A combination map, consisting of the information in both the population density map and the modified-distance map, was also created (Figure 2c). This map is simply based on a grid cell by grid cell multiplication of the reclassified values from the population density and the modified distance map. In this paper, I will refer to this map as the combination map.Data points for measuring soil P and associated factors were stratified by zone and randomly located within each zone according to the combination map—with approximately 70 data points per zone. Location and address of each point were determined using the Madison and Dane County parcel GIS layers. Permission was requested from landowners to take a soil sample, and the precise location of the sample on the property was determined using standard randomizing techniques. If permission was denied (2 cases out of 330) or if there was no one present at the location, a coin toss was used to determine movement one parcel to the right or to the left along the same road. Landowners were also asked about their fertilizer use practices, manure use, dog ownership, and the date the house was built, if known.Approximately 400 soil samples were taken in the top soil horizon to a depth of 13.5 cm with a standard soil corer (diameter = approximately 1.6 cm). This depth was always within the surface horizon and any grass thatch was removed from lawn samples. Other data collected include percent slope, convex or concave nature of the slope, land use, land-cover type, and percent vegetative cover. The visually perceived zone was also recorded. The visually perceived zone was determined by visual inspection using a predetermined set of definitions of each zone. For example, urban sites were those with the highest housing density or some industrial use; suburban sites were those of moderate housing density and residential character; suburban fringe were newer residential developments of low housing density and larger houses; agricultural fringe were older residential developments of low density; and agricultural were those areas that were actively farmed. A handheld global positioning device was used to determine the precise (± 1 m) location of the soil sample.Soil samples were stored for no more than 3 weeks at room temperature before analysis. They were then dried for 15–24 hours at 50–55degreeC and sieved (1.8-mm mesh). Soil samples were then analyzed for Bray-1 P at the University of Wisconsin Soil and Plant Analysis Lab. Bray-1, a measure of extractable P, is a commonly used measure of phosphorus available to plants in agricultural systems. While relationships between extractable soil P and dissolved P in runoff have been noted in some systems (Sharpley and others 1993, Sharpley 1995), these extractions were generally developed to estimate plant available P, not to reflect P storage in the soil or P runoff. Therefore, we tested a subset (60) of our samples for total P, a better measure of P storage in soils. A regression of our samples indicates a reasonably close relationship between Bray-1 P and total P in our study area soils (Figure 3), indicating that our Bray-1 P results are probably a satisfactory estimate of both extractable P and the sorbed P that tends to accumulate in agricultural soils.