Disentangling the processes lead to biotic homogenization is important and will guide conservation efforts. Temporal turnover in species composition (i.e., β-diversity) and changes in local α-diversity are two distinct processes drive biotic homogenization, but these effects may be masked by invasive dominance. After removing invasive dominance, we compared the changes in species diversity and species distribution patterns before and after invasion aims to measure the relative contribution of species turnover and local diversity changes to biological homogenization. Invasive dominance indeed had an important contribution to biotic homogenization, and the effects increased with an increase in the invasive species number. Species composition of native plots was not significantly different from invaded plots when removing invasive species. Invasion changed native species distribution patterns and promoted the wide spread of most species, which causes an increase in local richness and community evenness. We highlight that species redistribution combined with invasive dominance but not species turnover promotes biotic homogenization, especially at early invasion stages. Avoiding the ecological impacts of biotic homogenization following invasion will require much stronger proactive management to prevent invasive dominance as well as increase monitoring at the early stages of invasion.

(a) Data sources

This dataset was obtained from 300 5 × 5 m plots in the Brazilian Cerrado [22]. The main habitat types are the shrub-dominated habitat (campo sujo) and grass-dominated habitat (campo úmido); both types are progressively colonized by invasive species. In total, 114 plots were sampled in the campo sujo habitat, and 186 plots were sampled in the campo úmido habitat.

In each plot, a systematic survey was conducted to detect all grass layer species (incidence data). Eighty plant taxa that belong to 17 botanical families, including 76 native species and 4 invasive species, were recorded. The invasive species included one pine (Pinus spp.) and three African grasses: Urochloa decumbens, Melinis repens and Melinis minutiflora [23]. The number of invasive species per plot varied from 0 (native species only) to 4 (4 invasive species) and represented an invasion-intensive gradient.

(b) Data analysis

Biotic homogenization is quantified as taxonomic homogenization [7] based on the pairwise species dissimilarity (Jaccard dissimilarity) between invade plots using species presence or absence [3]. Species turnover is computed as the change in the pairwise species dissimilarity between native and invade plots using species presence or absence (Jaccard dissimilarity). Local diversity is measured using species richness (species number in each plot) and species evenness (Shannon index). Total invasive effects can be quantified by the change in local diversity or species composition before and after invasion. The dominance effects of invasive species can be quantified by estimating the difference among only native species and contrasting with invasive communities where invaders are moved. Species turnover can be further assessed by partitioning β-diversity into the two processes of species replacement and species loss [24].

Species distribution pattern changes were evaluated via the relative consistency of the loss or addition of individual species before and after invasion [25]. We calculated the proportion of plots, where each species was a new colonist, and the proportion of loss of species in invaded plots, and related these species frequencies to the native plots. The stable distribution of species before and after invasion displayed a line with a slope of 1 (X=Y), that is, its distribution range was not changed. Species above the X=Y line indicate that they are widespread in their distribution range after invasion, while those under the X=Y line indicate a narrowed distribution range.

We performed Kruskal-Wallis tests [26] to compare the differences in local diversity and species composition between groups with a different number of invasive species (e.g., 0, 1, 2, 3 or 4). Plots with three or four invasive species were grouped in each habitat due to their smaller numbers. All analyses were implemented using R [27].