Forest Policy and Economics 50 (2015) 200–209

Contents lists available at ScienceDirect

Forest Policy and Economics

journal homepage: www.elsevier.com/locate/forpol

Decision-making for the selection of a best wood extraction method: An

analytic network process approach
Abolfazl Jaafari a, Akbar Najafi a,b,⁎, Mónica García Melón c
a
Department of Forestry, Faculty of Natural Resources, Tarbiat Modares University (TMU), Noor, Iran
b
Department of Forest Harvesting Logistic and Amelioration, Faculty of Forestry, Technical University in Zvolen, Zvolen, Slovakia
c
DINGENIO (CSIC-UPV), Universitat Politècnica de Valencia, Camino de Vera s/n, 46022 Valencia, Spain

a r t i c l e i n f o a b s t r a c t

Article history: The importance of environmental and economical aspects in sustainable forest management (SFM), demands a
Received 26 January 2014 continuing need for the application of efficient methods in this field. In this paper, we applied analytic network
Received in revised form 4 August 2014 process (ANP) with benefits, opportunities, costs, and risks (BOCR) to evaluate four existing wood extraction
Accepted 13 September 2014
methods (rubber-tired skidder, tracked skidder, cable logging and animal skidding) and select the best one
Available online 8 October 2014
based on SFM strategies in the Caspian (Hyrcanian) Forest, in northern Iran. The required data were obtained
Keywords:
either from the previous studies or through pairwise comparisons conducted by a panel of forest engineering ex-
ANP perts. Based on the BOCR subnetworks and their corresponding criteria, four alternatives were synthesized. ANP
BOCR analysis indicated that the benefits and risks were more important in terms of decision-making, compared to the
Caspian Forest opportunities and costs. Finally, the alternative of ‘rubber-tired skidder’ was selected as the most suitable method
Forest harvesting for meeting SFM strategies. The decision framework proposed in this study offers a means for the
Sustainable forest management reorganizing and planning of timber harvesting operations as an integral step toward SFM. By making
minor adjustments for local conditions, this decision framework can be adapted for many other regions and
countries.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction
Timber harvesting is one of the main objectives of sustainable forest
management (SFM), the overall efficiency of which depends heavily on
the selection of an appropriate wood extraction method. Decisions for
selecting a method is based on a wide variety of criteria including site
condition, density of skid trails, suitable equipment and machines,
skilled forest labors and an effective resource management strategy.
Such a decision-making process can cause difficulties when reacting to
change in terms of benefits, opportunities, costs, and risks (BOCR) of
the alternatives. To date, economic efficiency has been the most impor-
tant criterion for selecting harvesting and/or extraction methods. How-
ever, the non-consideration of ecological and social criteria may impose
negative side-effects and risks that revoke this economic advantage
(Kühmaier and Stampfer, 2012). There are many quantitative and qual-
itative criteria related to harvesting operations that should be measured
or estimated and then integrated. This makes forest management plan-
ning processes more complex, creating a need for effective decision
analysis methods that are able to take into account criteria that have
been to date ignored in conventional planning methods (Laukkanen
⁎ Corresponding author at: Faculty of Natural Resources, Tarbiat Modares University,
P.O. Box: 64414-356, Noor, Mazandaran, Iran.
E-mail address: a.najafi@modares.ac.ir (A. Najafi).
et al., 2005). Various research can be found in the literature that
shows the use of multi-criteria decision-making (MCDM) methods
that greatly improve the efficiency of the forest management planning
process (e.g., Pukkala, 2002; Ananda and Herath, 2005, 2008; Kangas
and Kangas, 2005; Hayati et al., 2013; Pellegrini et al., 2013). The con-
cepts and methods of MCDM present a framework that incorporates
several conflicting criteria into planning (Miettinen, 2006). The MCDM
approaches range from addressing simple technical issues to complicat-
ed socio-economic problems (Lu et al., 2007). These approaches provide
a set of systematic procedures for analyzing complex decision problems
involved in SFM (Samari et al., 2012).
A forest management solution needs to be unique and based around
capitalizing on the strengths and mitigating weaknesses. In terms of for-
est management, it cannot with certainly be indicated what type of wood
extraction method is best, since the selection of a certain method involves
the consideration of different conditions such as slope, terrain shape,
skidding distance, soil, tree size, volume per hectare, weather, cost of
road construction, cost of logging, and productivity goals. In this sense,
selecting a wood extraction method is a MCDM problem. In this study,
we attempt to evaluate wood extraction methods and select the best
one by applying analytic network process (ANP) with BOCR based on
SFM strategies in the Caspian Forest, in northern Iran. The research un-
derlines the importance of providing a multi-criteria decision-making
tool for more efficient forest management in the given conditions.

http://dx.doi.org/10.1016/j.forpol.2014.09.010
1389-9341/© 2014 Elsevier B.V. All rights reserved.
 A. Jaafari et al. / Forest Policy and Economics 50 (2015) 200–209
2. ANP and related studies
As one of the decision-making techniques, analytic hierarchy pro-
cess (AHP) appears to be a comprehensive technique that aims to re-
move, to a great extent, the problems associated with traditional
methodologies (Kangas and Kangas, 2005; Samari et al., 2012). How-
ever, many decision-making problems cannot be structured as hierar-
chical, because they involve the interaction and dependence of higher-
level elements in a hierarchy on lower level elements. Therefore, by dis-
solving the structural boundaries toward a network structure covering
interdependencies and feedback between elements of different clusters,
ANP was proposed (Saaty, 1999). In the ANP, relations among criteria
and sub-criteria are involved in the evaluation process, allowing for de-
pendencies both within a cluster (inner-dependence) and between
clusters (outer-dependence) (Saaty, 2001). Pairwise comparison for
both weighting the clusters (i.e., criteria) and for estimating the direc-
tion and importance of influences between elements is conducted and
numerically presented as ratio scales in a so-called supermatrix
(Saaty, 2001). The supermatrix represents the influence priority of an
element in the left of the matrix on an element at the top of the matrix,
with respect to a particular control criterion.
Although the AHP has been successfully applied to a wide range of
decision-making problems during the past number of years, the ANP is
yet a promising approach with limited number of applications and pub-
lications in the literature (Niemiraa and Saaty, 2004; Erdoğmus et al.,
2005). However, the number of applications is continuously increasing.
Some of the recent ANP applications regarding decision-making prob-
lems include analysis of SFM strategies in Austria (Wolfslehner et al.,
2005), evaluation of high-tech alternatives in Turkey (Erdoğmus et al.,
2005), development of a model for supplying poplar wood to Iranian
paper and wood factories (Azizi, 2008), the appraisal of farmland
in Spain (García-Melón et al., 2008); and selection of the best
management alternative for entertainment businesses and the expansion
of Disney amusing parks (Saaty, 2009). Gómez-Navarro et al. (2010)
assessed the sustainability of tourism strategies using ANP approach.
They argued that “although the ANP procedure is not free of criticism, it
is a suitable tool for assessing sustainable tourism development strategies
for the coastal national parks of Venezuela.” Ghajar and Najafi (2012) ad-
dressed the issue of evaluation in the context of the harvesting systems in
the Caspian Forest. They concluded that “ANP provides a suitable frame-
work to incorporate all calculable and incalculable variables in the deci-
sion model and helped to achieve the goal of formulating forest policy.”
To the best of the authors' knowledge, there is no evidence in the litera-
ture regarding the application of ANP for choosing the best wood extrac-
tion method. Therefore, this paper aims to fill this gap in the research.
3. Materials and methods
3.1. Case study: the selection of a best wood extraction method in the
Caspian Forest
The Caspian Forest is located south of the Caspian Sea. This region
covers approximately 18,500 km2 and comprises 15% in the total of
the Iranian forests. Due to its humid temperate, climate and fertile soil,
the region is highly productive and made up of a wide variety of plant
communities (Jaafari et al., 2014a). Approximately 60% of the Caspian
Forest is used for timber production. Trees (alive and dead) are harvest-
ed using a single-selection or group-selection system. Given that state
controls this area, forest's management objectives are conservation, re-
habilitation, development and utilization according to SFM.
The wood extraction alternatives used in the Caspian Forest are as
follows:
i. Animal skidding
In the Caspian Forest, animal skidding is traditionally limited to
small logs, lumber, pulpwood and fuelwood from steep areas and
201
wherever road networks have not been developed, and using ma-
chinery in these areas is not practical. Animal skidding in the
Caspian Forest usually includes a group of five or six people and
seven or eight mules. Recently, with increasing environmental con-
cerns and a focus on environmentally sound timber extraction and
small-scale forest harvesting, the proportion of animal skidding
has increased in the Caspian Forest (Jourgholami, 2011).
ii. Rubber-tired skidder
The rubber-tired skidder is the most common type of skidding
machine used in the Caspian Forest. This skidder is typically an ar-
ticulated steering, four-wheel-drive vehicle weighing 10–15 t and
engine power of 110–140 kW and a maximum speed of
25–30 km/h (Sessions, 2007). It is equipped with a small, adjust-
able, push-blade on the front for the pushing of light obstacles and
the stacking of logs. The rubber-tired skidders are a flat ground sys-
tem, but with winches they can be effectively used on flat to moder-
ate slopes. Thus, the rubber-tired skidder is normally used where
the ground conditions are moderate, log size is not excessive, and
the skidding distances are long.
iii. Tracked skidder
In the Caspian Forest, tracks are usually used in both road construc-
tion and wood extraction. Although the tracked skidder is slower
and more expensive than the rubber-tired skidder, it can develop
traction and pull vary large loads. Tracks are best used where
short, steeper slopes prohibit overland rubber-tired skidding. Due
to its slower traveling speed, skidding distance is limited and
roads should be either existing or inexpensive to construct. Tracked
skidders can be used on wetter sites or in moderately inclement
weather.
iv. Cable logging
In the Caspian Forest, cable logging systems are best applied where,
due to steep sloping, ground-based systems require excavated
skid roads to operate, when harvesting is necessary in inclement
weather and where environmental factors should be taken into
account.
Each of the methods introduced above differ in terms of cost, pro-
ductivity and environmental damage. Although many studies have
been carried out on all aspects of wood extraction operations in Iran
(e.g., Mousavi, 2009; Agherkakli et al., 2010; Ezzati et al., 2011; Jaafari
et al., 2014b), forest managers still need a framework for selecting the
best option. Moreover, to continue the progress of forest management
in terms of sustainable management, decision-making within a MCDM
framework is required. To address these problems, we propose a
multi-criteria-based approach for selecting the best wood extraction
method using an ANP with BOCR.
3.2. The proposed ANP model for the case study
In this study, development of the ANP model was supported by the
latest research results reported as part of PhD dissertations focused on
timber harvesting operations in the Caspian Forest (Mousavi, 2009;
KeivanBehjou, 2010; Jourgholami, 2010; Ghajar and Najafi, 2012). The
possible criteria that were included into the ANP model were deter-
mined according to SFM (Ghajar and Najafi, 2012; ITTO, 2002) and the
FAO model code of forest harvesting practices (Dykstra and Heinrich,
1996). The final set of criteria and connections used in the ANP model
were then decided with the cooperation of the management. The ANP
model for the decision problem was developed around connections be-
tween the elements of clusters. The developed ANP model was a cou-
pling of two parts. The first part, the ANP was designed as a control
hierarchy that consists of a goal (“choosing the best wood extraction
method”) and three strategic criteria (ecological, economical and so-
cial). Each cluster included a group of indicators and management strat-
egies that were each a fully functional ANP-model within the whole
model (Fig. 1). It should be noted that it is useful to involve a conceptual
202 A. Jaafari et al. / Forest Policy and Economics 50 (2015) 200–209
Fig. 1. ANP control hierarchy with BOCR.
range of all possible indicators in the model, even if some rarely oc-
curred and may not show up as statistically significant in related models
(Niemiraa and Saaty, 2004; Ghajar and Najafi, 2012).
The second part consisted of a network of influences between the al-
ternatives and criteria clusters. This network consisted of four subnet-
works each representing the relationship of its own clusters and
elements, that is, benefits, opportunities, costs and risks (BOCR). Since
every decision has several favorable and unfavorable elements, the
decision-makers consider the favorable criteria as benefits and the unfa-
vorable criteria as costs (Azizi, 2008). In addition, any decision naturally
may comprise some uncertainties. Under uncertainty conditions, al-
though positive results may provide opportunities, negative results
may cause risk. An analysis consisting of the elements benefit, opportu-
nity, cost and risk is known as a BOCR. The BOCR is also considered as
merits of a decision (Saaty, 2006). Since each wood extraction method
offers particular benefits, opportunities, costs and risks, the model for
finding the optimal method was designed based on the BOCR model.
The BOCR model, as well as the details of the primary criteria, sub-
criteria, and alternatives are illustrated in Fig. 2.
In the benefits subnetwork, the ‘benefits’ cluster contained the ben-
efit items that were favorable for forest managers and/or stakeholders.
The ‘benefits’ cluster affected the alternative cluster and was also
impacted by each alternative.
In the costs subnetwork, the ‘outlays’ and ‘costs of environmental
protection’ clusters referred to the cost items of each extraction method.
Choosing a wood extraction method was simultaneously affected by the
‘outlays’ and ‘costs of environmental protection’ clusters.
The structure of clusters and their respective elements in the oppor-
tunities model are shown in the opportunities subnetwork, where the
‘opportunities’ cluster refers to uncertain favorable opportunities that
could be created depending on the extraction method that was chosen.
The ‘opportunities’ cluster affected the alternative cluster and was also
impacted by each alternative.
In the risks subnetwork, risk factors that occurred more often in the
forest operations were presented. The structure of this model was more
complicated than others. The ‘risks’ cluster referred to the negative ef-
fects of wood extraction on the site, while the ‘other risks’ cluster re-
ferred to risks such as equipment damage and labor injury that might
occur as a result of applying a specific extraction method; furthermore
the weather sensitivity of each method was also considered. The loop
connection indicated inner dependence within the cluster. These two
clusters affected alternative cluster and was also impacted by each
alternative.
3.3. The ANP analysis
Following setting up the ANP model and the required connections,
the proposed model's steps were performed using Super Decision soft-
ware v.2.0.8 as follows:
In the first step, real quantitative data for the quantitative criteria
(e.g., skidding productivity, skidding distance, number of employees,
and initial investment) were inserted into the model based on the find-
ings of prior studies conducted in Iran (Mousavi, 2009; KeivanBehjou,
2010; Jourgholami, 2010; Ghajar and Najafi, 2012). For the qualitative
criteria (e.g., coordination to site condition and available equipment,
weather sensitivity, environmental pollution), a pairwise comparison
questionnaire was developed with respect to the final set of criteria
for obtaining a weight for every criterion under each merit. In the
ANP, pairwise comparisons are performed at two levels, i.e., element
level comparison and the cluster level comparison. To do so, an alterna-
tive is chosen as base and pairwise comparisons among criteria are com-
pleted. This process is repeated for each alternative. In the same way, a
criterion is chosen as a base; pairwise comparisons among alternatives
are carried out until all the criteria are completed. A numerical funda-
mental scale, ranging from 1 to 9 was used to calibrate the quantitative
and qualitative performances of priorities (Table 1) (Saaty, 2003). In
 A. Jaafari et al. / Forest Policy and Economics 50 (2015) 200–209 203

Fig. 2. BOCR model.

this, a score of 1 indicated that the two options under comparison had
equal importance, while a score of 9 indicated the overwhelming
dominance of the options under consideration over the comparison
component (row component to column component). Therefore, if a
component had a weaker impact than its comparison component, the
scores would range from 1 to 1/9, where 1 had the same meaning as
mentioned above and 1/9 represented an overwhelming dominance
by a column component over the row component (Saaty, 2003).
Table 2 shows a sample of the questionnaire used for pairwise
comparison.
204 A. Jaafari et al. / Forest Policy and Economics 50 (2015) 200–209

Table 1
Scale of preference between two options.

Preference factor Degree of preference Explanation

1 Equally Two factors contribute equally to the objective

3 Moderately Experience and judgment slightly to moderately favor one factor over another
5 Strongly Experience and judgment strongly or essentially favor one factor over another
7 Very strongly A factor is strongly favored over another and its dominance is showed in practice
9 Extremely The evidence of favoring one factor over another is of the highest degree possible of an affirmation
2, 4, 6, 8 Intermediate Used to represent compromises between the preferences in weights 1, 3, 5, 7 and 9
Reciprocals Opposite Used for inverse comparison

The pairwise comparisons were then made by a panel of seven forest
engineering experts. Four of them were academics who had worked in
three different universities; two members of the panel were PhD candi-
dates, whose dissertation research focused on forest operations; and the
remaining member was the first author of this paper.
After receiving all the questionnaires, the responses were analyzed
and combined using Super Decision software. The inconsistency ratio
was used for identifying possible errors in judgments. To accept
pairwise comparisons, the inconsistency ratio had to be less than 0.10
(Saaty, 2003).
In the second step, an unweighted supermatrix was created by
inserting the obtained relative priorities of alternatives local priority
vectors, obtained from pairwise comparison, on suitable columns.
In the third step, a weighted supermatrix was computed by multi-
plying the values of the unweighted supermatrix with their related clus-
ter weights.
The fourth step of the analysis was to raise the entire supermatrix to
powers where the weights converged and remained stable, thus achiev-
ing the limit supermatrix. The final priorities for the alternatives under
each criterion could be observed when the supermatrix reached its
limit.
In the fifth step, the BOCR ratings were performed. In order to rate
the BOCR with respect to an ideal, intensity levels were created (very
high, high, medium, low, and very low) (Saaty, 2009). Following on
we took the best alternative under B and used it to consider the benefits,
as well as the best one under O (which may have been be different than
the one under B) and used it to consider the available opportunities; the
same way done for costs and risks. Finally we rated these four aspects
with respect to the strategic criteria using the appropriate intensity
levels. A score was calculated for each merit by multiplying the priority
of the selected intensity times by the priority of the criterion (global
weight) and summing for all the subcriteria (total rate). The priorities
of BOCR were obtained by normalizing the total score column through
respect to others. To convert the priorities so that less-favored alterna-
tives had lower values than more favored counterparts, the reciprocals
of each alternative's priority were calculated and presented in columns
1/Costs and 1/Risks. Then, these reciprocals were normalized as shown
in columns N1/Costs and N1/Risks (Erdoğmus et al., 2005).
The seventh and final step was to perform sensitivity analysis on the
final outcome and interpret the results of sensitivity by observing how
large or small these ratios were. In fact, sensitivity analysis allowed us
to monitor the effect by varying the influences on the stability of the
outcomes (Saaty, 2009).
4. Results
The final results of the ANP analysis are presented in three sections:
4.1. BOCR results
After comparing the degree of importance of the criteria affecting
the selection of the best wood extraction method by using the Super De-
cision software, the final relative weights of criteria were specified. At all
stages, the inconsistency ratio was less than 0.1 and all results and judg-
ments were accepted as consistent. The final relative weights of criteria
in BOCR subnetworks determined by pairwise comparisons and real
quantitative data are given in Table 3. It is seen that the most important
criteria associated with the benefits, opportunities, costs, and risks were
identified as being “maximum value of yield timber”, “coordination to
various harvesting systems”, “initial investment”, and “weather sensi-
tivity”, respectively.
Table 2
Sample of pairwise comparison questionnaire.
Questions Weight

dividing by the sum of value in it. The obtained normalized values Q1 With respect to “maximum value of yield timber” which 1
were used for the priorities of BOCR in order to conduct synthesis in alternative is more important? 3
“Rubber wheel skidder” is very strongly more important 5
the top-level network (Fig. 1).
than “animal skidding”. 7
The sixth step of the analysis was conducted to determine the rela- 9
tive importance of values for the alternatives and to identify the best 2, 4, 6, 8
one based on additive and multiplicative formulations. The first formu- Q2 With respect to “development of supporting equipment” 1
lation simply subtracted the sum of the weighted ‘bads’ from the sum of which alternative is more important? 3
“Cable crane” is strongly more important than “animal 5
the weighted ‘goods’, while the second was a tradeoff between a unit of skidding”. 7
BO against a unit of CR, a unit of the desirable against a unit of the unde- 9
sirable (Erdoğmus et al., 2005). The additive and multiplicative formula- 2, 4, 6, 8
tions can, respectively, be given as: Q3 With respect to “Preharvest inventory and mapping of 1
trees” which alternative is more important? 3
“Cable crane” is moderately more important than 5
bB þ oO–cC–rR ð1Þ “chain wheel skidder”. 7
9
n r o
b o
c
2, 4, 6, 8
B O ð1=CÞNormalized ð1=RÞNormalized ð2Þ Q4 With respect to “Environmental pollution” which 1
alternative is more important? 3
“Cable crane” is very strongly more important than 5
where B, O, C and R represent the synthesized results and b, o, c, and r “animal skidding”. 7
are the priorities for BOCR obtained from rating the strategic criteria. 9
Since the alternatives with the highest priorities under costs and 2, 4, 6, 8

risks were more costly and more risky, they were less favored with The assigned weight for each question is shown in bold.
 A. Jaafari et al. / Forest Policy and Economics 50 (2015) 200–209 205

Table 3 extremely higher than that of animal skidding.” Table 5 shows the
Final relative weights of criteria in BOCR subnetworks. weighted super matrix for the benefits subnetwork. The limit
Criteria Final relative weights supermatrix for the benefits subnetwork was then established with
Benefits
taken powers of the weighted supermatrix (Table 6). When the column
Maximum value of yield timber 0.4614 of numbers was the same for every column, the limit matrix had been
Maximum skidding distance 0.1458 reached and the matrix multiplication process is halted. The priority of
Skidding productivity 0.3260 each alternative and criterion is a dimensionless value that can be ob-
Slash distribution 0.0668
tained from the values in any of the columns of this supermatrix. The
Opportunities higher the value of the index, the better the prioritization for the alter-
Development of supporting equipment 0.0634 native and criterion. Values were normalized so that the total weight for
Coordination to various harvesting systems 0.1758
criteria was 1. Additionally the total weight for cases was 1. The Final re-
Coordination to topographic conditions 0.1668
Coordination to ecological conditions 0.1157 sults of priorities for alternatives under BOCR (Table 9) indicated that
Coordination to available equipment 0.1147 from a benefits point of view, ‘rubber-tired skidder’ will create more
Number of working days per year 0.0765 benefits than others. The ‘animal skidding’ alternative will offer the
Availability of skilled labor 0.0958 least amount of benefits.
Managing and organizing 0.1244
Number of employees 0.067
4.1.2. Opportunities subnetwork
Costs A synthesis of judgments in the opportunities subnetwork indicated
Preharvest inventory and mapping of trees 0.2591
that the ‘coordination to various harvesting systems’ criterion had the
Rehabilitation of sites of negative impacts 0.1619
Preharvest planning of skid trails 0.1364 highest relative weight (0.1758) among all the criteria (Table 3). Over-
Postharvest assessment 0.1305 all, the synthesis of judgments in the subnetwork showed that ‘cable
Preharvest vine cutting 0.0940 logging’ presented the highest positive alternative from an opportuni-
Directional felling 0.1637
ties point of view. The ‘rubber-tired skidder’ was identified as the
Initial investment 0.3137
Timber processing 0.1569 worse alternative (Table 9).
Skid trail building 0.0922
Wood extraction 0.2748 4.1.3. Costs subnetwork
Landing opening 0.0670 The results of the overall synthesis of judgments in the costs subnet-
Labors training 0.0441
work revealed the highest relative weight (0.3137) for the ‘initial in-
Labor's wage 0.1057
vestment’ criterion among all the criteria (Table 3). Additionally, ‘cable
Risks logging’ had the highest costs, while the ‘rubber-tired skidder’ had the
Environmental pollution 0.0882
smallest impact on wood extraction costs (Table 9).
Damage to equipment 0.245
Weather sensitivity 0.3865
Stand degradation 0.3086 4.1.4. Risks subnetwork
Soil degradation 0.3033 A synthesis of judgments in the risks subnetwork showed that the
Injury to labors 0.3685
‘weather sensitivity’ criterion had the highest relative weight
Runoff 0.1578
(0.3865). Additionally, the ‘rubber-tired skidder’ had the highest risk.
The highest weight is shown in bold. The least risky alternative was identified to be the ‘animal skidding’
method (Table 9).

4.1.1. Benefits subnetwork
The results of the synthesis of judgments in the benefits subnetwork
showed that the criterion ‘maximum value of yield timber’ had the
highest relative weight (0.4614) among the criteria of the benefits sub-
network (Table 3).
The obtained relative priorities of the criteria and alternatives were
then properly located into the unweighted supermatrix. Table 4 shows
the unweighted supermatrix for the benefits subnetwork. For example,
the priorities of ‘animal skidding’ and ‘rubber-tired skidder ‘with respect
to ‘skidding productivity’ are shown in the two bottom cells of the third
column, 0.0435 and 0.5876, respectively. This may be interpreted using
the statement, “The skidding productivity of rubber-tired skidder is
4.2. Rating and synthesis of the entire model
The five-step intensity level design for rating the strategic subcriteria
was very high, high, medium, low and very low. Table 7 shows the pri-
orities of intensities in ideal form which were normalized by dividing
each by the largest one. The BOCR were ranked by selecting the appro-
priate intensity for the ideal alternative from Table 7 for each strategic
subcriterion and making comparisons between them. The overall prior-
ities for the BOCR rating showed that the most important merit is ben-
efits at 48.11% followed by risks at 22.03%. This indicated that the
priorities of the alternatives under benefits were weighted more heavily
and that benefits at 48.11% drove the decision more than costs did at

Table 4
Unweighted supermatrix of the benefits subnetwork.

Advantages Alternatives

Skidding Slash Maximum skidding Maximum value Animal Cable Tracked Rubber-tired
productivity distribution distance of yield timber skidding logging skidder skidder

Advantages Skidding productivity 0 0 0 0 0.0673 0.1208 0.2163 0.5825

Slash distribution 0 0 0 0 0.1827 0.0548 0.0489 0.0632
Maximum skidding distance 0 0 0 0 0.0636 0.2570 0.0949 0.1285
Maximum value of yield timber 0 0 0 0 0.6864 0.5674 0.6399 0.2258
Alternatives Animal skidding 0.0435 0.5370 0.0671 0.0454 0 0 0 0
Cable logging 0.1139 0.0424 0.4146 0.3181 0 0 0 0
Tracked skidder 0.2550 0.2103 0.2134 0.3182 0 0 0 0
Rubber-tired skidder 0.5876 0.2103 0.3049 0.3183 0 0 0 0
206 A. Jaafari et al. / Forest Policy and Economics 50 (2015) 200–209

Table 5
Weighted supermatrix of the benefits subnetwork.

Advantages Alternatives

Skidding Slash Maximum skidding Maximum value Animal Cable Tracked Rubber-tired
productivity distribution distance of yield timber skidding logging skidder skidder

Advantages Skidding productivity 0 0 0 0 0.0673 0.1208 0.2163 0.5825

Slash distribution 0 0 0 0 0.1827 0.0548 0.0489 0.0632
Maximum skidding distance 0 0 0 0 0.0636 0.2570 0.0949 0.1285
Maximum value of yield timber 0 0 0 0 0.6864 0.5674 0.6399 0.2258
Alternatives Animal skidding 0.0435 0.6299 0.0517 0.0501 0 0 0 0
Cable logging 0.1139 0.0392 0.4278 0.3249 0 0 0 0
Tracked skidder 0.2550 0.1653 0.2599 0.3124 0 0 0 0
Rubber-tired skidder 0.5876 0.1656 0.2606 0.3126 0 0 0 0

Table 6
Limit matrix of the benefits subnetwork.

Advantages Alternatives

Skidding Slash Maximum skidding Maximum value Animal Cable Tracked Rubber-tired
productivity distribution distance of yield timber skidding logging skidder skidder

Advantages Skidding productivity 0.1630 0.1630 0.1630 0.1630 0.1630 0.1630 0.1630 0.1630
Slash distribution 0.0334 0.0334 0.0334 0.0334 0.0334 0.0334 0.0334 0.0334
Maximum skidding distance 0.0729 0.0729 0.0729 0.0729 0.0729 0.0729 0.0729 0.0729
Maximum value of yield timber 0.2307 0.2307 0.2307 0.2307 0.2307 0.2307 0.2307 0.2307
Alternatives Animal skidding 0.0405 0.0405 0.0405 0.0405 0.0405 0.0405 0.0405 0.0405
Cable logging 0.1236 0.1236 0.1236 0.1236 0.1236 0.1236 0.1236 0.1236
Tracked skidder 0.1375 0.1375 0.1375 0.1375 0.1375 0.1375 0.1375 0.1375
Rubber-tired skidder 0.1984 0.1984 0.1984 0.1984 0.1984 0.1984 0.1984 0.1984

13.62% (Table 8). The final results shown in Table 9 were obtained by
using the two formulations (multiplicative and additive). The ‘rubber-
tired skidder’ was identified as the best alternative with the highest
value, while the ‘animal skidding’ was the worst with the lowest
value, according to both formulations.
4.3. Sensitivity analysis
Since significant changes could not be obtained using the multiplica-
tive formulation, the additive formulation was used to analyze the sen-
sitivity of the alternatives to the weights of BOCR.
logging’ except in a little emphasis on costs. The next costly alternative
was ‘animal skidding’ followed by ‘tracked skidder’.
4.3.4. Risks
The sensitivity analysis for the risks subnetwork implied that the
ranking of the alternatives was sensitive to changes in the weight of
the risks and changes from rubber-tired skidder–tracked skidder–
cable logging–animal (for 0% risk weight) to a complete reverse order
(for 50% risk weight) (Fig. 6).
5. Discussion

4.3.1. Benefits
The benefit sensitivity analysis is shown in Fig. 3 and indicates that
the greatest benefit was achieved through the ‘rubber-tired skidder’.
While the weight of the benefits increased, it became clear that the pri-
ority of the ‘rubber-tired skidder’ increased and the other alternatives,
especially ‘animal skidding’, decreased.
4.3.2. Opportunities
When the opportunity graphic (Fig. 4) was analyzed, the priorities of
alternatives were rubber-tired skidder–tracked skidder–cable logging–
animal skidding for 0% opportunities weight. However, ‘cable logging’
reached first rank when weight arrived at 45% values. While the weight
of the opportunities increased, the alternatives moved toward conver-
gence, whereas the above-mentioned order remained constant.
4.3.3. Costs
When the sensitivity of costs was performed, the rank of ‘cable log-
ging’ changed from second to fourth due to an increase weight of costs
(Fig. 5). In fact, the highest costs will be presented through ‘cable
Table 7
Intensity levels in ideal form for rating BOCR.
Categories Very high High Medium Low
Values 1.000 0.5099 0.2515 0.1235
In the Caspian Forest wood extraction using skidders and animals
are the default systems except when: 1) logging is required in inclement
weather; 2) skidding distances are longer than 500 m due to the cost of
road construction; and 3) a dozed road is necessary for the skidder be-
cause the slope is excessive. Under these conditions, other methods, es-
pecially cable logging, should be considered. Without considering the
sensitivity of the results in terms of the control criteria, it can be stated
that the ‘rubber-tired skidder’ is the most suitable method for wood ex-
traction. However, the sensitivity analysis indicated that the priorities of
alternatives changed when the weight of each of the BOCR varied. In
fact, emphasis on each merit reflected forest management consider-
ations that should be applied by the managers who design model and
make decisions about harvesting operations and wood extraction.
When sensitivity analysis was realized according to the weight of bene-
fits, the ‘rubber-tired skidder’ was considered as the best alternative, ex-
cept in case when very low importance was given to benefits. This
means that if forest management is based on the most advantageous
wood extraction operations, this method will provide the maximum
benefits for each utilization unit (Ghajar and Najafi, 2012).
According to the results of the sensitivity analysis for opportunities,
‘cable logging’ offers the best opportunities among the alternatives.
Since cable logging is feasibly attuned to various harvesting systems,
Very low such as whole-tree harvesting, and to the topographic conditions of
mountainous areas such as the Caspian Forest, the greatest opportuni-
0.0650
ties will always be achieved through this method.
 A. Jaafari et al. / Forest Policy and Economics 50 (2015) 200–209 207

Table 8
BOCR rating and priorities.

Criteria Ecological (0.5499) Economic (0.2403) Social (0.2098) Total Priorities

rate

Subcriteria Minimizing Minimizing Minimizing Sustainable Maximizing Maximizing Minimizing Preference of Improving Maximizing
damage to residual soil production revenue production costs stakeholders livelihood number of
regeneration damage disturbance employees

Weight 0.1365 0.6250 0.2385 0.6165 0.1013 0.1498 0.1324 0.6250 0.1365 0.2385 – –
Global weight 0.0751 0.3437 0.1311 0.1481 0.0243 0.0360 0.0318 0.1311 0.0286 0.0500 – –
Benefits Medium Very high Very high High Very high High High Very high High Very high 0.8239 0.4811
Opportunities Very low Medium Low High High Medium Medium Medium Medium High 0.2782 0.1624
Costs Low Medium Low Medium Medium Medium High Medium Medium Medium 0.2333 0.1362
Risks Medium High Medium Low High Low Medium High High High 0.3772 0.2203

Table 9
Final priorities for alternatives under BOCR.

Alternatives Benefits Opportunities Costs Risks 1/Costs N1/Costs 1/Risks N1/Risks Final outcome using Final outcome using
additive formula multiplicative formula

Animal skidding 0.0808 0.2299 0.2851 0.0981 3.5075 0.2032 10.1937 0.4874 0.1848 0.1596
Cable logging 0.2472 0.3069 0.3459 0.2077 2.8910 0.1675 4.8146 0.2302 0.2544 0.2474
Tracked skidder 0.2751 0.2453 0.1925 0.2928 5.1948 0.301 3.4153 0.1633 0.2623 0.2910
Rubber-tired skidder 0.3969 0.2179 0.1765 0.4014 5.6657 0.3283 2.4913 0.1191 0.2985 0.3020
Total 17.2591 20.9149 1.0000

Best alternative under each merit is shown in bold.

According to the results of the sensitivity analysis for costs, while the
importance of costs increased past roughly 65%, ‘cable logging’ would be
the most costly option. This can be interpreted to mean that if forest
managers and/or stakeholders consider costs as the most important cri-
terion in making decisions or have limited financial resources, they will
not apply ‘cable logging’ method. Meanwhile, ‘animal skidding’ due to
its low productivity and the value of its timber yield, being limited to
small sizes and small-scale operations (Jourgholami, 2011), the difficul-
ties of animal maintenance and the method's low efficiency compared
to modern production systems supported by mechanical power, was
also considered as a costly wood extraction method. Picchio et al.
(2009) also reported that animal skidding had a higher cost than
machinery skidding. However, according to the results of the sensitivity
analysis for risks, ‘animal skidding’ was ranked as the least risky option
followed by ‘cable logging’, which is also already known to cause less
disturbance than skidders (Spinelli et al., 2010). Considering the envi-
ronmental risks of rubber-tired and tracked skidders (Ampoorter
et al., 2007; Zenner and Berger, 2008; Spinelli et al., 2010; Picchio
et al., 2012; Jaafari et al., 2014b), it is not surprising that the priority of
the two skidders decreased when the weight of the risks increased.
When introducing a wood extraction method, in addition to eco-
nomic issues, the potential damage to the environment caused by the
method should be taken into account. In other words, improper selec-
tion of an extraction method usually results in increased harvesting
costs and/or undesirable environmental impacts. Although the ‘rub-
ber-tired skidder’ and ‘tracked skidder’ were identified as the first and
second most appropriate (i.e., profitable) wood extraction methods
within the conditions of this study, the methods have high potentials

Fig. 3. Sensitivity analysis of benefits. Fig. 4. Sensitivity analysis of opportunities.

208 A. Jaafari et al. / Forest Policy and Economics 50 (2015) 200–209
Fig. 5. Sensitivity analysis of costs.
for causing unfavorable impacts on the forest environment. Although it
is not possible to completely avoid negative impacts in ground-based
timber skidding, impacts can be limited to an ecologically acceptable
and economically tolerable level by the application of favorable techni-
cal practices and technological solutions that minimize skidding-
induced damage (Susnjar et al., 2006). Several technical solutions can
be used to reduce the risks of unfavorable impacts, such as improving
winching technique (Picchio et al., 2012), decreasing the skid trail
slope in the design phase (Jaafari et al., 2014b), and employing
reduced-impact logging practices such as proper planning and layout,
directional felling, preharvest vine cutting, training of labors, and cor-
rect management and supervision (Dykstra, 2002; Mousavi, 2009).
Fig. 6. Sensitivity analysis of risks.
Additionally, damage might be avoided through the application of
more careful logging procedures and applying low-impact logging
methods, such as animal skidding, in combination with other methods.
Such approaches can promote techniques for both conserving and uti-
lizing the Caspian Forest in a more sustainable manner. In our study,
the ‘cable logging’ alternative was ranked after the two skidders. Since
cable logging's initial investment cost is very high, the system should
be run at a best performance level. Although there may be many reasons
for a system's performance to be affected (Sessions, 2007), one of the
important aspects that need planning is appropriate scheduling opera-
tions. Moreover, working with a maximum payload and increasing har-
vested volume can help reduce the total cost per cubic meter
(Ghaffariyan et al., 2010).
Although the model developed in this study contained a large num-
ber of criteria and involved the completion of very long questionnaires,
it improved the panel of experts' insight into the topic and forced them
to think about the weights of the criteria involved, and to analyze the
different available alternatives according to the objectives of SFM
(Samari et al., 2012). Moreover, incorporating all the effective factors
in the model was imperative in order to make a comprehensive decision
and to achieve the strategic goals of SFM (Ghajar and Najafi, 2012). All
the forest engineering experts felt that the ANP procedure had allowed
them to deal with prioritization in an organized and systematic manner.
They all agreed that the procedure was suitable for taking into account
criteria that have not been considered in conventional forest manage-
ment planning methods, but which should today be considered in near-
ly any forestry decision-making process. However, only two out of
seven experts considered the ANP procedure satisfactory and easy to
understand. However, although the model was thought to be too com-
plicated for forest management purposes, such complex models may
provide a basis for building and consolidating SFM strategies in the de-
veloping world.
6. Conclusions
Modern resource management involves the consideration of a wide
variety of factors and often seeks to optimize the attainment of multiple
objectives. The field of forest management has opened up an avenue for
an analysis not previously investigated in the literature. In this study, we
combined participatory data collected in the Caspian Forest with a
decision-making tool to develop a model that not only considered
local conditions in order to evaluate wood extraction methods, but
also to choose the most appropriate one in terms of fit benefits, costs,
opportunities and risks generated by the different situations involved
in these methods. Forest engineering experts who completed the survey
perceived that the benefits and risks associated with wood extraction
methods in the Caspian Forest are as more important than opportunities
and costs. This suggests that these aspects of wood extraction methods
may need to be addressed by managers when choosing a method. This
paper also indicates that through the use of ANP, the process of
decision-making is automatically documented and thus, transparent
decisions regarding SFM can be completed. Given the capability of
ANP in dealing with different types of feedback and dependencies and
feedback when modeling a decision-making environment, we believe
that our results are more reliable and therefore can be considered as a
basis for future harvest-scheduling efforts in order to meet SFM strate-
gies. Although the model has been adjusted for the Caspian Forest, the
key principles applied here can be used or developed in other countries
or regions to identify potential improvements in harvesting operations.
Future research may seek to evaluate the capability of other tech-
niques like TOPSIS, ELECTRE, VIKOR and their combinations. In addition,
fuzzy numbers can be introduced in the methods to more effectively an-
alyze cases with greater uncertainty in pairwise comparison matrices.
The comparison of the results obtained from other studies will help cre-
ate a better understanding of the criteria involved in the selection
process.
 A. Jaafari et al. / Forest Policy and Economics 50 (2015) 200–209 209

Acknowledgment Kangas, J., Kangas, A., 2005. Multiple criteria decision support in forest management — the
approach-methods applied, and experiences gained. For. Ecol. Manag. 207, 133–143.
KeivanBehjou, F., 2010. Economical Evaluation of Harvesting Operations Disturbance to
We would like to express our sincere appreciation to the experts that Forest(PhD dissertation) Faculty of Natural Resources, University of Tehran.
graciously accepted to participate in this study. We also thank Dr. Ismael Kühmaier, M., Stampfer, K., 2012. Development of a multi-criteria decision support tool
for energy wood supply management. Croat. J. For. Eng. 33 (2), 181–198.
Ghajar and Dr. Stefano Grigolato for their supportive comments and re- Laukkanen, S., Palander, T., Kangas, J., Kangas, A., 2005. Evaluation of the multicriteria ap-
views in early version of this paper. proval method for timber-harvesting group decision support. Silva Fennica 39 (2),
249–264.
Lu, J., Zhang, G., Ruan, D., Wu, F., 2007. Multi-objective Group Decision Making: Methods,
References Software, and Applications With Fuzzy Set Techniques. Imperial College Press.
Miettinen, K., 2006. International society on multiple criteria decision making mission.
Agherkakli, B., Najafi, A., Sadeghi, S.H., 2010. Ground based operation effects on soil dis-
Webpage http://project.hkkk.fi/MCDM/intro.html (27 September 2006).
turbance by steel tracked skidder in a steep slope of forest. J. For. Sci. 56 (6), 278–284.
Mousavi, R., 2009. Comparison of Productivity, Cost and Environmental Impacts of
Ampoorter, E., Goris, R., Cornelis, W.M., Verheyen, K., 2007. Impact of mechanized logging
Two Harvesting Methods in Northern Iran: Short-log vs. Long-log(PhD dissertation)
on compaction status of sandy forest soils. For. Ecol. Manag. 241 (1), 162–174.
Faculty of Forest Sciences, University of Joensuu, (Available at http://www.metla.fi/
Ananda, J., Herath, G., 2005. Evaluating public risk preferences in forest land use choices
dissertationes/df82.htm).
using multi-attribute utility theory. Ecol. Econ. 55, 408–419.
Niemiraa, M.P., Saaty, T.L., 2004. An analytic network process model for financial-crisis
Ananda, J., Herath, G., 2008. Multi-attribute preference modeling and regional land use
forecasting. Int. J. Forecast. 20, 573–587.
planning. Ecol. Econ. 65, 325–335.
Pellegrini, M., Grigolato, S., Cavalli, R., 2013. Spatial multi-criteria decision process to de-
Azizi, M., 2008. A model of supplying poplar wood for Iranian paper & wood facto-
fine maintenance priorities of forest road network: an application in the Italian alpine
ries. J. For. Res. 19 (4), 323–328.
region. Croat. J. For. Eng. 34, 31–42.
Dykstra, D.P., 2002. Reduced Impact Logging: Concepts and Issues. In: Enters, T., Durst, P.
Picchio, R., Maesano, M., Savelli, S., Marchi, E., 2009. Productivity and energy balance in
B., Applegate, G.B., Kho, P.C.S., Man, G. (Eds.), Applying Reduced Impact Logging to
conversion of a Quercus cerris L. coppice stand into high forest in Central Italy.
Advance Sustainable Forest Management. FAO, Bangkok, Thailand, pp. 23–39.
Croat. J. For. Eng. 30 (1), 15–26.
Dykstra, D.P., Heinrich, R., 1996. FAO Model Code of Forest Harvesting Practice. FAO,
Picchio, R., Magagnotti, N., Sirna, A., Spinelli, R., 2012. Improved winching technique to re-
Rome, Italy.
duce logging damage. Ecol. Eng. 47, 83–86.
Erdoğmus, S., Kapanoglu, M., Koc, E., 2005. Evaluating high-tech alternatives by using an-
Pukkala, T., 2002. Multi-objective Forest Planning: Managing Forest Ecosystems. Kluwer
alytic network process with BOCR and multiactors. Eval. Program. Plann. 28 (4),
Academic Publishers, Dordrecht.
391–399.
Saaty, T.L., 1999. Fundamentals of the Analytic Network Process. Paper presented at the
Ezzati, S., Najafi, A., Durston, T., 2011. Impact of animal logging on soil physical properties
ISAHP 1999, Kobe, Japan.
in mule trail in Hyrcanian forest. Transp. Res. Part D: Transp. Environ. 16, 316–320.
Saaty, T.L., 2001. The Analytic Network Process. RWS Publications, Pittsburgh.
García-Melón, M., Ferrís-Oñate, J., Aznar-Bellver, J., Aragonés-Beltrán, P., Poveda-Bautista,
Saaty, R.W., 2003. Decision making in complex environments: the analytic hierarchy
R., 2008. Farmland appraisal based on the analytic network process. J. Glob. Optim. 42
process (AHP) for decision making and the analytic network process (ANP) for
(2), 143–155.
decision making with dependence and feedbackRetrieved February 18, 2005 from
Ghaffariyan, M.R., Stampfer, K., Sessions, J., 2010. Optimal road spacing of cable yarding
http://www.superdecisions.com/demos_tutorials.php3.
using a tower yarder in Southern Austria. Eur. J. For. Res. 129, 409–416.
Saaty, T.L., 2006. Decision Making with the Analytic Network Process-economic, Political,
Ghajar, I., Najafi, A., 2012. Evaluation of harvesting methods for Sustainable Forest Man-
Social and Technological Applications With Benefits, Opportunities, Costs and Risks.
agement (SFM) using the Analytical Network Process (ANP). For. Policy Econ. 21,
Springer, New York.
81–91.
Saaty, T.L., 2009. Applications of analytic network process in entertainment. Iran. J. Oper.
Gómez-Navarro, T., García-Melón, M., Acuña-Dutra, S., 2010. An ANP approach to assess
Res. 1 (2), 41–55.
the sustainability of tourist strategies for the coastal national parks of Venezuela.
Samari, D., Azadi, H., Zarafshani, K., Hosseininia, Gh., Witlox, F., 2012. Determining appro-
Technol. Econ. Dev. Econ. 16 (4), 672–689.
priate forestry extension model: application of AHP in the Zagros area, Iran. For.
Hayati, E., Majnounian, B., Abdi, E., Sessions, J., Makhdoum, M., 2013. An expert-based ap-
Policy Econ. 25, 91–97.
proach to forest road network planning by combining Delphi and spatial multi-
Sessions, J., 2007. Harvesting Operations in the Tropics. Springer, New York.
criteria evaluation. Environ. Monit. Assess. 185 (2), 1767–1776.
Spinelli, R., Magagnotti, N., Nati, C., 2010. Benchmarking the impact of traditional small-
International Tropical Timber Organization (ITTO), 2002. ITTO Guidelines for the Restora-
scale logging systems used in Mediterranean forestry. For. Ecol. Manag. 260 (11),
tion, Management, and Rehabilitation of Degraded and Secondary Tropical Forests.
1997–2001.
No 13, (86pp.).
Susnjar, M., Horvat, D., Seselj, J., 2006. Soil compaction in timber skidding in winter con-
Jaafari, A., Najafi, A., Pourghasemi, H.R., Rezaeian, J., Sattarian, A., 2014a. GIS-based fre-
dition. Croat. J. For. Eng. 27, 1–14.
quency ratio and index of entropy models for landslide susceptibility assessment in
Wolfslehner, B., Vacik, H., Lexer, M.J., 2005. Application of the analytic network process in
the Caspian forest, northern Iran. Int. J. Environ. Sci. Technol. 11 (4), 909–926.
multi-criteria analysis of sustainable forest management. For. Ecol. Manag. 207,
Jaafari, A., Najafi, A., Zenner, E.K., 2014b. Ground-based skidder traffic changes chemical
157–170.
soil properties in a mountainous Oriental beech (Fagus orientalis Lipsky) forest in
Zenner, E.K., Berger, A.L., 2008. Influence of skidder traffic and canopy removal intensities
Iran. J. Terrramech. 55, 39–46.
on the ground flora in a clearcut-with-reserves northern hardwood stand in Minne-
Jourgholami, M., 2010. Comparison of Three Forest Harvesting Systems in Order to Sus-
sota, USA. For. Ecol. Manag. 256 (10), 1785–1794.
tainable Harvesting Management(PhD dissertation) Faculty of Natural Resources,
University of Tehran.
Jourgholami, M., 2011. Small-scale timber harvesting: mule logging in Hyrcanian forest.
Small Scale For. 11 (2), 255–262.