From Oligotrophic to Eutrophic States: Floristic Responses to Long-Term Organic and Mineral Fertilization in Mountain Grasslands

Abstract

High-Nature-Value (HNV) mountain grasslands are highly sensitive to nutrient enrichment, and understanding their responses to contrasting fertilization regimes is essential for safeguarding biodiversity and ecosystem functioning. This study compared the long-term effects of organic and mineral fertilization on floristic composition, vegetation types, and diversity in HNV grasslands in the Apuseni Mountains. After 15 years of differentiated inputs, moderate organic fertilization preserved a floristic structure closely related to oligotrophic grasslands and maintained a high degree of structural heterogeneity typical of habitat type 6520, while supporting a diverse assemblage of stress-tolerant species. In contrast, mineral fertilization induced marked shifts toward mesotrophic and eutrophic communities, reduced species richness, and promoted the dominance of nitrophilous grass. Multivariate analyses (PCoA, MRPP, and ISA) consistently distinguished the treatments, revealing distinct successional pathways shaped by nutrient sources and input intensity. Overall, the findings demonstrate the high sensitivity of HNV grasslands to fertilization and underscore the ecological importance of moderate organic inputs in maintaining community stability and characteristic floristic patterns in nutrient-limited mountain ecosystems.

1. Introduction

High-Nature-Value (HNV) grasslands are a central component of the European rural landscape, recognized for their exceptional biodiversity, role in maintaining ecological balance, and cultural and economic values in mountain and hilly regions [1,2,3]. These semi-natural systems have developed and persisted under traditional management based on mowing and extensive grazing with low external inputs, which has allowed for the long-term conservation of oligotrophic floras and distinct cultural landscapes [4,5,6]. In Romania, HNV grasslands occupy extensive areas, especially along the Carpathian chain, where they are integrated into the Natura 2000 network and play a major role in biodiversity conservation, maintenance of ecosystem services, and functioning of agro-pastoral rural economies [7,8]. In the Apuseni Mountains, long-term research has shown that the stability of oligotrophic grasslands depends on the maintenance of traditional management and the use of moderate organic inputs adapted to local soil and climate conditions [9,10]. HNV grasslands provide a broad range of ecosystem services. As provisioning services, they supply high-quality forage for ruminants, rich in proteins and minerals [11,12,13] and medicinal plants of high pharmaceutical interest, such as Arnica montana, emblematic of the Apuseni Mountains [14,15,16,17], together with honey resources and other non-timber products traditionally used in mountain households [18]. They also offer crucial regulating and supporting services, contributing to carbon sequestration, water regulation, erosion control, and soil fertility through complex ecological interactions [19,20,21,22]. Beyond their ecological functions, these mountain grasslands hold significant cultural and social value, shaping traditional landscapes and supporting local economies [23,24,25].

Despite these values, HNV grasslands are currently exposed to two major, partly antagonistic processes: agricultural intensification and land abandonment. Intensification, through the heavy use of mineral fertilizers and increased mechanization, simplifies the vegetation structure, enhances the dominance of competitive grasses, and reduces species diversity, particularly of oligotrophic and diagnostic species for habitat 6520, which is included in Annex I of Council Directive 92/43/EEC (Habitats Directive) [26,27,28,29]. Conversely, the abandonment of traditional management promotes rapid secondary succession, encroachment by woody species, and a reduction in forage value, leading to the loss of open habitats and profound transformations of cultural landscapes [30,31,32,33]. In the Apuseni Mountains, depopulation and declining livestock numbers have reduced both mown and grazed areas, which is reflected in shifts in species composition, progressive landscape closure, and the emergence of either simplified swards or shrub-invaded structures [34,35]. Within this context, fertilization remains one of the main drivers controlling the dynamics of plant communities in mountain grasslands, with the type and intensity of nutrient inputs determining both productivity levels and the direction of floristic succession [36,37,38]. International and national studies have shown that mineral fertilization, when applied in high doses or repeatedly, causes marked declines in species richness, favors nitrophilous species, and leads to the simplification of phytocoenoses [39,40,41]. Conversely, moderate organic inputs, particularly farmyard manure, can sustain productivity while maintaining high diversity, in line with the intermediate disturbance hypothesis, by reducing excessive competition and preserving a structural mosaic within the sward [42,43]. Organic fertilization also has beneficial effects on the soil microbiome and soil system stability, contributing to reduced nutrient loss and a lower environmental footprint [44,45]. Long-term fertilization has been shown to restructure soil functional microbiomes, thereby influencing nutrient cycling and vegetation responses [46]. In the Apuseni Mountains, several long-term experiments have separately investigated the effects of organic and mineral fertilization on Festuca rubra and Agrostis capillaris grasslands, documenting the transition from oligotrophic to mesotrophic and eutrophic communities with increasing inputs [47,48].

These studies showed that moderate doses of manure support the persistence of characteristic HNV species and maintain a heterogeneous vegetation structure, whereas mineral fertilization tends to homogenize the sward, reduce the proportion of legumes, and accelerate the loss of species sensitive to eutrophication. Recent work on indicator species and their relationship with soil agrochemical properties in the Apuseni Natural Park further confirmed that vegetation structure faithfully reflects the trophic gradient induced by fertilization and can be used to monitor the conservation status of habitat 6520 [49,50]. However, these valuable studies have largely examined organic and mineral fertilization in isolation or within organo-mineral systems, thereby not fully disentangling the distinct ecological contributions of each nutrient source. This lack of direct, integrated comparison within a single long-term experimental framework limits our understanding of the nuanced and potentially divergent successional trajectories driven by these distinct nutrient sources [51,52]. Such a comparative analysis is crucial, as organic and mineral fertilizers differ fundamentally in their nutrient release kinetics and impact on soil–plant interactions, potentially leading to ecologically distinct outcomes even at comparable nutrient loads. Fertilizers contribute to the release and increased availability of essential nutrients in the soil, thereby influencing plant growth and productivity [53,54]. In particular, there is a lack of integrated analyses combining phytosociological classification, ordination based on Bray–Curtis distances, Multi-Response Permutation Procedure (MRPP), and Indicator Species Analysis to describe subtle differences between fertilization regimes that may appear similar at the level of vegetation type, but are floristically and ecologically distinct. Filling this gap is crucial for developing precise, ecologically sound management strategies for HNV grasslands, as the distinct mechanisms of nutrient release and plant community response under strictly organic versus strictly mineral inputs are not fully understood when directly compared over extended periods. This direct, long-term comparative approach within a unified experimental design represents a significant advancement over previous fragmented studies.

In this context, the aim of the present study was to compare the long-term effects of organic and mineral fertilization on the floristic composition, vegetation types, and diversity of HNV mountain grasslands in the Apuseni Mountains. Specifically, our objectives were to:

• Identify distinct vegetation groupings and successional pathways induced by varying organic and mineral fertilization regimes over 15 years.
• Quantify the spatial differentiation of plant communities along a trophic gradient using ordination techniques.
• Determine characteristic indicator species for each fertilization treatment to refine ecological diagnostics.
• Assess the impact of different fertilization types and intensities on key plant diversity metrics (species richness, Shannon, Simpson, and evenness).

We hypothesize that (1) organic and mineral fertilization will induce distinct floristic successional pathways, with mineral inputs leading to a more rapid loss of diversity and homogenization compared to organic inputs at similar nutrient levels, and (2) moderate organic fertilization will maintain higher species richness and structural heterogeneity characteristic of HNV grasslands than any level of mineral fertilization.

2. Materials and Methods

This study builds upon previous research conducted in the Apuseni Mountains and aims to quantify the long-term ecological response of HNV mountain grasslands to organic and mineral fertilization. Unlike earlier work, which focused mainly on mineral inputs, the present analysis evaluates the cumulative trajectory of floristic change under two parallel long-term experiments—one organic and one mineral—established in 2001 and assessed after 15 consecutive years of annual fertilization.

2.1. Study Area

The research was conducted near Ghețari (Gârda de Sus, Apuseni Mountains, Romania), at 46°29′26.4″ N, 22°48′53.7″ E, on semi-natural mountain grasslands situated at ~1130 m altitude with a south-eastern exposure. The slope is approximately 5%, and the climate is cool and humid, with a multiannual mean temperature of 5.1 °C and an average annual precipitation of 1042 mm. In 2015, the study year, conditions were warmer (7.7 °C) and drier (706.4 mm) than the long-term average (

Table 1. The monthly average temperatures recorded at the Gheţari weather station (2015).

Year	Months												Average
	I	II	III	IV	V	VI	VII	VIII	IX	X	XI	XII
2015	−0.8	2	4.7	3.8	10.8	13.5	16.5	16.5	12.7	7.7	3.4	−0.2	7.7
Average values for the period 2001–2017
2001–2017	−4.5	−2.7	0.2	5.3	10.5	14.5	15.9	15.4	11.3	6.0	1.4	−3.1	5.8

and

Table 2. The monthly average precipitation recorded at the Gheţari weather station (2015).

Year	Months												Total
	I	II	III	IV	V	VI	VII	VIII	IX	X	XI	XII
2015	32.6	14.2	23.6	48.6	69.4	78.2	33.8	95	124	38.8	98.4	49.8	706.4
Average values for the period 2001–2017
2001–2017	67.2	55.9	81.8	77.2	102.4	100.3	137.3	98.4	92.6	86.6	86.5	55.8	1042.1

).

The grassland lies within the Natura 2000 network, in the ROSCI0002 Apuseni site, belonging to habitat type 6520–mountain hay meadows. This habitat is recognized for its exceptionally high floristic richness and cultural landscape continuity [55] and includes rare species such as Lilium jankae, Allium victoriale, Centaurea kotschyana, Trollius europaeus, and Anemone narcissifolia. In this complex and nutrient-sensitive floristic setting, the results of the present study are particularly relevant for defining adaptive management practices that reconcile productivity with the long-term conservation of HNV grasslands in the Apuseni Mountains.

Soil Characteristics

The soil in the experimental area was originally classified as brown eu-mesobasic rendzinic soil, corresponding approximately to modern Cambic Rendzic Leptosol. The soil physical and chemical properties were determined using standard Romanian and international methods for particle size distribution, bulk density, pH, humus, total N, and available P and K. The soil is shallow, moderately skeletal, slightly to weakly acidic, and has low to moderate fertility—conditions typical of oligotrophic F. rubra grasslands. Key diagnostic properties (from the on-site soil profile) include high humus content in the organic A_t horizon (18.2%, including non-humified material), slightly acidic reaction (pH 5.0–6.3, increasing with depth), moderate to high clay content in deeper horizons (up to 68.6%), moderate bulk density (0.98–1.28 g cm⁻³), low available P, and moderate K in surface horizons. These edaphic conditions favor stress-tolerant oligotrophic species and high floristic diversity, making the site particularly sensitive to nutrient enrichment.

2.2. Experimental Design

Two parallel long-term experiments were established in 2001, 15 m apart from each other, using a randomized block design with four replications per experiment: Experiment 1: Organic fertilization (O-series) and Experiment 2: Mineral fertilization (M-series). Each plot measured 10 m² (2 × 5 m), resulting in 32 permanent plots (eight treatments × four replicates). Both experiments shared identical environmental conditions and management but differed exclusively in the fertilization regime.

Fertilization Treatments

Organic fertilization experiment (O-series):

• T1_O–control, unfertilized;
• T2_O–10 t ha⁻¹ manure;
• T3_O–20 t ha⁻¹ manure;
• T4_O–30 t ha⁻¹ manure.

Mineral fertilization experiment (M-series):

• T1_M–control, unfertilized;
• T2_M–N50P25K25;
• T3_M–N100P50K50;
• T4_M–N150P75K75.

The fertilization levels correspond to typical low-to-high-input regimes used in grassland ecology and remain below or around the thresholds established by the Nitrates Directive for organic N inputs. Manure was sourced from local cattle and horses in the region. Its average composition (dry matter basis) was 0.40% N, 0.39% P, and 0.45% K. Fertilizers were applied annually in early spring, before vegetation onset. The mineral fertilization generally came from Azomures. The plots were mown once a year in early July, and all the biomass was removed. The entire experimental area was permanently fenced to exclude the grazing animals. Before installation (in 2001), the grassland was managed traditionally: one annual mowing followed by extensive cattle grazing in autumn (0.2–0.4 LU/ha), with no synthetic fertilization; nutrient inputs came solely from grazing animals [56,57].

2.3. Floristic Surveys

The floristic data used in this study were collected solely in June 2015, at the peak biomass stage when grasses were in full bloom, ensuring maximal discrimination among species after 15 consecutive years of differentiated fertilization (2001–2015). For each 10 m² plot, all vascular plant species were identified, and their cover-abundance was recorded using the Braun–Blanquet method (Figure 1), refined according to the detailed scale for mountain grasslands developed by [58]. Braun–Blanquet cover–abundance classes and subclasses were converted to central percentage values (

Table 3. Modified Braun-Blanquét scale for assessing the abundance–dominance of plant species based on classes and subclasses (after [58]).

Class	Coverage Interval (%)	Class Central Value (%)	Sub-Note	Sub-Interval (%)	Central-Adjusted Value of Sub-Interval (%)
5	75–100	87.5	5c	92–100	96
			5b	83–92	87.5
			5a	75–83	79
4	50–75	62.5	4c	67–75	71
			4b	58–67	62.5
			4a	50–58	54
3	25–50	37.5	3c	42–50	46
			3b	33–42	37.5
			3a	25–33	29
2	10–25	17.5	2c	20–25	22.25
			2b	15–20	17.5
			2a	10–15	12.5
1	1–10	5	1c	6–10	8
			1b	4–6	5
			1a	1–4	2.5
+	0.1–1	0.5	-	-	0.5
r	0.01–0.1	0.05	-	-	0.05

Note: a, b, and c indicate the subclasses of each class.

) prior to analysis to allow quantitative comparison among treatments. This adaptation of the Braun–Blanquet method uses refined subclasses of cover–abundance, thereby increasing the accuracy of cover estimates in highly diverse communities. A similar approach was applied in recent studies from the Eastern Alps, where the Braun–Blanquet scale was subdivided into finer classes to improve the precision of cover estimation in mountain grasslands [59]. Assessments were conducted at peak biomass, when grasses were in full bloom, to ensure maximal discrimination among species. Botanical nomenclature followed Plants of the World Online and the Euro + Med PlantBase.

2.4. Multivariate Analysis of Community Composition

Multivariate analyses were performed using PC-ORD v.7 [60], a statistical package widely used in community ecology for classification, ordination, and testing the differences among groups. This software has proven to be robust and versatile for analyzing complex floristic structures [61].

2.4.1. Cluster Analysis

To classify vegetation types, a hierarchical cluster analysis was applied using the Bray–Curtis (Sørensen) distance and UPGMA agglomeration. This approach is widely used in phytosociology and grassland community ecology [62]. The cluster cut-off level was chosen to retain approximately 75% of the information, yielding ecologically meaningful grassland groups.

2.4.2. Principal Coordinates Analysis (PCoA)

PCoA was performed using Bray–Curtis distances to visualize gradients in floristic composition. This method provides a unique and stable ordination solution [63]. In the resulting ordination:

• Axis 1 represented the fertilization-induced trophic gradient (oligotrophic → eutrophic);
• Axis 2 captured the differences between organic and mineral nutrient sources.

Treatment vectors (O_10t, O_20t, O_30t, M_50N, and M_100N) were overlaid using Pearson correlations, and their significance was tested via Monte Carlo permutations (999 runs), as recommended by [64].

2.4.3. Multi-Response Permutation Procedure (MRPP)

MRPP was used to statistically test the differences among treatments. It calculates:

• T (within-group agreement; more negative values indicate stronger separation among groups);
• A (effect size; positive values indicate groups more homogeneous than expected by chance);
• p-value (permutation-based significance).

MRPP is a standard in vegetation ecology and is widely applied in HNV grassland studies [65].

2.4.4. Indicator Species Analysis (ISA)

Indicator Species Analysis [66] was applied to identify diagnostic species for each fertilization treatment. Indicator Values (IV) were computed from species fidelity and constancy within groups, and their significance was tested using 999 permutations. ISA is frequently used to detect species responses to nutrient gradients and habitat conditions [58,67].

2.5. Diversity Indices

For each plot, α-diversity metrics were computed as follows:

• Species richness (S);
• Shannon–Wiener index (H′);
• Evenness (E = H′/ln S);
• Simpson index (D = 1 − Σp_i²).

These indices are standard in grassland ecology [68,69,70,71]. Differences among treatments were tested using one-way ANOVA followed by LSD post hoc tests (p < 0.05). Diversity indices and ANOVA/LSD tests were computed using SPSS, with a significance threshold of p < 0.05 [72,73].

3. Results

3.1. Cluster Analysis of Vegetation Under the Influence of Organic and Mineral Fertilization

The application of organic and mineral fertilization produced clear shifts in the floristic composition of mountain grasslands, as reflected by the separation of the experimental plots into six distinct vegetation groups corresponding to a well-defined trophic gradient (Figure 2). The hierarchical classification (Sørensen–Bray–Curtis index, UPGMA method) accurately reflected both the intensity and type of fertilization inputs, with the number of clusters being selected based on ecological interpretability and further supported by congruent patterns revealed by ISA and PCoA analyses. Multivariate analyses were performed using PC-ORD v.7.

Cluster 1 (Festuca rubra type) comprises all control plots from both experiments, organic and mineral (T1R1–T1R4_O and T1R1–T1R4_M). These communities are dominated by F. rubra and represent typical oligotrophic mountain grasslands, characterized by high floristic diversity, structural stability, and a large proportion of perennial stress-tolerant species adapted to nutrient-poor conditions. Although T2_O (10 t ha⁻¹ manure) and T2_M (N50P25K25) were grouped together in Cluster 2 owing to their overall floristic similarity, the MRPP results indicated that the two treatments still differed significantly in species composition (T = −3.257, A = 0.166, p = 0.0089). This confirms that similar vegetation types may emerge from different nutrient sources while retaining subtle, but statistically relevant, floristic distinctions. Cluster 2 (Subtype Trisetum flavescens–Festuca rubra cod. Agrostis capillaris) included both variants fertilized with 10 t ha⁻¹ manure (T2_O) and those fertilized with N50P25K25 (T2_M). This grouping indicates that moderate fertilization, whether organic or mineral, produces similar ecological effects on dominant species and community structure. In these plots, T. flavescens became dominant, F. rubra remained co-dominant, and A. capillaris was strongly promoted, becoming an additional co-dominant species. The coexistence of these three species defines a derived subtype within the broader F. rubra type, representing a transition toward mesotrophic grasslands with increased productivity but without major losses in floristic diversity.

Cluster 3 (Type Trisetum flavescens cod. Centaurea pseudophrygia) corresponds to the variant fertilized with 20 t ha⁻¹ manure (T3_O). Higher organic input strengthened T. flavescens as the dominant species and promoted the consistent presence of C. pseudophrygia as a codominant species. This pattern reflects a clear shift toward mesotrophic grasslands, characterized by increased biomass production and a more uniform, yet balanced, floristic structure. Cluster 4 (Subtype T. flavescens–Festuca pratensis cod. C. pseudophrygia) includes the variant fertilized with 30 t ha⁻¹ manure (T4_O). Intensification of organic input enhances the dominance of productive species, such as F. pratensis and T. flavescens, whereas C. pseudophrygia remains a characteristic element. This subtype represents a logical development of the T. flavescens type, indicating a progression toward higher mesotrophic conditions and a decline in the proportion of conservative species.

Cluster 5 (Subtype A.s capillaris–T. flavescens cod. C. pseudophrygia) groups the plots fertilized with N100P50K50 (T3_M). These communities exhibit a structure similar to that observed under moderate organic fertilization but show a stronger influence of nitrophilous species owing to the rapid availability of mineral nutrients. Increased competitive pressure favors A. capillaris and contributes to reduced floristic diversity. This subtype is also derived from T. flavescens. Cluster 6 (Subtype A. capillaris–F. pratensis cod. C. pseudophrygia) consists of the plots with the highest level of mineral fertilization (T4_M). These communities are dominated by A.s capillaris and F. pratensis, whereas C. pseudophrygia is a characteristic species of eutrophicated grasslands. Reduced diversity and increased floristic uniformity reflect the effects of strong eutrophication. Overall, the dendrogram highlights a coherent trophic gradient, ranging from oligotrophic F. rubra grasslands in the control variants to mesotrophic and eutrophic grasslands dominated by A. capillaris and F. pratensis under intensive mineral fertilization conditions. These results were fully supported by PCoA, which confirmed the same floristic groupings along the fertilization gradient.

3.2. Spatial Distance in Plant Community Projection Due to Long-Term Fertilization Organic and Mineral

The coherent trophic gradient identified by cluster analysis was further corroborated and spatially visualized through Principal Coordinate Analysis (PCoA) based on the Bray–Curtis distance, which revealed a clear separation of grassland communities according to fertilization intensity (Figure 3). The first two axes explained 93% of the total variation in floristic composition (

Table 4. Importance of axis.

Axis	Degree of Participation (r)	Cumulative
1	0.876	0.876
2	0.057	0.933

Note: r is the correlation coefficient between the ordination and original distances in the n-dimensional space.

), confirming that the trophic gradient induced by long-term fertilization is the primary factor structuring the mountain grassland vegetation. Axis 1, which accounted for the overwhelming proportion of variance (r = 0.876), accurately described the floristic transition from oligotrophic grasslands toward highly eutrophicated communities. On the negative side of Axis 1, the control variants (T1_O, T1_M) and treatments with low organic fertilization (T2_O) clustered together, corresponding to the F. rubra grassland type. These communities are characterized by low-productive mountain grasslands dominated by species adapted to nutrient-poor conditions. Representative species include F. rubra, Carex pallescens, Luzula multiflora, Leontodon autumnalis, Polygala vulgaris, Briza media, Plantago media, Plantago lanceolata, Scabiosa columbaria, and Thymus pulegioides. Their compact aggregation on the left side of the ordination reflects floristic stability and relative homogeneity of weakly fertilized grasslands.

As manure application increased (20–30 t ha⁻¹; T3_O, T4_O) or moderate mineral fertilization was applied (N50P25K25; T2_M), the plant communities shifted toward the central region of ordination. Here, the T. flavescens–F. rubra and T. flavescens–F. pratensis–C.a pseudophrygia subtypes were distinguished, indicating intermediate stages between oligotrophic and eutrophic grasslands, respectively. Floristic composition gradually became more complex and competitive, with higher participation of species such as T. flavescens, Pimpinella major, Crepis biennis, Colchicum autumnale, and Hypericum maculatum. This shift reflects increased nutrient availability and a gradual change in the competitive balance in favor of mesotrophic species. The treatments with intensive mineral fertilization (N100P50K50 and N150P75K75; T3_M and T4_M) were positioned on the positive side of Axis 1, forming well-defined groups corresponding to A. capillaris–T. flavescens and A. capillaris–Festuca pratensis–C. pseudophrygia communities. These grasslands are characterized by the dominance of nitrophilous and highly competitive species, such as A. capillaris, F. pratensis, T. flavescens, C. pseudophrygia, Veronica chamaedrys, and Rumex acetosa. Their distinct separation reflects the simplification and homogenization of the vegetation structure typical of intensively fertilized ecosystems, where oligotrophic species gradually disappear. Axis 2 (r = 0.057) captured secondary variation primarily associated with differences between purely organic and mineral fertilization. Manure-treated variants (T2_O, T3_O, and T4_O) tended to cluster in the upper part of the ordination, whereas mineral-fertilized variants (T2_M, T3_M, and T4_M) were located lower along the axis. This pattern reflects the subtle influence of fertilizer type on floristic composition, with slightly more heterogeneous communities under organic fertilization than under mineral fertilization.

Overall, the ordination revealed a coherent floristic gradient from oligotrophic F. rubra-dominated grasslands to eutrophic A. capillaris-dominated swards with increasing fertilization intensity. This spatial differentiation confirms the decisive role of long-term fertilization in shaping mountain grassland community structure.

The correlation coefficients illustrated how each fertilization treatment contributed to the separation of grassland communities along the PCoA axes, reflecting the trophic gradient generated by different levels of organic and mineral inputs (

Table 5. Correlation of treatment variants with the ordination axis.

Experimental Factors	Axis 1 (r)	Significance	Axis 2 (r)	Significance
O_10t (10 t ha−1 manure)	−0.173	ns	0.333	*
O_20t (20 t ha−1 manure)	0.133	ns	0.417	*
O_30t (30 t ha−1 manure)	0.331	*	0.240	ns
M_50N (N50P25K25)	−0.129	ns	0.375	*
M_100N (N100P50K50)	0.340	*	−0.090	ns

Note: r is the correlation coefficient between the ordination and original distances in the n-dimensional space. Significance. p < 0.05—*; ns—not significant. Treatment codes: O_10t–10 t ha⁻¹ manure; O_20t–20 t ha⁻¹ manure; O_30t–30 t ha ¹ manure; M_50N–N50P25K25 (moderate mineral input); M_100N–N100P50K50 (high mineral input).

). The application of 10 t ha⁻¹ of manure (O_10t) showed a weak, negative, and non-significant correlation with Axis 1, indicating that low organic inputs did not strongly influence the primary gradient of the ordination. However, the positive and significant correlation with Axis 2 suggests that this low-input treatment produces detectable floristic changes along the secondary ecological gradient, associated with the emergence of slightly mesotrophic species in the second year. For 20 t ha⁻¹ of manure (O_20t), the correlation with Axis 1 remained non-significant, whereas the positive and significant correlation with Axis 2 became stronger than that in the previous variant. This pattern indicates that medium organic fertilization more clearly influences the secondary community structure, reinforcing the transition toward mesotrophic grassland assemblages. Under the highest level of organic fertilization, 30 t ha⁻¹ of manure (O_30t), a marked shift occurred, and the correlation with Axis 1 became positive and significant. This demonstrates a clear association between high organic inputs and the primary trophic gradient, where communities with higher productivity and a more uniform floristic composition are separated. The correlation with Axis 2 remained non-significant, indicating that the main ecological response was captured predominantly by Axis 1. For moderate mineral fertilization, N50P25K25 (M_50N), the pattern resembled that of moderate organic inputs: a non-significant negative correlation with Axis 1 and a positive, significant correlation with Axis 2. This indicates that moderate mineral inputs influence secondary floristic differences without strongly affecting the main trophic gradients.

In the case of high mineral fertilization, N100P50K50 (M_100N), the correlation with Axis 1 was positive, significant, and the highest among all treatments. This confirms the strong effect of intensive mineral fertilization in shifting plant communities toward the eutrophic end of the primary gradient, which is consistent with the dominance of competitive, nutrient-responsive species. The correlation with Axis 2 was not significant, suggesting that most of the ecological responses to high mineral fertilization were embedded in Axis 1. Overall, the table demonstrates a progressive increase in positive correlations with Axis 1 as fertilization intensity increases, particularly with mineral inputs. This pattern reflects the transition from oligotrophic F. rubra grasslands to mesotrophic and eutrophic communities dominated by A. capillaris and F. pratensis. In contrast, Axis 2 captures more subtle floristic responses associated with low and moderate inputs, highlighting the secondary patterns of variation within the grassland system.

The MRPP results (

Table 6. Comparison of changes in floristic composition under organic and mineral fertilization (MRPP).

Treatments	T	A	p Value
T1 vs. T2_O	−7.2111	0.5163	p < 0.001 **
T1 vs. T3_O	−7.2866	0.6354	p < 0.001 **
T1 vs. T4_O	−7.3152	0.7054	p < 0.001 **
T1 vs. T2_M	−7.2353	0.5522	p < 0.001 **
T1 vs. T3_M	−7.3230	0.7256	p < 0.001 **
T1 vs. T4_M	−7.3271	0.7349	p < 0.001 **
T2_O vs. T3_O	−4.3638	0.4992	p = 0.0058 **
T2_O vs. T4_O	−4.4576	0.6868	p = 0.0056 **
T2_O vs. T2_M	−3.2573	0.1656	p = 0.0089 *
T2_O vs. T3_M	−4.4516	0.7009	p = 0.0056 **
T2_O vs. T4_M	−4.4622	0.7407	p = 0.0056 **
T3_O vs. T4_O	−4.3602	0.5307	p = 0.0056 **
T3_O vs. T2_M	−4.4095	0.5551	p = 0.0057 **
T3_O vs. T3_M	−4.4337	0.6459	p = 0.0056 **
T3_O vs. T4_M	−4.4584	0.7077	p = 0.0056 **
T4_O vs. T2_M	−4.4445	0.6756	p = 0.0056 **
T4_O vs. T3_M	−4.4370	0.7359	p = 0.0056 **
T4_O vs. T4_M	−4.4274	0.6923	p = 0.0056 **
T2_M vs. T3_M	−4.4402	0.6928	p = 0.0056 **
T2_M vs. T4_M	−4.4649	0.7529	p = 0.0056 **
T3_M vs. T4_M	−4.4398	0.6578	p = 0.0056 **

Note. T—test statistic; A—agreement statistic. Significance. p < 0.01—**; p < 0.05—*. The results of the Multi-Response Permutation Procedure (MRPP) test highlighted significant differences among all analyzed treatments (p < 0.01). The negative values of the T statistic and positive A coefficients indicate a clear separation between groups and high internal homogeneity. The clearest differentiation occurred between the control (T1) and heavily fertilized variants, particularly the mineral treatments (T3_M, T4_M), demonstrating the pronounced effect of nutrient enrichment on floristic composition. Treatment codes: O_10t–10 t ha⁻¹ manure; O_20t–20 t ha⁻¹ manure; O_30t–30 t ha ¹ manure; M_50N–N50P25K25 (moderate mineral input); M_100N–N100P50K50 (high mineral input).

) clearly show that both organic and mineral fertilization resulted in significant shifts in the floristic composition of mountain grasslands. All pairwise comparisons between treatments were statistically significant (p < 0.01), indicating that each fertilization level generated a distinct plant community. The consistently negative T values combined with positive A coefficients reflect a strong separation among treatments and high within-group homogeneity. The most pronounced differences were observed between the control (T1) and all fertilized variants, both organic (T2_O, T3_O, T4_O) and mineral (T2_M, T3_M, T4_M). The highly negative T statistics (between −7.21 and −7.33) and high A values (0.55–0.73) reveal a major ecological divergence between the oligotrophic F. rubra grasslands and fertilized plots characterized by increased productivity and altered floristic structure. This pattern is fully consistent with the PCoA ordination, where the control plots occupy the negative side of Axis 1, clearly separated from all fertilized treatments. Differences among organic fertilization levels (T2_O vs. T3_O; T2_O vs. T4_O; T3_O vs. T4_O) were also significant, indicating that increasing manure inputs drive gradual but well-defined floristic transitions. In the PCoA diagram, organic treatments followed a coherent spatial sequence, from positions closer to the control (T2_O) to progressively more positive values on Axis 1 (T3_O and T4_O), matching the increasing trophic status.

Mineral treatments (T2_M, T3_M, T4_M) were also distinct from one another, with all pairwise comparisons showing significant differences. High A values (0.65–0.75) indicate strong internal consistency and clear separation among the mineral input levels. In the ordination space, mineral treatments clustered on the positive side of Axis 1 and were arranged in an order reflecting the increasing intensity of mineral fertilization, from T2_M to T4_M. Comparisons between organic and mineral treatments (e.g., T2_O vs. T2_M; T3_O vs. T3_M; T4_O vs. T4_M) were also statistically significant, demonstrating that the source of nutrients (organic vs. mineral) affects floristic composition differently, even when nutrient levels are comparable. In the PCoA, organic treatments tended to occupy intermediate positions, whereas mineral treatments extended further along the eutrophic end of Axis 1. Even in cases where cluster analysis grouped organic and mineral variants together, such as T2_O and T2_M in Cluster 2, the MRPP test revealed significant floristic differences between them (T = −3.257, p = 0.0089). This indicates that, despite forming a similar vegetation subtype, the two treatments maintain distinct community signatures shaped by the source and release dynamics of nutrients.

Overall, the MRPP analysis confirmed the existence of a well-defined trophic gradient, ranging from oligotrophic control grasslands to increasingly productive mesotrophic and eutrophic communities shaped by rising levels of organic and mineral fertilization. This gradient aligned perfectly with the structure revealed by PCoA, demonstrating the consistency and robustness of the results.

3.3. The Response of Plant Species to Organic and Mineral Inputs

To identify which species drive the ordination patterns, we analyzed species–axis correlations (

Table 7. Correlation of species with ordination axis.

Species	Axis 1				Axis 2
	r	r-sq	tau	Signif.	r	r-sq	tau	Signif.
Agrostis capillaris L.	0.741	0.549	0.575	***	−0.381	0.145	−0.022	*
Anthoxanthum odoratum L.	−0.872	0.760	−0.719	***	−0.016	0.000	−0.034	ns
Briza media L.	−0.799	0.639	−0.622	***	−0.496	0.246	−0.415	**
Cynosurus cristatus L.	−0.675	0.456	−0.570	***	0.547	0.299	0.415	**
Dactylis glomerata L.	0.203	0.041	0.109	ns	0.207	0.043	0.224	ns
Festuca pratensis Huds.	0.894	0.799	0.800	***	−0.002	0.000	0.101	ns
Festuca rubra L.	−0.921	0.849	−0.784	***	0.149	0.022	0.132	ns
Poa trivialis L.	0.897	0.805	0.858	***	−0.316	0.100	−0.030	ns
Trisetum flavescens (L.) P. Beauv.	0.515	0.265	0.228	**	0.759	0.576	0.602	***
Carex pallescens L.	−0.744	0.554	−0.587	***	−0.212	0.045	−0.200	ns
Luzula multiflora (Ehrh.) Lej.	−0.833	0.694	−0.667	***	−0.216	0.047	−0.203	ns
Anthyllis vulneraria L.	−0.799	0.639	−0.622	***	−0.496	0.246	−0.415	**
Lotus corniculatus L.	−0.850	0.722	−0.760	***	−0.026	0.001	0.072	ns
Trifolium pratense L.	−0.312	0.097	−0.317	ns	0.381	0.145	0.295	*
Trifolium repens L.	−0.697	0.485	−0.563	***	0.402	0.162	0.284	*
Vicia cracca L.	−0.175	0.031	−0.182	ns	0.020	0.000	−0.159	ns
Achillea millefolium L.	0.219	0.048	0.194	ns	0.348	0.121	0.263	*
Alchemilla vulgaris L.	0.678	0.460	0.558	***	−0.174	0.030	−0.138	ns
Campanula patula L.	−0.439	0.193	−0.343	*	−0.284	0.080	−0.227	ns
Carlina acaulis L.	−0.675	0.456	−0.544	***	−0.297	0.088	−0.285	ns
Carum carvi L.	0.022	0.000	0.038	ns	0.308	0.095	0.226	ns
Centaurea pseudophrygia C.A. Mey.	−0.185	0.034	−0.196	ns	0.597	0.356	0.480	**
Cerastium glomeratum Thuill.	−0.492	0.242	−0.387	**	−0.315	0.099	−0.251	ns
Colchicum autumnale L.	−0.242	0.058	−0.280	ns	0.631	0.398	0.515	***
Crepis biennis L.	−0.318	0.101	−0.307	ns	0.506	0.256	0.423	**
Euphrasia officinalis L.	−0.799	0.639	−0.622	***	−0.496	0.246	−0.415	**
Hypericum maculatum Crantz	0.300	0.090	0.159	ns	0.343	0.117	0.398	*
Leontodon autumnalis L.	−0.834	0.696	−0.799	***	−0.395	0.156	−0.139	*
Leucanthemum vulgare Lam.	−0.484	0.235	−0.626	**	0.007	0.000	−0.173	ns
Pimpinella major (L.) Huds.	−0.138	0.019	−0.008	ns	0.734	0.539	0.538	***
Plantago lanceolata L.	−0.650	0.422	−0.690	***	−0.305	0.093	−0.245	ns
Plantago media L.	−0.657	0.431	−0.496	***	−0.370	0.137	−0.302	*
Ranunculus bulbosus L.	0.497	0.247	0.452	**	−0.253	0.064	−0.220	ns
Rhinanthus minor L.	−0.037	0.001	0.003	ns	−0.360	0.130	−0.316	*
Rumex acetosella L.	0.082	0.007	0.019	ns	0.499	0.249	0.475	**
Stellaria graminea L.	0.364	0.133	0.266	*	0.460	0.212	0.351	**
Taraxacum officinale Weber	0.063	0.004	0.027	ns	−0.596	0.355	−0.466	**
Thymus pulegioides L.	−0.628	0.394	−0.592	***	−0.407	0.165	−0.411	*
Veronica chamaedrys L.	0.386	0.149	0.326	*	0.668	0.447	0.483	**

Note: r is the correlation coefficient and r-sq is the determination Significance. p ˂ 0.001—***; p ˂ 0.01—**; p ˂ 0.5—*; ns—not significant. Negative r-values were associated with conservative, oligotrophic species (F. rubra, C. pallescens, and L. multiflora), whereas positive values corresponded to species favored by increased nutrient availability (A. capillaris, P. trivialis, and F. pratensis). Consequently, Axis 1 reflected the intensity of fertilization, whereas Axis 2 captured secondary variation related to differences between organic and mineral inputs.

), which revealed a well-defined trophic gradient represented by Axis 1. Positive r values were associated with mesotrophic and nitrophilic species favored by increased nutrient availability. These included A. capillaris (r = 0.741, p < 0.001), F. pratensis (r = 0.894, p < 0.001), P. trivialis (r = 0.897, p < 0.001), and T. flavescens (r = 0.515, p < 0.01). Such species typically dominate grassland communities subjected to mineral fertilization or higher organic inputs and reflect a shift toward more productive, nutrient-enriched conditions. Conversely, strongly negative correlations on Axis 1 characterize oligotrophic and stress-tolerant species typical of unfertilized or low-input organic treatments. Among these, F. rubra (r = −0.921, p < 0.001), C. pallescens (r = −0.744, p < 0.001), L. multiflora (r = −0.833, p < 0.001), and Lotus corniculatus (r = −0.850, p < 0.001) are indicative of HNV grasslands, where low soil fertility supports a diverse and stable floristic structure. Axis 2 captures secondary ecological variation related to the type of fertilization. Significant positive correlations were recorded for species such as T. flavescens (r = 0.759, p < 0.001), V. chamaedrys (r = 0.668, p < 0.01), P. major (r = 0.734, p < 0.001), and C. autumnale (r = 0.631, p < 0.001), suggesting a preference for fertilization regimes associated with increased nutrient availability. In contrast, moderately negative correlations (e.g., Trifolium pratense, Trifolium repens, C. pseudophrygia, Rhinanthus minor) indicate species that are more closely associated with lower trophic conditions or organic inputs.

Overall, Axis 1 reflects the gradient of fertilization intensity, showing a floristic transition from oligotrophic F. rubra grasslands to more productive, nutrient-responsive communities. Axis 2 differentiates secondary patterns related to the type and level of nutrient input, which is in agreement with the patterns observed in the PCoA diagram.

The distribution of species along the trophic gradient closely aligned with the trends identified through PCoA, showing a consistent association of mesotrophic species with moderate and high fertilization levels and a clear distinction of oligotrophic species in the control plots. These findings provide a coherent ecological framework for Indicator Species Analysis (ISA), where the diagnostic value of individual species is assessed in relation to the identified vegetation groups.

3.4. Indicator Species Analysis to the Gradient of Applied Inputs Organo and Mineral

Based on the trophic gradient identified by the ordination analyses, Indicator Species Analysis (ISA) was applied using the seven fertilization treatments as grouping factors (T1–control; T2_O–10 t ha⁻¹ manure; T3_O–20 t ha⁻¹ manure; T4_O–30 t ha⁻¹ manure; T2_M–N50P25K25; T3_M–N100P50K50; T4_M–N150P75K75). The results (

Table 8. Indicator Species Analysis (ISA) highlighted species with a significant affinity for each fertilization regime.

Species	Max Group	Indicator Value (IV)	Mean	S.Dev	p-Value
Agrostis capillaris L.	7	29.7	20.7	2.39	** (p = 0.0012)
Anthoxanthum odoratum L.	1	30.4	22.5	2.87	*** (p = 0.0010)
Briza media L.	1	100.0	18.4	8.29	*** (p = 0.0002)
Cynosurus cristatus L.	1	20.0	18.8	1.26	ns (p = 0.0824)
Dactylis glomerata L.	4	45.5	18.5	10.61	ns (p = 0.0698)
Festuca pratensis Huds.	4	28.7	21.6	3.98	* (p = 0.0422)
Festuca rubra L.	1	27.2	21.1	2.28	** (p = 0.0016)
Poa trivialis L.	7	37.0	23.1	4.79	** (p = 0.0020)
Trisetum flavescens (L.) p. Beauv.	3	19.8	19.5	1.62	ns (p = 0.4313)
Carex pallescens L.	1	47.1	18.5	8.15	*** (p = 0.0008)
Luzula multiflora (Ehrh.) Lej.	1	50.0	19.1	7.48	*** (p = 0.0002)
Anthyllis vulneraria L.	1	100.0	18.4	8.29	*** (p = 0.0002)
Lotus corniculatus L.	1	36.0	21.5	5.12	*** (p = 0.0006)
Trifolium pratense L.	3	33.7	23.3	3.48	** (p = 0.0088)
Trifolium repens L.	2	24.9	20.1	2.02	* (p = 0.0128)
Vicia cracca L.	5	36.4	23.4	6.32	** (p = 0.0030)
Achillea millefolium L.	6	21.6	18.0	1.55	* (p = 0.0358)
Alchemilla vulgaris L.	6	19.3	17.2	1.11	ns (p = 0.0530)
Campanula patula L.	1	20.0	18.8	1.21	ns (p = 0.0734)
Carlina acaulis L.	1	45.0	18.4	8.41	* (p = 0.0138)
Carum carvi L.	2	50.0	21.4	10.74	* (p = 0.0342)
Centaurea pseudophrygia C.A. Mey.	5	18.5	16.1	0.75	** (p = 0.0060)
Cerastium glomeratum Thuill.	1	23.5	19.6	3.75	* (p = 0.0214)
Colchicum autumnale L.	2	23.0	18.3	1.49	** (p = 0.0054)
Crepis biennis L.	2	36.4	19.4	7.45	ns (p = 0.0654)
Euphrasia officinalis L.	1	100.0	18.4	8.29	*** (p = 0.0002)
Hypericum maculatum Crantz	4	74.6	24.0	9.61	*** (p = 0.0002)
Leontodon autumnalis L.	1	70.4	23.3	7.32	*** (p = 0.0002)
Leucanthemum vulgare Lam.	2	66.7	22.5	9.65	*** (p = 0.0002)
Pimpinella major (L.) Huds.	4	22.3	18.4	1.35	** (p = 0.0082)
Plantago lanceolata L.	1	71.4	22.6	9.34	*** (p = 0.0004)
Plantago media L.	1	75.0	18.1	9.37	*** (p = 0.0006)
Ranunculus acris L.	1	14.3	14.3	0.20	ns (p = 1.0000)
Ranunculus bulbosus L.	4	25.0	23.3	4.66	ns (p = 0.5749)
Rhinanthus minor L.	2	18.6	18.2	0.65	ns (p = 0.4045)
Rumex acetosella L.	3	41.7	23.3	4.66	*** (p = 0.0006)
Stellaria graminea L.	2	23.8	21.1	2.56	ns (p = 0.1680)
Taraxacum officinale Weber	7	24.6	21.1	2.64	ns (p = 0.0780)
Thymus pulegioides L.	1	100.0	20.3	9.54	*** (p = 0.0002)
Veronica chamaedrys L.	4	26.6	20.7	2.35	** (p = 0.0106)

Notes. Max group–treatment with the highest indicator value; IV–indicator value (%). Significance. p ˂ 0.001—***; p ˂ 0.01—**; p ˂ 0.5—*; ns—not significant. Group 1–T1 (control, T1_O + T1_M), Group 2–T2_O (10 t ha⁻¹ manure), Group 3–T3_O (20 t ha⁻¹ manure), Group 4–T4_O (30 t ha⁻¹ manure), Group 5–T2_M (N50P25K25), Group 6–T3_M (N100P50K50), Group 7–T4_M (N150P75K75).

) revealed well-defined diagnostic species for each treatment, confirming the ecological differentiation generated by the increase in nutrient inputs. Low-input treatments—T2_O (10 t ha⁻¹ manure) and T2_M (N50P25K25)—were associated with early mesotrophic indicator species, particularly Trifolium repens and Trifolium pratense (p < 0.05). In this study, these species indicate moderate nutrient enrichment because they reached significant indicator values under low-input treatments while overall species richness, evenness, and community structure remained close to those of the unfertilized control. Their occurrence therefore reflects a moderate increase in nutrient availability without disrupting the structural balance and characteristic floristic composition of HNV grasslands. Control plots (T1_O + T1_M; Group 1) showed the highest number of significant indicator species (p < 0.01), including Festuca rubra, B. media, L. multiflora, A. vulneraria, L. corniculatus, P. media, P. vulgaris, and T. pulegioides. These species typify oligotrophic HNV grasslands and reflect the stable, low-fertility conditions of the unfertilized variants.

Low-input treatments—T2_O (10 t ha⁻¹ manure; Group 2) and T2_M (N50P25K25; Group 5)—were associated with early mesotrophic indicator species, particularly T. repens and T. pratense (p < 0.05). Their occurrence indicates moderate nutrient enrichment while maintaining structural balance and preserving the HNV character. Medium-input treatments, T3_O (20 t ha⁻¹ manure; Group 3) and T3_M (N100P50K50; Group 6), displayed indicator species characteristics of transitional mesotrophic grasslands, such as Vicia cracca and C. pseudophrygia (p < 0.01). These species signal a shift toward more productive communities with increasing competitive interactions and a reduced representation of conservative oligotrophic taxa. High-input organic fertilization (T4_O–30 t ha⁻¹ manure; Group 4) was associated with productive species such as F. pratensis and P. major (p < 0.05), indicating advanced mesotrophic–eutrophic conditions and a strong response to high manure inputs. High-input mineral fertilization (T4_M–N150P75K75; Group 7) showed strong affinities with nitrophilous and eutrophic species, including A. capillaris, P. trivialis, H. maculatum, and V. chamaedrys (p < 0.01). These taxa represent intensively fertilized, low-diversity communities that are typical of strongly eutrophicated mountain grasslands.

Overall, ISA revealed a coherent ecological progression from oligotrophic Festuca rubra-dominated communities with zero input (T1) to mesotrophic assemblages under moderate organic or mineral fertilization (T2 and T3), and finally to eutrophic Agrostis capillaris-type communities under high mineral fertilization (T4_M).

Although ISA distinguished seven treatment-based groups and cluster analysis identified six vegetation types, both approaches consistently reflected the same ecological gradient. The apparent discrepancy arises because ISA groups are defined strictly by fertilization treatments, whereas the cluster analysis groups vegetation types based on floristic similarity. For example, T2_O and T2_M formed two ISA groups owing to their different fertilizer sources, yet they belonged to the same vegetation subtype (Cluster 2), highlighting their similar floristic response despite different nutrient origins. These findings match the spatial patterns identified by PCoA and the group differences detected by MRPP, demonstrating strong analytical consistency across methods and confirming the robustness of fertilization-induced trophic gradients.

3.5. The Influence of Organic and Mineral Fertilizer on Plant Diversity

The analysis of diversity parameters (

Table 9. The influence of organic and mineral fertilizers on plant diversity.

Variant	Species No. (S)	Shannon (H′)	Evenness (E)	Simpson (D)
T1 (Control–combined)	33.63 ± 1.30	2.91 ± 0.06	0.827 ± 0.012	0.9245 ± 0.0054
T2_O (10 t ha−1 manure; Low-input organic)	31.50 ± 2.08	2.78 ± 0.03	0.805 ± 0.011	0.9135 ± 0.0040
T3_O (20 t ha−1 manure; Medium-input organic)	24.00 ± 0.82	2.73 ± 0.06	0.860 ± 0.021	0.9190 ± 0.0065
T4_O (30 t ha−1 manure; High-input organic)	22.75 ± 1.26	2.66 ± 0.02	0.852 ± 0.009	0.9147 ± 0.0012
T2_M (N50P25K25; Low-input mineral)	27.00 ± 0.82	2.64 ± 0.06	0.801 ± 0.016	0.9071 ± 0.0055
T3_M (N100P50K50; Medium-input mineral)	23.00 ± 0.00	2.45 ± 0.02	0.782 ± 0.006	0.8820 ± 0.0025
T4_M (N150P75K75; High-input mineral)	21.75 ± 0.50	2.29 ± 0.07	0.750 ± 0.017	0.8544 ± 0.0115
F test	72.43	14.14	1.01	5.6
p.val	p < 0.001	p < 0.001	p = 0.401	p = 0.004

Notes: Diversity indices (S, H′, and D) declined significantly with increasing fertilization intensity, particularly under mineral inputs. Low-input organic fertilization (T2) maintained relatively high diversity, whereas high mineral fertilization (T4) caused the strongest reduction in species richness and community complexity. These findings highlight the role of low-input organic management in preserving biodiversity in HNV mountain grasslands.

) indicated consistent and statistically significant effects of fertilization intensity on grassland diversity. ANOVA revealed significant differences among the treatments in terms of species richness (F = 72.43, p < 0.001), Shannon diversity (F = 14.14, p < 0.001), and Simpson index (F = 5.60, p = 0.004), whereas evenness did not differ significantly (F = 1.01, p = 0.401). These results indicate that fertilization primarily affects the number of species and overall community complexity, whereas the distribution of abundance among species remains relatively stable. The control treatment (T1–combined organic and mineral) displayed the highest species richness (S = 33.63 ± 1.30), Shannon diversity (H′ = 2.91 ± 0.06), and evenness (E = 0.827 ± 0.012). These values reflect the oligotrophic and structurally stable characteristics of HNV grasslands, a pattern further supported by the high Simpson index (D = 0.9245 ± 0.0054).

Under low-input organic fertilization (T2 organic), diversity remained relatively high (H′ = 2.78 ± 0.03), although species richness declined moderately (S = 31.50 ± 2.08). This indicates that moderate nutrient input may stimulate certain mesotrophic species without substantially altering the overall community structure. Medium organic fertilization (T3 organic) caused a marked reduction in species number (S = 24.00 ± 0.82), but evenness remained high (E = 0.860 ± 0.021), suggesting reorganization around a smaller set of competitive species. High-input organic fertilization (T4 organic) led to a further decline in diversity (H′ = 2.66 ± 0.02), although the evenness values continued to indicate relatively balanced species distributions.

In the mineral fertilization experiment, the decline in diversity was more pronounced. Under low-input mineral fertilization (T2), species richness dropped to 27.00 ± 0.82, and the Shannon index decreased to 2.64 ± 0.06. Medium and high mineral inputs (T3 and T4) produced substantial reductions across all indices, with the highest intensity treatment (T4) showing the lowest diversity (H′ = 2.29 ± 0.07), evenness (E = 0.750 ± 0.017), and Simpson index (D = 0.8544 ± 0.0115). These values reflect communities increasingly dominated by a small number of nitrophilous species typical of strongly eutrophic grasslands.

Overall, the results demonstrate that fertilization significantly alters grassland community composition, with mineral inputs producing much stronger shifts than organic fertilization. The control treatment consistently exhibited the highest diversity, highlighting the sensitivity of HNV grasslands to nutrient enrichment and confirming the ecological gradient detected by multivariate analyses.

The results clearly demonstrate the strong influence of fertilization type and intensity on the structure and diversity of mountain grasslands, revealing a consistent ecological transition from oligotrophic to mesotrophic and ultimately to eutrophic community types. All analytical approaches—MRPP, PCoA, ISA, and diversity indices—converged in depicting the same trophic gradient induced by increasing nutrient inputs. This coherence across multiple methods provides a solid quantitative basis for the ecological interpretation developed in the Discussion.

4. Discussion

4.1. Contrasting Effects of Organic and Mineral Fertilization on Floristic Composition

The present study demonstrates that fertilization remains the dominant ecological driver shaping the successional pathways of HNV mountain grasslands, with both the source and intensity of nutrient inputs determining the scale and direction of vegetation change. As shown by the ordination and MRPP results, organic and mineral fertilizers generated divergent floristic trajectories, confirming the patterns observed in other oligotrophic European grasslands [74,75,76]. Moderate organic fertilization maintained the structural heterogeneity typical of habitat 6520, allowing the persistence of conservative and stress-tolerant species such as F. rubra, B. media, L. multiflora, and T. pulegioides [77]. This aligns with earlier research in the Apuseni Mountains and Carpathians, showing that cattle manure applied at low–medium doses supports oligotrophic assemblages without triggering rapid eutrophication [78,79,80].

In contrast, mineral fertilization induced rapid simplification of the community, favoring competitive grasses (A. capillaris and Dactylis glomerata) and nitrophilous species (R. acetosa and P. trivialis). This pattern is consistent with the long-term nitrogen enrichment effects reported across European mountain regions [28,81]. PCoA confirmed that the primary trophic gradient was largely determined by the divergence between mineral and organic treatments, a pattern similarly reported in alpine and subalpine systems [82]. Indicator Species Analysis refined this interpretation by identifying a distinct set of diagnostic species for each fertilization level: oligotrophic communities dominated by F. rubra in T1, mesotrophic assemblages with Trifolium spp. and V. cracca in T2–T3, and eutrophic, mineral-driven communities dominated by A. capillaris and C. pseudophrygia under high mineral input (T4). These transitions closely mirror previous findings from long-term Apuseni grassland studies [83,84,85] and European mountain ecosystems [86,87], confirming the predictability of vegetation responses to fertilization.

4.2. Differential Responses of Functional Groups and Underlying Ecological Mechanisms

Functional groups exhibited markedly different sensitivities to fertilization, reflecting the underlying mechanisms of competition, stress adaptation, and nutrient acquisition. Legumes increased under moderate organic fertilization, likely due to improved soil microbial activity and gradual nitrogen mineralization [88,89]. This maximum abundance in T2–T3 aligns with the intermediate disturbance hypothesis, where moderate nutrient inputs prevent competitive exclusion, thereby promoting species coexistence [90,91]. In contrast, high mineral fertilization favored tall, fast-growing species and reduced the presence of hemicryptophytes and oligotrophic specialists [92]. Similar competitive shifts toward productive grasses under sustained nitrogen addition [75,93] are consistent with findings showing that fertilization strengthens the dominance of Festuca arundinacea over T. pratense, leading to reduced legume presence at higher nutrient levels [94]. Species negatively correlated with PCoA Axis 1 (C. pallescens, L. multiflora, and A. vulneraria) persisted mainly under organic regimes, indicating that manure does not homogenize microhabitats to the same extent as mineral inputs. This highlights the role of nutrient release kinetics as determinants of successional direction.

4.3. Directional Shifts in Community Structure Under Long-Term Fertilization

The vegetation groups identified through cluster analysis displayed a coherent trophic gradient fully consistent with the ISA and PCoA, confirming the robustness of the classification framework. As observed in long-term grassland studies worldwide [94,95], Apuseni grasslands exhibit a predictable directional response to fertilization. However, in contrast to the classical predictions of the intermediate disturbance hypothesis, the highest plant diversity in this study was recorded in the unfertilized control (T1) rather than under low-input treatments (T2). This pattern reflects the long-term ecological stability and conservation value of oligotrophic High-Nature-Value grasslands, where minimal disturbance supports high species richness. Low-input fertilization induced detectable floristic shifts without exceeding ecological thresholds, indicating a management compromise between biodiversity conservation and productivity rather than a diversity maximum. A notably informative case is the pair T2_O versus T2_M: although grouped within the same cluster owing to overall similarity, the MRPP test confirmed statistically significant floristic differences. This illustrates how hierarchical clustering may obscure fine-scale ecological distinctions, whereas distance-based ordinations and permutation tests reveal underlying group divergences [96,97].

These patterns suggest two contrasting successional mechanisms:

• Organic fertilization → gradual nutrient release → maintenance of microsite heterogeneity and slower competitive shifts;
• Mineral fertilization → rapid nutrient availability → accelerated competitive exclusion and reduced diversity.

This dual dynamic explains the pronounced divergence between the two sources.

4.4. Diversity Responses Along the Nutrient Input Gradient

Diversity responses followed a nonlinear pattern, consistent with the intermediate disturbance hypothesis. Moderate fertilization (T2_O and T2_M) maintained high richness and evenness, as observed in temperate and alpine grasslands [68,98]. In contrast, high mineral inputs caused sharp declines in richness and Simpson diversity, indicative of dominance by a few nitrophilous species, a process well documented in European eutrophication studies [81,99,100]. Notably, organic fertilization at high doses (T4_O) maintained substantially higher evenness than the equivalent mineral treatment (T4_M), suggesting that the gradual mineralization of manure buffers competitive imbalances [45,53]. Thus, the ecological effects of organic and mineral fertilizers with similar nitrogen contents are not equivalent, and their impacts must be distinguished in management and policy frameworks.

4.5. Implications for Adaptive Management of HNV Grasslands

The findings provide several key guidelines for conservation and agri-environmental management.

• Mineral fertilization should be avoided or strictly limited in habitat 6520 grasslands because of its rapid eutrophication effects and community homogenization.
• Moderate organic inputs (10–20 t ha⁻¹) are compatible with HNV conservation, maintaining floristic integrity and vegetation structure, with potential implications for ecosystem functioning [51].
• Multivariate tools, such as ISA and PCoA, can serve as early warning indicators of nutrient-driven degradation in protected meadows.
• Agri-environmental schemes should clearly distinguish between organic and mineral nutrient sources, as their ecological outcomes differ fundamentally from each other.

These conclusions support adaptive management approaches aligned with the EU biodiversity strategies and the Nitrates Directive. Monitoring of floristic changes can be enhanced through modern tools such as near-infrared (NIR) hyperspectral analysis, which has demonstrated a high capacity to discriminate grassland species and classify them into botanical families with high accuracy [36]. These technologies may become essential components of monitoring schemes for HNV grasslands.

4.6. Limitations and Future Research Directions

Two main limitations should be acknowledged.

• Floristic data were collected in a single year, although fertilization was applied long-term. Multi-year vegetation surveys would improve our understanding of interannual variability and stability.
• Soil chemical and biological parameters (e.g., mineral N fractions, C:N ratios, microbial community composition, and root dynamics) were not integrated and would strengthen mechanistic interpretations.

Therefore, future studies should incorporate the following:

• Long-term monitoring under fluctuating climatic conditions.
• Functional trait analyses (e.g., SLA, plant height, CSR strategies).
• High-resolution soil microbiome profiling under organic and mineral regimes.
• Remote sensing (multispectral, hyperspectral, and LiDAR) for the early detection of vegetation structural changes.

An integrated soil–plant–microbiome perspective is essential for identifying ecological thresholds beyond which HNV grasslands may shift irreversibly toward eutrophic and low diversity states. Future research should integrate modern tools for assessing root–fungal interactions, including sensitive microscopic approaches capable of detecting structural shifts in mycorrhizal colonization under fertilization conditions [100]. These methods can provide deeper insights into soil biological thresholds and ecosystem resilience.

5. Conclusions

This long-term experiment demonstrates that organic and mineral fertilization generate fundamentally different ecological trajectories in HNV mountain grasslands. Moderate organic inputs preserved the oligotrophic character of the vegetation, maintained structural heterogeneity, and supported the persistence of characteristic species typical of habitat type 6520. In contrast, mineral fertilization produced rapid shifts toward mesotrophic and eutrophic community types, as reflected by reduced species richness, increased dominance of nitrophilous taxa, and more homogeneous vegetation structures.

Multivariate analyses (PCoA, MRPP, ISA) consistently differentiated the treatments, confirming that both nutrient source and input intensity play decisive roles in shaping floristic reorganization. Diversity responses further supported these patterns, revealing a clear gradient from species-rich, structurally stable, oligotrophic grasslands to simplified communities under intensive mineral fertilization.

Overall, this study highlights the high sensitivity of HNV mountain grasslands to nutrient enrichment and demonstrates that organic and mineral fertilizers with comparable nitrogen contents exert markedly different ecological effects. These findings provide an essential empirical basis for understanding the dynamics of vegetation in nutrient-limited mountain ecosystems.

6. Implications for Policy

The findings of this long-term experiment provide clear evidence to support more differentiated and input-sensitive management policies for High-Nature-Value (HNV) mountain grasslands. The contrasting effects of organic and mineral fertilization demonstrate that the nutrient source and intensity must be explicitly considered in agri-environmental decision-making, as they influence biodiversity, vegetation structure, and functional stability in fundamentally different ways. Although European policy frameworks increasingly acknowledge nutrient-sensitive management in semi-natural systems [101,102,103] national regulations often overlook the ecological divergence between organic and mineral nutrient input. Our results show that low-to-moderate organic fertilization is compatible with the conservation objectives of Habitat 6520, supporting the persistence of oligotrophic species and maintaining structural heterogeneity while ensuring adequate productivity. This aligns with the priorities outlined in the EU Biodiversity Strategy for 2030 and the Natura 2000 management guidelines, which promote extensive, biodiversity-friendly practices in mountain grasslands [104,105]. In contrast, high mineral fertilization rapidly increases eutrophication, homogenizes vegetation, and leads to substantial losses in species richness, effects that have been repeatedly documented in European grassland monitoring programs [106].

The set of diagnostic species identified through Indicator Species Analysis provides a practical tool for field monitoring and is highly relevant for the implementation of Results-Based Payment Schemes (RBPSs) and GAEC biodiversity conditions under the Common Agricultural Policy. Indicator species, such as Festuca rubra, B. media, and T. pulegioides, can serve as early warning markers for nutrient enrichment, offering a cost-effective mechanism for adaptive management and compliance verification [105]. The strong agreement among multivariate analyses (cluster analysis, PCoA, MRPP, ISA) highlights the importance of long-term grassland experiments for calibrating management thresholds. Such evidence is essential for developing policies that accurately reflect ecological tipping points and prevent irreversible degradation, a priority emphasized in recent assessments by the Joint Research Centre [107] and the European Environment Agency [102].

Overall, these results underscore the need for adaptive and differentiated fertilization policies in HNV grasslands.

• Low-input organic fertilization should be encouraged;
• High mineral nutrient inputs should be avoided or strictly regulated;
• Indicator species should be incorporated into routine monitoring systems;
• Biodiversity targets should be explicitly linked to fertilization regimes.

Such policy alignment would enhance the ecological resilience of HNV grasslands, sustain their biodiversity, and support the socioeconomic functions of traditional mountain agriculture.