Introduction

Mixed-species forests have been promoted as a major pathway of forest management to mitigate the effects of global change, to increase the productivity of forests, and to enable a sustainable provision of forest ecosystem services in general because they can be more resistant and resilient towards climate change than monocultures (Zhang et al. 2012; Grossiord et al. 2014; Forrester and Bauhus 2016; Barbeito et al. 2017; Jactel et al. 2017; Ammer 2019; Forrester 2019; Messier et al. 2022).

Complementarity is a reduction in competition due to interspecific differences in resource use, or a facilitative effect, where plants interact in such a way that at least one species positively influences another (Vandermeer 1989). Mixing tree species can lead to complementarity through physiological, phenological, and morphological differences (within and between species) that result from interspecific interactions in mixed-species forests.

Light dynamics in mixed-species forests are directly driven by the spatiotemporal complementary use of canopy space. Mixing tree species with contrasting phenology (e.g. deciduous vs. evergreen) or differences in crown shape, crown size, crown leaf area density, as well as different tolerances to shade, influences light interception, which influence tree growth (Ishii and Asano 2010; Niinemets 2010; Sapijanskas et al. 2014; Forrester et al. 2018; Williams et al. 2021). For instance, a well-studied model of mixed planted forests is mixtures of Eucalyptus and nitrogen (N)-fixing species, where growth, APAR, and LUE were all greater in mixtures than in monocultures (Bauhus et al. 2004; Binkley et al. 1992; Forrester et al. 2012a, b). The drivers behind increases in APAR were greater crown sizes, higher leaf area index (LAI), and canopy stratification resulting in a reduction of light competition for one or both species (Bauhus et al. 2004; Binkley et al. 1992; Forrester et al. 2012a, b). Increases in LUE were associated with higher rates of photosynthesis (Forrester et al. 2012a, b), which could have been caused by higher availability or use of other resources such as nitrogen, phosphorous, and water, and shifts in carbon allocation from below ground to above ground (Forrester et al. 2005, 2006, 2010). In contrast, on sites where Eucalyptus outcompeted the N-fixing species, Eucalyptus monocultures had the highest biomass growth and LUE (le Maire et al. 2013). APAR was intermediate between the mixtures (50% mix of each species) and monocultures of the N-fixing species, which had the lowest growth and APAR. In these stands, Eucalyptus trees were much larger than the N-fixing species (le Maire et al. 2013). The APAR of individual Eucalyptus was about 169% greater in mixtures than in monocultures, while that of the N-fixing species in mixtures was only about 68% of that in monocultures (le Maire et al. 2013).

Differences in stand density and thinning can also influence APAR and LUE. Individual tree growth and APAR of Eucalyptus species increased following thinning in monocultures, and in some cases LUE also increased (e.g. Forrester et al. 2013; West and Osler 1995). Similarly observations were made for mixtures where density effects led to an average 14% increase in APAR for mixed stands of Pinus sylvestris and Fagus sylvatica (Forrester et al. 2018).

More recent studies about the contribution of light-related interactions or structural characteristics on APAR and LUE found that in addition to the structural and spatial characteristics (e.g. crown architecture, positioning of the different trees, leaf angle distributions, etc.), phenological differences between species can lead to complementary use of light (Forrester and Albrecht 2014; Sapijanskas et al. 2014; Forrester et al. 2018, 2019). However, the relative contribution of these structural and phenological factors to canopy APAR depends on the species and site characteristics, as well as how silvicultural interventions influence stand structure. Therefore, the relative contribution is difficult to predict given the small number of studies. A better understanding could aid the design of silvicultural interventions that further improve APAR, LUE, and hence forest growth and tree vitality. In addition, a better understanding of light-related tree species interactions in relation to canopy structure can improve the modelling of photosynthesis and productivity in process-based forest growth models (Pretzsch et al. 2015; Bravo et al. 2019) and Earth system models (Braghiere et al. 2019).

Two tree species that are currently of great interest economically and ecologically in Central Europe are Fagus sylvatica and Pseudotsuga menziesii. The native F. sylvatica has a broad ecological amplitude, a high growth performance, abundant natural regeneration, a high degree of shade tolerance and produces dense canopies (Brumme and Khanna 2009). Owing to these characteristics, F. sylvatica is a very competitive and dominant tree species that produces almost monospecific stands in most situations where site conditions are not extreme (Leuschner and Ellenberg 2017). P. menziesii, which is native to western North America, was introduced to Western and Central Europe, where it has become an important tree species due to its adaptability, wood quality, and high productivity (Kleinschmit and Bastien 1992). It can have a higher drought tolerance compared to European conifer species (Eilmann and Rigling 2012; Vitali et al. 2017) and is therefore viewed as an alternative to the most widely cultivated conifer Picea abies, which is experiencing a massive reduction in its cultivated area owing to its susceptibility to drought, wind storms, and bark beetles (e.g. Mezei et al. 2017). The mixing of these two species provides several advantages. Mixing reduces potential risks associated with the loss of a species owing to new pests or pathogens. Mixing an introduced tree species with a native species also buffers any potential negative effects the species may have on native biodiversity at the stand scale (Bauhus et al. 2017). In addition, mixing the shade-intolerant P. menziesii with the very shade-tolerant F. sylvatica prevents the spread of the non-native species through natural regeneration (Bindewald et al. 2021). From a forest growth perspective, P. menziesii can benefit from mixing with F. sylvatica in terms of drought stress release and time of growth recovery after drought events, and can also grow faster when mixed with F. sylvatica (Thurm et al. 2016). This was attributed to the height stratification of these two species and complementary crown architectures (Thurm and Pretzsch 2016; Thurm et al. 2016), which may lead to a higher stand-level light interception. Therefore, the objective of this study was to quantify the light absorption and light use efficiency of F. sylvatica and P. menziesii stands, and how these are influenced by the crown and canopy architecture. We also took advantage of a nearby stand density trial to examine whether differences in stand density (at a given time) can have comparable effects on light absorption of P. menziesii compared with mixtures. With measurements of tree dimensions, crown architecture, and stand structure, as well as the establishment of allometric equations, we tested the hypotheses:

1) Increases in APAR and LUE are both responsible for complementary effects on tree growth in mixtures of Fagus sylvatica and Pseudotsuga menziesii.

2) Intra-specific differences in crown architecture (in addition to interspecific differences) contribute to increases in APAR and LUE in mixtures.

3) Thinning or low stand densities increase individual tree growth of Pseudotsuga menziesii because they increase tree APAR, but only influence LUE on sites where thinning or low density also increases the availability of soil resources (e.g. not on sites that are nutrient-rich and moist).

Methods

Site description

The study sites were located in Bavaria, Germany (Table 1, Fig. 1). Mixing effects were examined at three sites (Ensdorf/Hirschwald; Gramschatzer Wald; Walkertshofen), while the effects of planting density and thinning on P. menziesii were examined at two additional sites (Ansbach; Heigenbrücken).

Table 1 Site characteristics: Climate data are annual means based on monthly means for the last three decades from 1988 to 2017 (ERA-Interim reanalyses from ECMWF)
Full size table
Fig. 1
figure 1

Location of study sites in Bavaria. The inserted map shows the location within Germany

Full size image

At each site of the mixture experiment, there were three plots, forming a “triplet”, consisting of a F. sylvatica and P. menziesii mixture and their respective monocultures. The plot size of the monocultures ranged from 170 m2 to 500 m2 and the mixtures from 725 m2—1115 m2. Within each triplet, the site conditions were as similar as possible, and the stands had not been thinned for at least a decade. All plots were close to even-aged (mean age 54 years) and had been regenerated conventionally through a mix of planting and natural regeneration. Plots had not been established according to a strict experimental design with fixed species proportions and spatial patterns.

To examine the effects of planting density and thinning on P. menziesii, two similarly designed experiments were used. The Heigenbrücken experiment contained 22 plots and the experiment in Ansbach contained 21 plots. All plots had an area of 40 × 40 m2 including a 30 × 30 m2 core area and a five metres wide buffer strip. The entire site was surrounded by a 10 to 20 m wide buffer zone. The planting density experiments were planted in May 1973 using three-year-old P. menziesii with 1000, 2000 or 4000 trees per ha as the initial planting density treatments. The experimental plots were thinned between planting and the time of fieldwork for this study such that an average of 33 trees per plot remained, regardless of initial planting density. Thus, the initial density of 1000 trees ha−1 corresponds to a light thinning treatment, 2000 to a medium, and 4000 to a heavily thinning schedule (Nickel and Rais 2013, 2015). At each of the sites, two heavily thinned, two moderately thinned, and two lightly thinned plots were sampled.

The methods described below were applied to all plots, with the exception of relative productivity calculations that were done exclusively for the plots in the mixing experiment.

Data collection

Estimation of leaf area, the vertical distribution of leaf area, and the vertical distribution of leaf area density

To estimate the individual tree leaf area, we used the Adelaide technique, described by Andrew et al., (1979) and validated by O'Grady et al., (2000). In this field campaign, we used a frame, containing nine horizontal lines, mounted on a tripod to divide an individual tree crown into ten horizontal layers. For each species, we selected 25 trees at each site that covered the range in tree size. On each of these trees, an individual crown unit was determined as a reference. For each tree, we counted the number of units per layer and sampled this reference afterwards to determine the weight, specific leaf area, and hence leaf area of a single unit. For the leaves of F. sylvatica, we used a leaf area metre (LI‑3050C Transparent Belt Conveyor Accessory (LI-COR Biosciences GmbH, Bad Homburg, Germany)) and for P. menziesii we used a flat-bed scanner combined with the WinFOLIA™ Reg 2013a software (Régent Instruments, Quebec City, Canada). In addition, we measured the diameter at breast height (DBH, 1.3 m), height, and live crown length of target trees and their neighbours within a radius of 10 m. Using these data and the specific wood density (WD) of both species, we calculated a competition index (Forrester et al. 2017),

$$CI= \frac{{\sum }_{t=1}^{n}{ba}_{P.menziesii}*WD+{\sum }_{t=1}^{n}{ba}_{F.sylvatica}*WD}{{r}^{2}*\pi }$$
(1)

where ba is the cumulative basal area of all individual tree stems at 1.30 m per species within a radius (r) of 10 m and WD is the species-specific wood density (g cm−3). This index was used to develop an allometric equation (Eq. 2) to calculate the crown diameter as a function of the competition index (CI), DBH, and relative height (rh) (height of the target tree divided by the mean height of its neighbours), and we used the mean wood density from values found in the literature and in the Global Wood Density Database (Zanne et al. 2009), for P. menziesii (Alden 1997; CABI, 2017) and F. sylvatica (Barbaroux 2002; Cienciala et al. 2005; Gryc et al. 2008; Skovsgaard and Nord-Larsen 2012)

$$Ln\left( Y \right) = \ln \beta + {\beta_1}\ln \left( {DBH} \right) + {\beta_2}\ln \left( {rh} \right) + {\beta_3}\ln \left( {CI} \right) + \varepsilon$$
(2)

where Y is crown diameter and ε is a correction ratio used when back-transforming the ln-transformed Y-variables. The correction ratio was calculated as the sum of the measured values divided by the sum of the (back-transformed) predicted values (Snowdon 1991).

The inclusion of a competition index was important because the allometric relationships used to predict some tree data, such as crown diameter in this case, can be influenced by stand structure and composition, as expressed by characteristics such as competition, relative height, and species richness (Forrester et al. 2017). This study also found that interspecific differences in response to competition can be related to the wood density of tree species such that species with higher wood density responded to increased competition by developing larger crowns, whereas species with lower wood density developed smaller crowns (Forrester et al. 2017). Since wood density may moderate the influence of competition, we included it in our calculations.

The leaf area density (LAD, m2 m−3) was calculated for the 10 crown layers. Each layer represented 10% of the live crown length and was assumed to have a volume of a semi-elliptical cone for the top layer or a frustum of a semi-elliptical cone for the other nine layers,

$$V\; = \frac{2}{3}\pi *h*{r^2} + \mathop \sum \limits_{l = 2}^{10} h\;*\left( {d_{{\text{basal}}}^2 + d_{{\text{top}}}^2 + {d_{{\text{basal}}}}*{d_{{\text{top}}}}} \right)*\frac{\pi }{12}$$
(3)

The vertical distribution of LAD was fitted to a beta distribution (Wang et al. 1990),

$$Y = a\left( {x^b} \right)*{\left( {1 - x} \right)^{cc}}\;$$
(4)

where a, b, and cc are fitted parameters, and Y is the leaf area density (units) at a relative height of x within the canopy. The LAD of one layer was a function of its relative height within the crown. The parameters of this equation are inputs required to run the MAESTRA model, which was used to calculate individual tree APAR.

Field measurements and further calculations to describe the structure and growth at the tree and stand levels.

We mapped stem positions and measured the DBH of all trees within the triplets, all trees in the density plots, and the next two tree lines around each plot. For 10 trees per species in each plot, we also measured height (h), live crown length (lcl) and crown diameter (cd). These variables were then determined for all other trees by fitting these variables as a function of DBH,

$${\text{ln}} (Y) = {{\text{ln}}\beta }_{0}+{\beta }_{1}{\text{ln}}(DBH )+\varepsilon$$
(5)

where Y is h, lcl, or cd, and ε is a correction ratio used when back-transforming the ln-transformed Y-variables. The correction ratio was calculated following Snowdon (1991). Individual crown volumes (m3) were calculated for each tree from their crown diameter (cd) and lcl by assuming semi-elliptical shapes (Eq. 3). Separate equations were developed for each species and each site.

The basal area increment of individual trees (cm2 year−1) was calculated using the DBH measurements of the triplet inventory in 2013 (see Thurm et al. 2016) and the density inventory in 2015 (see Nickel and Rais 2015). For the calculation of the individual leaf area (LA) (m2), we used allometric equations according to the general format as in Eq. 6 with corresponding variables (X1, Xn), parameter values (ß0, ß1), and correction ratios (ε) provided in Appendix (Table A1). These were developed in similar stands to this study and also covered the range of our data in terms of DBH, basal area (m2 ha−1), tree density (trees ha−1), and age.

$${\text{ln}}\left(Y\right)={{\text{ln}}\beta }_{0}+{\beta }_{1}{X}_{1}+{\beta }_{2}{X}_{2}\dots +\varepsilon$$
(6)

The specific leaf area (SLA in m2 kg−1) was calculated for each species as a function of age using equations (Eq. 7). The respective leaf area results were then averaged. When leaf mass was the response variable, it was previously divided by the SLA,

$$SLA\left(t\right)={SLA}_{1}+\left({SLA}_{0}-{SLA}_{1}\right){e}^{-\left(ln2\right){\left(t/{t}_{SLA}\right)}^{2}}$$
(7)

The parameters used in Eq. 7 for F. sylvatica and P. menziesii are provided in Appendix A (Eq. A1 and Eq. A2). Stand variables calculated included the relative height, mean stand height (m), tree density (trees ha−1), leaf area index (m2 m−2), basal area (m2 ha−1), basal area increment (m2 ha−1 year−1), canopy depth (m), canopy volume (m3 ha−1), and canopy leaf area density (m2 m−3). The relative height of a given species was calculated as the mean height of the target species divided by the mean height of the total stand.

Climate data

All meteorological data were obtained from the ERA-Interim reanalysis daily dataset produced by the European Centre for Medium-Range Weather Forecasts (ECMWF) with a high spatial resolution (0.125° × 0.125) (Dee et al. 2011). The climatic variables included monthly and daily mean temperature, precipitation, solar radiation, and number of frost days. For all plots, we extracted daily data of surface solar radiation downwards (SSRD), from 2012 to 2017 and monthly climate data from 1988 to 2017.

MAESTRA light modelling to estimate APAR

Individual tree APAR was calculated using the MAESTRA model (Grace et al. 1987; Wang and Jarvis 1990; Medlyn 2004; Duursma and Medlyn 2012). This 3D tree-level model calculates the APAR of individual trees based on their crown architecture (e.g. crown dimensions and leaf area and leaf angle distributions), leaf optical properties (LAD and distribution, leaf transmittance, and reflectance), and the shading from neighbouring trees based on the positions of other trees within the plot (defined by x and y coordinates and slope and aspect of the site) and their crown architectures. The growing season was defined by the site-specific phenology of the leaf-bearing period of F. sylvatica. For P. menziesii, APAR was calculated for the growing season (defined by the leaf-bearing period for F. sylvatica) and also for the whole year. The parameterization data are provided in Appendix (Table A3). To validate the APAR predictions, we compared the stand APAR calculated using MAESTRA with that calculated from hemispherical photos that were analysed using the Hemisfer software (Schleppi et al. 2007; Thimonier et al. 2010). The APAR values of both methods were highly correlated (R2 = 0.91) and indicated that MAESTRA underestimated Hemisfer-APAR by only 6.8% (Appendix Fig. A1). This level of accuracy is relatively high, and the slight underestimation by MAESTRA was considered to be acceptable. MAESTRA APAR predictions have also been validated in several other studies of mixtures (Charbonnier et al. 2013; le Maire et al. 2013; Forrester et al. 2018, 2019). The detailed 3-D representation of foliage distributions by MAESTRA enables it to accurately account for the influence of the sun zenith angle on canopy clumping. For example, it has been found to accurately reproduce the direct transmittance zenith profile, gap fraction, between solar zenith angles of 0 to about 70° (Braghiere et al. 2020). Individual tree APAR (GJ) was calculated for the growing season and the entire year. These were averaged (per year) for the entire study period (beginning of 2012 to the end of 2017). The stand APAR for the whole plot or for each species was calculated as the sum of the individual tree APAR of a given plot or the respective species.

Estimation of LUE and relative productivity (RP)

The LUE of individual trees (cm2 GJ−1) was calculated as the basal area increment (cm2 year−1) per unit of APAR (GJ year−1). Stand-level LUE (m2 GJ−1) was calculated as the stand basal area increment (m2 ha−1 year−1) per unit of APAR (GJ ha−1 year−1).

The relative productivity (Eq. 8 and 9) quantifies whether mixing led to an increase (RP > 1) or decrease (RP < 1) in a given variable compared to a monoculture at the stand level per species (Eq. 9) or the whole plot community (Eq. 8) (Williams and McCarthy 2001; Forrester and Pretzsch 2015). The RP calculation also accounts for the proportion of the species in the mixtures, which was quantified in terms of LAI or stand basal area, depending on the response variable. For response variables including tree density, basal area, and basal area increment, we quantified the species proportions in terms of stand basal area. All other proportions for the canopy and light variables (canopy depth, canopy volume, proportion of canopy filled, LAD, APAR, and LUE) were quantified in terms of LAI because it is considered to be more representative of the species’ contributions to light dynamics, and species proportions by leaf area have been shown to provide realistic estimates of species proportions on an area basis (Dirnberger and Sterba 2014).

$${{\text{RP}}}_{{\text{community}}}= \frac{{p}_{{mix}^{sp1, sp2}}}{{\colon\colon }_{{\text{sp}}1}{p}_{{mono}^{sp1}}+{\colon\colon }_{{\text{sp}}2}{p}_{{mono}^{sp2}}}$$
(8)
$${{\text{RP}}}_{{\text{species}}}= \frac{{p}_{{mix}^{sp1, (sp2)}}}{{\colon\colon }_{{\text{sp}}1}{p}_{{mono}^{sp1}}}$$
(9)

Equation 8 (\({p}_{{mix}^{sp1, sp2}}\)) calculates the productivity (or the performance of other response variables) of the mixed-species stand in relation to the productivity of their related monocultures (\({p}_{{mono}^{sp1}}\), \({p}_{{mono}^{sp2}}\)) and their proportional (\({\colon\colon }_{{\text{sp}}1}\), \({\colon\colon }_{{\text{sp}}2}\)) amount of LAI or stand basal area. Equation 9 is used to determine the relative productivity by species where (\({p}_{{mix}^{sp1, (sp2)}}\)) is the productivity of a given species (sp1) in a mixture with (sp2).

Results

Tree-level effects of mixing, density, and stand structure

For F. sylvatica, crown variables such as crown diameter, live crown length, and crown volume were larger in mixtures than in monocultures (Table 2). F. sylvatica trees were also significantly taller in the mixed stands than in monocultures even though the taller P. menziesii in the mixtures reduced the relative height of F. sylvatica in the stand (Table 2).

Table 2 Mean tree variables in 41–63-year-old monoculture and mixture stands for each species across all three sites in Bavaria
Full size table

In contrast to F. sylvatica, mixing did not affect the crown parameters for P. menziesii. The growth and APAR of individual P. menziesii were much higher in the mixture than in monoculture, while the growth and APAR of F. sylvatica trees were similar in both treatments (Table 2). As a result, there were no treatment effects on LUE (Table 3).

Table 3 Differences between 41- and 63-year-old mixtures and monocultures across all three sites in Bavaria
Full size table

There were no significant differences between P. menziesii density treatments for any variables (Table 4).

Table 4 Mean tree variables in 46-year-old stands of P. menziesii across two sites in Bavaria
Full size table

Individual tree APAR was positively correlated with tree leaf area and relative height for both species (Fig. 2). For a given tree leaf area, APAR was greater in mixtures than in monocultures for P. menziesii, but for F. sylvatica APAR remained the same as in monoculture (Fig. 2). There was no influence of LAI on individual tree APAR (Appendix Table A5).

Fig. 2
figure 2

The effect of mixing on the relationship between tree leaf area (a) and relative height (b) for F. sylvatica and P. menziesii and the predicted APAR of individual trees per growing season. Data were fitted to (Appendix Eq. A4), and the statistical information is provided in (Appendix Table A5). While there were no significant effects of mixing (M in Table A5) for F. sylvatica, the lines in the graph could differ slightly because we used the actual means of each variable from the given treatments to plot these lines

Full size image
Table 5 Differences between the 46-year-old stands of P. menziesii across two sites in Bavaria
Full size table

Stand-level effects of mixing and density

There were no significant differences between mixtures and monocultures for any stand variables except annual APAR, which was highest in P. menziesii monocultures and lowest in F. sylvatica monocultures (Table 3). There were no significant differences between P. menziesii density treatments for any variables (Table 5).

Relative Productivity at the stand level

The relative mixing effects, in terms of RP, showed that F. sylvatica benefited from the mixture in terms of nearly all variables, except for APAR (Table 3). P. menziesii also benefited in terms of canopy depth, LAD, growth, and LUE, but not in terms of basal area and canopy volume. The results indicate that the respective proportions of the two species make much better use of the light (LUE) in mixed stands compared to their monocultures (RP = 2.91 F.syl | RP = 1.61 P.men) and the LUE of the total stand was slightly higher in mixtures. In addition, there was a small increase in growth (BAI) for each species in the mixture and the total stand (Table 3). For the variables Trees per hectare and Basal area, F. sylvatica also achieved values with RP > 1, while P. menziesii had RP < 1, which can be attributed to the proportions of the two species in the mixture, as there were many smaller F. sylvatica and fewer but larger P. menziesii (Table 3). When APAR was evaluated, F. sylvatica and P. menziesii showed nearly the same relative productivity in mixed stands as in the corresponding monocultures (Table 3).

Discussion

P. menziesii and F. sylvatica made better use of the light in the mixture, as indicated by the relative productivity index (Table 3). The F. sylvatica trees were able to increase their height and crown dimensions below the P. menziesii to the extent that they grew and absorbed light as well as the F. sylvatica trees in the pure stand. The trees of P. menziesii, which were much better exposed in the mixture compared to their pure stand, achieved a significant increase in APAR (85.2%) and basal area growth (185.2%), but this was not sufficient to achieve a higher productivity for the whole mixed stand. Therefore, we could not confirm hypothesis 1: When growth is higher in the mixture, it is associated with increases in both APAR and LUE. Very few other studies have quantified mixing effects on growth and APAR or LUE along site gradients (Forrester and Albrecht 2014; Forrester et al. 2018). One of these studies examined Picea abies and Abies alba mixtures and found that for a given tree size, APAR (for both species) and LUE (of A. alba trees) were greater in mixtures than in monocultures and this effect increased as climatic conditions improved and on sites with faster-growing trees (Forrester and Albrecht 2014). The second study examined Pinus sylvestris and F. sylvatica mixtures and found that mixtures had greater APAR than monocultures, on average, and this effect increased with site productivity. However, this was mainly due to mixing effects on P. sylvestris (Forrester et al. 2018). Mixing or complementarity effects on growth can increase with increasing site quality because light-related interactions are likely to become more important as site quality increases. That is, as soil resources or climatic conditions improve, stand density or stand leaf area may increase, thereby increasing competition for light and the potential benefit of species interactions that improve APAR or LUE. However, in this study, the site quality was similar across all sites, which may explain the small influence of site mixing effects.

Our hypothesis 2 was that intra-specific differences in crown architecture (in addition to interspecific differences) contribute to increases in APAR and LUE in mixtures. Several studies have found that individual tree APAR is strongly related to individual tree LA, relative height and stand LAI and that mixing could modify these relationships (Forrester et al. 2018, 2019). However, there were no mixing effects on APAR for P. menziesii and F. sylvatica in mixtures, nor were there any mixing effects on relationships between individual tree foliage area and APAR, or between individual tree stem diameter and foliage area, or between individual tree stem diameter and crown diameter or crown length.

In mixtures, we observed that for a given tree foliage area, P. menziesii had a greater growing season APAR than F. sylvatica (Fig. 2). F. sylvatica had lower APAR per foliage area because it was the shorter species in the mixture, although F. sylvatica had higher foliage area for given stem diameter. The differences in phenology (deciduous vs. evergreen) did not affect the APAR of P. menziesii, because P. menziesii was taller than the deciduous F. sylvatica and hence its crowns also received high amounts of radiation during the growing season. Similarly, Pinus sylvestris benefited little from the deciduous characteristics of F. sylvatica in mixtures across Europe (Forrester et al. 2018). In contrast, differences in phenology were shown to play a significant role in the mixing effects on APAR in tropical forests where there was less distinct canopy stratification (Sapijanskas et al. 2014).

For both species, relationships between individual tree foliage area and APAR were linear indicating that all trees had a similar APAR relative to their foliage area. A review including data from several hundred plots and at least seven species found a range of linear and exponential relationships, showing that while competition for light is sometimes asymmetric (exponential foliage area – APAR relationships), this is often not the case (linear relationships) (Forrester 2019).

Our hypothesis 3 was that thinning or low stand densities, while increasing individual tree growth owing to increased tree APAR, only influence LUE under site conditions where these treatments also increase the availability of soil resources (e.g. unlikely on sites that are nutrient-rich and moist). However, the P. menziesii spacing experiment showed no significant differences in stand growth, APAR, or LUE between these two sites, and therefore, this hypothesis could not be tested. The lack of planting density effects is probably because the large initial differences in planting density were gradually lost over the 46 years due to thinning, which was more intense in the plots with initially higher planting densities, thereby removing the density effect. Studies on Eucalyptus nitens and Eucalyptus regnans showed that the positive thinning effects on LUE are suppressed when sufficient resources are not available (e.g. nutrients, water) or the competition for resources becomes more intense (West and Osler 1995; Forrester et al. 2012a). Thus, it remains an open question whether thinning or low stand densities can increase stand LUE on sites with a limited supply of soil resources.

This study shows the value of comparing tree- and stand-level growth, APAR and LUE in mixtures, and at different densities, to understand how species interact in mixtures. Mixing can modify APAR and LUE by combining, or modifying, various structural and physiological characteristics, including contrasting phenology (e.g. deciduous vs. evergreen) or differences in crown shape, crown size, crown leaf area density, as well as different tolerances to shade (Ishii and Asano 2010; Niinemets 2010; Sapijanskas et al. 2014; Forrester et al. 2018; Williams et al. 2021; Plaga et al. 2023). These studies also indicate that there is a strong context dependency on the effects. That is, the light-related interactions that have the greatest influence on growth depend on the species that are mixed, as well as age, stand density, soil, and climatic characteristics. This implies a potential for using forest growth models that incorporate and hence summarize the influence of crown and canopy structure on APAR and how this is influenced by site characteristics and management. For this purpose, it will be important that such models have been validated for their ability to predict mixing effects in forests (e.g. Bouwman et al. 2021).

Conclusion

The complementarity of canopy structure and related effects on APAR and LUE in these two species benefits P. menziesii concerning APAR, and most importantly, mixing does not lead to decreased vitality or growth in either of the two species. Since mixing is spreading potential risks from stress or disturbances and may also improve the nutrient use of these two species (Pretzsch 2014; Thurm et al. 2017), these forests might be more resilient, which is becoming an increasingly important attribute. However, this should be tested for other contexts in which this species combination can occur in terms of site conditions, age, stand density or mixing pattern. Since P. menziesii will likely increase in importance in Central European forests, there is a need to also explore other and more diverse species combinations with this species to reduce potential risks of its cultivation.