Physical Ripening Indices Improve the Assessment of Mechanical Harvesting Time for Olive Cultivars Resistant to Xylella fastidiosa subsp. pauca

Abstract

Xylella fastidiosa subsp. pauca (Xfp) is a significant threat to Mediterranean agriculture, particularly impacting olive trees in southern Italy, causing Olive Quick Decline Syndrome. Resistant olive cultivars, such as ‘Leccino’ and ‘Fs-17’, have been identified as alternatives to restore the oliviculture within the infected areas. ‘Frantoio’ and ‘Cipressino’ are included in ongoing studies on genetic resistance to Xfp. The mechanization of olive harvesting is essential for reducing production costs in the olive oil sector. Two systems, trunk shakers and over-the-row machines, are used depending on the tree density and canopy structure, with super-high-density systems offering advantages in terms of cost and efficiency. This study investigates the feasibility of using simple and non-destructive indices to assess the optimal mechanical harvesting time. Different physical ripening indices, including detachment force, fresh weight, pigmentation, and firmness, were measured on four olive cultivars (‘Fs-17’, ‘Leccino’, ‘Frantoio’, ‘Cipressino’) in southern Italy over two years. The study found that the pigmentation index had a strong relationship with the detachment index, particularly for ‘Fs-17’, and ‘Leccino’, providing a reliable non-destructive measure for optimal harvesting time. The results indicate that the optimal harvesting times for mechanical harvesting are early September for ‘Cipressino’, early October for ‘Fs-17’, and mid-October for ‘Frantoio’ and ‘Leccino’.

1. Introduction

Xylella fastidiosa subsp. pauca (Xfp) has been demonstrated to be a polyphagous bacterium that is seriously threatening the Mediterranean agriculture [1]. Found on olive trees in southern Italy in the 2010s, Xfp is a quarantine bacterium that causes OQDS (Olive Quick Decline Syndrome), leading to leaf scorching, branch wilting, and, lastly, tree death [2]. Olive cultivation is historically a key element of southern Italian agriculture, especially in Apulia, where centuries-old olive groves are an integral part of the region’s landscape, culture, and economy [3]. The impact of Xfp is particularly severe because it affects an area where fruit tree crops are already under significant stress due to climate change, with irrigated crops facing growing difficulties due to water scarcity and rising temperatures [4,5,6]. To overcome this severe economic crisis due to Xfp infections in infected areas, it is now only possible to plant host olive trees with resistant cultivars: ‘Leccino’ and ‘Fs-17’ [1,7]. More recently, ‘Lecciana’ and ‘Leccio del Corno’ were added to this list [8]. However, new olive orchards with resistant cultivars should be accompanied by appropriate sustainable agricultural practices [7].

The mechanization of olive fruits’ harvest is a key aspect of reducing the high costs of production which characterize the olive oil sector [9]. Furthermore, it results in several benefits, for instance, facilitating early harvesting and then lowering the risk of insect damage [10]. Now, olive harvesting could be mechanized by trunk shakers and over-the-row machines, depending on the cropping system: intensive or super-intensive, respectively. The former works well with tree densities of 400–600 trees ha⁻¹ and lower vase-shaped canopies [11]. The latter, known as super high density (SHD), performs with tree density of 1200–2000 trees ha⁻¹ and trees in hedgerows to form continuous fruit-bearing canopies [12,13]. Compared to intensive orchards, SHD cropping systems represent a valuable option not only for olive orchard profitability but also for all other fruit crops [14,15] through a strong reduction in production costs, thanks to full mechanization from planting to harvesting, and higher constant crop levels [12,16,17]. By employing trunk shakers, one hectare of an intensive olive grove is harvested within 2–5 days [18,19], while one hectare of a super intensive olive orchard is harvested within just 2–3 h using continuous harvesters [12]. Indeed, the speed of current straddle-harvesting machines could go from 1 to 2 km per hour, covering an area of 4–5 ha per day [16,18]. Consequently, as the harvest is concentrated in a short period, it is crucial to know the optimal harvesting time of each cultivar for this cropping system to optimize the mechanical harvesting efficiency and yield harvestable crops [12,20,21,22]. In any case, from cultivar to cultivar, choosing the correct harvesting time is mandatory to extract the highest amount of olive oil with the best quality [23,24,25,26].

To reach this goal, ripening indices are a valuable instrument to understand the most significant variations during the fruit ripening process. On this topic, different trials have been already carried out on ‘Fs-17’ and ‘Leccino’, with satisfying results [27,28,29,30]. For the cultivar ‘Lecciana’ ripening process, times and indices have already been assessed [31]. Maximum mechanical efficiency is the main criterion for choosing the harvesting time [32]. Physical ripening indexes, such as the detachment index and pigmentation index, are relevant for this purpose. However, they are often overlooked and little studied compared to chemical indices like oil and polyphenol content, which are more expensive and not easy to detect [33,34]. Indeed, a good ripening index should be related to ripening, simple and easily detectable, objective, and mostly economic [35].

In olive growing, the efficiency of mechanical harvesting changes in time, especially as a function of the detachment index (DI), defined as the ratio between detachment force and fresh weight [36,37]. This parameter depends on the cultivar and stage of maturity [38]. It assumes optimal values equal to less than 2.0 N g⁻¹ using trunk shakers [39,40] and equal to less than 2.7 N g⁻¹ using continuous harvesters [31,41]. On the other hand, if the DI is too low, an increment in the fruit drop can be observed, leading to fruit losses [42]. Moreover, cultivar and harvesting time can be the driving force in mechanical harvesting by improving its efficiency and reducing the broken or damaged axis [43]. However, the DI is a destructive, time-consuming ripening index, so it would be very useful to replace it with a more direct, simple index. For instance, Camposeo et al. [41] built a model, based on sigmoid regression, to find out the DI from the quick colorimetric index (CI). According to this model, the DI threshold value of 2.7 N g⁻¹ corresponds to a CI value of 12.5. However, the sigmoid model is applicable only for cultivars with a shorter maturation period; in the case of cultivars that show long maturation periods, this model does not fit, requiring the traditional destructive indices, DI and firmness. The pigmentation index (PI), or the Jaen index, is a simple, non-destructive index widely used in olive growing to determine the exact harvesting moment [44,45]. In the past, the Jaen index has already been related to other variables, including atmospheric temperatures, to predict the best moment for harvesting [46]. The study also considered the relationship between ripening stages and accumulated GDD to determine optimal harvesting periods.

Finally, the olive cultivars resistant to Xfp show different levels of suitability to the modern planting system, both intensive and super-intensive. ‘Fs-17’ (=Favolosa^®) is a new medium–low-vigor cultivar that has a certain level of suitability in SHD orchards [47,48,49]. On the contrary, ‘Leccino’, a traditional medium–high-vigor cultivar, is prone to only being planted in intensive orchards [50]. ‘Frantoio’ and ‘Cipressino’ are conventional high-vigor cultivars: the first one is widespread in all olive-growing regions and characterized by excellent oil quality [51]; the latter is present mainly in southern Italy and in other regions for oil production, but it is also used as a living windbreak [52]. Both ‘Frantoio’ and ‘Cipressino’ are included in ongoing studies on genetic resistance to Xfp [53], but physical ripening indices’ relationships to improve mechanical harvesting management have not yet been studied.

This two-year field research is aimed to investigate (i) the variations in the physical ripening indices during the fruit ripening process and (ii) the relationship between the DI and PI, in order to assess the mechanical harvesting time for four olive cultivars: ‘Fs-17’, ‘Leccino’, ‘Frantoio’, and ‘Cipressino’.

2. Materials and Methods

2.1. Experimental Design

A two-year field experiment was conducted in an olive orchard situated in Cassano delle Murge, located in Apulia, southern Italy (coordinates: 40°54′ N, 16°4′ E, 307 m above sea level). The orchard is established on clay–loam soil, is non-calcareous, and is classified as Haploxeralf–Xerothent according to USDA standards or Luvisols–Phaeozems based on the FAO classification [54]. The area experiences a Mediterranean climate, with an average annual rainfall of 560 mm, primarily occurring from autumn to winter, and an average annual temperature of 15.6 °C [55]. The olive trees, planted in 2002, are trained with a central leader system and spaced 4.0 m by 1.5 m (1667 trees per hectare) in a north–south row orientation (Figure 1). Drip irrigation was scheduled every 3.5 days on average, with irrigation volumes of 300 m³/ha in the first year and 500 m³/ha in the second year. Each year, the orchard received nutrient applications of 100 units of nitrogen (N), 80 units of phosphorus (P), and 60 units of potassium (K). This research focused on four cultivars: ‘Fs-17’, ‘Leccino’, ‘Frantoio’, and ‘Cipressino’. All the trees were bearing, being in their 15th and 16th year after planting (YAP). For each cultivar, a randomized block design with 3 replications was adopted; each replication included a row of 50 trees; within each row, 3 trees were used for the experimental measurements (Figure 1). Fruits were sampled when olives reached 90% of the final size, corresponding to code 79 of the BBCH phenological scale [56].

2.2. Physical Ripening Indices

For each sampling date and each cultivar, 100 olives were used for determining the following physical ripening indices: detachment force (DF; N) measured with a dynamometer (mod. Somfytec, Figure 2); fresh weight (FW; g); and detachment index (DI; N g⁻¹), calculated as (1) follows:

DI = FW/DF

(1)

Fruit color was determined both as a pigmentation index (PI) and as a colorimetric index (CI). PI was calculated by Equation (2) according to Camposeo et al. [41], as follows:

$$\mathrm{P}\mathrm{I}=\sum _{\mathrm{i}=0}^5(\mathrm{i}\times \mathrm{n}\mathrm{i})/\mathrm{N}$$

(2)

In this context, ‘i’ refers to the group number, ‘ni’ denotes the number of fruits in each group, and ‘N’ represents the total number of fruits in the sample. The method involved categorizing the olives into six distinct groups based on specific traits: group 0 for fruits with green skin; group 1 for fruits with less than 50% black skin and white flesh; group 2 for those with 50% or more black skin and white flesh; group 3 for fruits with 100% black skin and white flesh; group 4 for fruits with 100% black skin and less than 50% purple flesh; and group 5 for fruits with 100% black skin and 50% or more purple flesh (with a PI ranging from 0 to 5). The color index (CI) was measured on both sides of the fruit’s equatorial region using a tristimulus colorimeter (model CR300, Minolta Co., Ltd., Osaka, Japan) with an 8 mm aperture, diffused lighting, and a viewing angle of 0°. The CI was determined using Equation (3), which has been shown to be closely related to the olive degreening process [57] and is as follows:

CI = L(b − a)/100

(3)

where ‘L’, ‘a’, and ‘b’ are the color–space output measures. Lastly, a penetrometer (mod. TR) was used to measure fruit firmness (FF; N), using a Ø 2 mm tip on the equatorial zone.

2.3. Statistical Analysis

Data were analyzed by one-way analysis of variance (ANOVA) followed by post hoc testing (SNK-protected test) to separate means; standard error (SE) was also calculated. Interaction between each physical ripening index and cultivar and year was determined through two-way ANOVA. Further statistical analyses were carried out to evaluate the relationship between DI and PI. The relationship between DI and PI was analyzed using an inverse function (Equation (4)) in two ways: first, individually for each cultivar, and then as a whole using the aggregated data from all four cultivars. This analysis was performed using the first-year data. The inverse function is expressed as follows:

y = y₀ + (a/x)

(4)

where y = DI and x = PI. This inverse function model was chosen to capture the non-linear relationship observed between DI and PI, where DI decreases rapidly at low PI values and then approaches an asymptote as PI increases. The goodness of fit was assessed using the coefficient of determination (R²) and its associated p-value, calculated as follows:

$${\mathrm{R}}^2=1-\left({\displaystyle \frac{\mathrm{S}{\mathrm{S}}_{\mathrm{r}\mathrm{e}\mathrm{s}}}{\mathrm{S}{\mathrm{S}}_{\mathrm{t}\mathrm{o}\mathrm{t}}}}\right)$$

(5)

where SSres is the sum of the squares of the residuals, and SStot is the total sum of the squares. The second-year data were used to validate the model. The significant levels used for all the statistical analyses were * p < 0.05, ** p < 0.01, and *** p < 0.001. All the statistical analyses were carried out in R environment using the RStudio IDE (Version 2024.04.2 + 764 for Windows).

2.4. Climatic Pattern

Rainfall and temperature were tracked monthly throughout both years of the study (as shown in Figure 3). Agro-climatic data were provided by the Research Unit for Climatology and Meteorology Applied to Agriculture [54] and were recorded at a weather station located just a few kilometers from the experimental site. In 2017, the average monthly temperatures during the ripening period were over 1 °C lower, and the monthly rainfall was more than 50 mm higher compared to 2016. As a result, fruit sampling in the second year was delayed by approximately one week, occurring on 22 September compared to 14 September the previous year.

3. Results and Discussion

These are the first completed data on the physical ripening process of these cultivars, making it difficult to discuss them.

3.1. ‘Fs-17’

Indices trends were different for ‘Fs-17’ across the two years of the experiment. During the first year, FF decreased steadily, going from 7.9 N in the last decade of September to 2.9 N at the beginning of November (Figure 4). During the second year, FF observed a lower decrease after 30 days from the beginning of the sampling. The FF could be influenced by different variables: Diarte [58] observed that ‘Arbequina’ showed different firmness related to irrigation. Moreover, this parameter is highly relevant in some cultivars that are more susceptible to bruising, such as ‘Manzanilla de Sevilla’, compared to ‘Manzanilla Cacereña’ when they are cultivated in super-high-density orchards [24,59].

In the first year, DI values lower than 2.7 N g⁻¹ were recorded at the beginning of the sampling time (middle of September, 2.4 N g⁻¹), and then, they slowed down until the 45th day (end of October, 0.9 N g⁻¹). As for F, the DI had a different trend in the second year; it started at higher values, with 3.5 N g⁻¹, falling below 2 after 30 days (1.9 N g⁻¹). The first-year PI was 0.8 at the beginning of maturation, when the epicarp, starting from the calyx area of the fruit, started to turn soft pink, then red, and finally light violet (PI = 2.8 on the 45th day). During the second year, the PI showed a slower increase with time, exceeding 2 after only 45 days. The values remained identical until ripening. The CI in the first year decreased drastically in the first 15 days, going from 15.4 to 5.7, and then remaining almost constant. At the beginning of November, the fruits were overripe and drops were abundant, for which the maturation indices were measured up to that date. The second-year CI fell later than the first year; in fact, a significant difference was observed between 15 and 30 days of maturation, reaching 1 after 60 days.

3.2. ‘Leccino’

During the first year, FF decreased continuously from the first sample (middle of September, 10.5 N) to the 45th day (early November, 3.6 N), and then, it did not show significant differences (Figure 5). ‘Leccino’ exhibited a relatively consistent and gradual ripening pattern, with clear differences in behavior between the two years of observation. These data are also confirmed by previous research, with a firmness that went from 7.2 to 2.9 N before harvesting [60,61]. In the second year, the softening was more linear, with FF continuing to decline steadily until the 45th day, after which, it remained stable at lower values. This suggests that the softening process for Leccino can vary slightly between years, potentially due to environmental conditions, although the overall pattern remains predictable. The DI trend followed the FF trend in both years. The DI in the first year sharply decreased under the recommended threshold by the 45th day, while in the second year, the DI observed a fall in the first 3 samples (3 to 1.7 N g⁻¹), showing no statistical differences afterward. Similar results were also obtained in central and northern Italy, in which ‘Leccino’ showed a DI around 2 N g⁻¹ at harvesting [39,60,61]. The PI increased from 1.4 to about 4 by the 60th day in the first year, with a similar trend during the second year, indicating consistent pigment change [60]. The CI decreased from 5.1 to about 0 after 15 days in the first year of the experiment, while in the second one, it decreased from 13 to 0 after 30 days, remaining stable, reflecting a sharper decline in color change in the second year.

Different trends were observed during the second year. In contrast, PI showed the same trend as in 2016.

3.3. ‘Frantoio’

During the first year of the experiment, FF decreased very quickly after 15 days, going from 10.4 N to 5.1 N, and then slowly continued to decrease to about 3 N in early December (Figure 6). FF’s trend was reflected in the second year, with delayed ripening, probably due to higher precipitation. This behavior is confirmed by the literature, in which firmness values fell from 8 N in DOY 260 to 2.1 N in DOY 310 [58]. During the first year, the DI decreased more slowly, reaching values lower than 2.7 N g⁻¹ after 15 days of maturation (early October); then, the DI remained steady, at values of about 2 N g⁻¹ until the 45th day (early November). In the second year of the experiment, the DI continued to decrease to 0.9 N g⁻¹ by the last sample day, confirming the trends in previous work [60,62]. ’Frantoio’ had a prolonged ripening period, with the FF, DI, PI, and CI showing clear, progressive changes. Moreover, Farinelli et al. [39] observed a linear relationship between mechanical harvesting yield and DI, with a value for ‘Frantoio’ lower than 2.3 N g⁻¹, which ensured the harvesting. The PI was 0.4 at the beginning of maturation when the fruit epicarp was green. After the initial 15 days of maturation, the fruit’s skin began to change color, starting at the calyx area. The progression followed a sequence from pale yellow-green to red, then purple, and ultimately blue-black. By the 75th day, the pigmentation index (PI) had reached 3.6. During the second year, it showed a more linear increase, exceeding 4 after 60 days, suggesting a gradual color change. These data were in part confirmed by the literature, with a PI of 4.2 in the latest harvesting time [60], while, they in part differed from previous research, in which, at harvesting time, olives showed a PI of 1 [62]. The CI decreased steadily in both years, reflecting consistent color change patterns.

3.4. ‘Cipressino’

FF showed similar behavior during the two years, except for on the 45th day of maturation, when F was lower during the second year than the second (2.6 and 3.3 N, respectively) (Figure 7). During the first year, the DI showed a sharp decrease from 2.2 to 0.5 N g⁻¹. A different trend was observed during the second year, in which, after a decrement from 2.2 to 1.0 N g⁻¹, the DI rose to 1.4 N g⁻¹. During the first year, the CI remained unchanged, with values close to 0 during the entire observation period. The measurement of the data was completed during the fourth survey (beginning of November). During the second year, the CI decreased sharply from 13.4 to about 0 by mid-October in the second year, reflecting a quicker maturation process. The PI showed no statistical difference between the years, but the trends varied: in the first year, the PI increased steadily, while in the second one, it surged from 0.4 to 4.1 by the 45th day, indicating accelerated pigment development. ‘Cipressino’ is known for its relatively rapid ripening process, which can vary between seasons, as reflected in the calculated indices. These findings are in line with what was reported by Fidalgo-Illesca et al. [63], in which a PI of 3 was observed as the right moment for harvesting. The differences observed between years for ‘Cipressino’ could be attributed to the hotter first-year climatic pattern, which caused the earlier coloration of drupes. This behavior could suggest that different stress factors or delayed physiological responses can influence this parameter.

3.5. Relation Between Parameters

The analysis focused on the relationship between the PI and the DI and aimed to understand how changes in fruit pigmentation can be related to fruit detachment. No data have been published yet on this relationship.

The results of the two-way ANOVA show that the effect of cultivars was statistically significant for the FF (F = 24.683, ***), DI (F = 56.891, ***), and PI (F = 43.218, ***), while it was not significant for the PI (F = 1.403, n.s). The year factor was not significant for the FF (F = 0.326, n.s.) and PI (F = 0.512, n.s.), but it was statistically significant for the DI (F = 72.045, ***) and CI (F = 11.6, ***). The cv × year interaction was significant for the FF (F = 7.338, ***), DI (F = 8.697, ***), and CI (F = 3.212, *), while it was not significant for the PI (F = 0.239, n.s.) (

Table 1. Results of the two-way ANOVA considering cultivar (cv) and year as factors for firmness (FF), detachment index (DI), colorimetric index (CI), and pigmentation index (PI); The table reports F-values and their statistical significance levels: n.s. p > 0.05, * p < 0.05, *** p < 0.001.

	FF		DI		CI		PI
cv	24.683	***	56.891	***	43.218	***	1.403	n.s.
year	0.326	n.s.	72.045	***	11.6	***	0.512	n.s.
cv × year	7.338	***	8.697	***	3.212	*	0.239	n.s.

). In the Supplementary Materials, the boxplot of the ripening indices during the two years of the experiment per each cv are reported (Figures S1 and S2).

To use a non-destructive index to predict a destructive one, the relation between DI and PI was investigated (Figure 8). ‘Fs-17’ showed the most pronounced relationship between pigmentation and detachment (R² = 0.91), indicating a highly significant and strong negative relationship. In this case, as the PI increased, DI decreased sharply. For ‘Leccino’, the analysis revealed a fairly strong negative relationship (R² = 0.60). As with the other cultivars, increased pigmentation was associated with a decrease in the DI, although the strength of this relationship was less robust than that observed in ’Fs-17’. In the case of ‘Frantoio’, the negative relationship between PI and DI was slightly stronger (R² = 0.55). This suggested a more consistent trend where increasing pigmentation was associated with a reduction in the detachment index. Finally, for ‘Cipressino’, the analysis revealed a moderate negative relation between PI and DI (R² = 0.48). However, the strength of this relationship was relatively weak, implying that other physiological or environmental factors might also play significant roles in influencing fruit detachment for this cultivar [42]. The spread of data points around the regression line further supported the notion of variability in this relationship. When the data of the four are considered collectively, the general model built for the prediction of the DI with the PI as a predictor showed a significant relationship between these two variables (R² = 0.45; Figure 8). This model provided moderate explanatory and predictive power, as also suggested by the validation on the second-year data (R² = 0.57, Figure 9). However, the reduction in model performance when data from different cultivars are taken together compared to models built with data from individual cultivars suggests the need for in-depth cultivar-specific studies, as the full mechanism of olive ripening is still not fully understood due to its complex functioning and genetic variations across cultivars [64].

4. Conclusions

‘Fs-17’ demonstrated a relatively rapid softening and ripening profile, making it well suited for mechanical harvesting in the early stages of ripening, typically around the first decade of October. The variability observed in the FF, DI, PI, and CI between the two years highlights the influence of environmental conditions on the ripening behavior of this cultivar. ‘Leccino’, with its longer ripening season, is suitable for harvesting with continuous harvesters 15–20 days after the beginning of ripening (first to second decade of October). ‘Frantoio’ demonstrates the most prolonged ripening period compared to other cultivars, with the FF, DI, PI, and CI showing clear, progressive changes. The extended ripening process, coupled with the cultivar’s sensitivity to climatic conditions, makes ‘Frantoio’ well suited for mechanical harvesting approximately 15–20 days after the onset of ripening, typically during the first or second decade of October. In contrast, ‘Cipressino’ had a more rapid softening profile, making it suitable for harvesting with continuous harvesters in the first days after the beginning of ripening (second to third decade of September).

However, the reliability of PI as a predictor of DI could be particularly useful for determining the precise timing of harvest to optimize both yield and fruit quality for ‘Fs-17’ and ‘Leccino’. The study found that the pigmentation index had a strong relationship with the detachment index, particularly for ‘Fs-17’ and ‘Leccino’, providing a reliable non-destructive measure for optimal harvesting time.

For ‘Cipressino’ and ‘Frantoio’, PI cannot be used as a predictor of DI.

These findings underscored the importance of developing cultivar-specific strategies in managing harvesting time. The variability in the strength of the PI-DI relationship across cultivars suggested that a one-size-fits-all approach might not be effective. Instead, understanding the unique maturation and detachment processes of each cultivar could lead to more precise and efficient harvesting practices. Further research would be necessary to address these issues and better understand the physiological mechanisms underlying these cultivar-dependent differences with more long-term trials.