Calibration of bicarbonate-extractable phosphorus on wheat
van Kessel, C.
Nine locations were selected throughout Saskatchewan so that a summerfallow an oilseed stubble and a wheat stubble experimental site were adjacent to each other. Each site was characterized by analyses carried out on a composite soil sample obtained from the whole site. A randomized complete block (RCB) design with three treatments (control, 10 lb P2O5/ac , and 20 lb P2O5/ac) and four replicates was established at all sites. Subsequently, composite soil samples from each plot were taken. Comparison between the two sampling schemes suggested that characterizing a site on the basis of one composite sample may not result in a true representation of the P fertility status of individual plots within the site. Analysis of variance for a RCB design resulted in non-significant response of spring wheat to P fertilizer application independent of previous crop history. However, examination of soil P levels in each plot of every experiment revealed that sites were extremely variable. Hence, comparison of mean grain yields in many cases was based on averaging "non-responsive" parts of the field with "responsive" ones. When the probability of grain yield response was plotted on a per plot basis, a 65, 72 and 78% probability of positive response to P fertilization was obtained on the fallow, wheat stubble and canola stubble, respectively. Boundary-line analysis indicated that spring wheat had very small possibility of positive yield response to P fertilization when soil bicarbonate-P was over 25 lb/ac. Spatial variability of a site must be determined prior to carrying out an experiment and number of replications must reflect the differences sought.
Soils and Crops Workshop