Participatory mapping and geographical patterns of the social landscape values of rural communities in Zanzibar, Tanzania NORA FAGERHOLM AND NIINA KÄYHKÖ Fagerholm, Nora & Niina Käyhkö (2009). Participatory mapping and geographi- cal patterns of the social landscape values of rural communities in Zanzibar, Tanzania. Fennia 187: 1, pp. 43–60. Helsinki. ISSN 0015-0010. People attach commonly approved social values subjectively to landscape. These values vary spatially and can be studied in geographical context. In sus- tainable management of cultural landscapes, social values should be taken into account as professionally as the analysis of physical landscape features. This case study applies participatory and GIS techniques in the mapping and geo- graphical analysis of social landscape values in a multifunctional cultural land- scape in Zanzibar island, Tanzania. Social landscape data were collected with single-informant interviews using participatory GIS (PGIS) techniques. Four dif- ferent social landscape values (subsistence, traditional, aesthetic and leisure) were mapped on an orthophotoimage individually by 149 informants. Data were spatially and statistically analysed to construct understanding of the com- munity level patterns of the social landscape values. Results show geographical differences between individually and collectively held values in their distribu- tion and clustering across the landscape. These patterns reflect local culture and its interpretation of different social landscape values. Results address the impor- tance of local stakeholder participation when spatial planning and management of multifunctional cultural landscapes are realized. The paper discusses these management implication and methodological challenges of using participatory GIS techniques in studying cultural landscapes. Nora Fagerholm & Niina Käyhkö, Department of Geography, FI-20014 Univer- sity of Turku, Finland. E-mails: nora.fagerholm@utu.fi, niina.kayhko@utu.fi. Introduction Social values in cultural landscapes Most of the current problems in the management of natural resources in cultural landscapes lie in the interface between people and the environ- ment. Sustainable management can only be achieved if pluralistic land uses under the umbrel- la of long-term social, economic and ecological values are appreciated and taken into account in land use planning (Luz 2000; Potschin & Haines- Young 2006; Raquez & Lambin 2006). There is a need for broader understanding of the complex human-nature interaction in contemporary cultur- al landscapes especially in political decision mak- ing. Solutions for many of these management chal- lenges lie in the actions of the people and the ways they value and use the land. However, there is a great imbalance in how the knowledge and needs of different stakeholders are taken into account in spatial planning. It has been argued that far too little emphasis is still given to the expertise of the local residents and communities in relation to, for example, patterns and qualities of vegetation, soils, species and land cover (Williams & Patterson 1996; Luz 2000; Brown et al. 2004; Black & Liljebald 2006). It is well known that local com- munities play a crucial role in sustainable land- scape management. They possess valuable knowl- edge of the functions and social values attached to cultural landscapes, and this social knowledge is essential when tackling land use and land man- agement issues for better future development. Social landscape values are subjectively experi- enced, place-related and contextual, and tend to vary spatially (Tuan 1977; Zube 1987). Capturing social knowledge in landscapes requires the dedi- 44 FENNIA 187: 1 (2009)Nora Fagerholm and Niina Käyhkö cated participation of local inhabitants (Sauer 1925). In societal processes, the meeting and con- trol of the physical and social environment is cru- cial (Kaltenborn 1998; Luz 2000; Brown 2005; Black & Liljeblad 2006). Spatial data on the social landscape can depict how communities are using the environment and how they perceive and expe- rience it and, as Soini (2001) sees it, mapping ena- bles understanding of differences between the so- cial values of landscapes and natural scientific as- sessments made on them. It is necessary to gain knowledge of the complex social-ecological sys- tems within contemporary landscapes and provide techniques which enable collection, retrieval and analyses of social landscape values in a spatial form (Alessa et al. 2008). Recently, the social meanings of places have started to gain wide inter- est in the context of the geographical analysis of landscapes (Brown et al. 2004; McIntyre et al. 2004; Brown 2005; Black & Liljeblad 2006; Kyttä & Kahila 2006; Brown & Raymond 2007; Gunder- son & Watson 2007; Tyrväinen et al. 2007). As landscapes can be understood both as complex mosaics of the physical environment and social constructions and processes experienced by peo- ple with their senses, contemporary landscape re- search faces a challenge to integrate these ap- proaches, especially for practical landscape man- agement needs (Potschin & Haines-Young 2006). Cultural landscape research has also been criti- cised for concentrating merely on the textual inter- pretation of landscapes which tends to lack the necessary applicable knowledge to landscape management and planning (Olwig 1996; Soini 2001). Potschin and Haines-Young (2006) identify needs for transdisciplinary models and tools in landscape analysis which would serve practical needs in society and support the sustainable man- agement of cultural landscapes. Physical landscape patterns and distribution of natural resources can be quite effectively mapped with the aid of various spatial data sets, such as aerial photographs and satellite images. Recently, the combined use of spatial data has produced an increased understanding of the dynamics and de- velopment of landscapes from the perspective of land use and land cover patterns (e.g. Lambin et al. 2003; Pontius et al. 2004; Käyhkö & Skånes 2006; Hartter et al. 2008). Social values, which are associated with various places in landscapes, on the other hand, are more difficult to measure, since they are built on individually perceived, sub- jective and qualitative information. When people become acquainted with a specific space, this space develops into a place and values are at- tached to it (Tuan 1977). Hence, it can be said that social landscape values emerge from environmen- tal experience (Brown 2005). These values can be e.g. aesthetic, religious, cultural or recreational. Social landscape values have commonly approved meanings, as they are socially constructed. The concept of landscape value can be seen to act as an operational bridge in applied landscape man- agement and planning (Brown 2005). It connects the geography of place, i.e. the location of specific places, with the psychology of place, which refers to the underlying place-related perceptions. It is important to remember, however, that people have different expectations, needs and desires and these influence the ways they attach values and set pref- erences to various places (Relph 1976; Zube 1987). For example, aesthetic values do not con- centrate on the same places for all people because of individual differences in perception and experi- ence. Participatory GIS and mapping of social landscape values Participatory GIS (participatory geographical in- formation systems, PGIS) techniques combine community participation with the use of digital ge- ospatial techniques and enable the collection, storage and analysis of stakeholder data in a geo- graphical form. In practice, PGIS solutions are various, depending on the aims of the application, the level of the information needed and the knowl- edge of the participants. PGIS practices have been commonly used in urban planning and in the al- location of natural resources (e.g. Kingston et al. 2000; Craig et al. 2002; Voss et al. 2004). For the mapping and geographical analysis of social val- ues attached to landscapes, the use of participa- tory GIS techniques is a useful approach. Land management challenges are typical examples where stakeholder participation is needed in a geographical form. In many developing countries, for example, information on the local social values on cultural landscapes is completely missing, and natural resources are under constant pressure from various stakeholders (MA 2003; FAO 2006). For rural developing communities, a sustainable land- scape has a multiple social and economic impor- tance by providing e.g. life support, energy, shel- ter, food and means of income (Sitari 2005; Käyh- kö et al. 2008). Hence, understanding of the geo- FENNIA 187: 1 (2009) 45Participatory mapping and geographical patterns of the social … graphical patterns and variation of social values on the land is urgently needed in the circumstanc- es where traditional and new stakeholders meet and share the use of natural resources. Such a baseline understanding is essential for the sustain- able management of multifunctional landscapes and could be potentially integrated with geograph- ical data of physical resources. Participatory mapping of social landscape val- ues has been approached both from the individual and group data collection perspectives (e.g. McIntyre et al. 2004; Gunderson & Watson 2007). Methods, such as sticker dots, point markers and polygon delineations on the maps have been ap- plied (McIntyre et al. 2004; Brown 2005; Black & Liljeblad 2006; Tyrväinen et al. 2007) and in some cases data have been collected via map interfaces through the Internet (Kyttä & Kahila 2006). One particular challenge for PGIS is the application of ambiguous data set structures in the participatory mapping efforts. For example, social values are of- ten continuums rather than discrete points and patches in the landscape. In previous studies, social value typologies have been approached from different starting points. For example, Alessa et al. (2008) used 14 land- scape values, namely aesthetic, biological, cultur- al, economic, future, historic, intrinsic, learning, life sustaining, recreational, spiritual, subsistence, therapeutic and wilderness, in the mapping of the social spaces. This value typology has been modi- fied by Brown and colleagues in several case stud- ies (Brown & Reed 2000; Brown 2005; Raymond & Brown 2006) and is originally founded on the work of Rolston and Coufal (1991). Tyrväinen et al. (2007) used 11 different values, such as valua- ble nature site, forest feeling and unpleasantness in mapping urban green areas in Finland. The val- ues were based on previous Swedish studies re- garding the social values of open spaces (Region- plane- och trafikkontoret 2001; Ståhle & Sandberg 2002, cit. Tyrväinen et al. 2007). In addition, Man- ning et al. (1999) have proposed 11 human prefer- ence-based values for national forests and Tarrant et al. (2003) a 12-point scale to measure public values of national forests. This case study explores possibilities of apply- ing participatory and GIS techniques for the map- ping and geographical analysis of social values in multifunctional cultural landscapes, especially in the context of a developing society. Through stud- ying social landscape values it is possible to estab- lish an understanding of the geographical patterns of the social values; what kind of patterns the so- cial landscape values form, what and where the most important areas in the social landscape are, how the values might change and modify cultural landscapes and how the social spatial data reflect the land cover data of the physical environment. The study has three main objectives. Firstly, to map social landscape values of the local inhabit- ants in the village of Matemwe, Zanzibar (Tanza- nia) based on single-informant interviews and par- ticipatory GIS techniques. Four social values, namely subsistence, aesthetic, traditional and lei- sure, were selected for the study because these were considered essential in the social landscape of the local community. Secondly, to analyse and compare the geographical patterns of these social landscape values, and thirdly, to identify the most important characteristics of the social values, which could contribute to the sustainable plan- ning and management of multifunctional cultural landscapes, such as Matemwe. Furthermore, the paper discusses methodological aspects of the use of participatory GIS techniques for the spatial analysis of social values. Description of the case study site The Zanzibar islands are located in the eastern coast of Tanzania, approximately 30 km north-east from Dar es Salaam. The population of the main island Unguja (Zanzibar) is estimated at about 700,000 people and is growing approximately with an annual rate of 3.1% (Tanzania Sensa 2003). The Zanzibar islands have a tropical mon- soon climate, with two rainy seasons from March to May and October to December and an average annual temperature of 26 C° (Hettige 1990). The contemporary Zanzibar landscape is a mosaic of indigenous and cultivated vegetation, which ex- presses the combined and long-term influences of different cultures and land use activities, such as spice farming and shifting cultivation. The Zanzi- bar islands have experienced dramatic changes in land use and land ownership throughout their his- tory (Lofchie 1965). Today, the “environment is being more heavily utilised than ever before” and the fast growing tourism since the early 1990s is one significant contributing factor to this (RGZ 2004). The administrative region (swa: shehia) of Matemwe is situated in the north-eastern coast of Unguja Island (Fig. 1). Matemwe consists of sev- 46 FENNIA 187: 1 (2009)Nora Fagerholm and Niina Käyhkö eral sub-villages with a total population of about 7300 (Tanzania Sensa 2003). Geologically, Matemwe lies in the coral rag area, where bedrock consists of exposed, porous coral and where loose soil deposits are generally shallow and mainly found in the crevasses of the bedrock. The major- ity of the coral rag forests are characterised by ferns, grasses, indigenous trees and scrubs (ZFDP 1997) with marginal opportunities for permanent agriculture (Commission for Land and Environ- ment 1995). Shifting cultivation is practiced wide- ly across the forested and scrub covered land as the traditional form of agriculture, and occasional permanent fields and agroforests can be found in the vicinities of the villages. The cultivation cycle is short (3–5 yrs) and in many places fields are shifted even annually. Coral rag forests provide important livelihood services, such as firewood, extraction of coral and building poles for construction, materials for hand- icrafts, medicinal plants and sites for practicing tra- ditional beliefs (Sitari 2005; Käyhkö et al. 2008). Thus, the socio-economic importance of the forest products is high for the local communities, but due to multiple uses, there is general concern over the long-term sustainability of the forest resources (ZFDP 1997; RGZ 2004). Sea resources bring ad- ditional sources of livelihood for the villagers, es- pecially through fishing and seaweed farming (Käyhkö et al. 2008). As tourism is rapidly intensi- fying along the coastal fringe, it influences both the forest and sea related livelihoods of the villagers. Tourism potentially creates new opportunities for employment and the market in general, but tourist facilities also push local people to migrate inland, sell their lands, change areas for cultivation and restrict access to beach areas and sea resources in particular (Gössling 2002; Mustelin 2008). Material and methods Overall study design The mapping and analyses of the social values of the Matemwe communities consisted of several Fig. 1. Study site on the eastern coast of Tanzania and land cov- er in the shehia of Matemwe in 2004. There is no clear bound- ary for the shehia and therefore, for the purpose of the study, a fixed boundary was established. FENNIA 187: 1 (2009) 47Participatory mapping and geographical patterns of the social … work phases. Initially, the theoretical considera- tions and practical preparations for the study were made, including a literature review of the practical approaches of PGIS and selection of the social landscape values. Four social values, subsistence, aesthetic, traditional and leisure, were chosen for the study of the social dimension in the landscape among the local communities of Matemwe. These values were based on commonly practiced land use activities as well as the values they attach to their village landscape which were grouped under a typology of four social landscape values (Table 1). Comparisons to similar research papers were also made when formulating the value typology applicable to the Matemwe case, even though most previous studies concern developed societies (Brown & Reed 2000; Brown 2005; Raymond & Brown 2006; Tyrväinen et al. 2007; Alessa et al. 2008). Once the theoretical setting was formulated, the research approach was locally adjusted for Matemwe through discussions with the village leader (swa: sheha) and planning officers in the Department of Commercial Crops, Fruits and For- estry (DCCFF), which is a government department under the Ministry of Agriculture, Livestock and Natural Resources in Zanzibar. A combination of participatory mapping with semi-structured inter- view questions was chosen for the data collection technique (see e.g. Black & Liljeblad 2006; Gunderson & Watson 2007). Social values were collected individually as geographical information from each of the informants because of the subjec- tive nature of the information, but the data were analysed collectively to identify the geographical patterns of the values across the whole landscape of the study area. Subsequently, the results were shown to a group of 20 informants in a reflective focus group with lively discussion. The focus group discussion played an important role in in- forming the community members and raising dis- cussion among them. It also assisted in the inter- pretation of the results. Together with the case study, these participatory methods and the signifi- cance of social data in landscape planning were also introduced to the planning sector in Zanzi- bar. Participatory mapping and the interviews The spatial collection of the social value data was organised through a participatory mapping cam- paign, which took place in Matemwe in Novem- ber–December 2007. This PGIS campaign was based on the use of the most recent digital geo- referenced aerial photographs (2004, 0.5 m pixel size), which were obtained from the Department of Survey and Urban Planning (DoSUP) in Zanzi- bar town. Aerial photographs have been found useful in PGIS campaigns since they are visually attractive without too much abstraction of the landscape (Corbett et al. 2006). The use of aerial photographs was tested in earlier PGIS campaigns in Zanzibar and found useful and reliable for loca- tion-specific tasks given to the community mem- bers (Makandi 2008). Aerial photographs were printed at a scale of 1:5000 on a laminated paper sheet for data collection. A total of 149 community members from all the 21 sub-villages of Matemwe were interviewed. Table 1. Social landscape values and their respective activities/indicators and interview questions used in the study. Social value Activity/indicator Interview questions to locate the activities subsistence shifting cultivation Do you or your family cultivate crops or sea- weed, where?seaweed cultivation grazing Where are your grazing areas? collection of firewood, construction materials, medici- nal plants, wild fruits/vegetables, building poles for sell- ing, coral rock for making lime stone, hunting areas Where do you collect forest products? traditional religious or sacred place Are there religious or sacred places for you in the landscape, where? aesthetic beautiful, attractive place Where are the most beautiful places here? leisure social interaction, recreation Where do you go on your spare time? Are there e.g. some important meeting places for you or do you go to the surroundings? 48 FENNIA 187: 1 (2009)Nora Fagerholm and Niina Käyhkö The boundaries of the recent census (2002) enu- meration areas were not available, and thus the amount of persons sampled in each sub-village was determined according to the relative amount of buildings in the sub-villages according to the 2004 aerial photograph interpretation. This pro- vided the only applicable spatial estimate of the population distribution in Matemwe. Informants were selected from the sub-villages by the sheha or his assistant who were given detailed instruc- tions of the amount of participants, their age (15–30 yrs, ≥31 yrs) and gender division for each sub-village. Because most of the people have sev- eral livelihoods (Sitari 2005), this was not included in the selection criteria. The informants were se- lected on the same day or the day before the inter- view situation. Each participant received a small monetary compensation since they were not able to attend their normal daily activities while wait- ing to be interviewed. The interviews were made by two local field as- sistants from the DCCFF. Local field assistants were used since it was considered essential that the in- terviews were conducted in fluent Swahili. Before the interviews, the questionnaire had been trans- lated from English into Swahili and the social value concepts and related interview questions (Table 1) were discussed with the field assistants to ensure that the same understanding of terms and concepts was shared. The semi-structured interviews lasted 20–40 minutes for each informant and started with an introduction to the topic of the interview, orien- tation with the aerial photographs and collection of the background information (e.g. age, house- hold details, source of living, and level of educa- tion). Then the participants marked their homes onto the printed aerial photographs and continued marking social landscape values one by one as polygon delineations using drawing ink. For each value mapped the participants were allowed to mark as many places on the map as they wanted to. Via supplementary questions, such as what crops they cultivate and how often they change the field location, additional attribute information was associated with the value polygons. All delin- eated polygons had unique identifiers in terms of the social values and these were linked with unique informant identifier codes to each person interviewed. Social value delineations were transformed into cell-based geographical information immediately after each interview since informants marked their values on the same map sheet but were not al- lowed to see each other’s delineations. The data transformation was done manually in the field as follows. The image map which contained inform- ant delineations was overlaid with a transparent grid (Fig. 2). Each grid cell (50x50 m, 0.25 ha) had a unique identifier number. If the delineated value polygon covered at least one third of a cell area, the cell code number was attached with the social value polygon. Informants’ homes were marked with one cell accuracy. Subsequently, unique identifiers could be used to transform manual cell data into digital data in GIS. Methods of analysis Analogue spatial data sets collected in the field were recorded on the computer after each inter- view day. This MS Excel 2003 database included informant background as well as data on the de- lineations for each value (Fig. 2). The general char- acteristics of the informants were analyzed with Fig. 2. Spatial data collection of social landscape values with participatory mapping and data transformation from manual field notes into digital form and GIS datasets. FENNIA 187: 1 (2009) 49Participatory mapping and geographical patterns of the social … SPSS 14.0 using descriptive statistics and cross tabulations. These statistics were used to obtain an overall understanding of the informants for the in- terpretation of the social landscape value data. Descriptive statistics were also analysed from each of the four values in MS Excel. However, given the data collection methodology and the approximate character of social value data in geographical form, these measures, especially data on the size of the delineated areas, should not be interpreted as exact but rather as uncertain geospatial data of social values (MacEachren et al. 2005). The geographical data on the informants’ homes and their individual social value delineations were converted into digital GIS data (in ArcGIS 9.2/9.3) on the basis of the unique cell identifiers (num- bers) (Fig. 2). Each value area (minimum one cell, 0.25 ha) and home data were stored as vector pol- ygons, which were rectangular in shape. Addition- al information related to the social landscape value delineations was stored as attribute data for poly- gons representing one delineated social value. To analyse the geographical distributions of each social landscape value, individual informant- specific delineations were grouped into social value layers in GIS for spatial analysis and visuali- sation. Since it was expected that the physical dis- tance between informant homes and value loca- tions might explain some of the variation in the geographical distribution of the values, the ap- proximate distances between informant’s home and corresponding social value delineations were calculated as the straight-line distance using Hawths’s analysis tools (Beyer 2004). The calcula- tions were made from the middle points of the polygons (0.25 ha). Each social landscape value layer (4 layers) contained two types of geographical data: the presence (cell value 1/0) and intensity (total number of informants’ entries for each cell) of the social value. The social value intensity data were spatially analysed with a selection of landscape metrics using Fragstats 3.3 software. The purpose of the analysis was to describe the geographical patterns of each social value in landscape (within the whole study area). The intensity values were classified into four classes based on natural breaks in the data and converted into raster format to be handled in the software. Total patch area, number of patches and patch area mean, range and stand- ard deviation were calculated with the 4-neigh- bour rule. Mean Euclidian nearest neighbour dis- tance was calculated for each intensity class based on the nearest straight-line neighbours of each patch. This analysis was used to measure patch context and isolation at landscape scale. To analyse the overall diversity of the social val- ues at landscape scale, all four values were com- bined into one GIS vector layer. The diversity and relative occurrence of the overlapping four social values were analysed with the Shannon diversity index which is a popular measure of species diver- sity and has also been used to study social data (Krebs 1989; Reed & Brown 2003). The index was calculated to every social value cell based on the relative amount of informant entries of each value in the cell. The index does not have a specific range but is dependent on the richness and occur- rence of social landscape values. From the vector intensity grids, the spatial distri- bution (clustering vs. distribution) of the values was analyzed using a hot spot analysis. Getis-Ord Gi* statistics were calculated for each cell based on the summed intensity (total amount of overlap- ping informants’ entries in every cell) (Haining 2003). Looking at the neighbouring cells, the sta- tistics calculate where features of high value and features of low value tend to cluster in the study area and compares this to random chance. Clus- ters of high values represent hot spots of social landscape value intensity. The outcome from the analysis is a statistically significant Z score. The larger the positive Z score is (i.e. how many stand- ard deviations away from the mean it is), the more intensive is the clustering of high values and vice versa. A Z score near zero indicates no apparent concentration of the intensity values. Based on the distance at which spatial autocorrelation peaks, a threshold distance of 100 meters was used in the analysis. In this study, the confidence level of 95% was used and the intensity value hot spots were identified as those areas where the Z score is more than 1.96 standard deviations away from the mean. Results Characteristics and livelihoods of the informants Altogether 149 informants were interviewed in this case study (Fig. 3). According to the gender and age, 50% (74) were men and 50% (75) wom- en with a mean age of 35 years. Approximately half of the informants (47%) were over 30 years 50 FENNIA 187: 1 (2009)Nora Fagerholm and Niina Käyhkö the youngest informant being 15 and the oldest 80 years old. The majority were married (72%) and around 20% were single, 1% divorced and 5% widowed. Households were large with 5 to 10 members (73% of informants) and 4 children on average (max 19). A quarter of the informants did not have children. The education level among adults was low. Half of the informants had no for- mal education, 12% had some elementary educa- tion and 5% had completed elementary school. Secondary education was completed by one third (32%) of the informants and one of the informants was a high school graduate. The informants commonly depended on several livelihoods. The main livelihoods were subsist- ence farming (71%), fishing (19%) and seaweed cultivation (43%). In general, men seemed to prac- tise fishing and women cultivation of seaweed. Only one of the male informants said he helped his wife with seaweed cultivation and only some women practised fishing. Subsistence farming was practised by both genders and the main form of farming was shifting (slash and burn) agriculture. Livestock, such as cows, goats, chicken or ducks were maintained by only 13% of the informants, primarily in the sub-villages towards the inland re- gions. Some (7%) practised small-scale business such as selling chicken soup. Only one out of ten informants earned a salary through working for the hotels, schools, or shops or as drivers and tailors. Social landscape values and associated activities The informants delineated a total of 989 areas on the aerial photographs during the participatory Fig. 3. Location of the informant homes (n = 147, cells = 111) and the total number and gender distribution of the informants per sub-village. FENNIA 187: 1 (2009) 51Participatory mapping and geographical patterns of the social … mapping campaign. These consisted of places which represented subsistence, traditional, aes- thetic and leisure values of the local inhabitants. The highest response rate (n %) was established for the traditional and lowest for the aesthetic values (Table 2). Subsistence value primarily concerns those land uses which satisfy the basic daily needs of the community members in Matemwe. Subsistence value consists of four land use activities, namely agricultural fields, grazing areas, seaweed cultiva- tion and collection of the forest products. On aver- age, the informants had one to two agricultural fields in cultivation with four to five different food crops. This cultivation pattern is typical for the coral rag areas, where fields are shifted regularly and where mixed crops are cultivated. The most common crops are maize, cassava, sorghum and different kinds of peas. Livestock consists mainly of goats, cows, chicken and ducks which are kept freely in the forest areas and around the home but gathered together for the nights. Various forest products, such as firewood, construction materials and medicinal plants, are collected from the sur- rounding environment. Some men also practice hunting and extract coral stone for lime making. Women cultivate seaweed in the tidal area, and sell dried harvest for industrial purposes. Traditional value primarily relates to religion, and plays an important role in the rural communi- ties. Hence traditional value was readily mapped by almost all of the informants (Table 2). Typical traditional places are graveyards (86%) and sacred sites for practising traditional natural religions (12%) alongside Islam. One of the drawn tradi- tional places also represented a place where a sor- cerer practises his healing traditions. Aesthetic value was attached to sites where the informants have the possibility for social interac- tion or finding business opportunities and shop- ping (together 53%). Beautiful views, the beach and occasional newly-built private buildings and the cooling sea breeze were mapped as aesthetic only by approximately one quarter of the inform- ants. Home was mentioned as a beautiful place by 15% of the informants, mainly women. The low response figure (Table 2) indicates that a large part of the informants found it challenging to map beautiful places. The places where the informants spend their lei- sure time were different between men and women. Men have their own meeting places (swa: maska- ni), where third of the informants (30%) meet to change ideas and play board games. These sites are usually located in central places in the sub- villages. Popular meeting places are also around the shops in inland Mfuru Matonga and at the fish market in coastal Kigomani (see Fig. 3). Half of the informants (50%), usually women, spent their spare time at home doing home chores and weav- ing baskets. Some attended Quran school classes (8%) in their leisure time and young men liked to play soccer on the soccer grounds (9%). Geographical patterns of the social landscape values Most of the informants marked one to two areas per value (for subsistence value per land use activ- ity) on the orthophotomap (Table 2). The average Table 2. Statistics on social landscape value delineations (n = number of informants, n (%) = relative number of informants, SD = standard deviation, intensity 1 (%) = relative amount of cells that have no overlapping area delineations). No. of No. of Areas/ Area size (ha) Aver dist. to Inten. n n (%) cells areas inform. aver min max SD home (m) 1 (%) Subsistence 110 73.5 640 492 4.47 0.45 0.25 2.25 0.27 685.5 69.8 Fields 138 92.6 256 158 1.14 0.42 0.25 1.00 0.21 523.8 96.5 Grazing areas 52 34.9 104 66 1.27 0.41 0.25 1.75 0.29 356.0 97.1 Seaweed cult. 107 71.8 130 111 1.04 0.41 0.25 1.00 0.19 1219.1 75.4 Forest prod. 141 94.6 303 157 1.11 0.56 0.25 2.25 0.37 643.0 96.7 Traditional 146 98.0 205 206 1.41 0.44 0.25 2.25 0.29 695.5 59.8 Aesthetic 93 62.4 166 144 1.53 0.51 0.25 2.00 0.35 1974.2 62.7 Leisure 131 87.9 133 147 1.12 0.36 0.25 1.50 0.24 394.5 72.9 52 FENNIA 187: 1 (2009)Nora Fagerholm and Niina Käyhkö area size of the delineated areas ranged between 0.4 and 0.6 hectares, which is equal to one to two cells. In general, leisure places seemed to be the smallest in size and areas for the collection of the forest products the largest. The standard deviations for the sizes of the areas were highest for the col- lection of forest products and aesthetic places. This is explained by the varying nature of the sites at- tached to these values (e.g. spot-like houses and sea front area). Seaweed cultivation and agricul- tural fields are generally small and compact, and they also have the smallest variation in size be- tween the informants. The subsistence value had the highest coverage (160 ha), number of patches (360) and largest geo- graphical distribution of the four mapped values (Fig. 4, Table 3). The proportion of the cells with intensity value 1 (only one informant entry in the cell) is highest (75–97%) (Table 2) and the nearest neighbour distance between subsistence patches was also notably shorter compared with the other values (Table 3). The components of subsistence value all tended to be individual delineations for subsistence uses with little overlap between the informants (Fig. 5). Coral rag forest has high indi- vidual subsistence value for the community mem- bers of Matemwe, which is why their geographical patterns in the forest areas surrounding the sub- villages were scattered and quite evenly distribut- ed. Geographically, distribution of seaweed farm- ing was distinct and concentrated on the tidal zone with long average distances to homes. Agricultural activities and use of the forest products, on the other hand, concentrated in the vicinity of the in- formant homes but were geographically fragment- ed (Table 2 and Fig. 5). Although all the values mapped in Matemwe included direct use of land, subsistence could be considered as the most important value for the livelihoods of the villagers. However, it was not located nearest to the informants’ homes. This is contrary to previous studies made in developed context, which have pointed out that values which include direct or active use of the land are located nearest to home sites (Brown et al. 2002). For the villagers in Matemwe, it is not always possible to obtain fields or keep livestock and to collect forest products in the immediate vicinity of the home be- cause of the limited space and the characteristics of the coral soil. Hence, these activities need to be scattered in the surrounding forest areas. During the mapping process, the informants de- lineated altogether 206 areas for traditional values with an average size of 0.4 ha (Table 2). Approxi- mately 40% of the traditional sites overlap be- tween the informants, which means that traditional value has the least amount of individual delinea- tions between the informants. This results from the fact that graveyards and scared places are normal- ly shared with several families. Traditional sites are spatially fragmented, but concentrate around the sub-villages within approximately 700 m from the informants’ homes (Fig. 4, Table 2). Aesthetic and leisure values concentrated on sub-villages. In general, leisure values tended to cover small areas, and there were many single cell entries, which were marked as appealing leisure spots on the aerial photographs (Fig. 4). In total, there are 82 leisure patches, which cover an area of 33.3 ha (Table 3). Because leisure time is usu- ally spent near homes, the mean home distance to delineated leisure places was short, around 400 meters. Beautiful places are often shared between the informants and thus had the highest geograph- ical concentration and intensity in the studied landscape (Table 3, Fig. 4). The amount of deline- ated aesthetic places was 144 and their size was on average 0.5 ha (Table 2). In total, the aesthetic places cover an area of 42.3 ha of Matemwe land- scape (Table 3). While leisure spots are found in most of the sub-villages, aesthetically appealing places seemed to be concentrated along the coast- Table 3. Landscape metrics on social landscape value patch mosaic (ENN dist. = Euclidian nearest neighbour distance). Tot. Patch Patch area (ha) ENN dist. (m) area (ha) no. mean range SD mean range SD Subsistence 160.00 360 0.44 2.00 0.32 162.9 3104.6 220.1 Traditional 51.00 111 0.46 1.50 0.31 226.6 1499.3 254.3 Aesthetic 42.25 71 0.60 4.75 0.77 309.4 2396.1 537.1 Leisure 33.25 82 0.41 2.00 0.33 248.3 1183.3 217.8 FENNIA 187: 1 (2009) 53Participatory mapping and geographical patterns of the social … Fig. 4. Geographical distribution of the delineated areas, the intensity of overlapping areas and inten- sity range for the four social landscape values in Matemwe. 54 FENNIA 187: 1 (2009)Nora Fagerholm and Niina Käyhkö Fig. 5. Geographical distribution of the delineated areas, the intensity for overlapping areas and inten- sity range for the components of subsistence value in Matemwe. FENNIA 187: 1 (2009) 55Participatory mapping and geographical patterns of the social … line with the longest distances (1974 m) from the homes of the informants (Fig. 4, Table 2). Howev- er, distribution of the aesthetic value overlapped with that of leisure value and it is evident that among the villagers of Matemwe beautiful places are for large part related to social interaction. Together, the four social landscape values cov- ered 262 ha (1049 cells) of Matemwe land and tidal area (Fig. 6A). Values seemed to have little overlap as over 90% of the cells illustrate the presence of one value and the diversity index was zero. The highest diversity class (1.50−2.00) rep- resented only less than one percentage of the cells and all four values coexist only in two cells (di- versity index 2.0). In general, the social value di- versity is low in Matemwe and the few cells of high diversity were scattered in the major sub- villages. When the intensities (total amount of informant entries/cell) from each social landscape value were taken into account, there were eight geo- graphical hot spot clusters of the social values (Fig. 6B). These were mainly sub-villages, such as Kigo- mani, Matemwe Mtakuja, Mbupurini and Mfuru Matonga, or graveyards (see Fig. 3). One specific hot spot was the area of an international hotel in the southern part of Matemwe with high aesthetic value. Hot spot areas varied from 1.25 to 4.25 ha with a mean size of 2.4 ha and a total coverage of 21.5 ha (8.2% of the total area covered by the so- cial values). All hot spot sites were located in or nearby the sub-villages, where the highest intensi- Fig. 6. A) Value diversity as Shannon index and B) hot spot map as the Gi Z score deviation from standard deviation for subsistence, aesthetic, traditional and leisure values. 56 FENNIA 187: 1 (2009)Nora Fagerholm and Niina Käyhkö ties of leisure, traditional and aesthetic values could be identified. These meeting places, grave- yards and beautiful places represent shared social space of Matemwe community and are mainly lo- cated on the coastal area. The subsistence value overlapped the hot spot areas only by one hectare (four cells). It must be kept in mind, however, that the hot spots do not alone depict important areas for the community since individual subsistence values are crucial for the livelihoods of the com- munity members. It is evident that, in contrast to subsistence val- ue, traditional, aesthetic and leisure values are collective in character and they are clustered into the same socially meaningful sites in the village. Such sites act as key nodes for social activities and are also aesthetically appreciated places among the community members and appeared as hot spots in the analyses. This fundamental difference in the meaning of subsistence and other social val- ues is the reason why there was little spatial over- lap between them. Discussion Social landscape values and management of multifunctional cultural landscapes The participatory mapping and geographical anal- ysis of four social landscape values has broadened the understanding of the social landscape of the rural community of Matemwe. These results have several implications for the development of multi- functional cultural landscapes. In the following section, the most important characteristics of these social values and their potential contribution to future landscape planning and management are discussed especially in the light of the situation in Matemwe and Zanzibar, but also in the wider con- text of developing societies. The most significant result of this paper is the difference in the geographical patterning of the four studied social values and how this influences landscape development and planning. While sub- sistence value was individual-based and scattered in the landscape, other values tended to cluster and be shared between the informants. These pat- terns of the social landscape values reflect funda- mental activities of the local inhabitants the long- term influences of the land uses to the present landscape patterns, both physical and social. Sub- sistence farming, grazing and other uses of the for- ests form the basis of the livelihoods of the local communities and create a constant element of change in the landscape. One of the most alarming influences of this dynamic land use pattern is the gradual diminishing, even loss of the essential for- est resources. Especially along the coastline, where the largest sub-villages are located, forests have clearly deteriorated during the last decades (Käyh- kö et al. 2008; Käyhkö & Fagerholm, submitted). For land use planning, scattered and dynamic subsistence farming is challenging to manage be- cause natural resources should be maintained not only for the purpose of the community livelihoods but also for the essential ecosystem services they provide at the local, regional as well as on a global scale (UNEP 2007). The changes in the state of for- est resources has been studied quantitatively to find the key areas of forest loss and recovery but in shifting cultivation landscapes there is a call for research in multiple disciplines to find the drivers of the changes (Hartter et al. 2008). This knowl- edge, together with a supportive policy environ- ment directing investments in local participation, alternative livelihood development, protection area allocation and technology and infrastructure investments can reduce the pressure of shifting cultivation on forests (Müller & Zeller 2002). In contrast to the scattered subsistence values, traditional, aesthetic and leisure values were clus- tered to places characterised by intensive social interaction and cultural traditions. The well-being of the local people is dependent on these meeting places, graveyards and aesthetically appreciated sites and changes in these key nodes would have a significant effect on the social landscape in Matemwe. These areas, where collective social landscape values meet, have stability established through the long-lasting past and present interac- tion of several community members and in devel- opment and planning processes the implication of this shared social landscape is essential. The study has shown that culture has a signifi- cant effect on the interpretation of social landscape values. In a rural developing society such as Matemwe, subsistence value is so crucial for the livelihoods of the local communities that it also affects the perception and experience of other so- cial values. For example, aesthetic value was re- lated more to social interaction than to the visual view. It is evident that most of the villagers in Matemwe do not experience nature in an aesthetic way as something they would interpret as beauti- ful with their eyes. In contrast, nature is mainly FENNIA 187: 1 (2009) 57Participatory mapping and geographical patterns of the social … seen as a resource because it has high utility im- portance in daily life, also noted by Gössling (2002) in a study in coastal Zanzibar. The results from this study are essentially different from social values studies made in a developed context where aesthetic places are among the most important and easily mapped (e.g. Brown & Raymond 2007; Tyrväinen et al. 2007). It seems that rural subsist- ence-based communities hold a different concept of aesthetics compared to a society where direct contact to nature is not predominant. Another landscape planning and management challenge emerging both from the results of the analyses as well as from other social value studies is the interaction between local communities and other stakeholders active in the area. The coastal area in Matemwe seems to be an essential part of the social landscape of the community where so- cial values cluster. However, the very same coastal space is today occupied by hotels and internation- al tourists as a substantial part of the eastern coast- line in Zanzibar is being sold to tourism entrepre- neurs from abroad (RGZ 2004). The coexistence of the traditional and new land uses is evident in Matemwe (Käyhkö et al. 2008) and raises ques- tions about sustainable landscape development and management. Due to hotel construction, some settlements have been forced to move inland and traditional religious places, such as graveyards have been transferred away from the coast. Addi- tionally, access to the sea resources has been re- stricted, especially in front of the hotel areas (Gössling 2001; Mustelin 2008). These develop- ments are against the national principles where regarding tourism development “prior and tradi- tional right of use and access to land” is recog- nised for the local communities (RGZ 2003). For the future planning of rapidly changing coastal landscapes, these local and multifunction- al needs and expectations should be taken seri- ously and efforts to increase positive interactions between stakeholders should be introduced. The local communities and tourism need to find a way to share the same land successfully to mutually benefit each other. An initial starting point would be to respect and appreciate the geographical meanings of the social landscape values. For the practical needs in landscape management, knowl- edge of landscape perceptions are needed (Soini 2001). Participatory techniques, where multiple stakeholder preferences are collected and ana- lysed in a spatial form, would be one possible way of trying to solve conflicting land uses. This case study included only the community people in Matemwe but it would have also been interesting to include other stakeholders, such as the hotel and private land owners. Such analyses would be interesting to implement in the future. There is al- ready evidence of successful efforts of combining social values held by varying stakeholders to allo- cate future tourism growth and development (Ray- mond & Brown 2007). It is a well acknowledged fact that sustainable management of multifunc- tional cultural landscapes can only be achieved if participation of local stakeholders with their mul- tidimensional and pluralist values are included in the process. Methodological considerations of participatory mapping of social landscape values Participatory mapping is a valuable tool to capture spatial information on social landscape values at local community level (Soini 2001; Brown 2005; Black & Liljeblad 2006; Kyttä & Kahila 2006; Gunderson & Watson 2007; Tyrväinen et al. 2007). The method used in Matemwe enabled the com- munity, who do not have experience in landscape management, to produce valuable data of the so- cial values attached to their land and living space. Furthermore, it can be concluded that the partici- patory mapping process itself had value in capaci- ty-building among the villagers. Participation is stated in several national documents of Zanzibar, such as the National Land Use Plan (Commission for Land and Environment 1995), the Indicative Tourism Master Plan for Zanzibar and Pemba (United Republic of Tanzania 2003) and the Zan- zibar Tourism Policy Statement (RGZ 2003), but thus far its practical implementation has remained modest. Some interview based participation has been realised but this case study is among the first to collect stakeholder data in spatial form. With spatial analysis of social landscape values, it was possible to establish understanding and produce data of the geographical patterns and the variation of social landscape values and depict areas that are especially important in the social landscape of Matemwe. However, the methodology used has some aspects regarding representation, informant sampling and data collection, which need to be considered and discussed. Mapping of social values was done with the aid of aerial photographs. In contrast to abstract map 58 FENNIA 187: 1 (2009)Nora Fagerholm and Niina Käyhkö representations, which have been found to be problematic in participatory GIS campaigns (e.g. Zurayk et al. 2001), use of the image map was suc- cessful, since participants were able to read and identify places and areas on it with little support (see also Taylor et al. 2006). It seems the people in Matemwe have good site knowledge and have come accustomed to their environment in their daily lives. Mapping scale has an effect on the size of the social value delineations made by the in- formants and hence area sizes vary in different studies (Black & Liljeblad 2006). In this study, the social value delineations made by the informants were in general quite small in size and seem to be area specific because of the large scale orthopho- toimage. Interpretation of individually identified social landscape values is challenging as one needs to estimate the representativeness of the samples used in the analyses both in terms of their geo- graphical distribution and content. Originally, the informant sampling was planned to be accom- plished according to the proportion of inhabitants on the basis of the 2002 census enumeration data. However, census areas were geographically inad- equate and thus, the sampling was based on the relative amount of buildings according to a 2004 aerial image. This method has shortcomings as the amount of buildings does not really reflect the number of inhabitants. Furthermore, selection of the informants was in the hands of the village lead- er, sheha, and his assistant. It is possible that some of the informants were their relatives. However, as most of the sub-villages are clan-based i.e. they are formed around the same family or a couple of families, it is not likely that such subjective selec- tion had too much influence on the geographical representation of the sample. Additionally, the monetary compensation for the informants could have generated some prejudice in the results, but had compensation not been given it would have certainly lowered the motivation to participate in the mapping process. The collection of data with the grid sheet was experienced to be appropriate and effective in the context of Zanzibar. With a smaller amount of in- formants other methods of data collection (e.g. us- ing transparent sheet on which the mapping would be done) would also have been worth considering. The method of data collection for the social value delineations has not been used previously and it can be argued that some of the precision of infor- mation was lost when data were collected with the 50x50 m grid sheet. It is obvious that some of the areas drawn on the map were smaller than the 0.25 ha grid cell which was the minimum resolu- tion in the study. In participatory GIS approaches, we should, however, question the necessity of ac- curacy which, in general, is seen as an essential character of scientific data (McCall 2006). The de- lineations of the social values in the real world of- ten have an imprecise boundary like some natural features such as habitat boundaries. Reality is fre- quently ambiguous (McCall 2006) and we should consider if it is misleading to represent it in a pre- cise and accurate way. In addition, ethical aspects were considered as data were not collected with such a precision that it could be connected to in- dividual informants and their delineations on the map. This study represents one approach for han- dling uncertainty in geospatial data and it can be said that there is a need for further research and technical development in analyzing ambiguous and continuous data sets. ACKNOWLEDGEMENTS The authors would like to thank all the Finnish and Tanzanian members of the research project “Sustain- able landscapes in Zanzibar” for their commitment and interest in researching Zanzibar. We would like to thank Dr. Bakari Asseid, Mrs Miza Khamis and Mr Abass Juma Mzee of the Department of Commercial Crops, Fruits and Forestry (DCCFF) for guidance and field facilitations of the study, and the Department of Survey and Urban Planning, the Government of Zan- zibar for allowing the use of spatial data. The work was financially supported by the Geography Gradu- ate School, the Ministry for Foreign Affairs of Finland (project 50173), the University of Turku, Department of Geography and the Laboratory of Computer Car- tography (UTU-LCC) and the Department of Com- mercial Crops, Fruits and Forestry, Ministry for Agri- culture, Livestock and Natural Resources, Zanzibar, Tanzania. REFERENCES Alessa L, A Kliskey & G Brown (2008). Social-ecolog- ical hotspots mapping: a spatial approach for identifying coupled social-ecological space. Landscape and Urban Planning 85: 1, 27−39. Beyer HL (2004). Hawth’s Analysis Tools for ArcGIS. < h t t p : / / w w w. s p a t i a l e c o l o g y. c o m / h t o o l s > . 31.03.2008. Black A & A Liljeblad (2006). Integrating social val- ues in vegetation models via GIS: the missing link FENNIA 187: 1 (2009) 59Participatory mapping and geographical patterns of the social … for the Bitterroot National Forest. Final report JFSP project no. 04-2-1-114. 31 p. Aldo Leopold Wil- derness Institute, Misoula, MT. Brown G (2005). Mapping spatial attributes in survey research for natural resource management: meth- ods and applications. Society & Natural Resources 18: 1, 17−39. Brown G & P Reed (2000). Validation of a forest val- ues typology for use in national forest planning. Forest Science 46: 2, 240–247. Brown GG, P Reed & CC Harris (2002). Testing a place-based theory for environmental evaluation: an Alaska case study. Applied Geography 22: 1, 49−76. Brown G, C Smith, L Alessa & A Kliskey (2004). A comparison of perceptions of biological value with scientific assessment of biological impor- tance. Applied Geography 24: 2, 161−180. Brown G & C Raymond (2007). The relationship be- tween place attachment and landscape values: toward mapping place attachment. Applied Ge- ography 27: 2, 89−111. Commission for Land and Environment (1995). Na- tional Land Use Plan. 140 p. Revolutionary Gov- ernment of Zanzibar, Ministry of Water, Construc- tion, Energy, Lands and Environment. Corbett J, G Rambaldi, P Kyem, D Weiner, R Olhson, J Muchemi, M McCall & R Chambers (2006). Overview: mapping for change − the emergence of a new practice. Participatory Learning and Ac- tion 54, 13−19. Craig WJ, TM Harris & D Weiner (eds) (2002). Com- munity participation and geographic information systems. 383 p. Taylor & Francis, London. FAO (Food and Agriculture Organization of the Unit- ed Nations) (2006). Global forest resources as- sessment 2005 – progress towards sustainable forest management. FAO Forestry Paper 147. 350 p. Gössling S (2001). Tourism, economic transition and ecosystem degradation: interacting processes in a Tanzanian coastal community. Tourism Geogra- phies 3: 4, 430–453. Gössling S (2002). Human–environmental relations with tourism. Annals of Tourism Research 29: 2, 539–556. Gunderson K & A Watson (2007). Understanding place meanings on the Bitterroot National Forest, Montana. Society & Natural Resources 20: 8, 705−721. Haining RP (2003). Spatial data analysis: theory and practice. 432 p. Cambridge University Press, New York. Hartter J, C Lucas, AE Gaughan & LL Aranda (2008). Detecting tropical dry forest succession in a shift- ing cultivation mosaic of the Yucatán Peninsula, Mexico. Applied Geography 28: 2, 134–149. Hettige PML (1990). Land evaluation and land suita- bility classification – Unguja & Pemba Islands. 142 p. Food and Agriculture Organization of the United Nations, Zanzibar. Kaltenborn BP (1998). Effects of sense of place on re- sponses to environmental impacts. Applied Geog- raphy 18: 2, 169–189. Käyhkö N & H Skånes (2006). Change trajectories and key biotopes – assessing landscape dynamics and sustainability. Landscape and Urban Planning 75: 3–4, 300–321. Käyhkö N, M Orjala & J Mustelin (eds) (2008). Sus- tainable landscapes in Zanzibar. Publicationes Instituti Geographici Universitatis Turkuensis A 172. 152 p. Käyhkö N & N Fagerholm (submitted). Dynamic land use and land cover changes behind declining for- est resources in a coastal village of Matemwe, Zanzibar, Tanzania. Land Use Policy (LUP- D-09-00178). Kingston R, S Carver, A Evans & I Turton (2000). Web- based public participation geographical informa- tion systems: an aid to local environmental deci- sion making. Computers, Environment and Urban Systems 24: 2, 109−125. Krebs CJ (1989). Ecological methodology. 654 p. Harper & Row Publishers, New York. Kyttä M & M Kahila (2006). PehmoGIS elinympäris- tön koetun laadun kartoittajana. Yhdyskuntasuun- nittelun tutkimus- ja koulutuskeskuksen julkaisuja B 90. 175 p. Lambin EF, J Helmut, HJ Geist & E Lepers (2003). Dy- namics of land-use and land-cover change in tropical regions. Annual Review of Environmental Resources 28, 205–241. Lofchie MF (1965). Zanzibar: background to revolu- tion. 316 p. Princeton University Press, New Jer- sey. Luz F (2000). Participatory landscape ecology – a ba- sis for acceptance and implementation. Land- scape and Urban Planning 50: 1–3, 157–166. MA (Millenium Ecosystem Assesment) (2003). Eco- systems and human well-being: a frame work as- sessment. 245 p. Island Press, Washington. MacEachren AM, A Robinson, S Hopper, S Gardner, R Murray, M Gahegan & E Hetzler (2005). Visual- izing geospatial information uncertainty: what we know and what we need to know. Cartography and Geographic Information Science 32: 3, 139–160. Makandi H (2008). Mapping land use patterns using remote sensing, GIS and participatory methods. In Käyhkö N, M Orjala & J Mustelin (eds) (2008). Sustainable landscapes in Zanzibar. Publicationes Instituti Geographici Universitatis Turkuensis A 172, 55–61. Manning R, W Valliere & B Minteer (1999). Values, ethics, and attitudes toward national forest man- agement: an empirical study. Society & Natural Resources 12: 5, 421–436. McCall M (2006). Precision for whom? Mapping am- biquity and precisiness in (participatory) GIS. Par- ticipatory Learning and Action 54, 114–119. McIntyre N, M Yuan, RJ Payne & J Moore (2004). De- veloping values-based approach to managing rec- 60 FENNIA 187: 1 (2009)Nora Fagerholm and Niina Käyhkö reation on Canadian Crown Lands. Working pa- pers of the Finnish Forestry Institute 2, 285−293. Finnish Forest Research Institute, Helsinki. Mustelin J (2008). Tourism’s effects on the coastal communities in the east coast of Unguja island. In Käyhkö N, M Orjala & J Mustelin (eds) (2008). Sustainable landscapes in Zanzibar. Publicationes Instituti Geographici Universitatis Turkuensis A 172, 105–111. Müller D & M Zeller (2002). Land use dynamics in the central highlands of Vietnam: a spatial model combining village survey data with satellite im- agery interpretation. Agricultural Economics 27: 3, 333–354. Olwig KR (1996). Recovering the substantive nature of landscape. Annals of the Association of Ameri- can Geographers 86: 4, 630–653. Pontius RG Jr, E Shusas & M McEachern (2004). De- tecting important categorical land changes while accounting for persistence. Agriculture, Ecosys- tems and Environment 101: 2–3, 251–268. Potschin M & R Haines-Young. (2006). “Rio+10”, sus- tainability science and landscape ecology. Land- scape and Urban Planning 75: 3–4, 162–174. Raquez P & EF Lambin (2006). Conditions for a sus- tainable land use: case study evidence. Journal of Land Use Science 1: 2, 109–125. Raymond C & G Brown (2006). A method for assess- ing protected area allocations using a typology of landscape values. Journal of Environmental Plan- ning and Management 49: 6, 797–812. Raymond C & G Brown (2007). A spatial method for assessing resident and visitor attitudes toward tourism growth and development. Journal of Sus- tainable Tourism 15: 5, 1–22. Reed P & G Brown (2003). Values suitability analysis: a methodology for identifying and integrating public perceptions of ecosystem values in forest planning. Journal of Environmental Planning and Management 46: 5, 643–658. Regionplane- och traffikkontoret (2001). Upplevel- sevärden – Sociala kvaliteter i den regionala grön- strukturen. Rapport 4. Katarina Tryck AB, Stock- holm. Relph E (1976). Place and placelessness. 156 p. Pion, London. RGZ (Revolutionary Government of Zanzibar) (2003). Zanzibar tourism policy statement. 29 p. . 04.08.2008. RGZ (Revolutionary Government of Zanzibar) (2004). State of the environment report of Zanzibar. 38 p. Rolston H & J Coufal (1991). A forest ethic and multi- value forest management. Journal of Forestry 89: 4, 35–40. Sauer C (1925). The morphology of landscape. Uni- versity of California Publications in Geography 2, 19−54. Reprinted in Leighly J (ed) (1969). Land and life: a selection from the writings of Carl Or- twin Sauer, 315−350. University of California Press, Berkley. Sitari T (2005). Forestry in the community lifeworld in Unguja island, Zanzibar. In Sitari T (ed) (2005). Forestry, Community and Biodiversity in Zanzi- bar. Turku University Department of Geography Publications B 3, 37–58. Soini K (2001). Exploring human dimensions of mul- tifunctional landscapes through mapping and map-making. Landscape and Urban Planning 57, 225–239. Ståhle A & A Sandberg (2002). Sociotop som redskap i grönområde planeringen. Stockholms Stadsby- ggnadskontor, rapport 2002: 4. Tanzania Sensa 2002 (2003). Population and housing census. . 31.03.2008. Tarrant MA, HK Cordell & GT Green (2003). PVF – a scale to measure public values of forests. Journal of Forestry 101: 6, 24−30. Taylor J, C Murphy, S Mayes, E Mwilima, N Nuulim- ba & S Slater-Jones (2006). Land and natural re- sources mapping by San communities and NGOs: experiences from Namibia. Participatory Learning and Action 54: 79–84. Tuan Y (1977). Space and place: the perspective of experience. 235 p. Edward Arnold, London. Tyrväinen L, K Mäkinen & J Schipperjn (2007). Tools for mapping social values of urban woodlands and other green areas. Landscape and Urban Planning 79: 1, 5–19. UNEP (United Nations Environment Program) (2007). Global environment outlook, GEO 4, environ- ment and development. 540 p. Progress Press Ltd., Malta. United Republic of Tanzania (2003). Indicative tour- ism master plan for Zanzibar and Pemba – final report. 114 p. . 01.08.2008. Voss A, I Denisovich, P Gatalsky, K Gavouchidis, A Klotz, S Roeder & H Voss (2004). Evolution of par- ticipatory GIS. Computers, Environment and Ur- ban Systems 28: 6, 635−651. Williams DR & ME Patterson (1996). Environmental meaning and ecosystem management: perspec- tives from environmental psychology and human geography. Society & Natural Resources 9, 507−521. ZFDP (Zanzibar Forestry Development Project) (1997). Zanzibar long-term forestry plan 1997–2006. Tech- nical Paper 72. 99 p. Oy Edita Ab, Vantaa. Zube EH (1987). Perceived land use patterns and val- ues. Landscape Ecology 1: 1, 37−45. Zurayk R, F el-Awar, S Hamadeh, S Talhouk, C Sayegh, A-B Chehad & K al Shab (2001). Using indigenous knowledge in land use investigations: a participatory study in a semi-arid mountainous region in Lebanon. Agriculture, Ecosystems and Environment 86, 247–262.