Release and transport dynamics of colloidal phosphorus in a small agricultural catchment
Abstract
Colloids (11000 nm) have been widely suggested as important carriers of phosphorus (P) in agricultural soils.
This suggestion is, however, so far based on colloids extracted from laboratory soil solutions, with no direct field
evidence. Here, we conducted the first field monitoring study of colloidal P (Pcoll) release and transport dynamics
in a small agricultural headwater catchment located in Western France, during one hydrological year. Results
evidenced that Pcoll is an important fraction of the total dissolved P (TP) fraction (<0.45micron) in both soil and
stream waters from this catchment (mean=39%). However, a decrease in the proportion of Pcoll was observed
along the soil-stream continuum, suggesting either retention of part of the released Pcoll during transport or a dilution
by deep groundwater with low Pcoll concentrations. The proportion and composition of Pcoll were studied
in detail along two soil transects (K and G), and in the stream immediately adjacent to them. A spatial consistency
was found in the proportion of Pcoll which were comparatively higher in soil and steam waters from transect K
(mean=52% and 32%, respectively), as compared to transect G (mean=42% and 19%, respectively). A detailed
survey of the Pcoll composition revealed a major difference between soil waters from transects G and K, the colloids
being equally composed of colloidal-MRP (molybdate-reactive P) and colloidal-MUP (molybdate-unreactive
P) in transect G, as against mainly in the form of colloidal-MUP (mean=70%) in transect K soil waters. However,
this difference was not observed in corresponding stream waters, in which Pcoll was mainly in the form
of colloidal-MUP (mean=75%), thereby suggesting the likely preferential retention of MRP-rich colloids at the
soil-stream interface. Concentrations of Pcoll were relatively stable throughout the hydrological year in transect
G soil waters, except as regards a peak occurring in phase with a Fe(II) release peak. Interestingly, this peak was
not transferred to the stream, suggesting again the occurrence of a retention process at the soil-stream interface. In
samples from transect K, concentrations of Pcoll were more variable in both soil and stream waters, which could be
due in part to higher analytical uncertainties as the concentrations of TP were very low in most samples from this
transect. Considering all soil and stream water samples, a negative correlation was found between the proportion
of Pcoll in TP and the P:Fe ratios in <0.45micron filtrates (r=0.497). Concentrations of Pcoll were also found to
correlate with total anion (r=0.627) and DOC concentrations (r=0.859), suggesting a potential influence of the soil
properties and local groundwater chemistry on Pcoll concentrations. Overall, this study confirms that Pcoll is an
important fraction of the TP released in agricultural soils and that this fraction can be efficiently transported to
streams by shallow groundwater flow. It also indicates that the composition of the released Pcoll may vary in space
and time and that retention processes probably occur at the soil-stream interface with the potential of modifying
the composition of the Pcoll.