Using AHP and PROMETHEE multi-criteria decision making methods to define suitable apiary locations

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tjar20

Journal of Apicultural Research

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tjar20

Using AHP and PROMETHEE multi-criteria decision

making methods to define suitable apiary

locations

Fatih Sari , İrfan Kandemir , Durmuş Ali Ceylan & Aziz Gül

To cite this article: Fatih Sari , İrfan Kandemir , Durmuş Ali Ceylan & Aziz Gül (2020) Using AHP and PROMETHEE multi-criteria decision making methods to define suitable apiary locations, Journal of Apicultural Research, 59:4, 546-557, DOI: 10.1080/00218839.2020.1718341

To link to this article: https://doi.org/10.1080/00218839.2020.1718341

Published online: 04 Feb 2020.

Submit your article to this journal

Article views: 217

View related articles

View Crossmark data

ORIGINAL RESEARCH ARTICLE

Using AHP and PROMETHEE multi-criteria decision making methods to define

suitable apiary locations

Fatih Saria
, _Irfan Kandemirb, Durmus¸ Ali Ceylancand Aziz G€uld

a

Engineering Faculty, Department of Geomatics Engineering, Selcuk University, Konya, Turkey;bFaculty of Science, Department of Biology, Ankara University, Ankara, Turkey;cC¸umra Vocational School Department of Plantal and Animal Production, Selcuk University, Konya, Turkey;;dFaculty of Agriculture, Department of Animal Science, Mustafa Kemal University, Hatay, Turkey

(Received 7 December 2018; accepted 28 August 2019)

Beekeeping activities have a rapidly increasing importance due to their derived products and their contributions to human health, biodiversity, agriculture, and pollination. Moreover, beekeeping can play a major role in supporting rural development and bringing the benefits of sustainable developments and productivity. Thus, the decision on correct api-ary locations via suitability analysis techniques seems essential to sustain and increase the yield and efficiency. However, this decision-making process requires a large number of parameters including the topography of the field surrounding environment and climate conditions. At this point, Multi Criteria Decision Analysis techniques can provide efficient solu-tions for decision making on suitable apiary locasolu-tions. In this study, suitable apiary site selection analysis for beekeeping activities via Analytical Hierarchy Process (AHP) and The Preference Ranking Organization Method for Enrichment of Evaluations (PROMETHEE) methods was proposed for Konya city located in Central Anatolia as a pilot application. Aspect, elevation, slope, flora, water resources, roads, railroads, settlements, electricity lines, buildings, precipitation, and natural disaster area criteria were assessed to generate beekeeping suitability for apiaries. The restrictions and requirements of beekeeping activities were specified considering expert beekeeper decisions and recent studies for the study area. The AHP and PROMETHEE suitability maps were validated with existing 461 apiary locations to be able to determine the reliability and applicability.

Keywords: AHP; beekeeping; Geographical Information Systems; multi-criteria decision analysis; PROMETHEE

Introduction

In many developing countries, beekeeping activities have importance to rural economic development through derived products (honey, propolis, pollen, beeswax, royal jelly, and bee venom) which are very important for human health and crop pollination (Damian, 2016; Estoque & Murayama, 2010, 2011). Among agricultural activities, there may be a considerable amount of eco-nomic income coming from honey bees due to their pollinator role for a variety of agricultural crops (approximately 33%) (Maris, Mansor, & Shafri, 2008; Oldroyd & Nanork, 2009).

Turkey has considerable potential in agriculture with its 38,328,000 ha arable land which corresponds to 30.8% of the total area and 7.4% of the gross national product value of the country (URL 1). Due to this importance, sustainable management and monitoring beekeeping activities are becoming more important to ensure productivity considering its pollinator role and economic income in addition to its derived products (Sarı & Ceylan, 2017). In this context, assessment of suitable locations for the apiaries should be determined considering the environmental, ecological, economic, and social perspectives.

Determining suitable locations can be assessed on the basis of spatial data of the physical environment (Jafari & Zaredar,2010; Zhang, Su, Wu, & Liang, 2015) which are included in Geographical Information Systems (Stephanie & Benno,2013). Suitability analysis should be considered to ensure optimum use of the resources and maximum benefit involving human needs and eco-system sustainability according to the expectations and requirements (Ahamed, Rao, & Murthy, 2000; Amiri & Shariff, 2012; Collins, Steiner, & Rushman, 2001; Malczewski, 2004; Sarı & Ceylan, 2017; Zolekar & Bhagat,2015).

The most suitable way to achieve this is Multi Criteria Decision Analysis (MCDA) techniques such as Analytical Hierarchy Process (AHP), The Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), Vise Kriterijumska Optimizacija I Kompromisno Resenje (VIKOR) and The Preference Ranking Organization Method for Enrichment of Evaluations (PROMETHEE) which are widely used in suitability concept. Determining the best solutions within multiple alternatives, user-defined approaches and requirements are included in MCDA concept (Joerin, Theriault, & Musy, 2001; Wang, Hall, & Subaryono, 1990; Yu, Chen, Wu, & Khan, 2011;

Corresponding author. Email:fatihsari@selcuk.edu.tr

ß 2020 International Bee Research Association

Zolekar & Bhagat, 2015). AHP is one of the most applied MCDA techniques which calculate the import-ance of each criterion contributed to the total beekeep-ing suitability (Saaty 1977, 1980, 1994, 2001; Saaty & Vargas, 1991). AHP uses a pairwise comparison matrix to make a comparison of all criteria (Chen, Yua, & Khan, 2010) to provide suitable solutions to the prob-lems (Arentze & Timmermans,2000).

In an outranking method PROMETHEE, it is possible to define different preference functions for criteria (Brans, Mareschal, & Vincle,1984; Brans & Vincle, 1985; Brans, Vincke, & Mareschal, 1986). PROMETHEE is a useful method for ranking several alternatives from suit-able to non-suitsuit-able and adaptsuit-able for ranking conflicting criteria (Albadvi, Chaharsooghi, & Esfahanipour, 2007). The simplicity of PROMETHEE provides an advantage in comparison to the other MCDA methods (like TOPSIS,

VIKOR, and ELECTRE) and the preference functions support which provides more realistic definition for real problems (Senvar, Tuzkaya, & Kahraman,2014).

There are quite a few studies focused on determin-ing the most suitable apiary locations via MCDA. AHP is used for beekeeping suitability based on maximum temperature, relative humidity, summer crop area, water resources, and land cover criteria (Abou-Shaara, Al-Ghamdi, & Mohamed, 2013). Another AHP study considered rainfall, topographic, hydrology, road net-work, nectar, and pollen classes as criteria in the method (Maris et al., 2008). Similarly, Estoque and Murayama (2010), in their study, considered distance to water and roads, elevation, nectar, and pollen class cri-teria in another AHP application. Solar and electromag-netic radiation criteria were also included in suitability analysis using AHP (Fernandez, Roque, & Anjos, 2016).

Geographical Information System (GIS) is used by Amiri and Shariff (2012) to determine the beekeeping suitabil-ity based on road and water availabilsuitabil-ity, temperature and amount of precipitation, vegetation, and land use. Camargo et al. (2014) detected the beekeeping suitabil-ity by calculating 3 km buffer zone for each established apiary location and evaluated land use, flora, and honey productivity. Zoccali et al. (2017) used AHP method to generate beekeeping suitability by using, road network, hydrologic features, temperature, and elevation criteria. All the studies that focused on locating and evaluating suitable areas involve similar criteria and characteristic features of beekeeping requirements. Although the con-sidered criteria are similar to recent studies, in this study, building density, slope, aspect, railroads, and nat-ural disaster probabilities are also included in the suit-ability analysis and PROMETHEE method is used in addition to AHP to make a comparison of methods for beekeeping suitability in Konya province in the Central Anatolia as a pilot study.

Materials and methods Study area

The study was carried out in Konya (38.873 km2) in Turkey including its 31 districts (Figure 1). Konya has the largest agricultural lands in Turkey with 18,854,582 ha (96.6% grain farming, 1.2% vegetable gar-dens, and 2.2% for fruits, beverage, and spices) accord-ing to the 2017 governmental statistics, 49% of Konya total area is arable lands (URL 1). The topography, meteorological conditions and flora are varying in the region due to its vast geographic area. The altitude of the city is 1020 m above sea level and has high moun-tains above 3000 m. The forests cover 13% of the Konya and are mostly found above 1200 m including predominantly Cedar, Black Pine, Oak, Fir, Red Pine, and Juniper.

The AHP and PROMETHEE methods were applied to generate the beekeeping suitability maps of the study area and the methods are described below.

Criteria selection

The criteria selection for beekeeping suitability involves defining meteorological, topographic, and environmental requirements and limitations of honey bees. Providing optimum beekeeping activities and increasing the yield requires locating apiary sites in an ideal interval for each criterion which is defined by bee experts and advanced beekeeper decisions. Based on the optimum beekeeping activities, selected criteria and data sources were given inTable 1.

Aspect (AS): Aspect map was derived from ASTER GDEM elevation model at 30 30 m resolution. This criterion was considered to enable determining direc-tions and their effects on beekeeping activities.

Elevation (EL): Elevation map was downloaded from ASTER GDEM project web site (URL 2). This cri-terion was related to regional flora, meteorological con-ditions, and topographic features. The elevation varied from 591 to 3419 in the study area.

Slope (SL): Slope criterion was also derived from ASTER GDEM elevation model and varied from 0 to 71.4%. This criterion was related to topographic fea-tures and aspect which has profound effect on beekeeping.

Flora (FL): Flora criterion was downloaded from CORINE project web site. Flora of the study area has a decisive role in honey production quality, type of honey, and yield quantity. Thus, flora criterion should be weighted higher than other criteria (Figure 2d). Beekeepers prefer forests, pastures, and natural plant areas to benefit from pollen and nectar sources. Urban and residential areas were excluded to avoid negative effects of urbanization on beekeeping activities. Although agricultural lands are important for beekeeping activities, pesticide, and insecticide usages for agricul-tural crops refer high risk for productivity and honey bees.

Distance to water resources (DtW): Water cri-terion included lakes, dams, rivers, and streams. The study area has a total of 2127 m2 water resources. Beekeeping requires enough clear water for bees both in cooling beehives and production.

Table 1. Summary of spatial data types (resolutions, scales, and sources).

Data Scale Resolution Source

Elevation – 30 30 m Aster Global Digital Elevation Model (ASTER-GDEM)

Aspect – 30 30 m Derived from ASTER-GDEM

Slope – 30 30 m Derived from ASTER-GDEM

Roads 1/1000 – Konya General Directorate of Highways

Railroads Konya General Directorate of Highways

Water Resources 1/5000 – General Directorate of State Hydraulic Works

Rivers 1/1000 – General Directorate of State Hydraulic Works

Flora 1/1000 – Corine 2012 Data

Settlements 1/1000 – OSM Database

Buildings OSM Database

Natural hazards Disaster and Emergency Management Presidency

Precipitation 1/1000 – Turkish State Meteorological Service

Distance to roads, railroads and settlements (DtR, DtRl, DtS): These criteria were derived by buffer analysis of point (settlements) and polyline (roads, rail-roads) vector data with an interval of 500 m. Being close to roads and settlements are unsuitable places for bee-keeping due to the greenhouse gases and air–noise pollu-tion. Settlement data were used as urban area cover.

Distance to power lines (DtP): This criterion was derived by buffer analysis of power lines with an interval of 200 m. Power lines were considered as a criterion to avoid electromagnetic field effect on honey bees.

Distance to buildings (DtB): Similar to settlements criterion, buildings were considered as a criterion to avoid the negative effects of buildings on beekeeping. In

Figure 2. (a) Aspect, (b) elevation, (c) slope, (d) flora, (e) water resources, (f) roads, (g) settlements, (h) power lines, (i) buildings, (j) railroad, (k) natural disasters, (l) precipitation.

addition, being close to the buildings can form an undesired and dangerous environment for people when bees foraging. Unlike settlement data, buildings’ data included greenhouses, stables, farms, and rare buildings.

Precipitation (PR): Precipitation expectation for optimum beekeeping was given between 1275 mm and 1800 mm annual rainfall (FAO, 1976; Maris et al.,2008) and had a close relationship with elevation, flora and its flowering season (Sari & Ceylan,2017).

Natural disasters (ND): Natural disaster areas are a threat to beehives. Sinkholes, probable flood areas, and landslide areas were included in this criterion to avoid loss of beehives.

Analytical hierarchy process (AHP)

The AHP method was introduced by Saaty (1977,1980) and proposes a scale to define the importance of each criterion (1¼ Equal, 3 ¼ Moderately, 5 ¼ Strongly, 7¼ Very, 9 ¼ Extremely). AHP uses a pairwise compari-son matrix to scale (ann) the importance of each

criter-ion in Formula 1.

Normalized matrix was generated by division of each element to the sum of its own column with Formula 2.

a1_ij ¼Panij i¼1aij

(2)

The average of the sum represented the weights (Formula 3). wi¼ 1 n   Xn i¼1 a0_ij, ði, j¼ 1, 2, 3, ::::, nÞ (3) AHP method includes a consistency assessment to decide the weight calculations are consistent or not. Consistency Index (CI) is a reliability indicator of given preference values and calculated weights. CI was calcu-lated with Formula 4 (Saaty,1994).

CI¼kmaxn

n 1 (4) Calculating CI required the kmax (eigen value) value

and Random Index (RI) value according to the matrix order (1.53 in this study) (Alonso & Lamata, 2006; Saaty,1980). kmax¼ 1 n Xn i¼1 Pn j¼1aijwj wi " # (5)

Finally, Consistency Ratio (CR) could be calculated with Formula 6. According to Saaty and Vargas (1991), values

higher than 0.1 require revision of the preference values included in pairwise comparison matrix (Saaty,1980).

CR¼CI

RI (6)

PROMETHEE

There were two requirements in PROMETHEE method for each criterion: a Preference function and a weight value. Preference functions were used to scale decisions into a preference ranging from 0 to 1. To achieve this, there were six preference functions proposed (Brans & Mareschal, 2005). The weight value defined the contri-bution rate of each criterion to the total suitability.

The preference function characterized the difference for a specific criterion between the evaluations gener-ated by two decisions with a preference ranging from 0 to 1 (Taillandier & Stinckwich,2011).

The preference function (Pi) translated the difference

between the evaluations obtained by two alternatives (a and b) into a preference degree ranging from 0 to 1. A preference index p(a,b) of a over b could be defined that considers all the criteria (Brans et al.,1986)

p a, bð Þ ¼ Pk i¼1wi x Pi ða, bÞ Pk i¼1wi , (7)

where wi>0 are weights of each criterion. p(a,b) index

gave a measure how a is preferred to b for all criteria; the closer to 1, the greater the preference. As a calcu-lation for the strength of the alternative, the leaving flow was calculated (Brans et al.,1986)

ɸþ_{aÞ ¼}X_{pða, xÞ (8)}

The entering flow measured the outranked character of a and used to determine the weakness of the alterna-tive a (Brans et al.,1986)

ɸ_{aÞ ¼}X_{pðx, aÞ (9)}

In PROMETHEE I, partial rankings could be generated. These rankings were used to specify rankings, incompar-ability and indifference situations. In PROMETHEE I, alter-native a was preferred to alteralter-native b if one of the elements of equations was satisfied (Dagdeviren,2008),

ɸþ_{ð Þ>ɸa} þ_{ð Þ and ɸb} _{ð Þ<ɸa} _{ð Þb}

ɸþ_{ð Þ>ɸa} þ_{ð Þ and ɸb} _{ð Þ ¼ ɸa} _{ð Þb}

ɸþ_{ð Þ ¼ ɸa} þ_{ð Þ and ɸb} _{ð Þ<ɸa} _{ð Þb} (10)

PROMETHEE I evaluation also allowed incomparabil-ity situations. If the following equation was satisfied, there was an indifference situation between a and b alternative (Dagdeviren,2008),

ɸþ_{ð Þ ¼ ɸa} þ_{ð Þ and ɸb} _{ð Þ ¼ ɸa} _{ð Þb (11)}

If one of the following conditions was satisfied, two alter-natives were considered as incomparable (Brans et al.,1986),

(1) A Criteria 1 Criteria 2 Criteria 3 … Criteria n

Criteria 1 a11 a12 a13 … a1n

Criteria 2 a21 a22 a23 … a2n

… … … …

ɸþ_{ð Þ>ɸa} þ_{ð Þ and ɸb} _{ð Þ>ɸa} _{ð Þb}

ɸþ_{ð Þ<ɸa} þ_{ð Þ and ɸb} _{ð Þ<ɸa} _{ð Þb} (12)

Complete ranking could be calculated via PROMETHEE II. Complete ranking calculation required determination of the net flow of the alternatives (Brans et al., 1986)

ɸ að Þ ¼ ɸþ_{ð Þ  ɸa} _{ð Þa (13)}

For each alternative a, p(a,b) was an overall prefer-ence index of a over b, ɸþ(a) and ɸ– (a). ɸ(a) repre-sented a value function, whereby a higher value reflected a higher attractiveness of alternative a and was called net flow (Brans et al., 1986)

a outranks b if ɸ að Þ> ɸ bð Þ

a was indifferent to b if ɸ að Þ ¼ ɸ bð Þ (14) Beekeeping suitability application

Generating suitability maps both with AHP and PROMETHEE required determining the importance of each criterion to be able to calculate weights. AHP pair-wise comparison matrix was used to calculate the weights of criteria. The values and comparison matrix were given inTable 2.

The consistency ratio was calculated as 0.004 for this comparison matrix. The CR value meant the weights are consistent and can be used in the suitability map generation process. AHP suitability map was calculated by overlay analysis with the following formula which was applied to each pixel of criteria maps;

Total Suitability ðTSÞ

¼ ðAS  0:129Þ þ ðEL  0:063Þ þ ðSL  0:048Þ þ ðFL  0:339Þ þ ðDtW  0:133Þ þ ðDtR  0:036Þ þ ðDtS  0:035Þ þ ðDtP  0:048Þ þ ðDtB  0:047Þ þ ðDtRl  0:035Þ þ ðPR  0:042Þ þ ðND  0:045Þ In the second stage, the calculated weights were used in PROMETHEE to calculate suitability with an evaluation matrix. For the purpose of constituting evalu-ation matrix, there were 1500 test points specified in the study area considering a homogenous distribution

to the study area and a ranking S1, S2, S3, N1, and N2 values were assigned to each test point. According to the FAO (1976), land suitability classes were divided into five classes as highly suitable (S1), moderately suit-able (S2), marginally suitsuit-able (S3), currently not suitsuit-able (N1), and (N2) to be able to classify land suitability from none suitable to highly suitable. Thus, the evalu-ation matrix included 1500 rows and 12 columns includ-ing rankinclud-ing values for 12 criteria. Each value for test points was assigned with extracting raster values to points. Test points distribution was given inFigure 3.

The ranking values for each criterion were assigned between 1 and 9 considering beekeeping requirements (Table 3).

PROMETHEE included six types of preference functions to compare alternatives. Linear (type V) preference func-tion was best suited for quantitative criteria. Therefore, due to the quantitative structure of criteria which were involved in beekeeping suitability, the linear preference function was used. Although it was easier to achieve the solution by using PROMETHEE II (complete preorder), PROMETHEE I (par-tial preorder) provided more realistic solution. Especially in terms of incomparabilities, PROMETHEE I could often be useful for decision making (Senvar et al.,2014). Leaving flow (ɸþÞ, entering flow ðɸÞ, and net flow ðɸÞ calculations val-ues were given inTable 4.

Validation of results

The results were evaluated in three different methods such as suitability ranking calculation, the intersection assessment of existing apiary locations and suitability values, and correlation analysis of beekeeping govern-mental statistics and suitability values. As an effective validation, the reliability and accuracy of generated suit-ability maps were calculated by using existing apiary locations retrieved from the Directorate of Provincial Food Agriculture and Livestock District Offices in 2017. Existing apiary locations were stored between May and September 2017 with beehive count, beekeeper name, address, and honey type information. In total, 461 exist-ing apiary location coordinates were included to assess the intersection. Another validation could be realized with correlation assessment between beekeeping

Table 2. AHP pairwise comparison matrix.

Criteria AS EL SL FL DtW DtR DtS DtP DtB DtRl PR ND Weight AS 1 2 2.7 0.3 1.1 4 4 2.5 2.5 4 3 3 0.129 EL 1 1.1 0.2 0.5 2 2 1.1 1.1 2 1.5 1.5 0.063 SL 1 0.2 0.3 1.2 1.2 1 1 1.2 1.1 1.2 0.048 FL 1 3.5 9 9 8 8 8 7 8 0.339 DtW 1 4 4.2 3 3 4.2 3 2.8 0.133 DtR 1 1 0.8 0.8 1 0.9 0.8 0.036 DtS 1 0.8 0.8 1 0.9 0.7 0.035 DtE 1 1 1.6 1.1 1.2 0.048 DtB 1 1.4 1.1 1.1 0.047 DtRl 1 0.9 0.7 0.035 PR 1 0.8 0.042 ND 1 0.045

Figure 3. 1500 test points for the study area. Table 3. Evaluation matrix for PROMETHEE. Criteria (unit) AS (Class) EL (m) SL (%) FL (Class) DtW (m) DtR (m) DtS (m) DtP (m) DtB (m) DtRl (m) PR (mm) ND (m) Weight 12.9 6.3 4.8 33.9 13.3 3.6 3.5 4.8 4.7 3.5 4.2 4.5 Indifference 1 1 1 1 1 1 1 1 1 1 1 1 Preference 9 9 9 9 9 9 9 9 9 9 9 9 Min 1 1 1 1 1 1 1 1 1 1 2 2 Max 9 7 9 9 9 9 9 9 9 9 9 9 Average 6.11 6.77 6.04 5.18 2.75 4.47 6.25 8.42 8.82 8.79 5.50 5.51 SD 2.96 0.74 1.96 2.33 2.62 2.87 2.49 1.78 0.94 1.18 1.78 1.78 TP1 3 7 7 4 1 4 5 8 9 9 5 9 TP2 9 7 6 4 1 3 7 9 9 9 5 9 TP3 3 7 5 4 1 9 4 9 9 9 5 9 … … … … TP1498 9 7 5 4 1 1 1 9 9 9 9 9 TP1499 9 7 7 7 1 2 5 9 9 9 9 9 TP1500 9 7 8 4 2 7 8 9 9 9 8 9

statistics and determined suitability values. Thus, total honey production, total beekeeper and beehive counts were retrieved from Turkish Statistical Institute web site for 31 districts (URL 1).

Results

The results indicated that 67.4% and 67.5% of the study area is calculated as suitable and 32.6% and 32.5% of the study area is not suitable according to the AHP and PROMETHEE calculations, respectively. Considering most suitable locations for beekeeping, 3.40% and 2.13% of the study area were calculated with AHP and PROMETHEE, respectively. The flora criterion had 34%, distance to waters 13.30%, and aspect had 12.9% weights in total weight calculation (Table 2). Thus, approximately 60% of suitability was defined by these classes. The suit-ability index maps were produced for AHP and PROMETHEE. The suitability index maps were generated by using TS andA values and were given inFigure 4.

Considering the total suitability index values, the highest suitability rates were calculated by AHP. There was a high correlation between suitable pixel counts and test points distribution rate to suitability index val-ues according to AHP calculation. The suitability ranking

rates were compared according to the intervals which are given in Table 5. Considering the S1 and S2 as suit-able rankings, 440 test points were assigned suitsuit-able by AHP and 325 test points by PROMETHEE. For N1, N2 rakings, 517 test points by AHP, and 972 test points by PROMETHEE were assigned unsuitable according to the intervals which are given inTable 5.

Considering the S1, S2, and S3 suitability index val-ues, AHP had 74.19% and PROMETHEE 76.56% over-lapping rates with the suitability maps (Figure 5).

However, existing apiary locations were mostly over-lapped with AHP method. N1 and N2 classes repre-sented beekeepers located on agricultural lands. Because agricultural lands were weighted quite a few due to the negative effects of pesticide usage, existing apiary locations on agricultural lands were clustered in N1 and N2 classes. S1 and S2 classes represented the natural plant areas, forests and grasslands.

Considering the correlation analysis results, there was a good agreement revealed (0.71) between PROMETHEE suitability and total beekeeper count sta-tistics. Because beekeepers are being registered to the Republic of Turkey Ministry of Food, Agriculture and Livestock provincial directorates to locate their bee-hives in any area, total beekeeper count correlation were determined higher than other statistics due to the higher reliability of total beekeeper count records. The correlation graphics were given inFigure 6.

Correlation r values and statistics involved in Table 5

for both AHP and PROMETHEE methods.

According to the correlation analysis of each method, there was a strong correlation between AHP and PROMETHEE. It was possible to say that all the cal-culations with AHP and PROMETHEE are consistent

Table 4. PROMETHEE rankings.

Test points ɸ ɸþ ɸ TP1 –0.2175 0.1016 0.3192 TP2 –0.0764 0.1628 0.2392 TP3 –0.2197 0.1064 0.3261 … … … … TP1498 –0.1099 0.1618 0.2717 TP1499 0.2315 0.3307 0.0992 TP1500 0.0361 0.2246 0.1885

and one of these methods can be used in beekeeping suitability. The correlation graphic was given inFigure 7.

Discussion

Considering the used methods in site suitability analysis for beekeeping, Maris et al. (2008), Estoque and

Murayama (2010), Abou-Shaara et al. (2013), Fernandez et al. (2016), and Zoccali et al. (2017) used only AHP method to generate suitability. However, in this study, PROMETHEE method was also applied in addition to AHP which provides a more comprehensive approach to the complex structure of determining a suitable site for apiary locations. Applying different methods

Table 5. Suitability ranking rates and existing beekeeper comparison.

Values S1 (9–7) S2 (7–6) S3 (6–5) N1 (5–4) N2 (4–1)

AHP Test points 51 391 569 412 77

Min: 1.97 % 3.40% 26.07% 37.93% 27.47% 5.13% Max: 8.43 Mean: 5.37 Pixel 148318 1052215 1546137 1250546 249871 Pixel: 4251524 % 3.49% 24.77% 36.40% 29.44% 5.88% m2 1321.682 9601.631 14110.9 11545.28 2293.507 Existing beekeepers 24 141 176 104 15 % 5.21% 30.59% 38.39% 22.56% 3.25% Values S1 (0.5/0.4) S2 (0.4/0.2) S3 (0.20/-0.10) N1 (–0.10/–0.30) N2 (–0.30/–0.50)

PROMETHEE Test points 32 299 682 399 88

Min:–0.57 % 2.13% 19.93% 45.47% 26.60% 5.87% Max: 0.45 Pixel 34534 309815 650944 2088548 1167679 Mean:–0.01 % 0.02% 8.08% 64.44% 26.07% 1.40% Pixel: 4251524 m2 310.984 2829.954 5947.569 19125.52 10658.98 Existing beekeepers 0 51 302 105 3 % 0.00% 11.06% 65.51% 22.78% 0.65%

Figure 5. (a) Bozkır-Ahırlı/H€uy€uk/C¸umra-Karatay Beekeepers AHP suitability. (b) Bozkır-Ahırlı/H€uy€uk/C¸umra-Karatay Beekeepers PROMETHEE suitability.

increased the reliability of the project due to enabling us to make a comparison and verification of the results.

Another important stage of site suitability analysis for apiary locations was selected criteria which are included in site suitability analysis. Considering used criteria in site suitability for apiary locations in previous studies, most detailed criteria were used in this study. On the other hand, the most important stage of site suitability analysis is the determination of accuracy and reliability values. Amiri and Shariff (2012) and Abou-Shaara et al. (2013) did not examine the accuracy and reliability of the analysis. Camargo et al. (2014) compared the speci-fied suitable locations with average yield, average prod-uctivity, and beehive counts. Estoque and Murayama (2010) examined a correlation analysis between honey yield data of existing beekeeper projects and suitability rates. Maris et al. (2008) used only three existing apiary

Figure 6. Govermental statistics and AHP–PROMETHEE correlation.

locations for verification. In this study, in addition to honey yield statistical data and suitability values ation analysis, AHP and PROMETHEE method correl-ation analyses were included to verify the methods’ suitability and reliability. Although Maris et al. (2008) used only three existing apiary locations for verification, in this study, 461 existing apiary locations were used to calculate the intersections with suitability values. Considering the average 75% intersection rate of exist-ing locations with suitability maps, it was possible to say that the results of this study are quite satisfactory when decision making. The results indicated that all calcula-tions such as importance of each criterion, preference values and assigned from 1 to 9 scale were quite suc-cessful when evaluating the intersection rates and cor-relation analysis.

The results also indicated that the proposed model generated beekeeping suitability maps successfully. The results also indicate a good simulation of beekeeping requirements via preference values and importance of criteria given in this study. Specifying bee requirements and setting the optimum intervals to make decisions from complex alternatives were very difficult and inevit-able process. Thus, some limitations were encountered in this study. Increasing accuracy of this project could be succeeded by including additional criteria such as wind directions, flowering seasons and pesticide usage distribution. Due to the unavailability of temporal flora map and plant density information which are important for bees, there was a particular limitation when making a decision accurately.

Moreover, probable natural disaster data must be included in beekeeping suitability analysis. In order to avoid natural hazard effects, natural hazard risk maps (probable flood areas, forest fire risk, landslides risk, etc.) should be included in beekeeping suitabil-ity analysis.

The study could provide valuable and significant experience for beekeeping suitability projects when designing not only provincial but also national projects. As a validation, the results should be investigated with experts and the productivity of the existing beekeepers should be monitored in the next season to be able to compare the validity of this study. In addition, some api-ary locations could be established as unsuitable and most suitable locations to make a comparison of prod-uctivity. This study can be accepted as a conceptual model of beekeeping suitability in Turkey and using suit-able criteria, it can be also easily adaptsuit-able to other countries.

Acknowledgement

We would like to thank to Dr. Ergin S¸AH_IN for valuable efforts on proofreading of this manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Fatih Sari http://orcid.org/0000-0001-8674-9028

Aziz G€ul http://orcid.org/0000-0003-1158-5019

References

Abou-Shaara, H. F., Al-Ghamdi, A. A., & Mohamed, A. A. (2013). A suitability map for keeping honey bees under harsh environmental conditions using geographical informa-tion system. World Applied Science Journal, 22, 1099–1105. Ahamed, N., Rao, G. K., & Murthy, R. (2000). GIS-based fuzzy

membership model for crop-land suitability analysis. Agricultural Systems, 63, 75–95. doi:10.1016/S0308-521X(99)00036-0

Albadvi, A., Chaharsooghi, S. K., & Esfahanipour, A. (2007). Decision making in stock trading: An application of PROMETHEE. European Journal of Operational Research, 177(2), 673–683. doi:10.1016/j.ejor.2005.11.022

Alonso, J. A., & Lamata, M. T. (2006). Consistency in the ana-lytic hierarchy process: A new approach. International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, 14(04), 445–459. doi:10.1142/S0218488506004114

Amiri, F., & Shariff, M. A. (2012). Application of geographic information systems in landuse suitability evaluation for beekeeping: A case study of Vahregan watershed (Iran). African Journal of Agricultural Research, 7(1), 89–97.

Arentze, T. A., & Timmermans, H. J. P. (2000). ALBATROSS: A learning-based transportation oriented simulation system. Eindhoven: EIRASS, Eindhoven University of Technology. Brans, J. P., & Mareschal, B. (2005) PROMETHEE methods. In:

Multiple criteria decision analysis: State of the art surveys. International Series in Operations Research & Management Science (vol 78). New York, NY: Springer.

Brans, J. P., & Vincke, P. (1985). A preference ranking organ-ization method. Management Science, 31(6), 647–656. doi:

10.1287/mnsc.31.6.647

Brans, J. P., Mareschal, B., & Vincle, P. (1984). PROMETHEE: A new family of outranking methods in MCDM. In: Brans J. P. (Ed.). Operational research IFORS (vol 84, pp. 477–490). Amsterdam: North-Holland.

Brans, J. P., Vincke, P., & Mareschal, B. (1986). How to select and how to rank projects: The PROMETHEE method. European Journal of Operational Research, 24(2), 228–238. doi:10.1016/0377-2217(86)90044-5

Camargo, S. C., Garcia, R. C., Feiden, A., Vasconcelos, E. S. D. E., Pires, B. G., Hartleben, A. M., … Pereira, D. J. (2014). Implementation of a geographic information system (GIS) for the planning of beekeeping in the west region of Parana. Anais da Academia Brasileira de Ci^encias, 86(2), 955–971. doi:

10.1590/0001-3765201420130278

Chen, Y., Yua, J., & Khan, S. (2010). Spatial sensitivity analysis of multi-criteria weights in GIS-based land suitability evalu-ation. Environmental Modelling & Software, 25, 1582–1591. doi:10.1016/j.envsoft.2010.06.001

Collins, M. G., Steiner, F. R., & Rushman, M. J. (2001). Land-use suitability analysis in the United States: Historical devel-opment and promising technological achievements. Environmental Management, 28(5), 611–621. doi:10.1007/ s002670010247

Dagdeviren, M. (2008). Decision making in equipment selec-tion: An integrated approach with AHP and PROMETHEE. Journal of Intelligent Manufacturing, 19(4), 397–406.

Damian, G. C. (2016). GIS-based optimal localisation of beekeep-ing in rural Kenya (Master’s thesis). Department of Physical Geography and Ecosystems Science, Lund University, Lund. Estoque, R. C., & Murayama, Y. (2010). Suitability analysis for

bee-keeping sites in La Union, Philippines, using GIS and multi-cri-teria evaluation techniques. Research Journal of Applied Sciences, 5(3), 242–253. doi:10.3923/rjasci.2010.242.253

Estoque, R. C., & Murayama, Y. (2011). Suitability analysis for beekeeping sites integrating GIS & MCE techniques. Spatial ana-lysis and modeling in geographical transformation process. Dordrecht, the Netherlands: Springer.

FAO. (1976). A framework for land evaluation. Soils Bulletin 32. Rome: Food and Agriculture Organization of the United Nations.

Fernandez, P., Roque, N., & Anjos, O. (2016). Spatial multicri-teria decision analysis to potential beekeeping assessment. Case study: Montesinho Natural Park (Portugal). In: Sarjakoski, T., Santos, M. Y., Sarjakoski, L. T. (Eds.), 19th AGILE International Conference on Geographic Information Science—Geospatial Data in a Changing World, Helsinki, Finland.

Jafari, S., & Zaredar, N. (2010). Land suitability analysis using multi attribute decision making approach. International Journal of Environmental Science and Development, 1, 441–445.

Joerin, F., Theriault, M., & Musy, A. (2001). Using GIS and out-ranking multi-criteria analysis for land-use suitability assess-ment. International Journal of Geographical Information Science, 15(2), 153–174. doi:10.1080/13658810051030487

Malczewski, J. (2004). GIS-based land-use suitability analysis: A critical overview. Progress in Planning, 62(1), 3–65. doi:10. 1016/j.progress.2003.09.002

Maris, N., Mansor, S., & Shafri, H. (2008). Apicultural site zon-ation using GIS and multi-criteria decision analysis. Pertanika Journal of Tropical Agricultural Science, 31(2), 147–162. Oldroyd, P. B., & Nanork, P. (2009). Conservation of Asian

honey-bees. Apidologie Bee Apidologie, 40, 296–312. doi:10. 1051/apido/2009021

Saaty, T. L. (1977). A scaling method for priorities in hierarch-ical structures. Journal of Mathemathierarch-ical Psychology, 15(3), 234–281. doi:10.1016/0022-2496(77)90033-5

Saaty, T. L. (1980). The analytical hierarchy process. New York: Wiley.

Saaty, T. L. (1994). Fundamentals of decision making and priority theory with the analytical hierarchy process (pp. 69–84). Pittsburg: RWS Publications.

Saaty, T. L. (2001). Decision making with dependence and feed-back: The analytic network process (2nd ed.). Pittsburgh, PA: RWS Publications.

Saaty, T. L., & Vargas, L. G. (1991). Prediction, projection and forecasting. Dordrecht: Kluwer Academic Publishers. Sarı, F., & Ceylan, D. A. (2017). Site suitability analysis for

bee-keeping via analythical hyrearchy process, Konya Example. ISPRS Annals of Photogrammetry, Remote Sensing and Spatial Information Sciences, IV-4/W4, 345–350. doi: 10.5194/isprs-annals-IV-4-W4-345-2017

Senvar, O., Tuzkaya, G., & Kahraman, C. (2014). Multi criteria supplier selection using fuzzy PROMETHEE method. In: Kahraman C. & €Oztays¸i B. (Eds.), Supply chain management under fuzziness. Studies in fuzziness and soft computing (vol 313). Berlin, Heidelberg: Springer.

Stephanie, R., & Benno, S. (2013). Standard use of Geographic Information System (GIS) techniques in honey bee research. Journal of Apicultural Research, 52(4), 1–48.

Taillandier, P., & Stinckwich, S. (2011). Using the PROMETHEE multi-criteria decision making method to define new exploration strategies for rescue robots. 2011 IEEE International Symposium on Safety, Security, and Rescue Robotics, Kyoto, pp. 321–326.

URL 1, Turkish Statistical Institute Official Page. (2018). Retrieved fromhttp://www.tuik.gov.tr

URL 2, ASTER GDEM Official Page. (2018). Retrieved from

http://www.jspacesystems.or.jp/ersdac/GDEM/E/index.html. Accessed 18.05.2018.

Wang, F., Hall, G., & Subaryono, B. (1990). Fuzzy information representation and processing in conventional GIS software: Data base design and applications. International Journal of Geographical Inform Systems, 4(3), 261–283. doi:10.1080/ 02693799008941546

Yu, J., Chen, Y., Wu, J., & Khan, S. (2011). Cellular automata-based spatial multi-criteria land suitability simulation for irri-gated agriculture. International Journal of Geographical Information Science, 25(1), 131–148. doi:10.1080/ 13658811003785571

Zhang, J., Su, Y., Wu, J., & Liang, H. (2015). GIS based land suitability assessment for tobacco production using AHP and fuzzy set in Shandong province of China. Computers and Electronics in Agriculture, 114, 202–211. doi: 10.1016/j.com-pag.2015.04.004

Zoccali, P., Malacrino, A., Campolo, O., Laudani, F., Algeri, G. M., Giunti, G., … Palmeri, V. (2017). A novel GIS-based approach to assess beekeeping suitability of Mediterranean lands. Saudi Journal of Biological Sciences, 24(5), 1045–1050. doi:10.1016/j.sjbs.2017.01.062

Zolekar, R. B., & Bhagat, V. S. (2015). Multi-criteria land suit-ability analysis for agriculture in hilly zone: Remote sensing and GIS approach. Computers and Electronics in Agriculture, 118, 300–321. doi:10.1016/j.compag.2015.09.016