We sampled populations of 10 native (Calamagrostis epigejos, Cirsium arvense, Cirsium heterophyllum, Cirsium oleraceum, Filipendula ulmaria, Petasites hybridus, Phalaris arundinacea, Rubus idaeus, Tanacetum vulgare, and Urtica dioica) and nine invasive dominant plants (Aster novi-belgii agg., Heracleum mantegazzianum, Impatiens glandulifera, Lupinus polyphyllus, Reynoutria japonica, Reynoutria × bohemica, Rumex alpinus, Solidago canadensis, and Telekia speciosa; Suppl. material 1). We sampled plots of 4 × 4 m in size located within populations of studied species across the Czech Republic (Suppl. material 2) that ranged from hundreds to thousands of m² in size (see Hejda et al. 2021 for details on the sampling scheme); the populations were selected so as to include stands with high and low dominance of the target dominant species. Low dominance referred to 0–25% cover of the target species, and these low-cover plots were used as controls. On the contrary, high dominance encompassed >50% of the dominant species cover (see Hejda et al. 2021 for details). We then estimated the local impacts of these dominant species on species richness and Shannon diversity index of invaded communities, as well as on species composition.

To detect the changes in species composition associated with the dominant species, we calculated the Jaccard dissimilarity index, which is based on incidence data (Jaccard 1900) and regarded as robust to taxonomic error as well as both numerical and geographical undersampling (Schroeder and Jenkins 2018).

We calculated the Jaccard dissimilarity index (β_jac) at the (i) population-level (= dissimilarity of plots within populations); and (ii) between-population level (= dissimilarity of plots between populations), using the beta.pair function (index.family=”jaccard”) of the betapart package (Baselga 2013; R Core Team 2022). The values of the index range from 0 (maximum similarity or lowest dissimilarity) to 1 (minimum similarity or highest dissimilarity). The following formula was used:

${\text{β}}_{\text{jac}}=\frac{b+c}{a+b+c}$

where a is the number of species in common between two sites, b is the number of species unique to the site with the lowest number of species, and c the number of species unique to the site with the largest number of species.

Total Jaccard can be partitioned into turnover (β_jtu) and nestedness component (β_jne):

${\text{β}}_{\text{jac}}={\text{β}}_{\text{jtu}}+{\text{β}}_{\text{jne}}=\frac{2b}{2b+a}+\left(\frac{c-b}{a+b+c}\right)\left(\frac{a}{2b+a}\right)$

Nestedness refers to changes in species richness, in which the site with the lowest richness represents a subset of species of the richest site, and turnover, which refers to species replacement from site to site (Baselga and Orme 2012) (see Fig. 1 for a schematic representation of turnover and nestedness).

Figure 1.

Schematic representation of the partitioning of the Jaccard dissimilarity index into turnover and nestedness components (see text for the formula and more information). Turnover refers to the gain and loss of species (species replacement) between the areas (e.g. high dominance plots of a particular dominant species), whereas nestedness refers to the cases where the plot with the lowest number of species represents a subgroup of species of the richest plot/site. In the scheme, the species in blue are unique to plots A and C, the species in black are shared between both plots and the species in orange are unique to plot B.

We estimated the population- and between-population level impacts as differences in similarity between the plots with low vs. high dominance of the selected dominants, assuming that this represents the homogenizing effect of the dominant species (see Fig. 2 for a schematic representation of our sampling design and how the diversity metrics were calculated at distinct spatial scales).

Figure 2.

Scheme of our sampling design and how the impacts were estimated. The plot level is marked with blue arrows, the population level is marked with dashed green arrows, and the between-population level with the dashed orange arrows. In each population, the median value across all high-dominance plots and the median value across all low-dominance plots were computed; these median values of high and low-dominance plots were then compared in the analyses.

At the population level, we calculated Jaccard, turnover and nestedness amongst all high-dominance plots and amongst all low-dominance plots to each population of each species. We then recorded the median value in each population with high-dominance plots and compared them with the corresponding median values of the low-dominance plots.

Further, we compared the total Jaccard, nestedness and turnover of high-dominance plots with native vs. alien dominants to compare their homogenizing effect. In the case of the population-level impacts, we also calculated the Jaccard dissimilarity, nestedness and turnover between the high- and low-dominance plots within each population to express the population-level impact on species composition, assuming that the lower similarity between the low and high dominance plots (within populations) shows a larger impact on species composition. Here, to tackle the challenge of comparing different numbers of high and low dominance plots, given that the high vs low combination itself is substantially higher than high vs. high and low vs. low, we combined all high-dominance plots of a population as a single “high dominance plot”, and similarly for a single “low dominance plot”. Thus, in each population of each species, we recorded a single direct dissimilarity distance of high vs. low-dominance plots.

To express the plot-level impacts, we used LMM regression models to relate the plot-level species richness and Shannon diversity to the cover of selected dominants, accounting for the identity of dominant species and their populations (nested in “dominants”) by setting these as random effects. We quantified the plot-level impacts as the slope/intercept ratios of the corresponding LMM regression models, accounting for the a priori different species richness and diversity of different types of vegetation (see Hejda et al. 2021 for details on data processing and analyses).

Further, we used LMM models to compare the similarity, nestedness and turnover of low- vs. high-dominance plots at the population- and between-population levels and the effects of native vs. alien invasive dominants. Further, we used parametric and non-parametric correlations to test the relationships between the effects at different spatial levels.

At the population level, high-dominance plots with both native and alien dominants taken together showed higher nestedness and lower turnover compared to all low-dominance plots (p = 0.018 and p = 0.002, resp.). In other words, sites with high dominance were linked to a higher degree of nestedness (which in turn is related to species losses), consistently for both alien and native dominants (see Table 1).

Considering all plots with native dominants (high and low dominance plots taken together) vs. all plots with alien dominants, we found that the former were more similar to each other but also had a lower nestedness than plots with alien dominants (p = 0.039 and p = 0.043, respectively, Table 2, see also Suppl. material 3 for the details on statistical tests).

Comparing high-dominance plots with native and alien dominants, we found that the former showed lower Jaccard dissimilarity (i.e., were more similar), whereas the latter had marginally significantly higher nestedness (p = 0.045 and p = 0.072, resp). In other words, plots with a high native dominance had more species in common than plots with a high dominance of aliens, where the species loss was stronger.

Considering the low-dominance plots, no significant differences in Jaccard, nestedness or turnover were detected between native and alien dominants. At the population level, no significant differences between the effects of native vs. alien dominants (defined as differences in dissimilarity, nestedness and turnover between the low- and high-dominance plots) were found (Table 1).

High-dominance plots with alien dominants showed higher species turnover compared to high-dominance plots with native dominants, but this difference is only marginally significant (p = 0.051, Table 2). Comparing high and low-dominance plots within the same origin of dominants, native dominants show lower levels of turnover in high-dominance plots compared to low-dominance plots, whereas alien dominants show higher levels of nestedness in high-dominance plots compared to low-dominance plots (Fig. 3). Similarly to the population level, no differences in the effects of native vs. invasive dominants, defined as the dissimilarity differences between the low- and high-dominance plots, were detected at the between-population level (Table 2).

Figure 3.

Between-population level results for all species. Each dot represents the median value across all sites of the same species at a certain dominance category (high or low); each line connects the dominance category of a species. Alien dominants (a-c) show higher levels of nestedness in high-dominance plots compared to low-dominance plots, whereas native dominants (d-f) show lower levels of turnover in high-dominance plots compared to low-dominance plots. Alien species: An: Aster novi-belgii agg., Hm: Heracleum mantegazzianum, Ig: Impatiens glandulifera, Lp: Lupinus polyphyllus, Rj: Reynoutria japonica, Rb: Reynoutria × bohemica, Ra: Rumex alpinus, Sc: Solidago canadensis, Ts: Telekia speciosa; native species: Ce: Calamagrostis epigejos, Ca: Cirsium arvense, Ch: Cirsium heterophyllum, Co: Cirsium oleraceum, Fu: Filipendula ulmaria, Ph: Petasites hybridus, Pa: Phalaris arundinacea, Ri: Rubus idaeus, Tv: Tanacetum vulgare, Ud: Urtica dioica.

No significant relationships between the impacts recorded at the plot- and either population- or between-population levels were detected (Table 3). On the contrary, strong positive relationships between the population- and between-population level impacts were found for Jaccard dissimilarity, nestedness and turnover (Table 3). These strongly significant positive relationships were identified using both parametric (Pearson) and non-parametric correlations (Spearman, Kendall; see Suppl. material 3 for the results of non-parametric correlations).

We found strong positive relationships between the plot-level impacts on Shannon diversity and population-level impacts on species composition, expressed as the similarity between the low- and high-dominance plots within each population (Table 4). We also detected a strong negative relation between the turnover at the population level (the turnover component of the population-level compositional impacts) and the impacts on the turnover of species at the between-population level (expressed as the differences in turnover between the low- and high-dominance plots). Similarly to the previous finding, the relationship between the turnover component of the population-level compositional impacts and the turnover component of the between-population-level impacts was significant when both parametric (Pearson correlation) and non-parametric methods (Spearman and Kendal correlation; see Suppl. material 3) were used.

At the population level, high-dominance plots show higher nestedness and lower turnover than low-dominance plots. In other words, taking alien and native species together, the high-dominance plots lose more species in a nested pattern, and the species replacement is lower than in the low-dominance plots. However, no significant difference in Jaccard dissimilarity between the low- and high-dominance plots was recorded at the population level, suggesting that the high-dominance plots are not necessarily less diverse and more homogenous than the adjacent low-dominance plots. Apparently, distinctive dominants can lower the local, plot-level (alpha) diversity without affecting the large-scale diversity expressed by the beta diversity indices. However, Kortz and Magurran (2019) found a contrasting pattern: the presence of aliens was associated with an increase in the local diversity, as areas with more aliens tend to have more species, but decreased the large-scale (beta) diversity, by making the vegetation more homogenous due to adding commonly shared aliens amongst the areas. A similar pattern was detected by Nobis et al. (2016): the local richness of native and alien species was positively related. However, the richness of alien species was negatively related to native beta and gamma diversity, which especially concerned red-listed species. Importantly, the fact that aliens contribute to plot-level diversity precludes the competitive exclusion of native species by dominant aliens.

The homogenizing effect of alien dominants was described for some aliens, such as the amphibious Alternanthera philoxeroides (Wu et al. 2022). The large-scale impacts of this species were context-dependent, being stronger in invaded terrestrial rather than aquatic habitats and in the northern part of the invaded range in China. In other cases, the effects of invasive aliens on native diversity were detected to be consistently negative across different spatial scales. For example, Stotz et al. (2019) detected a consistently negative effect of the invasive Bromus inermis both within and across individual grasslands in Alberta, Canada, and Boscutti et al. (2020) detected a spatially consistent negative effect of the invasive Amorpha fruticosa in northern Italy. Interestingly, Bando et al. (2022) detected a negative effect of the invasive Urochloa arrecta on both spatial and temporal beta-diversity in Brazil.

We did not find studies comparing the large-scale effects of multiple invasive and native dominants, even though there are studies comparing the large-scale impacts of invasive aliens in their native and invaded ranges – see for example Lolis et al. (2019), who detected a negative effect of the invasive Eichhornia crassipes on both alpha and beta diversity in the invaded range, China, but not in its native range, Brazil.

When comparing plots with high dominance of native species with those of aliens, the former were more similar (i.e., showed lower dissimilarity), pointing to their stronger homogenizing effect. At the same time, high-dominance plots with native dominants also showed marginally lower nestedness than their alien counterparts. The same pattern was detected for all plots merged regardless of the degree of dominance: those with native dominants are more similar but show lower nestedness than plots with alien dominants.

It needs to be stressed that the tests on the differences between high- and low-dominance plots, as well as between the native vs. alien dominants at the between-population level, are weak due to the high residual variability. This is because the data include samples with different dominants, both native and alien, and with different types of vegetation, both within- and across dominants. Inevitably, this introduces a lot of residual variability that remains unexplained by our models.

No significant differences between the low- and high-dominance plots and between the native and alien dominants were recorded at the between-population level, except that high-dominance plots with native dominants showed a marginally significantly lower species turnover. This again suggests a slightly stronger homogenizing effect of native dominants, similar to that recorded at the population level. This is also in line with recent evidence that areas across the globe with alien plants have higher levels of species replacement than areas with native species only (Kortz et al. 2023).

When the population and between-population-level impacts were defined as differences in Jaccard similarity, nestedness, and turnover between the low- and high-dominance plots, no significant relationships between the impacts recorded at the plot level and those measured at either the population or between-population levels were revealed. However, we recorded strong positive relationships between the impacts at the population- and between-population levels, and this holds for all three indices used, i.e., Jaccard dissimilarity, nestedness and turnover. This clearly shows that the impacts at the population- and between-population levels are strongly related; however, it also confirms that plot-level impacts cannot be easily extrapolated to higher spatial scales.

On the contrary, we recorded strong positive relations between the plot-level impacts on Shannon diversity and the population-level impacts on species composition. This indicates that changes in the plot-level alpha diversity are strongly associated with compositional changes. Further, the turnover component of the population-level compositional impacts was strongly negatively related to the turnover component of the between-population-level impacts, defined as the differences between the low- and high-dominance plots.

The authors have declared that no competing interests exist.

No ethical statement was reported.

The collection of data was supported by grant no. 17-19025S, the analysis and paper writing by 23-05403K (both Czech Science Foundation). PPy and AK were supported by EXPRO grant no. 19-28807X (Czech Science Foundation). All authors were supported by the long-term research development project RVO 67985939 (Czech Academy of Sciences).

Conceptualization: AK, MH, JP, PPy. Data curation: MH, AK. Formal analysis: AK, MH. Funding acquisition: JP, PPy. Field work: JS, JP, JK. Methodology: AK, MH, JP, PPy. Project administration: JP. Supervision: PPy. Validation: PPy, AK, JP. Visualization: AK, JP. Writing - original draft: MH, AK. Writing - review and editing: AK, MH, JP, JK, PPe, JS, MVi, MVo, PPy.

Alessandra Kortz https://orcid.org/0000-0002-7473-1987

Martin Hejda https://orcid.org/0000-0002-0045-1974

Jan Pergl https://orcid.org/0000-0002-0045-1974

Josef Kutlvašr https://orcid.org/0000-0001-7486-8644

Petr Petřík https://orcid.org/0000-0001-8518-6737

Jiří Sádlo https://orcid.org/0000-0001-9723-3334

Michaela Vítková https://orcid.org/0000-0002-2848-7725

Martin Vojík https://orcid.org/0000-0001-9735-5120

Petr Pyšek https://orcid.org/0000-0001-8500-442X

All of the data that support the findings of this study are available in the main text or Supplementary Information.