**1. Introduction**

The Pannonian-Pontic environmental zone (PAN) occupies the major part of the Carpathian Basin. The area is characterized by natural forest-steppe and steppe vegetation [1,2].

Numerous studies were carried out investigating the sandy areas of the Pannon region. The first significant review was published by Zólyomi [3]. Szujkóné–Lacza [4] assembled and reviewed the literature

and collections of the Danube-Tisza Interfluve, while also using the data of the Botanical Department of the Hungarian Natural History Museum. However, surveys usually dealt with the central region of Hungary, which includes the most natural (i.e., relatively intact) parts. Owing to these studies, the temporal structure, the distinguishable aspects, and aspect-forming species of *Festucetum vaginatae* Simon 2000 are well known [5].

The zonal arrangemen<sup>t</sup> of the soil types and climate, which are characteristic of the eastern part of the continent, disbands completely and gives way to a mosaic-like landscape in the Carpathian Basin [6–8].

Both climazonal and edaphic mosaic habitats from steppe patches to forested areas [9–11], developed in sandy soil, which was altered and restricted as a result of landscape managements [12,13].

In these areas with calcareous soil in the central parts of the Carpathian Basin, the environmental factors are mosaic-like as well [14–19]. The present study examines the occurrences of the steppe and forest-steppe vegetation on the edge of the Pannonian forest-steppe region. The river Ipoly is one of the last rivers that have been preserved in their natural condition and have not been a ffected by water flow regulations. Not surprisingly, the Ipoly Valley is a protected nature reserve area of national significance, since it is part of the Danube-Ipoly National Park. In addition, it is also a nature conservation area (HUDI20026) and bird sanctuary (HUDI10008) and is subject to the Ramsar Convention in order to protect the migratory aquatic birds [20,21].

Despite its linear nature, Ipoly Valley harbors especially mosaic-like vegetation, mainly because it is a non-regulated, natural watercourse [22]. Close relations between the soil moisture level and degree of vegetation heterogeneity were also detected in other watercourses in the Pannonian region [23]. Based on earlier examinations of the Ipoly Valley, the changes of the ground water table clearly a ffect the spatial arrangemen<sup>t</sup> of vegetation types [24]. With the increased advance of agriculture, most of the grasslands were drained and plowed. The area along the Ipoly o ffers an ideal research terrain for studying the e ffects of various environmental factors such as soil on the distribution of native species. For these reasons, further examination of this area is needed.

Járdi et al. compared the coenology of the acidic sandy grasslands, steppes, meadows, and swamp meadows, which were grazed by Charolais and Hungarian gray cattle. The species pool and the cover of common species di ffered greatly in the examined grasslands, which clearly showed e ffects of the di fferent abiotic and pedologic factors, water supply, and landscape uses. In these sandy grasslands, *Festuca ovina* L. aggregate (Poaceae) and *Festuca rupicola* Heu ff. both appeared, and both *Festuca pseudovina* Hack. ex Wiesb. and *Stipa borysthenica* Klokov ex Prokudin were common in the more arid areas [25].

Examinations can be expanded with data from satellite images [26]. Sentinel-2A is the most useful satellite for this goal [27]. It was launched on June 23, 2015 as part of the European Copernicus Program.

Our questions were the following:


#### **2. Materials and Methods**

The study area covers 445 hectares and is situated in northern Hungary along the River Ipoly, in the municipality area of Dejtár, and a smaller proportion between Dejtár and Ipolyvece (Figure 1).

Manual GPS was used for recording coordinates of the sample points with consideration to the edges of the habitat patches, which were treated as separate units. For the identification of habitats, the protocol of the Hungarian Habitat Classification System ( Á NÉR) was used. This system was developed in

connection with the National Biodiversity Monitoring System (NBmR), which contains all vegetation types in Hungary [29]. It is the most frequently used complex system in the country and is under continuous development. Compared to coenological systems, NBmR is substantially simpler, as it contains fewer and broader categories. It groups associations into larger, more interpretable types and it is also feasible for practical use in nature conservation [30,31]. Association names were used according to [32]. Species nomenclature was used according to [33].

In general, one demarcated habitat patch belongs to only one habitat type (e.g., P2b see below in Table 1). However, there were habitat patches in which more than one habitat type was present. The main reason for this was that in some cases, habitat patches could not be allocated from the habitats in presentable size or they appeared as a mosaic of two or more Á NÉR categories, therefore these habitat patches were indicated as habitat complexes in decreasing order of share in the following manner: D34 × B1a × P2a. The relevant Á NÉR categories are shown in Table 1. While editing the habitat map, it was important to clearly trace and evaluate changes of habitat types.

**Figure 1.** The location of the study area in Hungary (prepared with Marosi and Somogyi 1990) [28].


**Table 1.** Habitats revealed in the study area, using the categories of the Hungarian Habitat Classification System ( Á NÉR).

Maps were created using QuantumGIS, an open source geoinformatical program, which can be downloaded from (www.qgis.org). Coordinates recorded in the field were visualized as a GSX file in the program. The Satellite images chosen for evaluation were recorded on 17th September 2019 as the cloud cover was relatively low and this date was as close to the field research's date as possible. All Sentinel-2A image was downloaded from the o fficial homepage of Copernicus. From the downloaded 12 optical bands, two optical bands were used to extract the Normalized Vegetation Index (NDVI) data. The visible red (RED) and near-infrared (NIR) were used to compute the numerical Normalized Vegetation Indices (NDVI) of each 10 × 10 m pixel [26], then the pixels were colored accordingly. NDVI is a non-dimensional value that reflects the vegetational activity of a given area. It is returned by the quotient of the sum and the di fference of the reflected intensity of NIR and RED [34]. NDVI shows the biological activity of the vegetation: the higher the reflection of the chlorophyll, the higher the value. In the absence of vegetation, NDVI will be negative, for example on water bodies in the early vegetation period. In the evaluation phase habitats mapped by using the classic field survey method were compared with NDVI of satellite images.

In order to visualize the latter, 20 points were selected randomly to each Á NÉR category, then the NDVI of the points were assigned to the categories. Data were analyzed using Microsoft Excel and PAST (PAleontological STatistics) software [35].

Mean NDVI data were compared by the non-parametric Kruskal-Wallis test, since raw data of 5 habitat types out of the 19 habitat types involved in this analysis did not fit the Gaussian distribution. Dunn's Multiple Comparisons test was used as a post hoc test, and di fferences at level *p* < 0.05 were considered significant [35].
