Environmental Management of Soil Phosphorus: Modeling Spatial Variability in Small Fields

management (Lemunyon and Gilbert, 1993). The P index accounts for source (soil P and rate, method, and The mapping of soil P concentration is necessary to assess the timing of applied P) and transport (surface runoff, erorisk of P loss in runoff. We modeled the distribution of Mehlich-3 sion, leaching, and landscape position) factors controlextractable soil P (M3P) in an east-central Pennsylvania 39.5-ha watershed (FD-36) with an average field size of 1.0 ha. Three interpolation ling P loss in surface runoff and ranks sites for their models were used: (i) the field classification model—simple field potential risk of P loss. means, (ii) the global model—ordinary kriging across the watershed, In areas with large fields, the mean or median soil and (iii) the within-field model—ordinary kriging within fields with test value is generally used as the best estimate of P a pooled within-stratum variogram. Soils were sampled on a 30-m concentration in a field, except in cases where precision grid, resulting in an average of 14 samples per field. Multiple validation sampling and fertilizer application are used. In areas runs were used to compare the models. Overall, the mean absolute with small fields, such as Pennsylvania, a single bulk errors (MAEs) of the models were 76, 71, and 66 mg kg 1 M3P for composite or the mean or median soil test value is tradithe field classification, global, and within-field models, respectively. tionally used as the best estimate of P concentration. The field classification model performed substantially worse than did Under these models, information on farmand fieldthe kriging models in five fields; these fields exhibited strong spatial scale variability is not used for the estimation of P distriautocorrelation. The within-field model performed substantially better than did the global model in three fields where autocorrelation bution. More complex interpolation methods, such as was confined by the field boundary. However, no differences in P those from the disciplines of geostatistics and precision index classification were observed between the three prediction suragriculture, incorporate spatial variability into estimates faces. The field classification model is simpler and less expensive of P distribution. Field-scale variability, which is conto implement than the kriging models and should be adequate for fined to field boundaries, may be caused by uneven applications that are not sensitive to small errors in soil P concentrafertilizer distribution or movement within fields. Farmtion estimates. scale variability, which is not confined to field boundaries, is likely caused by larger scale management factors such as distances to roads or manure storage facilities. P is an essential element for plant and Natural factors, such as variations in weathering, soil animal growth and its input has long been recognized parent material, erosion, and water movement patterns, as necessary to eliminate plant nutrient deficiencies and may also influence soil P distribution (Larson et al., to maintain profitable crop and livestock production. 1997). The influence of management is probably Excess P inputs, however, can increase the biological stronger than natural factors in fields with very high soil productivity of fresh waters by accelerating eutrophiP and a history of large P applications. The choice of cation (USEPA, 1996). Eutrophication is the natural an interpolation method should be based on an assessprocess of lake and stream aging through nutrient enment of the scale and strength of autocorrelation present richment, but may be unnaturally accelerated by humanand the costs associated with the sampling design. induced nutrient loadings. State and Federal authorities The variogram is an important tool to detect the presare moving towards stricter P management and inence of spatial autocorrelation and to estimate the varicreased pollution prevention support. Agriculture acability structure of soil properties (McBratney and Webcounts for the major proportion of total inputs of P to ster, 1986). A global variogram can be used to assess major freshwater systems in the USA (USEPA, 1996). the variability structure of a soil property across a waterThere is evidence that the great majority of agriculshed, but it does not account for smaller-scale factors tural P export originates from a small portion of the such as field boundaries. Variograms can also be devellandscape in humid, upland agricultural watersheds oped for each field individually (Goovaerts, 1997). Sev(Gburek and Sharpley, 1998). These areas have been eral researchers have used within-field, more generally termed critical source areas and are characterized by termed within-stratum, variography to estimate the spahaving high potential to release P into surface or subsurtial variability structure of soil properties (Stein et al., face runoff in conjunction with hydrologic connectivity 1988; Boucneau et al., 1998). However, data sparsity with streams or ditches. Targeting critical source areas may prevent the reliable estimation of spatial semivariwould increase the efficiency and reduce the economic ance functions within each stratum (Webster and Oliver, costs of control. In response, a site vulnerability assess1992). At least 50 to 100 data points may be necessary ment tool, the P index, has been developed to target P to achieve a stable variogram, depending on lag spacing and the smoothness of the spatial variation (Voltz and B.A. Needelman and G.W. Petersen, Dep. of Agronomy, The PennsylWebster, 1990; Burrough and McDonnell, 1998). vania State Univ., 116 A.S.I. Bldg., Univ. Park, PA 16802; W.J. Gburek and A.N. Sharpley, Pasture Systems & Watershed Management ReMost research concerning soil nutrient distribution search Unit, USDA–ARS, Curtin Rd., Univ. Park, PA 16802. Received 13 Sept. 2000. *Corresponding author ([email protected]). Abbreviations: Log M3P, logarithm of M3P; M3P, Mehlich-3 extractable soil P; MAE, mean absolute error. Published in Soil Sci. Soc. Am. J. 65:1516–1522 (2001).