The region of Coruche, Portugal, is characterized by a Mediterranean subhumid climate with hot and dry summers. According to the most recent climate normal (1981–2010) reported by the Portuguese Institute for Sea and Atmosphere, the average annual precipitation was 600.7 mm, and the average annual temperature was 16.2°C. Following the Regional Program for Forest Management (PROF) for Lisbon and Tagus Valley regions (Macedo et al. 2019b), the climatic aptitude for the region is considered superior for cork oak.

Before planting, soil profile evaluation was carried out at eight random locations, down to 2 m. Soil was characterized as unstructured with sandy texture, loose tenacity and friability, non-stickiness, no plasticity and minimal compaction. More than 75% of the particles were classified was gross sand and organic matter content was very low (0.32%). No significant differences in soil profile were observed between the locations and no weathered parent material was reached (C horizon). According to the FAO classification, the soil is of the Arenosol type. Arenosols exhibit constraints in soil water holding capacity, resulting in regular soil quality for cork oak development (Ferreira et al., 2001). Significant variations in tree growth rate were observed in relation to their location within the plot (Figure 1). Although soil profile evaluations did not provide an explanation for these differences, trees located on the backslope of the hill exhibited significantly lower growth compared to those on the footslope. Therefore, the plot was then divided into subplots according to dominant diameter at the base for the fertirrigated lines (Table 2).

Subplot with regular quality: the footslope of the hill, consisting of 22 plantation lines across the width, containing 1038 trees 8 years after planting.

Subplot with inferior quality: the backslope of the hill, consisting of 21 plantation lines across the width, containing 859 trees 8 years after planting.

Two years after planting, 20 profile probe tubes (PR2, Delta-T Devices®) were installed at 10 locations distributed across two planting lines, with four locations per treatment (Figure 1). The two planting lines were spaced 80 meters apart, with one located in the subplot of regular site quality and the other in the subplot of lower site quality. The tubes were installed between the plant and the irrigation tube, spaced 30 cm apart. Volumetric water content (%) was measured every other month at each location, at six depths down to 1 meter, using a profile probe device, totaling 35 data collection events. As irrigation tubes were placed 40 cm deep, weighted average of the soil moisture was calculated for two classes: sub-surface (0 down to 40cm) and deep (40 cm to 100 cm deep).

Soil field capacity was assessed by selecting only the measurements collected during the rainy periods. To avoid soil saturation, measurements were taken 2 days after the rainfall events. Nine independent measurements across the 20 locations were used for this assessment. For the statistical analyses, a full factorial general linear mixed model was applied—using the Restricted Maximum Likelihood (REML) method and the Satterthwaite approximation for degrees of freedom- to analyze whether soil volumetric water content—measured over time—varied according to (1) irrigation period—as opposed to the field capacity; (2) treatments; (3) site quality. The random factor in this analysis was each tube where soil moisture was repeatedly measured over time.

A PASI Electrical Imaging System Mod. 16G-N resistivity meter equipped with 32 electrodes was employed. Two parallel Electrical Resistivity Tomography (ERT) profiles were conducted, each extending 124 meters. One profile was located on the backslope, where soil moisture tubes were installed (site quality: inferior), and the other was positioned along a planting line at the footslope, intersecting the most productive cork oak patches (site quality: superior). These profiles were aligned between two planting lines and spaced 63 meters apart (Figure 1). Differential Spectra Precision Epoch 50 GNSS system was used to locate the profile ends accurately. The geoelectric surveys featured electrodes spaced 4m apart, and the system was configured to a Wenner-Schlumberger array with 12 scanning levels. The choice of the Wenner-Schlumberger array was motivated by its ability to provide results with good trade-off of horizontal and vertical resolution, particularly at shallower depths (Torrese et al., 2014; Hermawan and Putra, 2016). Data processing was executed using the RES2DINV inversion program from GeoTomo Software, that employ a smoothness-constrained least-squares method (Sasaki, 1992). The method was applied in June 2020, 9 days after the initiation of the fertirrigation period and a month following the spring rains.

All plants underwent annual measurements during the winter months (December–February). In the seedling phase, stem diameter at the base (Db) was measured with millimeter-precision using calipers, and total height (tH) was centimeter-accurate with a tape measure. As the trees grew, dendrometric measurements were performed using tree calipers and telescopic wands. Stand descriptors are in Table 2.

Perimeter at breast height (PB9) 9 years of age, only 19% of the trees had their PBH measured. Consequently, perimeter at the trunk base (PB) was used to infer the time until the first cork stripping. The relationship between PB and PBH was previously analyzed for trees within the experimental plot where both measurements were available (Supplementary Table S1). A general linear model was used to assess if irrigation treatments and site quality could potentially influence this relationship. The sampled trees were spatially scattered across the experimental area, therefore, spatial correlation within blocks was not expected. A significant (p < 0.001) and strong (R² = 0.855) allometric correlation was observed, independent of treatments and site quality (Figure 2; Supplementary Table S1). Based on this, the first cork stripping was projected to occur when the trunk perimeter at the base reached 98 cm. Additionally, it is important to note that the trunk perimeter above the cork is the key parameter for determining when trees are ready for debarking. Therefore, no estimation of cork thickness was required.

For the tree growth analyses, plants belonging to the blocks’ margins were discarded to reduce the influence of adjacent treatments on growth parameters. The remaining 663 trees from the regular site quality and 552 from the inferior site quality were included in the analyses. In a preliminary analysis, a general linear mixed model was applied using the Restricted Maximum Likelihood (REML) method and the Satterthwaite approximation for degrees of freedom, in order to evaluate differences in the perimeter of the 9 years ‘age trees based on treatments and site quality (Supplementary Table S2). Random variables were blocks and trees within blocks. Post-hoc pairwise contrasts, performed using the Least Significant Difference (LSD) method, indicated no significant differences between treatments (Supplementary Table S2.4). Therefore, in subsequent analyses, trees were grouped by site quality and by rainfed vs. fertirrigated. For each group, the time until the first cork stripping—defined as the period required for the tree’s perimeter at the base to reach 98 cm (serving as a proxy for 70 cm at PBH)—was estimated using a nonlinear regression analysis. This procedure was chosen considering that a growth curve is commonly divided into four stages, namely exponential, linear, logarithmic, and asymptotic (Figure 3A), related to the accelerated rate, constant velocity, decelerated rate, and no significant additional increment, respectively (Bahtiar and Iswanto, 2023). Graphical analyses showed that radial growth over time was in the exponential phase (Figure 3B). Given that the inflection point had not yet been reached (see Figure 3), it was explored four equations modeling exponential and linear growth, using the curve estimation regression procedure from SPSS software package:

TPB: Trunk perimeter at the base.

b0: constant representing the intercept.

b1: constant representing the slope.

All models were statistically significant (Equations 1–4).(p < 0.001); however, the power equation exhibited the highest R² value and F-statistics (Supplementary Table S3) (Equation 2). As a result, this equation was used to estimate the time until the first cork stripping at the stand level, by treatment and site quality. Using the nonlinear regression procedure (SPSS v.28 package), the parameters of the power equation were iterated to construct the models (Equation 2). Random effects (trees within blocks) were deliberately excluded from the model, so that the full observed variability—both among individual trees within blocks and across blocks—would be incorporated into the residual error. This approach allows the confidence intervals for the predicted year in which trees reach a perimeter of 70 cm to reflect not only the average growth trend, but also natural variation.

Soil volumetric water content below the sub-irrigation drip varied in relation to the field capacity and site quality (Table 3.1; Figure 4). No significant differences were observed across the treatments, nor was there any interaction between the factors. During the irrigation period deep soil moisture was about 9% less compared to the field capacity. Furthermore, locations at the backslope (inferior site quality) showed less 4% of water content when compared to the location at the footslope (regular site quality). Most of the unaccounted variance was primarily due to differences in the values measured over time (Table 3.2) and secondly across the 20 locations (Table 3.3).

Figure 5 displays the two ERT sections processed using the aforementioned method. Both sections exhibit resistivity values consistent with sandy soils (Pandey et al., 2015; Datsios et al., 2017; Gerscovich and Vipulanandan, 2023). At the section with regular site quality, the resistivity pattern is characterized by low values (depicted in green and blue colors) in a shallow layer with variable thickness, ranging from 2 to 10 meters. Within this layer, some spots with higher resistivity values (depicted in red and yellow colors) are detected but with no apparent correlation with treatments or tree size (Figure 5). The deeper layers of this section show high resistivity. On the other hand, the resistivity distribution at the section with inferior site quality follows a different overall pattern. It includes a superficial thin layer (down to 4 meters deep) with high resistivity values, irrespectively of treatments except where the road is located. An intermediate layer of low resistivity (falling within the range of values of the surface layer at the footslope section) with variable thickness, and finally, a third layer in the lower part of this section with high resistivity. This lower layer exhibits a geometry profile similar to the base layer of the previous section. In summary, the consistently high resistivity within the first 4 meters of soil in the section with inferior site quality indicates—for sandy soils—a very low soil moisture availability.

Overall, fertirrigation yielded similar outcomes across treatments, but its effectiveness appeared constrained under inferior site quality (Figure 3B). The absence of significant differences among the fertirrigation treatments allowed the development of only one model for fertirrigated trees by site quality (Table 4). The power model indicated that fertirrigated trees at the subplot with regular site quality are estimated to be debarked between 13 to 15 years of age (Figure 6A). In contrast, debarking for the control trees is expected to initiate, on average at 27 years of age (Figure 6A). As for the trees located at the inferior site quality, fertirrigated ones are expected to be debarked between 16- and 24-years of age. (Figure 6B). Additionally, the parameters’ 95% confidence interval indicated a high level of uncertainty for the control trees.

The REGASUBER plot was installed in an arenosol, a soil type considered to be of regular quality for cork production, due to its limited water-holding capacity. However, as the plants developed over time, differences in growth and survival indicated that the plot could be subdivided into two site quality classes: regular and inferior. Dominant diameter was chosen as a proxy for site quality, given that productive age is defined by the point at which radial stem growth reaches the threshold required for cork stripping.

Based on soil moisture retention curves from the literature (Evett et al., 2008), for soils reaching field capacity at 10 and 27%—such as those at the REGASUBER (Figure 4)—the wilting points are observed at 4 and 12%, respectively. Therefore, during the irrigation period, soil moisture above the irrigation tubes was close wilting point, while some water was available to plants below the tubes (Figure 4). However, differences between irrigation treatments could not be detected in the soil moisture profiles (Table 3.1). This can be explained by the low water-holding capacity of the sandy soil at REGASUBER. Previous measurements already had already indicated that the wet bulb fail to extend horizontally the 30 cm from the drip point to the soil moisture measurement tubes (Camilo-Alves et al., 2020). Minor variations in moisture retention were also detected based on site quality (Table 3.1), indicating improved soil texture for water retention down the hill. However, these differences were marginal, amounting to 4%. Given that cork oak roots can extend several meters deep to access groundwater (Otieno et al., 2006; David et al., 2007; Dinis, 2014), soil moisture measurements limited to a depth of 1 meter are insufficient to fully account for the water available to these trees. Indeed, during the excavation of six trees within the plot Šleglová et al. (2025a) observed that some roots extended beyond the 2-meter excavation depth. This highlights the need for the prospection of deeper water sources. The electrical resistivity method allowed the assessment of resistivity distribution down to a depth of 20 meters. Assuming similar soil composition in both areas, and considering the established correlation between soil moisture and resistivity (Ozcep et al., 2009; Jong et al., 2020), observed differences in resistivity can be interpreted as indicative of variations in soil water content. Accordingly, the higher resistivity observed down to 4 meters in the subplot with inferior site quality may be attributed to very low moisture. Conversely, this pattern was reversed in the subplot with regular site quality, possibly linked to higher surface water content in that section. These differences could be associated with variations in soil texture at deeper layers. Differences in ERT align with differences in mortality rate and tree development according to site quality. The surviving plants in areas with inferior site quality, facing drier conditions, may be allocating resources toward root system development (Padilla and Pugnaire, 2007; Ramírez-Valiente et al., 2019). Consequently, it is expected that their aerial development will occur once their root systems reach groundwater at depths below 4 meters. Further investigation through the long-term study within the experimental plot will confirm this hypothesis. Nevertheless, the use of electrical resistivity was a helpful tool to explaining differences in site quality across the plot. Water availability for cork oaks is not solely dependent on water supply; rather, the soil’s water-holding capacity and access to groundwater are also crucial factors (Pinto et al., 2014; Mendes et al., 2016). In this study, a preliminary prospecting test was conducted; however, further tests utilizing this technique should be performed over time to evaluate the spatiotemporal variability of soil moisture and optimize irrigation management.

Radial growth was lower in trees from the control lines; however, no statistically significant differences were detected across the irrigation treatments (Supplementary Table S1). This result contrasts with previous findings in the same plot (Camilo-Alves et al., 2020), as well as with results from the IRRICORK plot (Camilo-Alves et al., 2022). In both studies, growth increments were directly related to the applied water volume. A likely explanation lies in the temporal, structural and technical differences between the two experimental periods and sites, reinforced by the studies on root development in both plots (Šleglová et al., 2025b). While the earlier REGASUBER results reflected the initial response of young trees with localized root systems, the current analysis covers a later stage, when root systems have expanded further in depth. When water is supplied through subsurface drip irrigation in coarser soils, it is likely to result in water percolation and more homogeneous deep-water availability across treatments over time. In contrast, in IRRICORK, the loam to sandy clay loam soil texture combined with surface drip irrigation enables the formation of a stable superficial wet bulb near each tree, which supports a high density of fine roots within the irrigated area, allowing for a stronger treatment response, even in more mature trees. Further research using Electrical Resistivity Tomography (ERT) could provide valuable insights into water movement within the soil profile.”

Young trees from both the control and treatment groups were observed to be growing freely, showing no significant signs of having reached the inflection point. This is reflected in the power-type growth pattern, indicating continued and unrestricted growth at the age of 9 years. Thus, using the power model, estimates indicated that the first stripping will occur on average between 13 to 15 years’ age under the optimal conditions, i.e., 12 years earlier than rainfed trees in similar soil conditions. It is expected that, in areas with optimal and regular site quality, trees reach the appropriate dimensions for the first cork removal at around 20 years of age. In contrast, in areas classified as having inferior site quality, tree growth is diminished, and the cork stripping may be delayed by more than 10 years (Lacambra et al., 2010; Ribeiro et al., 2010; Sánchez-González et al., 2005). As comparison, diameter growth models based on stem analysis data from two of the main cork producing areas in Spain indicated that the age at which cork oak trees can be debarked for the first time is about 20 years for the best site quality (Sánchez-González et al., 2005), but considering a perimeter of 60 cm over virgin cork (Spanish legislation). Using the CORKFITS simulator, based or radial growth data from one of the main productive areas from Portugal (Ribeiro and Surovy, 2011), adjusting the potential modifier for the best site quality, and performing 100 simulations for a cork oak stand with a 4 ×4 spacing, the result indicated the trees could be debarked for the first time at an age of 19 ± 2 years.

These differences in the timing of the first cork stripping underscore the influence of year-round water availability and tree growth. Indeed, Camilo-Alves et al. (2022) demonstrated that radial growth is significantly sustained during summer drought in trees subjected to fertirrigation, even under conditions of high atmospheric vapor water deficit. Fertirrigation allowed the trees to achieve their full radial growth potential. Moreover, fertirrigation enhanced productivity in the subplot located on the backslope of the hill, effectively improving site quality from inferior to regular. In addition to the anticipation of the first cork stripping, another compelling outcome was the reduction of its uncertainty. Fertirrigation contributed to uniform tree growth throughout the stand, consequently enhancing productivity at the stand level. Conversely, on the control lines, variations in the access to water—and subsequently, nutrient availability—may lead to variations in tree growth. The first cork stripping is expected to be more gradual in rainfed conditions for the same soil conditions.

The findings from this study underscore the role of water availability—by means of irrigation during the drought periods—on site quality improvement in arenosols. This study also highlights the importance of soil water holding capacity and consolidate the mid-run impact of fertirrigation in tree growth. However, it is important to note that the model was specifically designed to estimate the time required to reach the first cork stripping. Future research should focus on developing tree growth and cork productivity models (e.g., Ferreira and Oliveira, 1991; Fonseca and Parresol, 2001; Ribeiro and Tomé, 2002; Ribeiro et al., 2003; Ribeiro et al., 2006; Tomé, 2004; Sánchez-González et al., 2007; Pasalodos-Tato et al., 2018) that incorporate modifiers for water availability, adapted to the conditions of the experimental plots.

Lastly, a note on the use of irrigation to enhance yield in sectors not directly linked to food production, particularly in regions prone to water scarcity: The Mediterranean climate, characterized by a summer water deficit, presents challenges for sustainable resource management. Therefore, the viability of cork oak irrigation is dependent on the existence of ponds to collect winter rainwater for use during the summer drought, thereby avoiding reliance on deepwater tables. Irrigation of approximately 700 m³/ha/year in the REGASUBER plot was sufficient to reach tree growth potential. This water volume represents 16% of the average annual precipitation in the stand and is considerably lower than the 2500 m³/ha/year used in intensive olive groves within the country. This approach is not intended to substitute rainfed stands. Instead, it is an alternative technique designed for specific conditions, aimed at providing a competitive edge over exotic tree species or over the intensification of the agro-pastoral systems associated with cork oaks. The enhancement of tree growth and canopy cover provides additional benefits, including increased carbon sequestration and biodiversity improvement. For instance, the rapid expansion of canopy cover was linked to greater floristic diversity in the REGASUBER plot (Raposo et al., submitted). Other studies have highlighted the role of high canopy cover in enhancing tree resilience and protecting regeneration in the face of climate change (Ribeiro et al., 2024a; Ribeiro et al., 2024b). Future analysis of a limited irrigation period (15–25 years) aimed at establishing productive cork oak stands more rapidly should also consider its effects on ecosystem services, such as carbon sequestration, water cycle, soil recovery; biodiversity, etc., particularly under climate change scenarios.