Production of hybrid aspen on agricultural land during one rotation in southern Sweden

Nils Fahlvik · Lars Rytter · Lars-Göran Stener

Abstract High potential productivity together with short rotation periods have made hybrid aspen an interesting option for wood production on former arable land in Nordic countries. In this study, some of the oldest active experimental plots with hybrid aspen in Sweden were remeasured at 23-30 years of age. A main aim was to assess age and productivity at the time of maximum mean annual volume increment. In addition, the influence of commercial thinning on stand development and differences in genetic gain among clones were investigated. Data from five experiments in southern Sweden were used, including three genetic trials,one demonstration stand with a clone mixture and one stand regenerated from root suckers. The three genetic trials were treated as single plot experiments, subject to a standard thinning program. In the remaining experiments, different thinning strategies were tested in a balanced block design. Volume growth had culminated or was close to maximum at age 25-30 years. Mean annual stem volume increment at culmination was 20-22 m 3 ha −1 a −1 . Dominant height reached 30-35 m at 28-30 years of age. Mean diameter at breast height was 27-29 cm after 29-30 years in the genetic trials.Clonal ranking based on diameter at age 7-9 years was positively correlated with the ranking at the final measurement in the genetic trials, 20 years later. This indicates that clones can be selected for superior growth based on results from young trials. More intense thinning programs increased the mean diameter compared to light thinning. The study indicates that one or two early and relatively heavy thinnings can promote the development of crop trees, without jeopardizing total volume production during a rotation of 25-30 years.

Keywords Populus tremula × Populus tremuloides ·Rotation length · Thinning · Biomass

Introduction

Biomass is an important renewable resource as a substitute for less climate-friendly products, and forest species have the potential to sequester large amounts of carbon in northern latitudes. An increased use of woody biomass is probably a prerequisite to meeting future climate goals in the Nordic countries (IEA 2013; Nordic Energy Research 2016). Woody biomass can, for example, be used as biofuels and substitute for other materials in house building. To increase available amounts of biomass, sustainable forest productivity must also increase. One way of doing this is to introduce fast-growing tree species, and species within the genus Populus are among the fastest growing in the Nordic countries. An important representative of the genus is hybrid aspen (Populus tremula L. × P. tremuloides Michx.), a cross between European and North American aspens. It has been planted extensively for almost 70 years in southern Sweden(Johnsson 1953; Tullus et al. 2012).

Hybrid aspen is suitable for both biomass production and traditional production of roundwood. A management program for hybrid aspen has been proposed by Rytter et al. (2013). The program includes the possibility of both biomass and conventional roundwood production. It starts with a density of 1100 plants ha −1 and includes zero to two commercial thinnings during a rotation of 20-25 years. The rotation time may, however, be questioned if maximum yield is the focus of management. A recent study has shown that peak productivity was not reached until after 25 years of growth (Rytter and Stener 2014).

Productivity of conventionally managed hybrid aspen varies but has been reported to be in the range of 10-20 m 3 of stem wood ha −1 a −1 (Einspahr 1984; Tullus et al. 2012;Johansson 2013a). This corresponds approximately to a stem biomass (dry matter) of 4-7 Mg ha −1 a −1 when using a basic density of 350 kg m −3 for the wood (Stener 2010). An additional 20% should be added to the biomass in well-managed stands when branches are included (Rytter and Stener 2003).

Consideration has also been given to the use of hybrid aspen for short rotation coppice systems (Liesebach et al.1999; Tullus et al. 2013) where productivity can be at least as high as under more conventional management (Rytter and Rytter 2017).

The genetic improvement of hybrid aspen has been implemented in Sweden since the late 1980s and the focus has been on improvement of growth and resistance to canker. As a first step, plus-trees were selected in stands planted in the 1940-1960s, followed by clonal testing in field trials (Stener 2004; Stener and Karlsson 2004; Stener and Westin 2017).Today, there are approximately 20 selected clones recommended for planting in either southern or northern Sweden.The average stem volume production of the selected clones is estimated to be around 25 m 3 ha −1 a −1 , using a rotation of 25 years on fertile sites in southern Sweden (Stener and Karlsson 2004). This corresponds to a stem biomass of 9 Mg ha −1 a −1 . However, growth estimations for this and most other studies are based on growth data covering the initial 20 years of the rotation or less.

Until recently, information has been collected on the establishment, growth, management, diseases, and breeding potential of hybrid aspen. However, there is need for basic information on the growth and response of established stands in order to develop and ref ine management programs.Maximum sustainable yield has been an important criterion in selecting rotation length although optimum rotation might depend on other aspects such as economy and risk (i.e., Binkley 1987; Newman 2002). As yet we have no reliable estimation of the age at maximum growth and the associated productivity in stands under conventional forest management. The rational for thinning is to stimulate the growth of the remaining trees in order to increase the production of merchantable round wood. In contrast, total gross volume productivity might decrease with increasing thinning intensity (e.g., Mäkinen and Isomäki 2004). Knowing the balance between stand level and tree level growth response to thinning is important to designing management programs in line with overall silvicultural goals. It would also be important to see if the most productive clones selected based on early measuring were still among the best after 20-25 years at final harvest.

This study estimates age and yield of maximum mean annual volume increment under conventional forest management. The influence of thinning operations on stand development and the stability in clonal ranking over time were examined. Our hypotheses were that: (a) peak productivity will be reached between 25 and 30 years with a productivity greater than 20 m 3 ha −1 a −1 ; (b) stem diameter will increase but total volume production will decrease with increasing thinning intensity; and, (c) the clonal ranking will be stable over the rotation period.

Methods and material

Experimental sites

Five trial sites had been established on former agricultural land in southern Sweden and have been described in detail by Rytter and Stener (2014). Kavlås 1, Kavlås 2 and Ättersta were established as genetic trials, including 39, 72 and 64 clones, respectively (Stener 2004; Stener and Karlsson 2004). The initial design of the genetic trials followed a randomized block design using single tree plots (2.5 × 2.5 m)with no buffer between blocks and plots. The clones were mixed and randomly distributed within each block. The Ingelstad 2 site was established as a genetic trial for demonstration purposes with six plots and with the same six clones randomly mixed within each plot. It was later used as a thinning experiment. The clones included in the genetic trials were phenotypically plus-trees selected during the 1980s from older stands and trials in southern Sweden. The Dimbo site was established as a thinning experiment with nine plots in a 5-year-old second-generation root sucker stand, originating from a plantation in 1958 with hybrid aspen crossings from the 1940-1960s breeding program.

Plantations were established with one-year-old containerized hybrid aspen plants at a 2.5 m × 2.5 m spacing(1600 stems ha −1 ). The year before planting, the sites were treated with glyphosate herbicide to avoid vegetation problems. The total age used in the evaluation includes plant age at planting. On the Dimbo site, the year of final harvest of the first generation stand was considered as the time of establishment.

Research layout and stand management

In this study, the experimental sites were followed from plantings or root sucker initiation in 1990-1997 to the last measurement in 2018. The initial design of the genetic trials with small randomized blocks and single tree plots was considered unsuitable for long-term growth and yield studies. The Kavlås 1, Kavlås 2 and Ättersta sites were treated instead as single plot experiments (Table 1). The number of clones and replicates present at the first and last measurement of the genetic trials is shown in Table 2. A section of Kavlås 2 was seriously damaged by wind before the last measurement and was excluded in growth calculations. The initial 3600 m2area was reduced to 2250 m2after exclusion of this wind-damaged section.

The initial layout of six and nine plots in the Ingelstad 2 and Dimbo sites, respectively, were retained. Plots on both sites were distributed in a balanced randomized block design of three blocks. The reason for blocking was indications of gradients in properties within the sites. The Dimbo site was initially designed with a 5 m buffer zone surrounding each plot. There were no buffer zones in the genetic trials since they initially were evaluated before the onset of competition. A buffer zone of 5 m was introduced around Kavlås 1, Kavlås 2, Ättersta and Ingelstad 2 sites by excluding the two outermost rows in the analysis. Ingelstad 2 site was a rectangle of 2 × 3 plots without spacing between the plots.Except for the surrounding buffer zone of 5 m in Ingelstad 2,there were no additional buffer zones between neighbouring plots as it would have reduced plot size significantly. Net and gross areas of the plots are presented in Table 1.

Table 1 Site and stand characteristics

Table 2 Number of clones and replicates per clone present at first and final measurement in the genetic trials

Stand density of the initially dense root sucker regeneration (> 50,000 stems ha −1) at the Dimbo site was reduced by a pre-commercial thinning five growing seasons after the harvest of the first generation stand. The average density was 2300 stems ha −1 when it was established as a thinning experiment.

Thinnings of the single plot experiments followed common practice of one or two thinnings (Table 3). The first thinning focused on small dimensional trees, whereas the second thinning was more uniform (Table 3). In Kavlås 1, a third thinning was carried out at 24 years, affecting approximately 50% of the area. Thinning in the Ingelstad 2 site included two treatments: a light thinning program with asingle thinning and a moderate thinning program with two thinnings. The first thinning was the same in both treatments with similar thinning grades and a uniform removal over all dimensions (Table 3). The second thinning within the moderate program focused on small dimensional trees. Thinning on the Dimbo site included three treatments with different grades and timing (Table 3). Light thinning removed approximately 20% of the basal area; moderate thinning included four thinnings of approximately 25% basal area removed;and, the heavy treatment included two early thinnings with approximately 45% basal area removed. All thinnings on the Dimbo site focused on small dimensional trees.

Table 3 Average thinning grade (TG) and thinning quotient (TQ)

Measurements

Stem diameter and tree height measurements were repeatedly carried out from 6-10 to 23-30 years. Diameter at breast height (DBH, 1.3 m above ground over bark) was measured on all trees. Initial measurements on the Kavlås 1, Kavlås 2 and Ättersta sites included height measurements on all trees. For all other cases, heights were recorded from a random sample of trees (22-70 trees per experiment in 2018). Heights for all trees were estimated based on the statistical model:

where H is height (m), D BH diameter at breast height (cm),a and b are coefficients calculated separately for different experiments, treatments and measurement occasions. There was no indication of clonal differences in the DBD-H relationship. Dominant height (H dom) was estimated with Eq. 1 using the arithmetic mean for DBH of trees in the largest 100 DBH trees ha −1 .

Quadratic mean diameter was calculated as:

where D g is the quadratic mean diameter (cm), D BH diameter at breast height (cm) and N the total number of stems ha −1 . In addition, the arithmetic mean diameter was calculated for trees representing the 400 largest trees ha −1 (D 400 ).D 400 reflected current recommendations of stand density at the end of the rotation (cf. Rytter and Stener 2014). Mean annual increment in D 400 (iD 400 , cm a −1) was calculated for the period between the establishment of the thinning experiments and the final measurements at the Ingelstad 2 and Dimbo sites. Stem volumes were calculated according to Johnsson (1953):

where V is stem volume over bark (dm 3 ), D BH diameter at breast height (cm), H tree height (m). Mean annual volume increment (MAI) is the gross volume production, including dead and thinned trees. However, removals of root suckers at the pre-commercial thinning in the Dimbo site were not recorded and not included in the analysis.

Dry weight of stem wood was calculated by multiplying total stem volume with values of basic density for hybrid aspen. Separate values (dry matter) of density were used for different age classes; 1-5 years, 342 kg m−3; 6-10 years,329 kg m −3 ; 11-15 years, 339 kg m −3 , 16-20 years,352 kg m −3 and 20+ years, 369 kg m −3 (Stener 2010).

Statistical tests

Separate tests of thinning responses were performed in the Ingelstad 2 and Dimbo sites using:

where Y ij is observation ij, μ the mean value, t i the fixed effect of thinning, b j the random effect of block and ε ij the random error for observation ij. The significance level p ≤ 0.05 was used for testing the null hypothesis of no difference among treatments. Pair-wise comparisons among treatments were performed with Tukey’s studentised range test.The W-test of Shapiro and Wilk (Sabin and Stafford 1990)was used for tests of normal distribution of the response variables. The function “aov” in software R was used for statistical tests of thinning responses (R Core Team 2018).

The evaluation of stability in clonal ranking was based on the clonal least square means for diameter measured at an early age (Ättersta age 7; Kavlås 1 and 2 age 9), and at a late age (Ättersta age 29; Kavlås 1 and 2 age 30). Only trees present at both measurements were included in the calculations.The clonal means were transferred to relative values, i.e., in relation to the means of all clones, where 100 indicates the mean for all clones and a value above 100 represents clones growing better than average. All trees (incl. buffer zones)were included in calculations of clonal ranking.

Results

Stand characteristics at the final measurement were similar for Kavlås 1, Kavlås 2 and Ättersta (Table 4). At 29-30 years, the Dgwas > 26 cm and D 400 > 30 cm. The highest average DBH was found in Kavlås 1 which also had a lower stand density than Kavlås 2 and Ättersta. Significant differences in the D g were found between the light thinning and the more intense thinning on both Ingelstad 2 and Dimbo sites (Table 4). The D 400 tended to be larger at the final measurement for the more intense thinnings. Both D 400 and iD 400 were significantly larger for the heavy thinning treatment compared to the light thinning treatment onthe Dimbo site. On average, the D g and D 400 were 9.3 and 4.6 cm greater, respectively, and the iD 400 was 0.2 mm a −1 greater after heavy thinning compared to light thinning on the Dimbo site.

Table 4 Stand characteristics at the final measurement in 2018

The single plot experiments had a dominant height(H dom) of > 34 m at the final measurement (Table 4;Fig. 1). It was less in Dimbo site where it reached approximately 30 m at 28 years. Dominant height followed the height development curves of Johansson (2013b) during the first 20 years (Fig. 1), but thereafter Hdomtended to have a greater increase with age in the planted experiments. Dominant height on the Dimbo site increased rapidly in the early years but levelled out around 25 years.Arithmetic mean height increased with heavier thinning on both the Ingelstad 2 and Dimbo sites (Table 4).

Fig. 1 Average dominant height (m) in the study sites compared to height development curves; dashed lines, site index 6-27 and base age 20 years, of Johansson (2013b). The lines represent four planted(dark circles) stands and one originating from root suckers (white circles). In the Ingelstad and Dimbo sites, dominant height was calculated as an average for all thinning treatments

Fig. 2 Basal area by age for Kavlås 1, Kavlås 2 and Ättersta

Basal area development was similar in the single plot experiments (Fig. 2). At the final assessment, basal area in the moderately thinned plots on the Ingelstad 2 site had not recovered since the last thinning, and the absolute difference between the two treatments remained (Fig. 3). There was a reduced basal area development in the moderately thinned plots compared to the other treatments during the last decade on the Dimbo site. Moderately thinned plots also had the lowest basal areas and stem numbers at the final measurement (Table 4).

Maximum volume growth was achieved at 30 years on Kavlås 1 and Kavlås 2 sites and at approximately 25 years on the Dimbo site (Figs. 4 and 5). MAI of stem volume was approximately 22 m 3 ha −1 a −1 (8.2 Mg ha −1 a −1 )on Kavlås 1 and Kavlås 2 by and 20-22 m 3 ha −1 a −1(7.5-8.2 Mg ha −1 a −1) on the Dimbo site at culmination.MAI was still increasing by 29 years on the Ättersta site and also in the younger Ingelstad 2 site. Lightly thinned plots had the largest MAI on both Ingelstad 2 and Dimbo sites but significant differences were only observed between light and moderate thinning on the Dimbo site (Fig. 5).

Fig. 3 Basal area by age for different thinnings at the Ingelstad 2 and Dimbo sites calculated as plot averages within each treatment. Significant differences between thinning treatments at the final measurement are indicated by different letters (p< 0.05). Thinning treatments were performed according to Table 3

Fig. 4 Mean annual volume increment by age for Kavlås 1, Kavlås 2 and Ättersta

Fig. 5 Mean annual volume increment by age for different thinning treatments at Ingelstad 2 and Dimbo sites. Significant differences between thinning treatments at the final measurement are indicated by different letters (p< 0.05). Thinning treatments were performed according to Table 3

With regards to genetic stability, there was a positive relationship in ranking according to relative clonal least square means for diameters between early (ages 7-9) and late (ages 29-30) measurements (Fig. 6). All clones genetically selected for use in southern Sweden, but one, had values above 100 at both measurement occasions and averaged 115 (Kavlås 1), 112 (Kavlås_2) and 127 (Ättersta) at the final measurement.

Fig. 6 Relative clonal least square means for diameter at early (ages 7-9) and late (ages 29-30) measurements on Kavlås 1, Kavlås 2 and Ättersta sites, calculated in relation to the overall trial mean for trees left at the final measurement. The circles represent the average value for each clone included in the trials. Black circles represent clones genetically selected for use in southern Sweden. Pearson correlation(r) between early and late measurements is shown

Discussion

Experimental growth data covering an entire rotation of hybrid aspen are rare in the Nordic countries. The present study summarizes growth and yield of some of the oldest,still active, experiments in Sweden. The results indicate a maximum mean annual increment at 25-30 years on fertile sites. This is in line with recent forecasts by Rytter and Stener (2014). A maximum MAI of stem wood of 20-22 m 3 ha −1 a −1 (7.4-8.2 Mg ha −1 a −1) in this study agrees with estimates of MAI of 20 m 3 ha −1 a −1 or more for managed hybrid aspen in northern Europe (Stener and Karlsson 2004; Tullus et al. 2012, 2013; Johansson 2013a).The growth pattern on the Dimbo site was different, with a greater initial volume growth and an earlier culmination of MAI. This was possibly due to the regeneration method.The Dimbo stand originated from root suckers, whereas the other plots were planted. Studies in southern Sweden have reported rapid initial growth rates and height development of regeneration from root suckers (Mc Carthy and Rytter 2015).In this study, the potential production on the Dimbo site was underestimated; a large proportion of the early production was removed at a pre-commercial thinning after five growing seasons, and this volume and biomass were not included in our calculations. However, lower production could be expected on the Dimbo site as site fertility was lower compared to the other sites. The difference in growth patterns and growth levels between root suckers and planted stands was possibly also influenced by the origin of the materials.The root sucker stand on the Dimbo site originated from hybrid aspen crossings from the breeding programs of the 1940-1960s. The genetic trials were planted with material originating from plus tree selections carried out in the 1980s,and the genetic gain could be expected to be greater.

More intensive thinning programs slightly reduced the mean annual increment compared to lighter thinnings in both thinning experiments. A decrease in total production with increasing thinning grade has been observed in several studies of other tree species (e.g., Rice et al. 2001; Mäkinen and Isomäki 2004; Nilsson et al. 2010). However, the difference between treatments in this study was only significant for light and moderate thinning on the Dimbo site, and on the final measurement after the culmination of MAI. Basal area, following a moderate thinning on the Dimbo site (represented by repeated thinnings over an extended period),recovered poorly after the last thinning at 17 years. This could be due to an inability to fully utilize site resources in sparsely spaced stands or a failure of the remaining trees to react to late thinnings. This could be because the initial thinnings have been too light, resulting in reduced green crowns due to competition (cf. Miller 2000; Juodvalkis et al. 2005;Mc Carthy and Rytter 2015). An important objective of thinning is to increase the growth of the remaining trees. In fact, significantly higher average DBH occurred after more intensive thinning with early heavy removals compared to the light thinning in this study. The average diameter of the largest trees is usually less effected by thinning from below than the average for all trees (e.g., Penner et al. 2001; Mäkinen and Isomäki 2004). In this study, light and moderate thinning had only a minor effect on diameters of the 400 largest trees (D 400) in both thinning experiments, whereas a significant difference of 5 cm was noted between the light and heavy thinnings on the Dimbo site. Ingelstad 2 is the youngest stand in the study and the different thinning treatments might still influence stand development.

Hybrid aspen is suitable for the production of both biomass and timber. If the goal is to maximize biomass production, thinning regimes might be questioned (i.e., Steneker 1974; Bella and Yang 1991). However, management that balances the production of biomass and the development of individual trees could be a feasible alternative to maintaining the freedom to choose production objectives. The results in this study indicate that one or two early and relatively heavy thinnings might be a useful alternative in hybrid aspen, considering both production and diameter of remaining stems.

Dominant height development in this study was compared with height curves for hybrid aspen on farmland in southern Sweden (Johansson 2013b). The initial height development was in agreement with these height curves. However,in our study, planted stands older than 20 years had more rapid height development compared to these height curves.This might be due to few observations of stands older than 20 years in the data used for model construction by Johansson (2013b), especially on the most fertile sites. Another explanation might be differences in plant material. The stands in Johansson (2013b) originated from clone mixtures sampled in 1939-1960. The genetic gain should be greater in the more recent plant material used in this study.

Early and precise estimates of the genetically best performing clones are important in order to reduce the time of selection, as they directly affect the rate of genetic improvement. In this study, clonal ranking based on diameter at 7-9 years was positively correlated with the ranking of clones at the last measurement 20 years later (Fig. 6). Some of the hybrid aspen clones recommended for use in southern Sweden were included in the planted trials and they were all among the largest trees both at the early and late measurements. This indicates that an early selection is reliable,which agrees with previous studies. Strong genetic age ×age correlation, i.e., clonal correlations for growth at different ages, has been found for hybrid aspen (Li et al. 1993;Stener and Karlsson 2004). Other traits, such as stem quality and resistance to stem canker, usually are manifest at higher ages. Stener and Karlsson (2004) suggests that the final selection of clones should be done not before 10-15 years.

Of the 20 genetically selected clones recommended for use in southern Sweden, nine were included in the present study. Based on their individual growth performances, they had on average, 16% higher diameter at 29-30 years compared to all tested clones. This is in agreement with results from other genetic field tests in southern Sweden (Stener and Karlsson 2004) where genetic gain for diameter was 20% if the 10% best clones were selected from a total of 280 clones. In their study, a MAI for genetically selected hybrid aspen material in southern Sweden was estimated to be around 25 m 3 ha −1 a −1 . Based on the results in this study, this estimation is valid. However, the experimental design of the genetic trials might have implications on clonal ranking at later measurements. Different clones were randomly mixed and planted at a 2.5 m × 2.5 m spacing. The trials were initially intended for early evaluations of clones based on height measurements carried out before the onset of competition. Clonal ranking in this study was based on diameter measurements carried out after crown closure. It is likely that competition has resulted in an overestimation of the difference in grow rate between fast- and slow-growing clones. Fast-growing clones might have less competition,and slow-growing clones more in mixtures compared to monocultures.

Productivity in this study refers to well-managed stands,mainly on highly fertile sites. The sites of the planted stands were fenced and soil preparation was carried out before planting. Thus, mortality was generally low except for ones caused by wind damage in Kavlås 2 before the final measurement. A survey of hybrid aspen on farmland showed a wide variation in production levels in mostly unmanaged commercial plantations in southern Sweden (Johansson 2013a). A biomass (dry matter) of 42-219 Mg ha −1 was recorded in 15- to 23-year-old stands, corresponding to a MAI of 3-13 Mg ha −1 a −1 . The silvicultural history of the stands was not well-known but poor soil preparation before planting and damage by wildlife were plausible causes of heavy losses and low productivity in some stands (Johansson 2013a). Intense afforestation efforts with fencing and chemical or mechanical soil preparation are considered important to successfully establish hybrid aspen on agricultural land(Tullus et al. 2012). It is also important to select plant material adapted to the site, and clones recommended for different parts of Sweden should be used (Stener and Westin 2017).

Conclusion

Conventionally managed hybrid aspen stands on agricultural land in southern Sweden can reach a mean annual volume production of at least 20 m 3 ha −1 a −1 during a rotation of 25-30 years. Hybrid aspen represents a flexible management alternative, combining high biomass production with the development of merchantable dimensional wood over a short rotation. This study indicates that one or two early and relatively heavy thinnings (approximately 450 trees ha −1 in the final stand) will promote the development of crop trees,without jeopardizing total volume production. There was a tendency towards reduced volume growth after repeated thinnings during an extended thinning period. The data used for this study was limited by the scarcity of older experiments with hybrid aspen. Thus, there is a need for more data covering different site types and management alternatives.This study focused on growth and yield of hybrid aspen;economics and risks are other important factors to be considered in the ref inement of thinning recommendations and decisions upon rotation length.

AcknowledgementsThe authors are grateful for the financial support provided by the Swedish Forest Society Foundation. We also wish to thank the personnel at the Forestry Research Institute of Sweden(Skogforsk) for their valuable help with measuring and processing the hybrid aspen material. Furthermore, we are grateful to two anonymous reviewers for valuable comments on the manuscript.