Multisensor Analysis for Biostimulants Effect Detection in Sustainable Viticulture

1. Introduction

The agricultural sector is increasingly challenged by climate change, which has caused significant shifts in weather patterns, including rising temperatures and unpredictable rainfall. These changes have necessitated the adoption of sustainable farming practices that prioritize environmental compliance while maintaining crop productivity and quality [1].

In viticulture, climate change manifests through earlier harvest times, reduced yields, and alterations in grape composition, all of which adversely affect the quality of wine [2]. Key abiotic stresses, such as excessive sunlight, water deficits, and high temperatures during the growing season, can severely impact vine health and grape quality, posing a considerable challenge to sustainable viticulture, particularly in Mediterranean regions [3]. In these regions, including Italy, vineyards are often cultivated without irrigation support, relying solely on natural rainfall. This reliance makes them highly vulnerable to drought, which can lead to significant reductions in yield and deterioration of grape quality traits, thereby affecting vinification processes and the sensory attributes of the resulting wine [4,5].

Conventional farming practices, particularly the extensive use of chemical fertilizers, are commonly employed to maintain crop yields and economic viability. However, these practices have profound environmental consequences, including soil contamination and reduced fertility, water pollution, and eutrophication [6]. The persistent application of fertilizers and agrochemicals in vineyards often leads to the accumulation of organic pollutants and heavy metals, further degrading soil quality and posing serious environmental and toxicological risks [7,8]. This degradation, coupled with erosion and sustained tillage, exacerbates the challenges associated with maintaining soil health in viticultural systems. Effective management of these issues necessitates the availability of timely, precise, and comprehensive data regarding vineyard conditions and stress factors, underscoring the importance of integrating advanced monitoring technologies in viticulture [9].

Biostimulants have emerged as a promising alternative to conventional agrochemicals, aiming to enhance plant growth and resilience while minimizing environmental impacts. These substances, applied to soil or foliage, stimulate natural plant processes independent of their nutrient content, thereby improving nutrient use efficiency, enhancing resistance to biotic and abiotic stresses, and promoting overall crop quality [10,11,12,13,14]. The European Union has recognized biostimulants as fertilizing products, reflecting their potential to support sustainable agricultural practices. They can be broadly classified into categories such as humic substances, seaweed extracts, complex organic materials, amino acids, antitranspirants, beneficial elements, inorganic salts, and chitosan derivatives [15]. Among these, seaweed extracts and humic acids have demonstrated significant benefits in viticulture. Seaweed extracts, applied as foliar treatments, have been shown to promote root growth, enhance nutrient uptake, and improve vegetative growth, contributing to increased vine vigor and grape quality [16]. Humic acids, on the other hand, enhance chlorophyll content, protein levels, and macro nutrient absorption in leaves, leading to improved plant efficiency and productivity, and rising anthocyanin content in the juice [17]. Fulvic acids, another category of humic substances, positively influence several growth parameters, including bud burst, vegetative development, leaf area expansion, and grape composition, while maintaining acidity levels that are crucial for wine quality.

Despite the growing interest in biostimulants within the agricultural industry (specifically on multi-compound products), scientific research has lagged behind commercial developments, partly due to the complexity of understanding their modes of action and the multifaceted nature of their effects on plant physiology. There is a pressing need for comprehensive studies that elucidate the mechanisms by which biostimulants influence plant health and productivity [18]. The advent of digital technologies, such as remote and proximal sensing, offers new opportunities to accurately monitor and quantify the effects of biostimulants on crop performance at different spatial and temporal scales.

Traditional methods for assessing plant stress and physiological status, such as the Scholander pressure chamber for Stem Water Potential (SWP) measurement, are limited by their destructive sampling requirements and labor-intensive nature, making them unsuitable for large-scale applications [19]. Similarly, tools like the SPAD 502 meter for chlorophyll content estimation and the Multiplex Research 3 (MFA) fluorimeter for assessing physiological status are valuable but constrained by their inability to cover extensive vineyard areas efficiently [20,21,22]. Remote sensing technologies, particularly those integrated into Unmanned Aerial Systems (UASs), have become increasingly relevant in precision viticulture. These systems provide non-invasive, scalable solutions for monitoring various vineyard parameters, including plant vigor, nutrient status, and disease incidence, enabling a comprehensive understanding of vineyard dynamics [23,24,25].

The ability to conduct multi-temporal analyses is a key advantage of remote sensing, as it allows for the monitoring of changes in vineyard parameters over time, facilitating the development of historical data series that can improve crop management decisions. By integrating remote sensing with traditional ground-based measurements, it is possible to obtain a more comprehensive and accurate picture of how biostimulants affect vineyard performance, ultimately contributing to more sustainable and effective viticulture practices [26,27,28].

This study aims to evaluate the sensitivity of remote and proximal sensors, alongside traditional manual observations, in detecting the effects of varying biostimulant application rates on vineyard biomass production, water stress resistance, and grape yield and quality throughout the vegetative and production phases. By employing a combination of these technologies and mainly focusing on the possibility of detecting such effect on a large scale relying only on UAS multispectral technology, this research seeks to enhance the understanding of biostimulant efficacy and provide valuable insights into the development of sustainable precision viticulture strategies that promote environmental sustainability and economic viability.

2. Materials and Methods

2.1. Experimental Site

The study was carried out at the "Columbu" winery in Magomadas (OR), Sardinia, Italy (coordinates: Easting 457074.761, Northing 4456775.910; UTM 32 N, EPSG 32632), in an organic vineyard of cultivar "Malvasia bianca" (controlled designation of origin Malvasia di Bosa DOC), covering an area of 0.81 ha (Figure 1). The vineyard, grafted onto 420 A rootstock and located at 220 meters above sea level (ASL), is characterized by Guyot training system with a 0.9 × 2.1 m spacing configuration, corresponding to 5290 plants per hectare, oriented northwest-southeast. The terrain exhibits a slope perpendicular to the vine rows (Figure 1b). Field data collection was conducted under sunny, clear sky conditions during the 2020 (3 July, 29 July, and 24 August) and 2021 (18 June, 2 August, and 24 August) seasons.

2.2. Experimental Design and Agronomic Management

The experiment consisted of four treatments, each containing different concentrations of Biopromoter biostimulant (Eurovix S.P.A.), replicated in two blocks, and a single plot size of 5.5 m (5 rows) x 60 m, arranged perpendicular to the terrain slope using a randomised complete block design (Figure 1a). The vineyard was managed uniformly, except for biostimulant application, which followed four specific supply treatments outlined in Table 1 and Table 2.

The organic biostimulant fertilizer utilized in the survey originates from the methodical maturation and stabilization of selected organic substrates, augmented with microbial strains and enzymes. Humic and fulvic acids inclusion, in conjunction with macro and micronutrients, microbial communities, and advantageous enzymes, renders the biostimulant an agent for restoring soil fertility (Table 3).

2.3. UAS Platform and Implemented Sensors

The UAS Phantom 4 Pro (DJI Technology Co., Ltd, Shenzhen, China) equipped with an RGB (Red Green Blue) CMOS 1" 21-megapixel (MP) resolution sensor alongside two 12 MP Survey 3 multispectral cameras (Mapir, San Diego, USA) performed remote sensing data acquisition. The MAPIR calibration target and Camera Control calibration software (MAPIR, San Diego, CA, USA) enabled raw data surface reflectance correction. UAS flights were conducted at an altitude of 45 m Above Ground Level (AGL), with a speed of 2.5 m s^-1, maintaining 85% front and 75% side overlaps, using the DJI Pilot android app for flight planning and execution, totaling 4 minutes and 15 seconds per flight. To ensure accurate orthomosaic orthorectification, an RTK GNSS Reach RS+ (Emlid Tech Kft., Budapest, Hungary) recorded coordinates for 9 Ground Control Points (GCPs), incorporated inside the structure from motion software Agisoft Metashape (Agisoft LLC, St. Petersburg, Russia), used for the vegetation index digital models obtainement together with QGis open-source software (ver. 3.30.0, QGIS Development Team) by incorporating Equations 1 and 2 [24].

$$NDVI={\displaystyle \frac{NI{R}_{(850 \mathrm{nm})}-Re{d}_{(550 \mathrm{nm})}}{NI{R}_{(850 \mathrm{nm})}+Re{d}_{(550 \mathrm{nm})}}},$$

(1)

$$NDRE={\displaystyle \frac{NI{R}_{(850 \mathrm{nm})}-RedEdg{e}_{(725 \mathrm{nm})}}{NI{R}_{(850 \mathrm{nm})}+RedEdg{e}_{(725 \mathrm{nm})}}},$$

(2)

2.4. Fluorometer Proximal Analysis

The proximal sensing tool implemented for fluorometric analysis was the portable UV-Visible MFA (Force A, Orsay, France), able to detect information about the physiological status of vegetation and grapes in a non-destructive way. The system is composed of three light detector channels, able to detect the Far-Red Fluorescence (FRF), the Blue Green Fluorescence (BGF), and the Red Fluorescence (RF) emitted by the tool and reflected by the plant. The light emission derives from an LED system that works at 375 nm (UV Radiation) and six RGB operating at 470 nm (Blue Radiation), 516 nm (Green Radiation), and 635 nm (Red Radiation). The decision to include the MFA in the trials derives from its ability to monitor the vegetation status and bunches’ growth-ripening phase without causing any damage to the crop. For this purpose, three indices were selected to recognize any physiological differences between plots [29], focusing on the analysis of stress (Equation 3) and flavonoids (Equation 4) on the upper page of adult leaves and the bunches. The fluorometric indices used and their mathematical function are listed below.

$$BRR-FRF={\displaystyle \frac{(YF-UV)}{(FRF-G)}},$$

(3)

$$FLAV={\displaystyle \frac{log(FRF-R)}{(FRF-UV)}},$$

(4)

2.5. Plant Water Stress Estimation

Scholander pressure chamber (SPC) (PMS Instrument Company, Albany, USA) was used to measure the water tension associated with plant water stress (SWP) [30]. The procedure involved sampling 16 leaves, each enclosed in an SKB bag for 1 hour, with two leaves collected from each plot. The samples of the leaves were then placed in the pressure chamber, gradually increasing the applied negative pressure until the xylem tension was overcome, indicated by the appearance of water on the stem of the leaf. Higher pressure values corresponded to greater leaf tension, reflecting elevated levels of water stress in the sampled plants.

2.6. Leaves Chlorophyll Estimation

SPAD 502 (Minolta, Osaka, Japan), a non-destructive portable tool to measure the chlorophyll concentration and the derived nitrogen content in leaves [31], was implemented according to the experimental design reported in section 2.2 to sample two plants (five leaves each) in each parcel as an indicator of vegetation health status, crop productivity, and photosynthetic efficiency [32].

2.7. Quantitative and Qualitative Harvesting Evaluation

To assess the impact of biostimulant applications on grape production, field sampling was conducted on 23 September 2020 and 28 September 2021. During each of these sampling events, the production quantity was measured as kg per plant for six randomly selected vines from each experimental plot. The individual vine yields were averaged to provide a representative value for each plot’s overall productivity.

In addition to yield measurements, grape quality assessments were performed on three dates prior to harvest for each year. Specifically, samples were collected on 9, 17, and 23 September in 2020 and on 7, 13, and 28 September in 2021. For each sampling date, random bunches were harvested from five plants per plot, resulting in a combined sample weight of approximately 1 kg per plot (Figure 1a). This sampling strategy ensured that the grape quality analysis was comprehensive and representative of each plot’s overall condition.

Following collection, the grape samples underwent a preparation process in the laboratory [28]. The entire sample from each plot was crushed, and the resulting must was filtered to separate coarse components such as skins and seeds. A 50 ml portion of the clarified liquid was then transferred into Falcon-type plastic tubes for further analysis. After allowing sufficient time for the solid particles to settle within the tubes, the liquid fraction was subjected to a quantitative and qualitative analysis using a Fourier transform mid-infrared spectrometer (WineScan™ Flex, Foss Italia srl, Italy).

The WineScan™ Flex spectrometer facilitates the precise determination of various grape quality attributes critical to evaluating the qualitative characteristics of Malvasia bianca [33]. In this study, titratable acidity, pH, and total soluble solids (°Brix) were reported as the primary indicators of grape quality, given their significant influence on sensory properties and overall quality profile of the resulting wine. Titratable acidity and pH provide insights into the balance and stability of the wine, while °Brix value, representing the concentration of soluble sugars, is a key determinant of grape ripeness and potential alcohol content. Together, these parameters offer a robust framework for evaluating the effects of biostimulant treatments on grape quality under varying environmental conditions.

2.8. Statistical Analysis

Survey features were evaluated using Python (Python Software Foundation) scripts through Pandas, Matplotlib, and Seaborn packages for data manipulation and plotting.

Analysis of Variance (ANOVA) was performed for each season separately using the statistical package SPSS 16^®. When needed, Least Significant Differences (LSD) were calculated at the 5% significance level to discriminate between treatment means and reported inside the plots in the following Results section.

3. Results

3.1. Vegetative Status Establishment

3.1.1. UAS Monitoring

The NDVI results (Figure 2) for both growing seasons demonstrated a higher vegetative response during the early stages, with values ranging from 0.70 to 0.75 in July. These values indicate substantial vegetative vigor, which progressively declined as the growing season advanced, reaching lower levels between 0.60 and 0.65 in August. Despite receiving higher nutrient rates, the blue treatment consistently exhibited lower performance compared to the red and yellow treatments throughout both seasons. This anomaly is likely attributable to the specific characteristics of the plot’s location in the northwest section of the vineyard (Figure 1a), an area marked by limited nutrient availability and suboptimal soil conditions, predisposing the vines to reduced vegetative growth and vigor.

In 2020, the NDRE results (Figure 3a) depicted a general decreasing trend across all treatments. Both the blue and yellow treatments initially started from marginally lower NDRE values and converged toward similar values as the season progressed. However, the blue treatment’s performance remained significantly lower, especially on the final assessment day, highlighting the considerable influence of the field’s heterogeneity, which disproportionately affected the blue treatment. A comparable trend was observed in the 2021 season (Figure 3 b), although overall performance decreased further compared to 2020.

This reduction in vegetative indices can be attributed to adverse climatic conditions, including elevated temperatures and water scarcity, to which the vines were unable to adequately adapt. Additionally, the detrimental impact of nearby wildfires, which reached the vineyard’s borders, likely exacerbated the stress conditions experienced by the plants. This hypothesis is supported by the NDVI data from the same year (Figure 2b), where only the red treatment maintained a stable trend throughout the season, whereas the green and yellow treatments displayed a sharp decline towards the end of the season. This overall vegetative decline is further corroborated by the SWP results from 2021 (Figure 6b), which showed a marked reduction in water status compared to the previous year.

Throughout both years, none of the vegetation indices were able to distinctly differentiate between treatments that displayed similar behavior over the growing seasons, with the exception of the blue treatment. This uniformity in response across treatments might be due to the consistent application of biostimulant rates (Table 1), which did not produce the expected differentiation in vegetative response, particularly for the green and blue treatments, which were hypothesized to outperform the others under increased nutrient supply.

3.1.2. MFA Leaves Analysis

The MFA results for the fluorescence indices (Figure 4) displayed a generally consistent trend among the four treatments, with notable variation between the two years. The indices indicated a more stable, albeit critical, vegetative condition in 2021, mirroring the trends observed in the UAS-derived vegetation indices and indicating a general decline in vegetative vigor in the second year of the study.

In 2020, the BRR-FRF index results for leaves (Figure 4a) exhibited a pronounced decreasing trend across all treatments, with values declining from approximately 0.75-1.0 at the beginning of the season to 0.2 by the end. The green treatment initially recorded high values (0.7) on the first sampling date but converged to similar lower levels as other treatments by the final survey date. In contrast, the 2021 season (Figure 4b) displayed a more consistent BRR-FRF response across treatments, with no significant differences observed and values stabilizing around 0.4 across the two monitoring dates.

The FLAV index results followed a similar pattern to the BRR-FRF index in both years, reflecting the physiological maturation of the vine canopy. In 2020, the total flavonoid content exhibited a steady increase, indicating typical canopy development during the growing season (Figure 5a). Conversely, the 2021 season (Figure 5b) displayed a nearly constant trend in FLAV values, maintaining elevated levels (around 1.5) compared to the previous year, in line with the BRR-FRF index results. These observations suggest a more critical but stable physiological condition, consistent with the intensified environmental stresses experienced during the second year.

The MFA indices clearly demonstrated significant differences in the first 2020 survey, which is inconsistent with the applied biostimulant rates. In stark contrast, the 2021 season showed no significant difference between treatments.

3.1.3. Water Stress Measurement

Early in the 2020 season, SWP values were uniform across all treatments, declining as water availability diminished (Figure 6a).

However, as the season progressed, the green treatment displayed a distinct advantage in maintaining higher water status (-1.2 MPa) compared to other treatments (-1.4 MPa), as showed in the last date of 2021, which reported a significant distinction between treatments, comparable to the applied biostimulant rates.

SWP divergence, not captured by the vegetation indices, likely reflects the ability of certain vines to better withstand water stress in a non-irrigated vineyard system, where only vines in optimal condition can sustain higher SWP values in late phenological stages. The observed variation may also be related to the biostimulant application rates, which were possibly insufficient to support increased biomass production, particularly in the initial year of treatment.

A similar trend was observed in the second year (Figure 6b), with all treatments starting from a better initial condition but converging to similar results. The green treatment, however, again demonstrated superior water stress resistance, highlighting the cumulative effects of biostimulant applications and their potential role in enhancing drought resilience over time. Despite the continued use of biostimulants, all treatments exhibited heightened water stress sensitivity in 2021, likely due to extreme weather conditions and the unanticipated occurrence of wildfires near the field, as previously discussed.

3.1.4. Leaves Chlorophyll Content Estimation

SPAD measurements, indicative of leaf chlorophyll content, revealed a declining trend without significant differences between the four treatments in the first year (Figure 7a). The green treatment, however, exhibited a variable pattern and improved performance on the final survey date, suggesting a delayed response to biostimulant application.

In 2021, both green and blue treatments showed markedly better SPAD values compared to red and yellow treatments (Figure 7b), consistent with observations from the previous year. These results suggest that the cumulative effect of biostimulant applications becomes more evident after two consecutive years, or alternatively, that the SPAD meter demonstrated a higher sensitivity to subtle differences in vegetative vigor not captured by other remote and proximal sensing methods employed in this study.

3.2. Production Results

3.2.1. Production Quantitative Analysis

As anticipated, the production outcomes in 2020 (Figure 8a) were closely aligned with the fertilizer supply patterns described in Table 1. The green treatment yielded the highest grape production, averaging 1.25 kg of grapes per plant, followed in descending order by the blue, yellow, and red treatments.

In the second year (Figure 8b), a more pronounced divergence in production was observed, with the green and blue treatments yielding approximately 1.75 and 1.5 kg of grapes per plant, respectively, outperforming the other treatments. This trend underscores the potential influence of biostimulant products on fruit production, which appeared to have a greater impact on yield than on vegetative vigor alone, or the well-known seasonal variability, influenced by the nutrient availability and meteorological conditions relationship.

It is important to note that the production data represents a single measurement taken before harvest and therefore does not capture the dynamic progression of grape growth throughout the season.

A key observation in the second year is that all treatments, except for the blue treatment, exhibited reduced variability in yield (excluding outliers), signifying an overall improvement in plant condition compared to 2020. This reduction in variability suggests that, over time, the application of biostimulants may contribute to a more consistent and stable production outcome across the different treatments.

3.2.2. Grapes Quality Assessment

Brix results exhibited a fluctuating pattern in the first year (Figure 9a), with no significant differences between treatments and no notable improvements throughout the growing season. The initial Brix values for 2020 ranged between 20° and 22° Brix, and by the final survey date, the treatments displayed homogeneous sugar content.

However, the second year (Figure 9b) revealed a substantial upward trend in Brix levels throughout the season, again with no significant differences between treatments. This improvement in sugar content may be more closely linked to the critical temperature changes observed in 2021 [34], rather than the effects of biostimulant products.

The titratable acidity measurements showed a comparable decreasing trend in both growing seasons. Early assessments in 2020 (Figure 10a) revealed substantial differences between treatments, but these differences diminished over time, with all treatments converging to a similar acidity level of 6.75 at the end of the season. In 2021 (Figure 10b), the treatments started under more favorable conditions, with the blue treatment unexpectedly achieving the highest acidity levels initially. However, as the season progressed, trends shifted, and by the final assessment, the green treatment emerged as the best-performing treatment, paralleling the results observed in grape production.

The pH readings exhibited a rising trend in both seasons, with the first season yielding the most favorable results for the evaluated cultivar. The 2020 results (Figure 11a) indicated no significant differences between treatments, except for the yellow treatment. In 2021 (Figure 11b), the treatments demonstrated higher, but still acceptable, acidity levels compared to the previous year, with the blue treatment initially attaining the highest acidity levels, before eventually reaching levels comparable to the other treatments.

3.2.3. MFA Grapes Analysis

The BRR-FRF index (Figure 12), measured on the grape bunches, displayed a gradual decline in the first year and a significant increase in the second year. This shift suggests that grape bunches experienced higher levels of environmental stress in 2021, confirming similar findings reported for leaf stress on August 2, 2021 (Figure 4b). When comparing the two growing seasons, the BRR-FRF values on the first monitoring date in 2021 were lower than the corresponding values from July 29, 2020, indicating reduced stress early in the season. However, by the final monitoring date, the BRR-FRF values had increased sharply, reflecting a sudden and unfavorable change in environmental conditions at the experimental site.

In contrast, the FLAV index (Figure 13) exhibited an opposite trend to the BRR-FRF index, with FLAV values generally increasing as BRR-FRF values decreased. This inverse relationship remained consistent across all four treatments, indicating that the biostimulant application had little influence on the variability detected by the FLAV index. In 2021, the FLAV index showed a slightly decreasing but stable trend, starting from a much higher initial value than in 2020. This behavior mirrors the trends observed in leaf fluorescence indices, indicating a general improvement in grape quality during the second year.

The difference in trends between the BRR-FRF and FLAV indices in 2021 could be attributed to the specific grape variety used in the study, typically harvested late in the season, which may have led to increased stress on the grapes due to dehydration. The relatively stable FLAV levels combined with the elevated BRR-FRF index suggest that the optimal harvest time for this grape variety, particularly for wine production, coincides with the period of higher stress and stable flavonoid content.

4. Discussion

This study investigated the sensitivity of both remote and proximal sensors, alongside traditional observations, in detecting the effects of different biostimulant application rates on canopy development, water stress resistance and grape yield and quality during various vegetative and production phenological stages exploring the possibility to rely only on UAS multispectral technology to improve the sustainable management of viticultural systems. In other words, the goal was not only to explore the opportunity to evaluate the impact on plants’ vigor using multiple tools, but also to understand if such discrepancy between the analyzed treatments could rely only on UAS remote analysis, more suitable on a large scale and less prone to labor fatigue.

The UAS-derived results highlighted the potential of UAS for reliable large-scale evaluations of the vegetative status of vineyards [35]. Consistent findings over two years underscore the practical utility of vegetation indices for ongoing plant monitoring and spatial variability assessment, as proved by M.V. Ferro et al. (2023) [36]. NDVI and NDRE indices effectively reported treatment trends, although differentiation between treatments was limited, except for blue treatment, which consistently underperformed over the two years, in contrast to what has been reported from the SPAD and in a more limited way by the SWP. The limited UAS ability to differentiate between treatments may derive from the lower sensitivity of NDVI and NDRE indices in subtling variations related to canopy status or the insufficient biostimulant application rates.

Proximal hand-held monitoring devices, such as the Pressure Chamber for SWP and the MFA for biochemical indices, mirrored vegetation index trends indicating similar treatment responses in both seasons. SWP patterns were consistent with cumulative stress observed by NDVI and NDRE, although they did not predict yield differences between treatments and evidence any difference between treatments, except for the green one, which showed a generally slightly smaller water stress sensitivity in both seasons. It is crucial to highlight that because of the length of measurement operations and the fatigue of obtaining such stress values, SWP was derived from punctual sampling, relying only on four measurements for each treatment. The limited sample size affected the final results, reporting a less representative overall condition than the NDVI and NDRE indices averaged over all canopy pixels of each plot.

MFA analysis showed a gradual increase in flavonoids in leaves and clusters, indicative of berry ripening and leaf senescence, along with a slight decrease in the BRR-FRF index in 2020, while in 2021 adverse weather events resulted in a constant level of stress on vegetative parts, while clusters showed a greater increase in stress during ripening. The FLAV index measured in 2021 was higher and more stable in the canopy and almost double in the berries compared to 2020, reflecting delayed ripening as observed by increased titratable acidity. The homogeneous behavior of fluorimetric indices (BRR-FRF and FLAV) on leaves and bunches across all treatments suggests that the MFA was unable to detect inter-parcel differences, probably due to the low sample size, with 20 leaves and 20 bunches per treatment, and the limited period analyzed in both seasons. However, it is notable that the fluorimetric sensor closely tracked the physiological responses of plants to seasonal meteorological changes, similar to the NDVI and NDRE results.

In contrast to vegetative indices and other portable devices, the SPAD meter detected significant differences in chlorophyll levels in the second year, probably due to its higher sensitivity and measurement methodology, demonstrating a direct correspondence with yield results, reinforcing its role in assessing vegetative vigor related to biostimulant application in viticultural scenarios. It remains to be clarified why SPAD is more sensitive than NDVI and NDRE in detecting differences between treatments and why these indices reported an inverse situation, with the blue treatment being the only underperformer (Figure 2). As can be seen from the SPAD results (Figure 5), the green and blue treatments have significantly greater nitrogen content than the red and yellow treatments, symptoms of a broadly better condition. In general, as already mentioned for the SWP and MFA, a great deal of responsibility could be attributed to the point nature of the SPAD measurement and the consequently limited sample size (with all the limitations described above), but also to the ability of the proposed indices (NDRE and NDVI) to provide generic information related to vegetative vigor, which can detect a worse performance of the blue plots, but not sensitive enough to detect the nitrogen content estimated by the SPAD.

After a literature search, the work of Antonucci et al. (2023) [18] was the only one to apply UAS-assisted remote sensing technology for detecting the effects of biostimulants. Specifically, the objective was to evaluate whether UAS imaging was able to detect biostimulant treatment effects on tomato (Lycopersicon esculentum L., 3406 Heinz) plants under water stress conditions to determine the critical time windows in which the effects of biostimulants were significant and to assess the impact of the treatments on yield and quality parameters at harvest. The absence of effects on yield and dry matter in their study suggests that the bottleneck may not lie in the PROSAIL inversion process used to recover biophysical traits (LAI, leaf, and canopy chlorophyll content) but in the efficacy of the biostimulant itself. It is important to note that their experiment was conducted on a different crop, using glycinebetaine-based biostimulants (30%) combined with variable water inputs, which makes a direct comparison with this study less applicable, raising the hypothesis that the effects of biostimulants in the field may be hardly detectable or inconsistent with other inputs such as fertilizers or agrochemicals .

The inherent complexity in interpreting results, particularly those derived from UAS and MFA indices, prevented the research team from definitively confirming the hypothesis that UAS-derived vegetation indices could reliably detect the effects of biostimulant applications. This led to mixed and, at times, contradictory findings when compared with hand-held monitoring devices, as previously discussed. These inconsistencies may be attributed to the study’s design limitations, including a limited number of replicates and the impact of topographical variations [37], both of which likely influenced the final outcomes. In particular, the results for MFA, SPAD, and SWP were skewed in areas prone to environmental stress, where differences in plant vigor and yield were more pronounced, as evidenced by the production data presented in Figure 8b. To enhance the robustness of future research, it is essential to address these methodological constraints. Increasing the number of replicates would reduce variability and provide a more representative dataset. Additionally, controlling for topographic influences, such as slope and soil composition, would help minimize their confounding effects on plant responses to biostimulants, allowing for more accurate treatment comparisons [18]. Furthermore, extending the study across multiple growing seasons would offer a more comprehensive understanding of the temporal dynamics of biostimulant efficacy [27,28]. Incorporating additional monitoring dates within each season would also facilitate a more detailed temporal resolution, enabling researchers to capture the nuanced, evolving effects of biostimulant applications on vineyard health over time.

UAS-derived multispectral vegetation indices provide a rapid and extensive means of monitoring large vineyard areas, proving particularly valuable for identifying zones susceptible to environmental stress, which may require more detailed examination using proximal sensing tools, and prove sufficient sensitivity for nutrient assessments with more extensive study, allowing a suitable and more readily available data stream for quantifying nutrients in vineyards than hyperspectral imagery [38,39]. Despite these advantages, the current limitations of UAS in consistently distinguishing between different biostimulant treatments highlight the need for further research. Advancements in developing novel vegetation indices or refining the use of specific wavelengths could significantly enhance the precision of UAS in detecting the subtle physiological changes induced by agricultural inputs.

Such improvements are critical to confirm the role of UAS as an indispensable tool in sustainable vineyard management, particularly in assessing the impacts of biostimulants and other inputs on crop health and productivity. In this context, the ability to monitor the effects of agricultural inputs through remote sensing represents a pivotal strategy. By facilitating the accurate evaluation of specific product benefits, growers can make more informed decisions about input application. This capability also allows for the comparison of multiple biostimulant brands or compositions under varying environmental conditions, providing a robust framework for optimizing vineyard management practices. Such a strategy is not only essential for improving agricultural efficiency and sustainability but also for enhancing growers’ awareness of product efficacy. In turn, this heightened awareness can accelerate the adoption of innovative technologies and practices, fostering a more data-driven approach to viticulture. Additionally, it supports research efforts by offering a scalable and efficient method for evaluating agricultural treatments, ultimately contributing to more rapid advancements in the field.

5. Conclusions

The combined use of UAS imagery, MFA, SWP measurements from a Scholander pressure chamber, and chlorophyll content from SPAD meter effectively monitored vegetative growth in Malvasia Bianca vineyard during 2020 and 2021. However, no differences were observed between treatments using remote or proximal sensing, indicating that the proposed indices were unable to differentiate vegetative vigor or water stress responses associated with varying biostimulant application rates. Only the SPAD meter, and to a lesser extent SWP, distinguished the treatments that received higher biostimulant doses and demonstrated a correlation with yield outcomes.

Future studies should investigate the impact of higher biostimulant application rates, alternative formulations, or extended monitoring periods to more accurately assess their influence on grapevine growth and yield. The successful integration of UAS imagery with conventional methods highlights the potential of these technologies as valuable decision-support tools for sustainable viticulture. By combining remote sensing with proximal assessments, vineyard managers can adopt a more comprehensive approach to monitoring vine status, optimizing management practices, and evaluating treatment effects on vineyard performance.