The effects of spatial variation of certain soil properties on the winter yield of a lucerne stand

Abstract

In general, agricultural fields are managed as uniform units, ignoring spatial soil heterogeneity and its effects on growth and yield of field crops. This study was conducted from June 2001—February 2002 and examines the effects of spatial variation of soil properties on the winter yield of a two-year-old lucerne stand on two soil types using geostatistical procedures. Seventy-two sampling points (nodes) were laid out on a 20 m square grid, with an additional 90 sampling points laid out on a 2.5 m square grid at six randomly selected node points to ensure that the total spatial structure would be identified. From initial soil sampling and analyses, the two experimental soils were classified as belonging to the Stella and Pyramid soil families with inter alia mean clay contents of 45% and 46%, pH(H2O) values of 7.8 and 8.8, and mean P status (Ambic) contents of 18.3 and 6.4 mg kg−1 respectively. Green biomass lucerne yield was determined on six occasions at all nodes, while soil sampling (0–300 mm layer) and analyses were done once in June 2001. Basic statistical analyses showed, for some soil properties, two distinct data populations, emphasizing the presence of two soil types. A yield prediction model (R = 0.55) contained pH(H2O), organic C, K and sand contents as variables. The geostatistical analyses of the yield model variables produced standard semi-variograms although with highly variable autocorrelation lengths. Making use of various kriging techniques, maps of soil properties and yield were compiled. These maps reveal that spatial variation of yield bears a fair resemblance to that of some soil properties and, therefore, supports the validity of the yield prediction model for this case study. This study has shown that the scale of variation of lucerne yield can be related to that of soil properties, a finding which can be useful when designing sampling schemes.