An evaluation of indices that describe the impact of ozone on the yield of spring wheat (Triticum aestivum L)

A comparison of the performance of different ozone indices in exposure-response functions was made using crop yield and ozone monitoring data from spring wheat studies carried out within the framework of the European open-top chamber programme. Indices were calculated for a twelve-hour period (0900-2100 h, local time). An attempt was made to incorporate a measure of absorbed dose into current indices by weighting with simultaneous sunshine hour values. Both linear and Weibull models were fitted to the exposure-response data in order to evaluate index performance. Cumulative indices which employed continuous weighting functions (allometric or sigmoid) or which censored concentrations above threshold values performed best as they attributed increasing weight to higher concentrations. Indices which simply summed concentrations greater than or equal to a threshold value did not perform as well as equal weight was given to all concentrations greater than the threshold value. Model selection was found to be very important in determining the indices that best describe the relationship between exposure and response. In general weighting hourly ozone concentrations with the corresponding sunshine hour values in an attempt to incorporate this proposed measure of plant activity into current indices did not improve index performance. Ozone exposure indices accounted for a large proportion of the variability in data (91%) and it is suggested that a strong link exists between exposure and dose. (C) 1997 Elsevier Science Ltd.