The Impact of Climatic and Agricultural Variables on the Frequency of Silage Pollution Incidents

Over the period 1969-1991, Northern Ireland silage pollution incidents were positively correlated (r2=0·40) with the quantity of silage made, giving an underlying silage pollution incident rate (SPIR) of 36·7 incidents per million tonnes of silage. Silage pollution was most prevalent in June, with 45·9% of all incidents, but the peak of silage production was in May (32% of silage produced). Analysis of SPIR with climatic variables indicated a statistically significant negative relationship (r2=0·36) with sunshine hours over the period 13 May to 9 June. However, the low correlation was due in part to a series of observations where SPIR was low and did not vary with sunshine hours. This group largely consisted of years with the lowest or highest rainfall for the period May-June. A polynomial to the fifth power of rainfall for this period (29 April-23 June) gave a good fit (r2=0·56) with SPIR, and this correlation could be improved by including a selection of four 2-week temperature variables from the months of May to August. The residuals from the final regression of SPIR vs climatic variables showed no trends with time, which suggests that the efficiency with which farmers were preventing silage pollution was not increasing. Silage dry matter (DM) was positively correlated (r2=0·29) with SPIR, with each 1·0% increase in DM associated with a decline in SPIR of 9·5 incidents Mt silage-1. DM was also positively correlated (r2=0·47) with sunshine hours for the period 13 May-9 June. The proportion of silages subjected to wilting was not significantly correlated (r2=0·07) with SPIR. Analysis of silage DM contents between 1980 and 1991 indicated that the DM content of wilted silages was only 2·2% DM higher than for unwilted silages (19·3% DM). This small difference indicates that, under current agricultural practices, wilting in Northern Ireland has been comparatively inefficient in increasing grass DM and decreasing silage effluent volumes.