The P dilemma: too much vs. too little
Despite the predictable oncoming of global P resources scarcity, the use efficiency of P fertilizers is still too low, with only 15-30% of the applied P fertilizers being taken up by harvested crop (FAO, 2006). Over the past two decades, the fertilizer industry, governmental institutes, and research organizations have actively supported more efficient fertilizer application practices, such as recycling of organic matter containing P, and using organic farming techniques to optimize soil conditions to increase soil phosphorus availability for plants (FAO, 2008). However, much of the world’s cropland (70%) is still subjected to P surpluses, due to the historical over-application of P onto soils (MacDonald et al. 2011).
Thus, the application of P fertilizer meets the human’s increasing food demand in western countries but meanwhile leads to the accumulation of high quantities of P in agricultural soils. This accumulation of P in soils generates a risk of P transfer from soils to surface waters through soil erosion and leaching (Bouraoui and Grizzetti, 2011). Once in streams, P will find its way to large rivers and estuaries and end up into oceans, it will be retransformed into rocks and eventually uplift to terrestrial layer over geological timescales, where it will form new mineable P rock reserves. Some parts of P in streams can also be deposited in wetlands, lakes, and reservoirs, which add time-lags to the delivery of P into oceans.
However, the excessive presence of bioavailable P in aquatic ecosystems due to human acceleration of the P cycle has damageable consequences for these ecosystems, because P together with nitrogen stimulates the growth of phytoplankton, algae, and aquatic plants, causing the process of water eutrophication (Smith and Schindler, 2009; see Figure 1.3). It is generally accepted that P is the limiting nutrient for eutrophication in freshwater ecosystems, while the coastal water eutrophication is limited by nitrogen (Schindler et al., 2008). Water eutrophication has many deleterious consequences for aquatic ecosystems and humans, including reduction of biological functions and biodiversity, decline of surface water quality for drinking water production and recreational purposes, development of toxin-producing algae like cyanobacteria, etc. In particular, the toxins produced by the proliferation of cyanobacteria in eutrophic freshwater can affect animal and human health through exposure during drinking or bathing (Serrano et al., 2015).
To limit P losses to waters is thus of critical importance, not only to avert the future P crisis by keeping the P cycling in the terrestrial ecosystem but also to secure human health by preventing the eutrophication of water bodies.
P losses to waters: point vs. diffuse sources
Phosphorus can be lost to waters from essentially two sources: i) point sources from urban and industrial wastewater treatment plants and ii) diffuse sources from agricultural landscapes, either through erosion of P-rich soil particles during overland flow, or P leaching during drainage and subsurface water flow (Figure 1.4). With the implementation of targeted measures such as banning P in detergents, precipitating P in sewage water treatment plants, or purifying industrial wastewater for P, P losses from point sources have generally been strongly reduced in many western countries (Collins et al., 2014; Schoumans et al., 2014). However, eutrophication is still occurring in many lakes and estuaries in these countries, pointing to the permanent contribution of diffuse sources that transfer P from agricultural soils to surface waters (Billen et al., 2007; Grizetti et al., 2012; Jarvie et al., 2017; Scavia et al., 2014; Schoumann et al., 2014).
The diffuse transfer of P from soils to waters occurs mainly through soil erosion and overland flow, which are the most important P loss pathways in hilly and mountainous areas, and more importantly from soil leaching and artificial tile drainage in flatter areas (Chapman et al., 2005; Chardon and Schoumans, 2007; Heathwaite et al., 2005b; Nelson et al., 2005; Ulén et al., 2010). Managing agricultural diffuse P loss is difficult because of the large spatiotemporal variations in water pathways and soil erosion processes and the complexity of P forms in both soils and waters (i.e. dissolved organic and inorganic P, colloidal P, and particulate P). Combined with the continuously increasing food demand, the percentage of P loss to watercourses originating from diffuse agricultural sources has increased over the past two decades in Europe and North America. It was estimated that agricultural P loss contributed 50% of the total P load to coastal waters in North America, Denmark, and Finland during 2000 (OECD, 2001). As shown in an assessment study at the scale of France, 46% of the total P input in watercourses came from diffuse agricultural sources between 2005 and 2009, versus 97% for nitrogen (Dupas et al., 2015a). Another study in Brittany, Western France, estimated that 70% of the phosphorus input in watercourses had an agricultural origin during the period 2007-2011 (Legeay et al., 2015). These proportions should be considered with caution due to the high uncertainty in flux data, but an average contribution of >50% of agricultural emissions to the total P input in watercourses is generally accepted in Western countries (Alexander et al., 2008; Dorioz et al., 2013).
Thus, research efforts and management options in many countries are currently focused on understanding and decreasing diffuse P losses from agricultural sources (Collins et al., 2014; Kleinman et al., 2007; Schoumans et al., 2014; Sharpley et al., 2013).
Knowledge and research trends on diffuse P loss processes
It is essential to limit diffuse P losses at all spatial scales, including the plot, field, catchment, regional and global scales. There are generally two strategies to study P transfer processes and dynamics in agricultural landscapes namely reductionist-based and complexity based approaches (Coveney and Highfield, 1995; Haygarth et al., 2005). The reductionist-based approaches are usually conducted at small spatial scales (plot, lysimeter, mesocosm batch) by inductive mechanistic, replicated hypothesis testing. The advantage of this approach is that it allows the isolation of P release and transfer mechanisms with high degree of precision, whereas it has the disadvantage of being far away from real field conditions (catchment, river or lake) and there are no actual means to connect them to the larger scale transfer processes and dynamics. In contrast, complexity-based approaches are usually conducted on large scales (catchment, river or lake) by empirical observation approaches. The advantage of these approaches is that the research sites cross the management scale and integrate the small-scale processes into natural, wider patterns. The disadvantage of these approaches lies in the difficulty in explaining the observed patterns and the great contingency and uncertainty (Haygarth et al., 2005).
Many studies have shown that the P loss mechanisms identified at the plot scale could not always cannot always produce an obvious signal when extrapolated to the scale of headwater catchments (Haygarth et al., 2012). In the past, most studies on agricultural P losses reduction have focused on batch experiments or rainfall simulations, which developed knowledge only on P release processes at the plot scale (millimeter to decameter), but not sufficient to study the catchment as a whole (1-100 km2). Thus, there is an essential need to integrate the two approaches for the development of the overall P transfer science.
Over the past two decades, there has been an increase in the number of catchment-based studies, especially at the small catchment scale (<10 km2) where the complexity and uncertainty regarding water pathways and dynamics are minimized (Aora et al., 2010; Dahlke et al., 2012; Dillon and Mollot, 1997; Haygarth et al., 2005; Heathwhite and Dils, 2000; Jordan-Meille and Dorioz, 2004; Mellander et al., 2012b, 2015, 2016; Moreau et al., 1998; Rodriguez-Blanco et al., 2013a, b; Siwek et al., 2013). The basic hypothesis of the catchment approach is that the P signal at the outlet of the catchment summarizes all the P mobilization and delivery processes that occur inside the catchment. As the one selected in this thesis, catchments in rural areas are preferred where point source P losses are minimized, with the purpose to focus on the diffuse P losses. As also shown in this thesis, the catchment approach does not exclude laboratory approaches at the small scale to fully assess the nature of P release mechanisms in agricultural soils.
Some key concepts about P mobilization processes in agricultural catchments
In agricultural landscape, it is generally accepted that not all catchment areas equally contribute to P losses. They occur predominantly from areas where high P concentrations and high transport capacities (water flow paths) coincide. These zones of coincidence are called critical source areas (Heathwaite et al., 2005b; McDowell et al., 2014; Shore et al., 2014). Although generally representing less than 20% of the total catchment area, these zones export most of the P to the catchment outlet (Pionke et al., 2000). Thus, in this critical-source-area concept, P losses may be managed by reducing the availability of P at the source level (e.g. decrease fertilizer P inputs) and/or by changing water flow pathways and trapping the exported P through the implementation of buffer zones at the interface between the hillslope and the stream.
Another concept to study the P transfer in agricultural landscapes is the “Phosphorus transfer continuum” which highlights the interdisciplinary and multi-scaled nature of the P transfer science (Haygarth et al., 2005). The P transfer continuum in agricultural landscapes assumes that the overall process can be decomposed into four steps, i.e. “sources”, “mobilization”, “deliver”, and “impact” (Figure 1.5).
The P “sources” can be of natural (indigenous soil-P mineralization and atmospheric deposition of P-rich micro-particles) or anthropogenic (fertilizers and manure applied to the soils) origins. Because the natural background concentrations of P in soils were generally insufficient to meet the increasing demand of farm yields, humans have applied fertilizers and imported animal feed to counterbalance the low natural P concentration in soils. This application and import are in many regions excessive, leading to the accumulation of P in soils, usually referred to as “legacy phosphorus” (Haygarth et al., 2014; Jarvie et al., 2013a; Sharpley et al., 2013). The “mobilization” describes the processes by which the P is separated from its sources in the soils, either by solubilization under the action of chemical/biochemical processes or by the physical detachment of P-rich particles. It has long been evidenced that the potential of P solubilization (indicated by the P concentrations in soil solutions) generally increases with the increasing concentration of extractable soil P (McDowell et al., 2001). Thus, agronomic P tests, which were originally designed to estimate the fertilization requirement of soils, are nowadays often used as environmental risk assessment test, as they have the capacity to evaluate the P solubilization potential in soils (Jordan-Meille et al., 2012; Wall et al., 2013). These soil P tests are even used as regulation tools in several European countries (Amery and Schoumans, 2014). The degree of P saturation (DPS) provides another tool to evaluate the risk of P solubilization in soils, which is currently used as a regulation and management tool in several countries (Khiari et al., 2000; Leinweber et al., 1999; McDowell et al., 2002; Schoumans and Chardon, 2015). The DPS can be calculated by adsorption isotherms or by chemical extraction of iron and aluminum oxides, which are considered as the two main P-sorbing phases in soils (Renneson et al., 2015). The physical detachment of soil P-rich particles is controlled by the capacity of soils to be eroded, either via surface runoff or through soil macropore water flows. The detachment capacity of soils can be estimated by structural stability.
Soil and water chemical analyses
Soil samples were air-dried, sieved to < 2 mm and analyzed for particle size fractions (NF X 31-107), organic matter/nitrogen/carbon contents (NF ISO 13878, NF ISO 10694), pH in water (1 :5 v :v water extraction NF ISO 10390) , extractable P (Dyer method, i.e. 1 :5 w/v extraction with citric acid 20 g l-1 NF X 31-160), total P (ICP-AES after total solubilization with hydrofluoric and perchloric acid NF X 31-147), Al and Fe (ICPAES after extraction with ammonium oxalate and oxalic acid, according to Tamm 1922) (Table 2.1). Equilibrium P concentration (EPCo) and maximum sorption capacity (Qmax) were estimated from 6-point batch isotherms (0, 0.1, 0.5, 50, 100, 200 mg P l-1; 1:25 w:v) in 0.01M CaCl2 according to Graetz and Nair (2000). One drop of chloroform was added to inhibit microbial activity. After 24 h equilibration at 20±2°C, samples were centrifuged (3000 rpm; 10 min), filtered (< 0.45 µm) and analyzed for MRP. Qmax was determined by fitting a Langmuir equation (Van der Zee and Bolt, 2001) to the last three points (50, 100, 200 mg P l-1): Q = (c * K * Qmax) / (1 + K * c).
where c is the concentration of P in the equilibrium solution (mg l-1), Q is the total amount of P sorbed (mg mg-1) and K is an a-nity parameter (l mg-1). EPCo represents the solution P concentration at which no net sorption or desorption of P would occur between soil and solution (Stutter and Lumsdon, 2008). EPCo was determined by fitting a linear equation to the first three points (0, 0.1, 0.5 mg P l-1). We consider EPCo as a reference MRP concentration in the soil solution, which we can compare to the actual MRP concentration of soil solution collected in-situ. Qmax served to calculate “Degree of P Saturation” (DPS), defined here as the ratio of Extractable P to Qmax. DPS is an index of P accumulation in the soil, either via direct application of fertilizers or enrichment via erosion (Schoumans and Chardon, 2015).
For each water sample collected in lysimeters or in the stream, MRP was determined colorimetrically by reaction with ammonium molybdate (ISO 15681). Because filtrates < 0.45 µm can contain colloidal forms of molybdate reactive phosphorus, we chose to use the term MRP rather than soluble reactive phosphorus (Haygarth and Sharpley, 2000). Precision of MRP measurement was ±4 µg l-1. Fe2+ was analyzed using the 1.10 phenantroline colorimetric method, according to AFNOR NF T90-017, with a precision of 5%. Nitrate concentration was measured by ionic chromatography (DIONEX DX 100), with a precision of 2.5%.
Table of contents :
Chapter 1- General Introduction
1.1 The global story of phosphorus
1.1.1 Why do we study phosphorus?
1.1.2 The global challenge of P resources
1.1.3 The P dilemma: too much vs. too little
1.2 P losses to waters: point vs. diffuse sources
1.3 Knowledge and research trends on diffuse P loss processes
1.3.1 The need for an integrated catchment approach
1.3.2 Some key concepts about P mobilization processes in agricultural catchments
1.3.3 Are dissolved and colloidal P species major components of diffuse P losses in agricultural landscapes?
1.3.4 Riparian buffer zones as potential sources of dissolved and colloidal P agricultural catchments
1.4 The long-term Kervidy-Naizin observatory: an ideal site for unravelling dissolved and colloidal phosphorus losses in agricultural catchments
1.5 General objectives and organization of the thesis
Chapter 2 Groundwater control of biogeochemical processes causing phosphorus release from riparian wetlands
2.2 Materials and Methods
2.2.1 Study sites
2.2.2 Soil and water sampling
2.2.3 Soil and water chemical analyses
2.3 Results and discussion
2.3.1 Soil P content and water table depths in riparian wetlands
2.3.2 Groundwater level controls P release in riparian wetlands
2.3.3 Linking MRP concentration in riparian wetlands and in the stream
2.5 Supplementary materials
2.6 Conclusion of chapter 2
Chapter 3 Release of dissolved phosphorus from riparian wetlands: Evidence for complex interactions among hydroclimate variability, topography and soil properties
3.2 Materials and Methods
3.2.1 Research site
3.2.2 Soil and water sampling
3.2.3 Soil and water chemical analysis
3.3.1 Hedley P fractionation
3.3.2 Rainfall, discharge and water-table variations
3.3.3 Soil water chemistry
3.3.4 Stream water chemistry
3.4.1 Influence of soil P content and soil P speciation
3.4.2 Key influence of interannual hydroclimatic variability on P release dynamics
3.4.3 Topography as the potential ultimate driver of dissolved P release in RW soils
3.4.4 Delivery and retention of mobilized P
3.6 Supplementary materials
3.7 Conclusion of chapter 3
Chapter 4 Drying/rewetting cycles stimulate release of colloidal-bound phosphorus in riparian soils
4.2 Materials and Methods
4.2.1 Soil properties and preparation
4.2.2 Experimental setup and conduct of DRW experiments
4.2.3 Leachate treatments
4.2.4 Chemical analysis
4.2.5 UF data treatment
4.2.6 Statistical analysis
4.3.1 P and DOC concentrations in RF samples
4.3.2 UF leachate results
4.4.1 Soil rewetting stimulates release of colloidal P
4.4.2 Co-existence of physically- and biologically-driven P release during rewetting
4.4.3 Influence of soil properties
4.4.4 Linking sources and production mechanisms of P forms released during soil rewetting
4.4.5 Environmental and ecological implications
4.6 Supplementary materials
4.7 Conclusion of chapter 4
Chapter 5 Release of dissolved phosphorus upon reduction of wetland soils: a laboratory study of the respective roles of soil Fe-oxyhydroxides dissolution, pH changes, sediment inputs and soil phosphorus speciation
5.2 Materials and methods
5.2.1 Sampling sites and soil preparation
5.2.2 Experimental setup
5.3.1 Soil/sediment composition
5.3.2 Anaerobic incubations of RW soils
5.3.3 Aerobic incubations of RW soils
5.3.4 Anaerobic incubations of sediment with and without RW soil addition
5.4.1 Controls of soil properties on concentration and speciation of released DP
5.4.2 Assessing the respective roles of reductive dissolution of Fe-oxyhydroxides and pH rise
5.4.3 Influence of sediment deposition in RWs on DP release under anoxic conditions
5.6 Supplementary materials
5.7 Conclusion of chapter 5 (laboratory simulation of reduction processes)
Chapter 6 General conclusions
6.1 Recall of thesis objectives
6.2 Summary of conclusions
6.2.1 Constraints from field monitoring on the mechanisms and factors causing DP releases in riparian wetlands
6.2.2 Highlighting how drying-wetting cycles stimulate the release of colloidal P in wetland using column leaching experiments
6.2.3 Constraints on the processes releasing P under anoxic conditions
6.3 Possible implications for management
6.4.1 Nature, source and significance of organic P fraction
6.4.2 Towards a better characterization of colloid composition and colloid properties regarding P transfer in soils
6.4.3 Test the generality of the conceptual model developed from the Kervidy- Naizin catchment
Chapter 7 General references