Interrelation Between Growing Conditions in Caucasus Subtropics and Actinidia deliciosa ‘Hayward’ Yield for the Sustainable Agriculture

Abstract

Kiwifruit is a high-value subtropical crop with significant nutritional and economic importance, but its cultivation faces growing challenges due to climate change, particularly in Caucasus. This study aims to study the impact of abiotic stressors such as temperature extremes, drought, and frost on the yield of the ‘Hayward’ cultivar over a 20-year period (from 2003 to 2022). Using a combination of agroclimatic data analysis, measurements of soluble solid content, and soil moisture assessments, this research identified key factors which limit kiwifruit cultivation productivity. The results revealed a high yield variability—68%, with the mean value declining by 16.6% every five years due to increasing aridity and heat stress. Extreme temperature rises of up to 30 °C caused yield losses of 79–89%, and the presence of frost led to declines of 71–94%. In addition, it is objectively proven that the vulnerability of kiwifruit is subject to climate-driven water imbalances. This highlights the need for adaptive strategy formation in the area of optimized irrigation for the sustainable cultivation of fruit in the subtropics. One of the study’s limitations was that it was organized around a single variety of kiwifruit (‘Hayward’). In view of the fact that there are significant differences in growth characteristics among kiwifruit varieties, future research should focus on overcoming this shortcoming.

1. Introduction

Actinidia deliciosa (A.Chev.) C.F. Liang and A.R. Ferguson kiwifruit has been widely cultivated in subtropical regions due to its unique biochemical composition. Kiwifruits are considered natural vitamin concentrates, containing more than 160 mg of vitamin C per 100 g—substantially higher than that of oranges, lemons, pineapples, or bananas. They are also rich in vitamin A, B-group vitamins, and various micro- and macronutrients. Additionally, kiwifruits are valued for their biologically active compounds, including the enzymes actinidin and ficin, which exhibit proteolytic activity comparable to papain. They also contain C- and P-active catechins in proportions that are highly beneficial to human health [1,2]. These fruits comprise a wide array of nutritionally important components, such as proteins, lipids, carbohydrates, vitamins, polyphenols, antioxidants, minerals, and dietary fiber. With a postharvest shelf life of up to six months and strong transportability, kiwifruit has gained prominence in international markets. The crop’s longevity—yielding fruit for up to 40 years—further enhances its industrial and economic value in meeting global food demands [3].

The problem of global climate change when ensuring sustainable kiwifruit cultivation is its disruption of the specific climatic conditions that kiwifruit requires for optimal growth, such as adequate winter chilling, moderate temperatures, and a reliable water supply. Climate change leads to warmer winters, which can reduce the chilling hours essential for bud development, while increasing the risk of spring frosts, droughts, and extreme weather events that damage vines and reduce fruit quality and yield. Additionally, shifting pest and disease pressures further threaten the health of kiwifruit orchards. These changes challenge the long-term viability and productivity of kiwifruit cultivation, necessitating adaptive strategies to maintain sustainability [4,5,6]. A comprehensive analysis of climate scenarios shows that shifting precipitation patterns, prolonged droughts, or increased winter cooling could further increase the volatility of the kiwifruit yield, while targeted adaptation strategies (e.g., irrigation optimization, breeding heat-tolerant varieties) could mitigate global negative impacts [7,8,9].

Due to its biological characteristics—including a shallow root system; large, broad leaves; and a fast-growing, liana-like morphology—kiwifruit requires substantial water input. Excessive precipitation can lead to soil waterlogging, which results in hypoxic conditions in the root zone, particularly affecting the roots and basal stem when sub-merged. According to Li et al. [10,11], the cultivar ‘Hayward’ shows lower tolerance to waterlogging stress compared to ‘KR5’, a difference that is attributed to both metabolic and morphological root responses. During the fruit development period (June to September), the transpiration demand significantly increases. Air temperatures frequently exceeding 30 °C from May through September increase water loss through evaporation. The plant’s root system is often unable to compensate for this high water consumption, resulting in greater energy demand and, ultimately, a reduction in yield. Although partial mitigation is achievable through calcium supplementation or summer pruning [12,13], such interventions are insufficient under prolonged stress conditions. Many studies claim [14,15] that the transcriptomic response in waterlogged plants is less pronounced compared to that observed under drought stress. Drought causes significant alterations across multiple metabolic pathways, with the most prominently affected including “plant hormone signal transduction”, “protein processing in the endoplasmic reticulum”, and the “mito-gen-activated protein kinase (MAPK) signaling pathway” [16].

Questions regarding the choice of physiological traits (related to water circulation) as an indicator for predicting expected plant yields still remain open [17]. So, for example, leaf thickness has been shown to correlate with environmental relative humidity and, by extension, soluble sugar concentrations [18], suggesting its potential as an indicator of plant water status. When evapotranspiration causes water loss from leaf cells at a rate faster than it can be replenished by soil moisture, a measurable reduction in leaf thickness may occur. In situations of severe or sudden water shortage, especially under high evaporative demand, leaves can become noticeably thinner within a short timeframe [19].

Plant physiological responses to drought can be categorized into passive (concentration effect) and active (osmotic adjustment) mechanisms. Water deficits reduce leaf water potential and relative water content. Sugar accumulation processes are relatively insensitive to moderate water deficits and are typically only affected when the total water potential of the leaf declines below −1 to −1.5 MPa [20]. Osmotic adjustment involves the accumulation of osmolytes such as sucrose, glucose, fructose, and proline, which reduce the osmotic potential and help maintain cell turgor. As a result, this adaptive response also manifests as increased Brix values (soluble solid content or SSC) in leaf sap, reflecting metabolic compensation under water-limited conditions. Given its sensitivity to both passive and active water stress responses, the leaf sap SSC may serve as a reliable proxy for assessing plant water status. So, for example, under water scarcity during flowering with tomato fruit setting, the soluble solid content correlated negatively with the temperature and yield [21]. Strong positive correlations were also found between the total marketable yield and Brix [22]. At the same time, the highest yields and soluble solids were achieved at low water stress levels (75% of the actual water demand) [23].

Elevated SSC values in leaves under such conditions are often reversible if moisture availability is restored either through rainfall or irrigation [24]. Given the vital role of leaves in nutrient assimilation and translocation, it is our opinion that SSC levels are closely associated with yield performance in kiwifruit. Water stress negatively affects key physiological functions, including photosynthesis and nutrient uptake, thereby impairing metabolic efficiency. Prolonged elevation of the SSC in leaf sap, as a result of such stress, may therefore correlate with reduced fruit setting and overall yield decline. Due to this, this study is based on the hypothesis that the SSC in leaf sap can serve as a physiological indicator of water stress.

Physiological assessments of water availability and the functional state of ‘Hayward’ kiwifruit under field conditions in Sochi, Russia have shown measurable changes throughout the growing season [25,26,27]. Water deficit in leaf petioles—used as indicator organs—was found to range from 4.9% in May to 19% in August. The leaf water content decreased slightly from 76% in May to 74% in August. Notably, the proportion of “bound” water declined from 57–63% in May to just 20% by late August, while the “free” water fraction increased correspondingly from 36% to 80%. These findings underscore the dynamic redistribution of water forms in leaf tissues over time and highlight the importance of continued investigations into water-related stress factors in kiwifruit cultivation.

Conventional trend-detection methodologies used in time series analysis—such as moving averages or exponential smoothing—suffer from inherent limitations. Moving averages are sensitive to the choice of time interval, while exponential smoothing presupposes a steady-state process, rendering it less effective in detecting abrupt or non-linear variations [28,29,30]. Under these conditions, the use of more complex approaches based on “time series data folding” in combination with the use of splines to form three-dimensional models of response function makes it possible to identify fundamentally new patterns [31,32].

Given the above, the aim of this study was to evaluate the impact of changing growing conditions (especially temperature and precipitation-induced stress) on kiwifruit yield, identify key limiting factors, and evaluate the possibility of using leaf sap SSC as a proxy indicator of stress to further formulate strategies for the sustainable cultivation of ‘Hayward’ in the Caucasus subtropics.

2. Materials and Methods

The field experiment was conducted at the Adler Experimental Station of the N.I. Vavilov All-Russian Institute of Plant Genetic Resources (VIR), located along the Black Sea coast of the Caucasus (43.438056, 39.909806, Figure 1).

Sampling protocol. The agrobiocenosis under study spans 5.5 hectares. The kiwifruit plantation was established in 1988 using a planting layout of 5 × 4 m and a three-tier palmette formation system. The research was focused on the commercial late-ripening cultivar ‘Hayward’. Sampling units were individual kiwifruit trees. Yield data were collected once per year, during harvesting, from 2003 to 2022. Replication and randomization: each year, 15 trees were randomly selected from each of the 5 ha orchards for yield measurement, for a total of 75 trees (there were approximately 450 trees in each 1 ha area). Each year, the procedure was repeated with different randomly selected trees.

Agroclimatic data. For such a small area, the formation of a high level of technical base and a staff of qualified specialists is very expensive. In this regard, it was more rational to use already obtained, valid data (for flight safety, proven with the advanced technical equipment employed for measurements) at a minimal cost. Agroclimatic data—including minimum, average daily, and maximum air temperatures, as well as monthly precipitation—were obtained from publicly available meteorological databases (http://www.pogodaiklimat.ru; https://rp5.ru; http://www.atlas-yakutia.ru/weather/climate_russia-III.html; and http://planet.rssi.ru). The date of last access was URL (accessed on 2 January 2022). These datasets were subsequently validated in collaboration with the Sochi branch of the Aviation Meteorological Station Civil (AMSG), a division of the Federal State Budgetary Institution “Aviamettelecom Roshydromet”, which is situated within one kilometer of the experimental site.

Obtaining data on soluble solids in leaf sap. A preliminary analysis of long-term data (see Section 3) revealed that yield metrics in 2019, 2020, and 2021 differed significantly from the average values of five-year intervals. Consequently, beginning in 2019, additional physiological studies were initiated, including measurements of soluble solid content (SSC) using refractometry. To assess soluble solids in leaf sap, a Bellingham and Stanley E-line Refractometer 44–891 was used. Leaf sap was sampled from kiwifruit plants following a systematic zigzag (Z-pattern) approach across the field to ensure representative coverage. Sampling was performed in the morning (around 11:00 AM) from the middle canopy zone. From May through September, samples were collected from ten plants during each of the three ten-day periods per month. For each time point, at least 12 leaf blades (including petioles) were harvested from vegetative shoots. Leaf sap was extracted using a garlic press. When leaf surfaces were damp, they were first blotted with a paper towel. One to two drops of sap were placed on the refractometer prism, the cover was closed, and readings were taken at the boundary between light and dark fields, as viewed through the eyepiece. The fruit yield was recorded for each bush (n = 10 per replicate) in three replicates.

Soil moisture. Since the use of the SSC as a stress proxy is not new, it was confirmatory of measurements made using another indirect method—the construction of contour lines with the same relative soil moisture. The soil moisture content was measured in 10 cm increments down to a depth of 60 cm within the root zone, using the gravimetric method described by [33]. Moist samples were weighed, dried in a 2B-151 oven (Odessa Experimental Plant of Medical Equipment, Ukraine), and reweighed using VSLT300/3A Pioneer electronic laboratory scales (Ohaus Instruments, Shanghai, China). The gravimetric water content was calculated as the difference between the wet and dry mass, divided by the dry mass. The soil’s field capacity (FC) was 32%. The optimal growing conditions for the kiwifruit were determined to be at 80% of the FC (corresponding to 26% soil moisture), while conditions below 65% FC (approximately 22% soil moisture) [19] were considered suboptimal for root nutrition and water uptake. In the Section 3, soil moisture data will be expressed as $K_{FC}$ (%)—the ratio of measured moisture to the field capacity.

Processing of measurement results. The validated data were summarized and formatted in Microsoft Excel. Standard statistical methods were applied to compute the sample mean ($\mu$), standard deviation of the sample data (SD), and coefficient of variation (CV) values:

$$\mu ={\displaystyle \frac{1}{H}}\sum _{i=1}^H{x}_i,$$

(1)

where: μ—sample mean; H—number of observations;

$$\sigma =\sqrt{{\displaystyle \frac{1}{H-1}}\sum _{i=1}^H{\left(x_i-\mu \right)}^2},$$

(2)

where: σ—standard deviation (applying Bessel’s correction);

$$CV={\displaystyle \frac{\sigma }{\mu }},$$

(3)

where: CV—coefficient of variation.

To visualize the response function—illustrating the dynamics of soil moisture reserves and sap concentration in relation to time and weather—a three-dimensional surface plot was generated using Gnuplot v5.4 with spline interpolation, following approaches similar to those described in prior studies [34,35,36]. The algorithm used for working with the data was an alternation of the following sequence of actions: in Vi Improved, a Python 3.11.0 script was written to filter the primary experimental data using the Savitzky–Golay method (https://docs.scipy.org/doc/scipy-0.16.1/reference/generated/scipy.signal.savgol_filter.html, accessed on 6 July 2025). Next, in gnuplot, a response surface was built for the process under study. Several operators implemented a standard b-spline interpolation procedure in the “gnuplot” program. To select a color, we used the “set palette gray” operator. The first is «set hidden3d» (The “set hidden3d” is a command that enables hidden line removal for surface plotting (see splot). The second is «set pm3d at bs» (pm3d is a splot style for drawing palette-mapped 3d and 4d data as color/gray maps and surfaces. It allows the plotting of gridded or nongridded data without preprocessing. The pm3d style options also influence the solid-fill polygons used to construct other 3D plot elements). The third is «set dgrid3d…» (the set dgrid3d command enables and can set the parameters for nongrid to grid data mapping). The fourth is the set view options and xrange, yrange, and zrange. The fifth is «splot “./kiwifrut_1.txt”»” [35].

3. Results

The warm, humid subtropical climate along the Black Sea coast promotes the early start of vegetative growth in thermophilic crops, including Actinidia. Growth usually begins in March, preceded by bud differentiation from December to February, followed by bud swelling and shoot emergence. However, periodically occurring spring frosts pose a significant threat to young shoots. For instance, in 2004, frosts events reached −5.0 °C on April, while in 2014, frosts of −1.8 °C to −2.4 °C persisted from March to April. As a late-season variety, ‘Hayward’ generally begins growth 4–5 days later than early- and mid-ripening varieties. Despite this delay, exposure to frost conditions during critical phenophases may result in yield reduction ranging from 6% to 26%, depending on the frost’s severity and duration. An analysis of agroclimatic factors’ influence in the Caucasus humid subtropics on kiwifruit yield during the study period are presented in

Table 1. Mean kiwifruit yield in the humid subtropics of the Caucasus.

Year	Yield (Y), kg/Plant	Stressors	Year	Yield (Y), kg/Plant	Stressors
2003	42 ± 4.1	-	2013	42.6 ± 5.0	-
2004	6.2 ± 0.7	Frosts from −0.6 to −5 °C in April	2014	1.3 ± 0.1	Frosts from −1.8 to −2.3 °C (March–May)
2005	29.5 ± 3.1	-	2015	35.2 ± 4.1	-
2006	32.6 ± 3.7	-	2016	7.3 ± 1.1	-
2007	28.7 ± 3.0	-	2017	8.6 ± 1.0	Heat above 30 °C (July–August)
2008	29.9 ± 3.0	-	2018	23.4 ± 2.6
2009	2.4 ± 0.3	Heat above 30 °C in May	2019	4.6 ± 0.5	Heat above 30 °C in May
2010	31.2 ± 3.6	-	2020	8.7 ± 0.8	Heat above 30 °C (June–September)
2011	21 ± 1.8	-	2021	23 ± 2.1	-
2012	35 ± 3.7	-	2022	21.4 ± 1.9	-

.

Analysis of these data suggests that the potentially achievable yield for this cultivar can be 42.6 kg/plant; under adverse environmental conditions, its levels declined significantly—ranging between 1.3 and 8.6 kg/plant. Across the 20-year period, the overall mean yield was 21.7 ± 13.5 kg/plant and the coefficient of variation 62%. For the unfavorable years group (2004, 2009, 2014, 2017, 2019, and 2020), the mean yield was only 5.3 ± 3.1 kg/plant or 4.1-times less than the 20-year mean. Kiwifruit in humid subtropical conditions usually bloom in the second ten days of May, and depending on the temperature intensity, pollination and fruit setting occur during this period. Temperatures above 30 °C at this stage of development significantly reduce pollen fertility [37]. As a result, plant productivity can sharply decrease [19]. For example, in 2009 and 2019, the ‘Hayward’ yield variety productivity was 2.4 and 4.6 kg/plant, respectively. Data presented in Table 1 indicate low yields in 2017, 2019, and 2020 (due to the extreme heat), amounting to only 4.6–8.7 kg/plant, approximately, which is 21–40% of the long-term average yield. The extended development cycle of kiwifruits, combined with their sensitivity to fluctuating weather conditions during peak growth (June–August), contributes to this variability. At the same time, significant fluctuations in precipitation are observed during this period.

The regional climate is classified as humid subtropical with Mediterranean characteristics, with warm, dry summers and cooler, wet winters. The humid subtropics of Russia are characterized by the presence of dry periods. The frequency of 10-day droughts from May to September is 80, 69, 83, 87, and 78%, respectively. In July and August, droughts lasting up to 20 days occur with a frequency of 36% and 30%, respectively. In August, droughts lasting up to 40 days have been observed with a frequency of 7%, and the moisture deficit is aggravated by a high radiation intensity. During the summer months (June to August), the total solar radiation is approximately 17.7, 18.0, and 15.9 kcal/cm². These conditions contribute to excessive evapotranspiration, increasing the effects of heat and moisture stress on plant productivity.

Table 2. Average data of mean monthly air temperatures and precipitations over a ten-year period.

Month	Mean Temperature (°C)		Mean Precipitation (mm/month)
	2001–2010	2011–2020	2001–2010	2011–2020
June	20.7 ± 0.4	21.9 ± 0.4	124 ± 22.7	96 ± 15
July	29.7 ± 0.4	23.9 ± 0.3	100 ± 15	116 ± 12
August	24.7 ± 05	24.7 ± 0.3	94 ± 19	63 ± 18
September	20.7 ± 0.3	21.2 ± 0.7	139 ± 19	155 ± 50

presents the average data on the mean monthly air temperature and precipitation for the period from 2000 to 2022.

Analysis of the results in

Table 3. Tabular arrangement of years at the “time series folding” for the period under study (years with stressors factor influence are shown in yellow).

m	N
	1	2	3	4
1	2003	2008	2013	2018
2	2004	2009	2014	2019
3	2005	2010	2015	2020
4	2006	2011	2016	2021
5	2007	2012	2017	2022

indicates a pronounced variability in moisture supply, especially in August and June, with decreases of −33% and −23% (94 to 63 and 124 to 96), while the mean monthly air temperature either remained unchanged or increased by 6% (20.7 to 21.9). Such changes affect not only the evapotranspiration rate, but also the soil water availability, influencing plant physiological responses.

Further research involved the use of “time series folding” [31], where a continuous long-term data series is divided into five-year sections and analyzed together to produce a three-dimensional dataset (Table 3 and

Table 4. Tabular arrangement of mean yield indicator at the “time series folding” for the period under study.

m	N				μ Y (mj)
	1	2	3	4
1	42	29.9	42.6	23.4	34.5 ± 8.2
2	6.2	2.4	1.3	4.6	3.6 ± 1.9
3	29.5	31.2	35.2	8.7	26.2 ± 10.3
4	32.6	21	7.3	23	21.0 ± 9.0
5	28.7	35	8.6	21.4	23.4 ± 9.8
μ Y (Ni)	27.8 ± 13.2	23.9 ± 13.1	19.0 ± 18.6	16.2 ± 8.9

). The number of five-year intervals, segments or “columns” is designated as “N”, and every 1st, 2nd…5th year inside each five-year interval or “line” is designated as “m”. Table 4 also shows the values of the mean yield in each five-year interval “μ Y (N_i)” (where the i$\in$ [1;4] and mean yield every 1st, 2nd…5th year is “μ Y (m_j)” and where j$\in$ [1;5]).

A graphical representation of the results of Table 4 is shown in Figure 2.

In the trend analysis of Figure 2, a downward trajectory downward trajectory in mean yields across four consecutive five-year intervals is revealed, probably due to broader climate shifts. Specifically, between intervals N 1 and 2, yields declined by 14%; between intervals N 2 and 3, by 21%; and between intervals N 3 and 4, by 15%. For a more detailed study of agroclimatic factors’ influence, it is more representative to consider the 2nd, 3rd, and 4th years within the five-year intervals, in which the 2nd years reflect only the consistent influence of stress factors, the influence of stress factors was present only in the last five-year period (3rd years), and the were no frosts or extreme heat in any of five-year intervals (4th years). In addition, if we exclude extreme values (1st and 5th years), the projection of the three-dimensional model (Figure 3) of the research processes will allow us to establish a number of important patterns.

As shown in Figure 3, fruiting exhibited a cyclical pattern across five-year intervals that was not changed from previously using two-dimensional models (Figure 2). Red zones highlight the local minima, which typically occur in the second year of each five-year period, while blue zones denote the conditional maxima. The 3D model enables a more detailed interpretation of yield trends by simultaneously capturing temporal (yearly) and interval-based (five-year) variations. In contrast to three-dimensional representations, this approach allows for a comprehensive spatial–temporal analysis. Notably, in the final interval (2018–2022), there was a marked expansion and temporal shift in the local yield minimum, which moved from year m = 2 (2019) to year m = 3 (2020), followed by a sharp yield increase in year m = 4 (2021). This anomaly is illustrated by the oval highlight in Figure 3 and red oval in Figure 2. These patterns underscore the sensitivity of kiwifruit yield to growing conditions and the growing impact of climate change.

At the next stage, the mutual influence of the mean monthly temperature and amount precipitation (Figure 4) during the growing period was compared for the following cases: only with a consistent influence of stress factors (2nd years); the influence stressors only being in the last five-year period (3rd years); and in the absence of stressors (4th years).

From the results analysis in Table 2, it follows that for the first case (2019), yields were higher by 27% than the mean yield every 2nd year “μ Y (m₂)” (4.6 kg/plant against 3.6 kg/plant), while from Figure 4, it follows that this year is characterized by favorable agroclimatic conditions (precipitation dynamics curve approaches the black line). Moreover, the curve of the SSC change dynamics seems to be approaching the “typical trend curve” for growing kiwifruit in these conditions—resembling a parabola branch. In contrast, for the second case during 2020, the yield was 67% lower than the mean yield every 3rd year “μ Y (m₃)” (8.7 kg/plant against 26.2 kg/plant), while from Figure 4, it follows that this is a climatically adverse year (precipitation dynamics curve is very low in comparison to the black line). Reaching the maximum average monthly air temperature, in combination with critically low precipitation levels, created a critical stress that caused a sharp and atypical increase in soluble solid content to the level of boundary values in the conditionally optimal 2019 year. For the third case during 2021, the yield was +10%, which can be considered close to the mean yield every 4rd year “μ Y (m₄)” (23 kg/plant against 21 kg/plant), while from Figure 4, it follows that there was an overabundance of soil moisture in these conditions—the precipitation dynamics curve exceeded by more than two times the black line. At the same time, there was a small increase in the soluble solid content to the level of values in a conditionally optimal 2019 year (the SSC trend curve completely coincides with the 2019 curve, but is slightly shifted upwards).

Since the use of the refractometric method as a stress proxy is not new, already validated in grapes and olives, it was necessary to carry out some confirmatory measurements by another indirect method. In addition, given the observed inter-annual differences in SSC trends between 2019 and 2020, and significant variations in precipitation, an additional analysis was conducted to determine whether SSC levels are directly supported by soil moisture within the root zone (0–60 cm). The standardized gravimetric method was used to quantify the soil water content. The measured soil moisture levels were then expressed as a percentage of field capacity (Figure 5) using the formula:

$$K_{FC}={\displaystyle \frac{u}{FC}}\times 100,$$

(4)

where: u—soil moisture; FC—field capacity (corresponding to 32% soil moisture).

As determined nin previous studies, a lower than 70% FC value characterizes the presence of stress from a lack of moisture, and a more than 95% FC characterizes the presence of stress from over-humidification. An analysis of the isolines in Figure 5 suggests that general trends in relative soil moisture profiles corresponded closely with SSC dynamics in 2019 and 2021. This association appears more consistent and convincing in 2020, which is characterized by stress from the presence of persistent heat from June to September, in combination with more than two-times less precipitation than the mean value (μ P (2011–2020). In the picture of deep profile changes, in August, an extended area (from 25 cm to 60 cm) with relative soil moisture of less than 65% of the FC was traced, which indicates an insufficient water supply to most of the root zone.

4. Discussion

Physiological changes in leaf water content from May to October were found to be influenced by both the moisture arability and plant developmental stage. Limited variation in SSC values, ranging between 4.4% to 5.0% in early summer, reflects the relatively stable leaf water content during this period. However, from July to August, the leaf water content begins to increase from 5.4% to 8.5% by late summer, coinciding with intense fruit growth and increased physiological needs of the plants. From August to September, the SSC level displaced a sharp and variable increase, ranging from 6.8% to 11.3%. This trend is attributed to a combination of abiotic factors, in particular, elevated temperatures and decreasing soil moisture, causing physiological regulation mechanisms inherent in the plant. Kiwifruit actively redistributes water and assimilates it from leaves to fruits during this phase, which is critical for fruit filling. The timing and extent of this redistribution are closely governed by prevailing weather conditions. The present findings highlight the vulnerability of ‘Hayward’ to such environmental variables, consistent with results from other kiwifruit-producing regions, including China, New Zealand, and Turkey [38]. Long-term observations conducted under humid subtropical conditions confirm that the yield of kiwifruit is highly sensitive to abiotic stress during key developmental stages [39]. As established in Section 3, recent climate trends include:

(a)

A notable increase in average air temperatures during the growing season over the past 22 years;

(b)

High interannual variability in moisture supply, particularly in August and September; and

(c)

Frequent drought events, including 10-day dry spells in 69–87% of cases between May and September, 20-day droughts in July–August (30–36%), and even 40-day droughts in August (7%). These shifts indicate an ongoing process of aridification, which is contributing to a growing water deficit.

Under warmer and drier conditions, questions about the impact of biotic factors are becoming more relevant. At the same time, the ecological consequences of global climate change significantly affect (causing changes in the nutritional status of plants) the complex interactions between plants and pests in natural ecosystems [40,41]. Certain pests (for example, Halyomorpha halys) may be more prevalent in the humid subtropics of the Caucasus, which may pose a significant threat to kiwifruit development, growth, and yield.

Aridification has caused a water resources crisis; the high water requirements of kiwifruit during the growing season highlight the need to develop more rapid and accessible methods (with subsequent refinement by standard indicators of water stress) to maintain optimal soil moisture levels [42]. High temperatures are known to have been shown to disrupt nutrient uptake. For example, Stefaniak [20] reported that temperature and precipitation jointly influence the rate of nutrient accumulation in plant tissues. Such changes in nutrient availability may further exacerbate yield loss in Actinidia deliciosa. During the flowering period—typically in the second half of May—temperatures above 30 °C have been found to significantly reduce pollen viability [36], causing a sharp decline in fertilization efficiency and fruit setting [43]. This phenomenon was evident in 2009 and 2019, when average yields dropped to 2.4 and 4.6 kg/plant, respectively.

Such changes in nutrient availability may further exacerbate yield losses. The combined effect of precipitation and air temperature [36,44] is very important for growing subtropical crops, especially in late March–early April. Temperatures above 30 °C have been found to significantly reduce pollen viability [36], causing a sharp decrease in fertilization efficiency and fruit setting [43]. This phenomenon was evident in 2009 and 2019, when yields fell by 89% and 79% to 2.4 kg/plant and 4.6 kg/plant, compared to the 21.7 kg/plant 20-year period mean yield.

While some fruit types can withstand frosts down to −25 °C [45], ‘Hayward’ is particularly vulnerable (its floral organs are damaged) at temperatures between −1 and −1.5 °C. Just like in the Caucasus, climate change in the north-eastern regions of Europe has led to an increase in spring frosts [46]. For example, in New Zealand [47], spring frosts are considered among the most damaging climatic events, capable of eliminating the entre season’s harvest [47,48]. However, for the studied Adler conditions, late-ripening ‘Hayward’ cultivar was able to maintain yields in 2009 and 2019 (in varying intensity conditions of frost) of between 1.3 to 6.2 kg/plant, decreased by only 94% and 71% compared to the 21.7 kg/plant 20-year period mean yield. Based on the analysis of projection of a three-dimensional response surface of the relative soil moisture level dynamics from the soil profiles’ depth, the optimal parameters for growing kiwifruit were determined in the Caucasus subtropical zone.

5. Conclusions

Yield loss of kiwifruit in Caucasus subtropical conditions is primarily driven by water stress caused by uneven precipitation (drought under conditions of maximum average monthly temperatures in August–September, as well as waterlogging). Insufficient levels can be indirectly diagnosed through increases in the soluble solid content of leaf sap. Maintaining optimal soil moisture is critical to minimizing yield variability, as failure to do so results in a sharp yield decrease of 4.1 times.

The main findings of this study indicate that:

-

The maximum ‘Hayward yield potential is 42.6 kg/plant, but under adverse environmental conditions, it levels significantly to between 1.3 and 8.6 kg/plant;

-

Yield variability over 20 years was exceptionally high (62% coefficient of variation), with mean values of 21.7 ± 13.5 kg/plant across the 20-year period;

-

Agroclimate change trends are contributing to a long-term decline in mean yield in five-years intervals by 14%, 21%, and 15% (or by an average of 16.6% between each 5-year interval for 20 years);

-

This region experiences frequent drought events (e.g., 10-day droughts occur 69–87% of the time May–September);

-

The consequent presence of high temperatures (>30 °C) severely reduces pollen viability and fruit, which causes yields to fall to 79–89% compared to the mean values of the 20-year period;

-

In frost conditions of varying degrees of intensity, the kiwifruit yield fell to 71–94% compared to the mean 20-year period values.

One of the limitations of this study is that it was organized around a single kiwifruit variety (‘Hayward’), and did not address comparisons with other kiwifruit varieties under the same climatic conditions. In view of the fact that there are significant differences in growth characteristics, tolerance to environmental stresses, and yield and quality performance between kiwifruit varieties, future research should focus on overcoming this shortcoming. It should be noted that the next significant limitation of study is the absence of any human intervention in those years when the parameters of climatic conditions were extreme. This was done in order to assess real losses in kiwifruit yields for the purpose of the further economic justification, the feasibility of capital investments in the implementation of irrigation and modern greenhouse complexes.