DIVISION S-4SOIL FERTILITY & PLANT NUTRITION

Ammonium Adsorption and Desorption in Sandy Soils

F. L. Wang and A. K. Alva* ABSTRACT
Leaching of fertilizer N in sandy soils is both an agricultural and environmental concern and depends, in part, on the N holding capacity of the soils in the vadose zone. We examined NH4 adsorption and desorption in surface (030 cm) and subsurface (3060 cm) samples of Wabasso (sandy, siliceous, hyperthermic Alfic Alaquod) and Candler (uncoated hyperthermic, Typic Quartzipsamment) sands using a batch technique. Samples of a 90- to 100-cm depth horizon of the Wabasso sand were also used in the study. The NH4 quantityintensity (Q/I) relations showed that the potential buffering capacity (PBC) of the soils ranged from 0.26 (Wabasso, 3060 cm depth) to 3.9 (Wabasso, 90100 cm depth) cmolc kg1 M1/2. Labile NH4, as determined from the Q/I, was 4.9 103 (Candler, subsurface) to 13.8 103 (Wabasso, surface) cmolc kg1. Positive linear relationships were observed between organic C content and Q/I plot parameters (potential NH4 buffering capacity and labile NH4) of all soil samples except in the Wabasso 90- to 100-cm depth horizon. Although NH4 adsorption capacity of the surface soils was greater than that of the subsurface soils, desorption was greater from the former soils than that from the latter. This study clearly demonstrated that the potential NH4 buffering capacity and labile NH4 for the sandy soils studied were much lower than those for clay and silt loam soils.

Odu, 1978; Rappaport and Axley, 1984; Evangelou and Blevins, 1985; Thompson and Blackmer, 1992; Lumbanraja and Evangelou, 1990, 1994). Little is known about adsorption and desorption of NH4 in sandy soils. Behavior of NH4 ions in sandy soils may differ significantly from that in clay or clay loam soils. Further our previous study using Florida sandy soils (Wang and Alva, 1996) showed a greater proportion of applied fertilizer N being retained by an Alfic Haplaquod than by a Typic Quartzipsamment. However, the mechanism for the observed differential NH4 retention was not clear. The objective of this study was to evaluate adsorption and desorption of NH4 in different horizons of sandy soils. Particular attention was given to NH4 Q/I relations and the relationship between selected soil chemical properties and Q/I plot components. MATERIALS AND METHODS
Candler fine sand and Wabasso sand were sampled from citrus groves in Polk and St. Lucie Counties, Florida, respectively. These soil series represent the soils in the citrus production regions along the Central and East Coast areas of Florida. The soils were sampled from 0- to 30-cm (surface) and 30- to 60- (subsurface) cm depths for both soils , and from a 90- to 100-cm depth horizon of the Wabasso sand. Our x-ray analysis (Whitting and Allardice, 1986) showed that the clay fraction of soil samples from every soil depth studied contained quartz and kaolinite. In addition, there was hydroxyinterlayered vermiculite in the Candler sand and smectite in the Wabasso sand. Selected properties of the soils are presented in Table 1. The approach of Beckett (1964a, 1964b) was used to study Q/I relations for NH4 in the soils. In Fig. 1, an ideal Q/I plot for NH4 is given, which includes the amount of NH4 adsorbed or released from the soil (NH4, cmolc kg1) and the NH4 activity ratio (ARNH4) expressed by

ost agricultural soils in Florida are sandy and are under intensive N fertilization programs. There is a potential for some fertilizer NO3 to enter the surficial aquifer. The use of NH4releasing fertilizers may be preferred because NH4 is less mobile than NO3. To improve the efficiency of N fertilizer application in the region, it is essential to understand fate of NH4 in these soils. In soil, N availability for plant uptake and leaching processes is related to NH4 adsorption and desorption capacity of the soil (Avnimelech and Laher, 1977; Fenn et al., 1982). The adsorption capacity of clay minerals appears to govern the rate of biologically mediated NH4 oxidation (Goldberg and Gainey, 1955; Kai and Harada, 1969). The nature of soil minerals also modifies the desorption capacity of the soil. For instance, fixation of NH4 can be partly attributed to the presence of vermiculite in the soil (Bohn et al., 1985). Similar results of NH4 exchange reactions have been demonstrated by several investigations conducted on clay and silt loam soils (Pasricha, 1976; Pasricha and Singh, 1977; Opuwaribo and
F.L. Wang, Dep. of Soil Science, Univ. of Saskatchewan, Saskatoon, SK Canada S7N 5A8; A.K. Alva, USDA-ARS-PWA, 24106 N. Bunn Rd., Prosser, WA 99350. Received 12 Apr. 1999. *Corresponding author (aalva@tricity.wsu.edu). Published in Soil Sci. Soc. Am. J. 64:16691674 (2000).

ARNH4 (NH4)/[(Ca) (Mg)] 1/2

[1]

where parentheses denote activities of the ions in the solution; a potential buffering capacity for NH4 (PBC, cmolc kg1M1/2), represented by the slope of the linear part of the Q/I plot; o , cmolc kg1) obtained by extrapolating the labile NH4 (NH4 the linear part of the Q/I plot to ARNH4 0; NH4 availability o (ARNH4 , M1/2) when NH4 0; and specific adsorption sites (NH4sas, cmolc kg1). Theoretical considerations with respect to Q/I relationships for cations in soils have been discussed elsewhere (Beckett, 1964a, 1964b; Talibudeen, 1981; Lumbanraja and Evangelou, 1990). Equilibrating solutions were prepared containing 5
Abbreviations: PBC, potential buffering capacity; Q/I, quantity intensity relationship.

1669

1670

SOIL SCI. SOC. AM. J., VOL. 64, SEPTEMBEROCTOBER 2000

103 M CaCl2 and various concentrations of NH4Cl (0 to 2 102 M ). Subsamples (5 g, in triplicate) were taken from soil samples of respective depths and placed into 100-mL polystyrene tubes, and 25 mL of equilibrating solution was added to each tube. The samples were shaken for 6 h at 90 strokes min1 and at 25 1C. A preliminary experiment did not show significant difference in NH4 adsorption as the samples were shaken for 2, 4, 6, 12, or 24 h. After shaking, the suspension was centrifuged and the supernatant filtered (Q2 filter paper, Fisher Scientific, Pittsburgh, PA). The filtered solution was analyzed for Ca and Mg ions, using an inductively coupled plasma emission spectroscope (Plasma 40, Perkin-Elmer Inc., Norwalk, CT), and for NH4 ion using a rapid flow analyzer (ALPKEM, 1989). The electrical conductivity (EC) of the filtered solution was determined and ionic strength (IS) calculated using the relationship: IS (M ) 0.013 EC (dS m1) (Griffin and Jurinak, 1973; Alva et al., 1991). Ion activities and activity ratio, ARNH4, were calculated from the Davies equation using ionic strength and ion concentration data (Bohn et al., 1985). The change in exchangeable NH4 (NH4) was taken as the difference between the initial concentration of exchangeable NH4 and that at equilibrium. Quantityintensity plots were constructed using NH4 and ARNH4 values, and regression analysis was applied to the linear portion of the Q/I plots. o , and Quantityintensity plot parameters (PBC, NH4 o ARNH4 ) were derived from these regression equations. The specific adsorption sites (NH4sas) of the soil were taken as the difference between NH4o and the Q/I plot value obtained at ARNH4 0 by extrapolation. Fixation capacity of the soil to added NH4 was quantified in terms of NH4 desorption percentage, which was calculated as follows:

Fig. 1. The ideal quantityintensity plot for NH4, where NH4 the amount of NH4 adsorbed or released from the soil; ARNH4 activity ratio, i.e. (NH4)/(CaMg)1/2 in solution; ARo NH4 equilibrium activity ratio; PBC linear potential buffering capacity for NH4; o NH4 labile NH4, and NH4sas specific adsorption sites.

than 99% of the extractable NH4 was removed in the first KCl treatment. All results are presented as means of triplicate treatments. Statistical analysis (ANOVA and LSD) on the difference between the means of the soils was conducted using a Super ANOVA program (Abacus Concepts, Berkeley, CA). In the regression analysis, regression coefficients were calculated and their significance was tested through general statistical procedures.

Percentage NH4 desorption [(amount of NH4 desorbed in 2 M KCl)/ (amount of NH4 adsorbed)]100 [2]

RESULTS AND DISCUSSION

Except for the 30- to 60-cm depth in Candler soil, NH4 Q/I relations for the two soils (Fig. 2A and 2B) showed curvilinearity at low activity ratios, suggesting a release of nonexchangeable (or specifically adsorbed) NH4 (Beckett, 1964b). The amount of nonexchangeable NH4 in the soils was estimated from NH4sas values (Table 2). Although all the soil depths studied showed the presence of specific NH4 adsorption sites, the fixation capacity of these sites (0.00270.0096 cmolc kg1) was much smaller than those (1.071.62 cmolc kg1) reported for clay loam soils (Thompson and Blackmer, 1992). The PBC for NH4 and the quantity of labile NH4 in the soils (Table 2) were also lower than those for soils containing higher silt and clay. A PBC value of 21.82 cmolc kg1M1/2 was reported for NH4 in a silt loam soil (Lumbanraja and Evangelou, 1990). In clay loam soils,

The desorption of NH4 was studied for selected treatments, that is, following the NH4 adsorption using 1 103, 5 103, and 1 102 M NH4Cl equilibrating solutions. After the adsorption reactions, the supernatant was separated from the soil suspension by centrifugation. The sediment was centrifuge-washed using 20 mL of 0.5 (m3 m3) methanol. The washing procedure was repeated until the supernatant was free of Cl based on the test using AgNO3. After final washing, 25 mL of 2 M KCl was added to the sediment. The suspension was shaken once more for 6 h at 90 strokes min1 at 25 1C. A preliminary experiment did not show any significant difference in NH4 desorption whether samples were shaken for 6, 12, or 24 h. Upon completion of NH4 desorption, the suspension was centrifuged, the supernatant filtered and the filtrate analyzed for NH4 as discussed above. The results showed that more
Table 1. Selected chemical properties of the soils studied.
Particle-size distribution Soil series Wabasso 030 3060 90100 Candler 030 3060 968 972 16 17 16 11 4.96 4.19 928 980 946 60 16 30 12 4 24 5.07 5.22 5.01 Depth cm Sand g Silt kg1 Clay pH

Exchangeable EC dS m 1 8.2 1.0 12.1 4.6 0.8 Organic C g kg1 0.48 0.15 0.45 0.39 0.15 5.16 0.05 2.58 1.29 0.03 Total N NH 4 mg kg1 22.1 22.1 13.3 13.3 22.1
NO3

0.49 0.33 0.82 2.03 0.49

pH (soil/water ratio 1:2); electrical conductivity, EC, (soil/water ratio 1:2); organic C was determined by a C analyzer; total N (Bremner and and NO by 2 M KCl. Mulvaney, 1982); exchangeable NH4 3

WANG & ALVA: AMMONIUM ADSORPTION AND DESORPTION IN SANDY SOILS

1671

Fig. 2. Quantityintensity plots for (A) surface and subsurface soils of the Wabasso and Candler sands and (B) the 90- to 100-cm depth horizon of the Wabasso sand. NH4 is the amount of NH4 adsorbed or released from the soil and ARNH4 is the activity ratio, i.e., (NH4)/ (CaMg)1/2 in solution.

the PBC was reported to range from 63 to 115 cmolc kg1M1/2 (Lumbanraja and Evangelou, 1990; Thompo son and Blackmer, 1992) and NH4 values from 0.053 1 to 0.144 cmolc kg (Thompson and Blackmer, 1992). Hence, the difference in Q/I plot parameters between the sandy soils examined in our study (Table 2) and the silt and clay loam soils reported by Thompson and Blackmer (1992) was approximately one to three orders of magnitude. On the other hand, ARNH4o values for the

sandy soils (Table 2) were up to 10-fold greater than o that for the clay loam soils. A high ARNH4 value suggests high availability of NH4 at equilibrium. However, low o values of NH4sas, PBC, and NH4 are indicative of poor NH4 holding capacities for these soils. The Q/I plot parameters varied between the soils and among soil depths. Among the surface samples, the Wabasso had a greater potential NH4 buffering capacity than the Candler sand. However, there was no significant difference between the two soil series with regard to PBC of the subsurface samples (Table 2). The labile NH4 pool decreased in the order: Wabasso, surface Candler, surface Wabasso, 90- to 100-cm depth horizon Wabasso, subsurface Candler, subsurface horizon. It appeared that the 90- to 100-cm depth horizon was unique with respect to its NH4 exchange reactions. Soil samples of this horizon showed greater PBC and NH4sas values than any of the other samples. Furthermore, curvilinearity of the Q/I plot for the 90- to 100-cm depth horizon was also observed at high activity ratios (ARNH4 values 0.020.16) (Fig. 2). This suggests that, compared with the other samples, the NH4 adsorption capacity of the Wabasso 90- to 100-cm depth horizon is more sensitive to the extent of surface coverage by NH4; that is, NH4 adsorption is expected to decrease at a faster rate for the 90- to 100-cm depth horizon than for the other horizons as adsorption sites become occupied by NH4. The 90- to 100-cm depth horizon and the Wabasso surface soil, which also had a substantial NH4 buffering capacity and labile pool, appeared to contribute to the greater retention of NH4 by the Wabasso sand compared with the Candler sand in our previous N leaching study (Wang and Alva, 1996). The increased NH4 retention by the Wabasso 90- to 100-cm depth horizon may not be exclusively due to its high organic C content, although potential NH4 buffering capacity and labile NH4 values tend to be greater for soils with higher organic C contents (Tables 1 and 2). The Wabasso 90- to 100-cm depth horizon sample, which had the highest organic C content among the soil depths studied, was an outlier in the linear relationship between organic C and Q/I plot components (PBC and o ). The correlation was established by excluding NH4 the data from the 90- to 100-cm depth horizon (Fig. 3A and 3B). The relatively high potential NH4 buffering capacity of the Wabasso 90- to 100-cm depth horizon cannot be attributed to its difference in organic C content only. The PBC of the Wabasso 90- to 100-cm depth

Table 2. Equations for the linear portion of quantityintensity (Q/I) plots and Q/I plot parameters for the soils.
Soil series Wabasso Depth cm 030 3060 90100 030 3060 y y y y y Equation 0.8843x 0.0138 0.2643x 0.0065 3.922x 0.00796 0.7208x 0.0123 0.3835x 0.0049 R2 0.999 0.988 0.983 0.989 0.990
o ARNH4 103

PBC cmolc kg1M1/2 0.88 0.26 3.9 0.72 0.38 0.13

o NH4 103

NH4sas 103 2.7 4.6 9.6 6.3 3.4 1.5

Candler LSD0.05

M1/2 15.6 24.6 2.0 17.1 12.8 2.0

cmolc kg1 13.8 6.5 7.9 12.3 4.9 1.1

o o ARNH4 is NH4 when NH4 0; PBC is potential buffering capacity; NH4 is the labile NH4 obtained by extrapolating the linear part of the Q/I plot to ARNH4 0; and NH4sas is specific adsorption sites. Least significant difference. Two means, all pairwise comparisons within each column, with difference greater than LSD indicate significantly different at P 0.05.

1672

SOIL SCI. SOC. AM. J., VOL. 64, SEPTEMBEROCTOBER 2000

Fig. 4. The relationship between exchangeable and labile NH4 for the Candler and Wabasso soils.

Fig. 3. The relationship between organic C and (A) potential buffering capacity and (B) labile NH4 for the soils. The circles represent the data used in the regression for the surface and subsurface soils of the Candler and Wabasso sand. The triangles represent the Wabasso 90- to 100-cm depth horizon and were not used in the regression analysis.

horizon sample was fourfold greater than that of the surface sample, while its organic C content was only 1.5-fold greater. This suggests that the mineralogy of the 90- to 100-cm depth horizon contributed more to the PBC than did the organic C. The clay mineralogy data suggest that there would be more smectite, a 2:1 expandable mineral, in the 90- to 100-cm depth horizon than in the surface soil of the Wabasso sand since clay content of the former soil was greater than that of the latter (Table 1). Further, despite its high C content, the size of the labile NH4 pool of the Wabasso 90- to 100cm depth horizon was smaller than labile NH4 pools of the surface samples of both soils. Such an observation may be attributed to the relatively high NH4 adsorption strength (the tightness with which the NH4 ion is held by the adsorption sites) and limited adsorption sites on organic components in the Wabasso 90- to 100-cm depth horizon. A study on soil organic matter accumulation (Evangelou et al., 1986) has shown that high organic matter content does not necessarily and proportionally contribute to an increase in buffering capacity when the affinity of a soil for NH4 is high. The PBC for 0- to 30- and 90- to 100-cm soil samples of the Wabasso sand was higher than those for the Candler sand. The higher PBC may be attributable to the presence of smectite in the Wabasso sand. Although the Candler sand contained hydroxyinterlayered vermiculite, it is well known that the cation-exchange ca-

pacity of this mineral is much lower than that of smectite because the internal cation adsorption sites of the vermiculite are occupied by a nonexchangeable hydroxyinterlayer. The quantity of smectite in the Wabasso sand (3060 cm) might have been too low to lead to a higher PBC. The correlation between organic C and PBC or labile NH4 for the surface and subsurface soils (Fig. 3) needs to be further evaluated using more soils. The correlation can be useful in studying the NH4 chemistry of sandy soils since organic C measurements are less laborious than are quantityintensity experiments. The correlation between labile NH4 and extractable NH4 was not significant (Fig. 4). This is contrary to what has been reported for clay loam soils (Thompson and Blackmer, 1992). The exchangeable NH4 concentration in the subsurface samples was low (Table 1), despite prediction of a relatively large labile NH4 pool (Table 2). More studies for sandy soils are required to redefine the nature of labile NH4 obtained from the intercept of Q/I plots. At a low concentration of NH4 in the equilibrating solution (1 103 M ), no additional NH4 was adsorbed by the surface and subsurface soil samples; however, there was desorption of NH4 when the soil samples were subsequently suspended in 2 M KCl (Table 3). This appears to be due to release of originally bound NH4 on the soils. At low levels of applied NH4, the Wabasso 90- to 100-cm depth horizon adsorbed and then desorbed a significant part of the applied NH4 due to its high buffering capacity. When the concentration of applied NH4 was sufficiently high (e.g., 5 103 M ), desorption of applied NH4 was evident from both the surface and subsurface horizons (Table 3). It appeared that the Wabasso 90- to 100-cm depth horizon served as an NH4 sink rather than a reservoir at low solution NH4 concentration. This is consistent with its low o value, which suggests little NH4 availability at ARNH4 equilibrium (NH4 0) for this horizon (Table 2). The amount of desorbed NH4 increased with increas-

WANG & ALVA: AMMONIUM ADSORPTION AND DESORPTION IN SANDY SOILS

1673

Table 3. Amounts of applied NH4 that were adsorbed and desorbed, and desorbed NH4 as percentage of the adsorbed NH4 in soil at different concentrations of NH4 in the equilibrating solution.
NH4 concentrations in solution (M ) 1 Soil series Wabasso Depth cm 030 3060 90100 030 3060 Adsorbed 2.2 1.1 8.3 1.8 0.2 0.7 103 % Desorbed 0 0 72 0 0 NC Adsorbed 10.8 3.2 49.5 8.6 5.4 2.3 5 103 Desorbed mg 4.7 0.6 36.2 6.0 2.4 1.5 kg1 44 19 73 70 44 1.7 23.7 10.8 75.3 12.9 10.8 1.7 9.5 2.2 57.0 9.0 4.5 2.2 40 20 76 70 42 2.0 % Desorbed Adsorbed 1 102 Desorbed % Desorbed

Desorbed 0.4 0 6.0 1.3 0.9 1.0

Candler LSD0.05

Percentage NH4 desorbed [(amount of applied NH4 that was adsorbed and then desorbed by 2 M KCl) (amount of applied NH4 that was adsorbed)]100. If the value for adsorbed NH4 is negative (no adsorption of applied NH4), the percentage NH4 is taken as zero. LSD0.05 least significant difference. Two means, all pairwise comparisons within each column, with difference greater than LSD indicate significantly different at P 0.05. NC not calculated due to zero values.

ing NH4 adsorption (Table 3). On the other hand, apparent NH4 adsorption strength changed little. This is shown by only small changes in percentage desorption for applied NH4, as the concentration of applied NH4 in the solution increased from 5 103 to 1 102 M. The stable NH4 adsorption strength across a range of applied NH4 values suggests that there was an equilibrium between labile and specific adsorption sites for this range of solution NH4 concentrations and NH4 fixation sites on the soils. It is interesting that adsorbing sites for the subsurface horizons had a greater potential to fix NH4 than for the surface soils, as indicated by smaller values for percentage NH4 desorbed for the subsurface horizons (Table 3). The Wabasso 90- to 100-cm depth horizon, although having the largest capacity for adsorbing NH4 among the soils, also released most of the applied NH4 that was adsorbed. Obviously, organic C content alone cannot explain these differences in NH4 desorption. There is a need to investigate the nature of the organic matter and minerals in these soils, which could impact the reactions of NH4.

is of particular interest for the Wabasso 90- to 100-cm depth horizon, which contributes to even greater retention of NH4 than does the surface horizon of this soil.
ACKNOWLEDGMENTS This study was made possible, in part, with funding from the Florida Citrus Production Research Advisory Council. Laboratory assistance of Debbie van Cleif is gratefully acknowledged.

REFERENCES
ALPKEM Corporation. 1989. RFA methodology for ammonium nitrogen. A303-S020. Alpkem Corp., Clackamas, OR. Alva, A.K., M.E. Sumner, and W.P. Miller. 1991. Relationship between ionic strength and electrical conductivity for soil solutions. Soil Sci. 152:239242. Avnimelech, Y., and M. Laher. 1977. Ammonium volatilization from soils: Equilibrium considerations. Soil Sci. Soc. Am. J. 41: 10801084. Beckett, P.H.T. 1964a. Studies on soil potassium. II. The immediate Q/I relation of labile potassium in the soil. J. Soil Sci. 15:923. Beckett, P.H.T. 1964b. Potassiumcalcium exchange equilibria in soils: Specific adsorption sites for potassium. Soil Sci. 97:376383. Bohn, H.L., B.L. McNeal, and G.A. OConnor. 1985. Soil chemistry. 2nd ed. John Wiley & Sons, New York. Bremner, J.M., and C.S. Mulvaney. 1982. NitrogenTotal. p. 595624. In A.L. Page (ed.) Methods of soil analysis. Part 2. 2nd ed. ASA and SSSA, Madison, WI. Evangelou, V.P., and R.L. Blevins. 1985. Soil-solution phase interactions of basic cations in long-term tillage systems. Soil Sci. Soc. Am. J. 49:357362. Evangelou, V.P., A.D. Karathanasis, and R.L. Blevins. 1986. Effects of soil organic matter accumulation on potassium and ammonium quantityintensity relationships. Soil Sci. Soc. Am. J. 50:378382. Fenn, L.B., J.E. Datcha, and E. Wo. 1982. Substitution of ammonium and potassium for added calcium in reduction of ammonium loss from surfaced-applied urea. Soil Sci. Soc. Am. J. 46:771776. Goldberg, S.S., and P.L. Gainey. 1955. Role of surface phenomena in nitrification. Soil Sci. 80:4353. Griffin, G.P., and J.J. Jurinak. 1973. Estimation of activity coefficients from the electrical conductivity of natural aquatic systems and soil extracts. Soil Sci. 116:2630. Kai, H., and T. Harada. 1969. Studies on the environmental conditions controlling nitrification in soil. II. Effects of soil clay minerals on the rate of nitrification. Soil Sci. Plant Nutr. 15:110. Lumbanraja, J., and V.P. Evangelou. 1990. Binary and ternary exchange behavior of potassium and ammonium on Kentucky subsoils. Soil Sci. Soc. Am. J. 54:698705. Lumbanraja, J., and V.P. Evangelou. 1994. Adsorption and desorption

CONCLUSIONS
The potential NH4 buffering capacity and labile NH4 values for the sandy soils studied were much lower than those for clay and silt loam soils reported in the literature. Although the soils should have a relatively poor NH4 holding capacity, this study also provided evidence of fixation of NH4 by these soils. The NH4 fixation capacity (i.e., NH4sas) of these soils should be taken into account when strategies for N fertilization are to be developed to maximize N use efficiency. Strong adsorption and desorption of NH4 was evident for soil horizons high in organic C. Organic C was positively correlated with potential NH4 buffering capacity and labile NH4 for the soils, with the exception of the Wabasso 90- to 100-cm depth horizon. This 90- to 100-cm depth and the surface soil of the Wabasso sand had a relatively high NH4 buffering capacity, which may be responsible for greater NH4 retention in this soil than in the Candler sand. Further studies are necessary to understand the role of composition and nature of soil components in sandy soils in governing the fate of applied NH4. This

1674

SOIL SCI. SOC. AM. J., VOL. 64, SEPTEMBEROCTOBER 2000

of potassium and ammonium at low cation concentrations in three Kentucky subsoils. Soil Sci. 157:269278. Opuwaribo, E., and C.T.I. Odu. 1978. Ammonium fixation in Nigerian soils: 4. The effects of time, potassium and wet dry cycles on ammonium fixation. Soil Sci. 125:137145. Pasricha, N.S. 1976. Exchange equilibria of ammonium in some paddy soils. Soil Sci. 121:267271. Pasricha, N.S., and T. Singh. 1977. Ammonium exchange equilibria in the unmanaged soils and forms of ammonium which are not water-soluble but are available to low land paddy. Soil Sci. 122: 9096. Rappaport, B.D., and J.M. Axley. 1984. Potassium chloride for improved urea fertilizer efficiency. Soil Sci. Soc. Am. J. 48:399401.

Talibudeen, O. 1981. Cation exchange in soils. p. 115177. In D.J. Greenland and M.H.B. Hayes (ed.) The chemistry of soil processes. John Wiley & Sons, New York. Thompson, T.L., and A.M. Blackmer. 1992. Quantityintensity relationships of soil ammonia in long-term rotation plots. Soil Sci. Soc. Am. J. 56:494498. Wang, F.L., and A.K. Alva. 1996. Nitrogen leaching from slow-release urea sources in sandy soils. Soil Sci. Soc. Am. J. 60:14541458. Whitting, L.D., and W.R. Allardice. 1986. X-ray diffraction techniques. p. 331362. In A. Klute (ed.) Methods of soil analysis. Part 1. ASA and SSSA, Madison, WI.

Soil Sampling Strategies for Precision Agriculture Research under Sahelian Conditions
J. W. van Groenigen,* M. Gandah, and J. Bouma ABSTRACT
The cost of soil samples to characterize field variability is a key problem in precision agriculture. This study was conducted to investigate whether yield maps can be used to optimize soil sampling for characterizing soil variables that determine yield variability. Using an inexpensive, low-tech scoring technique, yield maps of pearl millet [Pennisetum glaucum (L.) Br.] were produced for a zero-input farm in Niger. The soil was classified as a typic Plintustalf. The Spatial Simulated Annealing (SSA) algorithm was used to optimize three sampling schemes. Scheme 1 optimized coverage over the whole area. Scheme 2 covered the whole yield range. Scheme 3 covered the lowproducing areas. Yield varied from 0 to 2500 kg ha1, measured per planting hill. Using correlation coefficients, Scheme 2 found significant correlations between five soil variables and yield. Scheme 1 found only one significant correlation and explained 37% of the variation in yield using multivariate regression of yield on soil variables. Scheme 2 explained 70% of the variation in yield. Differences between Scheme 3 and Scheme 1 proved to be significant for distance to shrubs, relief, soil pH, and cation-exchange capacity (CEC). We concluded that shrubs are the main factor influencing millet yield by means of catching eroded materials and improving soil fertility. The possibilities of planting shrubs to improve soil fertility should be investigated. Variograms of relief and yield suggested that spatial correlation is largely confined to distances of 3 to 5 m. Since Scheme 2 was most effective in establishing soilyield relationships, we concluded that yield maps can be used to optimize soil sampling.

n recent years, an increased interest among soil scientists regarding soil variability on the field scale has been observed. Geostatistics plays an important part in the quantitative evaluation of such spatial variability. Interpolation techniques such as ordinary (block) kriging and indicator kriging have been applied for optimal interpolation of point observations (e.g., Van Uffelen et

J.W. van Groenigen, Soil Sci. Div., Int. Inst. for Aerospace Survey and Earth Sci. (ITC), P.O. Box 6, 7500 AA Enschede, the Netherlands; currently at Univ. of California-Davis, Dep. of Agron. and Range Sci., 1 Shield Av., Davis, CA 95616-0515; M. Gandah, Institut National de Recherche Agronomique du Niger (INRAN), B.P. 429, Niamey, Niger; J. Bouma, Dep. of Environ. Sci., Wageningen Agricultural Univ., P.O. Box 37, 6700 AA Wageningen, the Netherlands. Received 4 Dec. 1998. *Corresponding author (groenigen@ucdavis.edu). Published in Soil Sci. Soc. Am. J. 64:16741680 (2000).

al., 1997; Stein et al., 1997). Also, stochastic simulations have been used to reproduce the spatial variability that has been modeled from the observations on simulated mez-Herna ndez and Srivastava, 1990; maps (e.g., Go Goovaerts, 1997). Apart from these applications and adaptations of existing geostatistical tools, precision agriculture poses some more specific challenges to geostatisticians. One of these is the increasing availability of maps of data that can be helpful for purposes such as mapping of soil properties or yield prediction. Examples of such data are maps of soil tillage resistance (Van Bergeijk and Goense, 1997), remote sensed imagery (Booltink and Verhagen, 1997), and yield maps collected using lowtech (Stoorvogel, 1995) or high-tech (Bouma, 1997) approaches. Franzen et al. (1998) found that partitioning the area according to topography might decrease the number of samples needed to characterize the soil variability compared with a regular grid. In this study, we combined such partitioning of the area with geostatistical analysis and interpolation. Furthermore, we present an algorithm that optimizes spreading of the observations across the study area. Spatial variability of yield and soils is an important aspect of farming in the zero-input subsistence millet farming systems of the Sahelian Zone. Soils are highly variable in terms of both chemical and physical fertility. In a study in northern Niger, Bouma et al. (1996) mentioned five reasons for this: (i) relief and crusting, which cause crucial redistribution of water infiltration (Gaze et al., 1997); (ii) termites, which may enrich the soil locally; (iii) local effects of trees and shrubs; (iv) differences associated with landscape position; and (v) soil fertility gradients around villages. Brouwer et al. (1993) reported a yield stabilizing effect of this variability in fields in the same area over years. In relatively wet years, yields were highest at the highest sites, due to catchment of fertile eroded
Abbreviations: CEC, cation-exchange capacity; MMSD, minimization of the means of the shortest distance; OM, organic matter; SSA, spatial simulated annealing.