How important is plot relocation accuracy when interpreting re-visitation studies of vegetation change?

Background: Re-visitation studies are often based on phytosociological survey data where the precise location of the original plots is unknown. Attempts to evaluate the error associated with relocation uncertainty are rare, yet this is important in interpreting the results with any degree of confidence.
Aims: Using a 50-year re-visitation study of upland vegetation in the Scottish Highlands, we aim to assess the potential for, and implications of, uncertainty in relocating plots.
Methods: At nine sites, three to five replicate plots were surveyed within a stand of vegetation relocated to the nearest 100 m using the original plot location data.
Results: The compositional difference (measured by the Bray-Curtis distance) between the original plot and new replicate plots was greater than that among the replicate plots, both for the combined data and individual vegetation types. Temporal species turnover was greatest in the lower cover-abundance categories (< 5%).
Conclusions: We demonstrate that if temporal change in vegetation can be shown to be greater than local spatial heterogeneity today, patterns of change at the local scale detected by re-visitation data can be interpreted with some confidence if other sources of error are minimised. Recommendations for best practice in re-visitation studies are made.