and T shows the temperature in °C..
In addition to the simulation using the original bacteria module of SWAT, we have tested the modified version of SWAT bacteria module developed by Kim et al., (2017), which takes into account additional bacteria in-stream processes (Fig. 17). In order to simulate low concentrations of bacteria during the dry season with associated base flow, the hyporheic exchange process was implemented in the model. The amount of the bacteria released through the sediment pore fluid by hyporheic exchange into the stream was estimated by Grant et al., (2011). Once in the river, bacteria can either die off or regrow. Moreover, depending on the hydrological conditions of the reach, bacteria can settle with sediments or be released during sediment resuspension events.
Table of contents :
Chapter 1. General introduction
1.1 General context and problematic
1.2 Fecal contamination and water quality
1.2.1 Water quality standards and fecal indicator bacteria
1.2.2 Waterborne diseases
1.2.3 Water quality assessment
1.3 FIB sources
1.3.1 Primary sources
1.3.2 Point sources
1.3.3 Diffuse sources
1.4 FIB fate at watershed-scale
1.5 FIB occurrence in the environment
1.5.1 Environmental factors
1.5.2 Anthropic factors
1.6 Modelling studies for FIB fate and transport
1.7 Specific objectives and scientific approach
Chapter 2. Material and methods
2.1 Study site
2.1.1 General description of the Mekong basin
2.1.2 Main challenges on the Mekong basin
2.2.1 In-situ observations
2.2.2 Experimental approach
2.2.3 Modelling approach
Chapter 3. Effects of hydrological regime and land use on in-stream Escherichia coli concentration in the Mekong basin, Lao PDR
3.2 Material and methods
3.2.1 Study site
3.2.2 Sampling design and watersheds characteristics
3.2.3 Geographical analyses
3.2.4 Land use
3.2.5 Data on livestock and local populations
3.2.6 In-situ measurements and laboratory analyses
3.2.7 Rainfall and water level
3.2.8 Statistical analysis
3.3.1 Spatial surveys conducted during the 2016 dry and rainy seasons
3.3.2 Water quality monitoring of three northern watersheds during 2017 and 2018
Chapter 4. Apparent decay rate of Escherichia coli in a mountainous tropical headwater wetland
4.2 Material and methods
4.2.1 Study area
4.2.2 Experimental design
4.2.3 Mesocosms preparation
4.2.4 Analytical methods
4.2.5 Environmental variables
4.2.6 Apparent Decay Rates, T50 and T90 values
4.2.7 E. coli stock variations
4.3.1 Environmental variables
4.3.2 Physico-chemical and microbiological variables
4.3.3 Apparent decay rates and T50 and T90 values
4.3.4 E. coli stock variations
4.4.1 Particle attachment effect on E. coli apparent decay rates
4.4.2 Deposition effect on E. coli apparent decay rates
4.4.3 Solar radiation effect on E. coli apparent decay rates
4.4.4 Relative effects of light and sedimentation on E. coli apparent decay rates
Chapter 5. Impact of hydropower dams on hydro-sedimentary and Escherichia coli dynamics on watershed-scale, case of Nam Khan river in the lower Mekong basin
5.2 Material and methods
5.2.1 Study area
5.2.2 Monitoring data
5.2.3 Statistical approach
5.2.4 Hydrological modeling
5.3.1 Statistical analyses
5.3.2 Hydrological modeling
5.4.1 Observed hydrological alterations
5.4.2 Sediment trapping
5.4.3 Bacteria trapping
5.4.4 Hydrological modeling
5.4.5 Structural and parametric uncertainties
Chapter 6. Conclusion & perspectives