Forestry-2004-Mäkinen-349-64.pdf

Introduction

Control of stand density by thinning has been themajor tool in regulating tree growth and improv-ing timber quality. While thinning from belowmay increase the merchantable volume of astand, usually it does not increase the totalvolume increment per unit area (e.g. Assmann,1954; Carbonnier, 1967; Hasenauer et al., 1997;Zeide, 2001). Several studies have shown thatvolume increment of many tree species does notdecline with decreasing stand density within awide range of stand density (e.g. Hamilton, 1981;Horne et al., 1986). This indicates that thinning

from below redistributes the increment fromsmaller trees to larger ones, and a smaller numberof trees is able to produce the same volume incre-ment per unit area. In order to increase the mer-chantable volume, or not to decrease the totalvolume increment, the intensity, timing and fre-quency of thinnings have to be defined.

In Finland, Norway spruce (Picea abies (L.)Karst.) is economically and ecologically one ofthe most important tree species. The first growthand yield tables for Norway spruce were pub-lished by Blomqvist in 1872 (see also Heikkilä,1914), but they were not widely applied. System-atic growth and yield studies began in the 1910s

Thinning intensity and growth ofNorway spruce stands in FinlandHARRI MÄKINEN* AND ANTTI ISOMÄKI

Finnish Forest Research Institute, PO Box 18, FIN-01301 Vantaa, Finland*Corresponding author. E-mail: [email protected]

Summary

The effects of thinning intensity on the growth and yield of Norway spruce (Picea abies (L.) Karst.)were investigated in long-term thinning experiments on mineral soil sites in southern Finland. Themeasurement period was on average 27 years, and the intensity of the thinnings from below rangedfrom heavy thinning (45 per cent of basal area removed) to no thinning. Total stem volumeincrement and merchantable volume produced per hectare were the highest on the unthinned plots,but light and moderate thinning (<30 per cent removed) produced almost as much. Heavy thinning(>30 per cent removed) reduced the volume increment by about 10 per cent. However, a part of thetotal production of unthinned plots was lost through natural mortality. On the thinned plots, naturalmortality was considerably lower compared with the unthinned plots. The average diameterincrement of all the trees, as well as the diameter of the largest trees, clearly increased withincreasing thinning intensity. In contrast, dominant height increment was not affected by thinning.The stand age at the time of establishment of the experiments had no major effect on the growthreactions after thinning. Thus, heavy thinning results in earlier thinning yields and a higherproportion of larger-sized stems at the expense of a somewhat lower total yield.

© Institute of Chartered Foresters, 2004 Forestry, Vol. 77, No. 4, 2004

12 cph037 MATHS 9/9/04 9:58 am Page 349

by guest on August 29, 2015

http://forestry.oxfordjournals.org/D

ownloaded from

http://forestry.oxfordjournals.org/

and the first results for Norway spruce were pub-lished by Ilvessalo (1920). They described thestructure and development of fully stocked,naturally regenerated pure Norway sprucestands. A decade later, Cajander (1933) studiedthe development of planted, unthinned Norwayspruce stands.

After World War II, selective thinnings, i.e.removing only the largest trees, were prohibitedand thinnings from below were promoted (Anon.,1948; Kalela, 1948). At the same time, a researchproject was initiated to study the effects of thin-nings on the structure and development ofnaturally regenerated Norway spruce stands(Vuokila, 1956; Kallio, 1957). However, theintensity of the thinnings was low and theremoved trees were heavily suppressed or dying.In the early 1960s, the thinning intensity inpractical forestry increased mainly due to themechanization of harvesting operations. Becausethe short- and long-term growth effects of heavythinnings were not known, a new research projectwas established and growth and yield tables, aswell as thinning guidelines, were prepared forplanted Norway spruce stands (Vuokila andVäliaho, 1980). During the last decades, growthmodels have been developed for forest manage-ment planning (e.g. Hynynen et al., 2002).

All the Finnish studies cited above were basedon temporary sample plots (apart from the recentgrowth models that were based on remeasuredinventory growth plots) because there was anurgent demand for growth and yield tables, butno data available. However, permanent sampleplots provide more reliable results on long-termgrowth and stand dynamics. In Finland, the firstthinning experiments were established in the1920s and 1930s (Ilvessalo, 1932). Some of theresults of these experiments have been published(Vuokila, 1960, 1977), but the experiments have,in the main, not fulfilled their expectations due tonatural damage, changes in measurementmethods, subjective classification of many par-ameters, experimental design, etc. (Vuokila,1965). In addition, the thinnings applied havebecome out-of-date.

In the 1960s, new thinning experiments onNorway spruce, Scots pine and silver birch (theso-called HARKAS series) involving more inten-sive treatments and a better experimental designwere established (Vuokila, 1983). In the 1970s,

the establishment of experiments was continuedand the first results on Norway spruce were pub-lished on the basis of five experiments (Vuokila,1975). In later publications, Vuokila (1980,1985) described the further development of thesame experiments as in the first paper (seeTable 1; numbers Nyn1–Nyn5). Mielikäinen(1978) demonstrated the effects of thinning onthe diameter increment of Norway spruce duringa 7-year period in one experiment (Table 1;Nyn1). In addition, the results of the experimentshave been presented in seminars and excursions(e.g. Isomäki, 1993, 1995). However, no system-atic presentation of the whole material exists.

In the other Fennoscandian countries, recentresults from Norway (Braastad and Tveite, 2000,2001) and Sweden (Eriksson and Karlsson, 1997)have shown that the volume increment inNorway spruce stands stays constant or decreasesonly slightly with increasing thinning intensityover a wide range of stand density. Earlierfindings from central Europe by Assmann (1954),Hamilton (1976), Schober (1979, 1980) andKramer and Jünemann (1985) have also shownthat the volume increment of Norway sprucedoes not decrease proportionally with decreasingstand density. However, the benefits anddisadvantages of thinnings are still the subject ofheated debate (Braastad, 2001; Elfving et al.,2001).

The objective of this study was to relatethinning intensity with diameter, height andvolume increment on the basis of permanentlong-term experiments with thinnings frombelow in planted Norway spruce stands. Thisstudy is a sequel to the reports of Vuokila (1975,1980, 1985), which were based on the fiveexperiments also used in this study. In this study,we expanded the database and used all thethinning experiments of the HARKAS series,established by the Finnish Forest Research Insti-tute in Norway spruce stands (21 experiments).Only two stands which were fertilized before theestablishment of the experiments were excludedfrom the data set. Many experiments havealready been studied for over 30 years (Table 1)and they approach maturity or are alreadymature. This gives us an opportunity to investi-gate total stem volume production and thinningremoval, as well as stand structure, during thewhole stand rotation.

350 FORESTRY

12 cph037 MATHS 9/9/04 9:58 am Page 350



ownloaded from


Materials and methods

The experiments

The material was collected from 21 thinningexperiments in southern Finland (Figure 1). Theexperiments were established by the FinnishForest Research Institute in the 1960s and early1970s, apart from three experiments which wereestablished in the late 1970s and early 1980s(Table 1). The stands were even-aged, pure oralmost pure Norway spruce (Picea abies (L.)Karst.) stands located on mineral soil, and wereestablished by planting with seed of local origin.The sites were classified as the Oxalis–Myrtillusor Myrtillus forest site type (Cajander, 1909),which corresponds to highly fertile or fertile sitestypical for Norway spruce. The site index (H100,dominant height at age 100 years in metres)ranged from 29 m to 36 m. Site indices werecalculated using the functions of Vuokila andVäliaho (1980). Due to changes in land use (e.g.the construction of houses and roads), some ofthe experiments had to be terminated earlier thanplanned (Table 1).

The principal aim of the experiments was toinvestigate the effects of thinning intensity ongrowth and yield of Norway spruce stands at

different stages of stand development. Thematerial consisted of two separate sets: (1) thin-ning experiments based on stem number (13experiments); and (2) thinning experiments basedon stand basal area (eight experiments). Theexperiments based on stem number wereoriginally planned to include three different treat-ments with varying thinning intensity and anumber of repeated thinnings from below, as wellas an unthinned control plot. The final stemnumber per hectare was planned to be the samein all the treatments, and it was to be reachedusing two heavy, three moderate or five light thin-nings from below, based on a fixed number ofstems remaining after the thinnings. However, itwas soon noticed that, in many cases, the treat-ments resulted in only small differences amongthe plots and the research plan was thereforemodified. The intensity of the later thinnings wasincreased in order to maintain the differences instand density between the plots. In the experi-ments based on basal area, a constant basal arearatio compared with the unthinned control plotwas maintained on the thinned plots, i.e. the plotswere thinned to basal areas of 90, 75 and 60 percent compared with the control plot.

Square or rectangular plots surrounded by a5-m-wide buffer zone were established in each

THINNING INTENSITY AND GROWTH OF NORWAY SPRUCE 351

Figure 1. Location of the experiments. For explanation of the onset stages, see Materials and methods.

12 cph037 MATHS 9/9/04 9:58 am 51



ownloaded from


stand. The average area of the plots was1000–1600 m2 (range 500–2500 m2). Most ofthe experiments had four, eight or 12 plots, i.e.the four treatments were replicated in a random-ized block design in the experiments with eight or12 plots (Table 1). In the experiments with five,six or 10 plots, one or two of the treatments wereadditionally replicated within the experiments orblocks. In experiments Vh002 and Vh098,having only three plots, the light thinning inten-sity was missing (see below).

Measurements and statistical analysis

Following establishment, the experiments weremeasured between two and seven times (Table 1).The measurement period was on average 27years. At the time of the last measurement, stand

age ranged from 39 to 85 years. Tree species, stemdiameter at breast height (d1.3), and possibledamage (fallen, broken, decaying, needle loss,etc.), as well as its cause (wind, snow, com-petition, insect or fungus species) and severity(temporary, causing permanent defects, lethal),were recorded for each tree on the plot. In theselection of sample trees, the probability for a treeto be selected was proportional to its diameter,but the sample trees were randomly located onthe sample plots. Height, height to crown base,and stem diameter at 6 m height (d6.0) weremeasured on each sample tree (on average, 54 perplot). The crown base was defined as the lowestwhorl with at least one living branch that wasseparated from the other living whorls above it byno more than one dead whorl.

Stand characteristics for individual plots were

352 FORESTRY

Table 1: Characteristics of the experiments at establishment

YearAt establishment

No. of of Establishment Last Treatment No. of Hdom‡ No. of

No. plots H100* planting of plots measured onset† measurements (m) stems ha–1§

Vh001 6 34.4 1918 1970 1998 3 4 24.3 1072Vh002 3 31.6 1925 1970 1998 2 5 16.9 1404Vh005 10 28.9 1931 1971 1994 1 5 13.5 2212Vh009 4 31.1 1931 1973 1998 2 5 17.0 1856Vh011 8 30.2 1914 1970 1999 3 5 22.4 1114Vh012 10 32.4 1916 1970 1998 3 6 23.4 1042Vh013 8 32.9 1932 1970 1998 2 6 18.1 2013Vh014 8 32.7 1918 1971 1998 3 5 23.1 935Vh017 4 32.9 1936 1971 1985 2 3 16.3 2932Vh040 5 30.0 1934 1973 1992 2 4 18.0 1526Vh048 8 31.0 1934 1977 2001 2 5 16.9 2137Vh097 12 34.0 1955 1981 1994 1 4 12.5 2973Vh098 3 36.4 1930 1979 1994 3 5 24.1 1626Ha001 5 30.2 1938 1965 1998 1 6 9.8 3335Pu041 4 33.0 1934 1964 1999 1 6 12.1 1689Pu042 4 33.0 1924 1964 1999 2 6 17.4 1168Nyn1 12 30.0 1922 1961 1998 1 8 14.2 2055Nyn2 4 34.5 1931 1962 1998 1 8 14.7 3247Nyn3 8 34.7 1926 1962 1999 1 8 16.0 2394Nyn4 4 33.0 1925 1962 1988 1 6 15.6 2295Nyn5 4 33.0 1930 1962 1992 1 7 14.3 3402

* H100 site index, dominant height at age 100 years (m).† 1, early; 2, medium; and 3, late onset of the treatment based on the Hdom/H100 ratio, see Materials andmethods.‡ Dominant height based on 100 largest trees ha–1.§ Before the onset of the treatments.

12 cph037 MATHS 9/9/04 9:58 am Page 352



ownloaded from


calculated using the KPL software developed atthe Finnish Forest Research Institute (Heinonen,1994). Stem volumes of the sample trees werecalculated using volume functions based on themeasured stem diameters (d1.3, d6.0) and treeheight (Laasasenaho, 1982). The heights of theother trees were predicted using Näslund’s heightcurve (Näslund, 1937) that was fitted for each plotwith the help of the tree heights measured on thesample trees. The volume of the other trees wascalculated using smoothing functions fitted to thesample tree data. The minimum length applied forpulpwood boles was 3.0 m and the minimum topdiameter was 8.0 cm over bark. The stem woodbelow this size was considered as wastewood. Theminimum length applied for timber logs was3.7 m, and the minimum top diameter over barkwas 19.5 cm. The minimum top diameter pro-gressively decreased with increasing log length,being 16.0 cm when the log length was over 4.3 m.

Annual increments were calculated as thedifference between successive measurementsdivided by the number of years between themeasurements. The periodic growth betweenthe measurements was corrected to correspond tothe average climatic conditions using annualradial growth indices. The indices for Norwayspruce in southern Finland were based on theincrement cores measured in connection with theNational Forest Inventory. The indices publishedby Henttonen (1986) were used for the period1961–1979, and the indices by Henttonen(unpublished data) for the period 1980–1999. Asno indices were available for the years 2000 and2001, they were assumed to be average years.The method for calculating the growth indices isdescribed in Henttonen (1986, 1990).

Because the number and intensity of the thin-nings varied among the plots and experiments,the plots could not be classified into differentgroups in accordance with the original researchplan. Therefore, the plots were grouped on thebasis of the average basal area on the plot duringthe whole measurement period BA_ i comparedwith that on the control plots of each experiment.The basal areas of the different measurementperiods were weighted by the period length, i.e.

* /

BA T

T BAr BAt 2i i i

i

n 1

1

1

tot=

+ +

=

-

!_ i

(1)

where BAr is the basal area of the remainingliving trees at the beginning of each measurementperiod, BAt the total basal area at the end of eachmeasurement period including dead trees andtrees to be thinned, Ti the length of each measure-ment period (years), Ttot the length of the totalmeasurement period (ΣTi), and n the number ofindividual measurement periods. The plots wereclassified as follows: (U) unthinned (average basalarea ≥95 per cent of that on the control plots) –note that some plots with light thinning intensitywere classified as unthinned because their basalarea exceeded 95 per cent; (L) light thinning(85–94 per cent); (M) moderate thinning (70–84 per cent); and (H) heavy thinning (≤70 percent). In class H, the minimum and averagerelative basal areas were 55 per cent and 63 percent, respectively.

Based on the ratio between dominant height(based on 100 by diameter of largest trees ha–1)at the time of establishment and the site index(Hdom/H100), the experiments were furtherdivided into three thinning onset stages asfollows: early onset (Hdom/H100 ≤ 0.49), mediumonset (0.50 ≤ Hdom/H100 ≤ 0.60) and late onset(Hdom/H100 ≥ 0.61). The mean Hdom/H100 ratiowas 0.42, 0.54 and 0.71 for the early, mediumand late onset stages, respectively. At the earlyonset stage, only light pre-commercial thinningswere carried out before establishment of theexperiments. The stands of medium onset stagehad already passed the first commercial thinningphase, but the thinnings carried out had beenlight. Accordingly, only light thinnings werecarried out in the stands of late onset stage beforeestablishment.

Statistical significance of the differences amongthe treatments was analysed using covarianceanalysis including random experiment and blockeffects. The model used to test the treatmenteffects was:

Y = µ + δD + βXsbp + us + usb + usbp (2)

where Ysbp is a dependent variable, µ is theoverall mean, δD the effect of stand density class,β the regression coefficient, and us, usb and usbpthe random effects for stand s, block b and plotp. The initial differences among the plots wereremoved by applying a continuous covariate(Xsbp) measured before the onset of the treatment,i.e. mean stem diameter, dominant height (100 by


12 cph037 MATHS 9/9/04 9:58 am Page 353



ownloaded from


354 FORESTRY

Figure 2. Mean annual diameter increment of all the trees (parts a–c) and the 400 by diameter largest treesha–1 (parts d–f) on plots of different thinning intensity (U, unthinned; L, light; M, moderate; H, heavythinning) and treatment onset stage (early, parts a and d; medium, parts b and e; late onset, parts c and f).Treatments marked with the same letter are not significantly different (P ≥ 0.05). Stem diameter before theonset of the treatment was used as the covariate in equation 2.

12 cph037 MATHS 9/9/04 9:58 am Page 354



ownloaded from


diameter of largest trees ha–1) or stand volume.The values shown in the figures and tables areadjusted by setting covariate effects to their meanvalues.

The pairwise comparisons were performed bycomputing generalized least-square means of thetreatment effects. Owing to the unbalanceddesign, adjusted P-values for the multiple com-parison were computed from the simulated distri-bution of a multivariate random vector. In thefigures and tables, the treatments marked withthe same letter are not significantly different (P ≥0.05). Restricted maximum likelihood (REML)estimation in the MIXED procedure of SAS (SASInstitute, 1999) was used in the analysis.

Results

The arithmetic mean annual diameter increment ofall the trees on a plot increased with decreasingstand density (Figure 2a–c). The differencesbetween the thinning intensities were statisticallysignificant at all treatment onset stages. Becausethe differences of the mean increment may becaused by different numbers of stems, meandiameter increment was also calculated for the400 (by diameter) largest trees ha–1. The differ-ences between the treatments in the diameter incre-ment of the largest trees were, however, similar tothose for all the trees (Figure 2d–f). On the otherhand, thinning intensity had no clear effect onannual dominant height increment (Figure 3).

On plots with light and moderate thinningintensity, the annual volume increment per hectarewas about the same as on the unthinned plots atall treatment onset stages (Figure 4). Comparedwith the unthinned plots, a heavy thinning inten-sity decreased the volume increment per hectareduring the whole measurement period by 8 percent at the early onset stage. At the medium onsetstage, a heavy thinning intensity decreased thevolume increment even more (11 per cent), but thedifferences between the treatments were notstatistically significant. At the late onset stage, thevolume increments among the individual plotswere very variable and, therefore, the differencesbetween the thinning intensities were not statisti-cally significant.

In addition to the analysis according to fixedtreatment onset stages, the whole material was


Figure 3. Mean annual dominant height incrementon plots of different thinning intensity and treat-ment onset stage. Dominant height before the onsetof the treatment was used as the covariate inequation 2. For explanation of the symbols, seeFigure 2.

12 cph037 MATHS 9/9/04 9:59 am Page 355



ownloaded from


also combined, i.e. the volume incrementsbetween the successive measurements wereanalysed according to the dominant height in themiddle of each measurement period (Figure 5).The unthinned plots and heavy thinning intensitydiffered from each other at all dominant heightstages. The annual volume increment of the lightand moderate thinning intensity lay betweenthose of the unthinned and heavy thinning plots,excluding the lowest and highest dominant heightstages. However, the mutual order of the lightand moderate intensities was variable at differentdominant heights and it was not possible to dis-tinguish between them.

Volume increments and basal areas on thethinned plots during the whole measurementperiod were also examined in relation to themean volume increment and basal area of theunthinned plots in each experiment (Figure 6).Because the experiments had several unthinnedplots, and their basal areas were also related tothe mean value, some relative basal areasexceeded 100 per cent. Figure 6 clearly showsthat the variation of relative volume incrementwas high among individual plots, but that themean relative volume increment decreased onlyslightly with decreasing mean relative basal area.

As expected, the current stem volume perhectare decreased with increasing thinningintensity at all treatment onset stages (Table 2).However, total yield, calculated by summing upthe current volumes of living trees and thevolumes of thinned and dead trees, did not differsignificantly between the unthinned plot and thelight and moderate thinning intensities (Table 2).The heavy thinning intensity decreased the totalyield by 7 per cent, 4 per cent and 7 per centcompared with the unthinned plots at the early,medium and late onset stages, respectively, butthe differences at the medium and late onsetstages were not statistically significant. On theunthinned plots, some of the trees had also beenremoved in connection with the thinnings(Table 2), because some of the thinned plots wereplaced in the ‘unthinned’ group if their basal areadid not differ from the control plots >5 per cent.In addition, light sanitary thinnings were carriedout in some experiments, i.e. dying trees wereremoved at the request of the landowner.

The current log volume was highest with theunthinned or light thinning intensity and lowest

356 FORESTRY

Figure 4. Mean annual volume increment on plotsof different thinning intensity and treatment onsetstage. Total volume before the onset of the treat-ment was used as the covariate in equation 2. Forexplanation of the symbols, see Figure 2.

12 cph037 MATHS 9/9/04 9:59 am Page 356



ownloaded from



�

�

�

�

Dominant height (m)

Volu

me

incr

emen

t (m

2 h

a–1

a–1 )

15 17 19 21 23 25 27 29 31

16

13

10

7

Figure 5. Mean annual volume increment between the successive measurements plotted against meandominant height during each measurement period on plots of different thinning intensity. For explanationof the symbols, see Figure 2.

Figure 6. Volume increment on plot i during the whole measurement period in relation to the mean volumeincrement on the unthinned plots of the same experiment (ivi/ivu) plotted against the relative average basalarea during the whole measurement period (related to that of the unthinned plots, BAi/BAu, see equation 1).The thick continuous line is a non-linear curve fitted to the data [ivi/ivu = (BAi/BAu)2/{(0.08 + 0.91BAi/BAu)2}, R2 = 0.16].

12 cph037 MATHS 9/9/04 9:59 am Page 357



ownloaded from


on the heavy thinned plots, even though thedifferences at the medium onset stage were notstatistically significant (Table 2). Accordingly, thecurrent volume of pulpwood and wastewooddecreased with increasing thinning intensity at alltreatment onset stages. In contrast, total volumeof merchantable-sized timber, i.e. the currentvolume of logs and pulpwood summed up withthat removed in thinnings, was about the same inall thinning intensities.

On the unthinned plots, volume of dead treeswas higher than that on the thinned plots,especially in the later phases of the experiments(Figure 7). On the thinned plots, the differencesin dead wood volume were small among thethinning intensities (Figure 7 and Table 2). At thelate onset stage, dead wood volume was,however, about the same on the unthinned andlight thinned plots. The initiation of the treat-ment did not result in a dramatic change in mor-tality. Instead, the mortality rate increased

slightly in the later phases of the measurementperiod. On the other hand, the volume of log-sized dead trees was relatively similar in allthinning treatments, i.e. the higher dead woodvolume on the unthinned plots was mainlycaused by the mortality of small-sized trees(Table 2).

Discussion

Total volume production per unit area washighest on the unthinned or lightly thinned plots,irrespective of the treatment onset stage.However, light and moderate thinning hadalmost no effect on volume increment, or totalvolume produced during the long measurementperiod. Only heavy thinning resulted in a clearreduction in volume increment. Even though thebasal area of the heaviest treatments was, onaverage, 55–70 per cent below that on the

358 FORESTRY

Table 2: Total yield and its structure on plots of different thinning intensity (U, unthinned; L, light; M,moderate; H, heavy thinning) and treatment onset stages

Early onset Medium onset Late onset

U L M H U L M H U L M H

Total volume (m3 ha–1)Log 349a 384a 374a 377a 365a 408a 396a 374a 560a 556a 552a 541a

Pulp 206a 178b 172b 140c 182a 161a 145a 150a 109a 99a 101a 84a

Wastewood 29a 27a 29a 28a 20a 20a 17a 18a 10a 9a 9a 7a

Sum 584a 589b 575b 545b 567a 589a 558a 542a 679a 664a 662a 632a

Current volume (m3 ha–1)Log 331ab 370a 339ab 314b 349a 354a 335a 286a 526a 480ab 442b 419b

Pulp 165a 119b 94b 45c 145a 103a 50b 46b 80a 55ab 42b 25b

Wastewood 15a 9b 8b 4c 12a 8ab 5b 4b 7a 5ab 4ab 3b

Sum 511a 498a 441b 363c 506a 465ab 390bc 336c 613a 540ab 488b 447b

Thinned volume (m3 ha–1)Log 7a 7a 25a 56b 2a 38ab 44ab 65b 8a 46ab 90b 108b

Pulp 13a 51b 72bc 93c 4a 44ab 90bc 101c 3a 28b 52c 62c

Wastewood 5a 16b 20bc 24c 1a 6ab 12b 14b 1a 3a 5b 5b

Sum 25a 74b 117c 173d 7a 88ab 146b 180b 12a 77ab 147b 175b

Dead volume (m3 ha–1)Log 8a 4a 8a 7a 10a 15a 14a 12a 28a 29a 17a 13a

Pulp 25a 7b 5b 2b 35a 19ab 7b 8b 20a 14ab 6bc 0c

Wastewood 9a 1b 1b 1b 7a 6a 1b 1b 2a 1ab 1ab 0b

Sum 42a 12b 14b 10b 52a 40a 22a 21a 50a 44ab 24b 13b

Total volume before the onset of the treatment was used as the covariate in equation 2.Treatments marked with the same letter (in the same line and within the onset stages) are not significantlydifferent (P ≥ 0.05).

12 cph037 MATHS 9/9/04 9:59 am Page 358



ownloaded from



Figure 7. Cumulative volume of dead trees on plots of different thinning intensity: (a) early; (b) medium;(c) late onset of the treatment. For explanation of the symbols, see Figure 2.

12 cph037 MATHS 9/9/04 9:59 am Page 359



ownloaded from


unthinned plots, the volume production was onlyreduced by 4–7 per cent. The results of this studyare similar to the results reported in thinningexperiments of Norway spruce in otherFennoscandian countries (Möller, 1954; Carbon-nier, 1957, 1974; Bryndum, 1967, 1969;Eriksson, 1987; Pettersson, 1993; Eriksson andKarlsson, 1997; Braastad and Tveite, 2000,2001) and in central Europe (Assmann, 1954;Hamilton, 1976; Schober, 1979, 1980; Kramerand Jünemann, 1985; Spellmann, 1986). Inaddition, according to the growth and yieldtables for Norway spruce developed in Norway(Braastad, 1975), Sweden (Eriksson, 1976),Finland (Vuokila and Väliaho, 1980) andGermany (Assmann and Franz, 1965), totalvolume production during a stand rotation is notclosely related to the thinning intensity. Thevolume of the trees possibly removed in precom-mercial thinnings and commercial thinningscarried out before the study was not known.However, the trees removed have no effect on thedifferences among the treatments.

The results clearly demonstrate that theremaining trees can rapidly occupy the growingspace released by the removed trees. The averagediameter increment of all trees, as well as that ofthe largest trees clearly increased (cf. Hamilton,1976; Abetz and Unfried, 1984; Kramer andJünemann, 1985; Eriksson, 1987). According toBraastad and Eikeland (1986) and Braastad andTveite (2001), thinning either did not or onlyslightly accelerated the growth of the largesttrees. Their results were based on delayedand light thinnings and, most probably, the dif-ferences in thinning intensity explain thesedifferences.

The treatments had no effect on dominantheight increment. On the heavily thinned plots,height increment was slightly faster than thatwith the other treatments. Thus, the results ofthis study are in accordance with most studies onNorway spruce, i.e. the height increment ofdominant trees is not affected by stand density(cf. Bryndum, 1967; Kramer and Jünemann,1985; Eriksson, 1987; Handler, 1990; Erikssonand Karlsson, 1997). According to Möller(1954), Bryndum (1969) and Hamilton (1976),height increment may, however, increase withincreasing thinning intensity. Those authors havespeculated that height growth could be stimu-

lated by the nutrients released from loggingresidue. On the other hand, reduced heightgrowth on Norway spruce has also been reportedafter heavy thinning (Abetz, 1976; Abetz andUnfried, 1984).

In this study, the sites of the experiments werefertile and this may contribute to the fast growthof the remaining trees after the release. On lessfertile sites, the recovery of the remaining treesmay take a longer time and heavy thinningsprobably result in larger growth reductions(Mäkinen and Isomäki, 2004a). On the otherhand, the treatment onset stage had no majoreffect on the recovery after the thinning, i.e.slower growing older trees could utilize the freegrowing space as fast as the trees in youngerstands. Even though the medium and late onsetstands were dense at the time when the experi-ments were established, they were not over-stocked and the tree crowns were vigorous(Mäkinen and Isomäki, 2004b). Therefore, theywere able to rapidly respond to the treatments.

The experiments used in this study wereoriginally planned to study the effects on growthand yield of heavy and seldomly repeated thin-nings (Vuokila, 1983). Even though the heaviestthinnings were considered exceptionally intensiveat the time of establishment, the results showedthat heavier treatments are needed to find outthinning intensities that clearly reduce the totalstem volume production. In Norway sprucestands growing on fertile soils, removing even50 per cent of the growing stock has not resultedin marked reductions in volume increment(Möller, 1954; Carbonnier, 1967; Abetz, 1976;Eriksson and Karlsson, 1997). New thinningexperiments that cover the whole stand densityrange from free-growing trees to unthinnedstands are needed in order to define the completerelationship between stand density and volumeproduction.

As was the case for total volume production,merchantable volume produced up to the lastmeasurement was highest on the unthinned andlight thinned plots. Only the heavy thinningresulted in a reduced merchantable volume. Highstand density may, however, keep tree size underthe merchantable limit. Therefore, after a certainstand density level, increasing the stand densitymay also decrease the total merchantable volume(e.g. Spellmann and Nagel, 1996; Zeide, 2001).

360 FORESTRY

12 cph037 MATHS 9/9/04 9:59 am Page 360



ownloaded from


At establishment, the experiments of this studywere not completely unthinned. Stand densitieson the unthinned plots were probably not highenough to reduce the merchantable volume perhectare.

Even though the differences in total volumeproduction between the unthinned plots andlightly and moderately thinned plots were small,a part of the total production was lost throughnatural mortality on the unthinned plots. Thevolume of dead trees was not taken into accountwhen calculating the merchantable volume. Inpractical forestry, a part of natural mortality maybe collected in sanitary cuttings or is still usableat the time of normal harvesting and, thus, thedifferences in merchantable volume may besomewhat larger. In the earlier measurements, thecause of mortality was not defined and it oftenremained unclear in the later ones. On theunthinned plots, the main reason was mostprobably the competition among trees. Accord-ing to Abetz and Unfried (1984) and Laiho(1987), thinning increases the risk of wind andsnow damage for a number of years after thetreatment. In contrast, the results of this studyindicate no dramatically increased damage riskon thinned plots. However, the plots were rathersmall and located within a closed stand and,therefore, they do not represent conditions onlarger thinned areas.

The differences in basal area per hectarebetween the thinned stands and their unthinnedcounterparts have commonly been used as anapproximation of competition and suppression inthinned stands (e.g. Pienaar, 1979). Growth afterthinning has been related to the basal arearemaining in the stand immediately afterthinning. In this study, the growth responses tothinning were evaluated against the differences inaverage basal area between the thinned andunthinned plots during the whole measurementperiod. The thinning intensity was thereforecalculated not only on the basis of the basal arearemoved in thinnings, but also on the growthresponse after the thinnings. Thus, the treatmentsand possible differences in site fertility among theplots may have an effect on average basal arealevel. In addition, the length of the measurementperiod varied among the experiments. Periodlength may not, however, introduce any large biasin the relative differences between the treatments

because all the treatments were included inalmost all of the experiments.

In those experiments where thinnings werebased on the number of stems per hectare, it wassoon noticed that the original research planresulted in a rather similar basal area per hectare,even though the differences in stem number werehigh. Thus, the original plan was modified andthe thinnings were intensified. Furthermore, insome experiments the original schedule could notbe completely followed because of high measure-ment costs, low thinning removals that were noteconomically viable for the land owner, etc. Thesame average basal areas were classified into thesame group regardless of the thinning schedule,i.e. whether the basal area level was achievedwith one intensive thinning or with several lightthinnings was not taken into account. All thesechanges during the long measurement periodresulted in different intensities and intervalsbetween successive thinnings, as well as adifferent number of thinnings. These irregulari-ties may diminish the potential differencesbetween the thinning intensities. At least, theresults cannot shed any light on the profitabilityof few intensive thinnings compared with severallight thinnings.

The results are based on experiments on fertilesites in southern Finland. The experimentalstands were even-aged and almost pure plantedNorway spruce stands with a homogeneousspatial structure. In addition, strip roads werelocated outside the plots. In southern Finland, thetotal production loss in Norway spruce standscaused by strip roads at a spacing of 30 m was,on average, 10 m3 ha–1 during the 15-year periodafter the first commercial thinning (Isomäki andNiemistö, 1990). In practical forestry, thedistance between strip roads is nowadays 20 mand most probably results in higher productionlosses. In this study, the trees were manuallyfelled in random directions and the tree crownsof the felled trees were not directed towards thestrip roads as in current mechanized thinnings.Therefore, the nutrients released from loggingresidues were spatially more evenly distributed.Thus, the results do not completely represent thedevelopment of normal commercial forests.

As described above, the material was ratherheterogeneous and it had some shortcomings.The long duration of the experiments has caused


12 cph037 MATHS 9/9/04 9:59 am Page 361



ownloaded from


technical problems, which became more evidentover the course of time. On the other hand, theexperimental series used in this study is rare inrespect to its extent and duration. Many papersreporting thinning effects are limited to singleexperiments with a small number of replicationsand no variation in tree age.

In conclusion, thinning in Norway sprucestands does not increase the total volume incre-ment of trees. The results of this study confirmedthe previous Fennoscandian results, based on tem-porary sample plots, that increasing thinningintensity results in only a small reduction in totalstem volume and merchantable volume produc-tion within a large range of stand densities.However, the diameter increment of the remain-ing trees was clearly increased by thinnings. Meanstem diameter of the stand is often used as thecriterion of final felling in practical forestry(Hyvän metsänhoidon, 2001). Thus, standrotation can be shortened many years withoutsignificant losses in volume yield by applyingmore intensive management regime (H. Mäkinen,J. Hynynen and A. Isomäki, 2004). This studyonly concentrated on wood production on a cubicmetre basis. Whether thinning costs are more thancompensated by the income from earlier thinningsand a higher proportion of larger-sized stemsremains a topic for further investigation (cf.Valsta, 1982, 1992; Hyytiäinen, 2003).

Acknowledgements

We are greatly indebted to many technicians, especiallyto Timo Siitonen, who have maintained and measuredthe experiments. We also thank Professor Jari Hynynenfor his support and advice, Pentti Niemistö and DrAnssi Ahtikoski for valuable comments on the manu-script, Dr Helena Henttonen for the annual incrementindices, John Derome for revision of the language andSointu Nenola for drawing Figure 1.

References

Abetz, P. 1976 Reactions to enlarged growing spaceand consequences for the selective thinning inNorway spruce. Allg. Forst-u. Jagdztg. 147,72–75 [in German with English summary].

Abetz, P. and Unfried, P. 1984 Thinning experimentsin Norway spruce in the forestry district Riedlingen/Danube. Allg. Forst-u. Jagdztg. 155, 89–110 [inGerman with English summary].

Anon. 1948 Julkilausuma. MetsätaloudellinenAikakauslehti 11, 315–316 [in Finnish].

Assmann, E. 1954 Grundflächenhaltung und Zuwach-sleistung Bayerischer Fichten-Durchforstungsreihen.Forstwiss. Cbl. 73, 257–271 [in German].

Assmann, E. and Franz, F. 1965 Vorläufige Fichten-Ertragstafel für Bayern. Forstwiss. Cbl. 84,13–43 [in German].

Blomqvist, A.G. 1872 Tabeller framställande utveck-lingen af jemnåriga och slutna skogbestånd af tall,gran och björk. Helsinki [in Swedish].

Braastad, H. 1975 Yield tables and growth models forPicea abies. Reports of the Norwegian ForestResearch Institute 31, 357–537 [in Norwegian withEnglish summary].

Braastad, H. 2001 Ogallrat lika bra som gallrat.Skogen 1/01, 38–41 [in Swedish].

Braastad, H. and Eikeland, H. 1986 Thinning experi-ment in Picea abies forest. Density, increment anddiameter distribution. Experiment 919, Majer, Ø.Toten. Norwegian Forest Research Institute,Research Paper No. 12/86 [in Norwegian withEnglish summary].

Braastad, H. and Tveite, B. 2000 Tynning i granbe-stand. Effekten på tilvekst, dimensjonsfordeling ogøkonomi. Norwegian Forest Research Institute,Research Paper No. 4/00 [in Norwegian].

Braastad, H. and Tveite, B. 2001 Tynning i gran- ogfurubestand. Effekt av tynning på volumproduksjon,middeldiameter og diameter av de 800 grøvste trœrper ha. Norwegian Forest Research Institute,Research Paper No. 10/01 [in Norwegian].

Bryndum, H. 1967 Thinning experiments in youngNorway spruce. Medd. Norske Skogfors. ves. 84(bd. XXII), 49–63 [in Norwegian with Englishsummary].

Bryndum, H. 1969 A thinning experiment in Norwayspruce in Gludsted plantation. Det ForstligeForsøgsvæsen i Danmark 32, 1–156 [in Danish withEnglish summary].

Cajander, A.K. 1909 Forest types and their significance.Acta For. Fenn. 56, 1–69.

Cajander, E.K. 1933 Tutkimuksia Etelä-Suomen vilje-lykuusikoiden kehityksestä. Commun. Inst. For.Fenn. 19(3), 1–101 [in Finnish with Germansummary].

Carbonnier, C. 1957 Ett gallringsförsök i planteradgranskog. Svenska Skogsvårdsföreningens Tidskrift55, 463–476 [in Swedish with German summary].

Carbonnier, C. 1967 Comments on thinningprogrammes for spruce stands. Medd. NorskeSkogfors. ves. 84 (bd. XXII), 139–155 [in Swedishwith English summary].

Carbonnier, C. 1974 Preliminary results from athinning experiment in a Norway spruce plantation.Department of Forest Yield Research, Royal College

362 FORESTRY

12 cph037 MATHS 9/9/04 9:59 am Page 362



ownloaded from


of Forestry, Stockholm, Research Notes 29 [inSwedish with English summary].

Elfving, B., Johansson, T. and Johansson, U. 2001Gallra inte av slentrian. Skogen 3/01, 58–59 [inSwedish].

Eriksson, H. 1976 Yield of Norway spruce in Sweden.Department of Forest Yield Research, Royal Collegeof Forestry, Stockholm, Research Notes 41 [inSwedish with English summary].

Eriksson, H. 1987 New results from plot no. 5 at Sper-lingsholm estate in southwestern Sweden in theEuropean stem number experiment in Picea abies.Scand. J. For. Res. 2, 85–98.

Eriksson, H. and Karlsson, K. 1997 Effects of differentthinning and fertilization regimes on the develop-ment of Scots pine (Pinus sylvestris (L.)) and Norwayspruce (Picea abies (L.) Karst.) stands in long-termsilvicultural trials in Sweden. Department of ForestYield Research, Swedish University of AgriculturalSciences, Report 42 [in Swedish with Englishsummary].

Hamilton, G.J. 1976 The Bowmont Norway sprucethinning experiment 1930–1974. Forestry 49,109–121.

Hamilton, G.J. 1981 The effects of high intensitythinning on yield. Forestry 54, 1–15.

Handler, M.M. 1990 Spacing experiment with Norwayspruce, Picea abies (L.) Karst. Experiment 626/627,Spikkestad. Communications of the NorwegianForest Research Institute 43(4) [in Norwegian withEnglish summary].

Hasenauer, H., Burkhart, H.E. and Amateis, R.L. 1997Basal area development in thinned and unthinnedloblolly pine plantations. Can. J. For. Res. 27,265–271.

Heikkilä, T. 1914 Tuotantotaulut pääpuulajeillemme.Suomen Metsäyhdistyksen julkaisu. Erikois-tutkimuksia 2 [in Finnish].

Heinonen, J. 1994. KPL, a computer programmepackage for computing stand and single tree charac-teristics from sample plot measurements: manual.Finnish Forest Research Institute, Research PapersNo. 504 [in Finnish with English summary].

Henttonen, H. 1986 Puun kasvun ilmastollinenvaihtelu. Raportti Suomen LuonnonvarainTutkimussäätiölle [in Finnish].

Henttonen, H. 1990 Variation in the diameter growthof Norway spruce in southern Finland. Ph.D. thesis,University of Helsinki, Finland [in Finnish withEnglish summary].

Horne, R., Robinson, G. and Gwalter, J. 1986Response increment: a method to analyze thinningresponse in even-aged forests. For. Sci. 32, 243–253.

Hynynen, J., Ojansuu, R., Hökkä, H., Siipilehto, J.,Salminen, H. and Haapala, P. 2002 Models forpredicting stand development in MELA system.

Finnish Forest Research Institute, Research PapersNo. 835.

Hyvän metsänhoidon suositukset, 2001 Metsätaloudenkehittämiskeskus Tapio [Recommendations for goodsilviculture]. Helsinki, Finland [in Finnish].

Hyytiäinen, K. 2003 Integrating economics andecology in stand-level timber production. FinnishForest Research Institute, Research Papers No. 908.

Ilvessalo, Y. 1920 Kasvu- ja tuottotaulut Suomeneteläpuoliskon mänty-, kuusi- ja koivumetsille. ActaFor. Fenn. 15(4), 1–94 [in Finnish with Germansummary].

Ilvessalo, Y. 1932 The establishment and measurementof permanent sample plots in Suomi (Finland).Commun. Inst. For. Fenn. 17(2), 1–39.

Isomäki, A. 1993. Harvennus puuntuotannonohjauskeinona. Faculty of Forestry, University ofJoensuu, Finland, Research Notes 7, pp. 83–92 [inFinnish].

Isomäki, A. 1995 Thinning as a wood production tech-nique in Norway spruce stands in southern Finland.In Inventory of Forest Resources and Small-scaleForestry. Post-congress excursion Guide 6. IUFROXX World Congress, August 1995. K. Mielikäinenand H. Mikkelä (eds). IUFRO-95 Congress Secre-tariat, Finland, pp. 10–11.

Isomäki, A. and Niemistö, P. 1990 Effect of strip roadson the growth and yield of young spruce stands insouthern Finland. Folia For. 756, 1–36 [in Finnishwith English summary].

Kalela, E.K. 1948 Luonnonmukainen metsien käsittely.Silva Fenn. 64, 16–32 [in Finnish].

Kallio, K. 1957 On the development of spruce forestsof the Oxalis-Myrtillus site type in the southwest ofFinland. Forest mensuration and managementresearch. Acta For. Fenn. 66(3), 1–155 [in Finnishwith English summary].

Kramer, H. and Jünemann, D. 1985 The influence ofheavy individual and schematic thinning on thegrowth of young spruce stands. Forstarchiv 56,253–258 [in German with English summary].

Laasasenaho, J. 1982 Taper curve and volume func-tions for pine, spruce and birch. Commun. Inst. For.Fenn. 108, 1–74.

Laiho, O. 1987 Susceptibility of forest stands to wind-throw in southern Finland. Folia For. 706, 1–24 [inFinnish with English summary].

Mäkinen, H. and Isomäki, A. 2004a Thinning intensityand growth of Scots pine stands in Finland. For.Ecol. Manage. (in press)

Mäkinen, H. and Isomäki, A. 2004b Thinning inten-sity and long-term changes in increment and stemform of Norway spruce trees. For. Ecol. Manage. (inpress)

Mielikäinen, K. 1978 Predictability of tree growth. FoliaFor. 363, 1–15 [in Finnish with English summary].


12 cph037 MATHS 9/9/04 9:59 am Page 363



ownloaded from


Möller, C.M. 1954 The influence of thinning onvolume growth. Part I. In Thinning Problems andPractice in Denmark. S.O. Heiberg (ed.). State Univ.New York, Coll. For., Tech. Publ. No. 76, 5–32.

Näslund, M. 1937 Skogsförsöksanstaltens gallrings-försök i tallskog. Meddelanden från Statens Skogs-försöksanstalt 29(1), 1–169 [in Swedish withGerman summary].

Pettersson, N. 1993 The effect of density after precom-mercial thinning on volume and structure in Pinussylvestris and Picea abies stands. Scand. J. For. Res.8, 528–539.

Pienaar, L.V. 1979 An approximation of basal areagrowth after thinning based on growth in unthinnedplantations. For. Sci. 25, 223–232.

SAS Institute 1999 SAS/STAT: User’s Guide, Version 8.SAS Institute, Cary, NC.

Schober, R. 1979 Massen-, Sorten- und Wertertrag derFichte bei verschiedener Durchforstung. Teil I. Allg.Forst-u. Jagdztg. 150, 129–152 [in German].

Schober, R. 1980 Volume, assortment and value yieldof Norway spruce under different thinning treat-ments. Allg. Forst-u. Jagdztg. 151, 1–21 [in Germanwith English summary].

Spellmann, H. 1986 The effects of various thinning tech-niques and intensities on the development of a widely-spaced Norway spruce stand. Allg. Forst-u. Jagdztg.157, 27–31 [in German with English summary].

Spellmann, H. and Nagel, J. 1996. Aspects concerningthe thinning of Norway spruce. Allg. Forst-u. Jagdztg.167, 6–15 [in German with English summary].

Valsta, L. 1982 Profitability comparison of growingdensities in spruce plantations. Folia For. 504,1–33 [in Finnish with English summary].

Valsta, L. 1992 An optimization model for Norwayspruce management based on individual-tree growthmodels. Acta For. Fenn. 232, 1–20.

Vuokila, Y. 1956 On the development of managedspruce stands in southern Finland. Commun. Inst.For. Fenn. 48(1), 1–138 [in Finnish with Englishsummary].

Vuokila, Y. 1960 On low thinning and crown thinning,and their effect on development of growing stock inplanted spruce stand of Rautio. Commun. Inst. For.Fenn. 52(8), 1–29 [in Finnish with Englishsummary].

Vuokila, Y. 1965 Functions for variable density yieldtables of pine based on temporary sample plots.Commun. Inst. For. Fenn. 60(4), 1–86.

Vuokila, Y. 1975 Thinning of young spruce plantationsas a problem of timber production. Folia For. 247,1–24 [in Finnish with English summary].

Vuokila, Y. 1977 Vaccinium type as a spruce site. FoliaFor. 324, 1–14 [in Finnish with English summary].

Vuokila, Y. 1980 The dependence of growth and yieldon the density of spruce plantations in Finland. FoliaFor. 448, 1–15 [in Finnish with English summary].

Vuokila, Y. 1983 Suomalaisen puuntuotostutkimuksenmenneisyys ja tulevaisuus. Finnish Forest ResearchInstitute, Research Papers No. 89 [in Finnish].

Vuokila, Y. 1985 The effect of growing stock level onthe development, growth and yield of spruce planta-tions in Finland. Folia For. 636, 1–18 [in Finnishwith English summary].

Vuokila, Y. and Väliaho, H. 1980 Growth and yieldmodels for conifer cultures in Finland. Commun.Inst. For. Fenn. 99(2), 1–271 [in Finnish with Englishsummary].

Zeide, B. 2001 Thinning and growth: a full turn-around. J. For. 99, 20–25.

Received 25 August 2003

364 FORESTRY

12 cph037 MATHS 9/9/04 9:59 am Page 364



ownloaded from


Forestry-2004-Mäkinen-349-64.pdf

Documents

Transcript of Forestry-2004-Mäkinen-349-64.pdf