Seed longevity: analysing post-storage germination data in R to fit the viability equation

For many decades, seed germination data have been modelled by probit analysis. In particular, it is the basis of the seed viability equation used, in the first instance, to describe the decline in germination of seeds in storage, but then also the rate of the decline, depending on seed moisture content and the temperature of storage. The underlying assumption of a probit model is that the response follows a normal distribution, in this case, loss of the ability to germinate over time. Probit analysis also takes into account the binomial error associated with germination data. Many statistical packages have probit analysis as an option within the generalized linear modelling framework; here, we present code for applying probit analysis in the free software, R. Codes are provided for fitting a single survival curve, for a single seed lot stored in a constant storage environment; for fitting multiple survival curves and evaluating the effect of constraining parameters for the different seed lots; and lastly, to model the moisture relations of seed longevity. The code bases provided could also be used in pollen and fern/bryophyte spore longevity modelling.