1 Introduction

Young dense stands are a significant resource of biomass for the bioenergy market (Grebner et al. 2009; Karlsson et al. 2015; Fernandez-Lacruz et al. 2015). For example, Sweden, a country with boreal forests, has 23.2 million hectares of productive forest land, of which 7.2 million hectares are <30 years old (32.4%) (Swedish statistical yearbook 2014). Depending on the constraints applied, there are between 2.1–9.8 million ha of unthinned, biomass-dense forest in Sweden, representing 9–44% of the country’s productive forest land area. The minimum yearly harvesting potential of these biomass-dense forests is 3.0 million dry-weight (DW) metric tonnes (t) when calculated from delimbed stemwood (including tops), and 4.3 million DW t if the trees are harvested whole with limbs (Fernandez-Lacruz et al. 2015).

Biomass/bioenergy thinning (BT) represents one alternative for young dense stands, but it is important to consider how this silvicultural practice may affect individual trees and stand development. For example, research shows that late thinning of young dense stands increases the risk of snow damage, such as stem breakage (Valinger et al. 1994; Päätalo et al. 1999). Furthermore, stands of smaller diameter trees (diameter at breast height, DBH < 80 mm), which result from the thinning of young dense stands, are more susceptible to damage (Päätalo et al. 1999; Abetz and Klädtke 2002). On the other hand, there is evidence that BT can precede a later harvest of crop trees without risking crop tree value (Ulvcrona 2011).

Boom-corridor thinning (BCT) is a harvesting operation method that increases efficiency and cost-effectiveness by thinning strips of a defined size with boom-tip harvesting technology. Different boom corridor patterns and how they are applied give different degree of selective tree selection. For example, using a fan-shaped pattern, “laid out” by the decision of the operator, will give higher degree of tree selection than of using a perpendicular pattern laid out strictly systematically (Fig. 1). The combination of BCT and new felling technologies during BT operations (Bergström et al. 2007; Bergström 2009; Sängstuvall et al. 2011; Bergström and Di Fulvio 2014a) can significantly increase the harvesting efficiency of young dense stands. Bergström and Di Fulvio (2014a) provided quantitative evidence, showing that supply costs can decrease by 12–27% when conventional selective thinning and standard handling procedures are replaced with a BCT system that includes innovative biomass compaction and transportation methods. Furthermore, Bergström and Di Fulvio (2014a) reported that BCT systems are most efficient at harvesting trees under 30 dm³, a size class that according to Fernandez-Lacruz et al. (2015) represents a significant portion of the potential annual harvest in young stands in Sweden.

Fig 1. Sketch of possible selective thinning and two boom-corridor thinning patterns between strip roads, as described in Bergström (2009).

Boom corridor thinning techniques facilitate cost-effective management of young stands and also allow aspects other than wood production to be taken into account. For example, there has been a growing concern about how forest management plans can include ecosystem services and environmental goals without jeopardizing the production of valuable biomaterials (Bengtsson et al. 2000; Führer 2000; Sullivan et al. 2011). It has been suggested that stand heterogeneity and vertical complexity promotes various ecosystem services (Puettman et al. 2009; Messier et al. 2014). Moreover, Saastamoinen et al. (2014) have begun to classify the various ecosystem services so they can be recognized by forest management plans. However, the stand structure resulting from a management regime cannot only cater to ecosystem services, but must present a cost-effective solution as well. In this way BCT is an attractive option from a forest management perspective, as it is both cost-effective and should create a more complex vertical stand structure than conventional thinning. This is because BCT methods, when strictly used, only cut trees from pre-defined areas, leaving certain areas, or strips, untouched and with their natural stand structure. On the other hand, commercial selective cutting, which includes the preclearance of any undergrowth with a stem size smaller than the commercial size, thins, in general, an entire stand to a homogeneous area containing only larger diameter trees for future harvest. In this way, the first thinning phase largely determines the stand structure for a majority of the stand rotation period. There are many thinning operations to choose from, and each can result in a different stand structure.

The objective of this paper was to measure how conventional thinning (pre-commercial and commercial) and BCT affect the stand structures of young, dense, conifer stands with various heights and compare treatments. This study was done immediately after thinning and included both low- and high-thinning BCT regimes, which target the smaller and larger trees, respectively, yet result in a similar number of trees per ha.

2 Material and methods

2.1 Description of sites

Two series of field experiments were established. The first, BCT 1, with three sites in south of Sweden (6 and 11 m in mean and dominating tree height) were all planted with Norway spruce (Picea abies (L.) H. Karst.) (Table 1). Other species found were Betula spp., Salix ssp., Populus tremula L. and Sorbus aucuparia L. and Norway spruce from natural regeneration.

The second, BCT 2, with three regions in Sweden (in the north, centre and in the south) (Table 2). Each region consisted of two sites with different tree heights (4/9 m and 5/10 m in mean/dominating tree height). Planted Norway spruce and Scots pine (Pinus sylvestris L.) dominated all sites, and Betula spp., Salix ssp., P. tremula and S. aucuparia from natural regeneration were also found. Betula spp. was the most common deciduous species in all sites.

2.2 Study design

The treatments in BCT 1 with stand height 6/11 m (mean/dominating height) were:

C: Control with no treatment.

Conv: Conventional selective thinning, including strip roads. The Norway spruce was favored; the target number for remaining trees was about 1000 stems ha^–1 with an estimated removal of about 50% of the basal area.

BCT_{1 m}: Strict boom corridor (width 1 m, length 10 m) thinning. One corridor in each direction, 90° from each machine position, with 2.67 m between machine positions.

BCT_{2 m}: Strict boom corridor (width 2 m, length 10 m) thinning. One corridor in each direction (in total 2 corridors), 90° from each machine position, with 5.3 m between machine positions.

BCT_sel: Selective boom corridor (width 1 m, length 10 m) thinning. Two corridors (about 90° and 60°) at each side of the road (in total 4 corridors per machine position) with 5.3 m between machine positions. The corridors were not pre-determined and could during execution of treatment be selected to favor future Norway spruce crop trees. Fig. 1 shows the principal design for boom corridor thinning.

In the BCT 1-series each treatment plot had dimensions of 30 × 40 m, with a 5 m buffer zone on each side. Each site was referred to as a block.

In the BCT 2-series, the treatments for stands with mean/dominant tree heights of 4/9 m were;

BCT: Selective boom corridor (width 1 m, length 10 m) thinning. Two corridors (about 90° and 60°) at each side of the road (in total 4 corridors per machine position), with 5.3 m between machine positions. The corridors were not pre-determined and could be selected during execution of treatment to favor future crop trees. The target number for remaining (possible crop) trees was about 2000 stems ha^–1, which resulted in higher total number of stems ha^–1.

PCT: Conventional motor manual pre-commercial thinning (PCT). The target number of remaining trees was about 2000 stems ha^–1 and cut stems were left in the plot.

The treatments for stands with mean/dominant tree heights of 5/10 m were;

BCT_low: Low-thinning regime with selective boom corridors (width 1 m length 10 m), i.e. thinning from below (the smallest trees were harvested), to harvest the smaller trees and maintain to ca 2000 stems ha^–1 possible crop trees ha^–1 (higher total number of stems). Two corridors (about 90° and 60°) at each side of the road (in total 4 corridors per machine position), with 5.3 m between machine positions. The corridors were not pre-determined and could be selected during execution of treatment to favor future crop trees.

BCT_high: High-thinning regime with selective boom corridors (width 1 m length 10 m), i.e. thinning from above, to harvest the larger trees and maintain 2000 stems ha^–1. Two corridors (about 90° and 60°) at each side of the road (in total 4 corridors per machine position), with 5.3 m between machine positions. The corridors were not pre-determined and could be selected to favor future crop trees. This treatment had the same target basal area as Conv.

Conv: Conventional selective thinning to 2000 stems ha^–1 with an additional manual PCT of the smallest trees, leaving cut stems in the plot.

In the BCT 2-series each treatment plot had dimensions of 30 × 30 m, with a 5 m buffer zone on all sides. Each stand site was referred to as a block.

For each stand site, in both series, the plots were marked with poles in each corner. Machine positions (5.3 m distance and 2.67 m) and strip roads were also marked. Furthermore, for strict boom corridor treatments the pre-determined corridors were marked. The treatments were randomly distributed between the plots.

All treatments, except Controls and PCT, contain cut strip roads.

2.3 Execution of treatments

In both treatment series 6-wheeled thinning harvesters with a mass of ca 12 t were used for the thinning. The PCT was performed with professional use manual motor brush-saws. The BCT 1-series employed a Valmet 911 harvester (http://www.komatsuforest.se) equipped with the Bracke 16 head (Bracke Forest AB, Sweden) (http://www.brackeforest.com) and operated by an experienced driver. The PCT was performed by local contractors normally hired by the land-owner, and forest management company, Sveaskog. The BCT 2-series employed a Valmet 911 harvester equipped with the Bracke MAMA Prototype head (see Bergström and Di Fulvio 2014b), which was operated by a less experienced driver. However, the driver’s work was supervised during thinning. All of the thinning and PCT operations were carried out carefully to ensure consistency and minimize the damage to the future crop trees. The BCT 1-series treatments were performed from August 2013 to May 2015 and the BCT 2-series treatments were performed from October 2013 to January 2014.

2.4 Stand structure measurements and statistical tests

The DBH (1.3 m height) was measured for all trees within the study plots before treatments began. To ensure that all trees were measured only once during each inventory, corridors about 5 m wide were created with tape measures through each plot. Once a tree was measured, the stem was marked. The same procedure was also used after the treatments were performed. Additionally, in the BCT 1-series all trees with a DBH > 45 mm were numbered and the DBH measurement position was marked before the second inventory. Randomly-selected trees were used to measure tree height so that all tree sizes would be included in the measurements. Additionally, the 9 largest trees from 900 m² plots (BCT 2) and the 12 largest trees from 1200 m² plots (BCT 1) were measured. The selection process of numbered trees for height measurements in BCT 1-series plots followed routines developed by the Swedish University of Agricultural Sciences for long term forest field experiments (Karlsson et al. 2012).

The stem volumes of trees with DBH < 45 mm were estimated using volume functions according to Andersson (1954), and those of trees with DBH ≥ 45 mm were estimated with volume functions according to Brandel (1990). Equations from Repola and Ulvcrona (2014) were used to estimate the DW of biomass.

Trees with DBH > 80 mm were analyzed for comparisons of the largest trees in each treatment and as a measure to forecast the possible amount of future crop trees in the plot.

Analysis of variance, using the GLM procedure in Minitab 16 (Minitab2010), was then applied to evaluate (fixed) effects of the treatments (α), with sites as blocks (random effects, b) using the following model (Eq. 1):

where Y_ij is the dependent variable, μ is the grand mean, α is the fixed effect of treatment, b the random effect of site and ε the residuals.

Prior to analyses, Levene’s test was used to ensure that data showed a homogeneity of variance (Tamhane and Dunlop 2000), and that no transformations were required (cf. Sabin and Stafford 1990). Tukey simultaneous tests for multiple comparisons of means were used to assess any significance of differences in mean values of measurements between treatments. For all statistical tests, detected differences were deemed significant if p ≤ 0.05.

2.5 Simpson index

Simpson index (D) was used to estimate the biodiversity among tree species (Simpson 1949), Eq. 2.

where n = the total number of organisms of a particular species, and N = the total number of organisms of all species.

Simpson index of diversity (1–D) represents the probability that two samples randomly selected from a sample will belong to different species (Simpson 1949). The value ranges between 0and 1, e.g. 0give no diversity (only one species).

3 Results

3.1 Stand structure parameters of BCT 1-series treatments

There were no significant differences between any plots in the stand parameters before the treatments began (Table 1). The stem number of the control plots (9635 ha^–1), which did not receive silvicultural treatment, was about eight-fold compared to conventional thinning (lowest stem number was 1168 ha^–1), and about three- and two-fold compared to BCT_{1 m} (3949 stems ha^–1) and BCT_sel (4568 stems ha^–1), respectively. There were significantly more stems remaining per hectare following BCT_{2 m} treatment as compared to the conventional thinning treatment (Table 1). There were no significant differences between treatments regarding either dominant height or mean height after treatment. Furthermore, conventional thinning showed a significantly higher basal area weighted diameter, 9.5 cm, (calculated from DBH) than the other treatments. When the average DBH was calculated for each plot from the same number of stems as were left in the conventional thinning plot, and based on the larger stems, there were no significant differences between treatments (Table 1). The two highest values for stem volume and total biomass (DW) were found in the control treatment (116.7 m³ ha^–1, 72.6 t ha^–1), and in the BCT_sel treatment (76.4 m³ ha^–1, 46.4 t ha^–1). The conventional thinning treatment had the lowest values for stem volume and biomass (39.6 m³ ha^–1, 23.7 t ha^–1), both of which were significantly lower than the respective control values. The removal, in percent of total biomass and number of stems, resulting from conventional thinning (biomass 61%; stems 82%) was significantly higher than that of BCT-treatments. Additionally, conventional thinning harvested significantly more deciduous species than any of the BCT-treatments, in terms of both biomass and stem number (Table 1, Figs. 2a–b).

Fig. 2a. Number of stems ha^–1 for conifer and deciduous species for diameter classes 0–20 cm, mean values for all sites. BCT 1-serie, mean and dominant tree height 6 and 11 m; treatment BCT_{2 m}, BCT_{1 m} and BCT_sel.

Fig. 2b. Number of stems ha^–1 for conifer and deciduous species for diameter classes 0–20 cm, mean values for all sites. BCT 1-serie, mean and dominant tree height 6 and 11 m; control and conventional treatments.

The only significant difference in the total number of stems with a DBH > 80 mm was between the control (1524 stems ha^–1) and conventional thinning (622 stems ha^–1) plots (Table 1). Also for basal area the highest value was found in the control (13.4 m² ha^–1). Significant higher only compared to the conventional thinning (5.8 m² ha^–1). The same pattern was also found for total biomass, 44.3 and 18.9 respectively. No significant difference was found for the basal area weighted diameter at breast height (1.3 m) (Dg), mean height or stem volume (Table 1).

3.2 Stand structure parameters of BCT 2-series treatments

The measurements from stands with the same dominant height were compared before and after the silvicultural treatments. Thus, in the three stands with a dominant height of 9 m the PCT and BCT treatments were compared, whereas in the three stands with a dominant height of 10 m the BCT_low, BCT_high and Conv treatments were compared. As in the BCT 1-series, there were no significant differences between plots in any of the measured parameters before the treatments began (Table 2). However, following the treatments in stands with a dominant height of 9 m there were a significantly higher number of deciduous stems in BCT treatment plots (1267 stems ha^–1, 31%) compared to PCT treatment plots (274 stems ha^–1, 10%). Furthermore, the PCT plots were characterized by a significantly larger mean height (7.2 m) than BCT plots (5.2 m). The PCT treatment also resulted in a higher removal, both overall and of deciduous trees, than the BCT treatment (Table 2, Figs. 2c–d).

Fig. 2c. Number of stems ha^–1 for conifer and deciduous species for diameter classes 0–17 cm, mean values for all sites. BCT 2-series, mean and dominant tree height 4 and 9 m, treatment BCT and PCT.

Fig. 2d. Number of stems ha^–1 for conifer and deciduous species for diameter classes 0–19 cm, mean values for all sites. BCT 2-series, mean and dominant tree height 5 and 10 m treatments BCT_low, BCT_high and conventional.

Following the thinning operations in stands with a dominant height of 10 m, conventional thinning resulted in the lowest stem number, but this value was not significantly less than either of the two BCT-treatments. Furthermore, there were no significant differences between treatments in dominant height, mean height, basal area based on DBH, or amount of large trees left following treatment. The BCT_low treatment resulted in the highest values for stem volume and total DW of biomass, but these values were not significantly higher than those of conventional thinning, which had the lowest values (Table 2). The conventional thinning also harvested more deciduous trees than the BCT-treatments, but the difference was not significant (Table 2).

For the largest trees (i.e. DBH > 80 mm), no significant differences between treatments were found in stem number, basal area, stem volume, total biomass, mean height or the Dg for both tree heights, 4/9 as well as for 5/10 (Table 2).

Treatments where boom-corridor thinning have been applied clearly show higher degree of tree size and species heterogeneity compared to selective treatments (Figs. 2c–d).

Simpson index for biodiversity resulted in higher values for BCT-treatments compared to selective thinning and PCT-treatments, with the same patterns for both series. For BCT 1, BCT_1 m (0.47), BCT_sel (0.46), BCT_{2 m} (0.46) and the control (0.45) there were significant higher values compared to the conventional thinning treatment (0.21). For BCT 2, neither were significant differences found between treatments in the stands with lower tree height (mean/dominant: 4/9 m); PCT (0.12) and BCT (0.42), nor in the stands with higher trees (5/10 m); BCT_low (0.26), BCT_high (0.17) and Conventional (0.12).

4 Discussion

The results show that overall, the BCT-treatments resulted in a more heterogeneous stand structure than other thinning operations, such as conventional thinning or PCT, shown by more trees per ha, somewhat smaller if mean tree height is considered, as well as more deciduous trees per plot. This heterogenic stand structure could create habitats for certain organism groups and thus positively affect ecosystem values (McElhinny et al. 2005). For each treatment the number of stems with a DBH > 80 mm was also counted to determine the amount of possible future crop trees, as previous studies have shown that trees in young stands with a DBH ≤ 80 mm have an increased risk of damage, such as broken stems or windthrow, especially after the thinning of dense stands (Päätalo et al. 1999; Abetz and Klädtke 2002). Results indicates that BCT could be a cost-effective thinning technique in young dense stands without risking that the number of larger trees in the stand decrease.

Furthermore, all BCT treatments provided similar results, as neither the geometric strictness of boom corridor layout nor the choice of high- or low-thinning approaches showed significant differences in stand structure parameters. Thus, BCT operations could provide a highly efficient and cost-effective technique for first thinning that can handle a variety of tree sizes (Bergström 2009) and simultaneously potentially increase local biodiversity through stand heterogeneity. Conventional thinning methods, on the other hand, attain higher cost efficiency when they are used in stands that have been pre-cleared or earlier subject to PCT. It remains to be seen whether the combination of conventional thinning and BCT techniques can increase biodiversity and ecosystem values.

Early thinning of dense stands through BCT results in harvested biomass that contains a higher share of branches and needles compared to pre-cleared stands. When this biomass is used as solid biofuel there is a small difference in quality between thinning methods, and thus, the values per DW t between thinning approaches should be similar. However, if the biomass is used for the extraction of chemicals, then the biomass resulting from BCT has a higher proportion of extraction-rich fractions, such as bark and needles (Backlund et al. 2014). These extraction-rich fractions mean that more valuable nutrients are removed from the stand. Technologies that compress and semi-defoliate biomass at the stand before extraction allow forest managers to control which fractions are extracted from a stand (Bergström et al. 2010).

All of the thinning treatments in both series were performed by a mechanized thinning harvester. Each series had one machine operator, who performed all the treatments included in that series. All of the cutting and bunching work was performed in a controlled manner to ensure that trees remaining after thinning not were damaged. In BCT 1-series plots the cutting of strip-road trees was strictly controlled, whereas in BCT 2-series plots the operator choose which trees to cut himself, resulting in a marginally higher strip-road area cut and thinning intensity than had been planned. Still there was no treatment where the area differed from the target width.

The next inventory will be performed in five years from set-up, and we expect that all the treatments will have similar production of future crop trees because the number of trees with a DBH > 80 mm was about the same for all treatments. Additionally, we expect that very few of the larger trees will be damaged due to snow and wind. Moreover, we expect that the smaller trees will show increased mortality due to high competition between trees in dense stands.

5 Conclusions

After the first biomass thinning, BCT and selective thinning approaches result in similar stand structures based on the amount and average sizes of remaining possible future crop trees. However, BCT also maintains diverse tree sizes, deciduous species, and leaves more trees per ha. Hence, BCT should be a cost-effective option for thinning that can increase vertical complexity in stands, and as a result possibly increase biodiversity and ecosystem service values.

In our study we only consider tree species for biodiversity analysis and in the studied forest the number of species are low and predominated (in terms of volume) by conifers (pine and spruce). The index shows that the diversity of tree species for all trial series is somewhat higher for boom-corridor treatments than selective ones. However, the differences are significant only for BCT 1. Thus, schematic thinning renders keeping the tree diversity at the same level as for the untreated stands.

Acknowledgements

We would like to thank the staff of the Unit for Field-based Forest Research at the Swedish University of Agricultural Sciences (part of SITES, Swedish Infrastructure for Ecosystem Science) for providing the data and conducting the field work. This study was financed by Formas, Norrskogs Forskningsstiftelse, Stiftelsen Skogssällskapet and the forest company Sveaskog. We are grateful to SEES-editing Ltd. for professional linguistic revision. Parts of this study has been conducted in the EFFORTE project which has received funding from the Bio Based Industries Joint Undertaking under the European Union’s Horizon 2020 research and innovation programme under grant agreement No 720712.v

References

Abetz P., Klädtke J. (2002). The target tree management system. Forstwissenschaftliches Centralblatt 121: 73–82. [In English with summary in German].

Andersson S.-O. (1954). Funktioner och tabeller för kubering av småträd. [Functions and tables for measuring the volume]. Meddelande från Statens Skogsforskningsinstitut 44(12). [In Swedish].

Backlund I., Arshadi M., Hunt A.J., McElroy R., Attard T.M., Bergsten U. (2014). Extractive profiles of different lodgepole pine (Pinus contorta) fractions grown under a direct seeding-based silvicultural regime. Industrial Crops and Products 58: 220–229. https://doi.org/10.1016/j.indcrop.2014.04.027.

Bengtsson J., Nilsson S.G., Franc A., Menozzi P. (2000). Biodiversity, disturbances, ecosystem function and management of European forests. Forest Ecology and Management 132(1):39–50. https://doi.org/10.1016/S0378-1127(00)00378-9.

Bergström D. (2009). Techniques and systems for boom-corridor thinning in young dense forests. Doctoral thesis. Acta Universitatis Agriculturae Sueciae 87. ISBN 978-91-576-7343-0.

Bergström D., Di Fulvio F. (2014a). Comparison of the cost and energy efficiencies of present and future biomass supply systems for young dense forests. Scandinavian Journal of Forest Research 29(8): 793–812. https://doi.org/10.1080/02827581.2014.976590.

Bergström D., Di Fulvio F. (2014b). Studies on the use of a novel prototype harvester head in early fuel wood thinnings. International Journal of Forest Engineering 25(2): 156–170. https://doi.org/10.1080/14942119.2014.945697.

Bergström D, Bergsten U, Nordfjell T, Lundmark T. (2007). Simulation of geometric thinning systems and their time requirements for young forests. Silva Fennica 41(1): 137–147. https://doi.org/10.14214/sf.311.

Bergström D, Nordfjell T, Bergsten U. (2010). Compressing processing and load compression of young Scots pine and birch trees in thinnings for bioenergy. International Journal Forest Engineering 21: 31–39.

Brandel G. (1990). Volume functions for individual trees: Scots pine (Pinus sylvestris), Norway spruce (Picea abies) and birch (Betula pendula and Betula pubescens). Doctoral thesis No. 26. Swedish University of Agricultural Sciences, Garpenberg. ISBN 91-576-4030-0.

Fernandez-Lacruz R., Di Fulvio F., Athanassiadis D., Bergström D., Nordfjell T. (2015). Distribution, characteristics and potential of biomass-dense thinning forests in Sweden. Silva Fennica 49(5) article 1377. https://doi.org/10.14214/sf.1377.

Führer E. (2000). Forest functions, ecosystem stability and management. Forest Ecology and Management 132(1): 29–38. https://doi.org/10.1016/S0378-1127(00)00377-7.

Grebner D.L., Perez-Verdin G., Sun C., Munn I.A., Schulz E.B., Matney T.G. (2009). Wood biomass feedstock availability, production costs and implications for bioenergy conversion in Mississippi. In: Solomon B.D., Luzadis V.A. (eds.). Renewable energy from forest resources in United States. Routledge, Oxford, UK. (Routledge Explorations in Environmental Economics 12). p. 261–280. ISBN 978-0-415-77600-4.

Karlsson K., Mossberg M., Ulvcrona T. (2012). Fältdatasystem för skogliga fältförsök. Unit for field-based Forest Research, Swedish University of Agricultural Sciences, Umeå. Report No. 5. 29 p. [In Swedish, summary in English].

Karlsson L., Nyström K., Bergström D., Bergsten U. (2015). Development of Scots pine stands after first biomass thinning with implications on management profitability over rotation. Scandinavian Journal of Forest Research 30(5): 416–428. https://doi.org/10.1080/02827581.2015.1023351.

McElhinny C., Gibbons P., Brack C., Bauhus J. (2005). Forest and woodland stand structural complexity: its definition and measurement. Forest Ecology and Management 218(1–3): 1–24. https://doi.org/10.1016/j.foreco.2005.08.034.

Messier C., Puettman K.J., Coates K.D. (2014). Managing forests as complex adaptive systems: building resilience to the challenge of global change. The Earth Scan Forest Library, Oxford, UK. 368 p.

Minitab (2010). Minitab 16 Statistical Software. [Computer software]. Minitab Inc., State College, PA. http://www.minitab.com.

Päätalo M.-L., Peltola H., Kellomäki S. (1999). Modelling the risk of snow damage to forests under short-term snow loading. Forest Ecology and Management 116(1): 51–70. https://doi.org/10.1016/S0378-1127(98)00446-0.

Puettman K.J., Coates K.D., Messier C. (2009). A critique of silviculture: managing for complexity. Island Press, Washington D.C. 189 p.

Repola J., Ulvcrona K. (2014). Modelling biomass of young and dense Scots pine (Pinus sylvestris L.) dominated mixed forests in northern Sweden. Silva Fennica 48(5) article 1190. https://doi.org/10.14214/sf.1190.

Saastamoinen O., Matero J., Horne P., Knivilä M., Haltia E., Mannerkoski H., Vaara M. (2014). Classification of boreal forest ecosystem goods and services in Finland. Publications of the University of Eastern Finland, Reports and Studies in Forestry and Natural Sciences 11. 196 p.

Sabin T.E., Stafford S.G. (1990). Assessing the need for transformation of response variables. College of Forestry, Forest Research Lab (1990:20), Oregon State University.

Sängstuvall L, Bergström D, Lämås T, Nordfjell T. (2011). Simulation of harvester productivity in selective and boom corridor thinning of young forests. Scandinavian Journal Forest Research 27(1): 56–73. https://doi.org/10.1080/02827581.2011.628335.

Simpson E.H. (1949). Measurement of diversity. Nature 163: 688. https://doi.org/10.1038/163688a0.

Sullivan T.P., Sullivan D.S., Lindgren P.M.F., Ransome D., Bull J.G., Ristea C. (2011). Bioenergy or biodiversity? Wood debris structures and maintance of red-backed voles on clearcuts. Biomass and Bioenergy 35(10): 4390–4398. https://doi.org/10.1016/j.biombioe.2011.08.013.

Swedish statistical yearbook of forestry (2014). Official Statistics of Sweden. Swedish Forest Agency, Jönköping. ISBN 978-91-87535-05-5.

Tamhane A.C., Dunlop D.D. (2000). Statistics and data analysis. Prentice-Hall Inc., Upper Saddle River, NJ. ISBN 0-13-744426-5.

Ulvcrona K.A. (2011). Effects of Silvicultural treatments in young Scots pine-dominated stands on the potential for early biofuel harvest. Doctoral thesis no. 2011:79. Faculty of Forest Sciences, Swedish University of Agricultural Sciences, Umeå. ISBN 978-91-576-7623-8.

Valinger E., Lundqvist L., Brandel G. (1994). Wind and snow Damage in a thinning and fertilization experiment in Pinus sylvestris. Scandinavian Journal of Forest Research 9(1–4): 129–134. https://doi.org/10.1080/02827589409382822.

Total of 30 references.