Multifactorial spatial analysis of mycotoxin contamination of winter wheat at the field and landscape scale

Abstract

An assessment of contamination of cereals with the Fusarium mycotoxins deoxynivalenol (DON) and zearalenone (ZEA) is difficult because of the complexity and interactions of various driving factors. We study the mycotoxin accumulation in winter wheat growing after maize or oilseed rape at different topographic positions within fields consisting of alternating hilltop and depression positions across a transect in the Uckermark region in the Northeast of German Lowlands. A linear mixed-effects model and censored regression are used to account for spatial autocorrelation and non-detects in determining factors controlling DON content in 304 samples from 19 fields in 2006 and 2007. Environmental factors (actual rainfall during the anthesis, long-term mean annual precipitation), topographic factors (relief position, topographic wetness index TWI) and management factors (tillage, preceding crop, susceptibility ranking of wheat cultivars to Fusarium spp.) as well as the year of investigation were included as explanatory factors in this analysis. The spatial linear mixed-effects model indicates that DON on wheat fields with maize as preceding crop was one order of magnitude higher than in wheat grown after oilseed rape (significant at the 5% level). Tillage had a similar effect: Ploughing significantly reduced DON by more than one order of magnitude. Humid areas (mean annual precipitation >550 mm) showed a significantly higher DON contamination (3.7-fold increase) than the drier areas. TWI and topographic position had significant influences on DON levels. Their combined effect lead to a 68% higher DON at depressions with high TWI compared to hilltops with low TWI. The results of the censored regression were very similar to those of the mixed model. The susceptibility ranking score of the wheat cultivars to Fusarium infection did not have a significant influence on DON in these models. Overall, the statistical models revealed a combined influence of climatic, topographic, and crop management factors on DON concentration in winter wheat on different spatial scales for the environmental setting of our study area.