Release of soil phosphorus during runoff as affected by ionic strength and temperature

Dissolved reactive phosphorus (DRP) from two cultivated clay soil samples (Vertic Cambisols) was extracted under conditions simulating the variation in the properties of surface runoff water in the field. DRP was extracted at three temperatures (5, 15 and 25°C), and at different ionic strengths by using deionized water and CaCl2 solutions (0.00005-0.005 M) as extractants. The solution-to-soil ratio varied from 50 to 2000 1 kg 1 . Sorption to and desorption from the soils were studied at different temperatures and ionic strengths by determining quantity-intensity (Q/I) plots at the solution-to-soil ratio of 50 1kg 1 , and the results were fitted to a modified Langmuir equation: Q = Q maxl/( 1/K + I)-Qo where Qis P sorbed or desorbed, Qmax =maximum P sorption, I= P concentration in the equilibrium solution, K = sorption/desorption equilibrium constant, and Q 0 = instantly labile P. The desorption of DRP was depressed by increases in the CaCl2 concentration of the extractant and promoted by widening of the solution-to-soil ratio. At the solution-to-soil ratio of 50 1 kg ', the increase in the temperature from 5 to 25°C raised the DRP release to water from 12.6 to 20.7 mg kg* 1 in the Aurajoki soil and from 1.8 to 3.4 mg kg ' in the Jokioinen soil. In the Aurajoki soil, the constant Q 0 of the Langmuir equation responded to the changes of ionic strength and temperature in the same way as did DRP extracted at wide solution-to-soil ratios. However, the Prelease capacity of both soils was underestimated by the constant Q O.

Introduction and in the particulate material. In areas of frozen soil and snow coverage in winter, the volume of surface runoff water peaks in spring during the thaw, another peak often occurring in autumn. Erosion and removal of dissolved P are Surface runoff water transports phosphorus (P) from soil to watercourses in the dissolved form also most intensive during these periods (Turtola and Jaakkola 1995). The equilibrium P concentration (EPC) in the solution where the particulate material collected from river water neither released or adsorbed P was shown to be lower than 0.010 mg P f which is decisively below the EPC obtained for cultivated soils (e.g. Hartikainen 1982b, Yli-Halla 1991, Yli-Halla et al. 1995. Pietiläinen and Ekholm (1992) showed that 90% of particulate material in a small agriculturally loaded river in southern Finland was recently eroded from the surface soil of the fields of the drainage basin. Comparison of the results above thus reveals that P must be effectively desorbed during the erosion process.
Dissolved P in the runoff water originates partly from the bulk of the soil remaining in the field and partly from suspended particles. Soil P status, to some extent, explains the average dissolved reactive P (DRP) concentration in runoff water but not the temporal variation of DRP (Yli-Halla et al. 1995). The temperature during the runoffpeaks in autumn and particularly in spring is much lower (between 0 and 5°C) as compared to temperatures prevailing during the occasional summer rainstorms (commonly around 15°C) and to room temperatures at which laboratory experiments are usually done. There is also a marked seasonal variation in other external conditions (ionic strength and volume of runoff water) prevailing during runoff events.
In this study, a set of desorption tests was carried out to quantify the impact of environmental factors on the P loading risk due to the surface runoff from cultivated soils. The temperature as well as ionic strength and solutionto-soil ratio were varied to simulate their changes during the runoff and erosion process and material transport in watercourses. The impact oftemperature and external ionic strength on the dynamic equilibrium between solution and solid material was investigated by means of the quantity-intensity (Q/I) plots. The instantly labile P derived from these graphs was used as one estimate for the P loading risk due to runoff water. The suitability of the Q/I plots to predict P release from soil was also evaluated.  (Vuorinen and Mäkitie 1955) Chang and Jackson fractions of inorganic P (Hartikainen 1979) Material and methods

Soil samples
The soil samples ofAurajoki and Jokioinen (Table 1) were taken from experimental fields set up for studies on surface runoff. They represent fields with high and low level of P concentration in runoff, respectively (Yli-Halla et al. 1995). Both soils, located in southwestern Finland, are classified as Vertic Cambisols (FAO 1988). The soil samples were taken from the 0-10 cm layer. After air-drying and homogenization, the samples were rewetted to a moisture content of 20% and stored at 5°C for several weeks before the analyses. Air-dry samples were analyzed for pH in a 0.01 M CaCl 2 suspension and for water-extractable P(lg of soil, 50 ml of deionized water, 17 hours of equilibration; Harlikainen 1982 a) 194 AGRICULTURAL AND FOOD SCIENCE IN FINLAND Vol. 5 (1996): 193-202.

Desorption tests
To study the effect of ionic strength on P desorption from soil, deionized water, 0.0005 M and 0. M CaCl 2 were used as extracting solutions at room temperature (25°C). The selection of CaCl 2 is based on the dominance of Ca among the exchangeable cations of the experimental soils (results not presented). For the extractions, moist soil samples (four replicates) were weighed to give dry soil concentrations of 0.5, 1,2, 5 and 20 g I' 1 of the extractant (solution-tosoil ratio 2000-50 1 kg '). The soil suspensions were shaken for 17 h in an orbital shaker at a speed of 250 rotations min According to preliminary experiments, this reaction time was well sufficient to reach a semi equilibrium. The supernatant solutions were filtered through a membrane filter (0.2 pm, Nuclepore polycarbonate) and analyzed for dissolved reactive P (DRP) by a molybdenum blue method using ascorbic acid as the reducing agent. The effect of temperature on P desorption was studied by extracting soil with deionized water at 5, 15 and 25°C. At each temperature, the extractions were carried out at solution-to-soilratios from 50 to 2000 I kg 1 . The soil samples and the solutions to be added were adapted to the respective temperatures before the extraction. Phosphorus was determined as mentioned above. and 0.00005 M CaCl 2 as the supporting electrolytes. The ionic strengths of soil extracts were estimated from the electrical conductivity according to Griffin and Jurinak (1973).
Sorption to or desorption from soil (Y) in mg P kg' 1 was calculated from the changes in P concentration of the contacting solution and fitted to a modified Langmuir equation (Hartikainen and Simojoki 1994): where Qis P desorbed or sorbed, Q = maxi-'-max mum P sorption, I = P concentration in the final equilibrium solution, K = sorption/desorption equilibrium constant, and Q 0 = instantly labile P. Mathematically, the isotherm will intersect the y-axis when 1 = 0. According to Beckett and White (1964), the intercept (term Q 0 in the equation) represents what was termed instantly labile P that would have to be removed from the soil to reduce I to zero at a given solution-tosoil ratio. The intersecting point of the graph on the x-axis (Y = 0), the equilibrium phosphate concentration (EPC), represents the zero point of P exchange at which no net desorption from or sorption to soil occurs. The slope of the sorption-desorption curve at the EPC was referred to by Holford and Mattingly (1976) as the equilibrium buffer capacity (EBC).

Q/l plots
The Q/I plots were applied to express the sorption or desorption as a function of the P concentration in the equilibrium solution. They were determined in three replicates by adding 50 ml of KH,P0 4 solution (0-4 mg PT 1 for Aurajoki samples, 0-5 mg P I' 1 for Jokioinen samples) to moist samples corresponding to 1 g of dry soil (solution-to-soil ratio 50 1 kg' 1 , or concentration of suspended solids 20 g I' 1 ). The extracts were obtained as described above and analyzed for DRP. The Q/I plots were determinedat three temperatures: 5, 15 and 25°C. At 25°C, the plots were also determined using 0.005 M, 0.0005 M

Desorption tests
At every solution-to-soil ratio, the two CaCl 2 solutions extracted much less P than did water ( Fig. 1). At the solution-to-soil ratio of 50 1 kg 1 the P concentration in the extract was lowered from 0.42 to 0.14 mg 1 1 in the Aurajoki soil and in the Jokioinen soil from 0.068 to 0.015 mg 1 1 when the ionic strength of the soil extract, shown in Increasing the volume of water around the soil particles lowered the ionic strength and the DRP concentration in the extract (Table 3). Owing to the strong P buffer power of the soil, however, the decrease of DRP concentration was even less linear than that of the ionic strength.
Consequently, the desorption of P, expressed as mg kg -1 , was strongly promoted (Fig. 2). Desorption increased from 20.7 to 119 mg kg' 1 in the Aurajoki soil and from 3.4 to 45.0 mg kg' 1 in the Jokioinen soil at 25°C when the solution-tosoil ratio increased from 50 to 2000 1 kg 1 .

Q/l plots
The Q/I plots crossed from net desorption to net sorption, and the results conformed accurately to the modified Langmuir equation (r 2 >0.99) ( Fig. 3 and Fig. 4). However, for Jokioinen soil, the graphs intersected the x-axis close to the origin and the desorption remained very small at the solution-to-soil ratio of 50 1 kg' 1 at which the Q/I plots were determined. In both soils, sorption increased and desorption decreased when CaCl 2 solutions were used as extractants (Fig. 3).
The ÉPC values obtained in 0.005 M CaCl 2 were less than one fifth of that measured with out a supporting electrolyte (i.e., in water), and the EBC increased substantially upon increase of the ionic strength (Table 4). Despite the similar level of ionic strengths (0.3 mmol 1 1 in water extracts, 0.5 mmol I' 1 in 0.00005 M CaCl 2 extracts) in the Jokioinen soil, the Q/I plot in the CaCl 2 solution was markedly steeper. The Q/I plots determined at 5, 15 and 25°C (Fig. 4) showed that both desorption and sorption of P were promoted by gradual elevation of temperature. Both EBC and EPC increased upon increase in temperature (Table 4). The plots intersected at 1.2 and 0.2 mg P I ' in the Aurajoki and Jokioinen soil, respectively, i.e. clearly above the respective EPC values (see Table 4). In Jokioinen soil, dominated by a marked sorption tendency, the effect of temperature on the P exchange was small at the low P concentrations in the equilibrium solution. Therefore the graphs  Table 4.  are very close to each other and the cross-over points cannot be clearly seen in the scale used in Fig. 4.
In the Aurajoki soil, the constant Q 0 increased upon elevations of temperature and in general decreased upon increasing ionic strength (Table   4). The changes of Q 0 were thus in accordance with the influence of temperature and ionic strength on desorption obtained in the desorption tests. In the Jokioinen soil, on the contrary, the response of Q 0 to the changes of temperature or ionic strength was less consistent. In both soils, the values of Q 0 of the Q/I plots determined without a supporting electrolyte were less than half of the observed desorption to water at the widest solution-to-soil ratio. The values of Q 0 determined in CaCl 2 were 56-94% of the measured maximum desorption to the respective CaCl 2 solution (0.005 M or 0.0005 M; 0.00005 M not used in the desorption test).

Discussion
The quantity of labile adsorbed P on soil particles is the ultimate reserve of P which can be desorbed, but the DRP concentration in runoff water is also controlled substantially by ionic strength and temperature, and to some extent by the solution-to-soil ratio. The P buffer power of soil tends to maintain a constant DRP concentration in water, and therefore more voluminous runoff markedly increases the total quantity of DRP removed from the field. The ionic strength of the water extracts at the solution-to-soil ratio 50 1 kg' 1 corresponded to that of rain and snowmelt water (0.3 mmol I 1) while those of the extracts obtained at wider solution-to-soil ratios were even lower in salts. As for ionic strength, the soil extracts obtained with 0.0005 M CaCl 2 were similar to the surface runoff waters of the Jokioinen and Aurajoki fields (Yli-Halla et al. 1995), and those obtained with 0.005 M CaCl 2 corresponded to soil solution (Wiklander and Andersson 1974). As for the solution-to-soil ratio, it was observed in an earlier study that the average DRP concentration in the surface runoff in these particular fields was the same as that of soil extracts obtained at the solution-to-soil ratio between 250 and 500 1 kg 1 (Yli-Halla et al. 1995) but according to Ekholm (1994) the concentration of suspended solids in coastal river waters of Finland is lower than was applied in the present study. The equations in Fig. 1 describing the dependence of P desorption on ionic strength and solution-to-soil ratio thus cover the range of these factors occurring in the hydrologic environment in the field.
On the ionic strength scale used, the decrease in the salt concentration proved to effectively promote the P release from soil. In the water extracts, all dissolved salts originated from the soil sample, resulting in the decrease of the ionic strength of the extracts upon widening the solution-to-soil ratio. At a constant solution-tosoil ratio, decreasing ionic strength promotes P desorption (Hartikainen and Yli-Halla 1982). Therefore, it can be concluded that upon widening the solution-to-soil ratio the decrease of DRP concentration (mg I 1) may have been more substantial and the increase of P desorption (mg kg ') less marked if the ionic strength had been kept constant. The present results thus give the net effect of two factors promoting P desorption; widening solution-to-soil ratio and decreasing ionic strength.
Besides DRP released in the field, surface runoff water transports P which is adsorbed onto the suspended soil material and which can be released as DRP in the recipient watercourse. As for the total DRP loading from the eroded material, the temperature during the runoff event may be unimportant because in the water body, the eroded material is subject also to higher temperatures during the summer months. Therefore, results obtained at a low temperature are needed to assess the DRP release in the field during the cool and wet season, while those obtained at higher temperatures are applicable to DRP release by the summer rains and to desorption taking place in a watercourse during the warmer period of the year.
Increased speed of diffusion at elevated temperatures explains the cross-over of the Q/I plots.
Below the EPC, net diffusion occurs from soil to solution, resulting in a higher P concentration in the extract when the temperature is elevated. Above the EPC, net diffusion is from solution to soil, and an elevated temperature leads to a higher sorption. If diffusion were the only factor affected by the temperature, the curves should cross at the EPC. However, in both soils, elevation of the temperature seemed to shift the EPC to a higher concentration. At higher temperatures, a higher P concentration was required for sorption to start, or, vice versa, desorption continued to a higher P concentration. As a consequence, the crossing of the curves occurred at a P concentration above the EPC. At the EPC the net diffusion is zero. Based on the shift of the EPC, conclusions can be made on the temperature-dependency of the P exchange equilibrium. The shift of EPC to a higher concentration indicates that a high temperature favors desorption. This suggests sorption to be an exothermic reaction, as presented by Barrow (1979), and desorption to be an endothermic one. The parameter Q stands for P sorption sites available. Its increase as a response to the elevated temperature can be taken to indicate that the saturation of the sorption sites is kinetically controlled.
The Q/I plots can in principle be utilized to quantify the instantly labile P of the soil (Pionke and Kunishi 1992). The physical relevance of the constant Q(| as a measure for instantly labile P can be assessed by comparing it with the observed desorption at wide solution-to-soil ratios, e.g. at 2000 1 kg' 1 and with other estimates of P release from soil. Phosphorus bound to hydrous oxides of Al and Fe are the major reserves of bioavailable P in the watercourse (Dorich et al. 1985), and they control the level of water-extractable P in soil (Hartikainen 1982 a). Maximum desorption in this study (solution-to-soil 200 AGRICULTURAL AND FOOD SCIENCE IN FINLAND Vol. 5 (1996): 193-202. ratio 2000 1 kg ', 25°C) corresponded to 9 and 18% of the secondary P fractions (NH 4 CI-P + NH 4 F-P + NaOH-P, see Table 1) in the Jokioinen and Aurajoki soils, respectively, while the constant Q 0 of the respective Q/I plots amounted to only 3 and 8% of the secondary P reserves.
In an exhaustive pot experiment with one soil, Yli-Halla and Renlund (1990) measured a 30% decrease in these P fractions. If this decrease is taken to represent a measure of the maximum bioavaliability of soil P reserves, the Q 0 values markedly underestimate the potential P loading. Even the desorption measured at the widest solution-to-soil ratio (2000 1 kg 1 ) at 25°C may be smaller than the P amount that can be released from eroded soil in a watercourse. However, it should be mentioned that, particularly in the Aurajoki soil, Q 0 and the P release in the desorption tests responded similarly to changes of ionic strength and temperature. This shows that the Q/I plots qualitatively reflect the dynamic P exchange even though quantitative interpretation of the Q 0 values may be questionable.
The present results show that depending on prevailing experimental conditions, a wide variation of P desorption results can be measured in a given soil, leading to different estimates for P loading. In most studies on soil samples, the Q/I plots have been determined using a 0.01 M supporting electrolyte (e.g. Barrow 1979). The information obtained from those Q/I plots is applicable to P fertilization and the nutrition of plants. Phosphorus release to surface runoff or water bodies needs to be assessed by experiments performed at a low ionic strength and a wide (above 200 1 kg ') solution-to-soil ratio.