Preliminary Results of the Introduction of Dicotyledonous Meadow Species

Abstract

The reintroduction and introduction of native plant species is becoming more and more important in the restoration of plant communities. The study aimed to determine the possibility of predicting the effectiveness of the introduction of dicotyledonous species into impoverished patches of meadows in the landscape nature reserve in the proglacial valley of the Vistula River (Poland). Fourteen species planted into the soil from seedlings, after growing them from seeds in pots, in pure stands were assessed. Field studies were carried out in 2015–2017 on post-bog soil. Parameters of plant development and growth that were analysed included, among others, range, condition and height of shoots (vegetative and generative). Based on the biometric parameters, a statistical analysis (PCA, analysis of variance, decision tree) was performed. It was found that the range, i.e., the spread of the population, did not determine the classification of species into groups with a different nature of development after introduction. This classification was mainly determined by the plant condition in the following years after the introduction (over 3.4 on a 5-point scale), and the occurrence of generative shoots in the second year after planting. The group with the highest potential efficiency of introduction included three species: Achillea millefolium, Hypericum perforatum, Veronica longifolia. The failure of the introduction of other species resulted from their life form (two years old) and unfavourable weather conditions in the third year of study (2017), due to the high level of groundwater.

1. Introduction

Currently, permanent grassland ecosystem services are being increasingly appreciated in nature protection, in shaping the environment, landscape and culture [1]. Therefore, various measures are being taken to assess their condition, protection and restoration. According to the European Environment Agency (EEA), in the 2013–2018 report, in most part of the European countries, the conservation status of many monitored sites of semi-natural grassland communities important for Europe is in an unfavourable state of preservation—from U1 (Unfavourable inadequate) to U2 (unfavourable bad) [2]. Among them are the ones found in Poland, in the continental region, among others: Molinia meadows Molinion (habitat code 6410), Cnidion dubii meadows (6440), Lowland hay meadows (habitat code 6510) and Festuco-Bromatea (6210), the existence of which depends on extensive management. Unfortunately, many of these habitats and their species are now highly endangered. It has been shown that the most important threats to meadow and pasture communities in Europe include: abandonment of traditional utilisation (mowing and grazing), especially in the less-favoured areas (LFA)—particularly waterlogged, excessively moistened, marshes, in upland and mountain areas; and changing permanent grassland into arable land and temporary grassland, or afforestation [3,4]. Recently, due to the deepening climate change, environmental stresses have become a serious threat to these communities—mainly drought and high temperature, especially in the Pannonian and continental regions, as well as the spread of alien invasive plant species [5]. The excessive drying of the habitats, especially on the organic soils, influences stopping the peat forming process, the intensification of the decomposition process, the mineralisation of organic matter and the decreasing in physico-aquatic properties of the soils [6].

The impoverishment of floristic diversity is observed all over the world in various types of communities. This process has been progressing faster and faster since the second half of the last century. Individual species and even entire plant communities are disappearing, mainly due to habitat fragmentation, agricultural intensification, habitat degradation and climate change. The threat to the flora of Poland is growing rapidly, especially in highly urbanised regions or with intensive agriculture. The risk of extinction concerns mainly rare and endangered wild plant species. Currently, according to the World Conservation Union (IUCN), about 20,000 of the 300,000 vascular plant species are endangered. The current list of extinct and threatened plant species in Poland includes 765 plant species whose conservation status has been determined following the recommendations of the IUCN [7]), which accounts for 30% of the Polish flora. In many regions of Poland, more than 30% of the native and domesticated vascular plant species are endangered.

Taking into account the problem of the disappearance of species and even entire plant communities in Europe and Poland, the restoration of multi-species meadow communities is an important task [8,9]. Various activities aimed at restoring ecosystems and protecting biodiversity for sustainable development have been carried out around the world, in different habitats, on a larger scale, for over twenty years. Restoration of meadow communities is a difficult task because there are many factors limiting the course of this process. Inadequate moisture conditions, nutrient content and soil pH can constitute a serious problem [10]. The number of diaspores in the soil bank is greatly reduced [11]; large arable fields or highways form barriers to their spread. In addition, in soil seed banks, annual dicotyledonous species, belonging to segetal weeds or ruderal plants, are often the most numerous [12,13].

There are many techniques for establishing and restoring environmentally valuable plant communities of grasslands excluded from use, and what is more, various methods of collecting seeds directly from semi-natural meadows have also been developed [14]. Oversowings [15] and direct drillings are performed using specialised seeders. Spreading hay with diaspores is applied [16], often in combination with the removal of the top layer of soil [17,18]. Green hay spreading is used less frequently—it is possible only at small distances from donor stands [19,20]. Another method involves the production of seedlings, which is recommended especially when introducing rare and endangered species [21]. This method is also used for species the dispersal of which is limited and species with low competitiveness and, as a result, low survival of seedlings and young plants in the presence of primary sward [10]. The choice of method depends on the species, size of the area being restored and financial resources. In the current decade (2021–2030), known as the “Decade of ecosystem restoration” [22], actions taken to restore degraded and damaged meadow ecosystems are of particular importance [23].

There are three main techniques for the conservation of rare and endangered plant species: in situ, ex situ and reintroduction. Planting (introduction or reintroduction) is the creation of new populations by intentionally establishing individuals of a species in a habitat/area where that species existed but disappeared. Plant individuals or their underground parts can be planted (e.g., rhizomes, bulbs—this applies to species with good vegetative reproduction and establishment abilities) [24]. Planting of plant individuals is a relatively expensive but effective method of population restoration [25] and improvement of species richness of plant communities, as it allows the omission of particularly sensitive periods, i.e., germination and early development phases, which take place in optimal conditions in the vegetation halls, which limits competition. Once seedlings are planted, they will encounter new abiotic and biotic conditions to which they are not well adapted and may be subject to strong selection as a result. Therefore, it seems that the key to successful introductions is minimising differences in habitats and mimicking the original conditions [21].

The authors of numerous articles have been wondering what factors affect the success of reintroduction. What is more important: species characteristics, area characteristics or reintroduction technique? Kaye [26] points out that all three groups of factors interact with each other. It has been found that reintroductions are most successful in habitats ecologically similar to those in which the introduced species are the most common. In addition, the methods used to introduce plants (sowing seeds vs. planting transplants) and the choice of appropriate microsites impact the results of reintroduction [26]. Research on various methods of restoring meadow ecosystems showed that the most effective method was green hay transfer, resulting in the restored community being the most similar to the donor meadow [27]. Schaumberger et al. [20] emphasise that the choice of restoration method ultimately depends on cost and various circumstances, it is important that a large proportion of the transferred species can become established.

Reintroduction and introduction of native plant species becomes more and more important in the restoration of plant communities. This is a standard technique used in conservation and restoration of rare and endangered plant species populations. Moreover, according to Török et al. [28], this technique is often used to improve species richness or propagule availability in seed sowing or hay-transfer restored fields. This accelerates the success of the restoration of planted species population, and at the same time the entire plant community. So far, there are few studies predicting the success of the development and maintenance of plant species after introduction on the basis of their condition and morphological features. The measure of the final fate of the introduction is the ability of the planted seedlings to flower and set fruit [29], which indicates that the population is self-sufficient through the development of successive generations. Godefroid et al. [30] found that survival, flowering and fruiting rates of reintroduced plant species were generally quite low (on average 52%, 19% and 16%, respectively). In addition, they showed a decrease in the success rate of introductions over time. Planting seedlings (compared to sowing seeds), increasing the number of planted seedlings, using grafts from stable source populations and careful site preparation, including the use of fences, had a positive effect on the results of introduction [30].

Lowland hay meadows (Arrhenatherion, habitat code 6510-1) are becoming increasingly rare across the Polish landscape, located beyond the reach of river flooding, in eutrophic and mesotrophic habitats, fresh, i.e., not too wet and not too dry [31]. Extensively used, floristically rich, they usually occur in the form of small patches. Particularly valuable are the wetter fragments of these meadows with a large number of orchids. The main threat to their habitat is conversion to arable land or vegetable cultivation, as well as overseeding and intensive use.

The aim of the study was to determine the possibility of predicting the effectiveness of the introduction of meadow plant species into impoverished patches of meadows dominated by Arrhenatherum elatius (30%) and Urtica dioica (40%) in a landscape nature reserve. The working hypothesis assumed that the parameters determining the condition, growth and development of plants after the introduction of species from the dicotyledonous class indicate their survival in the meadow sward and the improvement of the natural values of the “Ursynów Escarpment” landscape reserve.

2. Materials and Methods

2.1. Study Area

The research was carried out in the area of “Ursynów Escarpment” landscape reserve (Figure 1). It is one of the twelve nature reserves located in the area of Warsaw (Poland). It was established in 1996 to protect the part of the high Vistula Escarpment (the highness of the slope in the reserve area is 5–18 m, gradient 24–60°) altogether with the meadows and peat bogs of high natural value. The area of the reserve covers 20.80 ha, and the buffer zone covers 134.6 ha [32,33]. It was found that this reserve is characterised by the presence of 126 taxa of the vascular plants [34]. The reserve serves as the migration corridor, provides the nesting or breeding sites for the amphibians, birds and mammals. It is the refuge for the invertebrates with many rare species, including daylight butterflies, such as Lycaena dispar, Polyommatus bellagrus and Cupido argiades [35].

The meadows being part of the reserve cover 12 quarters with a total area of 10.7 ha. The area was drained and managed in the years 1952–1953. In the 1970s and 1980s, intensive pasture–meadow management was carried out on that area. After the system transformation, the management was limited to one summer cut. Since 1998, the meadows have not been mowed. The studies carried out in 2013–2015 showed the simplification of the species composition, the development of herbaceous communities (from the Filipendulo-Petasition alliance), the invasive, segetal and ruderal plant species and the shrubs, mainly the willow ones [36] (Figure 2).

2.2. Botanical and Phytosociological Characteristics of Species

The research covered 14 plant species of the non-forest habitats (

Table 1. Characteristics of tested species and number of seedlings planted in 2015.

Species	Family	Life Form	Durability	Phytosociological Units	Seedlings (No.)
Achillea millefolium L.	Asteraceae	H	B	Mol-Arr Cl	48
Artemisia campestrisL.	Asteraceae	Ch	B, S	Fest-Brom Cl	17
Centaurea stoebe L.	Asteraceae	H	B, D	Fest-Brom Cl	22
Dianthus deltoides L.	Caryophyllaceae	C, H	B	Coryneph O	14
Eryngium planum L.	Apiaceae	H	B	Fest-Brom Cl	27
Hypericum perforatum L.	Hypericaceae	H	B	Fest-Brom Cl	25
Inula britannica L.	Asteraceae	H	B	Agr-Rumi All	7
Lysimachia vulgaris L.	Primulaceae	H	B	Aln glu Cl	48
Plantago lanceolata L.	Plantaginaceae	H	B	Mol-Arr Cl	17
Potentilla erecta (L.) Raeusch.	Rosaceae	H	B	Nard-Cal Cl	42
Scabiosa ochroleuca L.	Dipsacaceae	H	B, D	Fest val O	22
Tragopogon pratensis L.	Asteraceae	H	D	Arrhen O	42
Verbascum thapsus L.	Scrophulariaceae	H	D	Atrop O, Onopord O	5
Veronica longifolia L.	Plantaginaceae	H	B	Filipen All	36

Life form: H—hemicryptophyte, Ch—woody chamaephyte, C—herbaceous chamaephyte. Durability: B—perennial, S—subshrub, D—biennial plant. Phytosociological units: Cl—Class, O—Order, All—Alliance, Mol-Arr Molinio-Arrhenetheretea, Fest-Brom Festuco-Brometea, Coryneph Corynephoretalia canescentis, Agr-Rumi Agropyro-Rumicion crispi, Aln glu Alnetea glutinosae, Nard-Cal Nardo-Callunetea, Fest val Festucetalia valesiaceae, Arrhen Arrhenatheretalia, Atrop Atropion belladonnae, Onopord Onopordion acanthi, Filipen Filipendulion ulmariae.

). The nomenclature of the species was assumed according to Mirek et al. [37]. For each species, the following parameters were specified: family, life-form using the Raunkiaer scale, biological-stability (durability) and main phytosociological units in which the species usually grows in accordance with Matuszkiewicz [38]. Almost all of the studied species (except two) are hemicryptophytes, their buds are near the ground and are protected by live or dead leaves, mulch, topsoil or snow cover. These species are more susceptible to freezing or prolonged drought than species from other groups. The majority are perennial species. Nine plant species represented extensively used lowland hay meadows (Cl. Molinio-Arrhenetheretea, five species) and xerothermic calcareous grasslands (Cl. Festuco-Brometea, four species) and five belonged to other phytosociological units. Most of the species (11 out of 14 species) tolerate periodic soil moisture deficits well; according to Ellenberg and Leuschner [39] their soil moisture values are in the range of 2–4, i.e., dry, and moderately wet and periodically dry sites.

2.3. Growing Seedlings from Seeds

The diaspores (fruits and seeds) were collected in the Lower Pilica Valley in September 2014. This place is in the Special Area of Habitat Conservation (SAC, PLH 140016), located at the Mniszew in the Mazovian voivodeship. Next year, in the spring, on 10 April 2015, the diaspores were sown in pots, each species separately, in three replications (three pots per species). The number of diaspores was the same, i.e., 25 per pot (one replication). The germination capacity was differentiated. In 8 species, the germination capacity was very good (over 80%): A. millefolium, D. deltoides, E. planum, H. perforatum, L. vulgaris, P. pratensis, V. thapsus and V. longifolia. A lower germination capacity of 60 to 79% was found in three species: A. campestris, P. lanceolata and P. erecta. And the lowest germination capacity (40–59%) was found in: C. stoebe, I. britannica and S. ochroleuca. The room temperature during pot experiment varied over time. During the first three weeks it oscillated between 18–22 °C during the day, while at night it was lower and ranged from 13 to 15 °C. In the following weeks, the temperature was a little higher and ranged from 20 to 26 °C during the day and between 15 and 17 °C at night. The pots were often moistened with distilled water. Efforts were made to ensure the optimal soil moisture content for the development of plants of each species (Figure 3).

2.4. Planting Seedlings and Their Assessment

On 17 July 2015, in the investigation site (red rectangle, Figure 1), seedlings of 14 native plant species were planted. Site preparation consisted of low mowing of meadow sward and removal of mown biomass and dead parts of vegetation (litter) on the previous day (16 July 2015). The one-factor field experiment was designed, with three replications, in 1 m² micro-plots. All the seedlings that developed and maintained were planted. Their number varied per species (from 5 to 48; Table 1); this was the result of differences in seed germination capacity, establishment and survival of seedlings in the initial period of growth and development of the tested species. Seedlings of each species were planted on separate micro-plots. Seedlings from one pot were planted in one micro-plot; therefore, the pot was a replication. The distance between the seedlings was different, depending on their number in the pot. In some cases, the seedlings were not separated to avoid damaging their roots. (Figure 4 and Figure 5). At the time of planting, all the seedlings were in very good condition.

In 2015–2017, growth and development of the plants were monitored on the established plots. The following features were determined:

(1) The plants’ growth and development—once a month (during vegetation season) the height of vegetative (variable hv) and generative (variable hg) shoots was measured, in three points on the diagonal of the area covered by peculiar species and their development phase was determined.

(2) The population size of the planted species (expressed as the degree of ground cover with plants of planted species) was estimated (variable Range in cm²).

(3) The condition of the population of planted species (variable Condition) was determined on a scale point from 1 to 5 (

Table 2. Point scale for assessing the condition of the population of planted species.

Scale	Description
1	very poor condition, more than 80% of plants wither or turn yellow
2	poor condition, 50–80% of plants wither or turn yellow
3	medium condition, 30–50% of plants wither or turn yellow
4	good condition, 5–30% of plants wither or turn yellow
5	very good condition, brightly green plants, traces of withering or yellowing

).

The study analysed the results of biometric measurements in characteristic dates of the growing season, corresponding to the dynamics of plant development (

Table 3. Biometric measurements in data analysis in selected research dates.

Date of Measurements	Dynamics of Plant Development	Tested Parameters
17 July 2015	planting seedlings	h2015o
21 September 2015	before winter in the year of planting	h2015i
21 April 2016	after overwintering	h2016v, R, C
5 July 2016	in full vegetation—summer	h2016g, R, C
29 September 2016	before winter in the first year after planting	R, C
6 May 2017	spring regrowth	h2017v, R, C
20 July 2017	in full vegetation—summer	h2017g, R, C

).

h2015o—height of vegetative shoots at the time of planting, h2015i—height of vegetative shoots in the year of planting, h2016v—height of vegetative shoots in the first year after planting, h2017g—height of generative shoots in the second year after planting, R—Range, C—Condition.

2.5. Rainfall Conditions—Realtive Precipitation Index

The weather conditions during the research period were illustrated with the relative precipitation index (RPI expressed as a percentage). This index represents the drought status at the specific place of interest [40]. The RPI expresses a quotient of the average sum of precipitation and the long term-average of this meteorological parameter. The RPI classes proposed by Łabędzki [40] can be written as follows:

-

0–24.9 extremely dry;

-

25–49.9 very dry;

-

50–74.9 dry;

-

75–125.9 average;

-

>125.9 wet.

The precipitation sums were recorded at the Warsaw University of Life Sciences (WULS-SGGW) meteorological station, which is located about 500 m from the experimental site (

Table 4. Sum of precipitation (mm) in growing seasons 2015–2017.

Year	Month						Growing Season
	April	May	June	July	August	September
2015	31.4	57.7	37.5	66.2	11.3	73.1	277.2
2016	34.2	23.3	56.9	116.5	71.7	9.8	312.4
2017	62.1	68.3	106.2	111.0	68.5	146.8	562.9
1960–2017 *	36.7	59.4	68.7	78.6	64.6	48.9	356.9

* according to Oleszczuk et al. [42].

). The monthly base sum of precipitation was used for the RPI index calculation (Figure 6). The high temporal variation of precipitation was visible especially in 2015 and 2016, where the precipitation ranged from 11.3 to 73.1 and from 9.8 to 116.5 mm, respectively. Nevertheless, according to the RPI, half of the months of these years have been classified as average. August 2015 and September 2016 have been classified as extra dry. Relatively uniform in terms of rainfall was 2017, in which most of months indicated the wet conditions. Analysing the whole growing seasons of 2017, it can be stated that relative precipitation index classifies this year as wet, similarly to hydrothermal index discussed in the studies of Janicka et al. [41]. The RPIs for 2015 and 2016 were 77.67 and 87.53, respectively, indicating that these seasons were average in terms of rainfall. Temperature conditions in the 2015–2017 growing seasons were discussed in an earlier article [41].

2.6. Water and Soil Conditions

The moisture condition of the research site is controlled by the groundwater level, which depends on the water inflow from the surrounding plateau and drainage outflow, regulated by the water level in the nearby flood channel. The study area is dominated by mucky soil in the surface layer (0–40 cm), which is proven by the high content of phosphorus, not less than 1050 mg P∙kg⁻¹, and relatively high pH in the ranging from 5.5 to 6.8 [36]. The mucky soil is underlayered by the highly decomposed peat with a high amount of ash content, oscillating around 30% of absolutely dry matter. Gyttja formation also occurs locally. The organic soil layers developed on the loose sand. The groundwater level varied significantly depending on the year of the study. The lowest was in the 2015 growing season, on average about 101 cm. The groundwater level varied significantly depending on the year of the study. It was the lowest during the 2015 growing season, with an average of around 101 cm during the growing season. Already in the spring, it oscillated around an average of about 99 cm, and in July and August it ranged from 96 to 123 cm, remaining in this range almost until the end of the growing season. The groundwater table was rebuilt this year in November (70 cm). During the 2016 vegetation period the groundwater level was higher compared to the previous year and maintains mainly at 47–108 cm, after heavy rainfall in July it increased to 47 cm, but after the end of August it decreased to about 100 cm. During the 2017 growing season, due to the high amount of precipitation, the groundwater level was high, on average it was 56 cm below the ground surface, in May it rose to 25 cm, and at the end of June it was just below the soil surface.

2.7. Statistical Analysis

The statistical data analysis was performed in three stages. In the first part, the set of the variables (Range, Condition, h2015i, h2016v, h2016g, h2017v, h2017g) and observations was used to allocate the species into a similar group. For this reason, we used the hierarchical agglomerative method, so-called clustering. Clustering belongs to a group of methods enabling the classification of observations (species) into groups based on similarity (distance) and distinctly different from the other groups. The hierarchical agglomerative clustering method results in an approach, which successively joins the most similar observation [43]. Before the cluster analysis, the standardisation procedure (Equation (1)) is required because values of explained variables (Range, Condition, h2015i, h2016v, h2016g, h2017v, h2017g) vary over the wide range.

Z_ij = (O_ij − O_avg)/σ_i

(1)

where: Z_ij standardised value of i-th variable for j-th plant species; O_ij value of i-th variable as above for j-th plant species; O_avg average value of i-th variable; σ_i standard deviation of i-th variable.

The similarity between every pair of observations was determined using the Euclidean distance. Then based on the distance matrix, the Ward’s method was applied to minimise the distance between individuals as well as to maximise the sum squares between agglomerate groups [44]. This successively repeated procedure results in the distance between clusters in form of (Equation (2)):

d(Sa,Sk) = ((ni + nk)d(Si,Sk) + (nj + nk)d(Sj,Sk) – nkd(Si,Sj))/(ni + nj + nk)

(2)

where d is the distance between two established clusters; Si, Sj, Sk are agglomerated clusters i, j and k; Sa is a new agglomerated cluster combining the Si and Sj clusters; ni, nj, nk are the numbers of individuals included in the clusters i, j and k.

The cluster analysis was performed using the R-statistical software [45]. The optimal number of clusters was chosen through the application of the Silhouette test [46] similarly to the data analysis made by Janicka et al. [41]. Before the clustering procedure, the significant variables, which describe most of the variation in the dataset, were selected. Based on the correlogram in the PCA method [47], we visually excluded the Range variable from the data set, which was not significant for the explanation both first and second component. Therefore, we have decided to perform the cluster analysis without this variable.

In the second stage of the analysis, the establishment of the homogenous group was plotted using a phylogenic graph, which nicely illustrated the structure of the grouped data. Then the PCA biplot [48] was used to show plant species group and variables. The differences in groups of observations are illustrated by the box plots drawn for the selected most important explanatory variables. In the testing procedure, it was checked whether the means of the established groups were equal (null hypothesis) or significantly different. For this reason, the nonparametric Wilcoxon rank sum test enabling the calculation of an exact permutation null distribution was applied. The Wilcoxon test can be performed for a limited number of observations, resulting in the test statistics and p-values using the wilcox.test routine of the stats-package in R [49].

Finally, based on the biometric measurements (h2015i, h2016v, h2016g, h2017v) and average state parameters (Condition and Range), the surviving model of the plant species was developed. For each plant species which developed the generative shoots during 2017 were treated as survived plants and labelled as factorial data equal to 1 the rest of the plants were treated as died and marked as 0. The RWeka package with J48 algorithm [50] was used to establish a decision tree. The first step includes dividing the entire set of observations by selecting the most significant variable among all (Conditions, Range, and height during 2015 and 2016) and then binary split into at least two child sets.

3. Results

3.1. Shoot Height

A large diversity of species was found in terms of growth rate and height of seedlings in the sowing year. In the initial period, the fastest growth and the highest seedlings before planting were noted for Tragopogon pratensis, Plantago lanceolata and Achillea millefolium (

Table 5. Height (cm) of seedlings, vegetative and generative shoots of introduced species in subsequent years (2015–2017).

Species	2015		2016		2017
	17 July	21 September	21 April	5 July	6 May	20 July
	Seedling		Vegetative	Generative	Vegetative	Generative
Achillea millefolium	21.6	29.0	11.9	79.3	15.0	52.0
Artemisia campestris	12.3	18.3	7.5	57.3	-	-
Centaurea stoebe	14.6	26.3	7.2	61.0	-	-
Dianthus deltoides	10.8	16.5	8.3	40.7	-	-
Eryngium planum	11.2	12.7	5.5	79.0	-	-
Hypericum perforatum	14.0	19.2	8.0	59.6	14.0	46.7
Inula britannica	2.7	11.0	6.9	27.3	-	-
Lysimachia vulgaris	17.7	19.2	-	35.2	-	-
Plantago lanceolata	22.9	24.5	10.0	47.3	16.0	-
Potentilla erecta	7.9	12.0	10.5	39.8	-	-
Scabiosa ochroleuca	14.1	18.0	8.0	83.4	-	-
Tragopogon pratensis	27.1	36.7	13.7	88.6	-	-
Verbascum thapsus	9.0	27.5	16.5	80.9	-	-
Veronica longifolia	17.9	27.3	15.3	87.2	21.0	89.2

). In the first three months after planting, in addition to the above-mentioned species, tall seedlings were also developed by Verbascum thapsus, Veronica longifolia and Centaurea stoebe. In the spring of the following year, Verbascum thapsus, Veronica longifolia, Tragopogon pratensis and Achillea millefolium stood out with the highest vegetative shoots. In the following months, these species, as well as Scabiosa ochroleuca and Eryngium planum also developed the highest generative shoots. After reaching the phase of setting and ripening of seeds, the process of dying of two-year-old species: Verbascum thapsus, Tragopogon pratensis, Scabiosa ochroleuca and Centaurea stoebe began. Unfavourable weather conditions, especially the long-term drought prevailing in the second half of the growing season in 2015, resulted in the disappearance of, among others, Inula britannica, a species characteristic of floodplain riverside grasslands, and Lysimachia vulgaris, which is found mostly in wet thickets. Artemisia campestris, Dianthus deltoides, Eryngium planum and Potentilla erecta, which were well tolerant of drought conditions, but usually grow on loose grasslands, also disappeared, failing to withstand the competition from the vegetation of the old turf. Of the species planted in the third year, only four remained: Achillea millefolium, Hypericum perforatum, Veronica longifolia and Plantago lanceolata, of which the first three developed flower shoots and set seeds.

3.2. Range Population of Introduced Species

The degree of land cover by populations of individual species depended to a large extent on the way they grew, and the number of seedlings planted. Regardless of the date, rhizomatous species grew rapidly and strongly, especially Achillea millefolium and Veronica longifolia, which formed long underground rhizomes (

Table 6. Range population (cm²) of introduced species in subsequent years (2015–2017).

Species	2015	2016			2017
	21 September	21 April	5 July	29 September	6 May	20 July
Achillea millefolium	1085.3	1175.0	2217.0	4116.0	586.0	1782
Artemisia campestris	484.7	216.2	191.0	-	-	-
Centaurea stoebe	2421.3	478.0	217.0	-	-	-
Dianthus deltoides	637.5	240.5	3500.0	-	-	-
Eryngium planum	277.3	214.7	222.0	-	-	-
Hypericum perforatum	-	94.0	387.5	512.0	442.0	1145.5
Inula britannica	706.0	2295.0	3249.0	3249.0	-	-
Lysimachia vulgaris	-	242.0	736.0	1472.0	-	-
Plantago lanceolata	2050.0	204.0	560.0	552.0	150.0	-
Potentilla erecta	218.3	163.5	375.7	-	-	-
Scabiosa ochroleuca	360.3	117.3	446.7	1017.0	-	-
Tragopogon pratensis	1543.3	593.0	1040.0	341.0	-	-
Verbascum thapsus	2836.5	922.0	3185.5	-	-	-
Veronica longifolia	708.0	254.3	1236.7	4116.7	576.7	3293.3

). In the first year, Plantago lanceolata, Centaurea stoebe and Verbascum thapsus, which formed wide rosettes of leaves, were also distinguished by a large cover. In the second year, they grew back quickly, and, in the summer, the greatest range was characterised by Dianthus deltoides, Inula britannica and Verbascum thapsus. In the last year, only three species of Achillea millefolium, Hypericum perforatum and Veronica longifolia survived, which developed tall shoots and best tolerated both competition from old sward species and unfavourable weather conditions (both moisture deficiency and excess) and systemically expanded their range. In addition, all these species flowered profusely and produced many seeds.

3.3. Condition Population of Introduced Species

Throughout the research period, Veronica longifolia was in a very good condition (

Table 7. Condition population (scale 5°) of introduced species in subsequent years (2015–2017).

Species	2015	2016			2017
	21 September	21 April	5 July	29 September	6 May	20 July
Achillea millefolium	5.0	5.0	5.0	5.0	4.0	2.0
Artemisia campestris	4.7	4.7	3.7	2.0	1.0	1.0
Centaurea stoebe	4.8	5.0	3.3	2.0	1.0	1.0
Dianthus deltoides	5.0	5.0	5.0	3.0	1.0	1.0
Eryngium planum	3.5	3.7	2.0	1.7	1.0	1.0
Hypericum perforatum	4.3	4.7	5.0	5.0	3.3	3.2
Inula britannica	5.0	5.0	4.5	4.0	1.0	1.0
Lysimachia vulgaris	4.0	2.0	3.7	4.2	1.0	1.0
Plantago lanceolata	5.0	3.0	4.5	4.7	4.0	2.0
Potentilla erecta	3.7	3.7	3.0	2.0	1.0	1.0
Scabiosa ochroleuca	4.5	4.0	3.3	3.0	1.0	1.0
Tragopogon pratensis	4.3	3.7	2.5	3.7	1.0	1.0
Verbascum thapsus	5.0	5.0	5.0	1.0	1.0	1.0
Veronica longifolia	5.0	5.0	5.0	5.0	5.0	5.0

), which is confirmed by the fact that this species develops well on wet, periodically flooded meadows. In the second year of the study, and especially until its middle, Achillea millefolium, Dianthus deltoides and Verbascum thapsus were also in very good condition. In the last year, most of the planted species were in very poor condition, which was probably due to too high groundwater level and hypoxia of the root layer of the planted plants. In that year, Veronica longifolia, Hypericum perforatum and Achillea millefolium, as well as Plantago lanceolata, were in the best condition.

3.4. Grouping of the Plant Species

The preliminary analysis of the measurement data using PCA showed that the explanatory variables fall into three groups (Figure 7). The first one is related to the condition of vegetation (Condition) and the height of vegetative and generative shoots in the second year after planting (2017). The second group of variables is the measurement data concerning the height of shoots at the end of vegetation in the year of planting and the height of shoots in the first year after planting. The third group of explanatory variables included one feature—Range. The first group of variables was most correlated with the first PCA principal component. The correlation coefficient (r) was very high and ranged from 0.90 to 0.95. The second group of variables was less correlated and concerned both the first and second components of PCA. The “Range” of the correlation coefficient variation ranged from 0.75 to 0.80. Range had the smallest share in explaining the variance of the data set. This was due to the low values of the correlation coefficient in the description of both PCA principal components, especially in the second one (r- close to zero). Therefore, a matrix containing 14 observations (plant species) described by 6 explanatory variables (without Range variable) representing the growth and development of species was analysed to isolate homogenous groups of vegetation.

Based on the Silhouette test, it was established that the optimal number of clusters allowing for the separation of plant groups with similar growth and development characteristics after introduction was equal to 3 (Figure 8).

The analysis of the examined parameters determined by Ward’s method resulted in three groups of plants, which were illustrated in the form of a phylogenetic tree in Figure 3. The first group (marked in blue in Figure 9) consisted of four species: Achillea millefolium, Hypericum perforatum, Plantago lanceolata and Veronica longifolia. The first three are species typical for meadow communities, moderately wet and periodically dry habitats, and the fourth (V. longifolia) for tall herbs (Filipendulion ulmariae). The second group of plants (marked in green) was diverse. One branch included tall species (S. ochroleuca, E. planum, C. stoebe), typical of dry habitats (Festuco-Brometea, Festucetalia valesiaceae). The second branch included species of grassland habitats, that spread rapidly by rhizomes (Potentilla erecta, Artemisia campestris, Dianthus deltoides) and two species from herbaceous communities of moist habitats, also forming rhizomes (Lysimachia vulgaris, Inula britannica). The third group (marked in red) was represented by two tall species with a two-year life cycle (Tragopogon pratensis and Verbascum thapsus).

The three distinguished groups of plants were depicted against the background of the two main components of PCA (Figure 10). This allowed the presentation of the relationship between the factors determining the effects of introduction and individual species. The “blue” group was associated with high condition values, these species best tolerated adverse weather conditions and in the last year of the study (2017) produced generative shoots, which indicates the success of the introduction—the possibility of survival of these species in subsequent years. The “green” group, on the other hand, was characterised by features indicating a low chance of survival, due to low condition values in the 1st and 2nd year of the study, these species quickly disappear at higher sward density. The “red” group was quantified due to the occurrence of high generative shoots in the first year after planting, these are typical biennial species. The formation of seeds by them creates the possibility of their maintenance in the form of a soil seed bank.

The results on selected traits of plant population characteristics presented in the form of a box-plot (Figure 11) showed that the best indicator differentiating the distinguished groups of species was the indicator describing their development condition (Condition). It is noteworthy that the first group of species (“blue”) had the condition values above the overall mean value of 3.25. The remaining groups of species were characterised by a significantly lower value of this parameter (Condition) (Figure 11A). The range feature was not an indicator distinguishing individual groups of species. However, with probability of p < 0.1, it can be concluded that the second group of plants (“green”) had a smaller range compared to the other two groups (Figure 11B). In the case of the variable describing the height of vegetative shoots of species in the first year after planting (h2016v), it can be seen that the second group of species (“green”) developed lower vegetative shoots compared to the other two groups (Figure 11C). Therefore, the obtained results confirm the specificity of these groups of plants that were objectively separated using the hierarchical clustering method and the PCA method (Figure 9 and Figure 10).

Figure 12 shows the course of vegetative development, and thus the degree of land cover (Range) by the populations of three species characteristic of each hierarchical group of species distinguished in the cluster analysis, against the background of changing thermal conditions and the level of the groundwater table. For the first group (“blue”), it was Achillea millefolium, which was characterised by a clear increase in its range, especially in the first year after planting, with an average groundwater level of 85 cm (2016). This species, characteristic for moderately wet and even dry habitats, developed much less in the following year (2017), with a higher groundwater level in the early months of the growing season. This indicates a fluctuating response of this species to moisture conditions. Species from groups 2 and 3 (“green” and “red”, respectively) were characterised by a significantly lower Range parameter than A. millefolium. Potentilla erecta (the species characteristic of the second group) did not dynamically increase its range, despite the fact that the groundwater level in the 2015 and 2016 growing seasons was low (average 95 cm) and could have been conducive to the development of the species from the Nardo-Callunetea phytocoenosis. Tragopogon pratensis—a species of moderately wet and periodically dry habitats (Arrhenatheretalia) was chosen for the characterisation of species from the third group (“red”). The population of this species dynamically increased its range in the year of planting, i.e., in the conditions of ground water level of 105 cm on average. In the following year, its range was not so large, which was associated with the strong development of generative shoots.

The main objective of the study was to determine the possibility of species survival after introduction. For this type of analysis, the main variable describing the analysed data set was separated. The main feature allowing the classification of the species survival was their condition (Condition). The developed decision tree contains two “leaves” (Figure 13). The Condition variable separating the data set was 3.417 (on a scale of 1–5). In practice, this means that the above-average condition of species determines their survival. It can therefore be concluded that within the considered group of variables, one feature is sufficient to determine the survival of the introduced species—namely, Condition. The developed model positively identified 3 species that persisted for three years after planting (2015–2017) and 10 species that disappeared after two years. In addition, it should be noted that one species was misclassified (Plantago lanceolata). It was assigned to the group of surviving species in the model, although in 2017 it did not produce generative shoots, which were the basis for classification, but only had vegetative organs. This means that in order to positively classify this species, the range of variables in the model should be extended.

4. Discussion

Due to the high diversity of meadow habitats, not all methods of restoring plant communities can be applied everywhere. With limited seed resources, better results are obtained after planting seedlings compared to sowing seeds. The success of the introduction may be determined by the number of introduced individuals (seedlings). The results of our field research (micro-plots) showed that a small (compared to that recommended in the literature) number of seedlings of species with greater potential allowed them to be successfully introduced. In the research literature, the suggested number of introduced seedlings should range from 500 [51] to 5000 [52]. However, in most experiments of this type less than 100 seedlings were used, which results from the high labour and cost of such research. At the same time, it should be noted that large range in the number of planted seedlings per species (from 5 to 48) could have determined the maintenance of the population of the tested species and the final success of the introduction, especially species with a low number of seedlings. This is in line with earlier research by Reed et al. [53] who found that the greater the number of seedlings, the better their competitiveness in relation to primary sward species, the greater the survival rate, and the greater the ability of the population to adapt to new habitats as a result of adaptive genetic variation (genetic drift).

The results of discussed studies showed that species developing rhizomes easily spread and quickly increase the population range. Therefore, in assessing the success of the introduction, more emphasis should be placed on the biology of the species, which was also emphasised by Godefroid et al. [30]. It is necessary to expand knowledge about the population dynamics of individual species, habitat selectivity and their response to stress factors, including meteorological factors. Achillea millefolium and Veronica longifolia easily occupied new localities. In the changing climatic conditions of Poland, these species have been characterised by an increase in the number of new localities in recent decades [54].

The ability of species to produce seeds should also be assessed, because the persistence of perennial species introduced into a community is the result of vegetative and generative reproduction. During summer of the second year of study (first year after planting), the flowering of 9 out of 14 planted species was observed (Figure 14). Achillea millefolium, Centaurea stoebe, Dianthus deltoides, Hypericum perforatum, Scabiosa ochroleuca and Veronica longifolia flowered profusely and produced seeds. This suggested the possibility of maintaining and further developing in a new place (place of introduction), because, as stated by Menges [29], the measure of the final fate of the introduction is the ability of seedlings to flower and set fruit. Two low species, Plantago lanceolata and Potentilla erecta bloomed less profusely and were easily suppressed by the plants of the original sward. These species are better maintained in a low sward, more often mowed or grazed. Verbascum thapsus did not finish flowering, the plants in the flowering phase were knocked over by animals. Lysimachia vulgaris did not bloom because its shoots were bitten by animals. This fact confirms previous reports that the area where introductions (reintroductions) are carried out should be fenced [30]. The other three species (Artemisia campestris, Eryngium planum, Inula britannica) did not bloom either. The biology of flowering and seed setting of many “wild” species in meadow habitats is still poorly understood. Godefroid et al. [30] found that, in most cases, introduced seedlings only occasionally set seeds or did not set seeds at all, indicating that most introductions are not successful in the long term.

A significant decrease in the success of the introduction has already been observed in the third year of the study (second year after the introduction). It was probably related to unfavourable weather conditions. Due to progressive climate change, the frequency of years with unfavourable climatic conditions increases, which we observed during the research period—droughts in 2015 and 2016, and excessive rainfall and, consequently, high groundwater levels in 2017. The high groundwater level reduced the oxygen supply to the roots and caused the death of well-developed plants of Centaurea stoebe, Dianthus deltoides and Scabiosa ochroleuca, which are drought-tolerant species. The influence of environmental factors, such as drought, day–night variation, shedding, soil disturbance and irrigation on the functional traits of meadow species was emphasised by, among others, Isselstein et al. [55]. In central Poland, weather extremes (more often prolonged droughts) are becoming more frequent and more prominent. Therefore, in subsequent years, the range of species on muck soils may be affected by the groundwater level (both very low and high). In favourable hydrological systems, they can develop and maintain in the sward, provided that the vegetation is systematically mown and thus limiting the competitiveness of the old sward. It is difficult to predict the fate of planted species, as it requires a much longer research period than three years. Godefroid et al. [30] found that often too optimistic assessment of the success of introductions is based on the short-term results, monitoring of the development of the introduced species usually ceases after 4 years.

In addition, the ecological awareness of the society should be constantly raised through various types of educational activities. They make it possible to stimulate the local community to management in accordance with the requirements of valuable grassland communities. In addition, in order to support farmers managing areas of special habitat protection (SACs), appropriate subsidies are necessary, e.g., under climate, agricultural and environmental programs, which will allow to preserve valuable habitats. Activities in this area should be prioritised on the basis of the EU Restoration Plan 2030, which aims to stop deterioration rate of protected habitats and species by 2030.

5. Conclusions

The results of our study indicated that, among 14 dicotyledonous species introduced in the habitat of post-bog soils, a landscape nature reserve, three species Achillea millefolium, Hypericum perforatum, Veronica longifolia have the best introduction efficiency. This group of species was distinguished by the good condition and the presence of generative shoots in the second year after the introduction. Tragopogon pratensis and Verbascum thapsus are plants with the two-year life cycle; their seeds have the ability to persist in the form of a soil seed bank and can sprout in subsequent years. Most planted species have low introduction success, mainly due to the poor condition and low height. However, it should be emphasised that the studies are too brief to unequivocally judge about the success or failure of reintroduction, especially in the case of biennial species.

The decision tree model used showed that the condition is the main variable enabling the classification of species population survival. The cut-off value for survival was 3.417 on the 1–5 scale. Plant condition values higher than this threshold indicate the potential success of the introduction of species from the group distinguished in the analysis of hierarchical clusters (Achillea millefolium, Hypericum perforatum, Veronica longifolia). It has been suggested to include the assessment of the condition of plants in the systematic monitoring of restored plant communities.

In assessing the effectiveness of the species introduction, more attention should be paid to: (1) the biology of the species, (2) the critical development phases of the planted species against the background of weather conditions and the competitiveness of the old sward, and (3) the condition of the population should be monitored for a long time. These issues should constitute further directions of research in this field.