The Influence of Konik Horses Grazing and Meteorological Conditions on Wetland Communities

Abstract

Extensive management is one of the methods used in the conservation of fen meadows. The aim of the study was to characterize grassland communities grazed by Konik horses in the Biebrza National Park and to assess the influence of short-term grazing on the sward. During a 3-year period, phytosociological studies, measurements of the sward cover and height, as well as an assessment of the botanical composition of meadows were conducted in places grazed and excluded from grazing. The grasslands were mowed once a year at the turn of July and August, with the exception of mid-forest meadows. The area studied was dominated by communities of the Molinion alliance. Horses kept at a very low stocking rate reduced the height and cover of the sward. In the short-term, weather conditions, especially the duration of floods, had an even greater impact on the species composition of communities than grazing. Therefore, in order to protect the Molinia meadows, apart from extensive mowing, it is necessary to maintain appropriate habitat conditions, especially in terms of controlling the moisture, and conduct its regular monitoring.

1. Introduction

In the last 70 years, the area of meadows in Europe has gradually declined [1]. The most major threats to the multifunctionality and durability of grasslands in Europe are abandonment, heat, and drought stress as well as a conversion to temporary grassland [2]. To their loss also contribute N deposition, the conversion to arable land, and intensification [2]. Wetlands play a special role among grassland ecosystems in the landscape of Europe not only as the carbon fixation storage or for nutrient and contamination retention but due to their extreme importance for biodiversity conservation [3]. For this reason, they are protected in the EU under the Birds and Habitat directives (since 1979 and 1992, respectively). Additionally, the conservation of those of international importance is supported by the Ramsar Convention. One of the major problems regarding wetlands in protected areas such as national parks is the preservation of their natural value. The deterioration of these habitat conditions occurs as a result of either an intensification or abandonment of management practices because both of these may lead to changes in the botanical composition of grassland communities [4,5,6]. Furthermore, an additional critical threat to peatland habitats in Central Europe are changes in water conditions, within which shifts associated with the changing climate, such as higher temperatures, fluctuations in the precipitation amount, and distribution in the course of a year can be recognized [3,7]. Their conservation and/or restoration therefore often requires efforts aimed at restoring the environmental conditions and continuing the traditional management implemented thus far [8,9]. This is especially visible in the Biebrza National Park (BNP), the largest park in Poland, located in the north-eastern part of the country. It protects the Biebrza Valley and covers roughly 60,000 ha, of which more than 40% are semi-natural wet grasslands that require mowing once a year or every few years. Its international role is emphasized by additional protection under the NATURA 2000 network (Special Protection Area PLB200006 Ostoja Biebrzańska and Special Area of Conservation PLH200008 Dolina Biebrzy) and the RAMSAR Convention. The majority of grassland communities in the Biebrza Valley have been shaped by recurring floods as well as extensive mowing and grazing; thus, the habitats are very diverse [10]. They represent uncultivated meadows, mowed once a year, unfertilized and conserved as hay by local inhabitants. Since the establishment of BNP in 1993, the responsibility for the conservation of fen meadows has been resting with the National Park. This, in turn, requires not only financial expenses, but also physical work to be performed. Simultaneously, local farmers’ interest in keeping these grasslands in production has been declining due to their low yields and nutritive values under traditional management.

The mowing of semi-natural fen meadows as part of active protection is often carried out once a year, followed by hay collection, e.g., [9]. In the national park, such as in the Biebrza River Valley, there is a problem with the mowing of large areas and the management of the collected biomass. In contrast to traditional, manual mowing, the use of machines, e.g., tracked mowers based on snow grooming vehicles, can have a negative effect on functional plant diversity by disturbing micro-topography, leading to a decrease in the species richness, promoting forbs with clonal growth but suppressing the tussock or hummock species [11]. As part of active protection, extensive grazing is also possible [12]. On fen meadows, even moderate grazing can be recommended, but the optimal stocking rate has to be investigated [13]. This is to prevent negative effects such as the lowering of species richness resulting from foraging and trampling [13]. Animals kept even in a low stocking rate affect the swards by grazing, leaving manure and trampling, which may contribute to changing the biodiversity of the area [14]. Their impact on the plant communities has the character of small continuous disturbances, the first effects of which—depending on the stocking rate—can be noted even on a short time scale, e.g., [15]. In general, in comparison with only mowing, grazing can promote species diversity and favor species with a higher tolerance to trampling or rosette plants, e.g., [16,17]. Previous studies have focused on the impact of cattle grazing on fen meadows and showed that it can contribute to, e.g., suppressing reed expansion [17,18], but it can also cause a reduction in the species richness, e.g., [13].

In recent years, attention has been paid to the possibilities of using horses in the protection of grasslands. Even though both cattle and horses belong to the functional group of grazers and their diet is comparable, horses are characterized by a higher selectivity towards bitten species and eat more grasses [19,20]. It is worth emphasizing that, among others, primitive breeds of these species, due to their lower need for care and unique abilities allowing them to adapt to local environmental conditions, are particularly useful in nature-valuable habitats. An example of such a breed are Polish primitive horses (Koniks), the keeping of which is also supported by subsidies such as agri-environmental programs [21]. This breed is characterized among other things by a good fertility, gentle character, resourcefulness in looking for food, and straightforward foaling without human involvement [22]. Konik grazing is used in various regions of Europe (e.g., Germany, France, or Belgium) in order to protect grasslands. Previous studies have shown their usefulness in the protection of such habitats as the heathlands and calcareous grasslands, but still little is known about their maintenance on fen meadows [23]. As a consequence of all of the above, in 2004, the Centre of Conservation Breeding of Konik Polski Horses was founded and has been supervising the grazing of the horses in the wetlands ever since.

It was assumed that the plant communities in the area of the Centre of Conservation Breeding of Koniki Polski Horses in BPN are diverse due to their spatial arrangement and management. The aims of the study were: (i) to present the phytosociological and botanical characterization of grassland communities grazed by horses at the area of the Centre of Conservation Breeding of Konik Polski Horses in the Biebrza National Park, (ii) to indicate the main threats to grazed meadows, and (iii) to assess the short-term impact of Konik horses grazing on the sward.

2. Materials and Methods

2.1. Study Site

The study was conducted in the Biebrza National Park (BNP, Poland) for three years (2008–2010). The Park area is divided into three Basins: Upper, Middle and Lower. The Centre of Conservation Breeding of Konik Polski Horses and Rehabilitation of Wildlife is situated in the Middle one in the Grzędy forest district (53°61′88 N, 22°76′54 E) (Figure 1). The growing season in the Biebrza valley lasts 205 days on average. The climate is continental with hemiboreal features and belongs to the coldest in Poland, with a mean annual temperature of 6.6 °C [24]. Winters in this area are cold with regular frosts and snow, while summers are warm, with maximum temperatures up to 30 °C noted from May to August [24,25]. A minimal temperature in the Middle Basin of over 0 °C is recorded only in July [24]. The average annual rainfall for this region is about 575 mm [26], of which in summer—the critical period for wetlands—the average sum of precipitation equals 260 mm. The lowest precipitation is noted in January, whereas the highest is in July [24]. In 2009—the second year of the study—the mean temperatures in the warmest months (June–August) and the coldest winter month (January) were similar to the long-term averages (Figure 2). In contrast, in 2010, the winter was colder and the temperatures in individual summer months were higher (by 2 °C).

Due to the proximity of Woźnawiejski Canal, meadows situated in the enclosure are temporary flooded. Usually, the water stays over the ground level from early spring (thawing snow) until the beginning of June. Such conditions also occurred in 2009. In the following year (2010), the floods lasted until the beginning of August, which was caused by higher precipitation in the period from May to July (Figure 2). Thus, the annual precipitation this year was almost 200 mm higher than in the previous year and higher than the multi-annual average.

Since 2004, about 20–35 Konik horses have been kept in the enclosure in the Middle Basin of BPN (

Table 1. Total number of Koniks and stocking rate in the Biebrza National Park (2008–2010).

Year	Total Number of Horses * [Heads]	Stocking Rate * (LSU/ha) **
2008	34	0.11
2009	36	0.12
2010	26	0.09

* includes mares, stallions, and foals, ** LSU—livestock units represent the weight of the animal in terms of 500 kg.

). The fenced area for Konik horses was 209 ha, within which about 53.8 ha were meadows: one was located on the northern part of the enclosure, the second was a large complex in the southern part, and there were three small mid-forest meadows (Figure 1). The stocking rate of 0.1 LSU/ha was very low and comparable to other reserve breeding of Konik horses (0.19 LSU/ha on average in Poland) [23]. It allows not only for the preservation of plant communities in terms of overgrazing but also aims at adjusting the number of animals to the grazing capacity, especially at the critical period of feed shortage, i.e., late winter [28]. The horses stay in the enclosure all year round foraging mainly on grasslands. Only in winter are they provided with supplementary feed in the form of hay delivered from other meadows of the park. The construction of the enclosure (three rail fence) also allowed other wild animals (e.g., elks, roes and red deer, wild boars) to enter and leave. Wild herbivores are sporadically seen on meadows and are negligible in the analyses. According to the data available in BPN, the numbers of wild animals in the area are significantly low, e.g., 2.7 elk on average per 1000 ha.

The biggest meadow complex in the south of the enclosure is situated on typical slightly acidic peatland soil with a high soil organic matter content [5], classified as Eutric Rheic Hemic Histosols according to the World Reference Base for Soil Resources classification [29]. More information about the soil parameters have been provided by Sienkiewicz-Paderewska et al. [5]. The remaining part of the site is dominated by the Murshic Histosols [30] because of the gradual lowering of the water level, which causes peat degradation. Before the establishment of the park, the meadows within the Centre of Conservation Breeding of Konik Polish Horses and Rehabilitation of Wildlife belonged to private farmers and were mowed every 2–3 years. In 2008–2010, the grasslands grazed by Konik horses has been mowed once a year, as part of bird habitat conservation programs, at the turn of July and August (hay was collected), with the exception of mid-forest meadows, where no treatments were performed in the second and third years of the study.

2.2. Phytosociological Studies

Phytosociological studies were carried out in June 2008 on meadows situated at the area of the Konik horses’ enclosure in the BNP. A total of 33 relevés overall were recorded using the Braun-Blanquet method [31] based on the abundance of plant species. Plant species and their families were fully identified to the family level and named according to Mirek et al. [32], whereas there was a syntaxonomic affiliation after Matuszkiewicz [33]. The assessment of species diversity within the community was based on three indicators: the total number of species (floristic abundance), the average number of species in the relevé, and the Shannon–Wiener index (H’) [34]. The species diversity index (H’) was calculated according to the following formula:

$$\mathrm{H}\prime =-{\displaystyle \sum }_{i=1}^S{p}_i\times ln{p}_i$$

where S is the number of species and p_i is the proportion of individuals of each species belonging to the ith species of the total number of individuals.

Additionally, the fodder value score (FVS) defined by Filipek [35] was calculated. In order to establish the FSV numbers for each species, the following parameters have been taken into account: the fodder value, plant fertility, and poisonous properties. Plants with a very good fodder value have been assigned values within the range from 9 to 10, with good a value of 8–7, and with an average utilization value of 4–6. Low value plants have been assigned numbers between 1 and 3, while species with no fodder value, unreachable during mowing and omitted by animals, have been assigned a value of 0. Finally, all poisonous plants have been assigned a negative utilization value, ranging between −1 and −3, depending on the toxicity level. The FSV results are obtained from the multiplication of the percentage share of the current species in the yield by its assigned value, then by summarizing the products and dividing the total by 100. Sward of a very good value ranges between 8.1 and 10.0, good ranges between 6.1 and 8.0, mediocre between 3.1 and 6.0, and a poor value is below 3.0. Moreover, the Ellenberg indicator number method was used to describe the habitat conditions of the distinguished phytocenoses [36] and to emphasize the difference between them [37]. The weighted average values of climatic (L—light conditions, T—temperatures, and K—continentality) and edaphic (F—moisture, R—reaction, and N—nitrogen content in soil) indices were calculated for each of the relevés, taking into account the species coverage as a percentage.

2.3. Structure of the Sward

In order to evaluate the influence of horses on the sward in winter, fenced areas (5 m × 5 m) excluded from grazing were prepared in selected communities and were excluded from grazing. They were 12 such areas in total: three in the phytocenosis featured on the basis of phytosociological studies that are in the Carex panicea sedge communities, in the forest and northern meadow, and one in each mid-forest meadow. The destruction of sward by wild boars’ rooting activities made it impossible to establish a fenced area excluded from grazing in the Agrostis canina community. The distribution of the fenced areas is shown in Figure A1 (Appendix A). In the years 2009–2010, the sward height was measured using an HFRO sward stick [38] and the sward cover with a frame (0.5 m × 0.5 m) was divided into 25 squares (the Quadrat method) [39]. All measurements were made in grazing and non-grazing areas in four replications at random points of each phytocenosis three times per year: in the third decade of April, the second decade of June, and in the last decade of August.

2.4. Botanical Analysis of the Sward

In order to investigate the effect of Konik horses grazing on the botanical composition of the sward, the samples of 0.5 kg of fresh matter were collected at random from ungrazed (fenced-off) areas and the adjacent grazing areas of each community in three replications [40]. The herbage mass was harvested at the turn of July and August (at the same time as the mowing of the meadows) by using shears set to leave three centimeters of stubble. In total, 48 samples were taken out of 4 communities and in 3 replications from the places grazed by horses and those excluded from grazing, i.e., 24 samples per each year of the study. After the separation and drying of individual plant species (48 h at 60 °C), they were weighed with an accuracy of 0.1 g, and then the percentage share of the different groups of plants was calculated.

2.5. Statistical Analysis

Analysis of the similarity of relèves was carried out using the Jaccard measure. The grouping of the community was done using the Ward cluster analysis method, which is based on minimizing the variances within the resulting groups [41]. The mid-forest meadows were excluded from the analysis due to their similarity to the other relèves, which handicaps the interpretation of the dendrogram obtained. These communities were classified only on a class level. The cluster analysis and Jaccard measure was carried out using R 4.2.1 [42]. An analysis of the collected sward height data and sward cover data for the samples collected from grazed and un-grazed areas was carried out using a multi-factor analysis of variance (Tukey T test) performed in Statgraphics Plus 4.1.3.

3. Results

3.1. Phytosociological Studies

Non-forest communities grazed by Koniks comprise 86 vascular plant species. This represents 8% of the park’s vascular flora which, according to Werpachowski [43], amounts to ca. approx. 1015 species. Based on the floristic composition of the community, five plants communities were distinguished (Figure 3,

Table 2. Areas with non-forest plant communities grazed by Koniks in the BNP.

Meadow	Phytosociological Classification	Area *
		[ha]	[%]
1. Northern meadow	community with Potentilla erecta and Thalictrum flavum from the Molinion alliance	2.7	1.3
2. Mid-forest meadows	Molinio-Arrhenatheretea class	0.3	0.1
3. Sedge community	alliance Molinion var. Carex panicea	15.6	7.5
4. Near-forest meadow	association Molinietum caeruleae	5.4	2.6
5. Agrostis canina meadow	association Molinietum caeruleae var. Agrostis canina	3.4	1.6
Total area of non-forest communities		27.4	-

* The area with the communities measured by the Garmin eTrex Venture HC GPS and the percentage share was related to the area of the whole enclosure (209 ha according to BNP data).

). All of these communities are included in the Molinio-Arrhenatheretea class. The distribution of the distinguished phytocenosis is presented in Figure A1 (Appendix A). In the area studied, patches of the Molinia meadows (Molinion) dominated (Table 2). The floristic composition of the communities was quite similar (

Table A1. Detailed floristic composition (F—frequency, D—mean cover (sum of the mean percentage cover of the species) of plant communities grazed by Konik horses in the Biebrza National Park.

Community	1. Northern Meadow		2. Mid-Forest Meadows		3. Sedge Community		4. Near-Forest Meadow		5. Agrostis canina Meadow
Number of relevés	4		4		17		6		2
Sward cover [%]	95		87.5		65		75.8		80
Synthetic indicators *	F	D	F	D	C	D	C	D	F	D
Ch, Dass. Molinietum caeruleae, Molinion
Carex panicea L.	100	250	100	150	V	3368	V	2625	100	875
Molinia caerulea (L.) Moench	100	100	25	63	V	574	V	1875	100	1500
Potentilla erecta (L.) Raeusch.	100	4375	100	1188			V	1500	100	150
Carex flava L.					V	1047	V	875	100	250
Thalictrum flavum L.	100	2875	75	88	I	32	V	75	50	25
Lycopus europaeus L.	25	13			V	224	III	58	100	250
Gentiana pneumonanthe L.	25	13	25	13	III	21	III	25	100	10
Salix rosmarinifolia L.	50	25	25	63	IV	91	II	17	100	150
Inula britannica L.							I	8	100	150
Briza media L.	75	88	50	25			I	8
Iris sibirica L.	25	13
ChAll. Filipendulion
Filipendula ulmaria (L.) Maxim.	100	1188	100	875			V	83	100	50
Lysimachia vulgaris L.	25	13			V	168	IV	133	100	150
Valeriana officinalis L.	75	138	75	38	I	1	III	45	50	25
Lythrum salicaria L.	25	13			III	54	I	8	100	50
Stachys palustris L.					II	24	I	8	50	25
ChAll. Calthion
Lathyrus palustris L.	75	138	50	25	II	41	I	8	50	25
Cirsium rivulare (Jacq.) All.							I	8	50	25
Myosotis palustris (L.) L. em. Rchb.							I	2
Caltha palustris L.	25	13			III	43			50	5
Polygonum bistorta L.			25	63
ChAll. Cnidion dubii
Cnidium dubium (Schkuhr) Thell.			25	13	IV	136	V	117	100	250
ChO. Molinietalia
Deschampsia caespitosa (L.) P. Beauv.	100	513	100	513			V	77	100	875
Lychnis flos-cuculi L.	75	38	50	25			II	3	100	250
Cirsium palustre (L.) Scop.							III	25	50	25
ChAll. Arrhenatherion
Galium mollugo L.			25	63
Campanula patula L.			25	3
ChO. Arrhenatheretalia
Daucus carota L.	50	75	100	563
Taraxacum officinale F.H.Wigg.			25	13			II	10	50	5
ChO. Agropyro-Rumicion crispi
Ranunculus repens L.							I	2	100	875
Potentilla anserina L.					I	3			100	250
Lysimachia nummularia L.	50	25	25	13	I	15	I	8	50	25
Rumex crispus L.			25	125
ChCl. Molinio-Arrhenatheretea
Poa pratensis L.	100	1188	100	1338	I	1	III	25	100	10
Rumex acetosa L.	100	1188	75	188			I	2
Ranunculus acris L.	100	563	100	513	I	1	IV	20	100	250
Festuca rubra L.	100	513	100	1025	I	3	V	183	100	250
Cerastium holosteoides Fr. em. Hyl.	75	188	100	250			III	25
Plantago lanceolata L.	75	88
Cardamine pratensis L.	75	38	25	13	I	1	I	2
Vicia cracca L.	50	125	50	125
Holcus lanatus L.	50	25	25	13					100	10
Leontodon hispidus L.			25	63	III	44	IV	33	100	150
Phleum pratense L.									50	25
Avenula pubescens (Huds.) Dumort.			25	13
Companion species
ChCl. Alnetea glutinosae
Betula pubescens Ehrh.	75	88	25	63	II	26	IV	60	100	10
Salix cinerea L.	25	13			III	36	I	8
Salix aurita L.	25	13			III	30
Thelypteris palustris Schott.					II	59
Betula humilis Schrank							II	50
Frangula alnus Mill.							I	2
ChCl. Phragmitetea
Galium palustre L.	100	250	100	250	V	162	V	250	100	250
Carex buxbaumii Wahlenb.					V	385	V	292
Carex elata All.					IV	456	I	42	100	250
Phragmites australis (Cav.) Trin ex Steud					III	85	II	83
Scutellaria galericulata L.					II	21	I	8
Alisma plantago-aquatica L.					I	15
Lysimachia thyrsiflora L.					I	6
Poa palustris L.			25	63			III	267	100	150
Glyceria fluitans (L.) R.Br.									50	5
ChCl. Scheuchzerio-Caricetea nigrae
Agrostis canina L.	100	825	100	875	IV	35	V	175	100	3750
Carex nigra Reichard	100	250	50	125	IV	265	III	92	100	250
Viola palustris L.	50	438	25	63	II	12	IV	133	100	10
Comarum palustre L.	50	75	25	13	V	1044	V	117	100	150
Veronica scutellata L.	25	13	50	75			I	8	100	10
Juncus articulatus L. em. K. Richt.					V	219	II	83
Ranunculus flammula L.					V	84	I	8	100	1500
Calamagrostis stricta (Timm) Koeler					IV	121
Menyanthes trifoliata L.					II	34
Stellaria palustris Retz.			25	13	II	7	I	8	100	30
Eriophorum angustifolium Honck.					I	16
Other
Geum rivale L.	100	2063	100	2938			I	2
Anthoxanthum odoratum L.	100	1500	100	1438	I	3	V	42
Luzula multiflora (Retz.) Lej.	100	250	75	188			III	58
Viola sp.	100	100	50	75	II	50	IV	133	100	250
Glechoma hederacea L.	75	38	25	13
Veronica chamaedrys L.	50	75	50	125	I	15
Epilobium palustre L.	50	25			I	18	III	58	50	25
Stellaria graminea L.	25	63	50	75					100	10
Mentha aquatica L.			25	63	V	238	V	250	100	250
Dactylorhiza incarnata (L.) Soó					I	1
Cirsium arvense (L.) Scop.			75	138			V	142	100	250
Plantago major L.							III	12
Ranunculus sceleratus L.									50	25
Urtica dioica L.			50	125
Betula sp.			50	75

Ch—characteristic species of a particular syntaxon; D—differential species of a particular syntaxon, Cl—phytosociological class; O—order; All—alliance; Ass—association. * C—Constants (expressed on a five-point scale percentage of relevés of a given community in which its presence was recorded, where V—species occurs in 81–100% of relevés, and I—species occurs in 1–20%, calculated for each species in a community in which at least 5 phytosociological relevés were done), F—frequency (percentage of photos in which a given species occurred relative to the total number of photos of the community; in the case of a small number of the relevés), D—average coverage (sum of the mean percentage coverage of the species multiplied by 100 and divided by the number of relevés).

). All of the patches were dominated by the species characteristic for the Molinio-Arrhenatheretea class. However, they differed in the share of associated species from the Scheuchzerio-Caricetea nigrae, Phragmitetea classes, and non-specific taxa (Figure 3). Only mid-forest meadows were classified at the Molinio-Arrhenatheretea class level. Their floristic composition refers to Filipendulion, which indicates a progressive direction of their transformation.

The largest number of species was recorded in the near-forest community (59 in total), whereas in the remaining phytocenosis, there were no more than 50 in total (

Table 3. Species diversity in meadow phytocenosis grazed by Koniks in BNP.

Communities	Total Number of Species	Species Richness (S)	Shannon–Wiener Index (H’)	Fodder Value Score (FVS)
1. Northern meadow	46	32.0	2.58	4.5
2. Mid-forest meadows	50	27.5	2.53	4.0
3. Sedge community	48	22.5	2.17	3.2
4. Near-forest meadow	59	30.5	2.41	3.2
5. Agrostis canina meadow	48	42.5	2.77	4.6

). The species richness differed among the types of meadows. The highest average number of species per relevé and the value of the Shannon–Wiener index was determined at the Agrostis canina meadow (S = 42.5, H’ = 2.77), whereas the lowest was in the sedge community (22.5 and 2.17, respectively). All the distinguished communities were characterized by the poor fodder value, which was slightly worse for the sedge community and near-forest meadow.

The values of ecological indices L, T, and K for all of the communities were very similar (Figure 4). Lower values of the L index in the case of the northern and mid-forest meadows suggest a slightly larger share of species requiring less sunlight within those communities, which results from a greater abundance of tall herbs in the sward. The entire area under study was characterized by a high level of soil humidity (F), with the southern part of the enclosure—especially sedge phytocenosis—being the most humid. All of the discussed communities were located in areas with moderately acidic soils (R) and a low nitrogen content (N). The highest values of the latter were obtained in the Agrostis canina meadow as well as mid-forest meadows.

In the sedge community and near-forest meadow, we identified plant species protected in Poland according to the Regulation of the Ministry of the Environment [44]. Under strict protection is Carex buxbaumii, whereas Betula humilis, Gentiana pneumonanthe, and Iris sibirica requires an active protection for its maintenance. The last one—Dactylorhiza incarnata—is subject to partial protection.

3.2. Sward Structure

The sward height in the area under study changed not only during the vegetative season but was also dependent upon the manner of management and weather conditions during a given year (

Table 4. Average sward height [cm] and sward cover [%] in grazed areas and in areas excluded from grazing in selected grassland communities during the vegetative season.

	Month			Year		Community				Management
Trait	April	June	August	2009	2010	1	2 *	3	4	Grazed	Excluded from Grazing
Sward height [cm]	3.51 a	20.00 c	14.75 b	8.82 a	16.69 b	10.64 a	18.19 c	11.95 b	10.24 a	11.25 a	14.26 b
Sward cover [%]	34.74 a	78.46 c	61.63 b	57.80 a	58.75 a	64.48 c	83.88 d	34.73 a	50.02 b	54.92 a	61.642 b

Tests were performed for each factor separately; different letters indicate a significant difference, p ≤ 0.05 * meadows unmowed in 2009 and 2010; 1—northern meadow, 2—mid-forest meadows, 3—sedge community, 4—near-forest meadow. The destruction of the sward by wild boars made it impossible to establish a fenced area excluded from grazing in the Agrostis canina community.

). The vegetation on grazed sites was on average 3 cm lower than on sites excluded from horse grazing. The long inundation of 2010 promoted plant development, with the sward being almost 8 cm higher on average than during the first year of the study.

The grazing of Konik horses had a significant impact on the height of sward in three communities, mid-forest meadows, the sedge community, and near-forest meadows (Figure 5a), at the beginning of the vegetative season (third week of April). It needs to be emphasized, however, that the northern part of the enclosure was rarely visited by horses [45]. The greatest differences between the grazed and ungrazed sites were observed in June. Moreover, it was found that the sward was higher in 2010, which could have been caused by better soil moisture conditions (Table 4, Figure 5b).

The sward cover in all the communities changed during the growing season (Table 4). The lowest average sward cover was noted in early spring (April), when the level of ground water was still high, whereas the highest was in June, before mowing (Figure 5c,d). Periodically flooded communities situated in the southern areas were characterized by a relatively low sward cover, significantly lower than in the northern areas and on unmowed mid-forest meadows. The management (grazing/no grazing) also affected the sward cover, which was less in grazed sites (by 6% on average) compared to areas excluded from grazing.

With the exception of the northern meadow, horse grazing changed the sward cover of all communities in a significant manner (Figure 5c). The impact of Konik horses on the sward in a sedge community and mid-forest meadows was noted in early spring, whereas in June, their influence was especially noticeable on mid-forest meadows. The sward cover of the northern meadow foremost depended upon the weather conditions in individual years, as observed in April and June (Figure 5d).

3.3. Botanical Analysis of the Sward

The communities differed significantly in terms of their botanical composition (

Table 5. Botanical composition of the sward [% DW] in selected grassland communities.

Characteristic	Group of Plants [% DW]
	Sedges	Grasses	Other Monocot and Dicot Species
Community
1. Northern meadow	26.49 c	25.89 c	47.62 a
2. Mid-forest meadows	13.95 c	36.33 d	49.71 a
3. Sedge community	77.16 b	15.13 b	7.71 b
4. Near-forest meadow	52.68 a	27.02 a	20.30 b
Year
2009	37.05 a	25.39 a	37.56 a
2010	48.09 b	26.80 a	25.12 b
Management
Grazed	43.92 a	24.45 a	31.64 a
Excluded from grazing	41.22 a	27.74 a	31.04 a

Tests were performed for each group of plants and factors separately; the letters indicate a significant difference, p ≤ 0.05.

). The monocotyledonous species from the Carex and Poaceae families dominated in the sward in wet habitat with the sedge community (almost 93%, of which Carex sp. constituted 77%). It was similar in the near-forest meadow, were species belonging to Carex accounted for 53% (in total, monocotyledonous approx. 80%). The percentage share of sedges and grasses in the sward on other meadows amounted in total to only ca. 50%. It was not observed that the grazing of Konik horses would have an impact on the percentage share of plant groups.

The weather conditions had a significant impact on the floristic composition of plant communities (

Table 6. The share of particular groups of plant species [% DW] in communities under study in 2009 and 2010.

Community	Year	Carex sp.	Poaceae sp.	Other Species	Dominating Species (Two-Year Average)
1. Northern meadow	2009	46.06 a	25.53 a	58.66 a	Carex panicea, C. elata, C. flava (46%), Molinia caerulea (18%), Potentilla erecta (12%), Agrostis canina (4%)
	2010	59.30 ac	26.26 a	36.59 b	Carex panicea, C. elata, C. flava (59%), Molinia caerulea (8%), Potentilla erecta (2%), Agrostis canina (13%)
2. Mid-forest meadows	2009	15.82 b	33.61 ab	55.95 a	Carex panicea, C. elata, C. flava (16%), Agrostis canina (24%), Molinia caerulea (4%), Potentilla erecta (3%)
	2010	37.15 a	39.06 b	43.48 ab	Carex panicea, C. flava (37%), Thalictrum flavum (20%), Molinia caerulea (3%), Agrostis canina (5%)
3. Sedge community	2009	75.89 c	16.38 a	7.72 c	Carex panicea, C. flava (76%), Molinia caerulea (11%), Agrostis canina (3%)
	2010	78.43 c	13.87 a	7.70 c	Carex panicea, C. flava (79%), Molinia caerulea (4%), Agrostis canina (2%)
4. Near-forest meadow	2009	10.44 b	26.04 a	27.90 b	Carex panicea, C. elata, C. flava (10%), Potentilla erecta (11%), Molinia caerulea (9%), Agrostis canina (15%)
	2010	17.46 b	28.01 ab	12.70 bc	Carex panicea, C. flava (16%), Molinia caerulea (8%), Agrostis canina (8%), Thalictrum flavum (18%)

Tests were performed for each plant group separately; p ≤ 0.05 the letters a, b, c indicate a significant difference, p ≤ 0.05.

). In 2010, in all phytocenoses, high water levels promoted sedge species, that is Carex panicea, C. elata, and C. flava, which are associated with moist habitats. It was especially visible in the northern meadow and mid-forest meadows. In the second year of the study, a smaller share of Molinia caerulea and Agrostis canina in the sward were noted. In 2010, also the growth of other dicotyledonous species was limited.

4. Discussion

The communities from the Molinion alliance, which dominated at the studied area, belong to the most nature-valuable communities in Poland [46,47,48,49] and Central Europe [50,51]. Their conservation is important not only in terms of plant species, but also for the protection of meadow-nesting birds [52,53]. Unfortunately, the area of Molinia meadows in Europe has gradually decreased due to the intensification of management or abandonment of less productive types of meadows [54,55,56]. So far, the extensive grazing of Polish primitive horses has been used to protect heathlands or sandy calcareous grasslands [23]. A large share of Molinia meadows distinguishes reserve breeding in the BPN from others in Poland and Europe. The maintenance of these communities, in order to prevent succession processes, requires the extensification of management, with a one-time mowing until the end of the vegetation period and without fertilization [5,46,47,48,55], whereas grazing is rather not recommended (even if it is extensive), because it may be a threat to the stability of the meadows [57]. The aim of establishing the reserve breeding in the BPN was to enhance the biodiversity of the plant species and inhibit the process of secondary succession. In this regard, horses are kept at the very low stocking rate in order to prevent the degeneration of the present phytocoenosis. After 6 years of Koniks horses grazing, the Molinia meadows are considerably well preserved, as evidenced by the Shannon–Wienner Diversity Index which for all of the distinguished communities, exceeded the 2.1 value. The greatest diversity of the species was noted in mid-forest meadows. This might have resulted from their small area and the edge effect, e.g., [58] and the cessation of mowing, which consequently meant that they contained species characteristic of both maintained and abandoned meadows [48].

Animals on a pasture influence the sward not only by grazing, but also by trampling, rolling, and leaving dung. The year-round grazing of Konik horses kept at a low stocking rate in the BPN has little effect on the grasslands communities. The impact of the presence of horses was primarily manifested by changes in the structure of the sward, i.e., on average, a 3-centimeter reduction in its height and a 6% reduction in the sward cover. The lack of such changes in all communities in the late summer period results from the annual mowing of meadows performed at the turn of July and August. Konik horses in BPN affects the phytocenosis only locally. Animals had the greatest impact on regularly grazed communities (near-forest meadows); as well as on smaller meadows grazed more intensively, but only periodically (mid-forest meadows; c.f. [45]). The reduction in the sward cover of a meadow in the BPN in conditions of grazing should be considered as unfavorable as the degradation of the sward and the exposure of the soil surface accelerates the peat degradation processes. The lack of the effect of grazing on the percentage share of particular plant groups in the communities could result from the short period of the study and the low stocking rate of horses.

The maintenance of an extensive management (mowing performed once a year by the end of the growing season) is necessary to preserve the current state of the non-forest community located in the area of the BNP enclosure. The abandonment of the Molinia meadows would initiate a secondary succession process, leading to the encroachment of alder woodlands with willow and birch (Alnetea glutinosae class) [5,47]. Moreover, it is necessary to control the residues, which are mainly herbaceous plants that are not preferred by Konik horses or avoided by them due to dung contamination. In the long term, the lack of mowing may also result in the formation of patches dominated by unfavorable nitrophilous species, e.g., Urtica dioica or Cirsium arvense, which in turn could lead to a reduction in the biodiversity of the area [59].

The most important threats for the conservation of the Molinia meadows, apart from maintaining extensive mowing management, are changes in the water conditions. The Molinia meadows in the BNP, including these grazed by Konik horses, were created as a result of extensive management and formed mainly from wet communities belonging to the Scheuchzerio-Caricetea nigrae and Phragmitetea classes [5,60]. An additional factor contributing to their formation were changes in the water proportions, with the construction of the Woźnawiejski Canal in the first half of the 20th century [61]. The results of the botanical analyses showed the importance of the maintenance of appropriate habitat moisture as well as the amount of precipitation in a given year, in terms of delaying shrub encroachment. The higher sward of plant communities recorded in 2010 proves that the high level of groundwater not only contributes to the increase in the availability of nutrients for plants, but also promotes the growth of taxa associated with highly wet habitats (c.f. [62]). Soil moisture, in turn, may affect the distribution of plants by changing their competitive ability [63]. This is important in light of global climate changes, which in themselves contribute not only to the increase in average temperatures, but also to changes in the precipitation distribution and the increasing number of dry days per year [64]. The final pattern of peat-meadow succession depends not only on soil moisture but is also limited by the availability of nutrients [54]. The drying process accelerates peat decomposition and leads to changes in the fertility of habitats. The subsequent succession stages additionally contribute to the desiccation of the habitat as a result of an increased evapotranspiration of trees and shrubs in comparison to grassland communities [65,66]. Additionally, as demonstrated by Kołos and Banaszuk [67], in the case of disturbed hydrological conditions, extensive mowing itself would not be adequate for conserving fen meadows because it cannot suppress the encroachment of rush species, such as Calamagrostis canescens or Phragmites australis.

The values of the Ellenberg’s Indicators determined in our research showed that the habitats grazed by horses were still very moist, acidic, and nitrogen-poor. This is consistent with the literature data obtained for other communities from the Molinion alliance, which are usually found in nutrient-poor soils that are moist or sometimes dry during the summer season [48,51,68]. This, in turn, raises the question of whether the vegetation fulfils the macro- and micro-element needs of horses, especially due to the poor utility value of the sward. The Polish primitive horses seem to be well adapted to the reserve breeding because during the growing season, the animals were in good welfare and only needed supplementary feeding during the winter. Providing horses with hay from other areas (not only from the mowed grasslands located in the enclosure) poses a risk of an increase in the amount of nitrogen and phosphorus present in the habitat. Although, a one-cut system with a limited fertilizer application for the production of hay or fresh biomass or extensive grazing may be an appropriate way to preserve the Molinia meadows and increase the nutritive value of the crop [52]; however, they are not recommended in BNP due to bird protection (e.g., black grouse) and the risk of changes in the botanical composition.

5. Conclusions

The reserve breeding of Polish primitive horses in the Biebrza National Park is an example of the use of low stocking grazing as an element of active protection. This is especially important in wetlands where access is difficult but animals can stay there all year round. In the short-term aspect, the weather conditions, especially the duration of floods, have a clear impact on the species composition of communities, which is even greater than grazing. Thus, in order to protect the Molinia meadows, apart from mowing once a year, it is necessary to maintain appropriate habitat conditions, especially in terms of the level of moisture. Further research is needed on the behavior, habitat, and diet preferences of Konik horses, the knowledge of which will allow for the wider use of grazing in wetlands.