Seven sites that differed in soil types and textures (1x sand-, 1x loam-, 3x silt-, 2x clay-dominated) and water balance (PAWC gradient from 44 to 101 mm) were evaluated ( Table 1 ). These sites are located in southern Germany (Baden-Württemberg, 48°27’N, 8°58’E) in close proximity to each other (within a radius of 1 km). They are located at an elevation of 500–520 m a.s.l. and represent different sites and soils typical for the mountain range of the Triassic “Keuperbergland”. Due to the close proximity of the study sites and the comparable elevation, there are practically no, or at most minimal, differences in weather conditions. The area is characterized by a warm, temperate climate with warm summers and no recurrent dry season (Rubel et al., 2017). The average annual temperature from 1991 to 2020 was 9.3°C, recorded at the nearest weather station of the German Meteorological Service (Hechingen, elevation 517.5 m a.s.l.), with an average annual precipitation of 806.4 mm (DWD, 2021).

One soil profile was dug at each of the seven sites to determine the soil and rooting characteristics. To ensure that the results of root counting were comparable between sites, each soil profile was placed according to Arbeitskreis Standortskartierung (2016) so that the profile wall was located at the boundary between the outer and the middle third of the crown radius of a Douglas-fir. Fine and coarse roots were counted at the profile wall using 5 × 5 cm counting squares. Roots with a diameter < 2 mm were classified as fine roots (Böhm, 1979). For further calculations, values for two rooting characteristics were derived from the results of root counting: effective rooting depth and mean fine root density in a 0–40 cm soil depth ( Table 1 ). A threshold of three fine roots per dm² was used for the definition of effective rooting depth, in accordance with Arbeitskreis Standortskartierung (2016).

In addition, soil texture, skeletal and humus content, and bulk density were determined in the soil profile for each soil horizon up to a depth of 1 m, in accordance with Arbeitskreis Standortskartierung (2016) and Ad-hoc-AG Boden (2005). From these soil physical properties, PAWC (mm) was calculated for each soil horizon according to Ad-hoc-AG Boden (2005). For this, PAWC values were used based on samples of the forest soil survey in Baden-Württemberg (Puhlmann and von Wilpert, 2011; Arbeitskreis Standortskartierung, 2016) and were therefore well adapted to the local conditions. The total PAWC for the respective site was calculated down to the effective rooting depth using the PAWC values from the different soil horizons and considering the thickness of the horizons (Arbeitskreis Standortskartierung, 2016). Information on the superordinate soil texture was attributed to soil depths of 0–1 m (sand, silt, clay, and loam). The soil type of each soil profile was also determined and classified according to the World Reference Base for Soil Resources (IUSS Working Group WRB, 2015).

Meteorological data [air temperature (°C), precipitation (mm)] were measured every half hour at a height of 2 m. For this purpose, a WatchDog 2700 weather station (Spectrum Technologies Inc., Plainfield, IL, USA) was placed in an open, unshaded area in the immediate vicinity of the study sites Sand, Silt1, Silt2, and Clay1.

To investigate the influences of site and rooting during a prolonged drought in the growing season (hypotheses 1 and 2), a main growing season with a prolonged drought (i.e., several months of drought) was identified. For this purpose, the soil moisture index (SMI) was analyzed, which was determined and provided for Germany by the Helmholtz Centre for Environmental Research–UFZ (UFZ, 2021). SMI was determined based on meteorological data from the German Meteorological Service as input parameter for soil moisture calculations via the mesoscale hydrologic model (MHM) (Samaniego et al., 2010). In this process, the local soil moisture was estimated for the entire root zone considering hydrological processes, such as interception, soil water dynamics, groundwater recharge, and storage (Zink et al., 2016). SMI represented the percentile of the simulated soil moisture value (moving average of the preceding 30 days) compared to 60-year soil moisture reconstruction (1954–2013) (Zink et al., 2016). Values vary between 0 and 1, and an increasing SMI indicates a decrease in drought (Samaniego et al., 2013). The SMI values we used were interpolated grid data for the topsoil (uppermost 25 cm) and total soil (down to approximately 180 cm, depending on soil properties) for the area of our study sites (Samaniego et al., 2013; Zink et al., 2016; UFZ, 2021). May–August 2018 proved to be a period with a pronounced drought. In all 4 months of this main growing season, the SMI for both the topsoil and the total soil was below the drought threshold of 0.2 ( Figure 1 ). From May to August 2018, the SMI for the total soil decreased continuously and was below the threshold value of 0.05 in July and August, meaning that extreme drought occurred in these 2 months.

The next step was to identify particular drought periods lasting many days within the main growing season of 2018. There was no standard definition of a drought period (Slette et al., 2019). A frequently used meteorological definition based on precipitation defines a drought period as a certain minimum number of consecutive dry days. There are different precipitation thresholds for defining dry days, depending on the region and research question. In this study, the threshold values were defined so that TWD increased over the course of the drought period. We considered an occasional small amount of precipitation reaching the ground and thus a temporary slight decrease in TWD for a few days to be acceptable if the TWD generally increased over the entire drought period. The threshold value also considered interception losses due to the canopy. The canopy water storage capacity for different Douglas-fir forests was determined to be between approximately 1 and 4 mm (Pypker et al., 2005). Based on the age and structure of the stands, we assumed that canopy water storage capacity was no more than 3 mm for our study stands (Klaassen et al., 1998; Pypker et al., 2005; Andreasen et al., 2023; Spittlehouse and Maloney, 2023). However, the resulting interception losses were often higher due to evaporation of the intercepted water (Spittlehouse and Maloney, 2023).

Two complementary threshold values were used in this study to define the drought periods within the summer of 2018: (i) A threshold of 5 mm was applied for the maximum daily precipitation, as also used by König and Mayer (1989) for a drought index for the investigation of climatically induced forest damage. (ii) The average daily precipitation during the drought period was below 1 mm per day and thus below the assumed evapotranspiration in the summer (Jassal et al., 2009; Thomas et al., 2022). These two threshold values ensured that only consecutive dry days were selected as drought periods on which no or only a small amount of precipitation reached the soil and that the tree water deficit generally increased over the course of the drought period. The duration of the respective drought periods was also considered when selecting the studied drought periods. According to our hypotheses, we selected prolonged drought periods that caused a high level of drought stress for trees and reduced the influence of random effects, such as site-dependent differences in surface runoff during short intense rainfall events.

Based on these selection criteria and on the data from our weather station, two periods were identified as pronounced drought periods for the extremely dry summer of 2018. The first drought period lasted from June 13 to July 4 (in the following referred to as the “June drought period”, total precipitation during these 22 days amounted to 2.0 mm), and the second drought period occurred from August 8 to August 30 (“August drought period”, total precipitation during these 23 days was 16.6 mm, daily precipitation on each day < 5 mm). There was a stronger precipitation event directly before and after both drought periods, in which more than 15 mm of precipitation occurred within less than one day. TWD and MDS data starting from the morning after the end of the stronger precipitation event before the drought period were included in the analyses of the two drought periods. At this time, after the rain, the nightly replenishment of water storage tissues is complete (TWD usually achieves its minimum), and the contraction phase, which is decisive for calculating MDS, has not yet started. The TWD and MDS data included in the analysis of the drought periods ended with the onset of precipitation at the end of the drought period. This point in time was always clear due to a sudden stronger rainfall.

The stands at the investigated sites were Douglas-fir stands or in three cases mixed Douglas-fir and Norway spruce stands. While Norway spruce was spatially separated at sites Clay1 and Clay2, Douglas-fir was mixed with Norway spruce individually or in groups at the Silt3 site. Understory beech was mainly present at Clay2 and Loam sites. At each of the seven study sites, five Douglas-firs were selected per site for the dendrometer measurements (“dendrometer trees”). These trees had a social tree class of 1 or 2 (predominant or dominant trees) (Kraft, 1884) and had been selected as crop trees during previous thinnings. To minimize possible small site differences in the soil, the representativeness of the tree sites was examined for each tree using soil coring. The age of the trees was determined by tree ring coring and was approximately 40–55 years at the end of 2020 [with increment borers (Haglöf, Sweden)] ( Table 2 ). The diameter at breast height (DBH) was determined, and the tree height and height of the crown base were measured with a Vertex (Haglöf, Sweden) for each dendrometer tree. Furthermore, stand characteristics were determined (stand density, stand basal area, DBH, and tree height refer to the tree with quadratic mean diameter), based on a DBH-full inventory of a representative stand area for each site ( Table 2 ). We also derived the yield classes for Douglas-fir for each site from the growth tables for Baden-Württemberg (Forstliche Versuchs- und Forschungsanstalt BW, 2001). Input parameters were the measured heights and the tree age (determined from the number of years derived from the cores plus 5 years, as cores were collected at 1.3 m).

As the origin of a Douglas-fir tree may have a significant influence on drought tolerance (Eilmann et al., 2013; Jansen et al., 2013; Bansal et al., 2016; Chauvin et al., 2019), a genetic analysis was carried out for each of the five dendrometer trees per site. The cambium samples we collected for this purpose were genetically analyzed in the molecular genetic laboratory of the Institute of Silviculture at BOKU University in Vienna (Austria). The methods used were previously described in Slavov et al. (2004); van Loo et al. (2015); Hintsteiner et al. (2018); Neophytou (2019), and Spangenberg et al. (2024). As a result, all dendrometer trees were clearly assigned to the coastal variety of Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco var. menziesii). The Douglas-fir trees of the seven study sites were genetically very similar at the 13 microsatellite loci analyzed, as no significant genetic differentiation was detected among them (Neophytou, 2019). Therefore, Douglas-fir trees from the seven study sites were pooled for all subsequent analyses.

Dendrometer trees were equipped with a high-resolution point dendrometer at a height of 1.6 m in March 2017. We used spring-loaded linear displacement potentiometers (type MMR 10_11 R5K, MEGATRON Elektronik GmbH & Co.KG, Munich, Germany) with analogue stepless resolution. Stem radius changes (SRCs) from 2017 to 2019 were recorded every half hour using a data logger (DL2e, Delta-T Devices Ltd., Cambridge, UK; DL18, Ecomatic, Dachau, Germany). Before mounting the dendrometers, as much as possible of the outermost dead bark layer was removed at the point where the contact head of the dendrometer touched the bark, without damaging the inner living cortex and cambium (Zweifel et al., 2006). The compensation or minimization of the possible weather-related sensitivity of the dendrometers (Zweifel et al., 2006) was carried out metrologically and mechanically. For example, the maximum resistance was simultaneously measured, and dendrometers were protected from rain and sunlight with a cover.

Raw SRC data were checked carefully for each tree. Clearly incorrect measurements (e.g., due to cable damage) were omitted. Due to a data logger defect in spring 2018, the data of site Clay2 could be used for evaluating drought periods but not for growth duration analysis. Characteristic values needed for further analyses and figures (e.g., daily maximum values, daily precipitation sum) were determined from SRC and weather data using the R package “dendrometeR” (van der Maaten et al., 2016).

Two water status parameters (see Section 1) were derived from SRC values for the two drought periods analyzed in summer 2018. Maximum daily shrinkage (MDS, in μm) of stem radius was calculated for each tree as the difference between the morning maximum and the subsequent minimum value (Herzog et al., 1995; King et al., 2013; Dietrich et al., 2018; Schäfer et al., 2018) ( Figure 2 ). On very few days with a permanently rising stem radius, MDS was set at 0. In addition, tree water deficit (TWD, in μm) was calculated for each tree as the difference between the maximum value of the stem radius reached until time t (fully hydrated stem) and the stem radius at time t according to Zweifel et al. (2016). This assumes that no growth occurs during periods of stem shrinkage (zero growth concept) (Zweifel et al., 2016). The calculated daily values for MDS and the half-hourly calculated TWD values were averaged for each tree within one drought period (see Section 2.2).

SRC values, separately for each tree, were also the basis for estimating the days of the year (DOY) of growth onset and cessation from 2017 to 2019. Growth duration was derived for each tree by determining the number of days between growth onset and cessation. The reference base for these estimates was the total annual variation in stem radius. Growth onset and cessation were defined as reaching the 5% and 95% thresholds of total annual SRC, as used by van der Maaten et al. (2018) and Krejza et al. (2021). For these thresholds, sometimes other values were used in other studies, e.g., 2.5 or 3.5% for growth onset (van der Maaten et al., 2012; Miller et al., 2023). Since an increase in SRC in early spring may initially be caused by tissue swelling before the actual onset of growth, we used 5%. The year 2018, which was decisive for our study, showed a rather atypical course of the SRC curve. From mid/late June, many trees showed only little SRC increase and repeated periods of pronounced TWD ( Supplementary Figure 1 ). We therefore used raw SRC data for the calculations of onset and cessation of growth.

To analyze the influence of soil and rooting characteristics on water status parameters and growth duration, we used linear mixed-effects models (LMMs). Our measurements were recorded at seven sites and were not independent from each other. LMM considers the non-independence of the data and enables the estimation of possible random effects. Model optimization was carried out using the maximum likelihood method. First, a null model (also referred to as the intercept-only model), which only contained random effects, was tested. Then, the fixed effects of the respective model equation were included, and the best-fitting model was determined using a model simplification procedure. For this model selection, the Akaike Information Criterion (AIC) was used to evaluate the relative goodness of fit of the models (Akaike, 1974). Normality and homoscedasticity were checked visually using plots of residuals vs. fitted values (Zuur et al., 2010). Possible problems with collinearity in the models were checked using the variance inflation factor (VIF) (Zuur et al., 2010). As a threshold for critical collinearity, we used VIF > 10 (Dormann et al., 2013). If a model showed high collinearity, the biological reasons for removing predictors from the model played a decisive role (Zuur et al., 2010; Harrison et al., 2018).

In Equations 1-5, water status parameters (WSP, i.e., MDS or TWD) and growth duration (GD) indicate the response variables, and a₀ is the overall intercept. We included the predictors soil texture (ST), PAWC, effective rooting depth (RDepth), and mean fine root density at 0–40 cm soil depth (RDens) in the LMM with Equations 1, 2, 4, 5, and thus data at site level s. In another model (Equation 3), we included the water status parameters MSD and TWD and thus data at tree level t as predictors. The coefficients b₁ to b₁₀ were the coefficients to be estimated of these fixed effect predictors at the site or tree level. Soil texture was a categorical variable with four levels (clay, loam, sand, silt). The other predictors were quantitative variables. The predictor soil texture was analyzed in a separate model without integrating PAWC or rooting characteristics (Equations 1, 4) because of the high collinearity between soil texture and the other variables (VIF > 10). In all LMM we used, VIF is below 10 for all predictors. In addition, possible random effects were considered. In Equations 1-5, α indicates the random effects at site level s (seven sites), and ϵ is the independent error term. The variances of these random effects were estimated during model fitting.

The LMM with Equation 1 was used to analyze the influence of soil texture on the water status parameters during the June drought period and the August drought period in 2018 (hypothesis 1).

To study the influence of PAWC and rooting characteristics on the water status parameters during the two drought periods in 2018 (hypothesis 2), we used Equation 2.

The model for estimating the influence of the two water status parameters on growth duration (hypothesis 3) was as follows:

To analyze the influence of soil texture, PAWC, and rooting characteristics on growth duration (hypothesis 4), we used Equations 4 , 5.

Growth onset and duration shown in Figure 3 were tested for significant differences between several soil textures separately for each year in multiple comparisons by the Kruskal–Wallis test and a subsequent post-hoc test (p < 0.05). The post-hoc test used was the Dunn test. All statistical analyses were carried out using R version 4.3.2 (R Core Team, 2023). Statistical analyses using LMMs were conducted with R packages “lme4” (Bates et al., 2015), “lmerTest” (Kuznetsova et al., 2017), and “performance” (Lüdecke et al., 2021).

Both the stronger precipitation events before and after the two drought periods in 2018 and the drought periods themselves are clearly visible in the SRC diagrams for the loamy, sandy, and clayey sites ( Figures 4 , 5 ). These events are less clearly recognizable in the SRC curves of the silty sites. A comparable soil texture-specific ranking was observed in both drought periods. The clay- and loam-dominated sites had lower SRC values than the silt- and sand-dominated sites, both during the drought periods and in the subsequent precipitation-induced water saturation ( Figure 5 ). Up to about half to two-thirds of the drought periods, the silt- and sand-dominated sites showed a similar SRC. In the further course of the drought periods, the SRC values of the sandy site decreased slightly compared to the silty sites.

Averaged across the four soil textures, growth duration was 124 days (2017), 108 days (2018), and 122 days (2019).

In both drought periods studied in 2018, a higher TWD showed a shortening of the growth duration 2018 ( Table 4 ). This influence was highly significant (p < 0.001) in the August drought period, while the TWD influence in the June drought period was slightly above the significance level (p = 0.058). MDS had no significant influence on growth duration in either drought period.

PAWC and the two rooting parameters had no significant influence on growth duration in the extreme drought year 2018 ( Supplementary Table 2 ). In contrast, we found significant influences of soil texture on growth duration in 2018. Compared to the silt-dominated sites, the loam-, sand-, and clay-dominated sites significantly reduced the growth duration ( Table 5 , Figure 3 ). The loamy site led to the greatest reduction. Significant differences in growth duration in 2018 between Loam and Silt were also evident in the relative SRC curve ( Figure 4 ). From the beginning of the first drought period on June 13, 2018, until reaching the 2018 SRC maximum, the silty sites showed 32% of the relative annual growth, while the loamy site only showed 9% (Sand: 19%, Clay: 22%). The sandy and clayey sites showed a similar relative SRC curve in 2018.

For 2017, 2018, and 2019, no significant differences were found in growth onset between trees grown on different soil textures within one year ( Figure 3 ). Drought year 2018 also did not cause a significant delay in growth onset in 2019 compared to 2018 for any of the soil textures. Averaged across all soil textures, the growth onset in 2019 occurred two days later than that in 2018. The SMI for the study area showed very low soil moisture until April 2019 and a recovery only from May 2019. Miller et al. (2023) found a later growth onset for Douglas-fir in 2019 compared to 2020 and saw this as the legacy effect of extreme drought in 2018. However, several studies have indicated that other factors, especially temperature, are crucial for controlling growth onset (Rossi et al., 2007; Gruber et al., 2010; King et al., 2013; Rathgeber et al., 2016; Miller et al., 2022). The influence of age and stem size on the onset of cambium division and other phases of cell differentiation found by Rossi et al. (2008) probably had no significant influence in our study, as the trees studied had a similar age, and no significant difference in growth onset was found between the different sites despite different DBH values.

The extreme drought in 2018 shortened the growth duration across all soil textures by an average of 15 days compared to 2017/19. This was primarily due to earlier growth cessation. Significant correlations between soil moisture and the date of growth cessation were found for different tree species (Gruber et al., 2010; Eilmann et al., 2011; Saderi et al., 2019; Miller et al., 2022). In our study, a higher TWD in the June drought period, but especially in the August drought period, reduced the growth duration ( Table 4 ). The zero-growth concept assumes that no growth occurs during periods of stem shrinkage (Zweifel et al., 2016). Our investigations showed that even after the end of the pronounced summer drought periods, only very small radial growth occurred at some sites, resulting in a shortened growth duration as a legacy effect of a pronounced TWD.

There were, however, significant differences depending on the soil texture. The drought in 2018 significantly shortened the growth duration on the loam-, sand-, and clay-dominated sites compared to the silt-dominated sites ( Table 5 ). However, resistance studies carried out based on annual increments showed different results (Spangenberg et al., 2024). There, the same silt-dominated sites showed the lowest resistance among the four soil textures and thus the greatest relative radial growth reduction in 2018 compared to the 2016/17 reference years (see also Supplementary Figure 1 ). A study by van der Maaten et al. (2018) evaluated three deciduous tree species and showed that growth duration was not a decisive determinant of the size of the annual radial stem increment. The Douglas-firs we examined showed that, in summers with more precipitation (2017 and 2019), they achieved larger radial increments even in August at certain sites ( Supplementary Figure 1 ). In the extreme drought year 2018, however, a major part of annual radial growth was already reached by the middle of June ( Figure 4 ). Little precipitation in July and August 2018 was not sufficient for larger radial increments. However, the plant availability for this little rainfall during drought periods varies depending on soil texture. Silt-dominated soils store a high proportion of the water available to plants and are characterized by a high PAWC (Puhlmann and von Wilpert, 2012). Clayey soils have high field capacities, but store relatively high proportions of water not available to plants (Arbeitskreis Standortskartierung, 2016). Sandy soils have a low field capacity as a large proportion of the water seeps away (Puhlmann and von Wilpert, 2012; Arbeitskreis Standortskartierung, 2016). This could explain why the Douglas-firs on the silt-dominated soils still showed radial growth at a low level in July and August and had therefore not yet reached growth cessation, while the other three soil textures showed shorter growth durations in 2018 ( Figures 3 , 4 ).

As shown in Section 4.1.1, the loamy site was also noticeable in terms of growth duration. The Douglas-firs there showed the greatest absolute and relative decrease in growth duration in 2018 compared to 2017/2019 and the shortest growth duration in 2017 and 2019 ( Figure 3 ). At this loamy site, which had a comparatively high PAWC, the location on the south-eastern upper slope resulted in very low water inflow, exacerbating the water shortage. As a result, the Douglas-firs there presumably had periods of drought stress also during summers with higher precipitation. The trees at this site showed the highest resistance in 2018 and thus the smallest relative decrease in tree ring width in the drought year compared with the other study sites (Spangenberg et al., 2024). This indicates an adaptation to recurring drought, e.g., in the form of a deeper rooting system. The second deepest effective rooting depth was found at this site.

Based on our results on the influences of PAWC and the two rooting parameters on TWD and MDS (Section 4.1.2), effects on growth duration—caused by effects on growth cessation—would also have been expected. However, we did not find any significant influence of these three parameters. This confirms other studies in which different or less clear influencing factors on growth cessation were found. Several factors were investigated and discussed, such as temperature, drought stress, and photoperiod (Gruber et al., 2010; Eilmann et al., 2011; Rathgeber et al., 2016; Rossi et al., 2016; Saderi et al., 2019). Their influence on growth cessation, however, is less understood than that on growth onset (Rossi et al., 2013; Rathgeber et al., 2016; Miller et al., 2022).

In the LMM with Equation 5, on which the conclusions in the previous paragraph are based, the ratio of measured trees to predictors is 8:1 and therefore less favorable than in our other LMM. To prevent such models from becoming too complex, there are different recommendations for the minimum ratio of data points to the estimated parameters (Harrison et al., 2018). Although our model with Equation 5 did not quite reach the conservative recommendation of Harrison et al. (2018) of 10:1, all reduced models with only one or two of the three predictors of Equation 5 led to comparable results.

Studies that examine both above- and belowground biomass are important for a better understanding of the mechanisms of drought tolerance in trees. Tree roots can respond to severe drought with different strategies to avoid or tolerate drought stress (Brunner et al., 2015). These processes can also have an impact on aboveground tree growth. However, until now, there have been comparatively few studies on these complex relationships among soil, rooting, and drought tolerance on Douglas-fir. Therefore, further studies should be carried out to close existing knowledge gaps. The results of our study confirm that the rooting depth and, thus, the maximum water uptake depth is an important criterion for the possibility of stress reduction in trees during pronounced drought periods (Bréda et al., 2006).

Our study was carried out under average climatic conditions of SW-Germany on approximately 40 to 55 year old Douglas-fir trees. According to the results of the genetic analysis, the origin of the trees of all Douglas-fir stands studied is located in the central area of the natural range of the coastal variety between central Washington and northern California (Neophytou, 2019). Climatic conditions, tree age, tree species mixture, and origin of the Douglas-fir trees can influence the results (Eilmann et al., 2013; van der Maaten-Theunissen et al., 2013; Thurm et al., 2016; Vitali et al., 2017, 2018; Bréda and Brunette, 2019; Nicolescu, 2019); thus, further studies with other conditions are necessary to validate our findings. It should also be considered that our determination of growth onset and cessation is based on SRC estimates. Our results on growth duration should be verified by studies based on, for example, weekly sampling of microcores, which allow us to more precisely determine the onset and cessation of wood formation (Rossi et al., 2006; Güney et al., 2015; Rathgeber et al., 2016; Miller et al., 2022).

Despite the extreme strength of the drought in 2018, all trees studied survived and recovered well in subsequent years with higher summer precipitation in 2019 and 2020 (Spangenberg et al., 2024). There was no dieback of Douglas-fir in any of the study stands. Therefore, we do not know how the magnitudes we investigated for TWD and MDS should be assessed in comparison to the maximum possible stress level. Even within a site, there is, in some cases, a noticeable variance in magnitudes between the trees we measured. Since we examined very similar tree collectives, we interpreted this as different stress levels for individual trees. However, this can be overlaid by other tree-specific characteristics, such as different trunk and bark morphology (Brinkmann et al., 2016). Normalization based on the maximum values from the main growth period does not resolve this possible overlap. In our study, the magnitudes of TWD and MDS are the main study parameters and not onset of TWD (see Brinkmann et al., 2016). In our case, the use of normalized, i.e., relative TWD and MDS values, only brings an improvement if the reference value (e.g., a certain maximum value) can be associated with a certain stress level. This requires further studies, e.g., at the drought tolerance limit or through the parallel measurement of other stress parameters. Our approach was to investigate a comparable research question using two different methods at the same sites and trees. In addition to the dendrometer measurements described above, we also examined tree ring widths (Spangenberg et al., 2024). The conclusions from both studies fit and complement each other well. This shows that investigations using different methods can help to close knowledge gaps.