Residual tree damage along skidding trails in beech stands in Greece

Petros A.Tsioras?Diamantis K.Liamas

Residual tree damage along skidding trails in beech stands in Greece

Petros A.Tsioras1?Diamantis K.Liamas1

We studied the damage caused to unfelled trees during skidding operations in four beech stands of northern and central Greece that were managed under differing harvesting systems.Aftertimberwas harvested we recorded and analyzed alldamage to a width of 2 m along both sides of the skid trails.The percentage of damaged residual trees was 17–28%.Highernumbers ofsaplings were uprooted or destroyed in parts of the stands with high natural regeneration.Most damaged trees suffered 1.33–1.90 wounds thatwere inflicted up to a heightof 1 m above ground level. Atallsites mean wound sizes were large,89–1190 cm2,and caused elevated risk of future fungalinfection.The damage caused by skidding during timber harvesting can be minimized by better planning the harvest operations,and training forest workers in reduced impact logging.

Residual tree damage·Beech·Skidding trails·Wood harvesting·Greece

Introduction

Beech is a valuable tree species in Greek forestry because of its potential for wood production.Beech forests are located at sites of high production potential in terms of woodquality and quantity when the proper silvicultural treatments are implemented.Beech stands cover 336,640 ha (5%of the total forest area)in Greece,and the total volume of standing timber is estimated at 82 m3ha-1comprising 20%of the nationaltotalwood volume(Efthymiou 2009).

Selective harvesting is the predominant method in Greece.Single-tree selection is a silvicultural system in which individualtrees are removed throughouta harvesting site at regular time intervals(Nyland 2002).Although selective harvesting results in limited change to stand structure compared to other more intensive harvesting methods (Caspersen 2006),some damage to unfelled trees is inevitable.Excessive wounding of the residual trees can reduce the expected benefits of selective harvest by compromising tree vigor and stand quality,and by degrading the aesthetic and recreational values of some stands(Bustos et al.2010).Harvesting can cause injuries to unfelled trees and this results in further economic losses from forest management.Injuries often become an inputportfor fungal infection(Vasiliauskas 2001).According to the size of the wound,fungal decay can cause wood depreciation(devaluation)or even reduced growth increments(Heitzman and Grell 2002;Limbeck-Lilienau 2003).Forest management faces quantitative losses that are not reflected in short-term accounts,but become apparent over longer time periods(Dvorˇa′k and Cerny 2003).

The amount and characteristics of skidding damage are affected by site factors,tree species and logging season (Froehlich 1976).Skidding during winter causes less damage to soil and residual trees as a result of the protection provided by snow cover(Limbeck-Lilienau 2003). Slope has been correlated with increased skidding damage (Stampfer et al.2001).Although site-specific and seasonal factors are important,the amount of damage is to aconsiderable degree affected by the level of training and consciousness of the forest workers and contractors(Tavankar et al.2013).

Wound dimensions and wound area and location on the trunk are important determinants of the development of fungal infections,and these differ by timber species.In beech,complete closure without infection was reported in all wounds＜5 cm width,in 70%of wounds 5–8 cm width and in 50%of wounds＞8 cm width(Hosius 1967;Vasiliauskas 2001).In spruce(Picea abies(L.)Karst),the stage of decay increases in relation to the size of injury when the wound area exceeds 10 cm2(Meng 1978).Also,in the cases of fir(Pseudotsuga Menziessi)(Aho etal.1983)and spruce (Meng 1978),higher moisture levels close to the ground favor decay compared to lower moisture levels higher on the stem,making wounds near the root collar more susceptible to infection than wounds higher on the trees. However,this finding is not valid for beech,as wounds on beech trees that occur near ground level are less harmful compared to wounds higher on the stem(Nill et al.2011).

We present here the results of our study of pure and mixed beech stands in four areas of Greece with different harvesting systems.We recorded and analyzed all damage caused during the skidding operations with different types of skidding equipment.Our results show the residual tree damage levels along skidding trails in beech stands in Greece and represent a critical step toward developing management plans to minimize stand damage caused by logging activity.

Materials and methods

Study site

Field work took place in four areas in northern and central Greece:Arnaia(site 1),Stefanina(site 2),Karitsa(site 3) and Perivoli Grevenon(site 4)(Fig.1;Table 1).Two main reasons for choosing the above areas were their representativeness with regard to stand conditions and the differentiation of the harvesting systems in terms of timber extraction methods.The trees were marked by experienced forest officers of the responsible Forest Service Office for each site.Data collection started at sites 1,2 and 3 in July 2009 immediately after logging,and at site 4 during August 2011.However,we visited all sites before and during the forest operations in an effort to better understand the working environment and the timber harvest conditions.

Fig.1 Map of the study areas

At all study sites the trees were felled,limbed and topped motor-manually.Fallen trees were bucked and processed with chainsaws into logs of the desired length at the stump.Roundwood and fuelwood extraction to roadsides or landings took place with different combinations of forest machinery equipment(Table 2).The equipment types under study were characteristic for Greek conditions. In most cases agricultural tractors were used for wood extraction and were equipped with a single or double drum winch,with a cable length of 70–100 m.There is a trend of replacing old tractors with newer models of skidders such as LKT81 skidders that were typically also equipped with double drum winches.At site 1,a BOBCAT 733 equipped with a single drum winch was used in addition to a tractor. According to the members of this forest worker cooperative(hereafter FWC)this piece of equipment‘contributed substantially with its versatility in more efficient wood extraction’.Finally,mules were used in fuelwood extraction at sites 1,2 and 3.

Harvesting was conducted by FWCs assigned by the responsible Forest Service Office in each area.Equipment operators used their own equipment in an effort to control performance results and avoid higher residual stand damage thatmighthave been caused by lack of familiarity with new equipment.The workforce characteristics(age distribution,work experience)were recorded in each area to account for performance differences between experienced and inexperienced forest workers.All FWCs with members having work experience of less than 5 years were excluded from further consideration.

Data collection

We observed and recorded residualtree damage along skid trails(primary and secondary)after felling.In order to determine the percentage and types of damage,we examined all trees along the skid trails at a distance of two meters from both edges of each skid trail.The width of skid trails was not fixed but ranged in 3.4–4.1 m.We recorded all damage to trees of DBH≥7 cm.We also counted and recorded destroyed or uprooted saplings (young trees of DBH＜7 cm).

Data from sites 2 and 3 were from two stands in each case.However,the close proximity,the identical stand conditions and the fact that they were harvested by the same FWCs with the same harvesting methods allowed us to treat them as one site.This approach was supported by the statistical similarity of means for site 2(stands 3c and 3d)and site 3(stands 20a and 19b).

Tree wounds were documented according to the method described by Meng(1978).All damage to residual trees was recorded with regard to location and height on thestem(Fig.2)and wound size.Wound depth was also recorded in cases where wood tissues had been destroyed.

Table 2 Dimensions and specifications of the wood extraction equipment

Fig.2 Classification of damages according to location on the root and stem(Meng 1978)

The segregation of wounds into severity categories (Table 3)permitted assessment of risk of fungal infection (Limbeck-Lilienau 2003).With increasing severity of damage the probability of infection increases.According to Meng(1978),damage in the size-class DC 1 is irrelevant and there is generally no risk of an infection by wooddestroying fungi.When external damage to bark occurs,a fungal infection is not expected(Butora and Schwager 1986).If the bark is squeezed,the tree would only rarely be infected by fungi.Infection and the resulting decay mostly occur when the wood becomes visible due to removal of bark(DC 3).In the case of real wood injury(DC 4),the probability,that wound rot fungi appears,increases by 40–50%compared with DC 3.The highestrisk of decay is faced by trees with injuries in the area of the felling cutand the root collar.Damage to superficialroots and to the trunk above the root collar(＞0.3 m)less frequently leads to infection by wood-destroying fungi(Meng 1978).

Statistical analysis

We used one-way analysis of variance(ANOVA)and Duncan’s post hoc test to assess differences in wound attribute means between the four study sites.We tested normality and homogeneity of variance by applying Kolmogorov–Smirnov and Levene’s tests,respectively.We used Somer’s d ordinal measure of association,included in the cross tabulation module of SPSS ver.20,to evaluate possible correlation between the slope class,the number of inflicted wounds per tree,and the size class of inflicted wounds(p＜0.05).

Results

Damage frequency and distribution

A total of 1789 live stems of DBH≥7 cm were sampled at all four study sites of which 388 were wounded (Table 4).The percentage of damaged residual trees varied from 17%(site 4)to 28%(site 3).Most wounded stems were of beech trees at all study sites.The number of damaged saplings varied greatly,from 5 at site 1 to 141 at site 4.This resultis analogous with the stand condition,as, especially at site 4 and to a lesser degree at site 3,more saplings were wounded or uprooted in parts of the stand where natural regeneration was most successful.

Trees suffered different numbers of wounds across the study sites(Table 5).This varied from 1 to 4 at the first three sites,and from 1 to 5 atsite 4.This resulted in a mean of 1.33–1.44 wounds per tree for the first three sites and 1.90 wounds per tree at site 4.Trees at site 4 sufferedsignificantly more wounds than did trees at sites 1–3 (F=9.641,df=3,p＜0.0001).Numbers of wounds did not vary by tree species.

Table 3 Description ofdamage categories(DC 1–4)

Table 4 Number and percentages of damaged trees and saplings

Table 5 Wounds per tree statistics across the different study sites

Mean damage distance was 38 cm to 81 cm (F=33.497,df=3,p＜0.0001)from the edge of skid trails(Table 6).At all sites most damage(74–95%)occurred within a distance of 1 m from the edge of the skid trails.

Damage characteristics

The tree stem at heights of 30–100 cm above ground level was most susceptible to damage and more than 40%of all wounds were inflicted here(Table 7).The exception was site 1,where damage to roots was most common and occurred at a frequency double that at the other sites (χ2=58.734,df=9,p＜0.0001).The mean heightabove ground level of wounds varied by site from 33 to 62 cm (F=9.973,df=3,p＜0.0001).

Wounds were unevenly distributed by area class (χ2=109.794,df=9,p＜0.0001).Most wounds were categorized as DC3 or DC4 with regard to size(Table 8). Mean wound size at sites 1,3 and 4 ranged from 466–1190 cm2.At site 2,78%of wounds had an area of 10–200 cm and averaged 89 cm2.Many exposed roots were evident at site 1.For this reason,we analyzed the wounds on the tree stems(Table 9).Mean wound area values were considerably lower,ranging from 80 to 552 cm2(F=15.592,df=3,p＜0.0001).The largest means for wound width and length were recorded at site 1 and the highestmean for wound depth was recorded at site 3.

The large majority of wounds were recorded on slopes up to 20%in gradient(Table 10).At sites 1–3,the frequency of wounds decreased as slope gradient increased from 0–10%.At site 4 the highest wound frequency was recorded on slopes gradients of 10–20%,followed by gradients of 20–30%.There was no correlation between slope class,the number of wounds per tree and wound size class at any study area(Table 11).

Discussion

We recorded damage data for all trees of DBH＞7 cm. However,in other studies the limits for recording damaged stems were different[e.g.:5 cm in Tatsumiet al.(2013),2 in(5.08 cm)in Heitzman and Grell(2002),8 cm in Modiget al.(2012),4 in(10.16 cm)in Hassler et al.(1999)]. Another factor affecting the comparability of results is the choice of recording methodologies,which is discussed by Hassler et al.(1999)and Heitzman and Grell(2002). Despite these methodological differences from previous reports,we report here for the first time information on residualtree damage levels along skid trails in beech stands in Greece.

Table 6 Percentage occurrence of residual tree damages according to distance from the edge of the skidding trail

Table 7 Distribution percentages of the damages according to their location on the tree

Table 8 Wound area statistics

Table 9 Wound damage dimensions for trees with DBH≥7 cm(trunk measurements)

Table 10 Frequency distribution of the slope classes in the study areas

Table 11 Somer’s d statistic results for slope class association of the numbers of wounds per tree and wound size class

Many studies reported damage levels after timber harvest.Residual stand damage varies greatly from less than 5%reported by McNeel and Ballard(1992)after a thinning operation in a Douglas-fir plantation on flat to rolling terrain(0–17%terrain slope),up to 40%or even more,as reported for harvester-forwarder operations by Bettinger and Kellogg(1993)and by Han and Kellogg(2000).Trees along the trailside have been reported to suffer more damage than non-trailside trees(Han and Kellogg 2000; Heitzman and Grell 2002;Modig etal.2012).The damage percentages in our study varied between 17 and 28%and did not exceed the levels reported in other studies.For example,Heitzman and Grell(2002)in a similar study of eight stands reported damage percentages ranging in 30–60%for trailside trees.

At our study sites 1–3,we recorded few damaged saplings compared to site 4,where damaged saplings outnumbered the damaged trees.Damage to saplings was not reported by most similar studies although it could be regarded as another index of the environmental impacts of wood harvesting.

Every wounded tree suffered an average of more than 1.3 gouges and/or scars,but in some cases,trees suffered up to five wounds,as at site 4,which had the highest standing volume and harvest intensity values(Table 1).At site 4 black pine stems of larger diameters than the beech stems were felled and this caused more disturbance to the remaining trees.Increased damage frequency might be due to unprotected trees along the skid trail,turning of the log loads and bumping of trees during winching(Limbeck-Lilienau 2003;Solgi and Najafi2007;Han and Kellogg 2000;Ezzati and Najafi2010).Han and Kellogg(2000) suggested that artificial tree protection rigging such as rub pads should be used to prevent damage to stumps and stems.Poor planning and implementation of harvesting operations may have also been important determinants of wound frequency.At most sites,many forest workers and machines were often concentrated at certain points.This resulted in unnecessary maneuvers of the skidding equipment and wounds that could have been easily avoided.

Wound frequency decreased with distance from the forest road,as expected(Nichols et al.1994;Spinelli et al. 2010;Picchio etal.2012).Han and Kellogg(2000)studied damage to young Douglas-fir stands by four timber harvesting systems and found that damaged trees were more concentrated along the skid trails for tractor logging than for skyline and cut-to-length logging.In their case, about 56%of the total damage(any size scar)in skyline, 64%in cut-to-length,and 80%in tractor thinning units was observed within 5 m of the centerline.Greatest damage occurred within the first 3 m from the centerline of the skid trails or the skyline corridors.In another study of skidding damage,Froese and Han(2006)found that 67.7%of damaged trees were located within 4 m from the center of the skid trail.In our study,skid trails ranged from 3.4 to 4.1 m in width,so we conclude that most damage occurred within the first 3–4 m from the skid trail centerline,or in the first two meters from the skid trail edge. However,it should be noted that at site 4,the machine operator often unjustifiably directed the tractor outside the skid trail at points where saplings were established,thus inflicting unnecessary damage.

The percentage of damage to roots and the stem below a heightof 1 m represented more than 80%of alldamage at all sites.Most of this damage was recorded during round wood extraction,where the log hit residual trees or the cable contacted the standing stems.This resultis consistent with several other studies that identified the cut area as the location with the highest injury frequency(Han 1997; Ezzati and Najafi2010;Froese and Han 2006;Mousavi 2009).At site 4,20%of damage occurred at heights of more than 1 m.This was due to the increased difficulty of felling large black pine trees.

The average area of damage to roots was greater than thatfor the tree stem atallsites.This was probably because of the proximity of tree roots to the trail edge and repeated traffic over the same exposed roots(Bustos and Egan 2011).The large number and size of exposed roots atsite 1 was probably responsible for the higher mean wound area compared to the other study sites.At sites 1,3 and 4,the most frequent damage class was DC4(area more than 200 cm2).However,in all areas,wound areas larger than 10 cm2were recorded in more than 90%of the cases.

The lowest average wound area was recorded at site 2. This could be partially attributed to the low traffic of forest machinery due to the low volume of roundwood harvested. Only 136.2 m3of the harvested volume(1647.7 m3)was extracted along the skidding trails,while the remaining1511 m3of fuelwood was extracted with mules,using in most cases random routes through the stand from the harvesting site to the upper landing.The forest workers team in this area was better organized,with new machinery and good coordination.These characteristics of the harvesting team indicated a higher level of professionalism and a better quality of forest operations that resulted in better protection of the remaining forest stand.

Harvesting systems seemed to have a profound effect on the locations of damage to residual trees.At site 1 the use of a BOBCAT 733 equipped with iron tracks,was,in many cases,responsible for excessive root damage that could have been avoided with the use of different machinery. Limbeck-Lilienau(2003)found that the combination of wheeled harvester and forwarder caused more damage to roots(38%in summer,6%in winter)than the combination of tracked harvester and forwarder(10%in summer, 1%in winter).Han and Kellogg(2000)after comparing four harvesting systems and damage to residualtrees found that the amount of scarring below 60 cm was 2%for helicopter logging,17.5%for skyline logging,29.3%for cut-to-length logging,and 64%for tractor units.

Steeper slope has been recognized as a potential factor leading to higher damage levels(Stampfer et al.2001). This assumption was notsupported by our findings as mean damage area class(wound area)and mean number of wounds per tree did not vary by slope class.However,this might have happened because the present study examined residual damage along skidding trails,and did not sample plots inside the forest stand,where the prevailing conditions are different.In order to elucidate this point,further research is necessary.

Conclusion

We examined the impacts of skidding operations on the residual trees along skid trails in beech stands.Skidding damage resulted in exposure of sapwood and could lead to degradation of future forest yields.Wounds caused by skidding might function as entry points for insects and diseases.Therefore,the potential loss of future wood volume underlines the importance of skidding damage from the managerial point of view.

Our findings agree with those of other studies with regard to the frequency of damaged trees and sizes of damaged areas per tree.The frequency of damage depends on the harvesting system used,while there was no relationship between slope gradient and damage area.These preliminary results should not be generalized for all beech forests in Greece,since the study is still ongoing in other areas.This study should be further expanded to other species nationwide.

A weakness of the presentand similar studies,is thatthe number of damaged saplings per study site has been recorded but not the total number of saplings along the skidding trails as well.Therefore,we suggestfullinventory of both trees and saplings in similar residual stand damage research.This will increase knowledge of the inflicted damage levels and enable enhancement of silvicultural treatments to the benefit of natural regeneration.

Useful suggestions to reduce skidding damage are provided by Han and Kellogg(2000),Solgi and Najafi(2007) and Tavankar et al.(2013).These include planning the trails,utilizing the optimum trail spacing,keeping the trails straightand directional felling of trees on an angle towards trails.It is also important to keep skidders on the trails, limit skidding operations during wet periods,use the correct type and size of skidder,and to utilize bumper trees where required.

Another very importantfactor is forestworkers.None of the above recommendations will make a difference unless the professional capacity of the people involved in forest operations is increased(Tsioras 2012).Only well trained forest workers and machine operators can guarantee increased productivity,safety during work,and reduced environmental impacts during the forest operations.

Aho PE,Fiddler G,Srago M(1983)Logging damage in thinned, young-growth true fir stands in California and recommendations for prevention.USDA Forest Service,Portland 12

Bettinger P,Kellogg LD(1993)Residual stand damage from cut-tolength thinning of second-growth timber in the Cascade Range of western Oregon.For Prod J 43(11,12):59–64

Bustos O,Egan A(2011)A comparison ofsoilcompaction associated with four ground-based harvesting systems.North J Appl For 28(4):194–198

Bustos O,Egan A,Hedstrom W(2010)A comparison of residual stand damage along yarding trails in a group selection harvest using four different yarding methods.North J Appl For 27(2): 56–61

Butora A,Schwager G(1986)Holzerntescha¨den in Durchforstungsbesta¨nden,Berichtder Eidgeno¨ssischen Anstaltfu¨r das forstliche Versuchswesen,Birmensdorf,p 47

Caspersen JP(2006)Elevated mortality of residual trees following single-tree felling in northern hardwood forests.Can J For Res 36(5):1255–1265

Dvorˇa′k J,Cerny S(2003)Injuries on forest stands in Krusne Hory mountains caused by utilization of logging systems.Paper presented at the FORTECHENVIConference,May 2003

Efthymiou PN(2009)Harvesting of forest products—university lectures.Aristotle University of Thessaloniki,Thessaloniki

Ezzati S,NajafiA(2010)Long-term impact evaluation of groundbase skidding on residualdamaged trees in the Hyrcanian forest Iran.Int J For Res 2010:8.doi:10.1155/2010/183735

Froehlich HA(1976)The effect of soil compaction by logging on forest productivity,part 1:USDI Bureau of Land Management. USDI Bureau of Land Management,Portland,p 118

Froese K,Han HS(2006)Residual stand damage from cut-to-length thinning of a mixed conifer stand in northern Idaho.West JAppl For 21(3):142–148

Han HS(1997)Damage to Young Douglas-fir stands from commercial thinning with various timber harvesting systems and silvicultural prescriptions:characteristics,sampling strategy for assessment and future value loss,Oregon State University

Han HS,Kellogg LD(2000)Damage characteristics in young Douglas-fir stands from commercial thinning with four timber harvesting systems.West J Appl For 15(1):27–33

Hassler CC,Grushecky ST,Fajvan MA(1999)An assessment of stand damage following timber harvests in West Virginia.North J Appl For 16(4):191–196

Heitzman E,Grell AG(2002)Residual tree damage along forwarder trails from cut-to-length thinning in maine spruce stands.North J Appl For 19(4):161–167

Hosius D(1967)Bark stripping consequences on beech.Allgemeine Forst Und Jagdzeitung 22:484–487

Limbeck-Lilienau B(2003)Residual stand damage caused by mechanized harvesting systems.Paperpresented atthe AUSTRO 2003—High Tech Forest Operations for Mountainous Terrain, Schlaegl,Austria,October 5–9,2003

McNeel J,Ballard TM(1992)Analysis of site stand impacts from thinning with a harvester-forwarder system.J For Eng 4(1):23–29

Meng W(1978)Baumverletzungen durch Transportvorga¨nge bei der Holzernte—Ausma?und Verteilung,Folgescha¨den am Holz und Versuch ihrer Bewertung.,Schriftenreihe der LFV Baden-Wu¨rttemberg,p 159

Modig E,Magnusson B,Valinger E,Cedergren J,LundqvistL(2012) Damage to residual stand caused by mechanized selection harvest in unevenaged Picea abies dominated stands.Silva Fenn 46(2):267–274

Mousavi R(2009)Comparison of productivity,cost and environmental impacts of two harvesting methods in Northern Iran: short-log vs.long-log.University of Joensuu,Joensuu

Nichols MT,Lemin Jnr RC,Ostrofsky WD(1994)The impactoftwo harvesting systems on residual stems in a partially cut stand of northern hardwoods.Can J For Res 24(2):350–357

Nill M,Kohnle U,Sauter UH(2011)Rindenscha¨den mit mutma?lichem bezug zur holzernte im spiegel der betriebsinventuren in Baden-Wu¨rttemberg(Status quo and changes in loggingcaused bark damages in Baden-Wu¨rttemberg reflected in inventory data).Forstarchiv 82(6):216–224

Nyland RD(2002)Silviculture:concepts and applications.McGraw-Hill Companies Inc,New York

Picchio R,Magagnotti N,Sirna A,Spinelli R(2012)Improved winching technique to reduce logging damage.Ecol Eng 47:83–86.doi:10.1016/j.ecoleng.2012.06.037

Solgi A,NajafiA(2007)Investigating of residualtree damage during ground-based skidding.Pak J Biol Sci 10(10):1755–1758

Spinelli R,Magagnotti N,Nati C(2010)Benchmarking the impactof traditional small-scale logging systems used in Mediterranean forestry.For Ecol Manag 260(11):1997–2001.doi:10.1016/j. foreco.2010.08.048

Stampfer K,Steinmu¨ller T,Svaton R(2001)Grenzen der Steigfa¨-higkeit.O¨sterreichische Forstzeitung(Arbeit im Wald)112:1–3

Tatsumi S,Owari T,Kasahara H,Nakagawa Y(2013)Individuallevel analysis of damage to residual trees after single-tree selection harvesting in northern Japanese mixedwood stands. J For Res 31:1–10.doi:10.1007/s10310-013-0418-x

Tavankar F,Majnounian B,Bonyad AE(2013)Felling and skidding damage to residual trees following selection cutting in Caspian forests of Iran.J For Sci 59(5):196–203

Tsioras PA(2012)Status and job satisfaction of Greek forestworkers. Small-scale For 11:1–14.doi:10.1007/s11842-011-9164-0

Vasiliauskas R(2001)Damage to trees due to forestry operations and its pathological significance in temperate forests:a literature review.Forestry 74(4):319–336.doi:10.1093/forestry/74.4.319

5 November 2013/Accepted:21 March 2014/Published online:25 April 2015

?Northeast Forestry University and Springer-Verlag Berlin Heidelberg 2015

The online version is available at http://www.springerlink.com

Corresponding editor:Yu Lei

?Petros A.Tsioras ptsioras@for.auth.gr

1Laboratory of Forest Utilization,Faculty of Forestry and Natural Environment,Aristotle University of Thessaloniki, POB 227,541 24 Thessaloniki,Greece