The study sites are located in valleys of the eastern Alps in South Tyrol, Italy ( Figure 1 ). This region is a major apple producer, boasting around 18,000 hectares of apple orchards that contribute roughly half of Italy’s apple production. In the past, pear trees were also widely cultivated, but this practice declined significantly by the 1970s. Land previously used for pears now primarily houses apple orchards. Current apple yields in the region average around 55 tons per hectare. Due to the relatively short lifespan of apple and pear orchards (around 20-30 years), they are not ideal for studying long-term climate impacts. To address this challenge, we collaborated with experts from South Tyrolean Fruit Tree Cultivation Museum to access two rather unique sites featuring old veteran apple and pear trees maintained for cultural heritage purposes. The apple orchard, within the municipality of Lana (46.60° N, 11.16° E, 310 m asl), featuring trees of the variety “Gravenstein” grafted on seedling rootstocks, was established around 1976. The pear orchard is located in the municipality of Prato allo Stelvio (46.63° N, 10.61° E, 884 m asl). Trees of the variety “Bartlett”, grafted on seedling rootstocks, were planted between 1928 and 1938. The pear site sampling also included a monumental tree of more than 200 years old belonging to the variety “Pala Birne”. Both orchards have received regular management practices since their establishment, including pruning, fertilization, irrigation, pest and disease control, and fruit harvesting. The soil in the sampling area has a sandy loam texture and is relatively fertile due to its high organic matter content.

Meteorological data from the Schlanders-Silandro station (46.62° N, 10.72° E, 698 m asl), located between the two sampling sites ( Figure 1 ), suggests that the sites are energy-limited ecosystems, where temperature remains optimal only during the short growing season (April – September). These records indicate annual precipitation (1981 – 2021) in the range of 400 – 800 mm, of which, the growing season months (April – September) receive about 60% and the rest mostly as snowfall during winter. The mean annual temperature (1988 – 2021) varies between -1.2 and 27.3°C, which remains above 13°C during the growing season ( Figure 1 ). The range of variation of minimum temperature is -3.5 to 14.0°C, which remains above 5°C during the growing season ( Figure 1 ). In recent decades, temperature trends have generally been upward, with an overall warming trend observed across the region. The average temperature in Europe has increased by around 1.5°C since the pre-industrial era, and the warming trend has been more pronounced in winter than in summer. According to the European Environment Agency, overall precipitation has remained relatively stable, with an increase of around 5% since the pre-industrial era.

For each species, five trees were randomly selected and from each tree, two cores were extracted at 0.5 m from the ground using a 10 mm diameter increment borer. These increment cores were air-dried and glued on wooden supports. The cross-sectional surface of the cores was sanded by increasingly fine sandpapers until growth rings were visible and finally digitalized with a high-resolution scanner (2400 d.p.i.; Epson Expression 10000XL, Long Beach). A standard image was created for each sample and all images were saved into a graphic file format for further analysis. Subsequently, the determination of the tree ring width of each sample was performed with the Coo-recorder software (Cybis Elektronik & Data AB, Saltsjöbaden, Sweden) at a precision level of 0.01 mm (García-Hidalgo et al., 2022).

We selected three individuals from each species based on the absence of biotic damages and on the chronological length for the isotopic analyses. Each year’s growth-ring was cut using a sharp razor blade under the binocular microscope and pooling was performed for the individual tree rings of corresponding age. We utilized the whole-wood for the stable isotope analyses (δ¹³C and δ¹⁸O) (Schollaen et al., 2017; Wieser et al., 2016). Stable isotope analyses were carried out at the Università degli studi della Campania “L. Vanvitelli” Dipartimento di Scienze e Tecnologie Ambientali Biologiche e Farmaceutiche, Caserta, Italy. The analytical precision was equal to or better than 0.2‰ for both the isotopes. The isotope ratios are presented in common δ-notation against international standard PDB and VSMOW respectively as:

Where, (¹³C/¹²C) sample and (¹³C/¹²C) PDB are heavy to light carbon isotope ratios in the wood sample and the standard (Vienna Pee Dee Belemnite), respectively. (¹⁸O/¹⁶O) sample and (¹⁸O/¹⁶O) VSMOW are heavy to light oxygen isotope ratios in the wood sample and the international standard (Vienna Standard Mean Ocean water) respectively.

Discrimination against ¹³C (Δ¹³C, ‰) during carbon fixation by trees was computed by using atmospheric (δ¹³C_atm) and tree-ring (δ¹³C_plant) δ¹³C as:

Where, (δ¹³C_atm) and (δ¹³C_plant) are fractional differences in isotopic composition (¹³C/¹²C) in atmospheric CO₂ and that of tree-ring wood. We compiled δ¹³C_atm from McCarroll and Loader (2004) up to the year 2004, and after that was derived from Belmecheri and Lavergne (2020) (https://scrippsco2.ucsd.edu/data/). A widely accepted procedure to correct tree-ring isotope chronology for the incorporation of isotopically light carbon released by the burning of fossil fuels and increasing CO₂ concentration was adopted (McCarroll and Loader, 2004). The correction procedure has the advantage of being an objective one as it effectively removes any declining trend in the δ¹³C series post AD 1850, which is attributed to physiological response to increased atmospheric CO₂ concentrations (McCarroll and Loader, 2004).

Carbon isotopic discrimination (Δ¹³C) is related to intercellular CO₂ (Ci) and atmospheric CO₂ (Ca) concentration as:

Where, ‘a’ is the fractionation factor during intercellular diffusion (−4.4‰), and ‘b’ is the fractionation factor during carboxylation (−27‰) (Farquhar et al., 1982). The ratio of Ci and Ca was determined as:

or,

Using Ci and Ca values, intrinsic water use efficiency (iWUE) was calculated as:

To substantiate our inferences, we computed a standardized carbon-to-oxygen isotope difference index for all species over the entire observation period, following the model of Scheidegger et al. (2000):

Here, n is the year. Both corrected δ¹³C and δ¹⁸O chronologies were transformed into z-scores (based on the long-term mean and std. dev.) and analyzed in pairs for each crop. This index allows tracking year-by-year changes in trees’ physiological conditions induced by changes in stomatal conductance and photosynthetic capacity. The index assumes values close to 0 when both isotope ratios show similar values, indicating either enhanced stomatal conductance (both isotopic values are negative) or reduced stomatal conductance (both isotopic values are positive). Positive index values indicate high photosynthetic capacity (high δ¹³C and low δ¹⁸O), while negative values indicate low photosynthetic capacity (low δ¹³C and high δ¹⁸O) (Scheidegger et al., 2000; Siegwolf et al., 2023).

To understand relationships between δ¹³C-based processes (Ci, Δ¹³C, iWUE) in the two species and the regional climate (Precipitation and Temperature: Mean, Max., Min.), simple Pearson correlations were applied with a response-function approach. The confidence intervals of correlations were analyzed at 95% and 99% levels. This helped to investigate the correlations with monthly climatic averages in the species. To corroborate and stabilize these relations further, we plotted 3-month moving correlation coefficients between physiological processes and monthly hydroclimatic data (precipitation: 1982-2021; temperature: 1988-2021). The response function analysis for the species was plotted from October of the previous growth year to September of the current year (pOct-Sep). To test the relative importance of climate parameters including atmospheric CO₂ levels, multiple linear regression models with iWUE as the response variables, and temperature (mean, max, min), precipitation, and atmospheric CO₂ as continuous predictor variables, were built.

To test the significance of the slopes (p-values), we first computed Mean Square Error (MSE) as: M S E = ( 1 − R 2 ) × V a r   ( y ) .

Then, the Standard Error (SE) of the slope was computed as: M S E ( n − 1 ) × V a r ( x ) ; and the t-statistic was calculated as: t = s l o p e S E .

Finally, using the t-distribution, we calculated the two-tailed p-value for the respective t-statistics and degrees of freedom (df = n – 2).

We used ‘lm’ function from the R statistical computing environment (R development Core team, 2015). The relative importance of significant terms was obtained by applying function ‘calc.relimp’ using default options ( Supplementary Table S1 ). The process reconstructions were standardized using Z-scores and smoothed with a 3-year running mean to assess common signals.

Over the study period, raw δ¹³C in tree-rings of pear (1943-2021) and apple (1976-2021) species exhibited a slight decreasing trend ( Figure 2 ). The break-point analysis identifies the year of change in carbon isotopic composition in the species as the year 1987-88. Therefore, 1990 is assigned as a reference year for the analyses. For the common period, inter-species correlation was moderate at the inter-annual scale (r = 0.398, p< 0.001), which indicates the predominance of both species-specific and site-specific local effects on the assimilation process. The mean δ¹³C value of apple trees was −25.6‰ (std. dev.: 0.40‰), while for the pear trees, it was ∼1.0‰ lower (−26.6 ± 0.32‰) ( Figure 2A ; Table 1 ). This difference suggests a higher level of assimilation rate (isotope discrimination) in pear trees ( Figure 2B ). However, pear and apple trees have reportedly similar daily radiation use efficiency (Auzmendi et al., 2013). The long-term mean of δ¹³C (corrected for atmospheric CO₂ increase) in apple trees was −23.5 ± 0.6‰, which increased slightly (−23.2 ± 0.52‰) after 1990. The long-term mean for the pear trees (−25.1 ± 0.72‰) increased by ~ 1‰ after 1990 (−24.3 ± 0.42‰) ( Supplementary Figure S1 ). Considering the changes in atmosphere-to-plant ¹³CO₂ discrimination (Δ¹³C), we found a higher (~2.0‰) level of discrimination in pear trees, which showed a similar temporal pattern to that of apple trees ( Figure 2 ; Table 1 ). The corrections of the carbon isotope series of the two species for the physiological responses to increasing concentrations of atmospheric CO₂ are illustrated in the supplementary figure ( Supplementary Figure S1 ).

A rising trend in intercellular CO₂ (Ci) in the two species was noted during the observation period. The mean Ci value of the apple tree was 223 ± 10.3 µmol mol^-1, while for the pear trees, it was 233 ± 13.6 µmol mol^-1 ( Table 1 ). Prior to 1990, the mean Ci of apple trees was 207 µmol mol^-1 (range: 181 – 230 µmol mol^-1), which increased to 227 µmol mol^-1 (211 – 247 µmol mol^-1). While, for the pear trees, the mean Ci before and after 1990 was 225 µmol mol^-1 (211 – 251 µmol mol^-1) and 245 µmol mol^-1 (231 – 264 µmol mol^-1), respectively ( Figure 3 ). The trends in intercellular CO₂ (Ci) to atmospheric CO₂ (Ca), i.e., Ci/Ca ratio in both species were analogous to the trends in Δ¹³C. Nonetheless, a declining trend in both species is noticeable after 1990. Prior to 1990, the ratio for the apple trees was 0.62 (0.59 – 0.66), which decreased to 0.60 (0.56 – 0.64). While, in the pear trees the ratio before and after 1990 was 0.69 (0.66 – 0.71) and 0.65 (0.63 – 0.69), respectively. Fitted regression slopes for the Ci/Ca ratio (p< 0.05) in apple and pear trees prior to 1990 were 0.0021 (R² = 0.2) and -0.0003 (R² = 0.087), respectively. After 1990, magnitude of the slopes becomes more negative (Apple: -0.0011, R² = 0.25; Pear: -0.0009, R² = 0.28). Out of three theoretical scenarios of Ci/Ca ratios to CO₂ rise (Panthi et al., 2020), a positive slope in apple trees suggests a ‘Ca – Ci = constant’ scenario prior to 1990 that changed to ‘Ci = constant’ scenario after 1990. Conversely, the response of pear trees appears to be a ‘Ci = constant’ scenario throughout the observation period. These results probably suggest a varying physiological response of tree species to atmospheric CO₂ rise ( Figure 3 ).

The Δ¹³C discrimination at the plant level is controlled by Ci/Ca ratio. This ratio could decline either because of low stomatal/mesophyll conductance to CO₂ associated with water stress or low temperatures, or due to a high assimilation rate (McCarroll and Loader, 2004). At our sites, in the alpine valleys with frequent irrigation, water cannot be assumed to be a limiting factor. Growth at higher latitudes is generally limited by suboptimal temperatures for xylogenesis (i.e., formation of water conductive tissue), which remains almost optimal during the growing season. In energy-limited ecosystems at higher latitudes, climate warming may improve tree-water status where xylogenesis is temperature-limited and given that sufficient water is available. Therefore, increasing assimilation rates due to rising CO₂ levels could be another reason for the observed declining Ci/Ca ratio in recent decades. The effect of CO₂ fertilization has been observed globally, which is quite pervasive in European forests, particularly over the northern ecosystems (Mathias and Thomas, 2021; Waterhouse et al., 2004; Saurer et al., 2014; Frank et al., 2015).

Hydro-climate response function: the Pearson correlation with a response function approach demonstrates the relationship between species’ physiological processes (δ¹³C, Δ¹³C, and Ci) and monthly hydroclimatic data (precipitation and temperature: mean, maximum, and minimum). To corroborate and stabilize these monthly relations, we plotted 3-month moving correlation coefficients (precipitation: 1982-2021; temperature: 1988-2021) ( Figure 4 ). The response function analysis for the δ¹³C – precipitation relationship from October of the previous growth year to September of the current year (pOct-Sep) revealed non-significant correlations. The relationship confirms that water is not a limiting factor in these orchards, having provision of irrigation, especially for the apple trees. For this reason, at the beginning of the growing season (March-May) we observed an enhanced positive correlation for the pear trees. In contrast and irrespective of the seasons, δ¹³C – precipitation relationship for the apple trees was non-significant ( Figure 4A ). Likewise, Δ¹³C – and Ci – precipitation relationship during peak growing season (June-October) was significant for the pear trees but non-significant for the apple trees having inverse correlations ( Figures 4B, C ). Yet, Pearson correlations and 3-month moving correlations with temperature indicated that the latter has a major control on the species’ ecophysiological processes ( Figures 4D–F ). The δ¹³C – temperature (mean) relationship was non-significant for both species ( Figure 4D ). Nevertheless, we observed significant, Δ¹³C – and Ci – mean temperature correlations across the months for both species ( Figures 4E, F ). Particularly, the influence of minimum temperature on Δ¹³C and Ci was prominent during the entire growing season (March – October) ( Supplementary Figures S2A–C ). Whereas, maximum temperature appears to affect these processes during the start of the growing season (April-June) ( Supplementary Figures S2D–F ). The relationships between temperature (Min., Max.) and δ¹³C, Δ¹³C, and Ci have been detailed in the supplementary figure ( Supplementary Figures S2A–F ).

Oxygen isotope in the tree-rings complemented with respective δ¹³C values, remains the only proxy with proven potential to decipher a comprehensive picture of past and current ecophysiological status (Battipaglia et al., 2013; Nock et al., 2011; Siegwolf et al., 2023). Consequently, we have taken into account the δ¹⁸O chronologies of the trees ( Figure 5 ). The mean δ¹⁸O value of pear trees (1943 – 2021) was 25.9 ± 0.99 ‰ (Coefficient of Variation (CV): 3.8%), while for the apple trees (1976 – 2021), it was ∼1.0 ‰ higher (26.3 ± 1.25 ‰) with a higher CV (5.6%) ( Table 1 ). Prior to 1990, the mean δ¹⁸O of apple trees was 25.8 ‰ (range: 24.3 – 28.0 ‰), which showed a rising trend after 1990 with a similar range of variation. While, for the pear trees, the mean δ¹⁸O before 1990 was 26.3 ‰ (24.5 – 28.4 ‰), which dropped to 25.2 ‰ (23.5 – 26.3 ‰) since then ( Figure 5A ). The mean difference between species indicates a higher level of oxygen isotope discrimination and evapotranspiration in pear trees relative to the apple trees. At the same vapor pressure deficit, leaves with high transpiration rates are known to become isotopically enriched in heavy isotopes as compared to leaves having low transpiration (McCarroll and Loader, 2004; Battipaglia et al., 2013). The optimal transpiration rate coupled with the use of enriched surface irrigated water by the apple trees and differences in the stomatal conductance could be responsible for such an ¹⁸O enrichment offset. Moreover, because of limited evaporation, groundwater is less enriched (compared to surface water) and the probable use of this less enriched groundwater by the pear trees (having greater tree height and rooting depth) is reflected in its stable time series having lower CV ( Figure 5A ) and δ¹³C – precipitation relationship ( Figure 4A ). However, a decline (after 1990) in ¹⁸O enrichment (~1.0 ‰) in pear trees could be linked to increasing precipitation trend and irrigation provisioning.

Due to the above reasons, response function analysis for the monthly δ¹⁸O – precipitation relationship (pOct-Sep) showed opposite correlations in the species ( Figures 5B, C ). Cross-correlations of δ¹⁸O chronology of pear trees with precipitation revealed a strong positive relationship during the start of the growing season (March-May), which becomes negative during peak growing season (June-October). The δ¹⁸O – precipitation relationship for the apple trees, on the contrary, was opposite to that of the pear trees and insignificant throughout the growing season (March-October) ( Figure 5B ). Similarly, correlations with mean temperature indicated opposite but a major control on species’ ecophysiological processes associated with ¹⁸O enrichment, particularly during peak growing season (June-October). For this period, the correlation was significantly positive and negative for apple and pear trees, respectively ( Figure 5C ). In contrast, during March-May, correlation was negative in both species but remained statistically insignificant. Both Pearson and 3-month moving correlations between minimum temperature and δ¹⁸O have been detailed in the supplementary figure ( Supplementary Figures S3A, B ), which reflects greater but opposite response for the two species during peak growing season (June-October). On the contrary, the impact of maximum temperature was minimal. Modelling studies indicate that tree-ring δ¹⁸O values integrate signals from three primary factors: source water δ¹⁸O, evaporative enrichment of ¹⁸O in leaf water, and biochemical fractionation during organic matter synthesis. Consequently, tree-ring δ¹⁸O signals vary as a function of temperature, relative humidity, precipitation, water sources, and regional climate conditions (Barbour, 2007; Gessler et al., 2014; Kahmen et al., 2011). Based on the premise that tree roots take up soil water without fractionation, a major part of the isotopic signature in tree rings should reflect the variation in precipitation or hydroclimatic conditions (Farquhar et al., 2007; Lehmann et al., 2018; Roden et al., 2005). However, depending upon the local humidity condition, species-specific ecophysiological processes and responses (e.g.: isotopic composition of soil water, leaf-water enrichment, and oxygen isotope exchange reactions of photosynthates with water) may have a considerable effect on tree ring δ¹⁸O values (Baker et al., 2016; Lehmann et al., 2018), as revealed in our study.

Broadly, an increasing trend in iWUE was observed during the study period in both species that raised sharply after the 1990s ( Figure 6 ). The mean iWUE of the apple tree was 88.5 ± 9.0 µmol mol^-1, while for the pear trees, it was lower at 70.3 ± 10.5 µmol mol^-1. Prior to 1990, the mean iWUE of apple trees was 80.2 µmol mol^-1 (72 – 86 µmol mol^-1), which increased to 93.6 µmol mol^-1 (80 – 108 µmol mol^-1). In contrast, for the pear trees, the mean iWUE before and after 1990 was 62.6 µmol mol^-1 (55 – 73 µmol mol^-1) and 81.8 µmol mol^-1 (69 – 91 µmol mol^-1), respectively. During the observation period, mean iWUE of the apple trees was higher than that of the pear trees (~18 µmol mol^-1), which after 1990 increased by 15.5 and 30.6% respectively for apple and pear trees ( Table 1 ). For the common period, inter-species correlation was significantly high (r = 0.82, p< 0.001), which probably indicates the predominant control of climate on species’ ecophysiological process.

Moreover, it would be informative to note that long-term gains in iWUE by vegetation are usually overestimated (Gong et al., 2022). To provide a more accurate assessment, it is crucial to account for post-photosynthetic fractionations and mesophyll conductance, which influence CO₂ diffusion to carboxylation sites (Gimeno et al., 2021; Gong et al., 2022; Ma et al., 2021). Our iWUE calculations, based on a linear model of photosynthetic ¹³C discrimination (Δ¹³C), do not fully capture long-term structural and physiological acclimations. Therefore, we advocate for the adoption of advanced models that should incorporate post-photosynthetic fractionations and mesophyll conductance as well as photorespiration to mitigate errors in estimating iWUE from Δ¹³C across vegetation types.

Nevertheless, to elaborate on the climatic controls of iWUE, we performed response function analysis on iWUE chronologies and monthly hydroclimatic datasets. For both species, major hydroclimatic variables (temperature and precipitation) showed positive relations with iWUE ( Figures 6B, C ). During the beginning of the growing season (March-May), iWUE – precipitation relationship was negative (p > 0.05) for the species that turned to positive correlations during the peak growing season (June-October) ( Figure 6B ). Cross-correlations with mean temperature revealed a strong positive correlation during peak growing season in both species, which was insignificantly low during March to May ( Figure 6C ). Similarly, with respect to minimum and maximum temperature, we noted a higher influence of minimum temperature on iWUE during peak growing season in both species ( Supplementary Figures S4A, B ).

Furthermore, stepwise regression between annual and 5-year mean iWUE (as dependent variable) and mean temperature (Tmean), precipitation (Ppt), and atmospheric CO₂ concentrations (CO₂) (as independent variables) were employed to access the explanatory power of climate variables. In both species, annual CO₂ explained more than 70% of the variability. Whereas, 5-year mean CO₂ explained more than 90% of the variability in iWUE. Analyses further reveal that Tmean is the second most important variable, while precipitation has a minimal impact on species’ iWUE ( Supplementary Table S1 ). Several studies have inferred that rising atmospheric CO₂ levels do not always imply enhanced photosynthetic rate and tree growth (Nock et al., 2011; Rahman et al., 2020; Van Der Sleen et al., 2015), with species-specific responses. We also observed species-specific responses and noted a differential response of species’ iWUE to atmospheric CO₂ (Ca) ( Figure 6D ). The regression coefficient in apple trees (R² = 0.76) was lower than that of the pear trees (R² = 0.91). Results further suggest a slightly higher sensitivity of apple trees to rising Ca (slope: 0.38). However, a lower regression coefficient in apple trees indicates the possible role of other environmental factors particularly soil moisture (via stomatal regulation) in influencing their iWUE response. On the other hand, pear trees’ iWUE appears more tightly correlated with atmospheric CO₂ levels (higher R²) possibly due to their access to relatively invariable source water (groundwater), but having a slightly lower slope (0.35). Despite this, both species have similar intercepts (Apple: -50.89 and Pear: -50.64), suggesting comparable baseline iWUE in the context of CO₂ response. This implies that both species started from a similar baseline, but their responses to rising CO₂ levels have diverged.

In conclusion, both species have responded to rising CO₂ levels, with apple trees showing higher sensitivity as well as variability to environmental changes. Pear trees, on the other hand, have exhibited a more predictable and consistent response. It’s possible that apple trees may have reached a threshold in their ability to increase WUE as CO₂ rises ( Figure 6D ). This information is valuable for predicting how these species may adapt to future environmental conditions, especially in regions with fluctuating water availability.

Tree-ring δ¹⁸O, in combination with δ¹³C, is a powerful proxy to decipher a comprehensive picture of past and current ecophysiological status (Battipaglia et al., 2013; Nock et al., 2011; Siegwolf et al., 2023). Therefore, we computed carbon-to-oxygen isotope difference index for both species ( Figure 7A ). This difference serves as an indicator of the trees’ physiological conditions related to changes in stomatal conductance and photosynthetic capacity (Scheidegger et al., 2000).

The C-to-O difference index for apple trees exhibits an increasing trend from the late 1970s onwards, with a more pronounced positive shift after the year 2000. The index also shows considerable year-to-year variability. In the early years (1976-1980), values fluctuate around the zero line, indicating a balance between low stomatal conductance and high photosynthetic capacity. From 1980 onwards, periods of high photosynthetic capacity (positive index values) appear more frequent and sustained, likely reflecting the irrigation effect. In recent years (2000-2021), the index mostly remained positive, indicating a generally high photosynthetic capacity for apple trees, despite occasional dips around 2010 and 2015.

The index for pear trees shows an overall increasing trend over the entire period. There is a clear positive shift around the late 1980s, aligning with the apple tree index. However, the species’ pattern briefly diverged during the 2010s when δ¹³C levels were stable and δ¹⁸O values were high in apple trees ( Figures 2 , 5 ). Similar to apple trees, pear trees exhibit considerable inter-annual variability. The period before 1980 shows more frequent negative values, indicating periods of low photosynthesis, likely induced by stomatal limitations given limited irrigation provisioning. Post-1980, positive values become more dominant, suggesting an improvement in photosynthetic assimilation. In recent decades (2000-2021), the overall trend remained positive, with a few dips indicating brief periods of moisture stress and reduced photosynthetic efficiency. The analysis suggests an overall improvement in photosynthetic capacity over the studied periods, with more frequent and sustained positive index values in recent decades in both species. This trend reflects adaptive physiological responses to changing environmental conditions including the CO₂ fertilization effect in the Italian Alps. However, both species exhibit notable year-to-year variability, highlighting the influence of inter-annual climatic variations on tree physiology and productivity.

Previously, we observed species-specific responses of iWUE to atmospheric CO₂ (Ca), with pear trees exhibiting lower iWUE and a higher regression coefficient (R² = 0.91) compared to apple trees (R² = 0.76) ( Figures 6A, D ). To further explore these differences, we plotted species’ C-to-O difference indices against Ca ( Figure 7A ). The slope of pear trees (0.056, p< 0.001) with respect to apple trees (0.019, p< 0.001) suggests a stronger physiological response to increased CO₂, with clear changes in stomatal conductance and photosynthetic activity ( Figure 7A ). Further, a much higher R² value for pear trees (69%) indicates that a significant portion of the changes in stomatal conductance and photosynthetic activity can be explained by changes in CO₂ concentration.

This suggests that pear trees’ physiological conditions, as indicated by the index, are more closely tied to rising CO₂ levels compared to apple trees. The weak slope and low regression coefficient suggest that the physiological processes in the apple tree are primarily related to the stomatal regulation via soil moisture. Overall, our analysis revealed a significantly higher correlation for pear trees (r = 0.83, p< 0.0001) than for apple trees (r = 0.32, p< 0.05). These results suggest that apple trees may have reached a threshold in their ability to increase water use efficiency as CO₂ levels rise, while pear trees continue to show a stronger response. This could explain the differential responses of these two tree species to changing environmental conditions, including the CO₂ fertilization effect.

Cross-correlations with hydroclimatic variables were further performed to gain insights into the combined responses for the index series of the species. Response analyses clearly indicate that precipitation has a minimal impact on isotope-inferred ecophysiological processes, except for the pear trees which show a significant positive correlation during peak growing season ( Figure 7B ). Similarly, index – mean temperature relationship was significant only for the pear trees that only during peak growing season ( Figure 7C ). Both Pearson and 3-month moving correlations between minimum/maximum temperature and species index series have been detailed in the supplementary figure ( Supplementary Figures S5A, B ). These results demonstrate hydroclimate and ecophysiological relationships in these energy-limited ecosystems.