Meteorological Conditions and Site Productivity Modulate Genetic Controls over Radial Increment, but Not the Sensitivity of Hemiboreal Scots Pine

Abstract

The increasing stresses imposed on forests by climatic changes require agile and adaptive management, in which tree breeding plays a crucial role. Assessment of G×E interactions has been highlighted as an explicit source of information for targeted breeding for resilience and productivity via better coupling of the demands of genotypes and locally anticipated climatic changes. Commonly, the genetic effects are evaluated for morphometric traits, which are a cumulative representation of the past conditions. In this study, the variability of increment and its relation to weather fluctuations were addressed as a climate-sensitive functional trait. Parallel Scots pine (Pinus sylvestris) progeny trials representing hemiboreal conditions and a local productivity (edaphic) gradient in Lithuania were studied to evaluate G×E effects on increment, combining tree-ring analysis and quantitative genetics. Scots pine progenies showed low environmental sensitivity of radial increment, which, however, showed complex region-specific relationships with winter temperature and summer moisture regimes. Negligible genetic controls were estimated for the weather sensitivity of increment, yet the genetic controls manifested in response to the co-occurrence of meteorological extremes. The genetic control over increment showed nonlinear relationships with site productivity, being strongest under oligotrophic and eutrophic rather than extremely poor or mesotrophic site conditions. Site productivity showed marginal effects on climatic controls over heritability estimates, indicating weak environmental interactions. Hence, the observed relationships suggest some limited potential for targeted pine breeding based on local populations.

1. Introduction

The observed and projected northward shifts in tree distributions in Europe imply considerable alteration in forest composition and challenges for management [1,2]. In the eastern Baltic region and the Baltics in particular, which are substantial producers of timber in the EU [3,4], the abundance of conifers is projected to decrease as they are gradually replaced by deciduous species [5]. However, given the economic importance of conifers [6,7], adaptive management is implemented to increase the resilience of tree populations and counteract such a shift [8,9,10].

Tree breeding is playing a crucial role in adaptive management, aiming to harness advantages provided by the warming under cool climates [1,9,11] via increasing forest resilience [6,7,12,13]. Due to accelerating environmental changes, the trade-offs between productivity and resilience of forests might require agile local adjustments, thus exploiting the adaptability of populations [11,13,14]. The adaptability, in turn, depends on local genetic adaptations and phenotypic plasticity [15,16,17], which are reflected in adaptability [13,14,18].

Comprehensive assessment of the adaptability of the breeding population(s) is essential for the sustainability of breeding programmes [1,7,11]. Assessment of G×E interactions has recently been shown to have a high potential for breeding for sustainability, thus complementing information on the general genetic controls [13,14,16]. In this regard, edaphic conditions, which affect the productivity and stress susceptibility of trees [19,20], have been related to adaptability [13,18]. As the site productivity is temporarily stable, fine-tuned targeted breeding could aid the resilience of future forests [14,16].

Site productivity can mediate tree responses to weather conditions, particularly moisture-related conditions, highlighting its relevance in the context of accelerating climatic changes [18,21,22]. Under drought-limited conditions, more productive forests on well-drained soils can be increasingly sensitive to stresses due to the boost–bust cycle, while trees in poor sites can be better adapted to higher levels of stress [23,24]. In contrast, more productive growth has been linked with higher resilience, likely due to greater reserves [25], particularly where temperature controls growth [26,27]. Spatially non-stationary linear effects of soil properties on growth sensitivity have been observed in temperate lowland conditions [28], highlighting regional specifics of the edaphic/site effects.

The morphometric traits are commonly used for the selection of the reproductive material [29]; however, they are the result of the conditions in the past [11]. Considering the accelerating environmental changes, morphometric traits can be rapidly outdated [30,31]; hence, climate-related physiological traits appear promising for improvements in adaptability [32,33,34]. Analysis of increment, which represents environmental preferences of trees, appears as an informative source of climate-related functional traits [35,36]. Tree-ring width (TRW) is a common and highly informative proxy of tree growth [37,38,39], providing data at fine temporal resolution [40,41]. Lagged and continuous environmental effects, though, can increase the complexity of the controls over increments [42,43], while simultaneously acting as sources of a spectrum of adaptability traits [18,36,44].

Time series variance deconstruction is commonly used for the disentanglement of the main drivers of increment [45,46,47,48]. Specific deconstruction methods allow analysis of the responses to normal, as well as extreme environmental conditions [46,47], providing a spectrum of climate-related traits from a limited dataset [36,45]. For breeding populations, assessment of genetic correlations between traits of adaptive significance can be harnessed to reduce the complexity of information [13,29]. Although such evaluation is a routine quantitative genetic analysis [49], the increment-derived traits have been scarcely explored, particularly for G×E interactions [15,16,18].

Scots pine (Pinus sylvestris L.) is a common, economically important species in Northern Europe and in the Baltics in particular [50,51], where the most explicit changes in its abundance and growth are projected [5,52]. Due to commercial importance, conservative breeding programmes are implemented [7,29,51], and networks of progeny trials across local gradients have been developed [51,52,53]. Despite the location in the mid-part of the distribution area in the Eastern Baltic region, Scots pine is disproportionally sensitive to weather and climatic conditions [42,54]. Nevertheless, some genetic controls over the sensitivity of increment to winter thermal regime and summer drought have been screened at the provenance level [55]. Still, these estimates neglect the combinatory potential of the local populations, nor have the effects of climate on genetic controls been addressed.

This study aimed to estimate genetic control over the environmental (climatic) sensitivity of radial increment of local populations across the site productivity gradient in the Baltics. We hypothesized that radial increment and its sensitivity to the anticipated climatic changes have been subjected to local genetic adaptation and are genetically controlled at a level suitable for operational breeding for resilience. We also hypothesized that the genetic controls over increment would be stronger under poor or fertile sites, where climatic and competitional constraints are the strongest, respectively.

2. Material and Methods

2.1. Trials and Genotypes

Five parallel Scots pine progeny trials (Figure 1) established in Lithuania in 1983 [56] were studied. The trials were established to evaluate the performance of progenies of the local population (local provenances) of Scots pine across the local site productivity (edaphic) gradient (Supplementary Material, Table S1), which is common for commercial stands in the Baltics. Hence, the site productivity gradient represented freely draining extremely poor (N and P extremely deficient) to eutrophic deep mineral soils with a deep water table that have formed since the last glaciation. The texture of soil ranged from sand to silty sand, with the gravimetric water content of 8 to 19% at the poor and eutrophic parts of the site productivity gradient, respectively. The soils were podzolic.

The reproductive material was obtained from open-pollinated local plus-trees of Scots pine from sites in seven vicinities of the test trials (four local provenances). Hence, the tested reproductive material represented geographic variation in the Scots pine population in Lithuania [57], allowing evaluation of the adaptive potential of native genotypes. Seed material at each of the seven sites (local provenances) was collected from 20 plus-trees (20 half-sib families). In total, 140 open-pollinated families were initially tested at each trial. The topography of the trials was flat, and the elevation ranged from 20 to 130 m a.s.l., thus explicitly representing lowland conditions common in the region.

The tested progenies were planted in randomized row plots of 10 trees (1 × 10) with a spacing of the grid of 1.5 × 1 m or 1.5 × 2 m. Each trial contained five replications. One-year-old containerized seedlings, raised in local production nurseries, were planted. Prior to planting, the soil was prepared by disc trenching. Weed control was performed a few years after the establishment in the mesotrophic and eutrophic sites. Veisieju, Ignalina, Nemencine, and Latezeris trials were not thinned before sampling. The Silute trial was thinned from below in 2016; ~30% of trees were cut, yet efforts were made to maintain the block structure (design). The latest inventory was conducted in 2016 (after the thinning in the Silute trial), and tree dimensions were measured.

The inventory showed that 134 of the initially planted 140 families had survived, yet the dimensions of trees varied greatly among as well as within the trials, indicating plasticity of the genotypes in terms of survival and growth [58]. According to the inventory, the mean tree stem diameter at breast height and height of the trials (±st. dev.) ranged from 7.8 ± 2.9 to 14.1 ± 4.1 cm and from 7.7 ± 1.4 to 15.2 ± 1.5 m in Ignalina (poor) and Nemencine (mesotrophic) trials, respectively, representing a productivity gradient. The survival of trees ranged from 38.6% to 56.4% in the Nemencine and Latezeris trials, respectively.

The climate in the studied trials was temperate moist continental [59] with coastal features, as determined by proximity to the Baltic Sea (Figure 1, Supplementary Material, Table S1). According to the gridded climatic dataset (CRU TS 4.09) [60], the mean annual temperature ranged from 7.0 ± 0.9 to 8.1 ± 0.9 °C in Ignalina and Silute trials, respectively. The annual precipitation ranged from 615 ± 71 to 798 ± 106 mm in the Veisieju and Silute sites, respectively. February and July were the coldest and warmest months, with the mean monthly temperature ranging from −2.9 to 18.8 °C, respectively. The highest monthly precipitation occurred during the summer months (May–August) with a mean value of 74 mm. The climatic changes were expressed as warming during the dormancy period, hence the extension of the vegetation period, which was coupled with increasing heterogeneity of the summer moisture regime and intensification of hot droughts [61,62].

2.2. Sampling and Measurements

Considering the labour intensity of the processing of the material (increment cores) and potential hazards from the invasive sampling for the remaining trees, a stratified subset of families was selected for sampling based on the trial inventories. In 2017, a random subset of 20 families, which were represented by at least 30 trees per trial in at least four of the trials, was selected for sampling. In 2025, an additional 27 families, which were represented by at least ten trees in at least four of the five trials yet were not sampled in 2017, were randomly selected, complementing previous material. For each family in each trial, one to four trees per replication (block) were selected for sampling to meet the initial minimum intention of sampling of 25 and ≥9 sampled trees per family per trial in 2017 and 2025, respectively. As the sensitivity of the increment derived from measurement time series for trees at the maturing age was scrutinized, the datasets from both campaigns were considered fully compatible.

In each replication, visually healthy straight trees were sampled. In 2017, one increment core at breast height was taken from a random side of the stem with a 5 mm increment borer. In 2025, two cores per tree from the west and east (±30°) sides of the stem were taken at breast height with a similar borer. In the laboratory, sampled cores were fixed in wooden mounts, and their surfaces were prepared for measurements (levelled) either by an automated mill or by grinding, using progressively finer sandpaper. In 2017, to measure TRW, high-resolution (9600 dpi) images were acquired by an automated microscope camera setup (LignoStation; RinnTECH, Heidelberg, Germany), and measurements were done using LignoVision v. 2.4 (RinnTECH, Heidelberg, Germany) software in a semiautomated (supervised) regime. In 2025, measurements were done manually, using the LINATB6 (RinnTECH, Heidelberg, Germany) measuring table.

2.3. Data Analysis

As the measured time series of TRW were rather short (up to 40 years), their quality was primarily checked via graphical and then statistical crossdating [63] at family and trial levels. Corrections were introduced in dating when necessary. Mean time series for trees were calculated for the data acquired in 2025, thus harmonizing the datasets of both sampling campaigns. Considering the limited length of the time series, age and disturbance effects were removed by detrending using a fixed-wavelength (25 years) flexible cubic spline at a 50% cut-off level. The fixed wavelength was used to reduce bias due to series length; 25 years were used as a wavelength, allowing for high flexibility of the spline to remove non-meteorological effects [64]. Hence, such detrending highlighted weather-driven high-frequency variation in TRW, as the timespan was not sufficient for the assessment of medium-frequency variation.

For the description of environmental fluctuations captured by the datasets (families by trial), sensitivity and agreement metrics (e.g., Gini coefficient, r-bar, signal-to-noise ratio, etc.) were calculated based on detrended series [63,65]. As the autocorrelation in TRW was generally low, time series were not prewhitened. The biweight robust mean was used to average the detrended time series. To evaluate the effects of local population (local provenance) and site productivity (trial), as well as their interaction on the sensitivity and agreement metrics of increment at the family level, linear models were used. Replication covariates (number of trees and number of tree rings per tree) were included to account for heterogeneity of the data.

To evaluate the sensitivity of radial increment to meteorological fluctuation, which could be used as a climate-related functional trait for heritability analysis [36,55], the degree of linear dependence between detrended TRW indices and meteorological variables was quantified using bootstrapped Pearson correlation coefficients (1000 iterations) [66]. Bootstrapped correlation was used to account for “normal” variation, omitting the effects of outliers, thus ensuring the stability of the coefficient estimates. As the correlation coefficient was used as the metric of linear dependence for variance separation, the statistical significance of the coefficients was not estimated.

The tested meteorological variables were monthly mean temperature, climatic water balance (CWB), and standardized precipitation evapotranspiration index (SPEI) [67], which represented drought intensity with respect to the preceding three months. The CWB was calculated as the difference between monthly precipitation and potential evapotranspiration [60] to represent the amount of available moisture. The monthly climatic variables were arranged according to the climatic window from June in the year preceding tree-ring formation to August in the year of tree-ring formation, thus accounting for the carryover climatic effects. Gridded meteorological data for the grid points closest to the trials were used [60]. The analysis was performed for the common interval from 1990 to 2024 (2017 for the earlier sampling) at the tree level.

Considering that meteorological extremes can trigger genetic effects [55], pointer year indices [46] based on relative changes in increment with respect to four years were calculated to assess the presence of “abrupt” systematic changes in increment at the trials (TRW series of families pooled). As the trees were growing under managed competition conditions of a trial, the threshold values for event years were kept low (35% for growth change and 55% for the proportion of trees showing changes). Rolling 30-year z-scores were calculated for meteorological variables to identify extremes (|z-score| > 2.0). The resilience indices, which quantify the resistance and recovery of increment to environmental stresses (“RRR” [45], were calculated for each tree time series of TRW for the assessment of genetic control over the responsiveness to environmental events. Namely, resistance, recovery, resilience, and relative resilience indices [68] were calculated. As plants are constantly subjected to environmental stresses [36,43], the indices were calculated for the entire series length with respect to four years before/after.

To evaluate the strength of genetic control over increment (tree-ring width and its indices), weather sensitivity (climatic correlation), and resilience (resilience indices) as climate-sensitive functional traits, genetic and environmental variance components were partitioned using simple linear mixed-effects models. The analysis was done for each trial, as the interannual patterns of increment had notable site specifics. The models for climatic sensitivity represented by tree-level climatic correlations were as follows:

$${\mu }_{ijk}=\mu +r_i+c_j+\left(f_k\right)+{\epsilon }_{ijk},$$

(1)

where μ_ijk is the trait value of an individual tree, μ is the overall mean, r_i is the fixed effect of the i-th replication within a trial, c_j is the climatic offset of the j-th tree (mean value of the respective climatic variable for the period represented by the TRW series, for which the correlation was calculated), f_k is the random effect of family, and ε_ijk is the residual error. Models were also supplemented with weights, which were the number of observations (years) for each tree individually, for which the respective correlation was calculated, thus accounting for the unbalancedness due to the differing length of the time series underlying the estimates of correlations. The climatic offset was used to account for the shifts in the baseline of meteorological variables due to climatic changes for the periods for which the correlations were calculated.

For other quantitative traits (increment and resilience), excluding climatic sensitivity (climatic correlations), the models were simplified, omitting the climatic offset and weights. For increment and resilience coefficients, the analysis was done for each year separately. As the time series showed low autocorrelation, previous growth was not accounted for. As a result, time series of annual heritability coefficients were estimated. Additionally, the analysis was conducted across trials, in which case the model was supplemented with a non-interacted fixed effect of trial and a random interaction between family and trial to account for G×E interaction [13].

$${\mu }_{ijkl}=\mu +t_l+r_{i\left(t_l\right)}+c_j+f_k+f_k\times t_l+{\epsilon }_{ijkl}$$

(2)

where t_l is the fixed effect of the l-th trial and f_k × t_l is the random family by trial interaction, which was used to account for genotype-by-environment interaction.

Narrow sense heritability coefficient (h²) was calculated as the ratio of family variance to the sum of family and random variance. As the progenies originated from open pollination of known plus-trees, families were treated as half-sib families, assuming predominantly outcrossed mating, unrelated or weakly related pollen parents, and negligible dominance, epistatic, and common environmental effects. Though some (minimal) maternal effects might have affected the estimates. The standard error for the heritability was calculated according to Dickerson’s approximation [49,69]. In case the h² coefficients were calculated across the trials, the G×E variance was included in the denominator.

To assess the correlations between climatic conditions and time series of annual heritability estimates (genetic controls) of radial increment and its resilience components, as well as the correlations among the heritability time series, bootstrapped Pearson correlation analysis was used. The tested set of meteorological variables was the same as that used in the weather-growth sensitivity analysis, including carryover climatic effects. The analysis was done at the trial level due to the local specifics of growth patterns. Thus, such analysis provided insight into potential modulating effects of weather/climatic conditions on the physiological mechanisms of coping with climatic stresses [36,43]. Data analysis was conducted in R (v. 4.5.2) [70], using libraries dplR [63], pointRes [46], and lme4 [71].

3. Results

3.1. Statistics of the Dataset

According to the inventory (2016, tree age 32 years), the mean tree stem diameter at breast height of the sampled trees (±st. dev.) ranged from 9.6 ± 2.3 to 16.3 ± 3.3 cm in Latezeris and Nemencine trials, respectively. The mean tree height ranged from 8.4 ± 1.1 to 15.8 ± 1.2 m in Ignalina and Nemencine trials, respectively; hence, the dimensions of sampled trees were representative of trial means. The phenotypic variation was two times higher for stem diameter than for tree height. The TRW of the studied trees showed generally low inter-annual variation, as indicated by the standard deviation of the index values (Figure 2). The replication of TRW of families mostly exceeded 10 trees per trial; however, data subsets with lower replication (from three to eight trees) were quite common, as lower replication was allowed in one of four sites.

The environmental sensitivity of TRW, as shown by the mean sensitivity and Gini coefficients, was moderate [63]. The effects of previous growth in the interannual variation in TRW, as indicated by the autocorrelation, peaking at 0.48 for a single dataset, were generally low (Figure 2). The interseries correlation was moderate to strong [63] and, similar to the signal-to-noise ratio, showed high variability among the families within the trials, indicating varying environmental forcing of increment. The EPS values varied around 0.85 [65], with lower values being quite common, and the signal-to-noise ratio ranged widely.

Trial (soil fertility) showed significant relationships with all of the descriptive metrics of TRW except the standard deviation of indices (

Table 1. The strength (F-values) and statistical significance (p-values; asterisk codes) of the fixed effects of population (local provenance), trial, their interaction, as well as replication (number of trees and time series length, as numeric covariates) on family estimates of tree-ring metrics for open-pollinated progenies of plus trees of Scots pine plus trees from seven locations across Lithuania representing local variability of genotypes for the period 1990–2024. The metrics are calculated based on a spline detrended time series of tree-ring width (model ANOVA-like table). SD—standard deviation, AR1—first-order autocorrelation, SENS—the mean sensitivity of the series, GINI—the mean Gini coefficient, r-bar—the mean interseries correlation, EPS—expressed population signal, and SNR signal-to-noise ratio. Significance codes, p-value: * < 0.05, ** < 0.01, *** < 0.001.

	SD	AR1	SENS	GINI	r-bar	EPS	SNR
Population	1.4	1.6	2.0	2.4 *	0.8	0.2	1.3
Trial	1.2	11.8 ***	74.9 ***	14.1 ***	9.2 ***	3.6 **	7.8 ***
Replication	0.2	1.2	0.2	0.3	5.0 *	180.3 ***	146.3 ***
Tree rings per tree	10.1 **	41.3 ***	88.6 ***	41.9 ***	218.7 ***	186.6 ***	213.3 ***
Population by trial interaction	0.5	1.2	1.5	1.4	1.0	1.0	1.2
R2	0.078	0.318	0.67	0.417	0.577	0.523	0.569

). Higher similarity between the TRW series of families, as indicated by higher r-bar, EPS, and SNR, was estimated in the oligotrophic trials (Figure 2), particularly in Latezeris. The interannual variability of TRW, as indicated by mean sensitivity and Gini coefficients, was higher in the Veisieju trial. Population (local provenance of the genotypes) had a consistent (non-interacted) effect only on the Gini coefficient, indicating genetic control over the environmental responsiveness of increment.

3.2. Interannual Variability and Weather-Growth Relationships

The family-level chronologies of TRW generally had low interannual variation, hence limited variability of increment besides the age trend (Figure 3). The variation in TRW showed local specifics, although mean chronologies of the oligotrophic trials (Latezeris and Veisieju) had common signatures. Within the trials, families showed similar variation in TRW, as indicated by the interseries correlation, which ranged from 0.54 to 0.78 in Nemencine (mesotrophic) and Latezeris (oligotrophic) trails respectively. Despite the synchrony, the spread of index values among the families was temporally and spatially variable, although it increased with age. Higher spread of index values occurred in some years with increased or decreased TRW, as well as following suppressed TRW, indicating varying recovery. The opposite, however, was observed in 2000 in the oligotrophic trials.

The family-level chronologies showed local correlations with the meteorological variables, which were generally weak (below the frequency significance level at α = 0.05), as growth showed low environmental sensitivity. The strength of the linear dependencies indicated by the correlations tended to be higher at the extremes of the site productivity gradient (Figure 4). The variability in correlation coefficients among the families, however, was higher in Ignalina and Nemencine trials. Across the trials, TRW showed correlations with variables representing the year of tree-ring formation, as well as the year before it, indicating direct and carryover effects of meteorological conditions.

In the Ignalina (poor) trial, the majority of families showed similar correlations with moisture and temperature-related variables at the end of summer and autumn before the formation of increment (Figure 4). Then temperature showed negative and moisture level showed a positive correlation. During the dormancy period, these relations tended to invert, with temperature showing positive and precipitation-related variables showing negative correlations. In contrast, in May, temperature showed mostly negative correlation with radial increment, and signs of water shortage in summer were indicated by correlations with precipitation-related variables. In the oligotrophic trials (Figure 4), negative correlations with temperature and positive correlations with precipitation-related variables in March–May were observed. During the February–August period, the correlations showed an inversion of sign, indicating fluctuating limitations. Correlations indicating carryover effects were weaker, with sole families showing negative effects.

In the mesotrophic and eutrophic trials (Figure 4), the strongest correlations were estimated for temperature in the dormancy period (December and March) and precipitation-related variables in the previous vegetation season, which, however, were highly variable among families. The correlations with conditions during the growing period were weaker than in the poor and oligotrophic sites; still, some signatures of summer water shortages (negative correlations with temperature and positive with precipitation-related variables) were evident. In the eutrophic (Silute) trial, summer water deficit was indicated by positive correlations with SPEI in July. Still, the effect of moisture availability had a carryover effect on TRW, as indicated by a negative correlation with SPEI in the previous June. Across the sites, the correlations with water balance and drought indices were not always coherent, indicating varying temporal scales of the effects of precipitation (SPEI and CWB) throughout the year.

The pointer year analysis conducted for the trials showed that common systematic changes (reduction) in TRW during the reference period occurred only in the oligotrophic traits (Figure 3 and Figure S1). In the remaining trials, none of the years were identified as a significant pointer year, indicating high variability of growth responses. The general lack of pointer years, as the radial growth was of low sensitivity, also indicated the lack of influence of environmental extremes. Nevertheless, the significant pointer years in the oligotrophic traits occurred in 2000, 2006, and 2013, when, according to the gridded meteorological data, March temperature was extremely low, or April was extremely warm (|z-score| > 2.0), which was preceded by extremely warm and dry late summers.

3.3. Genetic Control of Growth, Resilience, and Sensitivity Traits

The genetic control over the radial increment of local Scots pine, as indicated by the heritability coefficients, was low to moderate (h² ≤ 0.42; Figure 5). The mean heritability of TRW, which is a proxy of tree size, remained similar (0.07–0.09) irrespective of trial. The mean heritability of TRW indices, which represent the responsiveness of increment, ranged from 0.03 to 0.08 in Silute and Nemencine trials, respectively. In some years, the heritability of both TRW and its index exceeded 0.15 cf. [13,18], particularly in the oligotrophic trials. In most cases, the estimated Dickerson’s standard errors were lower than the heritability coefficients, indicating their relevance, particularly for the peaks. Still, in some years, the heritability of the increment was effectively zero. The uncertainty indicated by the standard error, however, was higher for tree-ring width than for its index, indicating higher phenotypic plasticity. The decreasing sample size for 2018–2024 tended to increase the standard errors, thus increasing the uncertainty of the heritability estimates of increment.

The time series of heritability estimates of radial increment showed locally specific variability (Figure 5). In the nutrient-poor Ignalina trial, the heritability of the TRW index showed the highest (moderate) values in 2001 and particularly 2012, years with cold December, February, and June, and preceded by warm late summer. In the oligotrophic Latezeris trial, the heritability of the TRW index (and TRW) peaked in 1997 and 2013, when the preceding December and April were cold, yet the preceding November was extremely warm. Also, these years had a dry August. The dating of the peaks in heritability in the oligotrophic Veisieju trial differed from Latezeris despite the similarity of the pattern of increment, and the peaks occurred in 2006 and 2016, as well as in 2012 and 2024 (for TRW only). These years were associated with extremes (both positive and negative) in January and February temperature, as well as May or June precipitation. For the more fertile part of the site productivity gradient, heritability of radial increment tended to be lower, and the time series lacked apparent peaks, irrespective of co-occurrence of meteorological anomalies (Figure 5). Still, in the mesotrophic Nemencine trial, the strongest peak in heritability (reaching 0.42) was observed in 1991, when trees were young, and a weaker peak (exceeding 0.15) also occurred in 2016 (for TRW only). These years were associated with extremes in temperature in March and the preceding October and December.

The heritability estimated for the resilience components was slightly higher than for increment, yet some peaks were stronger, indicating periodic engagement of genetic control in the resilience of increment (Figure 6). Similar to radial increment, the time series of heritability estimates for the resilience components were trial-specific. In the nutrient-poor Ignalina and Latezeris trials, the heritability estimates were low to moderate throughout the common interval and did not show explicit peaks. In the Latezeris trial, two weaker peaks in heritability of resilience indices were estimated in 1997 and 2013, and for recovery in 1997, 2013, and 2016, when temperature in the preceding November–December, March, and June had extremes (Figure 6). Although situated in oligotrophic conditions, considerably higher peak values of heritability of the resilience of TRW were observed in the Veisieju trial in 2010–2012. The period can be associated with an extremely cold winter (December–February) with extreme warmth in July and high precipitation in May (2010). The heritability index for recovery showed a peak in 2009, which coincided with an extremely cold June and August and a moist June.

In the more fertile conditions represented by the Nemencine trial, genetic controls of resilience showed weaker yet more constant throughout the reference period (Figure 6). The peaks were less expressed than under poor conditions. Recovery and relative resilience maxed in 2005 and 2008, as well as in 2014, likely in response to temperature extremes in preceding December, March, and June, as well as dry July. Heritability of resilience peaked in 1996 and 2016–2018, which had low temperature extremes in winter (December–March) and July, coupled with dry May and June. Resistance showed a sole peak at an early age in 1997, when the previous December and April were cold, yet the previous November was extremely warm. Under eutrophic conditions in the Silute trial, heritability estimates of resilience occurred in the latter part of the reference period. The strongest peak in heritability was estimated in 2018 for recovery, resilience, and relative resilience. For recovery, peaks also occurred in 2014. These years can be characterized by a warm previous December. In 2018, a cold March, a warm April, and a warm May occurred. For resistance, two peaks were estimated in 2020 and 2024, which coincided with extremes in February and May temperature and in April and June or July (high) precipitation.

The correlations between increment time series and meteorological variables, which are climate-related functional traits, showed extremely low heritability (h² < 0.03) regardless of trial/site productivity (Supplementary Material, Figure S2). When the data from trials were pooled together, the heritability estimates for the weather-increment correlations were similarly low (Supplementary Material, Figure S3). Likewise, the heritability estimates for radial increment and its resilience components were weak (h² ≤ 0.10), as the trials bore specific patterns and the climatic triggers of genetic control differed locally (Figure 2). The heritability estimates for tree dimensions were low, ranging from 0.05 to 0.13 and from 0.07 to 0.26 for stem diameter at breast height and tree height in the Nemencine and Ignalina trials respectively, indicating some relationships with site productivity.

The time series of heritability estimates showed mutual correlations as well as correlations with climatic variables, indicating climatic modulation over genetic controls and genetic associations between the components of increment (Figure 7). The heritability time series of the resilience components showed trial-specific correlations with weather variables, which did not show a relation with site productivity. In the Ignalina trial, the heritability of resilience components of radial increment showed correlations (mostly negative) with meteorological variables related to thermal and moisture conditions in late summer/autumn of the preceding year and with temperature in winter months. In Latezeris trial, heritability estimates of resilience components showed contrasting correlations with temperature in January, March, and May, and negative correlations with variables related to moisture conditions in the preceding autumn (positive).

In the oligotrophic trial (Veisieju), the meteorological regulation of resilience components was specific. The time series of heritability estimates for the resistance indices showed correlations (negative) with precipitation and CBW in the preceding October. Heritability of recovery was correlated with variables related to moisture availability during summer; for resilience, previous late summer conditions appeared as the primary drivers. The genetic controls over relative resilience correlated with the conditions in December. In the mesotrophic Nemencine trial, genetic controls over recovery and relative resilience showed correlations with variables representing moisture availability in March and May. In the Silute trial, temperature and precipitation in February and March, as well as CWB in May and the previous summer, showed correlations with the heritability of the resilience components.

4. Discussion

4.1. Informativity of Increment

Radial increment of the studied pine showed low interannual variability (Figure 2), which can be explained by low completion and application of containerized seedlings, which reduces the sensitivity of environmental water deficit at the early stage of development [72]. Despite the low interannual variability, the moderate Gini and mean sensitivity indices suggested that the increment was responsive and hence informative for time series analysis [63]. The autocorrelation of increment was considerably lower than observed for provenance trials and open-pollinated stands within the region [42,55], implying clearer responses to the environment [43]. The survival of trees was locally decreased [73], which resulted in a lower replication for some families (Figure 2) and likely in lower agreement statistics, particularly those affected by sample size (e.g., EPS and SNR) [65]. Hence, the EPS, a benchmark for the presence of meteorological variability, was often below the arbitrability threshold of 0.85 [65], and local variability of increment might sometimes be underrepresented. The moderate interseries correlations for the families, however, were moderate, suggesting notable effects of local conditions on the environmental sensitivity of growth [13,16].

The significant effect of site productivity (trial) on the metrics of increment (Table 1) suggested stronger environmental forcing of increment under oligotrophic conditions, likely due to stronger limitations under harsher growing conditions [19,43]. Still, the differences in effect strength among the oligotrophic trials implied modulating effects of local conditions [14,18]. In turn, the consistent effect of local provenance/population on responsiveness represented by the Gini coefficient (Table 1) implied genetically determined environmental sensitivity of growth [18,36,39].

4.2. Sensitivity and Climatic Drivers of Increment

The local (trial) specifics of the inter-annual variability of TRW (Figure 3) supported differing environmental forcing of radial increment, as characteristic for the region [42,54]. The similarity of TRW chronologies from the oligotrophic trials implied modulating effects of site productivity [19], which might also be due to the geographic proximity and sub-regional growth specifics [74,75]. The variation in TRW among the families within a trial was rather synchronous, indicating a common growth limitation, which likely limited the expression of the genetic effects under non-marginal conditions [76]. Nevertheless, the age-related increase in the spread among the chronologies suggested an increasing manifestation of genetically determined growth sensitivity, likely as environmental susceptibility increased [33,77]. Higher spread in years with wider TRW (Figure 3) implied that the studied genotypes differed in their ability to benefit from favourable conditions [42].

Even though the linear weather-increment relationships appear outdated [42], given the limited reference period, they are still informative for local comparisons [37]. The local specifics in inter-annual variation in TRW likely originated from differing sets of climatic drivers [55], as indicated by the weather-increment correlations (Figure 4). The local sets of significant meteorological variables, however, resembled those estimated for the region [42,54]. Accordingly, TRW was complexly controlled by the thermal conditions during the dormant period and moisture availability in summer, particularly its first part [42,54,76]. The complexity of the controls was also highlighted by the seasonally fluctuating correlations (Figure 3). The correlations with the conditions in the preceding vegetation season indicated carryover effects, likely due to the growth-reproduction trade-offs [78].

Stronger correlations under poor and eutrophic conditions (Figure 4) might be explained by stronger climatic limitations under harsher conditions [19] and higher plasticity of increment under favourable conditions [25], respectively. The differences in the correlations among families were higher under the mesotrophic trial, suggesting stronger expression of genetic effects under sub-optimal conditions [30,58]. Under the poor growing conditions in the Ignalina trial, which had sandy soil [73], water availability during growing showed growth limitations [40,41]. Effects of winter conditions on root dynamics and spring water availability [79] were indicated by correlations from the previous November through February (Figure 4). As in the poor sites, growth was slow, and the growth-reproduction trade-offs [78] were rather explicit. In the oligotrophic trials, meteorological conditions had direct effects on TRW. In contrast, in the more productive/fertile trials, carryover effects were stronger (higher correlations), likely because the better conditions allowed xylogenesis to compensate for direct adverse meteorological effects [43]. Hence, the effects of predetermining wood formations [78,79] were more pronounced.

The general absence of “significant” pointer years (Supplementary Material, Figure S1) suggested the ability of local genotypes to cope with the weather extremes in terms of increment [46]. Also, the lack of common tendencies of increment can be related to a spectrum of responses due to the diversity of the studied genotypes [18,36,55] in combination with low growth sensitivity. Coinciding events of extreme temperature in the previous late summer and early spring, which have been identified as a principal driver of increment [42,54,74], were also the likely underlying cause of the common changes in the increment.

4.3. Genetic Control of Climate Sensiitive Functional Traits

The estimated genetic controls (heritability) over the tree dimensions and TRW were generally low, likely due to local and microsite effects [13,16]. Nevertheless, the peaks in heritability of radial increment were occurring locally throughout the reference period, indicating effects of environmental/meteorological triggers [2,15,17,80], hence a potential increase with changing climate. The peak strength of heritability estimates for TRW and resilience components also showed some relation to the site productivity, indicating some modulating effects of soil properties [18,19,21].

Under optimal conditions, genetic differences are triggered by environmental extremes [13,14,16]; accordingly, the peaks in the heritability time series (Figure 5 and Figure 6) were linked with the co-occurrence of several weather anomalies. Apparently, the local population showed a high ability to cope with single environmental anomalies, thus supporting their adaptability [57,76]. Co-occurrence of weather anomalies was shown to cause some genetic differences among eastern Baltic provenances of Scots pine [81], particularly those involving rapid shifts in temperature during the dormancy period [18].

Cold stress has been a permanent partial trigger of genetic differences in increment and its resilience (Figure 5 and Figure 6), highlighting the persistent influence of cold [37,52], despite the warming [75]. Involvement of the conditions in the preceding late summer implied genetic controls over meteorological regulation of growth- reproduction trade-offs as observed for Norway spruce (Picea abies Karst.) [82]. Hence, the sets of meteorological variables associated with the peaks in heritability resembled those showing correlation with the annual variability of increment (Figure 4, Figure 5 and Figure 6), aligning with the high complexity of the climatic controls over growth [42,54,81].

The subtle change in genetic control over the resilience components of TRW across the site productivity gradient (Figure 6) advocated some modulating effects of edaphic conditions, likely via moisture availability and hence the strength of water deficit [19,22,79]. Higher peak values suggested that the genetic differences among the families were related to the recovery after stress, rather than the increment under the stress [45,81]. In contrast, the extremely low heritability estimated for the weather-growth correlations (Supplementary Material, Figures S2 and S3) explicitly contrasted the estimates based on the provenances, for which heritability was moderate [55]. Such discrepancies comply with high genetic diversity within the studied population, which shows minimal local specifics [56,57]. Still, low replication of some families might have also decreased the estimates.

The significant genetic correlations between time series of heritability estimates and meteorological variables (Figure 7) supported climatic controls over the expression of genetic differences on a regular basis [18,36]. Hence, the expression of genetic differences in analyzed traits would increase as the climate changes and becomes harsher for local genotypes [17]. The correlations between the heritability estimates of resilience components of increment implied common genetic control [17,83], suggesting some potential breeding for resilience based on native genetic material [6]. The meteorological variables moderating the genetic controls over the resilience were generally similar to those affecting increment [42,54]. This highlighted the relevance of responsiveness to meteorological variables representing thermal and moisture regimes as climate-sensitive functional traits when breeding for sustainability under accelerating environmental changes in the long term [14,15,17].

4.4. Limitations

A few limitations of the study are worth mentioning. Even though sampling between campaigns in 2017 and 2024 differed, the averaging of time series for a tree was performed to minimize the effects of the sampling technique, which, considering that trees grew in plantations, was likely negligible. As open-pollinated progenies were analyzed, the unknown pollen parents may cause deviations from the ideal half-sib structure. Related or unevenly contributing pollen donors and partial full-sib structure could inflate among-family variance, whereas high paternal diversity within families or strong microsite variation could reduce family differentiation; therefore, the estimates should be interpreted as approximate narrow-sense heritability [13]. Heritability estimates around 0.05–0.10 (Figure 5 and Figure 6) indicated weak additive genetic control and should not be interpreted as strong operational breeding parameters. Such values may still reflect detectable family differentiation, but their practical breeding relevance depends on the amount of additive variation, stability across years and trials, and achievable selection intensity [13,29].

Detectable heritability should be distinguished from operational breeding utility. While nonzero h² suggests some additive genetic differentiation among families, practical selection requires sufficiently high and stable heritability, adequate additive genetic variation, consistent ranking across environments, and a defined breeding objective, particularly when genotype-by-environment interaction is expected [13,16,29]. The interpretation of functional climate-sensitive traits is not as straightforward as for morphometric ones [42,81], implying their applicability as informative supplementary traits regarding climatic effects in the long term. Due to this reason and low heritability estimates, the genetic gains were not estimated at this stage.

5. Conclusions

The genetic controls over the increment and its resilience components (Figure 5 and Figure 6), and hence the potential of tree breeding to improve recovery of growth and competitiveness of Scots pine following climatic extremes, were generally low, though with peaks in some years. Such estimates might be partially attributed to high phenotypic plasticity [57,58], advocating some potential for targeted breeding [13,14]. However, site productivity, which modulates tree responses to climatic stresses [19,20], had marginal (at best) effects on the expression of genetic controls, limiting the potential for local fine-tuning of operative breeding [18,21,22]. The lack of genetic control over the climatic sensitivity of increment, likely due to similar local adaptation [57,58], highlighted the relevance of morphometric traits regarding the native population. Hence, the adaptive potential of local populations under anticipated changes appeared limited [8]. Nevertheless, the functional traits are an additional source for long-term evaluations.

The estimated climatic correlations with the expression of genetic controls (heritability) of increment (Figure 7) suggested weather-induced activation of genetic coping mechanisms [16,17,18], which might contribute to the resistance of growth of local populations in the long term as the climate shifts [6,7]. The linearity of these effects highlighted the proportionality of genetic controls in response to “normal” variation in weather, as well as its extremes. The climatic drivers of genetic controls and correlations among the traits (Figure 7) suggested weak relationships between increment and dimensions, implying differing underlying genetic control mechanisms [36,49], and a potential for simultaneous improvements. Hence, proactive, yet semi-conservative management, which centres on local populations, might still bear some limited adaptive potential at the strategic scale [1,12], aiding the sustainability of domestic timber sources in the EU in the long term [3,4].