Semi-natural grasslands are environments with a long history of traditional, low-intensity pastoral and agricultural activities (Gorris et al. 2025). These environments have largely developed as a result of various agricultural land-use practices that replaced forest areas – the local potential natural vegetation (Poschlod and WallisDeVries 2002). It is widely documented that seminatural grasslands possess high ecological value and serve as global biodiversity hotspots, providing essential ecosystem services ranging from food production to carbon sequestration (Katoch et al. 2025; Milić et al. 2024; Shipley et al. 2024). The interaction between management practices, like extensive grazing, and abiotic factors, such as precipitation patterns, determines the ecological stability of these habitats (McSherry and Ritchie 2013). Despite their importance, these ecosystems are increasingly threatened by land-use change, abandonment or intensification, and by climate change, both leading to species loss and degradation of grassland ecosystem (Lv et al. 2024) with a detrimental effect on the vital ecosystem services that they provide (Prangel et al. 2024). Understanding how grasslands respond to these stressors requires high-resolution data on plant community composition and plant functional traits. Research indicates that grasslands undergo rapid compositional reorganisation in response to drought and temperature shifts (Zhu et al. 2024). To isolate these effects, manipulative experiments using standardised protocols, such as the International Drought Experiment (IDE) framework, are essential for simulating altered precipitation regimes (Knapp et al. 2017). The dataset presented here was collected within the two-year project “CAROLINA: Impact of land-use change on climate resilience of semi-natural grasslands”. An important part of the research focuses on the short-term impact of abandonment, and variation in precipitation regime on soil biochemical properties and on taxonomic and functional diversity of vascular plants across different climatic zones. Based on the Köppen-Geiger climate classification (Beck et al. 2023, https://koppen.earth/), the selected areas encompass: hot-summer Mediterranean coastal plains (Csa), the hilly transition between Mediterranean and warm temperate climates in the Central Apennines (Csa/Cwa), and the high-altitude gradient of the Alps, ranging from humid continental (Dfb) to subarctic (Dfc) conditions. The project employs a field-based manipulative approach to simulate grazing exclusion (mimicking grazing abandonment) and precipitation reduction (mimicking future drought scenarios). The presented dataset integrates two dimensions of plant biodiversity: i) taxonomic, and ii) functional (including leaf morphological and plan ecophysiological traits). This paper provides a comprehensive repository of data collected as baseline measurements during the initial characterisation of three experimental sites (between 2024 and 2025) and one year after treatment application for two of them (2025). The dataset includes: i) Taxonomic diversity data, reported as floristic-vegetational surveys; ii) Plant functional diversity data, reported as leaf morphological functional traits, and carbon and nitrogen isotopes measurements.

The dataset aims to take an important step towards understanding the dynamics linking the two diversity layers and identifying the best management and conservation practices for these valuable grassland habitats and their characteristic plant communities. By providing a comprehensive multidimensional framework of plant structural and performance-based differences, these data offer a foundational resource for long-term monitoring and ecological modelling of Mediterranean and temperate grassland responses to global change.

The experimental phase of the project was designed to study the short-term effects of grazing suppression through fencing, as well as the impact of reduced precipitation, which was simulated using rain shelters. The experiment was conducted across three sites. San Rossore (43°44'06.5"N, 10°19'31.9"E, 5 m a.s.l.), located in the Migliarino, San Rossore, Massaciuccoli Regional Park extending along the coast of the Tuscany region (Northwestern Italy), is characterised by a mild and dry climate (mean annual temperature of 16.2 °C and approximately 540 mm of rainfall, Csa climate zone of the Köppen-Geiger climate classification), and lays on a recent alluvial geological substrate composed of sandy and silty deposits from the Serchio and Arno rivers. Within this setting, the alternation of wetlands and sandy ridges supports hygrophilous and mesophilous herbaceous vegetation, maintained through extensive rotational cattle grazing. San Venanzo (42°49'38.6"N, 12°15'52.6"E, 500 m a.s.l.), which is found on private pastures in the Umbrian Apennines (Central Italy), is characterised by a temperate continental climate (mean annual temperature of 12.9 °C and 801 mm of rainfall) as it is situated in a transition zone between hot-summer Mediterranean (Csa) and more temperate climates (Cwa), and it rests on a unique Pleistocene volcanic substrate known for the presence of rare rocks like Venanzite. This geological setting, shaped by the San Venanzo volcanic complex, supports a landscape mosaic where Turkey oak (Quercus cerris) forests alternate with open pastures, which were restored in the 1980s for extensive cattle grazing. Tesino (46°06'51.7"N, 11°42'11.7"E, 1667 m a.s.l.) is located in the Passo Brocon area (Cinte Tesino), within the Lagorai mountain range of the Eastern Alps in the Trentino-Alto Adige region at an elevation of 1667 m a.s.l. The site features a harsh subalpine climate spanning the humid continental (Dfb) and subarctic (Dfc) zones (mean annual temperature of 5.5 °C and 1866 mm of rainfall), and sits upon a geological substrate of metamorphic rocks, specifically the quartz phyllites of the Lagorai chain. These acidic, nutrient-poor soils support subalpine grasslands dominated by Nardus stricta and Norway spruce (Picea abies) forests, maintained through seasonal cattle grazing managed under local municipal leases. Climate data are provided by the Euro-Mediterranean Centre on Climate Change (CMCC) (Adinolfi et al. 2021; Raffa et al. 2023).

At each location, an area of 135 m² was fenced off to prevent cattle from grazing. Specifically, at the San Rossore site, grazing management involves 15 head of cattle occupying an area of 3.4 hectares. The San Venanzo pasture covers 60 hectares and supports a herd of 25 cattle, while in the Tesino subalpine pastures, a herd of 51 cattle grazes an area of 15 hectares. Three treatment levels were established, each with three replicate blocks (2 × 3 m): Control (G as per ‘grazed’), which was located outside the fence and exposed to ambient precipitation and extensive grazing; Grazing Exclusion (GE), which was located inside the fence and exposed to ambient precipitation; and Grazing Exclusion + Drought (GED). A schematic figure of the experimental design is available in the Suppl. material 1. Following the international Drought-Net protocol, rain-out shelters were used to simulate extreme drought, targeting the 1^st percentile of long-term local precipitation. This resulted in site-specific expected reductions in rainfall of 57% (308 mm) for San Rossore, 50% (400 mm) for San Venanzo, and 33% (616) for Tesino. The sites’ details are summarised in Table 1. Figure 1 shows the location of the sites on a map. The sampling design and methodology employed to collect the presented data are outlined in the following paragraphs.

This multi-proxy dataset provides robust data with which to investigate the interactions between land-use abandonment and climate stressors in semi-natural grasslands. By integrating taxonomic and functional layers, the dataset serves as a foundational resource for ecological modelling and predicting the climate resilience of plant communities in Mediterranean and temperate regions. While this dataset offers a multidimensional perspective on grassland resilience (Figures 2, 3, 4, 5), it is important to acknowledge its limitations, including the short-term nature of the post-treatment observations and the geographical restriction to three Italian sites. These limitations may prevent the dataset from capturing the full range of environmental variability across the Mediterranean and Alpine biomes. Nevertheless, this comprehensive repository could support the development of evidence-based management practices and serve as a critical asset for long-term monitoring and global meta-analyses.