Management of invasive alien plants is an increasing problem throughout the world. In some cases native rare or protected species can appear or even prefer habitats dominated by invasive alien plants, which raises questions about the optimal treatment of such areas. Erythronium dens-canis in Hungary is a protected species which only have several occurrences in the country and a number of these populations situated in Robinia pseudoacacia stands developed after harvesting native forests. In this study a total of five populations of E. dens-canis were surveyed between 2020 and 2022 in southwestern Hungary examining and comparing the ongoing demographic changes under native and Robinia stands by monitoring individual plants. Two populations were situated in forests composed of native tree species, two in Robinia pseudoacacia-dominated stands and one in a Robinia-native tree species mixed stand. We categorized the plants into five age-state categories: dormant, seedling, juvenile, vegetative adult, and reproductive adult. We found some considerable differences (e.g. leaf size, reproduction rate) between the populations situated in native and in Robinia stands, whereas the population in mixed forest showed intermediate character in most examined factors. Based on our results, R. pseudoacacia have a significant effect on the phenology and life history of E. dens-canis, and this effect is greater with higher proportion of R. pseudoacacia in a forest stand where the E. dens-canis occurs.

Endangered species, habitat transformation, Hungary, invasive alien species, population dynamics

One of the most problematic aspects of invasive alien plant species is their potential ability to transform ecosystems in which they are introduced to (Richardson and Rejmánek 2011). They can affect plant functional traits (Sitzia et al. 2018), influence community structure and composition (Daehler and Strong 1994; Hejda and Pyšek 2006; Gaertner et al. 2009; Nascimbene and Marini 2010) together with altering soil characteristics and nutrient cycling (De Marco et al. 2013; Medina-Villar et al. 2015). Biological invasions can lead to homogenization of native communities (McKinney and Lockwood 1999; Rooney et al. 2004), loss of biodiversity, including altered ecosystem functioning (Pimentel et al. 2001; Didham et al. 2007; Vilá et al. 2010).

In Hungary, Robinia pseudoacacia L. (black locust or false acacia) is one of the invasive alien plant species with the greatest impact on natural ecosystems (Mihály and Botta-Dukát 2006). R. pseudoacacia belongs to the legume family (Fabaceae), it is a light-demanding pioneer species native to North America. In its native range it rapidly colonises forest gaps and is gradually replaced by other tree species after 15–30 years. In contrast, in secondary habitats its populations can persist for longer periods, displacing native communities (Cierjacks et al. 2013) and therefore R. pseudoacacia is considered an invasive species throughout Europe, which resulted in the inclusion of this species in national blacklists and other lists summarising alien species (Norway: Gederaas et al. 2012; Czech Republic: Pyšek et al. 2012; Pergl et al. 2016; Germany: Seitz and Nehring 2013; Italy: Celesti-Grapow et al. 2009; Switzerland: FOEN 2010).

It was one of the first American tree species to be introduced to Europe in the early 17^th century (Vadas 1914; Vítková et al. 2016) due to its numerous economically positive properties (Vadas 1914; Göhre 1952; Straker et al. 2015). Since then, R. pseudoacacia has played a major role in forest management in Hungary, resulting in a significantly higher proportion of black locust forests than in other Central European countries (Vítková et al. 2017). The share of R. pseudoacacia in the managed forested areas is the highest (nearly 460 thousand hectares, about 24%; KSH 2023) of all tree species, and its area is still increasing (KSH 2023).

Several authors have shown that the herbaceous level of R. pseudoacacia-dominated forests differs significantly from that of native forests in Europe (Wendelberger 1954; Montagnini et al. 1991; Peloquin and Hiebert 1999; Von Holle et al. 2006; Taniguchi et al. 2007; Vítková et al. 2016). Under the canopy of R. pseudoacacia, conditions are more favourable for shade-tolerant and nitrophilous species (Hruška 1991; Cierjacks et al. 2013; Vítková et al. 2016).

In the first half of the growing season, two different phenological aspects of the herb layer are observed in R. pseudoacacia-dominated forests (Vítková et al. 2017). In early spring, before the appearance of Robinia leaves (March-April) geophytes and ephemeral annuals are observed. In late spring (May-June), shade-tolerant annuals, common nutrient-demanding plants and grasses appear. In the second half of the growing season, annuals and geophytes disappear and the herb layer often dries out (Vítková et al. 2017).

As agricultural intensification in Europe resulted in substantial loss of natural habitats (Tilman et al. 2001), there has been a gradual selection towards species that can survive in secondary habitats or on the edge of cultivated fields, especially when the management is extensive (Perrino et al. 2014). This process, although in a slower pace, is still in progress, and therefore the conservational value of semi-natural habitats is gradually increasing, as for example for annual meadows of the Thero-Brachypodietea, that give refuge to some rare and endangered plant species (Brana et al. 2014; Perrino et al. 2022, 2023), and are considered a priority habitat of the Directive 92/43/EEC. In Hungary, several rare and endangered species occur in Robinia-dominated stands, such as Erythronium dens-canis L., Crocus reticulatus Steven ex Adams or Sternbergia colchiciflora Waldst. Et Kit. (Bagi et al. 1998; Pacsai et al. 2022). As E. dens-canis only has a few large populations in Hungary and a number of these are situated in Robinia forests, the better understanding of the species biology and ecology in this specific situation is essential to plan suitable treatments of these habitats. Relatively little is known about the population biology of Erythronium species in general, a few long-term studies have been carried out so far with E. japonicum Decne. (Kawano et al. 1982) and E. americanum Ker Gawl. (Holland 1981), and two study concentrated on E. dens-canis, one in Italy (Pupillo and Astuti 2017), and one in Ukraine (Kricsfalusy et al. 1995) where the detailed morphological and demographic properties of multiple populations were monitored through five years.

In our study we started a long-term monitoring of five E. dens-canis populations occurring in forests with different compositions: natural forests consisting of native tree species, intermediate, mixed and semi-natural, Robinia-dominated stands to follow the demographic and structural changes taking place in each population. During the first year of the study we noticed considerable differences in phenology and demography of the E. dens-canis plants in native and in Robinia stands (significantly different-sized individuals in same age-states, different ratio of flowering and pollination), which prompted us to expand our study with more, intermediate type sites to investigate these differences between populations situated in these two types of habitats in more detail.

The number of individuals present at a sample site showed notable changes between years (Table 1), which is partly due to individuals which did not appear aboveground in some years (we found such plants in all sites and in all years after the baseline survey), but more influenced by the high fluctuations in the number of seedlings each year. In 2020 there were no seedlings present in the surveyed quadrats, we found only a very few around the sample sites. In contrast, at most sites, in 2021 a high number of seedlings appeared which was followed by a lower, but still considerable amount in 2022. Only the NR site showed a different trend, where the number of seedlings was higher in 2022 (Fig. 1).

Figure 1.

Population structure at each sample site between 2020 and 2022, seedlings included.

With the highly fluctuating seedlings category omitted, the population structure at each sample site during the three years show some uniform trends (Fig. 2). The fraction of juvenile individuals gradually increased at all sample sites during the study years. This was mainly caused by the substantial increase in the number of juvenile individuals, but in the case of N1, N2 and NR the decrease in the numbers of vegetative and reproductive adult individuals also contributed to this rearrangement in population structure. At sites with Robinia pseudoacacia dominance (R1, R2) the proportion of reproductive adults in the population and the ratio of vegetative/reproductive adults were notably higher than at sites with native tree species (N1, N2) for each year. The population structure of E. dens-canis population in native-Robinia mix stands (NR) was similar to native sites (N1, N2) in 2021, while in 2022 it was close to Robinia dominated sites (R1, R2).

Figure 2.

Population structure at each sample site between 2020 and 2022 with the exclusion of seedlings.

Although seedling lengths were quite similar in 2021 and 2022 at each sample site (Fig. 3), we found significant differences between different sites in both years. Sites separated into two groups by statistical analyses (one-way variance analysis followed by Tukey-tests), with the R1 and R2 sites together with NR forming one (p = 0.201 in 2021 and 0.177 in 2022) and N1 and N2 sites forming the other in 2021 (p = 0.966). In 2022 the latter group consisted of only N1 since the absence of seedlings at N2 in that year (Table 2).

Figure 3.

Length of seedlings at each sample site between 2021 and 2022.

Leaf areas of juvenile plants showed a similar decreasing trend in all sample sites during the three years (Fig. 4), most remarkably in R1. Although in 2021 and 2022 there were marked differences between the maximums of leaf areas, their means were much closer to each other (Table 2).

Figure 4.

Leaf area of juvenile (A), vegetative adult (B) and reproductive adult (C) individuals at each sample site between 2020 and 2022.

In the case of adult individuals (both vegetative and reproductive), their leaf areas were the largest at R1 and R2 sites, followed by NR, while the two sites with natural habitats had the smallest leaves in all three years. The extent of these differences varied between years (Table 2).

In all three sample sites which were surveyed over the three years, in 2022 we found adult individuals (10 vegetative and 2 reproductive) which were not recorded before, which suggests that E. dens-canis could become dormant for at least two years. Transition matrices also reveal that all age-states are prone to dormancy (Suppl. material 1). At sites with Robinia, reproductive plants had much higher tendency to flower again in subsequent years (32–86%) and juvenile plants developed into vegetative adult category in higher percentage. In terms of recruitment we did not observe notable trends as the rate of recruitment at each site varied greatly in some cases. The overall growth rates (λ) of the populations also show some differences between sites (Table 3). During the first transition (2020–21) the one site with Robinia cover (R1) had the lowest growth rate by far (0.792) of the three surveyed sites, the two populations situated in native forest stands (N1, N2) had a λ close to 1 which is a characteristic of stable populations. However, during the 2021–2022 transition, this trend reversed, the two populations under Robinia (R1, R2) had the highest growth rate, while the ones under native stands (N1 and N2) had the lowest (still close to 1) and the mixed stand (NR) had a λ between the values of these two groups.

Between 2020 and 2022 the number of individuals surveyed in the quadrats at sites with native tree species were more constant than at sites with R. pseudoacacia. These differences were mainly caused by the more pronounced recruitment in some years at the latter sites. Besides this difference in most sites the gradual increasing proportion of juvenile individuals in the populations was observed. Comparing the demographic characteristics of the studied populations with literature data we found that at the N1 and N2 site the demography of the Erythronium populations is very similar to what Kricsfalusy et al. (1995) described as “left-sided”, which means the dominance of young individuals, the highly dynamic seedling category and marginal proportion of adult plants in the populations. The populations situated in mixed forests showed somewhat similar trends, but with the growing proportion of Robinia in the forest stands (NR, R2, R1) the demographic distributions of the Erythronium populations gradually changing towards the “right-handed” state described by Kricsfalusy et al. (1995) with the difference that the seedling category is also significant and highly fluctuating. Despite the fact that prolonged dormancy (not producing aboveground shoots during one or more growing seasons even though the plant is still alive) has been reported in several genera of the lily family (Tyler and Borchert 2002; Delvallée et al. 1990; Miller et al. 2004; Tatarenko 2019), we found no mention in the literature about the observation of this phenomenon regarding any species of the genus Erythronium. We found only mention of ‘senescent’ plants by Kricsfalusy et al. (1995) as old, dying bulbs without leaves. In contrast, at all sites in both years after the baseline survey we found individuals from multiple age-states which were not present aboveground in the previous one or two years. The relatively high number of such plants indicates that this phenomenon is most likely not only the result of occasional damage occurring to plants, but a natural characteristic of the species.

The average size of adult individuals was significantly greater in the populations under Robinia than at the sites in native forests in all three years. Such a difference would hardly be explained by the location, exposure or geology of the sample sites, and it is therefore assumed that differences in the composition of the forest stands in the sample sites may be the cause of this phenomenon. It is known that R. pseudoacacia can significantly increase the amount of nitrogen available for uptake by Rhizobium bacteria (Rice et al. 2004), often resulting in a significant increase in the leaf area of species in its understory (Guo et al. 2021).

Although the mean of leaf areas of juvenile plants decreased in all areas throughout the three years, the lower limit of the leaf area of reproductive individuals did not change as much. Thus, even smaller plants became adults in each following year, which could be the result of gradual environmental changes or just a coincidence in weather patterns.

One transition does not tell much about recruitment or growth rate of the population (Crawley 1990), as these values are highly variable, likely depending on environmental factors among others as well. However, it is notable that although the λ value of the one population under Robinia (R1) during the 2020–21 transition was the lowest among all sites, in the next transition the λ values of Robinia stands were the highest by far. The populations situated in native stands had a similar growth rate (close to 1) in both transitions. As in 2020 we could not find any seedlings in any surveyed quadrats, and in 2021 and 2022 there were numerous in some sites, these great differences in recruitment rates likely caused by annual variances in weather. Great differences between recruitment rate at different sites have been observed in the case of this species (Kricsfalusy et al. 1987) but in our study these contrasts cannot be attributed to differences in geographical conditions as all our study sites are situated in the same landscape and the NR site which is just 50 metres in distance from the N1 site shows the highest similarities with the R2 site (9 km away). The low λ value of R1 in the 2020–21 transition means that the population without recruitment is rapidly declining even with the mortality rates omitted during the calculation of deterministic growth rates. In the 2021–22 transition the presence of recruitment made λ considerably higher than the 2020–21 value, which suggests that in Robinia-dominated habitat type E. dens-canis is highly relying on recruitment and these populations could be characterized by a more dynamic demography which is generally uncommon in long-lived herbaceous species (Eriksson 1989), but not unknown for E. dens-canis (Kricsfalusy et al. 1995).

In contrast with the R1 and R2 populations, the E. dens-canis populations situated in native forests have more stable demographical characteristics, their growth rate was close to 1 with or without recruitment. The Robinia-native tree species mixed stand (NR) showed an intermediate growth rate, which also suggests that the greater the ratio of Robinia in a stand the greater is its effect on the Erythronium population as well. Since these values were also calculated without the mortality rates of most age-states, the growth rates are likely lower. It could easily change the growth rate of populations in native forest stands from stable or slowly growing to a declining category.

Nitrogen pollution originating from agricultural activities (fertilization, production of leguminous crops) is a common potential threat to biodiversity, especially to endangered species (Hernández et al. 2016). The habitat-transforming capability of R. pseudoacacia is already known, which is mostly caused by increasing nitrogen input into the ecosystem (Buzhdygan et al. 2016; Vítková et al. 2016), however its effect on endangered plant species is a less studied topic. Our results agree with previous studies (Buzhdygan et al. 2016) that R. pseudoacacia accelerates environmental processes, as E. dens-canis populations under R. pseudoacacia show much more volatile deterministic growth rates and higher turnover compared with populations situated under native tree species and this effect was greater in sites with greater ratio of Robinia in the tree cover. Such variations in the life cycle of plants may support the dynamic heterogeneity of populations which in turn ensures their stability in different environmental conditions and management regimes (Kricsfalusy 2016). In most aspects, where we found differences among populations, the extent of these differences was in correlation with the proportion of R. pseudoacacia in the forest stands, as the population at the NR site often showed intermediate characteristics between the native and Robinia-dominated stands.

We would like to express our thanks to all those who helped us in our fieldwork: Ádám Tarr, Fruzsina Hajdu and Friderika Szabó.