Identifying a preservation zone using multi–criteria decision analysis

Identifying a preservation zone using multi–criteria decision analysis.— Zoning of a protected area is an approach to partition landscape into various land use units. The management of these landscape units can reduce conflicts caused by human activities. Tandoreh National Park is one of the most biologically diverse, protected areas in Iran. Although the area is generally designed to protect biodiversity, there are many conflicts between biodiversity conservation and human activities. For instance, the area is highly controversial and has been considered as an impediment to local economic development, such as tourism, grazing, road construction, and cultivation. In order to reduce human conflicts with biodiversity conservation in Tandoreh National Park, safe zones need to be established and human activities need to be moved out of the zones. In this study we used a systematic methodology to integrate a participatory process with Geographic Information Systems (GIS) using a multi–criteria decision analysis (MCDA) technique to guide a zoning scheme for the Tandoreh National Park, Iran. Our results show that the northern and eastern parts of the Tandoreh National Park that were close to rural areas and farmlands returned less desirability for selection as a preservation area. Rocky Mountains were the most important and most destructed areas and abandoned plains were the least important criteria for preservation in the area. Furthermore, the results reveal that the land properties were considered to be important for protection based on the obtained preservation zone. However, these parts are not fully covered under the current protection plans for the area.


Introduction
The zoning of protected area is an approach to reduce conflict by designating areas into different management and land use units (Hjortsø et al., 2006;Geneletti & van Duren, 2008;Zhang et al., 2013).Identification and delineation of management zones is necessary for effective management of protected areas.Detailed strategies and activities for different zones can only be defined after management zones are delineated.Multiple land characteristics can be evaluated using zoning, a complex decision-making process (Zhang et al., 2013).Typical zoning sche� mes include a preservation zone with a high level of protection.This kind of zoning can largely exclude human activities surrounded by zones that allow for increasing levels of human activities.Most protected areas in developing countries are suffering from the lack of zoning (Sabatini et al., 2007;Hull et al., 2011), which has now become a challenge for governments and land managers.
The evaluation of multiple land attributes based on multiple objectives that inherently involve conflicts is necessary for decision-making regarding land use zoning.Multi-criteria decision analysis (MCDA) has been used to support complex decision making con� strained by multiple conflicting objectives and criteria (Massam, 1988).The field of multi-criteria decision aiding (MCDA) has been developed since the 1960s.Methodological work focused on discreet methods has been carried out by Roy (Roy & Vincke, 1981;Roy, 1985Roy, , 1991) ) who took the lead in using multi-criteria assessment with the ELECTRE family of methods.The PROMETHEE method has been created by Brans (Brans et al., 1986).A REGIME method has been developed by Hinloopen & Nijkamp (1990), while the DEFINITE package has been developed by Janssen (Janssen, 1993).The NAIADE method has been de� veloped by Munda (1995, 2008). Figueira et al. (2005) presented a survey of multi-criteria analysis methods.MCDA has the potential to be applied to a range of regional issues, such as.industrial development, waste management, and renewable energy.Moreover, the issues of sustainability assessment on the macro scale have been analyzed using MCDA methods.Guitouni & Martel (1998) offered an extensive survey of MCDA methods.Furthermore, a review of several MCDA sustainability applications was undertaken by De Montis et al. (2004).A good overview of existing approaches to multi-criteria evaluation of biodiversity in conservation planning has been done by Moffet & Sarkar (2008).The classical goal of finding an optimal solution is subject to a set of constraints that are characteristic of operations research, differing from the new paradigm in MCDA.The primary purpose of analysis in the MCDA paradigm is to search for a compromise solution that satisfies the decision maker, rather than some illusory optimum (Guitouni & Martel, 1998;Shmelev, 2012).
Development and implementation of zoning meth� odology for protected areas is a critical strategy to enhance the appropriate conservation system.The lack of zoning in protected areas in Iran can stop many conservations activities, which may cause ir� reversible damage to local biodiversity.Therefore, a practical quantitative method needs to be developed to design zoning for protected areas in Iran.This zoning is easy to implement and is transferable to various national parks and protected areas.
This study aimed to evaluate the utility of MCDA for zoning the Tandoreh National Park in Iran.Specifically, we tried to illustrate how an MCDA framework can identify the preservation zones of a protected area.

Study area
Tandoreh National Park is located in north-east Iran (58º 33' -58º 54' N and 37º 19' -37º 33' E) near the Turkmenistan border.It covers approximately 44,848 ha and includes the national park and protected area (fig.1).In view of the lack of river flows, springs are the main water resource in this area.Tandoreh National Park also has some waterways and streams, but there is no continuous flow in its stream bed all year round due to lack of rain and snow falls in the area.The park has a wide diversity of plants, encompassing 373 spe� cies from 60 families.The study area includes part of the highlands, hills and mountains of north Khorasan province.The existence of high mountains and deep valleys in Tandore National Park creates good habitats for mountainous wildlife.Five of seven feline species in Iran live in this area, such as the Persian leopard, Panther pardus saxicolor, the jungle cat, Felis chaus, the Pallas' cat, Felis manual, and the Eurasian lynx, Lynx lynx.Tandore National Park is one of the best habitats for Persian leopards in Iran.A total of 134 leopards were observed in the area in year 1991 and 60 in 2008.From the wild ungulates in the area, special mention is given to wild goats that live in herds of up to 100-150 in the highlands and wild sheep, Ovis orientalis arkali, that are the purest race of this species in Iran.Wolves can be observed in groups of two or five in the lowlands.The other mammals in this park are red fox, jackal, beech marten, hyena, and rodents such as pica.Birds can be seen in the lowlands, valleys, and near springs.Gypaetus barbatus, Eurasian griffon and Pheasanidae are considered endangered species in the park.
Tandoreh National Park is a good example of tradeoffs between achieving biodiversity and human activities.The main disturbance in the area is gra� zing, which has a high negative effect on vegetation.The area was used as a pastureland before it was designated as a national park.We still can see gra� zing in the park, but with less intensity.The other disturbances in the national parks are rural areas and roads that lead into or around the park.In this study we used three ways to show the conflict between human development and conservation goals: (1) the areas under high grazing were entered into the model as unsuitable areas for protection; (2) rural areas, roads and a 2 km buffer area around them have been removed from the model; and (3) rural areas and roads were used in the models as indexes with negative effects on the safe zone.

Defining the criteria
Semi-structured interviews and participatory meetings were designed to elicit knowledge, point of view, un� derstandings, interpretations, and experiences of diffe� rent stakeholders and academic experts in relation to the preservation zone.Criteria were identified according to the interviews and meetings.Table 1 shows criteria impacting the preservation zone.Although endangered species are included as important targets, the data for this kind of species are limited.This is especially true for the animals for whom knowledge of their their population size and distribution is poor.Ecologists and conservationists believe the habitat of endangered species is an essential criteria for survival.Therefore, in this study we used the map of habitat type rather than the distribution map of endangered species.

Selecting criteria
We used the Delphi method to select the correct criteria.Delphi is a systematic and interactive method which relies on a panel of independent experts (Ye et al., 2006;Guo, 2007).It is based on the principle that a structured group of experts achieve more accurate forecasts than unstructured groups or individuals (Rowe & Wright, 2001).The carefully selected experts an� swered questionnaires for criteria selection to evaluate preservation zone in three steps.After each step, the summaries of the experts' selection from the previous round and the reasons they provided for their judgments were fed back to the experts.The range of selected criteria decreased during this process and the group converged towards the 'appropriate' criteria.Finally, the process was ended after pre-defined stop criteria (e.g., number of rounds, achievement of consensus, and stability of results).

Weighting criteria
To obtain the important weights for each criterion, we used the analytic network process (ANP) approach to rank the criteria with respect to the objective.Many studies have used this method to determine the weight of a criterion (Mohanty et al., 2005;Ramik, 2006;Dag� deviren et al., 2008;Aznar et al., 2010;Catron et al., 2013;Tavana et al., 2013;Yeh & Huang, 2014;Tadić et al., 2014).The ANP method is an improved version of the AHP method and it is more accurate for many complicated models using many criteria feedback and interrelations among criteria.It evaluates all relation� ships systematically by adding potential interactions, interdependences, and feedbacks in the decision-ma� king system.The powerful side of this method is that it approaches a decision-making problem involving many complicated relationships in a simple way.This technique not only enables pair-wise comparisons of the sub-criteria under main criteria, but also allows  the decision-maker to independently compare all the sub-criteria within interactions.Figure 2 shows a com� parison of AHP and ANP methods.Decision-making problems that occur in firms cannot be explained by hierarchical structure alone.The criteria and alterna� tives in a problem can lie in interactions.Under such circumstances, a complicated analysis would be neces� sary to determine the weights of all the components.The ANP method is used for such problems and it is based on the same pair-wise comparisons as the AHP (Sevkli et al., 2012).For pair-wise comparisons, the 1-9 scale of Saaty (1980) is used as shown in Saaty (2008, table 1).In the ANP model, all the components and relationships are defined and the relationships are determined as two-way interactions.In the model, the network structure is used and all the relationships in a cluster (namely, relationships among sub-criteria in a cluster and relationships between sub-criteria under different clusters) are considered.Because of the in� volvement of relationships among sub-criteria under a cluster and interactions among different criteria, the ANP method is useful to obtain more accurate and more effective results such as those in a complex and crucial decision-making problem.The ANP method has three Tabla 1. Criteria and their weights for preservation.

Feedback
Cluster Elements XXX matrix analyses: super matrix, weighted super matrix, and limit matrix.The super matrix provides relative importance of all the components and the weighted super matrix is used to determine the value that is obtained by the super matrix values and the value of each cluster.In the limit matrix, the constant values of each value are determined by taking the necessary limit of the weighted super matrix.The results of the decision-making problem are gained from the limit matrix scores.It is important to value the criteria and alternatives of the experts and experienced people to achieve more consistent and reliable results (Saaty, 1999(Saaty, , 2003)).In our study, the relative importance weights were calculated from freely available software for academic purposes known as Super Decisions Software (http://www.superdecisions.com).

Determining preservation zone
A suitability evaluation using the GIS-based MCDA has been developed as a tool to support decision-ma� king systems for management policies and strategies (Malczewski & Jackson, 2000;Geertman & Stillwell, 2004;Malczewski, 2004;Gerber et al., 2008).Infor� mation about several criteria is combined by MCDA to form a single index of evaluation.To combine con� tinuous factors by applying a weight to each factor, a linear combination is used followed by a summation of the results to yield a suitability map (Malczewski, 2000;Eastman, 2001).
where S is the suitability, w i the weight of factor i, n the number of factors, x i the criterion score of factor i in continued range.Since the scales on which criteria are measured are different, standardization must be performed for all factors before combining them using Eq (1).Moreover, if necessary, all factors must be transformed so that they are positively or negatively correlated with suitability.In this study, standardization was performed using a GIS fuzzy set membership function on to a 0-255-byte scale through the IDRISI program (Eastman, 2001) with 0 as the lowest and 255 as the highest suitability.Herein, we categorized the continuous suitability map into suitable and unsuitable classes.

Results
Table 1 shows the list of selected criteria.The criteria were divided into three groups: (1) sources of distur� sources of distur� sources of distur� bance, (2) slope, altitude and spring, and (3) species habitat.Species habitat was the most important criteria for preservation (table 1).Rocky Mountain, meadow, 3 J Fig. 2. A high-level comparison of AHP and ANP (W 1 is a vector that represents the impact of the goal, namely, selecting the best strategy according to SWOT factors, W 2 is a matrix that represents the inner dependence of the SWOT factors, W 3 is a matrix that denotes the impact of the SWOT factor on each of the SWOT sub-factors, and W 4 is a matrix that denotes the impact of the SWOT sub-factors on each of the strategies).Strategies Strategies tall shrubs, forest and mixed forest were the second most important, and sources of disturbance such as rural area, road, and agriculture land and destroyed and abandoned plains were the third most important criteria for preservation.
A suitability map was created to show the areas with the highest priority for preservation based on the weighted criteria (table 1).Generally, the area with high suitability values for preservation was found at elevations above 1,300 m (fig.3).The highest suita� bility values were observed in the south and central parts of the park.In contrast, the lowest suitability values were found at lower elevations along the north and east boundaries of the park, which is close to rural areas (fig.3).

Discussion and conclusion
Tandoreh National Park is one of the most biologi� cally and culturally diverse protected areas in Iran.The area was designed for biodiversity conservation; however human activates effect the ecosystem of the park in a negative way.In other words, there is a huge conflict between conservation goal, and local economic development such as tourism, road cons� truction, cultivation, and grazing.Therefore, it is highly demanded to determine a safe zone for biodiversity conservation and ensuring that human activities are located outside the zone.Our results showed that the eastern and northern areas of the park, which are close to human activities such as villages and agri� cultural land, returned the lowest suitability.Because Tandoreh is one of the best habitats for the Persian panther, Panther pardus saxicolor (Ziaee, 2009) the park managers put a lot of effort into protecting their habitats.We used the habitats of panthers as an important factor to define the safe zone.As shown in figure 3, the habitats of this species are located inside the safe zone, indicating that the protection of these animals depends on protection of safe zone.
The results revealed the importance of land prop� erties for protection.However, these areas are not fully covered by the current protection plan (fig.3).As can be observed from figure 3, the new zoning is completely different from the old zoning.The area which was considered as safe zones in the old zon� ing retrieved lower protection priority in our results.However, this result was expected because the sys� tematic methodology did not play any role in the old zoning, and it was based on expert knowledge alone.The availability of habitat maps can be a solid base for the suitability analysis of the zone.In this sense we could make use of a good quality dataset and the experience of local and thematic experts, which has a very strong effect in the analyses.Future improve� ments will likely require more accurate distribution maps for individual speciesand assessment of the fragility and sensitivity of the different habitat types.
The IUCN recommends that the primary manage� ment objectives in a protected area are to preserve its natural ecosystems and species and their associated habitat in at least 75% of the land or water bodies therein (called 75% rule) (Dudley, 2008).Moreover, Dudley (2008) mentioned that 'hard' zones can be assigned to an IUCN category when they are clearly mapped, recognized by legal or other effective means, and have distinct and unambiguous management aims that can be assigned to a particular protected area category (the 75% rule is not relevant).In this study, about 42% of land in the national park is protected areas based on the 'preservation zone'.We presen�   ted a systematic method that combines participatory planning with a GIS-based MCDA technique to ob� jectively design land use zones in the protected area.The zoning of protected areas that considers both socioeconomic and biodiversity factors has moved to the forefront of conservation planning (Stewart & Possingham, 2005;Klein et al., 2008).Here we have described a method to evaluate zoning plans that shows the tradeoffs between biodiversity conservation goals and human developments.Tradeoffs between conservation and socioeconomic interests must be considered in any planning process in order to ade� quately conserve ecosystems.

Fig. 1 .
Fig. 1.Land cover map (left) and digital elevation map (right) of the study area.

Fig. 2 .
Fig.2.Comparación de alto nivel entre los procesos analítico jerárquico (AHP) y analítico en red (ANP) (donde W 1 es un vector que representa los efectos del objetivo, esto es, seleccionar la mejor estrategia de acuerdo con los factores de DAFO, W 2 es una matriz que representa la dependencia interna de los factores de DAFO, W 3 es una matriz que indica los efectos del factor de DAFO en cada uno de los subfactores de DAFO y W 4 es una matriz que indica los efectos de los subfactores de DAFO en cada una de las estrategias).

Fig. 3 .
Fig. 3. Suitability map for preservation in the Tandoreh National Park.