Vineyard Floor Management as a Multifunctional Strategy for Pollinator Communities and Grape Quality

1. Introduction

The decline of pollinating insect communities is a challenge to the implementation of sustainable agriculture and conservation of global biodiversity [1,2]. Reductions in population abundance and species diversity among groups such as honeybees and bumblebees, butterflies, and syrphid flies, have been widely observed over the past several decades, especially in agricultural environments [3]. Intensive agricultural practices such as monocultures, heavy soil tillage, or the widespread and frequent application of pesticides can negatively affect plant and pollinator diversity and their interactions, thereby impairing ecosystem functioning and stability [4,5].

Pollinators provide an indispensable service in terrestrial ecosystems by enhancing crop provision and the resilience of plant communities [6,7]. Impacts on crop provision, biological regulation, and the stability of ecological systems have been associated with losses in pollinator population and diversity [8]. Hence, preserving plant diversity and pollinator populations, and increasing habitat availability are key priorities for sustainable agricultural management [9].

Increasing landscape diversity and implementing agri-environmental strategies can enhance insect-plant interactions and promote pollinator diversity in agricultural ecosystems [10,11]. During the flowering period, plant community diversity provides pollinators with shelter and nesting sites, while also providing nectar and pollen resources [12]. Vegetation community composition can enhance functional diversity by increasing floral abundance [13]. Functional biodiversity refers to the diversity of species that support important ecosystem services such as pollination, biocontrol, and nutrient cycling [14]. Therefore, maintaining and promoting diverse cover crops is becoming increasingly important in sustainable development strategies aimed at conserving biodiversity and enhancing ecosystem function [15].

Vineyards are established for the production of grapes for wine or direct consumption, but they are also permanent agroecosystems that can support appreciable biodiversity when managed appropriately [16]. Inter-row vegetation and semi-natural habitats surrounding vineyards can act as ecological infrastructures that provide both trophic resources and habitat for pollinators, predatory insects, birds and soil organisms [17]. Grapevine is a plant with predominantly hermaphroditic flowers, and most wine grape cultivars are self-pollinating [18]. However, insect activity on grape clusters may contribute to pollen transfer on the stigmas and thereby contribute to more complete berry set, especially under low-wind conditions [19]. A pollinator-rich environment attracts and maintains natural predators of pests in the vineyard, reducing the need for chemical treatments [20].

The ecological value of vineyards is influenced by soil management within vine rows and inter-rows [21]. These strategies also influence vegetation, insect diversity and community structure [22]. Conventional bare-soil management or herbicide treatments minimise competition for nutrients between vines and spontaneous vegetation, but also decrease biodiversity and increase exposure to erosion [23]. In recent decades, in many vineyards, spontaneous vegetation or sown cover crops have increasingly been practised inter-row to improve soil quality by incorporating organic matter, to increase the above- and below-ground faunal biodiversity, or to reduce erosion [24,25,26]. Cover crops may additionally provide a more predictable and manageable source of floral resources, whereas spontaneous vegetation can support naturally occurring and potentially heterogeneous plant communities [27].

According to recent studies, inter-row vegetation improves biodiversity and ecosystem services in vineyards, but much of the available research results are from France, Spain, or Italy [28,29,30]. Few studies have simultaneously investigated how different inter-row vegetation management strategies impact vegetation characteristics, pollinator populations, and grape compositional attributes throughout several growing seasons in an Eastern European vineyard [31,32]. An integrated approach is relevant because vineyard biodiversity and grape production and quality may be linked to vegetation management; the ecological benefits generated by the diversity of vegetation can be influenced by the available resources and composition of the plant community.

Therefore, the purpose of this study was to assess how contrasting soil management practices influence vegetation biodiversity, pollinator communities, grape yield, and quality parameters and to examine the relationship between vegetation traits and insect abundance, in a western Romanian vineyard.

Specifically, the primary goals were to: (i) assess the impact of three vineyard floor management options on vegetation cover, floral resources, and plant diversity; (ii) evaluate the impact on bee, bumblebee, hoverfly, and butterfly populations; (iii) examine the correlation between vegetation attributes and pollinator abundance; (iv) investigate how vineyard floor management impacts grape yield and composition in four wine grape cultivars.

2. Results

2.1. Effects of Inter-Row Management on Vineyard-Floor Vegetation

Inter-row management and year significantly influenced vegetation cover, flower density, and plant species richness; the way these variables acted changed over time (Table 1). Treatment and year had a substantial impact on vegetation cover, flower density, and plant species richness. Treatment effects varied over years, as shown by significant treatment × year interactions.

Vegetation cover expanded throughout the research duration under both cover crop (CC) and spontaneous vegetation (SV) management, while bare soil (BS) had persistently low vegetation cover. Flower density followed a similar pattern, with the highest annual mean under CC, intermediate values under SV, and significantly lower values for BS throughout the study period. Plant species richness followed a similar treatment gradient, with CC maintaining the most diverse plant community, followed by SV, and BS consistently supporting the fewest species. CC consistently had the highest values for all three vegetation traits, followed by SV and BS (Tukey’s HSD, p < 0.05).

Figure 1. Vegetation cover (A), flower density (B), and species richness (C) associated with different inter-row management treatments (CC, SV, and BS) during the 2023-2025 seasons. The values are mean ± SD (n = 3 plots per treatment). Lowercase letters within each year show significant differences among treatments (Tukey’s HSD, p < 0.05).

Figure 1. Vegetation cover (A), flower density (B), and species richness (C) associated with different inter-row management treatments (CC, SV, and BS) during the 2023-2025 seasons. The values are mean ± SD (n = 3 plots per treatment). Lowercase letters within each year show significant differences among treatments (Tukey’s HSD, p < 0.05).

2.2. Pollinator Community Composition

The NMDS ordination based on Bray-Curtis dissimilarities provided a very good representation of pollinator community structure (stress = 0.022), revealing unique patterns across the three inter-row management systems (Figure 2). Samples from the BS treatment were distributed more widely in the ordination space, whereas CC and SV samples exhibited partially overlapping clusters with overlapping 95% confidence ellipses, indicating a higher degree of similarity in pollinator community composition within the two vegetation-based treatments.

A two-factor PERMANOVA based on Bray-Curtis dissimilarities revealed significant effects of inter-row management treatment, sampling year, and their interaction on pollinator community composition (Table 2). The largest proportion of variation in community composition was explained by inter-row management (R² = 0.765), followed by sampling year (R² = 0.181). The ‘Treatment × Year’ interaction explained a smaller proportion of variation (R² = 0.025). All three effects were significant at p ≤ 0.002. PERMDISP identified no significant differences in multivariate dispersion between treatments (p = 0.868) or years (p = 0.982), indicating that the significant PERMANOVA effects of treatment and year reflected differences in community composition rather than differences in within-group dispersion.

2.3. Responses of Pollinator Groups to Inter-Row Management

The abundance of pollinators increased from 2023 to 2025 for all three inter-row management treatments, with the highest abundance consistently observed under CC and the lowest under BS. Honey bees were the most abundant pollinator group and their mean numbers were always higher under CC than under SV and BS. Wild bees showed the strongest numerical response to vegetation-based management, although hoverflies, bumblebees, and butterflies were all substantially more abundant under CC and SV than under BS.

While the temporal influence on butterfly abundance was not significant (p = 0.053), year effects were significant for total pollinators, honeybees, wild bees, hoverflies, and bumblebees. Honeybees, bumblebees, and butterflies showed significant treatment × year interactions, with treatment differences varying by year. Total pollinators, wild bees, and hoverflies did not show any significant interactions (Figure 3).

2.4. Relationships Between Vegetation Characteristics and Pollinator Abundance

Linear regression analyses showed significant positive relationships between vineyard floor vegetation characteristics and total pollinator abundance for all inter-row management treatments and sampling years (Figure 4). Pollinator abundance was positively related to vegetation cover, flower density and plant diversity, but the strength of these relationships was highly variable across the explanatory variables. Vegetation cover and flower density both explained ~60% of the variation in pollinator abundance, and the relationships were highly significant (p < 0.001; Figure 4, A–B). The results suggest that increases in ground cover and floral resources were consistently associated with more pollinators.

However, plant diversity as measured by the Shannon index (H′) was the best predictor of pollinator abundance. This single variable accounted for over 92% of the variation observed, with a correlation coefficient above 0.96 (p < 0.001; Figure 4, C). The relationship was strong across the diversity gradient, indicated by consistently narrow confidence intervals. Together, these results show that both vegetation cover and flower density are relevant drivers of pollinator activity, but that plant diversity is the vegetation trait most closely associated with pollinator abundance in the studied vineyard environment.

To estimate the relative contribution of vegetation characteristics to total pollinator abundance, a multiple linear regression model was created with flower density and plant diversity (Shannon index, H’) as predictors (Table 3). The overall model was highly significant (F_2,24 = 155.382, p < 0.0001) and explained over 92% of the variance in pollinator abundance (adjusted R² = 0.922); the model had an RMSE of 12.805 individuals per sampling plot. Plant diversity was the primary predictor, with a considerable positive standardised coefficient (β = 0.890) and a significant effect (p < 0.001). This suggests that greater plant diversity was strongly related to increased pollinator abundance. Flower density did not have a significant independent influence when plant diversity was included in the model (β = 0.094, p = 0.269). Its 95% confidence interval included zero, showing that the positive bivariate connection between flower density and pollinator abundance did not persist as an independent impact after controlling for plant diversity.

2.5. Multivariate Relationships of Inter-Row Management and White Grape Quality and Yield

A principal component analysis (PCA) was performed to analyse multivariate relationships between soil management treatments and grape quality and yield parameters. The first two principal components explained most of the total variation with PC1 and PC2 explaining almost 89% of the variance (Figure 5).

Grape compositional characteristics were clearly distinguished by the first principal component (PC1). Samples with higher values on this axis were characterised by increased pH, soluble solids (Brix) and total phenolic content, while titratable acidity was strongly negatively associated with PC1. This component is considered to represent a gradient of grape ripeness and compositional quality, with higher PC1 values suggesting advanced ripening and improved phenolic composition. By comparison, the second main component (PC2) was mostly affected by physical yield-related traits, i.e., berry weight and total yield, distinguishing treatments based on productivity instead of chemical composition. The biplot showed different clustering patterns of soil management practices. Samples from bare soil (BS) were found to be situated on the positive side of PC2, in the direction of the berry weight and yield vectors, indicating a relationship with higher values of these parameters. Cover crop mixture (CC) samples had a stronger relationship with the compositional characteristics represented along PC1, notably phenolic content, whereas titratable acidity was oriented in the opposite way along this axis. Samples from spontaneous vegetation (SV) were more widely scattered over the PCA space, indicating more variability in combined quality and yield attributes.

2.6. Multivariate Relationships of Inter-Row Management and Red Grape Quality

Principal component analysis (PCA) was used to examine the multivariate relationships between grape quality and yield traits for the red cultivars (Merlot and Pinot Noir). More than 85% of the variance was explained by the first two principal components, PC1 and PC2 (Figure 6). Titratable acidity showed a negative loading on the first principal component (PC1), which was a gradient of technical maturity with positive values for pH and soluble solids (Brix). Thus, this component primarily reflects ripening-related compositional changes. On the other hand, samples were categorised by productivity and secondary metabolite accumulation using the second main component (PC2).

The sample distribution in the PCA biplot revealed variations across the inter-row management strategies, although the 95% confidence ellipses showed some overlap. Samples from bare soil (BS) were more likely to occur in the direction of the berry weight and yield vectors, showing a relationship with higher values of productivity-related attributes. On the contrary, samples from cover crop mixtures (CC) were situated closer to phenolic compounds, anthocyanins, and titratable acidity, indicating improved compositional quality. Spontaneous vegetation (SV) treatment fell between the other two management strategies, indicating a moderate response with regard to both production and quality components. The PCA showed distinct multivariate patterns for the inter-row management approaches. Bare soil (BS) was more closely related to berry weight and yield, but cover crop combinations were more strongly linked to phenolic and anthocyanin properties. These patterns show a difference between productivity-related and compositional traits among inter-row management treatments.

2.7. Explanatory Capacity of Models for Grape Quality and Yield

The complete factorial models had explanatory potential for all observed grape quality and yield parameters (Table 4).

The models explained 0.799–0.996 of the variance, with adjusted R² ranging from 0.701 to 0.994. Phenolic compounds had the highest explanatory capacity, whereas Brix had the lowest, although it still represented a significant proportion of the explained variance (Table 4). Diagnostic analyses showed no influential observations and no evidence of multicollinearity (VIF = 1.00). Some models showed deviations from residual normality, most notably for phenolic compounds. These diagnostic data were used to interpret the model estimates.

3. Discussion

3.1. Effects of Vegetation Management on Vineyard Floor Species and Pollinator Resources

The results of the study suggest that inter-row soil management practices significantly and consistently affected vegetation characteristics, with cover crop mixtures resulting in the highest vegetation cover, flower density and species richness in all three growing seasons. This is consistent with previous studies showing that sown cover crop mixtures increase the availability of floral resources and enhance plant diversity compared to spontaneous vegetation or bare soil management [33,34,35]. In a European survey of 63 vineyards, extensive vegetation management and enhanced floral resource availability were linked to increased wild bee abundance and diversity, whereas less intensive management resulted in much higher inter-row vegetation cover [36]. Similarly, studies comparing vegetated inter-rows have found positive relationships between plant diversity and arthropod abundance, indicating the function of plant communities as an essential component of vineyard habitat quality [37].

All vegetation metrics showed significant Treatment × Year interactions, implying that treatment effects were dynamic and became more pronounced throughout the study. The cover crop treatments exhibited incremental increases in vegetation cover from 84.5% in 2023 to 92.7% in 2025, suggesting that perennial or self-seeding combinations can develop deeper root systems and become more competitive over time. In contrast, bare soil treatments consistently had low values for all vegetation metrics, indicating that tillage or herbicide application suppressed spontaneous plant establishment. The observed treatment ranking—cover crop combination > spontaneous vegetation > bare soil—was consistent in terms of flower density and plant species richness, supporting the view that active vegetation management may significantly improve floral resources in vineyard settings.

These outcomes support recent research from Northern Italy, which found herbaceous vegetation cover and floral diversity as major factors positively affecting pollinator abundance and diversity [38]. Similarly, studies from southern France have revealed strong associations between Hymenoptera abundance, floristic richness, and plant cover, highlighting the consistency of these relationships across viticultural regions [39].

3.2. The Composition of Pollinator Communities Reflects the Availability of Resources Provided by Vegetation

The NMDS ordination showed a clear differentiation among pollinator communities between treatments, with the bare-soil samples forming a clear cluster compared with the partially overlapping communities under cover-crop and spontaneous-vegetation treatments. This pattern indicates that vegetated inter-row treatments support more similar pollinator communities, probably due to comparable plant resources. Inter-row management treatment was also the dominant factor structuring pollinator communities, explaining >76% of the variance in community composition, far beyond the contribution of year (18%) or the ‘Treatment × Year’ interaction (3%). The strong effect of local vegetation management is in agreement with other vineyard studies which showed that ground vegetation characteristics were major drivers of wild bee community structure and often more important than landscape context or local influences [36]. In this regard, substantial vegetation management and improved floral resource availability have been linked to enhanced wild bee abundance throughout vineyards in different European countries [40].

Responses of individual pollinator groups give further insight into the ecological mechanisms behind these community-level patterns. Cover crop and spontaneous vegetation treatments had higher numbers of honey bees, wild bees, bumblebees, and butterflies than bare soil; wild bees had the highest numerical response to vegetation-based management. This differential sensitivity may reflect the dependence of many wild bee species on a diversity of floral resources and adequate nesting sites, which are more abundant and varied in vegetated inter-rows. The uniform ranking of treatments over the years indicates that the benefits of vegetation management are temporally stable and such stability may have important implications for the adoption of pollinator-friendly practices in viticulture [41]. The stability of treatment differences over years indicates that the effect of inter-row vegetation management was rather steady across the three-year research period. This temporal stability is important for vineyard management because it shows that the ecological advantages of vegetated inter-rows extended beyond a single growing season.

3.3. Plant Diversity Is the Best Predictor of Pollinator Abundance

Both vegetation cover and flower density were positively associated to total pollinator abundance, but plant diversity, expressed as the Shannon index, showed the strongest relationship with pollinator abundance. Plant diversity alone explained more than 92% of the observed variation, while the multiple regression model retained plant diversity as the only significant predictor (β = 0.890, p < 0.0001). Flower density did not provide a significant additional contribution after plant diversity was included in the model (β = 0.094, p = 0.269). Thus, within the conditions of the present study, the diversity of the plant community was more closely associated with pollinator abundance than flower density considered independently.

This result is consistent with findings from vineyards and other agricultural systems that plant variety and floral-resource diversity can be major predictors of pollinator populations. Kratschmer et al. [41] found a favourable relationship between the functional richness of blooming plants and the functional richness of wild bees in vineyard inter-rows, and they suggested varied cover-crop mixes to increase flowering-plant variety.

The recognition of plant diversity as the most significant characteristic of vegetation has practical consequences for vineyard management [42,43]. Results suggest that management should focus on establishing and sustaining the maintenance of diverse plant communities rather than maximising vegetation cover or floral abundance. This is in accordance with previous research recommendations to use site-adapted combinations of native wild plants with a variety of flower colours, shapes, and phenologies to sustain diverse pollinator groups [44]. The tight confidence intervals along the diversity gradient suggest a strong relationship between Shannon diversity and pollinator abundance, which contributes to the evidence that increasing plant diversity, is a reliable way to promote pollinator conservation in viticultural landscapes. The ecological interpretation of the current findings is that a more diversified plant community may supply a greater range of floral resources that differ in species identity, blooming season, and floral features. The regression analysis shows strong associations rather than causal effects; therefore, the high R² should be understood within the study’s geographical and temporal context.

3.4. Differentiated Effects of Soil Management on Grape Yield and Quality

Multivariate evaluations of grape quality parameters revealed that soil management strategies had distinct impacts on white and red grape cultivars, although overall trends remained quite similar across both cultivar groups. For white grape cultivars, PC1 was a gradient of technical maturity and chemical composition quality, with cover crop samples closer to titratable acidity and phenolic concentrations, and bare soil samples related to berry weight and yield. PC2 had a stronger correlation with berry weight and yield, showing a distinction between compositional and productivity-related features. Within this multivariate space, bare-soil samples were more closely related to berry weight and yield, whereas cover crop samples were more closely related to acidity and phenol content. Spontaneous vegetation took an intermediate position, showing a less obvious relationship with either production or compositional characteristics.

A similar tendency was seen in the red cultivars Merlot and Pinot Noir. Bare-soil samples were more strongly related to berry weight, yield, pH, and Brix, whereas cover crop samples were more directly linked to phenolic and anthocyanin variables, as well as titratable acidity. Spontaneous vegetation once again held an intermediate place. These findings point to a consistent distinction between productivity- and maturity-related features under bare-soil management and compositional factors under cover-crop management. However, because PCA represents multivariate relationships rather than treatment effects alone, these patterns should not be regarded as proof of direct causality.

The observed connections are consistent with earlier vineyard studies demonstrating that the impacts of groundcover on grape production and composition are substantially influenced by cover-crop type, soil, climate, and the intensity and timing of grapevine competition. In a three-year Mediterranean vineyard experiment in Sardinia, Muscas et al. [45], discovered that cover crops lowered grape yield while increasing sugar, anthocyanin, and polyphenol concentrations in comparison to tilled soil. A three-year study in Mediterranean organic vineyards in Chianti indicated that spontaneous vegetation and a legume cover crop might increase yield compared to other groundcover treatments, although the impact on anthocyanins and polyphenols differed depending on management method and year [46].

The full-factorial models accurately explained all grape quality and yield variables, with R² values ranging from 0.799 for Brix to 0.996 for phenolic compounds. The high explanatory power for phenolic compounds and anthocyanins suggests that the combination of cultivar, management treatment, and year accounted for a significant fraction of the observed variance in these parameters. However, R² values should be considered as indications of model explanatory power, not as evidence that soil management alone impacted grape composition.

3.5. Implications for Sustainable Vineyard Management and Pollinator Conservation

The findings have a direct impact on developing long-term vineyard management strategies that balance grape quality, productivity, and biodiversity conservation. The association of cover crop management with higher pollinator abundance and grape compositional traits suggests that this strategy can provide diverse ecological services, including pollinator conservation and associated ecosystem services, including soil conservation, and improved wine quality. Similar research has demonstrated that keeping vegetated inter-rows can boost arthropod abundance, and that plant species diversity is a key vegetation trait connected with arthropod populations in Mediterranean vineyards [37].

However, there are challenges to adopting cover crop mixtures in vineyards. Competition for water and nutrients, particularly in areas with water scarcity, is an important barrier to successful implementation [47]. Managing alternating rows with spontaneous vegetation has been suggested as a suitable solution for erosion control, yield support, and grape quality improvement [48]. Research results indicate that spontaneous vegetation treatments may provide intermediate advantages to both pollinators and grape quality, but with greater variability and lower productivity than in cover crop mixtures. Future studies should investigate the long-term viability of these strategies and their interactions with climate variability, soil parameters, and cultivar-specific responses.

3.6. Limitations and Opportunities for the Future

Several limitations to this study should be addressed. First, the study was carried out across three growing seasons in a single vineyard, thus the applicability of the results to other viticultural locations with various climatic conditions, soil types, and management practices may be limited. Second, while the study focused on the effects of soil management on vegetation characteristics and pollinator communities, it did not account for the potential influences of the surrounding landscape and semi-natural habitats, or agricultural chemical use, all of which are known to have an impact on pollinator communities in farming environments. Third, the grape quality assessments were confined to basic enological indicators and did not include sensory evaluations or in-depth wine analyses, which could offer additional information about the impact of management strategies on wine quality. Fourth, the presence of heterogeneity in the model for phenolic compounds, as well as deviations from residual normality in some models, suggests that caution should be applied when interpreting the results, and future research should consider more advanced statistical approaches or larger sample sizes to address these issues.

4. Materials and Methods

4.1. Study Area and Plant Material

The study was carried out in a commercial vineyard located in the Minas-Maderat wine-growing area, Arad County, western Romania (46°05′ N, 21°41′ E), during the 2023–2025 growing seasons. The area has a mild continental climate with Mediterranean influences (an average annual temperature of 10-11 °C and annual rainfall of 600-650 mm). The vineyard soils are mainly dark brown forest soils with rendzinas forming on the gently sloping limestone horizon. These soils are rich in stable humic acids, have high cation-exchange capacity and have a neutral to slightly alkaline pH reaction (6.7-7.8).

The study included four wine grape cultivars, two red cultivars, Merlot (nine years after planting) and Pinot Noir (10 years after planting), and two white cultivars, Sauvignon Blanc (13 years) and Riesling Italian (10 years). The rows of the vineyard were oriented in a north–south direction and the grapevines were cultivated under rainfed conditions. All four cultivars were trained on a single Guyot system. Winter pruning, canopy management during the growing season, protection and fertilization treatments were carried out according to the same standard commercial practices, uniformly applied in the entire vineyard.

4.2. Field Experimental Design

The field experiment was designed as a randomised complete block design (RCBD). For each of the four grapevine cultivars, three spatially separated blocks were established along the predominant environmental gradient of the vineyard to account for potential spatial variation in soil characteristics, slope and microclimatic conditions. Within each block, the three soil management treatments were randomly assigned to individual plots within the block. Therefore, nine experimental plots (three soil management treatments × three blocks) per cultivar were included, generating a total of 36 experimental plots for the four cultivars.

The experimental plot size was 10 m x 2 m with seven successive vines (distance of 1.5 m between vines) corresponding to a planting density of approximately 4,444 vines ha-1. To minimise potential edge effects, grape yield and berry quality measurements were conducted on three central vines of each experimental plot with the lateral vines excluded from sampling. During the 2023–2025 study period the same experimental plots and assigned central vines were monitored.

4.3. Inter-Row Vegetation Management

Inter-row floor management was established as follows:

(i)

bare soil (BS): managed by superficial soil tillage (up to 10 cm, three times per year).

(ii)

Spontaneous vegetation (SV): naturally occurring vegetation was maintained throughout the growing season and controlled by mowing to a height of approximately 5 cm.

(iii)

Cover crops (CC): a multispecies cover crop mixture was established using the following species, expressed as percentages of seed weight: white clover (Trifolium repens, 20%), birdsfoot trefoil (Lotus corniculatus, 20%), alfalfa (Medicago sativa, 10%), red fescue (Festuca rubra, 15%), sainfoin (Onobrychis viciifolia, 20%), and yarrow (Achillea millefolium, 15%). The total seeding rate was 30 kg ha⁻¹. The seed mixture was sown in September 2022 and incorporated into the soil at a depth of 2–3 cm across the inter-row area.

Spontaneous vegetation and cover crops were managed by mowing three times a year in late May, mid-July, and early September. Removed vegetation was left in situ as mulch.

Vegetation Assessment

Vegetation was sampled twice during May–July, corresponding to the peak flowering period of the component plant species. Five quadrats (1 m × 1 m) were randomly selected in each plot, at least 0.5 m from the edge. Vegetation cover (%), flower density (flowers m^-2) and plant species richness (number of vascular plant species per quadrat) were noted. Plant diversity was further assessed using the Shannon diversity index (H′), calculated from the relative frequency of occurrence of each plant species within the quadrats. The five quadrat-level H′ measurements were averaged to calculate the average plant diversity value for each experimental plot. Spontaneous species were identified directly in the field using standard botanical identification criteria; the Pl@ntNet plant identification platform [49], was used as an alternative identification tool for specimens that could not be successfully identified in the field. The sown cover-crop mixture was kept throughout the experimental period and the same management practices were applied for all three years of study. Plant species richness included planted and naturally occurring spontaneous vascular plant species.

4.4. Pollinator Sampling

Pollinator communities were studied using fixed transects, established according to the plot dimensions (10 m × 2 m) to ensure spatial alignment between vegetation and pollinator assessments. Observations were conducted annually throughout the season during the peak flowering period (May to July), at two-week intervals. Sampling was limited to days with suitable weather conditions favourable for pollinator activity: air temperature >18 °C; wind speed <3 on the Beaufort scale, and no precipitation; surveys were undertaken between 09:00 and 16:00 h.

In each transect, the observer repeatedly walked the entire 10-m transect during a standardised observation period of 15 min and recorded all insects observed visiting or interacting with flowers in the inter-row vegetation. When necessary, individuals were captured using an entomological net for identification and subsequently released. Identification was performed in the field whenever possible and, when necessary, supplemented using the iNaturalist identification platform [50] (accessed during the study period). Pollinator abundance was recorded for each transect at each sampling occasion, and observations from the multiple sampling occasions within each year were averaged to calculate the annual mean abundance for each transect.

Pollinators were classified in five groups as follows: honey bees (Apis mellifera), wild bees (Anthophila; excluding Apis mellifera and Bombus spp.), bumblebees (Bombus spp.), hoverflies (Syrphidae), and butterflies (Lepidoptera). Community diversity was characterised using the Shannon diversity index (H′) and Simpson diversity (1 − D), and the relative composition of the five pollinator groups was assessed.

4.5. Grapevine Yield and Berry Quality

Grape yield and berry quality were evaluated at technological maturity during each growing season (2023-2025). The sampling date was adjusted according to the maturity stage of each cultivar. Grapes were collected from the three centrally located vines in each experimental plot to determine grape yield (kg vine^-1) and berry weight (g).

Total soluble solids (TSS, °Brix) were measured using a digital refractometer (HI96811, Hanna Instruments, Woonsocket, RI, USA). Titratable acidity (TA, g L⁻¹ tartaric acid equivalents) was determined by titration with NaOH (Hanna HI84502 mini-titrator), and juice pH was measured using a calibrated pH meter (HI2002, Hanna edge^®).

For red grape cultivars (Merlot and Pinot Noir), the accumulation of phenolic and anthocyanin compounds was determined in berry skins using spectrophotometric methods. Absorbance measurements were performed using a HI801 Iris visible spectrophotometer (Hanna Instruments, Woonsocket, RI, USA). Total phenolic content was determined using the Folin–Ciocalteu method by measuring absorbance at 765 nm and expressing the results as gallic acid equivalents (mg GAE L⁻¹). Total anthocyanin content was determined using the pH differential method in buffer solutions at pH 1.0 and 4.5. Absorbance was measured at 520 and 700 nm, and anthocyanin concentration was expressed as cyanidin-3-glucoside equivalents (mg L⁻¹).

4.6. Pollinator Community Indicators of Habitat Quality

Pollinator community attributes were utilised to indirectly assess habitat quality and ecological conditions in the vineyard agroecosystem. Although grapevine is primarily self-pollinating, pollinator abundance and community diversity can serve as important indicators of vineyard habitat sustainability. The indicators considered were total pollinator abundance, Shannon diversity (H′), Simpson diversity (1 − D), and the relative composition of the main pollinator groups.

4.7. Statistical Analysis

All statistical analyses were performed using R software (version 4.6.1; R Core Team, 2026) and XLSTAT (version 2023.5.1; Addinsoft, Paris, France). Statistical significance was set at p < 0.05. Data distributions and model residuals were assessed for normality and homogeneity of variance. When assumptions were not satisfied, data transformations were evaluated, and the model diagnostics were reviewed.

Vegetation characteristics (vegetation cover, flower density, and plant species richness) were analysed using linear mixed-effects models (LMMs) in XLSTAT, with inter-row management treatment, year, and their interaction as fixed effects, and block and experimental plot nested within block as random effects. Significant differences among treatments were assessed using Tukey’s HSD test.

Total pollinator abundance and the abundance of the five pollinator groups (honey bees, wild bees, bumblebees, hoverflies, and butterflies) were analysed using two-way linear models with treatment, year, and their interaction as fixed effects. Differences among treatments were assessed using ANOVA followed by Tukey’s HSD or Tukey-adjusted estimated marginal means (EMMs). Associations between vegetation parameters and pollinator abundance were assessed using Pearson’s correlation coefficients, followed by linear regression analysis to identify significant associations.

Pollinator community composition was analysed in R using the vegan package. The abundance matrix of the five pollinator groups was square-root transformed and used to calculate Bray–Curtis dissimilarities. Community composition was visualised using two-dimensional NMDS, and differences associated with treatment, year, and their interaction were tested using PERMANOVA with the adonis2 function and 999 permutations. Homogeneity of multivariate dispersion was assessed using PERMDISP (betadisper). When significant treatment effects were detected, pairwise PERMANOVA comparisons were performed using 999 permutations. SIMPER analysis was used to assess the contribution of individual pollinator groups to Bray–Curtis dissimilarities.

Grape yield and berry quality were analysed using full-factorial linear models with cultivar, soil management treatment, year, and their interactions as explanatory factors. Separate models were fitted for Brix, juice pH, titratable acidity, total phenolic content, total anthocyanin content, berry weight, and grape yield, using the model: “Response variable ~ Cultivar × Treatment × Year”.

Model performance was assessed using R² and adjusted R², while residual diagnostics were used to evaluate normality, homogeneity of variance, influential observations, and multicollinearity using the performance package in R.

PCA was used to investigate multivariate relationships among soil management treatments and grape yield and berry quality variables. The general PCA included Brix, pH, titratable acidity, phenolic content, berry weight, and yield. A separate PCA for Merlot and Pinot Noir additionally included anthocyanin content. PCA biplots were used to visualise variable loadings and relationships among treatments and response variables.

5. Conclusions

The results of this study indicate that inter-row soil management strategies were associated with differences in vineyard vegetation traits, pollinator groups, and grape quality during the growing seasons analysed. The cover crop mixture constantly resulted in higher vegetation cover, flower density, and plant species richness than the spontaneous vegetation and bare soil treatments. Pollinator abundance showed a strong association with plant diversity (Shannon diversity index) and showed the strongest association with the vegetation variables. These results emphasise the significance of inter-row vegetation composition for plant and pollinator diversity in the vineyard ecosystem. Differences in the quality parameters of the grapes were also linked to inter-row management practices. Cover crop mixtures were associated with higher phenolic and anthocyanin-related traits. Spontaneous vegetation performed moderately in general, although the magnitude and direction of treatment effects impact fluctuated seasonally and by cultivar. The findings suggest that inter-row vegetation management may involve a compromise between yield-related traits and grape compositional parameters.

The findings suggest that cover crops have the potential to be a multifunctional vineyard management tool, enhancing vegetation resources and supporting higher pollinator abundance as well as certain grape compositional traits. However, these benefits should be evaluated considering the potential for competition for water and nutrients between cover crops and vines, especially in areas where resources are limited. The consistency of numerous treatment responses across seasons and cultivar groups supports the use of varied inter-row vegetation in sustainable vineyard management. Additional multi-year studies in different environments and with a wider range of cultivars are necessary in order to assess the stability and applicability of these results. Furthermore, sensory studies and wine-level evaluations are essential to evaluate whether the observed changes in grape composition are translated into relevant effects on wine quality.