Introduction Top

The concept of plant available water capacity (AWC) of soil was proposed by Veihmeyer & Hendrickson (1931), who defined it as the water content of soil between an upper limit, called field capacity, and a lower limit or permanent wilting point. Thus, AWC characterizes the reservoir of the soil, which holds water that may be withdrawn by plants. AWC is a central property for soil fertility and thus of great relevance for forestry planning (AK STOK, 1996; Silva et al., 2014).

The influence of soil fertility on tree growth is a central paradigm of forest site classification (e.g. Barnes et al., 1982; Pyatt et al., 2001). Since the effort for soil examination is quite high, height growth has served as a major indicator for site fertility in forest classification systems in Europe and North America for a long time (Tesch, 1980). The foundation for this method was the observation that dominant height versus age relationship remains quite stable even at different management regimes (Huang & Titus, 1993). Although soil effects are obvious and undisputed, only a few more recent studies quantify soil effects on height growth of adult trees in Central Europe. Investigations on the country-level imply only a low contribution of soil parameters for describing the height growth compared to temperature and precipitation (Brandl et al., 2014; Burggraef et al., 2016; but see Boncina et al., 2023). On the other hand, some studies show marked soil effects along a large climatic gradient in the Bavarian Alps (Mellert & Ewald, 2014a,b), while others highlight effects of stand density and water availability on the height growth of trees (Toraño Caicoya & Pretzsch, 2021). Investigations involving microsite effects on single tree growth are scarce (Rabbel et al., 2018; Hartl et al., 2021). To our knowledge our investigation is the first study about the effect of small-scale variation in AWC on height growth of single trees.

However, single tree growth in seed crop stands is an important issue especially for forest genetics and therefore a central motivation of our study. To broaden the research base, the project sensFORclim aims to locate climate resilient populations of important main tree species by various methods (climate envelopes, dendroecology, physiology, and genetics) at different spatial scales (continent, region, stand) (Schmied et al., 2023). Within the framework of sensFORclim the current study encompasses actual and potential seed crop stands in Southern Germany. Working at the single tree level, it is important to know whether phenotypic differences, e.g. in tree growth, are based on individual genotype performance (Rellstab et al., 2015; Pfenninger et al., 2021) or microsite effects due to small-scale variation in soil parameters. This issue is not only interesting within population genomics but also of practical relevance for forest seed harvesting and breeding (Clark & Wilson, 2005), since soil effects could confound the effectiveness of plus-tree selection conducted by tree breeders (Cornelius, 1994).

Therefore, we investigated possible small-scale soil effects on achieved tree height in more or less even-aged monospecific stands of three important tree species (Abies alba (Aalba), Fagus sylvatica (Fsylv), Picea abies (Pabie)) within their niche space. For the characterisation of the niche zone (NZ), we used the climatic marginality towards the rear edge (Mellert et al., 2016), i.e. the dry and warm border of species distributions (Hampe & Petit, 2005), as within this climatic zone important demographic and genetic processes are taking place (Fady et al., 2016). This study comprises a large gradient from the cold and wet mountainous climate to the warmer and drier temperate climate in the lowlands of Southern Germany (Schmied et al., 2023). We surveyed the soil in close proximity to trees sampled by estimation of soil texture and gravel content, following the German soil survey guidelines (KA5, AG Boden, 2005; AK STOK, 1996). Based on these data, we calculated the AWC of the pedons (Johnson, 1963) associated with the specific target trees.

We hypothesised that trees growing on pedons with higher AWC reach greater heights and assume that this effect interacts with the NZ of the species. According to our assumption, achieved tree height should increasingly benefit from higher microsite-AWC with increasing marginality of the NZ. This effect should be most obvious at shallower soils with a high heterogeneity in AWC. If such an effect should exist this would be of general relevance in forest research. In our research area, genetic evaluations that focus on single trees and plus-tree selection would be most affected.

Regarding the site trophy, the traditional expectation is that height growth increases among sites from acidic to base-rich soils (Neebe & Hofmann, 1982). However, on base-rich calcareous soils with high carbonate content in the whole soil profile and a low loam content, tree growth can be limited due to a shortage of N, P, K and trace elements (Mellert et al., 2014a; AK STOK, 1996; Kolb et al., 2018).

Material and methods Top

Study area

The study region lies in Southern Germany comprising four federal states (Baden-Wurttemberg, Bavaria, Saxony, Thuringia) (Fig. 1). This area lies in the western centre of the (Hercynian) mixed mountain forest (Jeník, 1998), where the target tree species European beech (Fagus sylvatica L.), Norway spruce (Picea abies [L.] Karst), and silver fir (Abies alba Mill.) dominate. Our target region covers a large part of the main distribution area of this submontane to montane forest type in Germany. In this geographical area, the warm and dry climatic edge of the montane mixed forest is already exceeded in the lowlands, because the dominance of the investigated main tree species clearly decreases in the colline to planar area (Ellenberg, 1988; Jeník, 1998). In the lowlands and especially in warm, dry locations, marginal yield sites of these tree species are regularly found. The climatic spectrum representative of the study tree species can therefore be well represented in the target area.

Figure 1. Study sites for Abies alba, Aalba (a), Fagus sylvatica, Fsylv (b) and Picea abies, Pabie (c) in Southern Germany with symbols according to their niche zone (see legend). Background shading represent the macroclimatic niche zones according to the climatic marginality index (CMI: 1_O: optimal, 2_I: intermediate, 3_M: marginal, 4_R: rear edge) with lighter shading at more marginal sites. Geographic coordinates (WGS84) in degrees as white lines (N: northing, E: easting) and the position of cities as black letters (D: Dresden, E: Erfurt, F: Freiburg, K: Karlsruhe, L: Leipzig, M: Munich, N: Nuremberg, W: Wurzburg) are given for orientation.

Tree species and stands

Our project works on the main tree species of the (Hercynian) mixed mountain forest (see above). We preferably selected mature stands with mostly even-aged trees. Monospecific forest stands were preferred for sampling to minimize mixing effects. However, in the absence of monospecific stands, a small proportion of other admixed tree species was accepted, particularly for silver fir, which was often associated with Norway spruce. Further, forest stands that had been frequently or heavily thinned were purposefully avoided to reduce potential management effects.

Tree sampling

We measured the height of each tree using a Vertex IV ultrasonic hypsometer (Haglöf, Sweden). We selected only trees in forest stands with a closed canopy and trees that were dominant or at least co-dominant. Thus, trees near forest edges or gaps and trees with visible damage were excluded. Two cores were extracted for each tree with a 5-mm increment borer (Haglöf, Sweden) from eastern and northern cardinal direction, which allowed us to determine individual tree age. All core samples were stored, air-dried, glued onto wooden carrier boards and sanded with increasingly finer grit-paper. Subsequently, we used the digital positioning table LINTAB 5 and the software TSAPWIN (both Rinntech, Heidelberg) to count annual tree rings. Tree age was estimated from tree ring series where the tree pith was hit or estimated if the last rings counted were really close to the pith. In rare cases where it was not possible to detect the pith or estimate tree age, the average age of the other trees in the same stand was used. Basic characteristics of sampled trees (age, height and diameter at breast height (DBH) are outlined in Table S1[suppl].

Niche zone (NZ)

For defining the macroclimatic niche (Mellert et al., 2021), we used environmental niche models (ENMs) based on national forest inventory data (Mauri et al., 2017; Thurm et al., 2018) and WorldClim data (Fick & Hijmans, 2017) for the climate period 1970-2000. The applied ENMs relied on the BIOCLIM variables minimum temperature (Bio6, Tmin), average summer temperature (Bio10, Twq) and summer precipitation (Bio18, Pwq), acting as proxies for the limitation due to chilliness and frost (Tmin) as well as for summer heat (Twq) and drought (Pwq) (cf. Mellert et al., 2015). These ENMs are equivalent to the models used for decision support at the Bavarian State Institute of Forestry, especially with regard to cultivation risks under climate change (BayLWF, 2019; 2020). Usually, the outputs of ENMs are mapped based on the predicted probability (Pocc) as a measure for habitat suitability. However, in climate change the rear edge matters most (Hampe & Petit, 2005). Growth depression, forest damage and adaption are taking place predominately in marginal areas at the rear edge. As Pocc does not show, whether a low suitability is associated with the leading or the rear edge, we used the climatic marginality instead (Mellert et al., 2015, 2016; Dorado‐Liñán et al., 2019). We computed the climatic marginality index (CMI) for each pixel of the map (resolution 2.5 arc seconds, approx. 1 km; Fig. 1) by stepwise increasing the temperature until the rear niche border was reached. The CMI represents the length of the trajectory on the level of probability values. The climatic stratification is founded on the potential future marginality. Therefore, we applied a simple climate scenario assuming a moderate warming of 2.5 °C (EEA, 2017). As a result, pixels with CMI > 0.7 were attributed to the optimal macroclimatic NZ, values between 0.4-0.7 to the intermediate, and with CMI < 0.4 to the marginal NZ.

Similar to the two-step procedure of the German site classification (AK STOK, 1996), the study sites were assigned to the NZs by considering (1) the macroecological and (2) the local situation. To properly represent the forest regions of Southern Germany, we used the map of forest ecological regions of Germany (FER; Kolb & Göttlein 2014; resolution approx. 1:1,000,000) and the macroclimate (CMI) as described above (Fig. 1) for pre-stratification. The macroecological information were intersected with the local soil moisture regime according to the German site classification systems (AK STOK, 1996) derived from forest site maps with a high resolution (approx. 1:10,000). The classification according to these fine-grained maps was approved in the field to assure a correct description of the soil moisture and nutrient regime (NR) (see below subchapter ˈSoil nutrient regimeˈ) of each site. Soil characteristics involving hydromorphology played a key-role within these checks. To avoid influences of supplemental water, stands growing on soils with Gleysol characteristics were strictly excluded from the strata 3_M and 4_R (see below). In principle the classification of the four NZs followed the subsequent criteria:

1_O = “optimal”: optimal macroclimatic NZ (Fig. 1); zonal soil condition, at least moderately fresh soil moisture regime according to the forest site classification system; supplemental water (slope water) possible.

2_I = “intermediate”: intermediate macroclimatic NZ (Fig. 1); zonal soil conditions, moderately fresh soil moisture regime, no supplemental water (exclusion of soils with marked Gleysol hydromorphology).

3_M = “marginal”: marginal macroclimatic NZ (Fig. 1); zonal soil conditions, usually still (moderately) fresh soil moisture regime; moderately dry soil moisture regime possible; no supplemental water (strict exclusion of soil with Gleysol hydromorphology).

4_R = “rear edge”: marginal macroclimatic NZ (Fig. 1); minor soil moisture regime (moderately dry, dry) compared to 3_M due to unfavourable edaphic condition, i.e. higher gravel/sand content; no supplemental water (strict exclusion of hydromorphic soils). We consider this NZ as the potential future rear edge.

As the stands of the last two strata (3_M and 4_R) should grow under a similar macroclimate but different soil moisture regimes, they should be situated in the same forest ecological landscape (Table S1, symbolized by the capital letter of the forest region according to Kolb & Göttlein, 2014) characterising regions with similar climate and geology, or soil NR, respectively.

Because the macroecological classification maps (Forest Ecological Regions of Germany and CMI) provide only a spatially low-resolved estimate of site conditions (zonal site conditions), the final classification was primarily based on the local site characteristics in consideration of (extrazonal) site conditions (soil and micro-/mesoclimate). Due to a drier local soil moisture regime, one spruce stand, two beech and four fir stands were re-assigned to a more marginal NZ compared to macroclimate (CMI). Also, in these cases the central criterion was that the soil moisture regime of 4_R sites is lower and/or the gravel/sand content is higher than those of the respective 3_M sites.

The marginal zone (3_M, 4_R) converges largely with areas where beech has mostly been affected by droughts in recent years in Germany (Bolte et al., 2021; Michel et al., 2021; Martinez Del Castillo et al., 2022). This coincides with the observations during site selection in this study (Mellert et al., 2021).

Soil nutrient regime (NR)

Beside the climatic NZ, the soil NR represents the second axis of the stratification and is classified into three levels (Ba: acidic/poor, Mb: medium, Br: rich) following Kölling et al. (1996). These trophic levels assign (1=Ba) deeply weathered soils with low base saturation (BS < 20% from the topsoil to the subsoil); (2=Mb) soils with low BS (< 20%) in the topsoil and high BS (> 80%) in the subsoil; (3=Br) soils with high BS (> 80%) throughout the entire soil profile. The soil NR has been included to systematically represent the site units (AK STOK, 1996) of the study area with respect to the site trophy. The classification of the soil NR according to the map of forest ecological regions of Germany and local site maps was matched based on the humus forms, soil texture classes and soil types found in the field.

Available water capacity of the soil (AWC)

We surveyed possible small-scale soil heterogeneity by sampling the soil upslope the target trees within a distance of 1.5 and 2 m. The organic surface (O) was subdivided into the organic horizons (L (litter), Of (fermented) and Oh (humic) according to KA5, AG Boden, 2005) and their thickness was measured. The mineral topsoil was characterised to a maximum depth of 30 cm based on small soil pits. The subsoil was examined by using a Pürckhauer auger. Soil cores were taken down to 1 m depth (length of the auger) or the parent material. The depth reached ranged from a few centimeters (O/mC soils) to the maximum depth of the auger (well-developed soils). The mineral soil was subdivided into horizons. The texture class and the soil density were manually determined according to the German pedological field guide (KA5, AG Boden, 2005; Bormann, 2007; AK STOK, 1996), while the gravel content was estimated based on visual inspection of the soil pits and the core sample. Laboratory analyses were not conducted. AWC of each horizon was then calculated with the pedotransfer tables of KA5 (AG Boden, 2005) giving the AWC values of soil texture classes structured by the levels of soil density (KA5, AG Boden, 2005; AK STOK, 1996). The soil (bulk) density influences the pore volume and thus the AWC. We used three levels of density (low < 1.4 g cm-3; intermediate < 1.6 g cm-3; high > 1.6 g cm-3) indicated in KA5 (AG Boden, 2005). The gravel content was subtracted from this value. AWC of the organic layer was set to 30% (Hammel & Kennel, 2001). Finally, the resulting AWC values of the organic and mineral horizons were summarized to the cored maximum depth of the soil profile. Table S1 shows the mean and range of AWC of our study sites.

Study design

The stratification relies on climatic niche zone (Table S1, column “NZ”) and soil nutrient regime (Table S1, column “NR”). Based on these factor variables, we have partitioned the environmental space into two ordinal axes, the first one accounting for NZ of the species (see ˈNiche zoneˈ above). The second axis represents the NR (see subchapter ˈSoil nutrient regimeˈ) based on forest site classification system of the states of Germany (AK STOK, 1996). Within the 12 resulting strata per species (4 climatic niche strata × 3 trophic strata) we searched for mature stands fulfilling predefined stand characteristics with regard to species distribution, structure, disturbance, thinning, stand density (see ˈTree samplingˈ). For each stratum we selected one stand for sampling. Due to the high importance of beech forests in Southern Germany growing on sites with an optimal climate (1_O) and a medium trophy (m), we selected two sites for this stratum. This resulted in a total number of 37 stands (Fig. 1) with 565 sample trees in total and the same number of surrounding pedons (i.e. a roughly hexagonal area, with a surface area of ~10 m² and a depth of ~1 m; Johnson, 1963). For further methodological details see Schmied et al. (2023).

Data analysis

We examined our hypothesis by using a mixed regression model (lmer-function from the lme4 package; Bates et al., 2015) with the following formula:

where the target variable is y_ij (= achieved tree height) with m=565 individual observations (i=1,…, m), clustered by n=37 sites (j=1,…, n). Predictors are ln(age) = natural logarithm of age of the tree, AWC = available water capacity, NZ = niche zone, TS = tree species, NR = nutrient regime (site trophy). AWC*NZ describes the interaction effect between NZ and AWC (see Table 1). To account for possible bias due to site selection and/or stand history, we included a random effect y_0i for the sites (stand level) and assumed a Gaussian error distribution.

Inclusion of the term ln(age), assuming logarithmic growth, accounts for differences in age of single trees. This term avoids a bias due age trends and makes younger and older trees comparable. All continuous independent variables were included in the analysis in a scaled form (divided by the standard deviation and subtracted by the mean). Model residuals were checked by plots for individuals and aggregated by sites. These checks generally showed no hints of serious violations of modelling assumptions. The prediction error or model goodness was measured based on the Akaike information criterion (AIC), as well as the marginal and conditional R² derived by the function “performance” from the R-package “performance”.

Prior to the analysis we checked whether further climate parameters not included into NZ (annual average temperature, annual precipitation) contribute to the explanation of the single tree growth. Moreover, we checked whether there exists a significant interaction between NZ and tree species. Neither was the case. All statistics were computed using the free statistical software R (R Core Team, 2020). For the production of the map (Fig. 1) we used the raster package (Hijmans et al., 2015).

ResultsTop

The mixed regression model for testing AWC effects by NZ on height growth explained 43% of the total variance (marginal R², Table 1). Mean tree growth was similar at the first three niche zones (NZ: 1_O – 3_M) levels but significantly lower at rear edge (4_R) (Fig. 2a). This reflects the low soil moisture regime at rear edge (4_R) stands. With respect to studied hypothesis, we observed significant differences in achieved height between NZs when accounting for different levels of AWC at the small-scale. We found a negative effect of AWC on growth at optimal sites, but a significant positive effect on growth of stands at intermediate sites (2_I) and the rear edge (4_R) with unfavourable soil conditions. For marginal sites with zonal soil conditions (3_M), we did not find a significant influence of individual AWC.

Figure 2. Effect on height growth of single trees of the studied key variables according to the regression model (Table 1). Main effect of niche zone (=line, partial residuals=points) (a) and the interaction between niche zone and AWC the focal relationship of our study (b).

As we did not see a stepwise increase in the AWC influence with increasing marginality along the NZs from optima (1_O) to the rear edge (4_R), the result only partially confirms our hypothesis. Additionally, we did not expect a negative effect of AWC at optimal sites. However, under the most unfavourable site conditions with low water regime and water storage due to larger gravel contents and/or sandy soil texture (4_R), the promoting effect of AWC on single tree growth was maximal. The estimated mean partial effect on height growth over the whole range of AWC values was almost 8 m at those sites (Fig. 2). The marked contribution of the hypothesised interaction effect (21% of explained variance) indicates that AWC was particularly affecting growth on sites in the optimal zone (1_O) and the rear edge (4_R) representing sites with low water regime.

The growth performance of the three tree species was significantly different, with fir and spruce equal to each other and beech with markedly lower growth performance (Table 1, Fig. 3). The NR was not statistically significant.

Figure 3. Effect on the height growth of single trees of further covariables according to the regression model (Table 1): height growth (a) and nutrient regime (b). The effect of the nutrient regime was not statistically significant (p > 0.05). Ba: acidic/poor, Mb: medium, Br: rich.