Discussion

Our work highlights how abundance-based diversity indices can obscure changes in community assembly. We quantified this effect by measuring rarity shifts, and showing how this effect can counteract more familiar sources of diversity change such as selection (Figure 1). This insight extends the observation that richness based observations of diversity change can obscure changes in communities (Chase et al. 2019, Dornelas et al. 2019); (Urban 2015, Urban et al. 2016). Our results indicate that strong changes in relative abundances can lead to large shifts in species’ rarity. These shifts produce the illusion of negligible diversity change in dynamic communities.
Rarity shifts tended to oppose increases in selection and decreases in immigration (Figure 4). Increases in diversity from selection and decreases in diversity from rarity shifts oppose overall diversity change (Figure 1). This mismatch is commonplace across all studies (Figure 2). We have shown that rarity shifts are commonplace across many datasets (Figure 3), particularly in studies where changes in relative abundances are high studies such as studies experiencing environmental disturbances (Webb and Scanga 2001, Thorn et al. 2016) or regeneration (Venturoli et al. 2011). Additionally, studies of annual plant communities in the Arizonan desert by (Ernest et al. 2009) and (Lightfoot 2011) experienced strong rarity shifts because measurements matched or exceeded the seasonal lifespan of the community, meaning diversity measures reflected the high mortality and rapid emergence of plant species. Rarity shifts were uncommon in long-term forest plots where changes in community composition and diversity are slow (Bradford et al. 2014). Nevertheless, we elect to analyse the data using the observations, as they were collected, even if, in retrospect, the sampling periods used may have been comparatively brief. This choice is likely conservative, as picking longer intervals would likely increase shifts in relative abundances increasing the importance of rarity shifts. By analysing the sampling intervals in the original data, our results illustrate how gradual changes in diversity sometimes imply rarity shifts and sometimes indicate negligible overall change.
The observations in the previous paragraph highlight the benefits of focusing on diversity change in plants. Plants are sessile which lowers the chance that descendants of one species will be confused with immigrants of the same species. Long-term vegetation plots are available in many model systems. As a result, data in our study represents 5 continents - North America, South America, Asia and Australia and Europe. It is of course possible that other sources of change dominate in other taxa. We suspect for example that immigration will be more important in other systems.
Our results are one of several recent applications of diversity partitions. These include analyses of why empirical observations show seemingly slow shifts in diversity over time (Dornelas et al. 2019), how to quantify the effects of biotic homogenization (Tatsumi et al. 2020, Tatsumi et al. 2021, Godsoe et al. 2022), and how to relate biotic interactions such as competition to shifts in diversity (Godsoe et al. 2021). We have added to this literature by showing that rarity shifts, a mechanism derived from partitioning (Godsoe et al. 2021), is an important source of diversity change.
For generations, ecologists have sought to understand the relationship between changes in diversity, ecosystem functioning, and the fates of individual species (MacArthur 1965, Blowes et al. 2019). This debate has become more urgent as we have confronted our limited ability to explain shifts in diversity in nature (McGill et al. 2015, Socolar et al. 2016, Urban et al. 2016, Vellend et al. 2017). To help resolve this problem, we have shown how abundance-based diversity indices can be influenced by shifts in species rarity. We have provided examples from individual plots that highlight how these rarity shifts can obscure selection (Figure 1). We have shown rarity shifts are comparable in magnitude to selection, and immigration across the 15 studies (Figure 2 and Figure 3). Finally, we have shown rarity shifts can oppose selection and immigration (Figure 4). Our results highlight how abundance based diversity measurements can give a disconcertingly slow impression of diversity change, even in communities with rapid shifts in relative abundances. Therefore, reliable predictions of diversity change will require a more nuanced understanding of the role of rarity shifts
Table 1: Summary of the 15 long-term vegetation plots analysed from the BioTIME database.