Public preferences for landscape features: The case of agricultural landscape in mountainous Mediterranean areas

Land Use Policy 26 (2009) 334–344

Contents lists available at ScienceDirect

Land Use Policy journal homepage: www.elsevier.com/locate/landusepol

Public preferences for landscape features: The case of agricultural landscape in mountainous Mediterranean areas ´ Samir Sayadi ∗ , M. Carmen Gonzalez-Roa, Javier Calatrava-Requena Andalusian Institute of Agricultural Research and Training (IFAPA), Dpt. Agricultural Economics and Rural Sociology, Apdo. 2027 18080 - Granada, Spain

a r t i c l e

i n f o

Article history: Received 8 November 2006 Received in revised form 27 February 2008 Accepted 17 April 2008 Keywords: Aesthetic value Rural landscape Conjoint Analysis Contingent Valuation Sustainable rural development South-eastern Spain

a b s t r a c t Provision of landscape amenities produced by farmers, in addition to their economic function of producing food and fibre, has contributed to a reassessment of the role of agriculture in society. In this paper, we examine whether agricultural landscape provision really responds to a social demand as is argued by those in favour of multifunctionality. Thus, the aim of the present work is two-fold. First, we evaluate rural landscape preferences of citizens from a range of choices in the mountain area of the Alpujarras (southeastern Spain), and second, we estimate their willingness to pay (WTP) to enjoy each of the landscape characteristics existing in the area. For the empirical analysis, based on a survey of public preferences due to the good public characteristics of landscape amenities, we applied two stated preference methods: Conjoint Analysis (CA) and Contingent Valuation (CV). Three landscape attributes were considered for this analysis: type of vegetation layer, density of rural buildings, and level of slope. Several levels were also considered for each attribute: abandoned fields, dryland farming, irrigated farming, and natural lands were included for the vegetation layer; three levels (low, intermediate and intense) were considered for the level of slope and three levels (none, little and intense) for rural buildings. The empirical findings from the CA and CV confirm that the agricultural-landscape component (first irrigated lands, followed by dryland farming, within the attribute “vegetation layer”), plays an important role in public preferences on the landscape and WTP. Maintaining local agricultural activities, preventing future migration from agricultural lands, recovering abandoned fields, and including elements of rural landscape observation and appreciation of existing recreational programmes for rural tourism in the area, were among the strategies to take full advantage of this aesthetic landscape potential, and to foster sustainable development of the region. © 2008 Elsevier Ltd. All rights reserved.

Introduction Society’s recreational demands for landscapes in rural Mediterranean areas have been increasing heavily in recent years, since the aesthetic contribution provided by these areas clearly increases ´ the welfare of the citizens (Dearden, 1980; DeLucio and Mugica, 1994; Santos, 1998; C. Hall et al., 2004). Society’s demands for new functions in rural landscapes are also rapidly changing and diversifying (Sarapatka and Sterba, 1998; Vos and Meekes, 1999; Gary, 2001; Musacchio et al., 2005). At the same time, the supply of high-quality landscapes is steadily declining both quantitatively and qualitatively as a result of the degradation caused by activities of diverse nature and magnitude (Bush, 2006; Mottet et al., 2006;

∗ Corresponding author. Tel.: +34 660 402344; fax: +34 958 895203. E-mail addresses: [email protected] (S. Sayadi), [email protected] ´ (M.C. Gonzalez-Roa), [email protected] (J. Calatrava-Requena). 0264-8377/$ – see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.landusepol.2008.04.003

Rao and Rekha, 2001; Verburg et al., 2006; Tasser et al., 2007). Landscapes have dramatically changed in the countryside as a result of both public subsidies and technological changes in agriculture and forestry (intensification/extensification, agricultural practices, afforestation, nature conservation, etc.) (Bush, 2006; Van Meijl et al., 2006; Westhoek et al., 2006). These changes have brought about a decline in the more traditional roles of agriculture as well as an increasing interest in new functions (Sayadi and Calatrava, 2001; C. ¨ a¨ and Kola, 2004; De Groot, 2006). Hall et al., 2004; Yrjol In response to social environmental concern and demand, and as a result of the growing consideration of environmental objectives in the new paradigm of sustainable agriculture, evaluation of environmental externalities of agricultural systems has become increasingly important, particularly since the mid-eighties. Among the externalities caused by agriculture, we should consider how this activity has shaped the landscape, analysing the aesthetic function of agro-ecosystems (Deffontaines, 1985, 1986; Thenail and Baudy, 1994). Different agro-ecosystems have different capabilities of shaping the landscape, and rural landscapes will

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

display a different degree of the agricultural component, depending on the composition of the agricultural systems. To identify various types of environmental externalities linked to agricultural activities, Kline and Wichelns (1996), Sayadi (1998), Sayadi and Calatrava (2001) and Mottet et al. (2006) consider their role to be key in shaping the landscape. It is therefore crucial to recognize and appraise this contribution to the rural spaces and to determine whether landscape provision really responds to a social demand. A comprehensive approach to the analysis and assessment of a certain agricultural landscape for rural development must take into account its aesthetic (Laurie, 1975; Hammitt et al., 1994; Hull and Revell, 1989; Arriaza et al., 2004); its ecological (Zonneveld and Forman, 1989) or geographical (Dunn, 1974; Blaschke, 2006), and its cultural (Vos and Meekes, 1999) aspects. This can be achieved only ´ if we understand the concept of perception. According to Gonzalez (1981), landscape is the “multi-sensory perception of a system of ecological and cultural relations”. People thus shape the landscape, are part of it, and also form perceptions of it. As Laurie (1975) points out, landscape evaluation may be defined as “the comparative relationships between two or more landscapes in terms of assessment of visual quality”. For such an evaluation, we consider the rural landscape as the final product, in visual and aesthetic terms, of a series of interacting factors, including climate, relief, water, soil, natural flora and fauna, and human actions. The result of this interaction is a specific spatial layout of agro-ecosystems which is a characteristic of each territory, this being its most perceivable dimension. Despite the many studies on alternatives for evaluating externalities (Daniel and Vining, 1983; Amir and Gidalizon, 1990; Adamowicz et al., 1994, 1997; Boxall et al., 1996; Blamey et al., 1998; Hanley et al., 1998a,b, 2001; Santos, 1998; Wherrett, 2000; ´ ¨ Bennet and Blamey, 2001; Hernandez et al., 2004; Kayhk o¨ and Skanes, 2006; among others), submitted to monetary evaluation methods (Contingent Valuation, Hedonic Price, Travel Cost method, ´ etc.) to estimate the value of open spaces (DeLucio and Mugica, ¨ 1994; Hammitt et al., 1994; Tyrvanen and Hannu, 1998; Scarpa et al., 1999; Wang et al., 2006; among others), studies of primarily agricultural landscapes are scarce (Dunn, 1974; Price, 1978, 1990; Drake, 1987, 1992; Lee, 1990; Willis and Garrod, 1993; Pruckner, 1995; Brunstad et al., 1999; Arriaza et al., 2004). An aesthetic valuation of agriculture is complex, and may be expressed directly in monetary values only in the extreme cases of homogeneous, specific landscapes, spatially localized and in a situation of evident aesthetic contrast. In Spain, the only study available (Calatrava, 1996) applies the Contingent Valuation Method to assess such landscapes, in the context of the sugarcane landscape in the ˜ valley (Granada, south-eastern Spain). Another Motril-Salobrena work compares and debates the results found using two commonly used preference techniques: ranking and rating in the application of Conjoint Analysis method for assessment of agricultural landscape preferences (Sayadi et al., 2005). The present paper adds to this literature by appraising the value of agricultural landscape amenities and by comparing estimates of this value obtained using Conjoint Analysis (CA), which is a non-monetary approach, and Contingent Valuation (CV), a monetary approach. Estimating the values for different attributes of rural landscape and the willingness to pay (WTP) using the latter technique is also a novelty with respect to earlier studies. This is particularly valuable in helping policy makers redesign sustainable rural-development programmes in order to take fuller advantage of the aesthetic potential and to increase social welfare. In this paper, we attempt, on one hand, to evaluate the agricultural attribute in the public enjoyment of the landscape, and, secondly, to quantify the monetary value of the different aspects of these landscapes. We provide a short overview in section ‘The

335

study area: landscape change in the Alpujarran mountain of southeastern Spain’ of the landscape evolution and changes in the mountain areas of south-eastern Spain. Section ‘Methodological framework’ explains the methodology in the empirical study of the contribution of agricultural landscape to public aesthetic preferences and willingness to pay. Two experiments using both the CA and CV methods were designed for comparative valuation of rural landscapes. The CA and CV methods were based on surveys performed to citizens from the provinces of Almeria and Granada (south-eastern Spain), near the study area and regions of potential visitors. This provides useful information on the importance of agriculture for public preferences and allows valuations of the aesthetic rural landscape. Section ‘Results’ presents the results of the empirical analysis and section ‘Discussion’ discusses the findings. The main conclusions and recommendations are offered in section ‘Conclusions’.

The study area: landscape change in the Alpujarran mountain of south-eastern Spain Since the 1950s, as a consequence of the rural exodus, many rural Spanish regions have undergone changes in their landscape structure due to the abandonment of agricultural activities and, in some cases to the proliferation of other economic activities, such as tourism. The Alpujarras of Granada (see Fig. 1), situated in the south of the massif of Sierra Nevada (south-eastern Spain), exemplifies this transformation, being typical of the Mediterranean high-mountain regions of Europe. The Alpujarras district, with a series of mountain valleys and gorges, has abrupt altitude gradients (almost sea level to 3500 m), and steep inclines impeding traditional farming systems. Irrigation systems, many dating from the 15th Century or earlier, are fed by streams and snowmelt from the Sierra Nevada summits and have permitted an intricate system of terraced agricultural land, which typifies the landscape around the mountain villages from 800 m a.s.l. to 1800 m a.s.l. This traditional agricultural landscape is at risk from agricultural abandonment. The irrigated terraces are labour-intensive and thus support a multi-cropping system which includes field crops, vegetables, trees, and, at lower elevations, vines and olives. In this study, agricultural landscapes are analysed, these being below 2000 m and having undergone steady anthropic activity over history. The landscapes in the zone have been described in detail by Calatrava and Molero (1983), Sayadi and Calatrava (2001) and among others. Local farming has been gradually abandoned since the beginning of the rural exodus in the fifties, and demographic changes in the second half of this century were dramatic. Most of the Alpujarran villages recorded population highs in 1950 and an exodus since then. The population declined by some 50% since 1950, with rates approaching 4% per year between 1960 and 1975, migrating towards other parts of Spain (especially Barcelona and Madrid) seeking employment in industry, in coastal tourism (e.g. Costa del Sol, Malaga), and also in intensive agriculture (particularly that of greenhouse horticulture along the Spanish Mediterranean coast). This migration, as in other European mountain areas, had a measurable environmental impact. Calatrava and Sayadi (1997, 1999) reported that small-irrigated multi-crop farms are particularly threatened. According to a more recent work by Calatrava and Sayadi (2003), farm abandonment has slightly decelerated as a consequence of the current European Rural-Development policies favouring non-farm activities (particularly rural tourism), non-agricultural activities, and, in some cases, the development of part-time farming. Nevertheless, small-irrigated farms are still

336

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

Fig. 1. The study region.

been abandoned to some extent. Furthermore, the specific climatic conditions of the area have accelerated terrace degradation through the collapse of terrace fronts. Douglas (1997), using air photographs to identify terrace collapses, demonstrated a clear association between high levels of degradation and abandoned land or de-intensified agriculture. In areas of active cultivation, where terrace wall maintenance continued, has apparently suffered less. The degradation trend of this mountain landscape for the future is clear. Growing income from tourism has not halted the decline in farming and consequently the attractive attributes of the mountain landscape will suffer further decline. The abandonment will affect not only the landscape structure but also the habitat suitability of many species (Burela and Baudryb, 1995; Scozzafava and De Sanctis, 2006; Woodhousea et al., 2005). The autochthonous landscape in these mountain areas and their ecological equilibrium are consequently menaced (Bakkera et al., 2005; Pecoa et al., 2006; Antrop, 2006; De Groot, 2006; Potschin and Haines-Young, 2006). Methodological framework The assessment of landscape visual quality assumes that landscapes have an intrinsic beauty (Shutteworth, 1980a) which, although being a subjective response of the observer (Palakowski, 1975), can be quantified. Buhyoff and Riesenmann (1979), Dearden (1980), Daniel and Vining (1983), Willis and Garrod (1993), Calatrava (1996), and Arriaza et al. (2004) offer a detailed review of landscape-evaluation techniques. The public’s preferences for landscape amenities can be quantified by different valuation methods. This study highlights two methods. The Conjoint Analysis method (CA), which is a nonmonetary approach, is used in combination with the Contingent Valuation method (CV), a monetary approach, in order to perform the relative appraisal of landscapes. Conjoint Analysis The Conjoint Analysis (CA), developed by Luce and Tukey (1964) and Krantz and Tversky (1971), was initially applied in commercial psychology and marketing literature by Green and Rao (1971). As CA subsequently adapted to new research trends, its practical application increased in environmental research (Wittink and Cattin, 1989; Ness and Gerhardy, 1994; Green et al., 1999; Alvarez-Farizo and Hanley, 2002). Conjoint measurement is based on the assumption that: (1) a product can be described according to levels of a set of attributes and (2) the consumer’s overall judgement with respect to that product is based on these attribute levels. Both assumptions are commonly made in economics and marketing. Conjoint mea-

surement seeks to quantify the consumer’s overall judgement (e.g. quality evaluation) on the basis of these underlying product attributes, providing a quantitative measurement of the relative importance of certain features over others. It starts with the consumer’s overall judgements concerning a set of product alternatives (i.e. combinations of attribute levels) and breaks down the overall judgements into the combination of each attribute level. The contribution of the various attribute levels to the overall judgement is called part-worths (Green and Srinivasan, 1990). The importance of an attribute “i”, Ii , is defined in terms of the range of the part-worths, aij , across the levels “j” of that attribute: Ii = {max(aij ) − min(aij )},

for each I

Where: i = attributes; j = levels for attribute i; aij = part-worth for attribute i and level j. The attribute’s importance index (Ii ) is normalized to ascertain its importance relative to other attributes, Wi : Wi =

I

ni

I i=1 i

n = number of attributes. So that: n 

Wi = 1

i=1

For more details on the basic principles of expressed preferences methods, the reader is referred to Green and Wind (1975), Fenwick (1978), Antilla et al. (1980), Steenkamp (1987), Louviere (1988a,b), Mackenzie (1990, 1992, 1993), Gilbert and Larkin (1995), Boxall et al. (1996), Adamowicz et al. (1998), Garrod and Willis (1998), Green et al. (1999), Blamey et al. (2000), Louviere and Street (2000), Bennet and Blamey (2001), Morisson et al. (2002), Ara (2003) and J. Hall et al. (2004). Contingent Valuation Contingent Valuation (CV) is probably the most widely used method for estimating the monetary values of public goods. This method is based on consumer surveys whose questions elicit the consumer preferences for public goods by constructing a hypothetical market based on a private goods market or a political market (Mitchell and Carson, 1989). The aim of a CV study is to estimate consumers’ maximum willingness to pay (WTP) for public goods

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

by asking them how much they would pay for certain government actions (Carson, 2000). The contingent analysis is a direct method of expressed preferences, based on the information stated by the interviewees when asked for their willingness to pay for certain types of goods, the market for which had been previously simulated (see section ‘Data Collection and analysis’). In a CV study, respondents are presented with commodities, usually in the course of a personal face-to-face interview, which involves:  A detailed description of the commodity(ies) being valued and the hypothetical circumstance under which the valuation is supposed to be made are presented to the respondents.  Questions which elicit the respondents’ preferences, and their maximum willingness to pay for the commodity(ies) being valued. The CV question can be asked in several different formats ¨ a¨ and Kola, 2004). (Navrud, 2000; Yrjol  Questions about respondent’s socio-economic status (e.g. age, income, etc.). This part is important to determine whether attitudes and willingness to pay is influenced by socio-demographic factors. For more general overview on economic valuation methods, refer to Navrud (1992), Bateman and Willis (1995), Bennett et al. (1998), Carson (2000) and Carson et al. (2002). Methodological design for landscape valuation Methodology for the Conjoint Analysis of landscapes In order to apply the Conjoint Analysis method, the Alpujarra’s landscapes features were identified by analysing the answers given by visitors in previous work (Sayadi, 1998; Sayadi and Calatrava, 2001) to some scale-rating questions involving the main attributes of the landscape in the area. Three landscape attributes were finally selected as follows: type of vegetation layer, density of rural building and level of slope. For each attribute, specific levels were chosen that were both relevant to visitors and representative of the landscape situation in the area. The attributes used and attribute levels that were finally selected are shown in Table 1. For the experimental design and analysis of the preference structure for Alpujarran landscapes, Bretton & Clark’s Conjoint Designer (Bretton-Clark, 1987a) and Conjoint Analyser (BrettonClark, 1987b), version 2.0, computer programs were used. The multiple regression model was fit using Statgraphics Plus, version 4.0.

Table 1 Features and levels of the landscapes used in the experiment Features

Levels

Description

Vegetation layer (feature 1)

Abandoned fields Dryland farming

Abandoned farmlands Almond orchard, vineyard, fig, olive tree Irrigated-orchard Lands that never were used for agriculture

Irrigated farming Natural lands Level of slope (feature 2)

Gentle slope Intermediate slope Steep slope

Less that 10% From 10% to 20% More than 20%

Level of building (feature 3)

No building

Without architectonic components Some typical houses, isolated houses Population centre

Little building Intense building

337

Had we used a full profile design (by including all the possible combinations of levels and attributes), an excessive number of possible landscapes (36 profiles) would have been generated for a respondent to be evaluated, some of which would not even be present in real life. Each combination of attribute levels represents a specific landscape alternative (profile). For this to be narrowed to a reasonable number of testable combinations, an orthogonal fractional factorial design was used (Louviere et al., 2000), allowing us to assess the relative importance of the different landscape attributes through a reduced sampling of the profiles. Since none of the attributes in an orthogonal array are related, the intercorrelations, or of-diagonal elements are 0.0 (Papoulis and Pillai, 2002; Bretton-Clark, 1987b). Thus, the orthogonal array permits the measurement of the main effects of attributes on an uncorrelated basis. This design assumes that all interactions are negligible. The Conjoint Designer provided 16 hypothetical landscapes or index cards, which comprise the total number of final stimuli which the respondents were shown. We also adopted the additive composition model as applied by Steenkamp (1987). This model, the simplest and by far the most frequently used, assumes that the overall evaluations are formed by the sum of the separate part-worth (partial standardised utility) of the attributes. Other models include the interactive model and the multiplicative model. However, the interactive model research indicates that these models seldom have a significantly better fit to the data than additive model (see, Emery and Barron, 1979; Steenkamp, 1987, p. 474). This can be formulated as: Total Value = U0 +

mi n  

aij

i=1 j=1

Where: i = 1, . . ., n: number of attributes; j = 1, . . ., mi : number of levels for attribute i; U0 : constant; aij : part-worth for attribute i and level j. The method allows us to estimate one model for each individual (individual model) and an aggregate one (mean model). The stimuli (attribute-level combinations) presented to the interviewees were photographs of real landscapes taken in the area, following the orthogonal fractional-array design shown in Table 2. Each photo represents a specific combination of attribute levels or specific landscape alternative. Several studies have attempted to assess the scenic preferences of observers using photographs of Table 2 Orthogonal fractional factorial design Landscapes

Vegetation layer

Level of slope

Level of building

Landscape 1 Landscape 2 Landscape 3 Landscape 4 Landscape 5 Landscape 6 Landscape 7 Landscape 8 Landscape 9 Landscape 10 Landscape 11 Landscape 12 Landscape 13 Landscape 14 Landscape 15 Landscape 16

Irrigated farming Irrigated farming Irrigated farming Abandoned farming Natural lands Irrigated farming Dryland farming Dryland farming Abandoned farming Abandoned farming Dryland farming Dryland farming Abandoned farming Natural lands Natural lands Natural lands

Gentle Intermediate Gentle Gentle Steep Gentle Intermediate Gentle Intermediate Gentle Gentle Steep Steep Gentle Intermediate Gentle

No building Little building Intense building No building Little building Little building Little building No building Little building Little building Little building Intense building Little building Little building No building Intense building

338

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

rural landscapes (Dunn, 1976; Shafer and Brush, 1977; Shutteworth, 1980b; Law and Zube, 1983; Wherrett, 2000; Arriaza et al., 2004). Descriptions of the use of pictures in public-preference models and other methods (field observation, written description, etc.), can be found in Stewart et al. (1984), Shelby and Harris (1985), Bernaldez et al. (1988), Hull and Stewart (1992), Silvennoinen et al. (2002) and Garc´ıa et al. (2006). Landscapes 1 and 9 had to be partially modified, since these combinations were incompatible and this type of landscape is difficult to find in the area. The only modifications made refer to the change in the level of slope from “steep” to “gentle” in landscape 1 and in the level of building from “intense” to “little building” in landscape 9. Although this reduces the statistical efficiency, it is justified by an increase in the validity of the data (Bretton-Clark, 1987a,b). Despite the presence of an inter-attribute correlation, the basic conjoint analysis premises have not been violated. For all effects, inter-attribute correlations should be kept to a minimum, although they need not be zero, as long as the allowance of minor correlations (close to 0.2) adds realism to the study. In our case, the maximum correlation was about 0.279. Although 16 landscape photos can be considered an excessive number of stimuli in order to establish preferences, the respondents showed no difficulty in managing the 16 photos. The experimental design described above (construction of the stimuli, orthogonal design for the final stimuli and presentation form) was similar to that used in other landscape-preference studies (Sayadi et al., 1999, 2000, 2005).

Methodology for the Contingent Valuation of landscapes To estimate the willingness to pay for landscape views, we created an artificial market. Each interviewee was asked: “Please imagine in the Alpujarras area a hypothetical rural hotel of standard category with rooms offering different landscape views for which the only differentiating criterion in the room price is the view to be enjoyed from the terrace of room. With all the other lodging conditions the same (category of the hotel, services, type of the room, etc.), please indicate, after carefully examining the following landscape photographs, your maximum willingness to pay (WTP) for a day of lodging to enjoy the different views represented in the photographs”. The people interviewed were asked only once to state their maximum WTP for the room including the view. The most useful information in the CV was the difference in WTP that the same individual expressed for rooms with different landscape views. The landscapes (stimuli) in the photographs shown to the individuals were selected according to the same orthogonal design used for the Conjoint Analysis. For the expression of the WTP, an open-ended question format was used because the market was private (room price) and the persons polled were familiar with the prices of the hotels and country houses of the zone (Sayadi, 1998). Also, given that the number of landscapes shown to each individual was rather high, the use of another type of format (dichotomous, auction, etc.) would have excessively prolonged the test. For subjects who expressed some difficulty in understanding the kind of “medium category hotel” in the area, prices, etc., a brochure was used to explain the type of accommodation existing in the area. These subjects, representing less than 5% of the sample, less familiar with the area and showed some difficulty in comprehending the objective of CV exercise. As stated above, people interviewed were asked only once to state their maximum WTP for the room to enjoy each of the landscapes; in this case, the payment vehicle was the price of lodging. The most useful information was the difference in WTP that the same individual expressed for the different landscapes views.

Data collection and analysis The questionnaire was split into three parts. The first part contained the CA exercise and was aimed at quantifying the individual landscape preferences of the interviewees. The second part included the CV exercise and attempted to find out the maximum daily WTP of interviewees for room in a medium Cottage/Rural Hotel with views of any of these different landscapes from the bedroom window. The third part of the questionnaire included socio-economic characteristics of the respondents (age, sex, income, education, etc.). The CA and CV exercises were administered by the same researcher but surveys were separated in time. The landscape preferences were measured on an ordinal scale from 1 to 16, by ranking the landscape cards or stimuli from most (1) to least (16) preferred. Thus, after examining each photograph, the interviewees were asked to rank each landscape according to their preferences. Sayadi et al. (2005) show that this ranking approach represents public preferences better than does a rating technique in the Conjoint Analysis of rural landscapes and reveals clearer differences between levels of attributes than does the rating method. Data were collected from June to August 2002 in 163 personal interviews to citizens from Granada and Almeria, who lived close to the area. These two provinces were the most likely origins for potential visitors (Sayadi, 1998; Sayadi and Calatrava, 2002). The sample size was acceptable, considering the standard deviation resulting for the WTP means giving a sample error of less than 2 D . The individuals were randomly selected. However, sample characteristics were compared to visitor characteristics on the basis of the distribution of age, gender, and level of study, and were found to be representative of tourists in the area. Both the CA and CV were administered to the same subjects, but separately in time and without any connection between the two tests. Persons were contacted in different places (some on the street, others in bars, workplaces, parks, etc.), but always where they were able to examine the photos (presence of tables, benches, etc.). The respondent was asked to observe all the photographs carefully in order to fill out the questionnaire. It was explained that the questionnaire consisted of two parts: the first part to be conducted immediately and the second afterwards on the date that the person preferred (after one day, two days, etc.) in the place that the person specified. Persons that did not agree to participate in the second part (less than 10% of the subjects), were excluded from the survey. The refusal to participate did not appear to be the content of the questionnaire, as the individuals were first asked for their willingness to co-operate before beginning the work. Only 9% of respondents that participate in the first part of the survey (CA) failed to fulfil their commitment to continue with the second exercise (CV), resulting in 163 valid questionnaires. After the survey, we analysed the data of the CA and CV by calculating: (i) the average ranking (AR) assigned by respondents to each landscape (input of CA), (ii) the average utilities (AU) obtained from the assessed utility model (output of CA), and (iii) the average of the WTP (AWTP) for each landscape stimuli used in the experimental design (input CV). Also, several correlation coefficients were calculated: the Pearson correlation and Spearman’s ordinal correlation coefficients between (i) the average WTP (AWTP) assigned by respondents to each landscape and the average ranking (AR), (ii) the average WTP (AWTP) assigned by respondents to each landscape and the average utility (AU) and (iii) the average ranking (AR) for each landscape and the average utility (AU) for each landscape. A multiple regression model was also fitted to data in order to identify the relationship between respondents’ WTP to enjoy each landscape, and their socio-economic profile.

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

339

Table 3 Definition of independent variables in the multiple regression model

Table 4 Preference test results for the total group

Variables

Description

Features

Constant Irrf Dryf Abaf Natl

Constant term “1” if irrigated farming, “0” if not “1” if dryland farming, “0” if not “1” if abandoned farming, “0” if not “1” if natural lands, “0” if not

Stpslp Interslp Gentslp

“1” if steep slope, “0” if not “1” if intermediate slope, “0” if not “1” if gentle slope, “0” if not

Intbuild Litbuild Nobuild

“1” if intense building, “0” if not “1” if little building, “0” if not “1” if no building, “0” if not

Age Educ Occup Prov Ruz Agr Fagr Degfagr Vis Ms Sex Nper Pci AWTP-ind

Interviewee’s age (years) Interviewee’s level of education “1” if in the labour market, “0” if not “1” if from the province of Granada, “0” if not “1” if subject has lived or lives in a rural zone, “0” if not Present or past association with agrarian activities “1” if any relative or ancestor is/was a farmer, “0” if not Degree of family association with agrarian activities Frequency of outings to the countryside (approx.) “1” if married, “0” if not “1” if male, “0” if female Number of persons in household Monthly per capita income Average of prices assigned to landscapes by each individual

Explanatory variables considered in the model (Table 3) were of two types. The first were the landscape characteristics appraised in the photographs (possible levels of each attribute): irrigated farming (Irrf), dryland farming (Dryf), abandoned fields (Abaf) and natural lands (Natl), of the “Vegetation Layer” attribute (variable); steep slope (Stpslp), intermediate slope (Interslp) and gentle slope (Gentslp) of the attribute “Level of Slope”; and intense building (Intbuilt), little building (Litbuild), and no building (Nobuild) of the attribute “Level of Building”. To avoid the dummy-variable trap, we include n − 1 levels for each attribute in the regression model. The second part describes the socio-demographic characteristics of interviewees: age (Age), education (Educ), occupation (Occup); province (Prov); living or having lived in a rural zone (Ruz); being or having been a farmer or farm worker (Agr); having or not having relatives/ancestors who were or had been farmers (Fagr); degree of family relationship with agricultural activities (Degfagr); approximate frequency of outings to the countryside (Vis); sex (Sex); marital status (Mst): number of persons in the household (Nper); monthly per capita income (Pci). Likewise, the average WTP assigned to landscapes by each individual was used as a co variable (Awtp-ind). The dependent variable used for the fitting of the model, was WTPij , which is the willingness to pay expressed by the individual i to enjoy the landscape j from the terrace of a hypothetical rural hotel.

Results The main characteristics of the citizens surveyed were as follows: 69% were 25–45 years old; most (62.35%) were male; about half (52.47%) were married; the most frequent household size (77.85%) was 2–4 people; the average monthly household income was about 1200 D /month; and more than 40% had a university degree.

Vegetation layer Level of building Level of slope

Relative importance (%)

Levels

Utility value (part-worth)

44.95

Irrigated farming Dryland farming lands Abandoned farming Natural lands

2.255 0.044 −1.916 −0.383

35.66

Intense Little None

1.630 0.050 −1.679

19.39

Intermediate Gentle Steep

1.034 −0.270 −0.765

Results of Conjoint Analysis of landscapes Table 4 shows the group utility function (part-worth) and relative importance (%) of the different attributes. From the results on the relative importance of the attributes (Table 4), the nature of the vegetation layer proved the most relevant attribute (44.95%) in forming public landscape preferences. This was followed by the level of existing building or construction (35.66%). The level of slope occupied the third place in importance (19.39%). Within the attribute Vegetation Layer, “irrigated lands” constituted the most relevant level in forming public landscape preferences, followed by “dryland farming”, showing a positive relative part-worth (Table 4). “Natural lands” and “abandoned fields” show less importance in forming landscape preferences, their part-worths being negative (Table 4). Therefore, the higher the agricultural component in Alpujarran landscapes, the greater the level of appreciation and worth on the part of respondents. It should be mentioned that, in the Alpujarra the landscape has been transformed from centuries of human habitation, so that the presence of purely natural lands is rare. Although we might expect a preference for the scarce natural lands that exist, this was not the case, since, due to the natural aridity of the zone, natural vegetation consists mostly of dry and low mountain scrub. Also, the area shows a growing proportion of abandoned agricultural lands. It should be noted that a negative value in part-worth does not necessarily mean that the respondents had negative preferences for a respective trait level, but rather that it was less preferable than the others, since we are dealing with relative zero totalling coefficients (Bretton-Clark, 1987b). In relation to the level of building, the presence of “intense building” (traditional and vernacular Alpujarran houses and villages) was the most important level in forming landscape preferences, followed by the presence of “little building”. Landscapes with “no building” were least preferred by respondents, with a negative partworth (Table 4). Thus, the higher the level of building within the landscapes, the greater the degree of appreciation by respondents. It should be mentioned also that the region is part of the Sierra Nevada Natural Park, where building is forbidden apart from the traditional architecture in the old inhabited villages. Regarding the attribute “slope”, the “intermediate level” was the most relevant in forming public preferences in the landscape, while the extremes “steep” and “gentle” slopes were less relevant (Table 4). The most highly valued landscape had an agricultural component of irrigated farming on an intermediate slope with a village visible in the landscape. On the other hand, the least preferred landscape involved old abandoned farmlands, with no villages and a steep slope.

340

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

Table 5 Ordinal scoring average values, average utility and average willingness to pay (AWTP) Landscapes (photos)

Average ranking (AR)

Average utility (AU)

Average WTP* (AWTP) (D )

Average WTP (AWTP) difference significance (p ≤ 0.05)**

13 8 4 10 15 14 11 16 5 9 2 1 12 7 6 3

13.03 12.01 11.70 10.59 9.66 9.05 8.86 8.10 8.46 7.65 7.38 6.63 6.47 6.21 5.14 5.07

5.99 6.72 4.76 6.49 7.59 8.02 8.45 9.60 7.52 7.79 11.96 8.93 9.53 9.75 10.66 12.23

21.48 23.29 23.56 24.36 25.40 26.18 26.71 26.76 26.78 27.70 28.49 28.97 29.03 30.24 31.52 31.60

a b b bc cd de def def def efg fgh gh gh hi i i

* Taking into account that the WTP estimated here is the price for the room including the view and the willingness to pay score was from 21.48 D /day to 31.60 D /day, the only difference (between 0 D and 10.12 D ) was due to the landscape characteristics. ** Same letter implies no significant difference between groups of landscapes (homogeneous groups).

Table 5 displays the average ranking (AR) and the average utility (AU) calculated for each landscape stimulus used in the experimental design. The ranking offers a maximum value of 5.07 for the highest valued landscape, and a minimum value of 13.03 for the least appreciated. The highest calculated average utility of the landscapes was 12.23, and the lowest 4.76.

Results of Contingent Valuation of landscapes As for the interviewees’ WTP for a room to stay in a standard hotel to enjoy the views of the landscapes, the average daily price expressed was 27.07 D /day, the absolute maximum price being of 54.09 D /day and the absolute minimum of 18.03 D /day. Most respondents (60.5%) were willing to pay between 24.04 D /day and 36.06 D /day. Taking into account that the WTP estimated here is the price for the room including the view and the WTP score was from 21.48 D /day to 31.60 D /day, the only difference (between 0 D and 10.12 D ) was due to the landscape views. No protest bids were identified (protest bids are zero answers from subjects deriving positive utility from the good subject to valuation). Although protest bids in contingent valuation are quite common (Mitchell and Carson, 1989), they refer to a normal protest rate percentage of around 30%. In our case the absence of protest bids can be explained by the valuation context. Firstly, the payment vehicle was not focused on public policy (i.e. taxes), which is ´ ´ related to most protest bids in Spain (Barreiro and Perez y Perez, 1999). Secondly, the landscape in Spain is not supported through specific public programmes and the payment vehicle is related to private benefits derived from accommodation, and thus incentives for free enjoyment and/or strategic behaviour, expressed as protest bids, are also minimised. Table 5 also lists the average prices (AWTP) expressed by respondents for the different landscape stimuli shown in the photographs. The last column of Table 5 shows the results of Duncan’s multiplerange test to determine which WTP means were significantly different from others. The landscapes with the same letters imply homogeneous groups (p ≤ 0.05) of WTP means. Different letters imply significance of the difference between WTP means of the corresponding landscapes. Table 5 shows an overlap in the central preference valuation section, and two clearly differentiated groups: that of the most highly valued landscapes, and that of the least valued ones.

Comparison of Conjoint Analysis and Contingent results The Pearson and the Spearman ordinal correlation coefficients between the average WTP value (AWTP), the average ranking (AR), and the average utilities (AU) reached from the Conjoint Analysis results for each landscape profiles were calculated, giving the following results: AWTP AR = −0.997 between AWTP and AR rAWTP AR = −0.9913 rAWTP AU = 0.8647 AWTP AU = 0.8735 between AWTP and AU rAR AU = 0.8604 AR AU = −0.8941 between AR and AU where ‘r’ is the Pearson correlation and ‘’ is Spearman ordinal correlation. All were significant for (p ≤ 0.0001), which shows the great ordinal similarity among the average preference, WTP, and utility of the respective landscape views. The corresponding linear regression functions are as follows: AWTP=6198.77−200.618AR; (∗∗∗)

(∗∗∗)

F = 801.29(∗ ∗ ∗), R2 = 98.28

AWTP = 2764.96 + 203.359AU; (∗∗∗)

AR = 17.00 − AU ; (∗∗∗)

(∗∗∗)

*** Significance

(∗∗∗)

F = 41.41, R2 = 74.77

F = 39.93(∗ ∗ ∗), R2 = 74.04

(1) (2) (3)

(p ≤ 0.001). All adjustments were highly significant. The difference between WTP functions (1) and (2) lies in the value of R2 . While the average ranking (AR) of the preference function expressed explained over 98% of the variance in respondents’ average WTP for the landscape views, the average utility (AU) explained about 74%. The results of the multiple regression model are displayed in Table 6, where only significant variables (p ≤ 0.05) are included. The dependent variable, as indicated in the methodology, was WTPij , which is the willingness to pay expressed by the individual ‘i’ to enjoy the landscape ‘j’ from the terrace of a hypothetical rural hotel. Notably, despite the relatively low variance in WTP explained by the independent variables considered (32%), the model assessment was highly significant (p ≤ 0.001). The variance unexplained by the model (68%) may be attributable to the highly subjective and intrinsic traits of respondents, not specified in the model, that influenced their aesthetic preferences (life style, membership of ecological association, psychographics, etc.). Socio-demographic variables that do not influence the WTP for the landscape (p ≤ 0.05) are as follows: province (Prov), living or

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

341

Table 6 Results of the multiple regression model Parameter

Standard error

t Statistic

P-value

Dependent variable WTPij : WTP of the individual i for the landscape j Constant 4.44 Natl 1.60 Irrf 5.26 Dryf 2.64 gentslp −1.46 stpslp −2.28 Nobuild −2.06 Intbuild 1.60 Agr 0.48 AWTP-ind 0.66

1.07 0.42 0.44 0.42 0.36 0.48 0.36 0.42 0.21 0.03

4.14 3.74 11.89 6.18 −3.97 −4.72 −5.62 3.73 2.27 20.54

0.0000 0.0002 0.0000 0.0000 0.0001 0.0000 0.0000 0.0002 0.0229 0.0000

Source

Sum of squares

d.f.

Mean square

F-ratio

P-value

Analysis of variance Model Residual

27291.2 57796.1

9 1638

3032.35 35.2845

85.94

0.0000

85087.3

1647

Total (Corr)

Estimate

R-squared = 32.07%; R-squared (adjusted for d.f.) = 31.70%; standard error of est. = 5.94; mean absolute error = 4.49; Durbin–Watson statistic = 0.78.

having previously lived in a rural zone (Ruz), having or not having family/ancestors who were/had been farmers (Fagr), degree of family relationship with agricultural activities (Degfagr), approximate frequency of outings to the countryside (Vis), occupation (Occup), sex (Sex), marital status (Mst), age (Age), number of persons in household (Nper), education (Educ), and monthly per capita income (Pci). These results coincide with those of Sayadi et al. (1999), in finding no relationship between the socio-demographic variables and agricultural landscape utility. Within the socio-demographic variables explained in the methodology, only the variable “present or past relationship with farming activities” (Agr) was significantly related (p ≤ 0.05) to respondents’ WTP for landscapes to be viewed while staying in a rural hotel. Respondents who had neither previously been farmers nor farm workers showed a higher WTP for the landscape views than those who were or had been farmers (p ≤ 0.05). This implies that people with a present or past connection with farming activities would more used to seeing landscapes with a strong agricultural component, and therefore would be less willing to pay money to enjoy them. In order to carry out significance contrasting among the different levels of the variables referring to the characteristics of the landscape stimuli, results of their respective changes in the model can be summarised as follow:

In reference to the vegetation layer: • The WTP for the views with irrigated farming was significantly higher than for dryland farming (p ≤ 0.05), natural lands (p ≤ 0.001), and abandoned fields (p ≤ 0.05). • The WTP for views with dryland farming was significantly higher than for natural lands (p ≤ 0.05) and abandoned fields (p ≤ 0.001). • The WTP for views with natural lands was significantly higher than for abandoned fields (p ≤ 0.001). Thus, the higher the agricultural component in the area’s landscape views, the higher the WTP for their enjoyment. This agrees with the findings on the preference ranking. In relation to the level of slope: • The WTP for landscapes of intermediate slope was significantly higher than for gentle slopes (p ≤ 0.001) or steep slopes (p ≤ 0.001). • The WTP for landscape views with extremely gentle or extremely steep slopes did not significantly differ (p ≤ 0.05).

In relation to the level of building: • The WTP for landscape views with intense degree of traditional architecture (traditional houses and farmhouses) is significantly higher than for those with little or none (p ≤ 0.001). • The WTP for landscape views with a little degree of building was significantly higher than for those with none (p ≤ 0.001). Thus, the higher the level of building within the Alpujarran landscape views the higher the WTP for them. This is logical in an area such as the eastern Alpujarras, where the local traditional architecture of the small villages spread out over the slopes above deep gorges constitutes one of the most admired aspects by visitors to the area (Sayadi and Calatrava, 2002). These results indicate that the most economically valued landscape view would be of irrigated farming on an intermediate slope where a village is visible in the landscape. On the other hand, the least-valued landscape in the area involves old abandoned farmlands on steep slopes with no views of villages. Paradoxically, the current rural-development policy in the area does not encourage the maintenance of the kind of landscape that the interviewees most valued; therefore, current policy could indirectly cause a negative future impact on the development of rural tourism activity in the area. These results agree with those of the aforementioned conjoint analysis. Preference is reflected in WTP for landscape (i.e. higher preference implies higher WTP). These results confirm stability regarding landscape preferences for individuals. Discussion This paper attempts to measure consumers’ preferences for rural landscapes. Estimating the value that people assign to a landscape is difficult because it is a non-market commodity, and therefore methods such as the conjoint analysis (CA) and contingent valuation (CV) are used. The results found here are useful to make a series of decisions within the framework of sustainable rural development and to support agricultural policy decisions, which directly affect the provision of landscape as a non-commodity output of agricultural multi-functionality. Moreover, this study shows that other functions of agriculture, other than mere production, also benefits society; it is therefore relevant to deal with the value of other functions of agriculture, in particular aesthetic utility of agricultural

342

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

landscape. This is significant due to the importance of agro-tourism in the region as in many European Mediterranean mountain areas. Furthermore, agricultural landscapes in mountain areas constitute an important component of society’s aesthetic utility function. Therefore, the abandonment of farming activities, particularly irrigated farming, would indirectly have a negative impact on rural tourism by reducing demand (both in quantity and/or WTP), which is enough to justify a policy aimed at maintaining this kind of rural landscape, particularly in the Mediterranean regions of Europe. Policy makers need information on public preferences for intervention that efficiently enhances the aesthetic externality of agriculture in such areas (Dearden, 1980; UE, 2000). Previous studies (Calatrava and Sayadi, 1998) performed in the same region, have shown that actions which avoid the fragmentation of farms, foster co-operativism for marketing agricultural products, and develop part-time farming, positively influence the maintenance of diversified irrigated farming lands, which was found here to be the most aesthetically valued component of the landscape. Policies promoting these factors, along with the improvement of agricultural infrastructure, are key in maintaining this most appreciated rural landscape. Notwithstanding these findings and the suitability of stated preferences methods to estimate the social value of non-market agricultural output and environmental amenities, we should not ignore some limitations and weaknesses related to the nature of the methods used and the robustness of parameters and values estimated. Limitations are linked to methodological biases which are both instrumental and non-instrumental. These methods cannot alone provide the definitive answer to any significant political decision (Carson, 1998, 2000; Carson et al., 2002). Conjoint analysis and contingent valuation are fraught with difficulties. The assumption of the additive and non-interactive utility function model in conjoint analysis, the way of presenting the stimuli for respondents, and the technique for quantifying public preference are the main limitations for this method. Biases related to the payment vehicle, question formats, interviewer, strategy, embedding, etc., are among the most cited contingent valuation limitations that may lead to unreliable results. Nevertheless, the nature of the item evaluated “market commodity” and the used payment vehicle related to “private benefits” derived from accommodation (price of a hotel room), minimise the most common biases which could generate results that would give non-optimal guidance for policy design due to invalid values estimated (Navrud, 2000; Bateman and Willis, 1995; Bennett et al., 1998; Hanley et al., 2003). Lastly, regarding the value concepts estimated in this study, the values of rural landscape found (utilities and WTP) should be interpreted as a lower boundary of the use value of the agricultural landscape in the region, as it measures only part of use value, the aesthetic value, based on data from citizens located in neighbouring provinces. Agricultural landscape amenities certainly also have other use value (recreational, cultural, etc.) and non-use value (e.g. conservation, existence, option, legacy, etc.) which have not been considered in our research.

to lowest relative importance in the expression of preferences and WTP by respondents. Within the attribute vegetation layer, the agricultural component (first, irrigated, followed by dryland farming) was the most highly esteemed vegetation layer aesthetically, and the one that most strongly stimulates respondents’ WTP for views that include it. Natural lands and abandoned fields were the least appreciated and the ones which inspired the lowest WTP of citizens. Thus, in general, the higher the appreciation of the landscape, the higher the individual’s WTP to enjoy its aesthetic qualities. We also found that, the WTP for a certain landscape was better explained through the average assigned range than through the average estimated utility, explaining over 98% of the variance in respondent’s average WTP for landscape views. Respondents who neither were nor had previously been farmers or farm workers showed a higher WTP for landscape views than did those who were or had been farmers. Except for the foregoing finding, non-socio-economic traits in respondents appeared to affect their WTP for the landscapes, a WTP evidently related to inner, highly subjective traits of respondents (life style, membership of ecological association, psychographics, etc.). The average WTP for accommodations with views of the most highly appreciated landscape (landscapes with an irrigation agricultural component on an intermediate slope with a village or traditional houses visible in the landscape) was 31.60 D /day, and the least valued (landscapes of abandoned agricultural lands without any village in view and with a steep slope) was 21.48 D /day. Paradoxically, current rural-development policies in the area do not encourage the growth (nor even the maintenance) of the most highly valued vegetation-layer component in the area. The opposite is occurring: the abandonment of agriculture is sharply increasing the surface area of the least-valued vegetation layer. Based on our findings concerning the aesthetic potential of the agricultural systems of the area, we offer some recommendations for designing agricultural policies and rural-development strategies: • Local agricultural activities should be maintained and, whenever possible, those involving irrigation. Future migration from agricultural lands must be prevented and previously abandoned fields recovered. • Rural landscape observation and appreciation should be included in existing recreational activity programs for rural tourism in the zone (hiking, etc.). • Agriculture should be maintained close to population centres, since there seems to be a positive impact, in landscape preferences, on the architecture–agriculture combination. • Further research will be needed to analyse preferences using other aesthetic criteria (biological, ecological, etc.), in order to gain other perspectives for sustainable development in the area. Acknowledgements

Conclusions Taking into account the above results, we found that the Conjoint Analysis and the Contingent Valuation results match in the ordinal preference structure and WTP of interviewees, both in relation to the attributes that make up the landscape profile and their respective levels. The attributes of vegetation layer, level of building, and level of slope were, in this order, the ones presenting the highest

The authors thank four anonymous reviewers and the journal’s editor for their constructive comments which have substantially improved the final version of this paper. The usual disclaimers apply. Financial support from the Spanish National Institute for Agricultural Research and Technology (INIA) and Andalusian Regional Ministry of Innovation, Sciences and Enterprises (CICE) through projects RTA2006-00055 and PAIDI P07-SEJ-03121, respectively.

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

References Adamowicz, W., Louviere, J.J., Williams, M., 1994. Combining revealed and stated preference methods for measures of environment amenities. J. Environ. Econ. Manage. 26, 271–292. Adamowicz, W., Swait, J., Boxall, P., Louviere, J.J., Williams, M., 1997. Perceptions versus objective measures of environmental quality in combined revealed and stated preference models of environmental valuation. J. Environ. Econ. Manage. 32, 65–84. Adamowicz, W., Boxall, P., William, M., Louviere, J.J., 1998. Stated preference approaches for measuring passive use values: Choice Experiments and Contingent Valuation. Am. J. Agric. Econ. 80, 64–75. Alvarez-Farizo, B., Hanley, N., 2002. Using Conjoint Analysis to quantify public preferences over environmental impacts of wind farms. Energy Policy 30 (2), 107–116. Amir, S., Gidalizon, E., 1990. Expert-based method for the evaluation of visual absorption capacity of the landscape. J. Environ. Manage. 30, 251–263. ¨ Antilla, A., Van Den Henvel, R., Moller, K., 1980. Conjoint Measurement for Marketing Management. Eur. J. Mark. 14, 397–408. Antrop, M., 2006. Sustainable landscapes: contradiction, fiction or utopia? Landscape Urban Plan. 75 (3–4), 187–197. Ara, S., 2003. Consumer willingness to pay for multiple attributes of organic rice: a case study in the Philippines. In: 25th International Conference of Agricultural Economics, Durban, SA. ˜ ˜ ˜ J.A., Ruiz-Aviles, P., 2004. Assessing the Arriaza, M., Canas-Ortega, Canas-Madue no, visual quality of rural landscapes. Landscape Urban Plan. 69, 115–125. Bakkera, M.M., Goversb, G., Kosmasc, C., Vanackera, V., Van Oostb, K., Rounsevella, M., 2005. Soil erosion as a driver of land-use change. Agric. Ecosyst. Environ. 105 (3), 465–481. ´ ´ Barreiro, J., Perez y Perez, L., 1999. Non-market benefits valuation of conservation policies in Spain. J. Econ. Agric. Environ. 1 (99), 4–14. Bateman, I., Willis, K. (Eds.), 1995. Valuing Environmental Preferences: Theory and Practice of the Contingent Valuation Method in the US, EC and Developing Countries. Oxford University Press. Bennet, J., Blamey, R., 2001. The Choice Modelling Approach to Environmental Valuation. Edward Elgar, Cheltenham, UK/Northamptom, MA, USA, 269 pp. Bennett, J., Morrison, M., Blamey, R., 1998. Testing the validity of responses to contingent valuation questioning. Aust. J. Agric. Res. Econ. 42 (2), 131–148. Bernaldez, F.G., Ruiz, J.P., Benayas, J., Abello, R.P., 1988. Real landscapes versus photographed landscapes: preference dimensions. Landscape Res. 13, 10–11. Blaschke, T., 2006. The role of the spatial dimension within the framework of sustainable landscapes and natural capital. Landscape Urban Plan. 75 (3–4), 198–226. Blamey, R., Bennett, M., Morrison, M., Louviere, J., Rolfe, J., 1998. Attribute selection in environmental choice modelling. Environ. Res. Econ. 23, 167–186. Blamey, R.K., Bennet, J.W., Louviere, J.J., Morrisson, M.D., Rolfe, J., 2000. A test of policy labels in environmental Choice Modelling Studies. Ecol. Econ. 32, 269–286. Boxall, P., Admowicz, W., Williams, M., Swait, J., Louviere, J.J., 1996. A comparison of Stated Preferences Methods for Environmental Valuation. Ecol. Econ. 18, 243–253. Bretton-Clark, 1987a. Conjoint Designer, Version 2. Bretton-Clark, 37 pp. Bretton-Clark, 1987b. Conjoint Analyser, Version 2. Bretton-Clark, 61 pp. Brunstad, R.J., Ivar, J., Erling, V., 1999. Agricultural production and the optimal level of landscape preservation. Land Econ. 75, 538–546. Buhyoff, G.J., Riesenmann, M.F., 1979. Experimental manipulation of dimensionality in landscape preference judgements: a quantitative validation. Leisure Sci. 2, 221–238. Burela, F., Baudryb, J., 1995. Species biodiversity in changing agricultural landscape: a case study in the Pays d’Auge, France. Agric. Ecosyst. Environ. 55 (3), 193–200. Bush, G., 2006. Future European agricultural landscapes—what can we learn from existing quantitative land use scenario studies. Agric. Ecosyst. Environ. 114 (1), 121–140. ´ economica ´ Calatrava, R.J., 1996. Valoracion de paisajes agrarios: Consideraciones ´ del metodo ´ ´ contingente al caso de la cana ˜ generales y aplicacion de valoracion ´ ˜ In: Azqueta, D., Perez, L. (Eds.), Gestion ´ de azucar en la Vega de Motril-Salobrena. de Espacios Naturales. McGraw-Hill, pp. 139–215. Calatrava, J., Molero, J., 1983. Consideraciones sobre la aplicabilidad de las directrices ´ penibetica ´ socioculturales agrarias de la CEE a la depresion de Las Alpujarras. Anales del Instituto Nacional de Investigaciones Agrarias. Serie Econom´ıa y Sociolog´ıa Agraria 7, 119–165. Calatrava, J., Sayadi, S., 1997. Economic marginality, social depreciation of agriculture and the disappearance of farming in depressed areas in South-eastern Spain. In: XXIIIth Congress of the I.A.A.E., Sacramento, CA, USA, August 10–18. Calatrava, J., Sayadi, S., 1998. De l’exploitation agricole a` l’exploitation rurale: Nouveaux regards sur l’agriculture des zones de montagne du Sud-Est espagnol. ´ ´ ` Agraires et le Developpement; Gestion Rev. Etudes Recherches sur les Systemes ´ ´ des exploitations et des ressources rurales: entreprendre, negocier, evaluer, 31, 387-396, INRA, France. Calatrava, J., Sayadi, S., 1999. Agrarian crisis, farming abandonment and social regard for agriculture in depressed mountainous areas of South-eastern Spain. In: European Congress of Agricultural Economists. European Agriculture Facing the 21st Century in a Global Context, Warsaw, Poland, August. Calatrava, J., Sayadi, S., 2003. Agrarian activity permanence and rural development policies: an analysis from a follow-up (1981–2001) of farms in mountain areas in South-eastern Spain. In: 25th International Conference of IAAE, Durban, South Africa, August.

343

Carson, R.T., 1998. Valuation of tropical rainforest: philosophical and practical issues in the use of Contingent Valuation. Ecol. Econ. 24 (1), 15–29. Carson, R.T., 2000. Contingent Valuation: a user’s guide. Environ. Sci. Technol. 34, 1413–1418. Carson, R.T., Nicholas, E.F., Norman, F.M., 2002. Contingent Valuation: controversies and evidence. Environ. Res. Econ. 19, 173–210. Daniel, T.C., Vining, J., 1983. Methodological issues in the assessment of landscape quality. In: Altman, I., Wohwill, J.F. (Eds.), Behaviour and the Natural Environment. Plenum Press, New York, pp. 39–83. Dearden, P., 1980. A statistical technique for the evaluation of the visual quality of the landscape for land-use planning purposes. J. Environ. Manage. 10, 51–68. Deffontaines, J.P., 1985. Etude de l’activite´ agricole et analyse du paysage. L’Espace ´ Geogr. 1, 35–47. ´ Deffontaines, J.P., 1986. Un point de vue d’agronome sur le paysage. Une methode ´ d’analyse du paysage pour l’etude de l’activite´ agricole. In: Foucher Editor. INRAP, “Lectures du Paysage”, Paris, pp. 33–52. De Groot, R., 2006. Function analysis and valuation as a tool to assess land use conflicts in planning for sustainable multifunctional landscapes. Landscape Urban Plan. 75 (3–4), 175–186. ´ DeLucio, J.V., Mugica, M., 1994. Landscape preferences and behaviour of visitors to Spanish national parks. Landscape Urban Plan. 29 (2–3), 145–160. Douglas, T., 1997. Land use changes in the Sierra Nevada of Southern Spain. The Globe 39, 9–10. Drake, L., 1987. The value of preserving the agricultural landscape: result from surveys. Rapport IESSL 289, 108 pp. Drake, L., 1992. The non-market value of the Swedish agricultural landscape. Eur. Rev. Agric. Econ. 19 (3), 351–364. Dunn, M.C., 1974. Landscape Evaluation Technique: An Appraisal and Review of the Literature. Centre for Urban and Regional Studies, University of Birmingham. Dunn, M.C., 1976. Landscape with photographs: testing the preference approach to landscape evaluation. J. Environ. Manage. 4, 15–26. Emery, D.R., Barron, F.H., 1979. Axiomatic and numerical conjoint measurement: an evaluation of diagnostic efficacy. Psychometrika 44, 195–210. Fenwick, I., 1978. A user’ s guide to conjoint measurement in marketing. Eur. J. Mark. 12, 203–211. ´ Garc´ıa, L., Hernandez, J., Ayuga, F., 2006. Analysis of the materials and exterior texture of agro-industrial building: a photo-analytical approach to landscape integration. Landscape Urban Plan. 74 (2), 110–124. Garrod, G., Willis, K., 1998. Using contingent ranking to estimate the loss of amenity value for inland waterways from public utility service structures. Environ. Res. Econ. 12, 241–247. Gary, F., 2001. Multifunctional landscapes—towards transdisciplinary research. Landscape Urban Plan. 57, 159–168. Gilbert, S., Larkin, S.H., 1995. Firm-level intermediate demand for pacific whiting products: a multi-attribute; multisector analysis. Can. J. Agric. Econ. 43 (3), 501–518. ´ Gonzalez, F., 1981. Ecolog´ıa y Paisaje. Ed. Blume, Madrid. Green, P.E., Rao, V.R., 1971. Conjoint measurement for quantifying judgmental data. J. Mark. Res. VIII, 355–363. Green, P.E., Srinivasan, V., 1990. Conjoint analysis in marketing: new developments with implications for research and practice. J. Mark. 4, 3–19. Green, P.E., Wind, Y., 1975. New way to measure consumer’s judgements. Havard Bus. Rev. 53, 107–117. Green, P.E., Wind, J., Rao, V.R., 1999. Conjoint analysis: methods and applications. In: Dorf, R.C. (Ed.), The Technology Management Handbook. CRC Press, Boca Raton, FL, pp. 12–72. Hall, C., McVittie, A., Moran, D., 2004. What does public want from agriculture and the countryside? A review of evidence and methods. Seminar. Multifunctionality agriculture and markets: understanding the critical linkage. Rennes, France, October 27–29. Hall, J., Viney, R., Hass, M., Louviere, J.J., 2004. Using stated preference discrete choice modelling to evaluate health care programs. J. Bus. Res. 57 (9), 1026–1032. Hammitt, W., Patterson, M.E., Noe, F.P., 1994. Identifying and predicting visual preference of southern Appalachian forest recreation vistas. Landscape Urban Plan. 29 (2–3), 171–183. Hanley, N., MacMillan, D., Wright, R., Bullock, C., Simpson, I., Parisson, D., Grabtree, B., 1998a. Contingent valuation versus choice experiment: estimating the benefits of Environmentally Sensitive Areas in Scotland. J. Agric. Econ. 49, 1–15. Hanley, N., Wright, R., Adamowicz, W., 1998b. Using choice experiment to value the environment. Environ. Res. Econ. 11 (3–4), 413–428. Hanley, N., Mourato, S., Wright, R., 2001. Choice modelling approaches: a superior alternative for environmental? J. Econ. Surv. 15, 433–460. Hanley, N., Ryan, M., Wright, R., 2003. Estimating the monetary value of health care: lessons from environmental economics. Health Econ. 12, 3–16. ´ Hernandez, J., Garcia, L., Ayuga, F., 2004. Integration methodologies for visual impact assessment of rural buildings by geographical information systems. Landscape Urban Plan. 88 (2), 255–263. Hull, R.B., Revell, G.R.B., 1989. Issues in sampling landscapes for visual quality assessments. Landscape Urban Plan. 17, 323–330. Hull, R.B., Stewart, W.P., 1992. Validity of photo-based scenic beauty judgements. J. Environ. Psychol. 12, 101–114. ¨ ¨ N., Skanes, H., 2006. Change trajectories and key biotopes-assessing landKayhk o, scape dynamics and sustainability. Landscape Urban Plan. 75 (3–4), 300–321. Kline, J., Wichelns, D., 1996. Measuring public preferences for the environmental amenities provided by farmland. Eur. Rev. Agric. Econ. 23–4, 412–436.

344

S. Sayadi et al. / Land Use Policy 26 (2009) 334–344

Krantz, H., Tversky, A., 1971. Conjoint measurement analysis of composition rules in psychology. Psychol. Rev. 78, 151–169. Laurie, I.C., 1975. Aesthetic factors in visual evaluation. In: Zube, E.N., Brush, R.O., Fabos, J.G. (Eds.), Landscape Assessment: Values, Perceptions and Resources. Dowden Hutchinson and Ross, Stroudsburg, pp. 102–117. Law, C.S., Zube, E.H., 1983. Effects of photographic composition on landscape perception. Landscape Res. 8, 22–23. Lee, T.R., 1990. Attitudes Towards and Preferences for Forestry Landscapes. Report to the Forestry Commission Recreation Branch, Edinburgh. Louviere, J.J., 1988a. Analysing Decision Making: Metric Conjoint Analysis. Sage University Papers Series in Quantitative Applications in the Social Sciences, No. 67. Sage, Newbury Park, CA. Louviere, J.J., 1988b. Conjoint analysis modelling of stated preferences: a review of theory, methods, recent developments and external validity. J. Trans. Econ. Policy 22 (1), 93–119. Louviere, J.J., Street, D., 2000. Stated preferences methods. In: Hensher, D.A., Button, K. (Eds.), Handbook in Transport I: Transport Modelling. Pergamon (Elsevier Science), Amsterdam. Louviere, J., Hensher, D., Swait, J., 2000. Stated Choice Models Methods: Analysis and Applications in Marketing, Transportation and Environmental Valuation. Cambridge University Press, London. Luce, R.D., Tukey, J.W., 1964. Simultaneous Conjoint Measurement: A New Type of Fundamental Measurements. J. Math. Psychol. 1, 1–27. Mackenzie, J., 1990. Conjoint analysis of deer hunting. North-eastern J. Agric. Res. Econ. 19 (2), 109–117. Mackenzie, J., 1992. Evaluating recreation trip attributes and travel time via Conjoint Analysis. J. Leisure Res. 24 (2), 171–184. Mackenzie, J., 1993. A comparison of contingent preference models. Am. J. Agric. Econ. 75 (3), 593–603. Mitchell, R.C., Carson, R.T., 1989. Using Surveys to Value Public Goods: The Contingent Valuation Method. John Hopkins University Press, Baltimore, 461 pp. Morisson, M., Bennett, J., Blamey, R., Louviere, J.J., 2002. Choice modelling and tests of benefit transfer. Am. J. Agric. Econ. 84 (19), 161–170. ´ N., Gibon, A., 2006. Agricultural land-use change an its Mottet, A., Ladet, S., Coque, drivers in mountain landscapes: a case study in the Pyrennees. Agric. Ecosyst. Environ. 114 (2–4), 296–310. Musacchio, L., Ozdenerol, E., Bryant, M., Evans, T., 2005. Changing Landscapes, changing disciplines: seeking to understand interdisciplinarity in landscape ecological change research. Landscape Urban Plan. 73, 326–338. Navrud, S. (Ed.), 1992. Pricing the European Environment. Scandinavian University Press, Oslo, 284 pp. Navrud, S., 2000. Valuation techniques and benefit transfer methods: strengths, weaknesses and policy utility. In: OECD. Valuing Rural Amenities. OECD Proceedings, Paris. OECD, pp. 15–38. Ness, M., Gerhardy, H., 1994. Consumers preferences for quality and freshness attributes of eggs. Br. Food J. 96, 26–34. Palakowski, K.J., 1975. Landscape assessment of the Upper Great Lakes basin resources: a macro-geomorphic and micro-composition analysis. In: Zube, E.N., Brush, R.O., Fabos, J.G. (Eds.), Landscape Assessment: Values, Perceptions and Resources. Dowden Hutchinson and Ross, Stroudsburg, pp. 203–219. Papoulis, A., Pillai, S.U., 2002. Probability, Random Variables and Stochastic Process. McGraw-Hill, p. 211. ´ ´ Pecoa, B., Sanchez, A.M., Azcaratea, M.F., 2006. Abandonment in grazing systems: consequences for vegetation and soil. Agric. Ecosyst. Environ. 113 (1–4), 284–294. Potschin, M., Haines-Young, R., 2006. Rio + 10”, sustainability science and landscape ecology. Landscape Urban Plan. 75 (3–4), 162–174. Price, C., 1978. Landscape Economics. MacMillan, London. Price, C., 1990. Forest landscape evaluation. Paper Presented to a Forestry Commission Economics Research Group Meeting, September 1990, University of York. Pruckner, G.J., 1995. Agricultural landscape cultivation in Austria: an application of the CVM. Eur. Rev. Agric. Econ. 22 (2), 173–190. Rao, K.S., Rekha, P., 2001. Land use dynamics and landscape change pattern in a typical micro watershed in the mid elevation zone of central Himalaya, India. Agric. Ecosyst. Environ. 86 (2), 113–124. Santos, J.M.L., 1998. The Economic Valuation of Landscape Change. Edward Elgar, Cheltenham, UK. Sarapatka, B., Sterba, O., 1998. Optimization of agricultural in relation to the multifunctional role of the landscape. Landscape Urban Plan. 41, 145–148. ´ Sayadi, S., 1998. Analisis de la potencialidad de los sistemas agrarios en el desarrollo local de zonas rurales: el caso de La Alta Alpujarra Oriental. Doctoral thesis, ´ ´ Escuela Tecnica Superior de Ingenieros Agronomos y de Montes, University of ´ Cordoba, 655 pp.

Sayadi, S., Gonzalez, M.C., Calatrava, J., 1999. Estimating relative value of agrarian landscape by Conjoint Analysis: the case of the Alpujarras (South-eastern Spain). In: IX Annual Conference of the European Association of Environmental and Resource Economists (EAERE), Oslo, Norway. ´ Sayadi, S., Gonzalez, M.C., Calatrava, J., 2000. Valuing Agrarian landscape: an application of Conjoint Analysis to specially sensitive Mediterranean mountain areas in South-eastern Spain. In: IAAE XXIVth Congress, Berlin, 13–19 August. ´ Sayadi, S., Calatrava, J., 2001. Analisis funcional de los sistemas agrarios para el desar´ Serie Estudios. rollo sostenible. Ministerio de Agricultura, Pesca y Alimentacion. ´ Secretar´ıa General Tecnica. Madrid-Spain, 336 pp. ´ actual, potenSayadi, S., Calatrava, J., 2002. Agroturismo y Desarrollo Rural: situacion ˜ del Sureste espanol. ˜ cial y estrategias en zonas de montana Universidad de Murcia. Cuadernos de Turismo 7, 131–157. ´ Sayadi, S., Gonzalez, M.G., Calatrava, J., 2005. Ranking versus scale rating in Conjoint Analysis: evaluating landscapes in mountainous regions in South-Eastern Spain. Ecol. Econ. 55 (4), 539–550. Scarpa, R., Hutchinson, W.G., Buongiorno, J., Chilton, S., 1999. Stated willingnessto- pay for forest recreation: Do forest attributes matter? In: Proceedings: 9th annual conference of the European Association of Environmental and Resource Economists, Oslo, June. Scozzafava, S., De Sanctis, A., 2006. Exploring the effects of land abandonment on habitat structures and on habitat suitability for three passerine species in a highland area of Central Italy. Landscape Urban Plan. 75, 23–33. Shafer, E.L., Brush, R.O., 1977. How to measure preferences for photographs of natural landscapes. Landscape Urban Plan. 4, 237–256. Shelby, B., Harris, R., 1985. Comparing methods for determining visitors evaluations of ecological impacts; Site visits, photography, and written descriptions. J. Leisure Res. 17, 57–67. Shutteworth, S., 1980a. The evaluation of landscape quality. Landscape Res. 5, 14–20. Shutteworth, S., 1980b. The use of photographs as an environmental presentation medium in landscape studies. J. Environ. Manage. 11, 61–76. Silvennoinen, H., Pukkala, T., Tahvanainen, L., 2002. Effect of cuttings on the scenic beauty of a tree stand. Scand. J. Forest Res. 17, 263–273. Steenkamp, J.B., 1987. Conjoint measurement in ham quality evaluation. J. Agric. Econ. 38, 473–480. Stewart, T.R., Middleton, P., Dowton, M., Ely, D., 1984. Judgments of photographs vs. field observations in studies of perception and judgment of the visual environment. J. Environ. Psychol. 4, 283–302. Tasser, E., Walde, J., Tappeiner, U., Teutsch, A., Noggler, W., 2007. Land-use changes and natural reforestation in the Eastern Central Alps. Agric. Ecosyst. Environ. 118 (1–4), 115–129. ´ ´ ´ agricoles et Thenail, C., Baudy, J., 1994. Methodes d’etude des relations entre activites paysages. In: Systems-Oriented Research in Agriculture and Rural Development. International Symposium, Montpellier, France, pp. 316–321. ¨ Tyrvanen, L., Hannu, V., 1998. The economic value of urban forest amenities: an application of the Contingent Valuation Method. Landscape Urban Plan. 43, 105–118. European Commission, 2000. European Landscape convention. Florence, 20.x.2000. Van Meijl, H., Van Rheenen, T., Tabeau, A., Eickhout, B., 2006. Agric. Ecosyst. Environ. 114 (1), 21–38. Verburg, P.H., Schulp, C.J.E., Witte, N., Veldkamp, A., 2006. Downscaling of land use change scenarios to asses the dynamics of European landscapes. Agric. Ecosyst. Environ. 114 (1), 39–56. Vos, V., Meekes, H., 1999. Trends in European cultural landscape development: perspectives for sustainable future. Landscape Urban Plan. 24, 63–68. Westhoek, H.J., Van den Berg, M., Bakkes, J.A., 2006. Scenario development to explore the future of Europe’s rural areas. Agric. Ecosyst. Environ. 114 (1), 7–20. Wang, X., Song, B., Chen, J., Zheng, D., Crow, T.R., 2006. Visualizing forest landscapes using public data sources. Landscape Urban Plan. 31, 111–124. Willis, K.G., Garrod, G.D., 1993. Valuing Landscape: a Contingent Valuation approach. J. Environ. Manage. 37, 1–22. Wittink, D.R., Cattin, P.H., 1989. Commercial use of Conjoint Analysis: an update. J. Mark. 53, 91–96. Wherrett, J.R., 2000. Creating landscape preference models using the Internet as a medium for surveys. Landscape Res. 25, 79–96. Woodhousea, S.P., Goodb, J.E.G., Lovetta, A.A., Fullerc, R.J., Dolmana, P.M., 2005. Effects of land-use and agricultural management on birds of marginal farmland: a case study in the Llyn Peninsula, Wales. Agric. Ecosyst. Environ. 107 (4), 331–340. ¨ a, ¨ T., Kola, J., 2004. Consumer preferences regarding multifunctional agriculture. Yrjol Int. Food Agribus. Manage. 7 (1), 78–90. Zonneveld, I.S., Forman, R.T. (Eds.), 1989. Changing Landscapes, An Ecological Perspective. Springer Verlag, New York, p. 286 pp.