Rarefaction profiles for repeated biodiversity surveys: integrating temporal variability and trend along sample-based accumulation curves

Repeated biodiversity surveys are often summarised by one diversity value per survey date, although temporal change may be expressed differently at the scale of individual sampling units, across the full sampled community, or along the accumulation process linking the two. We introduce rarefaction profiles for repeated biodiversity surveys, based on sample-based rarefaction applied independently to repeated survey dates. For each survey date, a rarefaction curve is calculated by accumulating comparable sampling units. Temporal variability and temporal trend are then estimated at each number of accumulated units, (m), from (m = 1), representing the expected diversity of one sampling unit, to (m = N), representing sampled gamma diversity within the defined sampling focus. The approach is metric-general and can be applied to taxonomic diversity, functional diversity and directional community descriptors, provided that the target quantity can be calculated for pooled subsets of sampling units. We provide a reproducible R implementation that returns survey-specific rarefaction curves, temporal variability profiles, temporal trend profiles, alpha- and sampled-gamma summaries, whole-profile summaries and contrasts among groups. We illustrate the method using data from 16 vegetation surveys conducted between 2001 and 2019 at the Monte Labbro long-term grassland experiment in southern Tuscany, Italy. Ungrazed plots showed higher temporal variability and stronger negative temporal change in richness and Shannon diversity than grazed plots, with resampling-based uncertainty supporting these taxonomic contrasts across most or all of the accumulation gradient. Exact within-block label randomisation showed that the taxonomic Grazed–Ungrazed contrasts were stronger than expected under random treatment assignment across the complete accumulation profile, whereas a species-allocation null model indicated non-random accumulation-dependent richness structure primarily in ungrazed plots. Functional responses differed from taxonomic responses: point estimates indicated higher temporal variability of functional richness in grazed plots and higher temporal variability and temporal increase of community-weighted mean trait displacement in ungrazed plots, but sampling uncertainty around these functional treatment contrasts was substantially greater. An alternative permanent-plot bootstrap produced the same overall uncertainty pattern. Rarefaction profiles for repeated biodiversity surveys therefore provide a reproducible way to analyse where, along sample-based accumulation curves, biodiversity change is variable, directional or both.