BIOASSAY97: a new EXCEL VBA macro to perform statistical analyses onherbicide dose-response data

This paper presents BIOASSAY97, a new EXCEL® macro add-in to perform non-linear regression analysis on bioassay data. This macro has been specifically developed to comply with all the peculiarities of herbicide bioassays, even though it can be used with any kind of bioassays, especially by users with limited knowledge in statistics and computer programming. Starting from experimental data, BIOASSAY97 estimates all the most important ED-levels, such as the ED10, ED50, ED90 and any other user specified ED-levels, which are very important as decision tools in defining rational Integrated Weed Management Systems. In particular, those indicators can be used as a basis to adjust herbicide doses according to pedological, floristic and meteorological conditions. This latter aspect is particularly important as the climatic component is frequently neglected when selecting herbicide doses. Simultaneous fitting of several dose-response curves to the same dataset is also possible, in order to estimate the relative efficiency of either several herbicides or the same herbicide in different formulations or environmental conditions or weed flora situations. Three basic response models are built-in BIOASSAY97: a log-logistic symmetric model, a Gompertz model and a peaked logistic model; constraints on parameters can be introduced in several ways, according to user specified needs, to increase the flexibility of BIOASSAY97 and be able to analyse data from any type of bioassay experiments. The Box-Cox-transform-both-sides approach was built in, for the cases where the assumption of variance homogeneity is violated. Estimates are always provided with standard errors and confidence intervals; graphical analysis of residuals and F test for lack of fit are also possible to evaluate the goodness of regression. BIOASSAY97 has been extensively tested and validated, it is freeware and can be easily downloaded from the author web-site.