The use of longitudinal data analysis to study the multi-seasonal growth responses of Norway and Sitka spruce to summer exposure to ozone: Implications for the determination of critical levels

A common feature of many pollutant exposure studies is that data from these experiments often consist of either plant biomass or yield, and the response to the pollutant is usually based on a simple comparison of means determined at the end of each growing season. This type of data is now being used to quantify critical levels for different types of vegetation. Such an approach is, however, inappropriate for the relatively short-term exposure studies with trees since, due to their longevity, it is not possible to determine a final yield. Instead these studies should be regarded as a type of intervention experiment in which only a small part of the life cycle of the tree is investigated. Moreover, critical levels as they are now defined focus on the cumulative exposure to ozone concentrations over time. Hence, any analysis of the effects of ozone on the growth of trees should similarly focus on the behaviour of growth functions over time and not on a comparison of biomass at the end of a growing season or experiment. Here we report on the statistical analysis of a longitudinal study, where the term 'longitudinal' refers to the analysis of repeated measurements over time, and which was used to investigate relative differences in the growth of Sitka spruce and Norway spruce seedlings during three summers' exposure to ozone over three (Norway) or four (Sitka) growing seasons. Measurements of total seedling height and stem diameter were made at frequent intervals over the period of the experiment and the above-ground growth of individual trees (as log d²h) analysed for each growing season using a statistical model of the form: m(t)= a+b(t-t₀). No statistically significant differences in the growth of Norway spruce were observed after three summers of ozone exposure. The growth of Sitka spruce was, however, reduced by ozone during the third growing season and in the following year, even in the absence of the pollutant. For the Sitka spruce, the fitted model was used to calculate the time at which a 10% reduction in growth had occurred in the ozone-exposed trees. In combination with ozone AOT40 indices for the relevant growing seasons, this was then used to determine a critical level of 21.3 ppm-h for this species. Since the growth of the Norway spruce was unaffected by exposure to ozone it was not possible to calculate a critical level other than to surmise that it is in excess of 30 ppm-h.