Investigating the effect of previous treatments on wheat biomass over multiple spatial frequencies

Abstract. In this study we use the maximum overlap discrete packet transform (MODWPT) to investigate residual effects on wheat biomass of fertigation treatments applied to a previous crop. The wheat crop covered nine subplots from a previous experiment on melon response to fertigation. Each subplot had previously received a different level of applied nitrogen. Many factors affect wheat biomass, causing it to vary at different spatial frequencies. We hypothesize that these will include residual effects from fertilizer application (at relatively low spatial frequencies) and the local influence of individual plants from the previous melon crop (at high frequency). To test this hypothesis we use the MODWPT to identify the dominant spatial frequencies of wheat biomass variation, and analyse the relationship to both the previous fertilizer application and the location of individual melon plants in the previous crop. The MODWPT is particularly appropriate for this because it allows us first to identify the key spatial frequencies in the wheat biomass objectively and to analyse them, and their relationship to hypothesized driving factors without any assumptions of uniformity (stationarity) of wheat-biomass variation.

The results showed that the applied nitrogen dominated the wheat biomass response, and that there was a noticeable component of wheat-biomass variation at the spatial frequency that corresponds to the melon cropping. We expected wheat biomass to be negatively correlated with the position of melons in the previous crop, due to uptake of the applied nitrogen. The MODWPT, which allows us to detect changes in correlation between variables at different frequencies, showed that such a relationship was found across part of the experiment but not uniformly.