UBC Theses and Dissertations
Soil variability and the use of conventional and innovative methods for assessing soil fertility Bank, Gary Michael
Soil sampling plays a major role in the process which leads to fertilizer recommendation. Conventionally, characterization of field chemistry is based on a single composite sample. This single value is used as the basis for a fertilizer recommendations which often amounts to many hundreds of dollars per field. In order to test the validity of the conventional composite sampling method a sampling program involving three different sampling methods was undertaken. The sampling program was designed to make a comparison of the information provided by and the costs associated with the three different sampling methods. In addition, factors having the largest influence on variability were determined. Characterization of the chemical variability of several agricultural fields in the Lower Fraser Valley was also achieved. Four fields were sampled in the spring and three of them in the fall. Soils were sampled in the plow layer, at 30-60 cm, and at 60-90 cm. Random stratified sampling, random stratified composite and conventional sampling was carried out on each field. Soil analyses included exchangeable bases, pH, %C, NO[sub 3]-N, and Bray's extractable P. Analysis of variance showed that differences among fields and depths were the most important sources of variance. Contributions of time and interaction effects to variance were generally small. A comparison of mean values produced by the different sampling methods showed that mean values resulting from the use of conventional sampling for NO[sub 3]-N, P, and K, were within 20% of the stratified random mean 65% of the time, and within 50-125% of the stratified random mean 17% of the time. Random stratified composite mean values differed from the random stratified mean by less than 50% in all cases. The value of stratified random sampling (detailed sampling) becomes apparent when a comparison of the chemical value within each of the field strata is made with the conventional mean. In the worst case, only 22% of a field's area fell within plus or minus 15 ppm of the conventional mean for P. The high cost of analysis for detailed sampling, compared to the cost associated with conventional sampling, was shown to be offset by savings in the amounts of fertilizer recommended. The results of detailed sampling could be used to provide optimum fertilizer recommendations to all parts of the field and also to reduce field chemical variability. Cluster analysis carried out on plow layer samples of individual fields was able to classify fields into units with significantly different chemistry. The multivariate approach was able to distinguish among areas of fields with significantly different levels of CEC, %C, pH, and the major fertilizer variables P and K. The lack of contiguousness of unit members, for some fields, and the need for more than one P and K recommendation within many of the cluster units would reduce the possible practical usefulness of these results. Cluster analysis run on plow layer data was highly successful in discriminating among three fields. This suggests a high degree of chemical identity of fields, even if found on the same mapping unit.
Item Citations and Data