Floristic changes in the British Isles: comparison of techniques for assessing changes in frequency of plants with time

Botanical recording data are often used to assess changes in the frequencies of plant species over time, but are subject to marked variations in recording activity. We compare and evaluate some general methods that can be used to detect changes in species' frequencies taking into account the recording variations. Models for 15 species that have been studied in detail previously were compared using the numbers of individual records, sites, hectads, or vice-counties at different time scales (year, decade, moving averages, and pre-/post- specific dates), with or without correction for recording variation. The best methods had a correction for the amount of recording over time, summarized records by decade or moving average, and used an extrapolation between first and last records for sites or hectads. Increasing the geographical and temporal scales can decrease the influence of recording variations, but leads to a loss of sensitivity and under-estimates the true extent of change. The choice between sites and hectads will depend on the detail of the records available; cruder data sets should use the latter.