Analyses from stand to tree level allow disentangling the effects of age, size, origin and competition on tree growth sensitivity to climate in natural and afforested Scots pine forests

Agricultural and Forest Meteorology 355 (2024) 110148 Available online 9 July 2024 0168-1923/© 2024 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/). Analyses from stand to tree level allow disentangling the effects of age, size, origin and competition on tree growth sensitivity to climate in natural and afforested Scots pine forests ´ Alvaro Rubio-Cuadrado a , b , * , Fernando Montes c , Iciar Alberdi c , Isabel Ca˜ nellas c , Isabel Aull´ o-Maestro c , Raúl S´ anchez-Salguero d , Andrea Hevia d , e , C´ esar P´ erez-Cruzado f , Juan Alberto Molina-Valero g , J. Julio Camarero b a Departamento de Sistemas y Recursos Naturales, Escuela T´ ecnica Superior de Ingeniería de Montes, Forestal y del Medio Natural, Universidad Polit´ ecnica de Madrid. Ciudad Universitaria s/n, 28040, Madrid, Spain b Instituto Pirenaico de Ecología (IPE-CSIC), Avda. Monta˜ nana 1005, 50080, Zaragoza, Spain c Departamento de Din´ amica y Gesti´ on Forestal, Instituto de Ciencias Forestales (ICIFOR-INIA), CSIC. Ctra. La Coru˜ na km 7.5, 28040, Madrid, Spain d DendrOlavide, Departamento de Sistemas Físicos, Químicos y Naturales, Universidad Pablo de Olavide, Crta. de Utrera km. 1, 41013, Sevilla, Spain e Departamento de Ecología, Evoluci´ on y Conservaci´ on de la Biodiversidad, Universidad de Sevilla, Facultad de Biología, Avda. Reina Mercedes, s/n, 41012 Sevilla, Spain f Proyectos y Planificaci´ on (PROEPLA), Escuela Polit´ ecnica Superior de Ingeniería, Universidade de Santiago de Compostela, Benigno Ledo s/n, Campus Terra, 27002 Lugo, Spain g Faculty of Forestry and Wood Sciences, Czech University of Life Sciences Prague (CZU), 16 500 Prague, Czechia ARTICLE INFO Keywords: Climate-growth relationship Old-growth forest Mature forest Individual tree analysis Dendroecology Pinus sylvestris ABSTRACT The sensitivity of tree growth to climate is conditioned by several variables, often intermingled, such as the origin of the forest (natural vs. artificial), tree age, tree size and tree-to-tree competition. The effect of these variables is usually inferred from average growth series obtained at the stand level, thus ignoring the differences at the individual tree level and their drivers. Our objective is to disentangle the effects of stand origin, age, size, competition and social status on the sensitivity of tree growth to climate, including the effects of extreme climatic events, such as droughts, on resistance and resilience, and also to show the advantages of using a tree-level approach. To this end, we compared four stands of Pinus sylvestris with contrasting characteristics: young afforested, young of natural origin, old afforested and old of natural origin. We analyzed differences in growth sensitivity to climate at both stand and tree levels to compare both approaches. Our results show the great complexity of the relationships of the variables considered with the sensitivity of growth to climate. All these variables are important and with strong interactions between them, which makes their effects not unidirectional and strongly dependent on the site conditions. While the stand approach hides these interactions, the tree-level approach makes it possible to analyze them in detail. In general, the sensitivity of growth to climate increases with age, diameter, afforested origin, decreasing competition and higher social status; growth resistance increases with age, decreasing diameter, afforested origin, decreasing competition and higher social status; and growth resilience increases with age, diameter, natural origin, decreasing competition and higher social status. We show the usefulness of the analysis of data at individual tree level, which, combined with the joint analysis at stand level, allows us to obtain more accurate and detailed information. 1. Introduction Ongoing climate change is leading to changes in forest growth responses. These changes may lead to adaptation to new conditions while maintaining stable growth (Rubio-Cuadrado et al., 2020a), or to a reduction in growth due to drought limitation, which could impact future forest dynamics and carbon uptake (Babst et al., 2019). Therefore, it is pivotal to disentangle the multiple drivers of tree growth response to climate to assess the vulnerability of forests to warming and drying conditions (e.g., S´ anchez-Salguero et al., 2015, 2013). The sensitivity of * Corresponding author. E-mail address: [email protected] (´ A. Rubio-Cuadrado). Contents lists available at ScienceDirect Agricultural and Forest Meteorology journal homepage: www.elsevier.com/locate/agrformet https://doi.org/10.1016/j.agrformet.2024.110148 Received 21 May 2024; Accepted 1 July 2024 Agricultural and Forest Meteorology 355 (2024) 110148 2 tree radial growth to climate (both average climate conditions and climate extremes), is determined by a set of primary variables, with an obvious and widely studied influence, such as tree species and provenances, and soil and climate characteristics (Camarero et al., 2015; D’Orangeville et al., 2018; Gea-Izquierdo and Ca˜ nellas, 2014; Griesbauer et al., 2011; Orwig and Abrams, 1997; S´ anchez-Salguero et al., 2018). But there are also other variables, whose influences on climate-growth relationships are less evident or may even have no influence at all under certain conditions. These variables include stand and tree characteristics, such as size, age, origin, stand structure, and competition, among others (Fritts, 1976), many of which are easily modifiable by management. The effect of these variables is poorly understood because they are often intermingled, making it difficult to disentangle the variables responsible for the different sensitivity of growth to climate (Carrer and Urbinati, 2004; G´ omez-Aparicio et al., 2011; S´ anchez-Salguero et al., 2015; Trouillier et al., 2019). Moreover, the studies that seek to analyze this effect are focused on one or some of these variables, but there are no studies that analyze all of them together with their interactions, and usually employ mean growth series for groups of trees, being these groups made up of trees from the same stand or with similar characteristics (e.g., size, age or species). However, considering the differences between individual trees allows us to study the whole range of individual responses among sampled trees and to understand those variables responsible for the different responses (Carrer, 2011; Galv´ an et al., 2014; Rozas, 2014). All these elements make this issue a field of study that requires greater efforts and new methodological approaches to unravel the effects of each of these variables on the sensitivity of tree growth to climate. Forest origin (natural or artificial) can have a great influence on the response of growth to climate (Camarero et al., 2021b; S´ anchez-Salguero et al., 2013), and a relationship has been found between this variable and drought-induced mortality (Rubio-Cuadrado et al., 2021). However, this influence is not direct, but it is due to the effect of other related variables. Afforestations are often pure stands with elevated competition, a low coefficient of variation of tree diameter, inadequate seed origin in these afforestations made at the beginning and in the middle of the twentieth century, poor and superficial soils and, in some cases, they were established beyond the current climatic range limit of the species, all of which affect the sensitivity of tree growth to climate (S´ anchez-Salguero et al., 2013). Likewise, cultural treatments carried out in nurseries, such as the repeated undercutting of the taproots of seedlings or different types of container, can affect the adequate development of the plant, modifying in this case the shape of the root system, and thus increasing the sensitivity of growth to climate (Zadworny et al., 2019). The number of variables interrelated with the origin of forests means that the results (in terms of their sensitivity to climate) obtained in a given afforestation may not necessarily be valid for another afforestation. This, together with the importance (e.g., ecological, social, and environmental) and extent of afforestations (FAO, 2006), and the current conditions of climate change, make it especially important to increase the research effort in this area. Tree size and age are, in general, related to climate sensitivity, which increases up to a certain value of these variables after which it remains constant (Carrer and Urbinati, 2004; Rozas et al., 2009; Trouillier et al., 2019). However, relatively few studies have tested the influence of these variables in the climate–growth response, and conclusions vary among those that do exist (Dorman et al., 2015; Fritts, 1976; Galv´ an et al., 2014; Rozas, 2014; Rozas et al., 2009; S´ anchez-Salguero et al., 2015; Szeicz and MacDonald, 1994). In addition, age and size are strongly interrelated, which makes it difficult to disentangle their effects on the sensitivity of growth to climate. Most studies concerning this topic tend to focus on the age effect, however, age appears to have few direct effects on tree physiology as compared with size (Bond, 2000; Thomas, 2002). In contrast, size is related to hydraulic conductivity (Ryan et al., 2006; Ryan and Yoder, 1997), and changes in xylem anatomy due to tapering (Carrer et al., 2015). Therefore, there is ecophysiological background to support the role played by size on driving the sensitivity of growth to climate. Stem diameter could thus facilitate a more direct assessment of size effects (Trouillier et al., 2019). Competition is a key process driving forest growth due to the limitation that this variable produces on obtaining soil water and nutrients and acquiring radiation (Orwig and Abrams, 1997; Weiner, 1990). Thus, trees may have different climate-growth sensitivities, depending on the different competition levels, including different structural attributes and social statuses (Andersen et al., 2009; Cherubini et al., 1998; Linares et al., 2010). However, the influence of this variable on growth, and thus on the sensitivity of growth to climate, is often confused with others variables, especially with the influence of climate itself (G´ omez-Aparicio et al., 2011; Rubio-Cuadrado et al., 2020b; S´ anchez-Salguero et al., 2015). Tree-ring width data are widely used to assess climate-growth relationships (Fritts, 1976). One of the basic methodologies in dendroecology consists of extracting the common signal, which is considered to be mainly due to the climate, from the growth series of a group of trees of the same species and sampled in the same stand (stand approach), reducing as much as possible the unwanted variability (the noise, that is, the tree-to-tree variability not due to climate) (Fritts, 1976). For this purpose, researchers usually build robust averages of detrended individual growth series. This reduces the noise and considerably increases the climate signal, creating an artificial inflation to the climate sensitivity that can lead to unrealistic conclusions about the effect of climate on growth (Carrer, 2011). However, by using mean values we are losing all the information retained in the growth variability that occurs in each tree. On the contrary, by using individual growth series (tree-level approach) it is possible to analyze differences in sensitivity due to differences in competition, age, size, sociological class, genetic pool, microenvironmental conditions, etc., and analyze these differences statistically (Galv´ an et al., 2014; Rozas, 2014). In this study, we seek to disentangle the effect of these variables on the sensitivity of radial growth to climate. To this end, we compare four stands of Scots pine (Pinus sylvestris L.), one of the most important tree species in Eurasia in terms of extension, ecological value and timber production. The stands were selected based on their contrasting characteristics: young afforested, young of natural origin, old afforested and old of natural origin. We analyze the differences in growth sensitivity to climate both at the stand and tree levels. Our objectives are (i) to know the effect of origin, age, diameter and competition on the sensitivity of growth to climate, measured through growth-climate relationships and by considering the impacts of climate extremes such as droughts on growth, and (ii) to analyze the advantages or disadvantages of the stand and tree-level approaches. 2. Material and methods 2.1. Study area The study area was located in “Sierra de Guadarrama” National Park (central Spain), where Scots pine forests dominate the landscape at mid to high elevations (1350–2000 m a.s.l.). Within these forests we looked for 4 stand types with contrasting characteristics: planted stands where most of the trees were less than 70 years old (young afforestation stand hereafter); young stands of natural origin (young natural stand hereafter); planted stands where most of the trees were at least 100 years old (old afforestation stand hereafter); and stands similar in structure and age to an old-growth forest (old natural stand hereafter) (Table 1). All these stands have been historically managed by shelterwood cuttings, although in the old natural stand the interventions have been minor and it is currently assigned to a protection district. The climate in the study area is continental Mediterranean with average temperature of 9.2 ◦C and annual precipitation of 687 mm (average data for the period 1950–2015, see Section 2.4). Soils are siliceous and acid in the four stands. Shrubs such as Genista florida L., Juniperus communis L., ´ A. Rubio-Cuadrado et al. Agricultural and Forest Meteorology 355 (2024) 110148 3 Vaccinium myrtilus L. and Cytisus oromediterraneus Rivas Mart. dominated the understory. 2.2. Field sampling We carried out different sampling strategies in each stand type. For the young afforestation stand, we sampled 41 plots of 9.8-m radius (1.23 ha in total of sampled surface) in 2016 in a homogeneous area located between 40⁰ 47 ′ 54 ′ ’ – 41⁰ 1 ′ 50 ′ ’ N and 3⁰ 52 ′ 53 ′ ’ – 4⁰ 7 ′ 47 ′ ’ W, within the 1499–1810 m a.s.l. altitudinal range. All these plots were systematically distributed on a square grid of 250 m x 250 m and were located in young forests, although they were afforested in different years. For the old natural stand, we sampled 5 plots of 25-m radius (0.98 ha in total) in 2018 around the coordinates 40⁰ 47 ′ 21 ′ ’ N and 4⁰ 2 ′ 17 ′ ’ W, within the 1830–1960 m a.s.l. altitudinal range, located in areas with complete canopy cover and no apparent disturbance. Finally, for the young natural and old afforestation stands, we sampled a single plot in each stand of dimensions 60 m x 85 m (0.51 ha) and 70 m x 70 m (0.49 ha), respectively, located at coordinates 40⁰ 49 ′ 6 ′ ’ N, 4⁰ 1 ′ 24 ′ ’ W and 41⁰ 2 ′ 46 ′ ’ N, 3⁰ 51 ′ 37 ′ ’ W, respectively, and at an altitudes of 1355 and 1448 m a.s.l., respectively. The separation between the two furthest plots sampled was 35 km, having plots of young forest stands sampled along this entire length. Sampling designs and altitudes were very different from stand to stand, but this is due to their specific characteristics. While the young afforestation forests are widely distributed in the study area, the other stand types are found in very limited areas. In addition, the old natural stand was located in a forest area with many gaps and a highly variable structure, which prevents the creation of a large plot and makes it necessary to make a non-systematic selection of smaller plots. In this stand, plots were located in areas without gaps and with a homogeneous stand structure of adult trees (Fig. S1), making their results comparable to those of the other stands. In all plots, we measured diameters at breast height (at 1.3 m; DBH) and height for all trees, as well as azimuth and distance from the plot center, except in the young afforestation stand plots where we measured DBH only. In addition, we cored between 31 and 51 dominant or codominant trees within plots in each stand type, using a Pressler increment borer. In the young afforestation stand, we selected the bored trees from the plot center (1 tree per plot) so that the competition measured in the plot is representative of the competition experienced by the cored, focal tree. Two cores were extracted at 1.3 m from each sampled tree. 2.3. Growth data Cores were taken to the laboratory, air dried, mounted on wooden supports and carefully sanded until tree rings were clearly visible. Tree rings were visually cross-dated and measured to the nearest 0.01 mm, using the semi-automatic LINTAB device with the TSAP-Win software (RINNTECH, Heidelberg, Germany), obtaining the tree-ring width (TRW) series for each tree. Cross-dating was further verified with the COFECHA program (Holmes, 1997). TRW series were subsequently transformed into basal area increments (BAI) as this variable is better than tree-ring width for capturing growth trends and accounting for the increase in tree size and age (Biondi and Qeadan, 2008). The BAI was calculated as in the Eq. (1). When cores did not contain the pith, the missing core length was estimated to calculate the BAI by subtracting the core length to the tree radius measured in the field without bark. Bark thickness of Scots pine (BT Ps ; in mm) was estimated based on DBH (in mm) using the Eq. (2) fitted to the data of this species from the Second Spanish National Forest Inventory (DGCONA, 1998). The final BAI chronologies were built by averaging annual BAIs across all trees of the same stand types. BAI = π (r2 t−r2 t−1)(eq. 1) BTPs =0.0951 DBH +10.82 (eq. 2) where r t and r t-1 are the stem radius at the end and at the beginning of a given annual ring increment. Since the BAI series may contain long-term trends due to non-climate variables (i.e. age or changes in competition), to determine the main climatic drivers of tree growth we used ring-width indices (RWI), removing the long-term trends of the raw tree-ring widths (TRW) for each stand using the dplR package (Bunn et al., 2016) in the R statistical software (R Development Core Team, 2022). For this end, we fitted a 32-year cubic smoothing spline to each TRW series, subsequently the TRW series were standardized by dividing them by the fitted spline, and then detrended series were pre-whitened fitting an autoregressive time series model (AR1 model). The residuals of this model were divided by its mean to yield a series with white noise and a mean of one (RWI). Finally, to obtain the RWI of each stand, averages of the individual RWI series were calculated using bi-weight robust means. This method efficiently eliminates long-term growth trends, such as biological trends or trends due to gradual changes in competition, as well as the effect on growth of those events that produce multi-year changes, such as thinning, but also eliminates the effects of long-term climate trends while preserving the short-term effects of annual non-climatic events, such as pests. To estimate tree age, the ring widths in the missing part of each core were considered to be equal to the average width of the 5 innermost rings adjacent to this missing part. The missing core length was estimated in the same way as for the BAI calculation. The statistical quality of each chronology was checked via Expressed Population Signal (EPS) (Wigley et al., 1984) using dplR package (Bunn et al., 2016). Mean sensitivity of the detrended (but not pre-whitened) TRW series (Speer, 2012) and signal to noise ratio of the RWI series (Wigley et al., 1984) were also obtained. The current differences in BAI between sites were Table 1 Description of structural and dendrochronological variables (mean values) of cored trees for each stand type studied (young/old/natural/afforestation stands). Values in parentheses indicate standard deviations. Mean correlation with master refers to mean correlation of detrended growth series with master. Note that standard deviations are not shown for plot basal area in young natural and old afforestation stands because only a single plot was sampled in these stands. Young Natural Young Affor. Old Natural Old Affor. Number of sampled cores 63 74 102 62 Number of sampled trees 32 41 51 31 First year with growth data 1951 1970 1582 1890 Last year with growth data 2021 2015 2017 2021 Mean age at 1.3 m (years) 59 (8) 35 (7) 263 (162) 126 (6) Mean DBH (cm) 29.3 (5.5) 25.1 (7.2) 55.3 (14.7) 44.1 (5.2) Mean correlation with master 0.57 0.55 0.48 0.64 Mean inter-trees correlation 0.42 0.37 0.27 0.53 Mean tree-ring width (mm) 2.22 (0.86) 3.25 (1.63) 0.99 (0.73) 1.58 (1.08) Mean BAI (cm 2 ) 15.68 (6.99) 15.12 (11.02) 15.36 (10.93) 14.76 (8.01) Mean BAI 2006–2015 (cm 2 ) 16.73 (2.60) 14.66 (2.25) 11.67 (2.89) 9.34 (2.14) Mean sensitivity 0.18 0.24 0.24 0.26 Expressed Population Signal 0.95 0.94 0.90 0.97 Signal to noise ratio 17.41 16.05 8.81 30.86 Plot basal area (m 2 ha −1 ) 50.18 33.53 (13.03) 49.54 (3.77) 41.16 ´ A. Rubio-Cuadrado et al. Agricultural and Forest Meteorology 355 (2024) 110148 4 analyzed through ANOVA and Tukey post-hoc tests (after verifying that the normality assumption is met), based on the average growth for each stand in each year of the period 2006–2015. 2.4. Climate data Due to the lack of long and complete series of local meteorological data from stations located in the study area, we used daily precipitation (P) and maximum (Tmax) and minimum temperature (Tmin) data from the 0.11⁰-gridded Spain02-v5 dataset (Herrera et al., 2016, 2012). This dataset reproduces climatological features, including weather extremes, thanks to the dense network of weather stations it uses, and covers from 1950 to the present. The Spain02-v5 dataset uses more than 3000 precipitation stations and over 250 temperature stations located along the Spanish surface. We obtained climate data from Spain02-v5 for the central coordinates of our study area. To assess drought intensity we used the Standardised Precipitation- Evapotranspiration Index (SPEI) (Vicente-Serrano et al., 2010a, 2010b) calculated from the Spain02-v5 climate dataset using the Hargreaves equation (Droogers and Allen, 2002) to estimate the monthly reference evapotranspiration (ETo) within the SPEI package in R (Beguería and Vicente-Serrano, 2017). High and low SPEI values correspond to wet and dry conditions, respectively. The SPEI drought index considers the effect of temperature on evapotranspiration rates and calculates a cumulative water deficit. The period from the arrival of water input (precipitation) to availability of a given usable resource (water absorption by the roots) differs considerably. Thus, the time scale over which water deficits accumulate becomes extremely important to calculate the SPEI. For this reason we used a range of time scales (from 1 to 20 months) during which the water deficit and surplus are accumulated (Vicente-Serrano et al., 2010a). For the rest of the document and for simplicity, we use the initials of the variable together with the time scale. Thus SPEI1, SPEI2… and SPEI20 refer to monthly SPEI variables calculated with a time scale of 1, 2… and 20 months. 2.5. Stand-level analyses 2.5.1. Climate-growth relationships We calculated Pearson correlation coefficients for the common period between the RWI of each stand and mean monthly Tmax and Tmin, total monthly P and the SPEI using time scales from 1 to 20 months. The common period was from 1972, the first year with growth data in more than one tree in the young afforestation stand, until 2015, the last year with growth data in this same stand. Pearson correlation coefficients were calculated from October of the previous year to September of the growth year considering that: (i) tree growth is affected by the previous year climate, and (ii) growth of Scots pines may occur from April to September usually peaking in May-June (Aldea et al., 2017; Camarero et al., 2010, 1998; S´ anchez-Salguero et al., 2015). Radial growth is a continuous process that is not limited by monthly boundaries. Furthermore, the study of growth-climate relationships can give very different results depending on whether continuous periods of time are used or whether the climate of each month is analyzed separately (Rubio-Cuadrado et al., 2022). To address this issue, we calculated, for each climatic variable (Tmax, Tmin and P) and stand, the time period in which the growth-climate relationship is maximal (best climate window) using the climwin R package (Bailey and van de Pol, 2016; van de Pol et al., 2016), which offers important advantages in the field of dendroecology (Rubio-Cuadrado et al., 2022). To select the best climatic window between a RWI stand series and a given climate variable, all possible linear models relating them were first fitted. In each model a different climate window with daily temporal resolution was tested. Then, the model that minimizes ΔAICc was chosen (Burnham and Anderson, 2004). ΔAICc is the difference between the corrected Akaike information criterion (AICc) of the selected model and the AICc of the baseline model. For the RWI series of each stand, we fitted both simple models, including a single climate variable in each model (in this case the baseline model was the null model that contains only the intercept), and multiple models. To create the multiple models, we first introduced the variable and the climate window with the lowest ΔAICc (i.e., we started from the best simple model, which was in this case the baseline model) and then added one by one the rest of the possible variables, in order to fit all possible two-variable models, and for each variable we tested all the possible climatic windows. Three-variable models were also fitted by repeating the process and in which the baseline model for each RWI stand series was the best two-variable model. The limits established for the beginning and closing of the climate window (windows open and close, respectively) were the same as those established in the monthly climate analysis (from October 1 of the previous year to September 30 of the growth year). The models were also fitted for the same period used in correlation analyses, i.e., 1972–2015. We used the mean (for minimum and maximum temperatures) and the total (for precipitation) of each time window considered as the aggregate statistic. We established a minimum period of 15 days between windows open and close. The large numbers of models that were fitted to test all possible climatic windows increases the possibility of obtaining models with low ΔAICc by chance. To solve this problem, randomization tests were performed on all the models obtained using 1000 repetitions. The randomization test determines the expected distribution of ΔAICc values in a data set where no response to climate exists, and provides a probability value (p AICc) which determines the likelihood that the ΔAICc value of the selected model has occurred by chance (van de Pol et al., 2016). As randomization tests require a high computational power, we used the “Magerit” High Performance Computer of “Universidad Polit´ ecnica de Madrid” (www.cesvima.upm.es, Madrid, Spain) to derive these analyses. 2.5.2. Growth response We analyzed the sensitivity of the growth series to abrupt environmental changes, whether these changes are due to climate extremes (e. g., droughts) or to forest disturbances (that imply changes in competition). The responses to climate extremes was analyzed by defining pointer years and calculating growth resistance and resilience indices (Lloret et al., 2011), while the responses to disturbances was analyzed by detecting growth releases (Nowacki and Abrams, 1997). We considered as negative/positive pointer year those years with a TRW decrease/increase of at least 25 %, relative to the average TRW of the 4 preceding years –relative growth change method (Schweingruber et al., 1990)– in at least 60 % of the sampled trees in the stand. The growth response to these negative pointer years and the ability to recover pre-disturbance growth levels after the disturbance were estimated for the TRW series through the resistance (Rt) and resilience (Rs) indices (Lloret et al., 2011): Rt =TRWi/TRWi−4(eq. 3) Rs =TRWi+4/TRWi−4(eq. 4) where TRW i is the TRW value of the i year. These indices were calculated with 4-year pre- (i-4), and post-disturbance (i+4) periods, based on previous studies (Rubio-Cuadrado et al., 2020a, 2018), using the pointRes package in R (van der Maaten-Theunissen et al., 2015). The differences between sites were analyzed through Levene’s test (to study the differences in variance) and ANOVA test with Bonferroni post hoc test, after verifying that the normality assumption is met. To reconstruct forest disturbances, growth releases were detected through the study of abrupt growth increments (releases) using the TRADER package in R (Altman et al., 2014). Running yearly comparisons of sequential 10-year BAI means were used to quantify the growth increases using the formula (Nowacki and Abrams, 1997): GC = (M2− − M1)/M1(eq. 5) ´ A. Rubio-Cuadrado et al. Agricultural and Forest Meteorology 355 (2024) 110148 5 where GC is the growth change between preceding and subsequent 10- year BAI means, M 1 is the preceding 10-year mean BAI, and M 2 is the subsequent 10-year mean BAI. We consider that there has been a moderate or major release when GC is greater than 0.25 and 0.5, respectively. We used BAI instead of TRW series because in these comparisons spanning 20 years (10 years before and 10 years after a given year), the decreasing trend of ring widths with age, which is very pronounced in young trees, may affect the result. 2.6. Tree-level analyses Within each stand there are variations in the competition experienced by each tree, and in the size and age of trees. To control for this variability and analyze how it influences the sensitivity of growth to climate, we repeated the previous climate-growth relationships analyses, but this time using the RWI of each tree instead of those of each stand. The period 1990–2015 was considered for these analyses for all trees to make the results comparable and to be able to use the growth series of the largest possible number of trees in the young afforestation stand (of the 41 trees sampled in the stand, 34 have at least growth data since 1990 but in 7 of them their first growth is after this year). In this way, we obtained the ΔAICc and the slope of the best simple models for each tree (ΔAICc BM and Slope BM hereafter). ΔAICc was obtained in the same way as explained in Section 2.5.1, but in this case the RWI of each individual tree was related to climate and only simple models (with one climate predictor) were used. ΔAICc BM and Slope BM are therefore the series (one value per tree) of ΔAICc and slope, respectively, of the models with the best climate window (the one that minimizes the AICc). ΔAICc BM and Slope BM were obtained only for the climatic variable P, since it is the variable that shows the greatest relationship with growth in this case. We also obtained the correlations (this time by Spearman instead of Pearson due to the lack of fulfillment of the normality assumption in some trees) between the RWI of each tree and the different SPEI variables (from SPEI1 to SPEI20) and the slopes of the simple linear models relating these variables (with RWI as response variable). Within these correlations and models calculated with the different SPEI variables, we chose the results (correlation and slope) of that SPEI variable that maximizes correlation for each tree (SPEI max correlation and SPEI max slope hereafter) as representative of the sensitivity of the growth of that tree to drought. As variables that may influence on the climate-growth relationships, we have considered the diameter of the cored trees (DBH c ), their age and height, the ratio between DBH c and the average diameter of the neighboring trees (DBH n ) as a measure of their sociological status as well as of the historical competition that a tree has suffered, especially in evenaged stands, and the competition suffered by each tree measured as the total basal area of the neighboring trees (BA). For the calculation of BA and DBH n , all adult trees (DBH >7.5 cm) were considered, excluding the rest of the vegetation, which is scarce in the studied stands. Among the different plot sizes used in the field sampling, the smallest is the one used in the young afforestation stand, with a radius of 9.8 m, which is approximately twice the average crown diameter of Scots pine according to data from the Second Spanish National Forest Inventory (DGCONA, 1998). Therefore, this is the surface that has been considered to calculate DBH n and BA of the neighboring trees. Thus, DBH n and BA values were obtained directly for each of the plots sampled in the young afforestation stand. In the rest of the stands, within the sampled plots (all of them larger), 9.8-m radius subplots were established centered on each of the bored trees, and DBH n and BA values were calculated for each of the subplots. For the spatial analyses and calculations of the variables in the subplots, the spatstat package in R was used (Baddeley and Turner, 2005). To analyze the relationships between these variables and climate sensitivity, Pearson correlations were calculated for each stand by relating these variables to the ΔAICc BM , as well as to the related values of windows open, close and length (number of days between windows open and close), the SPEI max correlation, and the resistance and resilience values, both in each of the pointer years and the mean values (mean for each bored tree of the Rt and Rs values of the pointer years 1979, 1986, 1991, 2004 and 2013). In order to obtain generalizable results, we also fitted multiple regressions to evaluate the predictors that influence growth and climate sensitivity, including all the data from all the stands together. As response variables we used the log-transformed BAI, ΔAICc BM (in absolute values), Slope BM , SPEI max correlation, SPEI max slope, and mean resistance and resilience. BAI was log-transformed to meet normality assumptions since this variable has a gamma distribution. We used the absolute value of ΔAICc BM for ease of interpretation. In this way, |Δ AICc BM | and SPEI max correlation measure the sensitivity of growth to climate in the same direction, with greater sensitivity to climate when higher |ΔAICc BM | (but lower ΔAICc BM ) and SPEI max correlation values are obtained. As predictors, we used the same variables in all (saturated) models: BA, DBH n , DBH n /DBH c , age, stand origin (as dummy variable, taking values 0 and 1 for the natural and afforested stands, respectively) and the possible two-by-two interactions between them. We did not consider height because it was not measured for the Young Afforestation stand. For the BAI model, we used the growth values of the 10 years comprised in the period 2006–2015, since we consider that in general the current competition values will be valid for the last decade. Given the relationship between DBH c and age, for the models we have used the residuals of the potential function relating age and DBH c (Eq. (6)). Age =0.1222 DBHc1.8389 (eq. 6) To identify the best-supported model for each response variable we constructed all possible combinations of alternative models from the saturated model. Then we selected the best model by minimizing the AIC. The existence of multicollinearity among predictors was evaluated by calculating the variance inflation factor (VIF) (Dormann et al., 2013), which was always lower than 5. These analyses were performed in R software with the MuMIn (Barton, 2012), car (Fox and Weisberg, 2011) and effects (Fox, 2003) packages. 3. Results 3.1. Stand structure and tree growth The current basal areas in the young afforestation and young natural stands are 33.53 and 50.18 m 2 ha −1 , respectively, being the stands with the lowest and highest competition respectively (Table 1). In any case, it must be taken into account that in the young afforestation stand different plots were sampled in which the basal area varied between 7.46 and 69.68 m 2 ha −1 . Likewise, in the old natural stands, different plots were sampled and in them the basal area varied between 43.19 and 53.85 m 2 ha −1 , being, in any case, larger than in the old afforestation stand (41.16 m 2 ha −1 ). Trees of young and old afforestation stands showed larger TRW values than trees of young and old natural stands, respectively (Fig. 1), due to the lower age, size and competition of trees in these stands (Table 1). However, this growth (in TRW) is equalized within each age category (old/young) between natural and afforested stands during the last years despite differences in age, size and competition. The BAI values were similar among the four stands in the period 2006–2015, although growth rates were significantly higher in the young versus old stands (p <0.001 in both ANOVA and pairwise Tukey post-hoc tests, except for the young afforestation-old natural pair, in which p =0.069). Within each age category, natural stands also showed a higher BAI than afforested stands during this period, although these differences were not significant. The common growth signal varied greatly among the stands studied, with a maximum signal to noise ratio of 30.86 for the old afforestation stand and a minimum of 8.81 for the old natural stand, ´ A. Rubio-Cuadrado et al. Agricultural and Forest Meteorology 355 (2024) 110148 6 taking intermediate values in the young stands (Table 1). In all cases the Expressed Population Signal was larger than the 0.85 threshold, commonly considered in tree-ring studies to represent well-replicated chronologies. Stand structure varied considerably (Fig. S1). The structure was irregular in the old natural stand, with a similar density of trees for all DBH classes between 12.5 and 62.5 cm, decreasing in density above this DBH value up to the 90 cm class. In contrast, the other three stands showed a regular structure, with a greater presence of trees in the intermediate DBH classes, and with a distribution curve with positive symmetry (i.e., with values more separated from the mean to the right or to the higher DBH classes). The DBH classes present in the old afforestation, young afforestation and young natural stands are 35–65, 10–55 and 10–45 cm, respectively. 3.2. Stand-level analyses 3.2.1. Climate-growth relationships There has been an increase in drought conditions in recent decades, especially since 2000, according to the SPEI (Fig. S2). In the two young stands and also in the old afforestation stand the growth responses to climate were similar, whereas the old natural stand presented different responses to climate and correlations were lower (Figs. 2 and S3 and Tables 2 and S1). In general, in the young stands and in the old afforestation stand, growth is higher in years with rainy springs and summers and mild winters and summers (Fig. S3). On the contrary, in the old Fig. 1. Growth trends and patterns (tree-ring width, basal area increment and ring-width indices) for the four studied stand types since 1900 and sample depth (number of cored trees; right y axis in the upper panel). ´ A. Rubio-Cuadrado et al. Agricultural and Forest Meteorology 355 (2024) 110148 7 natural stand, there were hardly any significant (p <0.05) climategrowth relationships, and these only occurred during the autumn prior to the growth year (e.g., negative relationships with November precipitation and December minimum temperature). Modeling growth as a function of daily climate showed that the most important variables in all stands were spring and summer precipitation, which accounted for 37–40 % of the growth variability, except in the old natural forest where the explained variability dropped to 16 % and where the relationship of growth with precipitation (in this case of previous autumn and winter) was inverse (Tables 2 and S1 and Fig. S4). The second most important climatic variable was maximum temperature (Tmax), which explained 10–19 % of the variability, although it was only significant (p AICc <0.05) in the young afforestation stand. The period (climatic window) during which Tmax was positively related to growth varied among stands, encompassing winter and early spring in the young afforestation stand. In the other stands the climatic window was very small (about 20 days, a value close to the pre-established minimum), indicating that the result obtained may be due to randomness rather than to an actual effect of climate on growth, which is corroborated by the p AICc values. No model with more than two variables was shown because no significant p AICc values were obtained in any case. Correlation analyses between the SPEI drought index and mean RWI generally corresponded to inverse relationships for the January-April period (significant relationships in all stands except the young natural) and direct relationships with June and July SPEI (significant relationships in all stands except the old natural) (Fig. 2). In the young natural stand, the maximum correlation corresponded to July SPEI2 (r = 0.57, p <0.001). In the young afforestation stand April SPEI3 showed the minimum correlation (r = − 0.52, p <0.001), and July SPEI2 showed the maximum correlation (r =0.40, p =0.007). In the old natural stand, April SPEI4 showed the minimum correlation (r = − 0.41, p =0.006). Finally, in the old afforestation stand, February SPEI1 showed the minimum correlation (r = − 0.37, p =0.013) and the maximum correlation corresponded to July SPEI3 (r =0.41, p =0.006). 3.2.2. Growth responses to climate extremes and disturbances Considering the common time period, the afforested stands showed a greater occurrence of abrupt drops in growth (negative pointer years) compared to the natural stands (Fig. S5 and Table S2). Within the natural stands, the growth of the young stand showed greater stability, in which there was a single pointer year and where the variability of the resistance indices was smaller. Considering all the negative pointer years shared by the two afforested stands (1979, 1986, 1991, 2004 and 2013), both stands showed similar resistance indices, but the old afforested stand showed higher resilience despite also showing higher variability in Fig. 2. Pearson correlations calculated by relating mean ring-width indices (RWI) of the four studied stand types to the SPEI drought index from January to December (y axes) and for time scales between 1 and 20 months (x axes). Only the common period 1972–2015 was considered. Correlation values higher than 0.30, 0.38 and 0.48 are significant at 0.05, 0.01 and 0.001 significance levels, respectively. ´ A. Rubio-Cuadrado et al. Agricultural and Forest Meteorology 355 (2024) 110148 8 both resistance and resilience indices (Table S2 and Fig. 3). This stand showed even higher resilience indices in 1991 and 2013 than the young natural stand, which did not show abrupt drops in growth (negative pointer years) in these years. Considering the negative pointer years shared by the old natural stand with the two afforested stands (1979, 1986 and 2013), the three stands showed similar resistance indices and, again, the old afforested stand showed higher resilience, with the old natural and the young afforestation stands having similar resilience values. Due to the absence of negative pointer years in the young natural stand, it showed higher resistance values (Fig. 3), showing however a more variable behavior in the resilience indices. The negative pointer years of 1986 and 2013 occurred in dry years (Fig. S2). Regarding positive pointer years, whether related to changes in climatic conditions or to competitive releases, these were mainly detected in the old stands (Fig. S5 and Table S2). Considering the growth releases, these occurred in all four stands, although to a greater extent in the old stands, with the old natural stand having the highest percentages of releases, the highest percentages of major releases and the highest frequency of clustered release events (Figs. S6 and S7). 3.3. Tree-level analyses The tree-level analyses show the relationships between individual variables (age, DBH, height and sociological status of the bored tree and competition suffered by it) and the sensitivity of radial growth to climate. The sensitivity was measured as ΔAICc and slope of the best models relating growth with daily climate data, the highest correlation for each tree (and the slope of the model resulting from relating the same variables) obtained between growths and the different SPEI variables, and values of the resistance and resilience indices in the pointer years. In the correlation analyses between these variables, different results were obtained for each stand and, as for the resistance and resilience variables, there were significant correlations only in some of the pointer years (Table S3). The different results obtained for each pointer year indicate a high level of variability caused by other variables (e.g. the time of year when the drought occurred, the drought intensity or the climate of previous years), variability that cannot be statistically analyzed with our data due to the small number of pointer years. In these analyses (Table S3) sociological status did not influence the sensitivity of growth to climate in old stands or in the young natural stand, being only important in the young afforestation stand, where a worse sociological status (higher DBH n /DBH c ) was related to a lower sensitivity to the mean climate (lower correlations with SPEI) and to a lower resistance and resilience. In the young natural stand, an opposite relationship was found with resilience in 2013, but this was only marginally significant (p <0.10). In general, age, as well as DBH c and height, were directly related to greater response or sensitivity to mean climate conditions and greater resistance and resilience, except in the young afforestation stand, where higher ages were related to lower resistances, and in the height-resilience relationships, which had opposite directions depending on the year considered. Although age, DBH c and height are variables related to each other, however the relationships between these variables and 2004 resilience in the Old Natural stand were opposite (rvalues of 0.33, 0.32 and −0.34 for age, DBH c and height, respectively). Finally, competition (measured as subplot basal area) showed few and opposite relationships, being directly related to resistance in the old Fig. 3. Resistance and resilience indices for the four studied stand types considering the repeated negative pointer years among the different stands. For 2013, resilience values in YouAff are missing due to lack of growth data in this stand after 2015. YouNat, YouAff, OldNat and OldAff refer to the young natural, young afforestation, old natural and old afforestation stands, respectively. Letters show significant differences (using ANOVA with Bonferroni post hoc test) of resistance and resilience indices between the different stand types. ´ A. Rubio-Cuadrado et al. Agricultural and Forest Meteorology 355 (2024) 110148 9 natural stand but inversely related to resilience in young stands. Increased competition, poorer sociological status, older age or higher height were related to an earlier opening or closing of the climatic window, although these relationships strongly varied among stands (Table S4). We built seven models including the data of all trees of the four stands (Table 3). In addition to the six variables related to the sensitivity of growth to climate considered (ΔAICc BM , Slope BM , SPEI max correlation, SPEI max slope, and values of the mean resistance and mean resilience), we used growth (BAI) as a response variable to compare the results obtained with the climate sensitivity variables with the relationships of the variables on growth, since the amount of growth is not necessarily related to the sensitivity of growth to climate (Table S5). The fact that DBH and age covaried led us to use the residuals of one of the two variables. The DBH was left in its original form as this variable was more important in all the models that have been tested, and age was transformed, as this variable was less important than DBH. According to the BAI model, the higher the DBH c and age (residual), the lower the growth rate, especially in the afforested stands (Table 3 and Fig. 4). Furthermore, these two predictors interact with each other in such a way that the higher the DBH c the lower the growth in trees with low residual age, but the relationship was inverse in trees with high residual age. Finally, the higher the competition (BA), the lower the growth in large trees (with high DBH c ), but not in small trees. It must be taken into account that, working with residual age, a tree with a higher or lower value does not correspond exactly to a young or old tree, respectively. Resistance was influenced (the model explains 33.6 % of the total variability) by all the variables considered (Table 3). Competition from neighboring trees negatively influenced resistance when the sociological status was worse (higher DBH n /DBH c ) and vice versa, i.e., a worse sociological state reduced resistance when competition was high (Fig. 5). On the contrary, in trees with high residual age this effect of both competition and sociological status was compensated or even reversed. Finally, DBH c had a negative effect on resistance in stands of natural origin (in afforested stands the effect was not significant). The resilience model explained the greatest variability (38.9 %), where DBH c and its interaction with the stand origin were the most influential variables (with the highest coefficients; Table 3 and Fig. 5). The higher the DBH c , the higher the resilience, especially in afforested stands. Higher competition (BA) and worse sociological status decreased resilience, although the effect of the DBH n /DBH c variable was compensated or even reversed by age (residual). The greater the DBH and residual age, the greater the resilience, especially in afforested stands. If interactions are not considered, the influence of the predictors was similar in the AICc BM and SPEI max correlation models (Table 3). Sensitivity to climate increased with afforested stand origin, better sociological status, smaller DBH and greater competition. However, the AICc BM model does not include interactions and explains a very small percentage of variability (3.9 %), while the SPEI max correlation model explains 22.5 % of the variability. The interactions in this model show that increasing BA increases the climatic signal in small trees (but no significant relationship in large trees) but reduces it when the sociological status is poor (high DBH n /DBH c values) (Fig. 6). DBH c showed opposite effects depending on stand origin; in afforested stands the climatic signal increased as DBH c did, while in natural stands higher values of this variable decreased the signal. Slope BM and SPEI max slope models included different predictors but their results were concordant (Table 3 and Fig. 6). Diameter and residual age strongly increased the effect of climate on growth in afforested stands but decreased it in natural origin stands. Poorer sociological status and higher competition decreased the effect of climate, although in the latter case, only with medium or high DBH c values. 4. Discussion The models that included the strength of the growth-climate relationship as a response variable (ΔAICc BM and SPEI max correlation) and those that included slope (Slope BM and SPEI max slope) showed concordant results (Table 3 and Fig. 6). Although the predictors may vary between the models, the effect of the predictors on the response variables were similar, with the sign of the regression coefficients of each predictor being maintained in almost all cases between the four models. This means that the coefficient of the same predictor (e.g. DBH n /DBH c or origin), and also the same interaction (e.g. DBH c : origin), maintains the same sign (negative for DBH n /DBH c and positive for both origin and the interaction DBH c : origin) and therefore the same effect in the four Fig. 4. Effects of the interactions between predictors on radial growth. Origin refers to the stand origin of the plots (natural or afforestation), Age (residual) are the residuals of the nonlinear model relating Age and DBH c , BA is the subplot basal area, and DBH c is the diameter at breast height of the cored trees. The predictors are standardized. Shaded bands indicate the 95 % confidence intervals. Positions of the data along x axis are denoted by tick marks. ´ A. Rubio-Cuadrado et al. Agricultural and Forest Meteorology 355 (2024) 110148 16 Monserud, R.A., Sterba, H., 1996. A basal area increment model for individual trees growing in even- and uneven-aged forest stands in Austria. For. Ecol. Manage. 80, 57–80. https://doi.org/10.1016/0378-1127(95)03638-5. Navarro-Cerrillo, R., Rodriguez-Vallejo, C., Silveiro, E., Hortal, A., Palacios- Rodríguez, G., Duque-Lazo, J., Camarero, J., 2018. Cumulative drought stress leads to a loss of growth resilience and explains higher mortality in planted than in naturally regenerated Pinus pinaster stands. Forests 9, 358. https://doi.org/10.3390/ f9060358. Navarro-Cerrillo, R.M., Cachinero-Vivar, A.M., Ruiz-G´ omez, F.J., Camarero, J.J., Gonz´ alez-P´ erez, J.A., P´ erez-Priego, ´ O., 2023. Planted or natural pine forests, which one will better recover after drought? Insights from tree growth and stable C and H isotopes. Forests 14, 573. https://doi.org/10.3390/f14030573. Ni, Y., Xiao, W., Liu, J., Jian, Z., Li, M., Xu, J., Lei, L., Zhu, J., Li, Q., Zeng, L., Cherubini, P., 2023. Radial growth-climate correlations of Pinus massoniana in natural and planted forest stands along a latitudinal gradient in subtropical central China. Agric. For. Meteorol. 334, 109422 https://doi.org/10.1016/j. agrformet.2023.109422. Niinemets, Ü., 2010. A review of light interception in plant stands from leaf to canopy in different plant functional types and in species with varying shade tolerance. Ecol. Res. 25, 693–714. https://doi.org/10.1007/s11284-010-0712-4. Nowacki, G.J., Abrams, M.D., 1997. Radial-growth averaging criteria for reconstruction disturbance histories from presettlement-origin oaks. Ecol. Monogr. 67, 225. https:// doi.org/10.2307/2963514. Orwig, D.A., Abrams, M.D., 1997. Variation in radial growth responses to drought among species, site, and canopy strata. Trees 11, 474–484. https://doi.org/10.1007/ s004680050110. Development Core Team, R., 2022. R: A Language and Environment for Statistical Computing. http://www.R-project.org/. Ripullone, F., Camarero, J.J., Colangelo, M., Voltas, J., 2020. Variation in the access to deep soil water pools explains tree-to-tree differences in drought-triggered dieback of Mediterranean oaks. Tree Physiol 40, 591–604. https://doi.org/10.1093/ treephys/tpaa026. Rozas, V., 2014. Individual-based approach as a useful tool to disentangle the relative importance of tree age, size and inter-tree competition in dendroclimatic studies. IForest 8, 187–194. https://doi.org/10.3832/ifor1249-007. Rozas, V., DeSoto, L., Olano, J.M., 2009. Sex-specific, age-dependent sensitivity of treering growth to climate in the dioecious tree Juniperus thurifera. New Phytol. 182, 687–697. https://doi.org/10.1111/j.1469-8137.2009.02770.x. Rubio-Cuadrado, ´ A., Camarero, J.J., Aspizua, R., S´ anchez-Gonz´ alez, M., Gil, L., Montes, F., 2018. Abiotic factors modulate post-drought growth resilience of Scots pine plantations and rear-edge Scots pine and oak forests. Dendrochronologia 51, 54–65. https://doi.org/10.1016/j.dendro.2018.08.001. Rubio-Cuadrado, ´ A., Camarero, J.J., Bosela, M., 2022. Applying climwin to dendrochronology: a breakthrough in the analyses of tree responses to environmental variability. Dendrochronologia 71, 125916. https://doi.org/ 10.1016/J.DENDRO.2021.125916. Rubio-Cuadrado, ´ A., Camarero, J.J., G´ omez, C., Ca˜ nellas, I., Aull´ o-Maestro, I., Gil, L., Montes, F., 2020a. Scots pine plantations growth adaptation to climate warming in locations at the southernmost distribution limit of the species. Dendrochronologia 63, 125745. https://doi.org/10.1016/j.dendro.2020.125745. Rubio-Cuadrado, ´ A., Camarero, J.J., Gordaliza, G.G., Cerioni, M., Montes, F., Gil, L., 2020b. Competition overrides climate as trigger of growth decline in a mixed Fagaceae Mediterranean rear-edge forest. Ann. For. Sci. 77, 1–18. https://doi.org/ 10.1007/s13595-020-01004-5. Rubio-Cuadrado, ´ A., L´ opez, R., Rodríguez-Calcerrada, J., Gil, L., 2021. Stress and tree mortality in Mediterranean pine forests: anthropogenic influences. In: Ne’eman, G., Osem, Y. (Eds.), Stress and tree mortality in Mediterranean pine forests: anthropogenic influences. Pines and Their Mixed Forest Ecosystems in the Mediterranean Basin. Managing Forest Ecosystems, Vol 38 141–181. https://doi. org/10.1007/978-3-030-63625-8_9. Ryan, M.G., Phillips, N., Bond, B.J., 2006. The hydraulic limitation hypothesis revisited. Plant Cell Environ 29, 367–381. https://doi.org/10.1111/j.1365-3040.2005.01478. x. Ryan, M.G., Yoder, B.J., 1997. Hydraulic limits to tree height and tree growth. What keeps trees from growing beyond a certain height? Bioscience 47, 235–242. S´ anchez-Salguero, R., Camarero, J.J., Navarro-Cerrillo, M., Dobbertin, M., Fern´ andez- Cancio, ´ A., Vil` a-Cabrera, A., Manzanedo, R.D., Zavala, M.A., Navarro-Cerrillo, R.M., 2013. Contrasting vulnerability and resilience to drought-induced decline of densely planted vs. natural rear-edge Pinus nigra forests. For. Ecol. Manage. 310, 956–967. https://doi.org/10.1016/J.FORECO.2013.09.050. S´ anchez-Salguero, R., Camarero, J.J., Rozas, V., G´ enova, M., Olano, J.M., Arzac, A., Gazol, A., Caminero, L., Tejedor, E., de Luis, M., Linares, J.C., 2018. Resist, recover or both? Growth plasticity in response to drought is geographically structured and linked to intraspecific variability in Pinus pinaster. J. Biogeogr. 45, 1126–1139. https://doi.org/10.1111/jbi.13202. S´ anchez-Salguero, R., Linares, J.C., Camarero, J.J., Madrigal-Gonz´ alez, J., Hevia, A., S´ anchez-Miranda, ´ A., Ballesteros-C´ anovas, J.A., Alfaro-S´ anchez, R., García- Cervig´ on, A.I., Bigler, C., Rigling, A., 2015. Disentangling the effects of competition and climate on individual tree growth: a retrospective and dynamic approach in Scots pine. For. Ecol. Manage. 358, 12–25. https://doi.org/10.1016/J. FORECO.2015.08.034. Schuster, R., Oberhuber, W., 2013. Age-dependent climate-growth relationships and regeneration of Picea abies in a drought-prone mixed coniferous forest in the Alps. Can. J. For. Res. 43, 609–618. https://doi.org/10.1139/cjfr-2012-0426. Schweingruber, F.H., Eckstein, D., Serre-Bachet, F., Br¨ aker, O.U., 1990. Identification, presentation and interpretation of event years and pointer years in dendrochronology. Dendrochronologia 8, 9–38. Sohn, J.A., Saha, S., Bauhus, J., 2016. Potential of forest thinning to mitigate drought stress: a meta-analysis. For. Ecol. Manage. 380, 261–273. https://doi.org/10.1016/j. foreco.2016.07.046. Speer, J.H., 2012. Fundamentals of Tree Ring Research. The University of Arizona Press, Tucson, USA. Szeicz, J.M., MacDonald, G.M., 1994. Age-dependent tree-ring growth responses of subarctic white spruce to climate. Can. J. For. Res. 24, 120–132. https://doi.org/ 10.1139/x94-017. Thomas, H., 2002. Ageing in plants. Mech. Ageing Dev. 123, 747–753. https://doi.org/ 10.1016/S0047-6374(01)00420-1. Trouillier, M., van der Maaten-Theunissen, M., Scharnweber, T., Würth, D., Burger, A., Schnittler, M., Wilmking, M., 2019. Size matters—A comparison of three methods to assess age- and size-dependent climate sensitivity of trees. Trees - Struct. Funct. 33, 183–192. https://doi.org/10.1007/s00468-018-1767-z. Tyree, M.T., Zimmermann, M.H., 2002. Xylem Structure and the Ascent of Sap, Springer Series in Wood Science. Springer Berlin Heidelberg, Berlin, Heidelberg. https://doi. org/10.1007/978-3-662-04931-0. van de Pol, M., Bailey, L.D., McLean, N., Rijsdijk, L., Lawson, C.R., Brouwer, L., 2016. Identifying the best climatic predictors in ecology and evolution. Methods Ecol. Evol. 7, 1246–1257. https://doi.org/10.1111/2041-210X.12590. van der Maaten-Theunissen, M., van der Maaten, E., Bouriaud, O., 2015. pointRes: an R package to analyze pointer years and components of resilience. Dendrochronologia 35, 34–38. https://doi.org/10.1016/J.DENDRO.2015.05.006. Vicente-Serrano, S.M., Beguería, S., L´ opez-Moreno, J.I., Vicente-Serrano, S.M., Beguería, S., L´ opez-Moreno, J.I., 2010a. A multiscalar drought index sensitive to global warming: the Standardized Precipitation Evapotranspiration Index. J. Clim. 23, 1696–1718. https://doi.org/10.1175/2009JCLI2909.1. Vicente-Serrano, Sergio M, Beguería, S, L´ opez-Moreno, J.I, Angulo, M, Kenawy, A.El, Vicente-Serrano, S.M, Beguería, S, L´ opez-Moreno, J.I, Angulo, M, Kenawy, A.El, 2010b. A new global 0.5◦gridded dataset (1901–2006) of a multiscalar drought index: comparison with current drought index datasets based on the Palmer Drought Severity Index. J. Hydrometeorol. 11, 1033–1043. https://doi.org/10.1175/ 2010JHM1224.1. Weiner, J., 1990. Asymmetric competition in plant populations. Trends Ecol. Evol. https://doi.org/10.1016/0169-5347(90)90095-U. Wigley, T.M.L., Briffa, K.R., Jones, P.D., 1984. On the average value of correlated time series with applications in dendroclimatology and hydrometeorology. J. Clim. Appl. Meteorol. 23, 201–213. https://doi.org/10.1175/1520-0450(1984)023<0201: OTAVOC>2.0.CO;2. Zadworny, M., Jagodzi´ nski, A.M., Łakomy, P., Mucha, J., Oleksyn, J., Rodríguez- Calcerrada, J., Ufnalski, K., 2019. Regeneration origin affects radial growth patterns preceding oak decline and death – insights from tree-ring δ 13 C and δ 18 O. Agric. For. Meteorol. 278, 107685 https://doi.org/10.1016/j.agrformet.2019.107685. Zhang, Y.J., Meinzer, F.C., Hao, G.Y., Scholz, F.G., Bucci, S.J., Takahashi, F.S.C., Villalobos-Vega, R., Giraldo, J.P., Cao, K.F., Hoffmann, W.A., Goldstein, G., 2009. Size-dependent mortality in a Neotropical savanna tree: the role of height-related adjustments in hydraulic architecture and carbon allocation. Plant Cell Environ. 32, 1456–1466. https://doi.org/10.1111/j.1365-3040.2009.02012.x. ´ A. Rubio-Cuadrado et al.