Study species and area
Rumex alpinus is distributed throughout European high mountains, including the Apennines and mountains of the Balkan Peninsula and Caucasus (Meusel et al. 1965; Tutin et al. 1993). The plant has also been introduced to some European countries, including Great Britain, Scandinavian countries and areas of the Czech Republic (Št’astná et al. 2010), where it is often considered as alien invasive species (Hejda et al. 2009; Hejda and de Bello 2013; Šurinová et al. 2018). It is suggested that the natural habitats in central European mountains from which R. alpinus colonised secondary zoo-anthropogenic habitats are moist brushwood communities along mountain streams in the upper montane zone and open tall herb communities of the class Mulgedio-Aconitetea (Stachurska-Swakoń 2009). R. alpinus is a rhizomatous perennial with a horizontal rhizome growing at a depth of 5–10 cm. Because a new segment of the rhizome develops each year, the plant’s growth and age can be determined from the number of segments (Klimešová et al. 2013). It has been reported that a rhizome can be as much as 120 cm long and can even persist for 35 years (Št’astná et al. 2010; Št’astná et al. 2012). Each year, three to five big leaves grow from a rhizome segment, with petiole ranging from 70 to 80 cm and lamina up to 50 cm long and 20 cm wide, creating a dense canopy of robust, leafy shoots with a density of 3–8 m−2 and height of 30–200 cm. They can also produce 1500–5000 fruits (Št’astná et al. 2010). R. alpinus is a nitrophilic plant species associated with moist, nutrient- and base-rich soils (Rehder 1982; Bohner 2005). The plant is a strong competitor capable of forming species-poor stands, which are known to persist for several decades (Ellenberg 2009). Vegetation dominated by R. alpinus was described in phytosociology as the Rumicetum alpini Beger 1922 plant association, but a number of syntaxonomical ranks were later distinguished (Kliment and Jarolímek 1995; Stachurska-Swakoń 2009). Št’astná et al. (2010) provides a comprehensive review of the biology and ecology of R. alpinus in central Europe.
This study was conducted in the Tatra Mountains (Fig. 1) in the protected area of Tatra National Park in southern Poland. Five vertical vegetation belts are present in the Polish Tatra Mountains: the lower montane belt reaches up to 1250 m a.s.l.; the upper montane belt from 1250 to 1550 m a.s.l.; the subalpine belt from 1550 to 1800 m a.s.l.; the alpine belt from 1800 to 2300 m a.s.l.; and the sub-nival belt from 2300 m a.s.l. up (Mirek and Piękoś-Mirkowa 1992). The mean annual temperature decreases from approximately +6 °C at the foothills of the Polish Tatras (about 900–1000 m a.s.l.) to −2 °C at 2200 m a.s.l., and − 4 °C at the highest peaks. The mean annual sum of precipitation measured in the foothills at the weather station in the town of Zakopane (at an elevation of 844 m a.s.l.) is 1138 mm, and, at the weather station on the Kasprowy Wierch peak (at an elevation of 1991 m a.s.l.), 1876 mm (Hess 1996).
The beginning of pasturage management in the Tatra Mountains dates back to the thirteenth or fourteenth century. In the nineteenth and first half of the twentieth century, pasturage practices in the Tatra Mountains were intensive, with large flocks of sheep and cows causing serious devastation to mountain meadows, glades and surrounding forests (Śmiałowska 1962; Kołowca 1962). In 1947, with the establishment of a separate administrative unit, the Tatra Park, the first actions were undertaken to reduce the number of livestock in the Polish Tatras. After the establishment of the Tatra National Park in 1954, further pasturage limitations took place in the Tatra Mountains (Śmiałowska 1962). In subsequent years, pasturage was gradually ceased on alpine meadows and pastures, as well as in most glades within limits of Tatra National Park; from 1981 on, only a limited “cultural pasturing” was carried out in selected glades at lower elevations (Mielczarek 1984). A concise history outline on pasturage in Tatra National Park (with selected references) is provided by Stachurska-Swakoń (2008).
Vegetation sampling
Data was collected in 2015 from 20 glades and pastures located in the vertical zone of the lower and the upper montane vegetation belts, where formerly pasturage took place (Fig. 1, Table 1). At these sites, we examined 100 vegetation plots with R. alpinus of 4 m2 and 100 control plots of the same dimension. Plots with R. alpinus were selected to cover a gradient from very low (minimum 5%) to very high cover of the invader and to represent spatial distribution of the species within the study area. Control plots were located spatially very close to the paired plots with R. alpinus; most often they were adjacent or within distance of 2 m, with highly similar habitat conditions. The cover of the invader in the control plot was allowed to be at a maximum of 1%. Examined plots were distributed at elevations ranging from 991 m to 1497 m a.s.l. (interquartile range 1110–1295 m a.s.l.).
Table 1 List of localities in the Tatra National Park from which R. alpinus plots were sampled Species composition was surveyed for each plot, and phytosociological relevés were made using Braun-Blanquet’s method and a six-point plant cover scale (Kent 2012), whereas the abundance of R. alpinus was estimated in terms of percentage cover (Supplementary Table 1). The nomenclature of vascular plants followed Mirek et al. (2002). In the space-for-time substitution approach, 55 plots with R. alpinus cover of at least 50% (dominance of the species) were selected (hereafter, referred to as invaded plots) together with their paired control plots (hereafter, referred to as uninvaded plots); 36 sites were located in the lower montane belt and 19 in the upper montane belt. Analyses of diversity and vegetation composition in plots differing in the extent of R. alpinus cover were performed for all 100 plots with R. alpinus.
Data analysis
For each plot, we calculated the plant species richness (S) as the number of vascular plant species per plot, Shannon diversity index (H´) as H´ = − ∑pi × ln pi, Simpson diversity index (D) as D = 1 – ∑pi2 and Pielou’s evenness index (J) as J = H´ / ln S, where pi is the proportion of species i per plot (Hill 1973). For diversity calculations, Braun-Blanquet cover-abundance values +, 1, 2, 3, 4 and 5 were transformed to 0.1, 2.5, 15.0, 37.5, 62.5 and 87.5, respectively (Wildi 2010). R. alpinus was not included in the data set used in the calculation of the diversity indices in order to evaluate the impact of the invader on the remaining species. To assess the impact of the invasion by R. alpinus on resident vegetation, we compared invaded and uninvaded plots with respect to species richness, diversity and evenness measures. This comparison was made also separately for sites located in the lower and the upper montane belts. As statistical distributions of most of these diversity indices deviated from the normal distribution significance of differences in diversity measures between invaded and uninvaded plots, they were tested by the Wilcoxon signed-rank two-sided test. To examine the impact of an increasing invader cover on the diversity of plant communities, simple linear regression and quadratic regression were applied. In this approach, diversity indices were modelled, as a response variable, by percentage cover of R. alpinus in a plot. The effect of R. alpinus cover on species diversity was tested using simple linear regression: Y = a + b1 × (R. alpinus cover) and regression with a quadratic term added: Y = a + b1 × (R. alpinus cover) + b2 × (R. alpinus cover)2, where a denotes an intercept, and b1 and b2 denote regression coefficients. Linear and quadratic models were fitted to R. alpinus cover gradient, and the best model was selected based on an analysis of variance (ANOVA) test, where a significant F statistic expressed significant improvement of the linear model once the quadratic term was added (Dalgaard 2008). In analogous way, we examined whether the impact of R. alpinus on diversity of invaded vegetation differed across elevations; in this analysis diversity indices were modelled by elevation in simple linear and quadratic regressions, both run separately for invaded and uninvaded plots.
The impact of invasion by R. alpinus on species composition was examined with indirect gradient analysis, Detrended Correspondence Analysis (DCA), performed for 55 pairs of invaded-uninvaded plots. In DCA analysis, rare species occurring only in one or two plots (relevés) were excluded from subsequent analyses (Legendre and Legendre 2012). Also, R. alpinus was excluded from this analysis because the aim of the analysis was to examine changes in the composition of the remaining species of resident vegetation. Species abundances were expressed in the original Braun-Blanquet scale (+, 1, 2, 3, 4, 5), with transformation of ‘+’ into a value of 0.1. This approach was applied to down-weight the influence of accidental high abundances of species on the ordination result. To understand the ecological shift in the species composition of invaded plots in relation to uninvaded plots, we superimposed on the ordination diagram environmental factors, expressed by elevation of investigated sites and mean Ellenberg’s indicator values (EIVs), fitted post hoc to the DCA ordination. EIVs for light, soil nutrients, soil moisture and soil acidity (Ellenberg et al. 1992) were assigned to all species, if available, and their mean values for each plot were calculated. Only these factors, which proved to be significant in the permutation test (n = 999) at the 0.05 significance level, were presented on the ordination diagram. The effect of the invasion on the species composition was assessed by comparison of the DCA site scores of the pairs of invaded-uninvaded plots along the first and the second ordination axes, applying the Wilcoxon signed-rank test.
All statistical analyses were performed in R version 3.4.2 (R Core Team 2017). To compute diversity indices and perform DCA analysis, the Community Ecology Package ‘vegan’ was used (Oksanen et al. 2017).