Lake Palić and Lake Ludaš, located near the city of Subotica, are shallow, semi-static lakes of steppe areas that are among the rare and well-preserved steppe lakes of the Pannonian region (Table1; Figure 1). As such, they are protected natural areas of the Republic of Serbia. The wetland habitat complex and steppe fragments, along with Lake Ludaš, constitute the SNR “Ludaš Lake” (Official Gazette of the Republic of Serbia, 2006), a legally protected natural area (Official Gazette of Republic of Serbia, 2018). Since 1977, this area has been classified as a wetland of international importance (Ramsar Site 3YU002) (Ramsar, 2024). Lake Palić has been protected since 1996 and declared as NP “Palić” (Official Gazette of the City of Subotica, 2017). The SNR “Ludaš Lake” and NP “Palić” wereformally connected in 1982, establishing the Palić-Ludaš Regional Park.

The studied area is characterized by a temperate continental climate. The Digital Climatic Atlas of Serbia provided more extensive climatic data for the study sites (MEP, 2025). For the IPCC’s RCP 8.5 scenario, daily mean temperature and mean daily sum of precipitation were subsets of daily observations for 1951–2020 and projected values for the same parameters for 1951–2,100. Observed and predicted values of temperature and precipitation are presented in Figure 2, and the difference (Δ) is displayed in Supplementary Tables S1 and S2.

The studied area is part of the forest-steppe zone, characterized by a vegetation type which includes sandy and steppe components, interspersed with patches of salt marshes, meadows, wetlands, and remains of deciduous forests (Beloica et al., 2024).

Ecological restoration was carried out in the area of the SNR “Ludaš Lake” and NP “Palić”, following the principles of preserving forest-steppe ecosystems (Supplementary Table S6). During the planning of the restoration, all relevant legal and ecological regulations were taken into account, including the Nature Protection Act of the Republic of Serbia, the Ramsar Convention (Ramsar, 2024), the Natura 2000 network (EEA, 2021; European Commission, 2022a; European Commission, 2022b), as well as the Detailed Regulation Plan adopted by the City of Subotica.

Species selection for the buffer zone was based on ecological site assessment, the analysis of the natural potential vegetation, and observation of existing vegetation on-site. Information about natural vegetation was found in the EuroVegMap (Bohn et al., 2004) and information about existing vegetation for Palić-Ludaš region was obtained from field observations and from documents provided by Institute for Nature Conservation of Vojvodina Province (INCS, 2004; Institute for Nature Conservation of Vojvodina Province, 2011), and information about conditions of neighboring regions, specifically region of Nyíregyháza in Hungary (Török et al., 2017).

The EuroVegMap database (Bohn et al., 2004) identifies the predominant vegetation types in this region as G4 – Pannonian mixed Tatarian maple-pedunculate oak forests, L10 – South Pannonian herb-grass steppes, L11 – Pannonian sand steppes, and P32 – Pannonian salt meadows. Based on the study of natural potential vegetation and habitat types, and in accordance with Natura 2000 guidelines, three priority habitats have been identified: 1,530 – Pannonian salt steppes (Šefferová Stanová et al., 2008b), 6,250 – Pannonian loess steppe grasslands (European Commission, 2013), and 6,260 – Pannonian sand steppes (Šefferová Stanová et al., 2008a).

Establishing the mosaic structure characteristic of forest-steppe ecosystems, the restoration was planned to achieve a ratio of 60:40 or 70:30 in favor of grassland vegetation in forest-steppe mosaic (Erdős et al., 2022), which was the composition pursued throughout the project. This approach aligns with Erdős’s observation that forests typically occupy between 10% and 70% of such landscapes (Erdős et al., 2022).

The basic rules for planning the establishment of vegetation in buffer zone were:

At least 20% of the planned species must be from the EuroVeg list for the corresponding type of natural potential vegetation, i.e., type of habitat,

After collecting all available data, a list of species from all above-mentioned documents was extracted:

Tree layer: Acer tataricum, Acer campestre, Quercus robur, Quercus cerris, Quercus pubescens, Quercus petraea, Ulmus minor, Ulmus leavis, Frangula alnus, Fraxinus angustifolia, Pyrus pyraster, Salix alba, Salix cinerea, Populus alba, Populus x canescens, Populus nigra ‘Italica'

This project integrates the cultivar P. nigra ‘Italica’ planted in protective belts along the edges of agricultural fields, which are the historical element of cultural landscape of Vojvodina (Beloica et al., 2024). Supporting ecological principles by adding eco-cultural and biocultural elements into ecological restoration projects is important in the buffer zones, where local communities play a crucial role in maintaining and safeguarding landscape values (Sena et al., 2021).

As part of this project, a total of 2,877 tree and shrub seedlings were planted around Lake Ludaš in 2020 and 2021, and 3,433 seedlings were planted around Lake Palić, forming a narrow strip ranging from 5 to 30 m in width (Supplementary Table S3). A total of 6,310 planted individuals (n = 6,310) were monitored in this study. The planting was designed to include keystone species (species crucial for the ecosystem), as well as plants with significant ecological functions, such as pollinator and nectar species, and species whose fruits serve as winter food for birds (Ecolacus, 2022). For the establishment of the buffer zone, species that are naturally present in the area were used, based on the assumption that all plants would successfully adapt and achieve good results.

The planting material was nursery-grown, with seedling sizes specified in Supplementary Tables S4, S5. To ensure sustainability and success, and in accordance with the Detailed Regulation Plan and the Ecolacus project “Biodiversity and Water Protection for Lake Palić and Lake Ludaš” (Ecolacus, 2022) efforts were made to remove invasive species through clear cutting. All seedlings are protected from adverse nature and other influences after planting. To achieve this, it is necessary to provide the best possible maintenance (including watering, mulching, and pruning). Supplementary Table S6 shows terrain preparation, planting, maintenance, monitoring, and project documentation.

Observations were made on the success of 29 native tree and shrub species during 2022 and 2023 (Supplementary Table S6). The success of planted species was assessed based on visible signs of vitality, including the presence of leaves, flowering, or fruiting, and the absence of serious physical damage from wildlife or machinery. Seedlings that were dry, dead, or significantly damaged were classified as failing.

In this study, we used several coupled models that integrate climate parameters, soil characteristics, and vegetation dynamics (Figure 3). By integrating climate projections and site-specific data into models, scenario analysis helps predict vegetation responses, optimize restoration strategies, and enhance long-term ecosystem resilience.

At the NP “Palić” site, 85.37% of the 3,433 seedlings planted in 2020 and 2021 were found during the 2022 monitoring, and 72.15% were found during the 2023 monitoring. In 2022, 2,877 seedlings were planted in the SNR “Ludaš Lake” area, which had a planting success rate of 90.30%. There was a decline in 2023, with the success rate dropping to 69.89%.

The radial graph (Figure 4) illustrates the prevalence of different tree and shrub species in 2023, the final year of monitoring, highlighting differences in species performance across locations. Several species have successfully adapted to the conditions in the studied area. For example, Acer campestre had a survival rate of 92.2% at Palić and 88.8% at Ludaš. Acer tataricum had a high survival rate, with 90.4% in Palić and 94.9% in Ludaš, despite numerous instances of damage caused by wild animals.

Other notably successful tree species are Pyrus piraster (88.9% at Palić, 77.5% at Ludaš), Ulmus leavis (98.4% at Palić, 74.6% at Ludaš), and Ulmus minor (73.1% at Palić, 94.6% at Ludaš). Despite being planted in fewer numbers, Salix alba shown remarkable adaptation, with a 100% survival rate at Palić and 83.3% in Ludaš. Ligustrum vulgare demonstrated high survival, with 87.8% at Palić and 77.1% at Ludaš. Cornus sanguinea adapted well on this area, with 72.6% survival rate at Palić and 71.2% at Ludaš.

Populus alba showed a notable decrease, with survival rates falling to 50.51% in Palić as well as 39.04% at Ludaš. Quercus robur had significant losses, with a survival rate of 42.40% at Palić and 50.91% at Ludaš. Shrub species showed vulnerability, especially Euonymus europaeus, where the success rate on Palić was 60.90% and on Ludaš was 55.43%.

According to the analysis of the VSD + PROPS model, the species with the highest probability of occurrence (the maximum HSI) are shown in Figure 5, while specific locations are illustrated in Figure 1. Furthermore, the HSI distribution for each lake separately is provided in Supplementary Figure S2. In the area of the NP “Palić”, it is evident that the species Crataegus monogyna and Cornus sanguinea has the highest HSI index values, with a maximum value of 0.52. Also, species such as Ligustrum vulgare, Prunus spinosa, P. fruticose, Quercus petraea, Quercus pubescense and Fraxinus angustifolia record HSI values greater than 0.3. The most suitable locations are P1, P2, P11, P12, and P18.

At the Ludaš Lake region (Figure 5), Crataegus monogyna is distinguished as the sole species with maximum values of 0.52. Subsequently, elevated values over 0.3 are noted for the species Ligustrum vulgare, Prunus spinosa, P. fruticosa, and Quercus pubescens. The most suitable locations in this area are L6, L7, L9, L10, and L12.

The model projection indicates higher values of HSI for shrub species (Cornus sanguinea, C. monogyna, L. vulgare, P. spinosa, Prunus fruticosa) compared to tree species, indicating a shift towards shrub-dominated communities in future climate scenarios. Visual summary of the most successful species, based on survival and adaptation rates as shown in Figure 6.

Based on the RCP 8.5 climate scenario, the projected Habitat Suitability Index (HSI) values in the study area suggest a decrease in habitat suitability. The initial HSI values are approximately 0.3, declining to under 0.2 at the end of century (Figure 7). The average HSI demonstrates a progressive decline over time, with minimal variation observed in the Palić region. Similar projections are observable at the majority of monitoring sites. A progressive increase is evident at P2, whereas the most significant decrease is recorded at P19, with values falling below 0.2 by 2,100.

In the area of Ludaš, initial HSI values are around 0.3, and despite fluctuations over the years, values remain around 0.2 by the end of the century (Figure 8). A relatively stable trend is observed at several sites L9, L12, L6, L7, whereas greater variability is recorded at locations L4 and L7. Notably, location L4 demonstrates a more distinct upward trend in habitat suitability by the end of the century.

To minimize the influence of site-specific management practices, such as differences in maintenance intensity caused by limited accessibility of the Ludaš Lake site, correlation analyses were performed not only between the number of planted and surviving seedlings within each site, but also between the number of planted seedlings at one site and the total number of surviving seedlings across both study areas (SR). This cross-site approach allowed partial control for human-related maintenance effects and provided a more robust assessment of planting success.

Correlation values based on the full species list of modeled future occurrence probabilities (VSD Ludaš and VSD Palić) with SRP (surviving rates Palić), SRL (surviving rates Ludaš), or SR (integrated survival rates across both study areas) metrics are low and statistically non-significant (Table 2). In contrast, the reduced species list of modeled future occurrence probabilities (VSD Ludaš reduced and VSD Palić reduced) with SRP, SRL, or SR metrics shows strong and statistically significant correlations (0.59–0.90), demonstrating that removing unreliable species substantially improves the ecological signal and predictive strength of the VSD indicators. This indicates that the model provides reliable predictions for a specific subset of species, whereas certain other species may require model recalibration or the use of alternative models that incorporate additional ecologically relevant variables.

Model simulations indicate that five species (Acer campestre, Acer tataricum, Pyrus pyraster, Ulmus leavis, Ulmus minor) require a different set of predictor variables, suggesting that either the variable set or the model parametrization (or the input data themselves) should be improved. The discrepancy may stem from the observational datasets on which the model was originally trained, which predominantly represent conditions in Central and Northern Europe. We consider that incorporating relevé data from our study region could yield more accurate correlation patterns. Similar issues have been reported in previous studies with Ellenberg indicators (Dengler et al., 2023; Tichý et al., 2023), which subsequently led to the publication of country-specific indicator values precisely to account for such biogeographical differences. These inconsistencies suggest that additional species-specific ecological parameters may need to be incorporated to improve model performance and predictive robustness in future simulations.

Resolution 73/284 of the United Nations declared the Decade on Ecosystem Restoration for the period from 2021 to 2030, with the aim of preventing, halting and reversing ecosystem degradation. Although actively involved in cooperation with the Food and Agriculture Organization (FAO) and the UN Economic Commission for Europe (UNECE) and promoting ecosystem restoration, the Republic of Serbia has been delayed in presenting clear regulations on how it approaches or acts in achieving this goal. In accordance with the Law on Climate Change, the Ministry of Environmental Protection has formulated the “Program of Adaptation to Climate Change for the Period 2023-2030″with the Action Plan. Additionally, through numerous regulations and reports, the Republic of Serbia is working to improve its goals in line with the agendas of the European Union (Beloica et al., 2023).

During the project implementation, we encountered an important challenge with the lack of appropriate planting material suitable for this type of ecological restoration at national level. This deficiency highlights one of the major shortcomings in the practical implementation of the UN Decade on Ecosystem Restoration. National nurseries must be closely aligned with long-term restoration plans and adequately prepared to meet the projected needs at least a decade in advance. To improve the results of reforestation and ecological restoration itself, it is crucial to ensure the availability of high-quality planting material of native species and the implementation of effective monitoring and maintenance practices (Werden et al., 2024). After the technical restoration and species introduction phase, the ‘nature participation’ restoration phase can be expected and represents a natural and complementary stage of the restoration process. Such natural dynamics may significantly impact long-term ecosystem recovery (Kudryavtsev, 2007; Prach et al., 2013). It is crucial to monitor how restored areas continue to develop naturally, tracking the progression toward higher levels of naturalness and the dynamics of ecological succession.

An additional indicator of the long-term sustainability of such projects is the allocation of dedicated funding for site managers responsible for maintaining the restored areas. These financial resources must be carefully planned, as the costs associated with maintaining newly established restoration sites cannot be directly compared to those of previously existing, stabilized areas. Newly restored ecosystems typically require more intensive and frequent management activities, especially during the first 5 years, which are critical for the successful establishment and stabilization of these habitats. Therefore, long-term financial planning and secured budgets are essential to support the success and resilience of restoration initiatives. Short project timelines often conflict with the principles of ecological restoration, especially when complex administrative procedures and permits are required for protected or large restoration sites, a challenge, which is also widely reported in international restoration practice (Oluwajuwon et al., 2025). A further challenge in Serbia is the 5-year maintenance period: public enterprises usually lack the capacity for long-term management, which frequently results in project degradation. During the planning phase, we found that using older planting stock rather than seedlings can significantly improve establishment success and long-term survival. Solving these challenges will be essential for Serbia to achieve its ecological restoration goals during the current decade.