Site-occupancy models that predict habitat suitability for plant species in relation to measurable environmental factors are useful for conservation planning. Such models can be derived from large-scale presence-absence datasets on the basis of environmental observations or, where only floristic data are available, using plot-mean values for plant traits. However, the form of the modelled relationship between species presence and environmental variable depends on the form of the statistical model adopted and hence can introduce additional uncertainty. We used an ensemble-modelling approach to constrain and quantify the uncertainty due to the choice of statistical model, fitting multiple logistic regressions (MLR), generalised additive models (GAM), and multivariate adaptive regression splines (MARS) to very large floristic datasets. Niche models were derived for 1342 species of vascular plants, bryophytes and lichens, representing a large proportion of the British flora and many species occurring in continental Europe. Each model predicts habitat suitability for a species in response to annual rainfall, maximum and minimum temperature, and plot-mean trait scores for soil pH, fertility, wetness and canopy height. An R package containing the fitted models is presented which derives habitat suitability predictions using the three fitting methods. Bespoke functions are included so that these can be plotted in relation to individual explanatory variables, to illustrate responses and the uncertainty due to choice of model. Model-average values are also derived and can be mapped. The package is freely available in the supplementary material.