Building a common description of land cover in a tropical watershed plagued with intercultural conflicts: The value of participatory 3D modelling

This project took place in the Upper Bayano watershed in eastern Panama, a region covering a total area of 4844 km² (Fig. 1), with the entire area covering 6252 km². The watershed is bounded on the south by the Maje Cordillera and on the north by the San Blas Cordillera with the artificial Bayano Lake (350 km²) in the center (Wali 1989). The watershed is located in both the provinces of Panama and the Darien and includes indigenous territories: the Comarca Kuna de Madugandi (208 550 ha, created in 1996) and the collective lands of Ipeti Embera (3285 ha), Piriati Embera (3869 ha), and Maje (18 920 ha) created in 2014 and 2015, respectively (Panama 1996; Vergara-Asenjo et al. 2015). Maje is the only indigenous territory of the Upper Bayano that is not yet legally recognized.

In the 1970s, both indigenous groups were resettled after the construction of the Ascanio Villalaz dam and promised legal land titles for their new territories as well as financial compensation for the flooding of their ancestral territories (Wali 1989). Panama’s government failure to grant land titles and give financial compensation led Gunas and Emberas to bring the government of Panama to the Inter-American Court of Human Rights. The Court ruled a process that culminated in favor of the indigenous groups in 2014 (la Corte Interamericana 2014, 66) and the collective lands of Piriati and Ipeti were officially granted their collective land titles in 2014 and 2015, respectively.

To understand current land use within each sector and to improve the 3D model resolution through the use of these participatory maps, participatory mapping sessions were carried out for each sector over a period of 2 years. The mapping of Embera communities was carried out in 2013 in the Tierras colectivas of Ipeti, Piriati, and Maje as described by Vergara-Asenjo et al. (2015) (Fig. 1). The mapping of campesinos was carried out around the community of Loma Bonita (Fig. 1), where 70 households were visited and participated in the mapping, representing a group of 44 males and 26 females ranging from 24 to 72 years of age (Potvin et al., In Preparation). A 5 m resolution multi-spectral Rapideye® 2012 images map was used, and a plastic sheet (acetate) was placed over it to allow drawing over the satellite image.

Participatory mapping in the Comarca Kuna de Madugandi was carried out in July 2015 in a subset of 4 of 14 communities: Nargandi, Naca, Ipeti Kuna, and Wuagandi (Fig. 1). The four communities vary in population size, degree of remoteness from the highway, and traditional practices, and correspond to the three geographical sectors recognized internally by the Comarca: communities alongside the lake, rivers, and roads. Mapping built on the methodology used earlier for the Emberas’ communities, relying on a base map with main rivers and roads and a 30 m resolution satellite imagery map from the Environmental Systems Research Institute (ESRI) 2012. For each community, the exercise took 2 d.

Each exercise began with a meeting with the village authorities. During these meetings, one of the authors (LM) explained in the Guna language the mapping purpose and methodology. Once mapping received approval from the local authorities, they appointed a working group. (Note: at all steps of this project, we followed McGill’s Protocols and Ethics in Aboriginal Environment in Panama, and staff members were certified by National Institute of Health of the United States of America.) Although we indicated our preference for a balanced ratio of gender and age, groups were mainly composed of adult men. Mapping started by listing the landscape features of the territory (e.g., trails, rivers, and streams) and agreeing on land cover categories. By midday, participants started drawing the trails, roads, rivers, and streams on a plastic sheet (acetate) affixed to the satellite imagery map. Then, using the acetate and carbon paper, they copied it onto the base map. These features served as references to locate land cover categories. Relying on these features, the participants were able to recall travelling their territory by following the landscape features and to draw each chosen land cover category. Land cover categories were chosen depending on the elements previously listed by participants. For example, no monoculture was present in communities; rather crops consisted of beans, cacao, coffee, plantain, rice, and yucca, so these were grouped as mixed agriculture. Land cover categories were as follows: primary forest, intervened forest, sacred forest, mixed agriculture, pasture, and cemeteries (Fig. 2). In total, 27 participants (five in Nargandi, seven in Naca, six in Ipeti Kuna, and nine in Wuagandi) created the four land cover maps.

To map the landscape of the entire watershed, a P3DM approach was used, aiming to develop a consensus on land cover and land-use conflicts currently present in the region. The construction of the P3DM followed Rambaldi (2010) and consisted of three phases: preparatory, construction, and painting.

The preparatory phase lasted nearly 1 month. During this phase, participants were identified and materials were bought. Members of each sector selected a local coordinator to represent their sector throughout the process: LM representing the Gunas, BP representing the Emberas, and DC and IP representing campesinos. In turn, the local coordinators selected participants from their own sector. The work leading to the construction of the 3D model began with a workshop on 3D mapping, presenting photos and results of previous similar initiatives carried out by Rambaldi et al. (2002, 2007) in Vietnam and Kenya and by Tan-Kim-Yong (1992) in Thailand. The proposed methodology was discussed and adjusted to local circumstances based on participants’ comments. Ten participants, selected by the local coordinators, attended the workshop (four Gunas, five Emberas, and one campesino). After the workshop, efforts were made to increase campesinos’ participation.

Construction started with 12 participants (four Gunas, five Emberas, and three campesinos) but, as the exercise went on, people came and went depending on their availability. Additional participants joined as the process unfolded for a total of 39 participants. The first 12 were the core group who stayed for the majority of the modelling and painting phases. The other 27 were friends, relatives, authorities, or other inhabitants of the watershed. Most of them came for one or two sessions during the construction phase or only for the painting session of their sector.

The 3D model construction took 8 d separated into three sessions and resulted in a blank model of the Upper Bayano watershed with topography. We chose a scale of 1:45 000 because of the large area of the Upper Bayano watershed (6256 km²). The final 3D model is a rectangle of 1.52 × 2.03 m. A vertical exaggeration of 4.5 was implemented to increase the contrast of elevation and facilitate interpretation. The blank model was constructed using 5 mm thick laminated foamboard instead of cardboard to prevent it from degradation due to humidity. To reproduce the topography, we used an shuttle radar topography mission digital elevation model (SRTM DEM) 100 m interval contour line map from 2000 printed at the scale and size of the model, traditional white glue, carbon paper, gypsum paste, crepe paper, scissors, cutters, saw, nails, and a hammer. The construction crew was composed of six Emberas (two men and four women), six Gunas (six men and six women), and four campesinos (two men and two women).

The painting phase followed construction with the objective of reproducing the region’s landscape features and land cover categories on the blank model. This took 4 d divided into four sessions. In the first session, the participants who built the model worked together to identify and paint fixed landscape features of their territory, e.g., rivers, lake, and roads. Features related to water were painted in different shades of blue, whereas roads were indicated with yarn. Those features were then used as references for the participants during land cover painting sessions. Once the landscape features were painted on the blank model, participants began to paint the limits of the three indigenous territories as well as land cover categories for the entire watershed. This phase was done one sector at a time, moving the model from one village to another. The Embera’s painting was held in Ipeti Embera, the Guna’s painting in Akua Yala, and the campesino’s painting in Loma Bonita (Fig. 1). Information provided to participants for the painting phase included local participatory maps when available (see section above), a 2012 digitalized land cover categories map of the Panamanian Ministry of Environment (MiAmbiente), and an ESRI 2012 30 m resolution satellite images map.

The first sector to paint their land was the Emberas, using their three participatory maps. Second, the Gunas painted and delimited their Comarca using their four participatory maps, and the campesinos completed the unpainted area with their participatory map. When the participatory maps were not available, participants were asked if they knew what was on this land. If yes, they would paint it as they knew it, and if no, they would refer to the satellite map of ESRI and the digitalized map of MiAmbiente. Each sector chose its land cover categories, often following the land cover categories used in their participatory maps. The overall land cover categories retained for the 3D model were as follows: primary forest, intervened forest, sacred forest, pastures, agricultural land, plantations, and fallow. The end of this phase resulted in a complete 3D model of the Upper Bayano region (Fig. 3). In the indigenous territories, after each painting session, members of the local authorities came to see and approve the completed work. The participants involved in the painting were represented by eight Emberas (four men and four women), nine Gunas (four men and five women), and six campesinos (two men and four women).

A final exercise located historical and present land cover and land-use conflicts on the 3D model. After painting the land cover categories, meetings were held one sector at a time with the traditional authorities (chiefs and delegates from Ipeti Embera and Piriati Embera communities and the Comarca Kuna de Madugandi) and “influential” locals, such as previous community delegates, or for the campesino side, members of local organizations. In total, participants consisted of five Emberas, three Gunas, and two campesinos (see Discussion section for more on campesinos’ poor participation).

Prior to each session of conflict identification and mapping exercises, all participants were asked to look at conflicts identified by the other sectors and concur that the conflict exists. Additionally, they were asked to verify if any other conflicts were missing. The Emberas were the first sector to identify land cover and land-use conflicts; the information that they provided was verified by the Gunas and campesinos. We acknowledge that the campesinos were unlikely to add conflicts, as it might weaken their or a friend’s claim to land possession, but hypothesize that they may have wanted to remove some proposed conflict areas. Therefore, it should be noted that the two campesinos who participated only verified and confirmed conflicts listed by both indigenous sectors and did not add or subtract any. The campesinos, the last sector to participate in the conflict phase, also had to verify the information added by the Gunas. As the Gunas added no conflicts involving the Emberas, the Emberas were not asked to review the information of the Gunas. Likewise, as no conflicts were added by the campesinos, the Emberas and Gunas were not asked to review the information.

During the verification exercises, participants discussed the history of land cover and land-use conflicts involving their sector as well as present conflicts. The name under which the conflict is known in the region, the nature of the conflict (invasion or legal delimitation), the year the conflict started, the sectors involved in the conflict, and whether it is solved or not were all noted. Participants located conflicts on the model with a red pin for unsolved conflicts and a blue pin for those solved. All pins were numbered according to the numbering of the legend printed and attached to the 3D model. When a conflict involved a large area, it was indicated with a hatched red layer (Fig. 4).

The P3DM exercise resulted in the creation of a common land cover representation of the Upper Bayano and conflict areas therein. The representation included land delimitation, watershed delimitation, and land cover of each sector’s territory. In total, 11 conflict areas were identified by participants and allowed us to develop a timeline of key land cover and land-use conflicts in the region (Fig. 5). Four conflicts covered large areas of land (i.e., a territorial overlap of land delimitation between Gunas and Emberas (conflict 7; Fig. 5) and establishment of whole campesinos’ communities inside indigenous land (conflicts 1, 2, and 3; Fig. 5)). Seven of the conflicts identified were assigned an exact location and consisted of invasions of particular individuals or families (conflicts 4, 5, 6, 8, 9, 10, and 11; Fig. 5). Three conflicts were identified and acknowledged as having been resolved (conflicts 4, 5, and 10; Fig. 5). In the case of conflicts 4 and 5, the Emberas won back their land in court from the campesinos who invaded Ipeti. For conflict 10, the campesinos legally received the right to stay via an official land title. Three unsolved conflicts were between Piriati’s Emberas and campesinos, one conflict between Ipeti’s Emberas and campesinos, one conflict between Piriati’s Emberas and Gunas, and three conflicts between Gunas and campesinos. All conflicts involving campesinos were due to invasion of indigenous territories. These invasions were all located next to the Pan-American Highway (Figs. 1, 6). Participants mentioned that invasions of the Gunas’ Comarca dated from resettlement in the 1970s, whereas invasions of Emberas’ Collective Lands are more recent. Invasions are related to access of land for pasture or cattle ranching. Unlike the others, the conflict involving the two indigenous groups refers to an overlap in the official delimitation of the Guna’s Comarca and the Collective Land of Piriati.

Once the 3D model was built, the traditional Embera authorities agreed to officially present it to Panamanian officials (i.e., the then-director of Autoridad nacional de administracion de tierras (ANATI), and the mayor of the district), television reporters from TVN channel, and local inhabitants (Arcia 2015). This presentation was held during the celebration of the Official Land Title Entry of Ipeti Embera at the end of August 2015. Banners were displayed explaining the steps of participatory mapping and P3DM exercises as well as other activities included in the project. Several members of the Neotropical Ecology Laboratory including authors (MG, BP, and IP) were present to answer questions and explain the project. The presentation was a success, as was the celebration. In fact, government interest was expressed during the official presentation of the 3D model, where the then-Director of the National Land Authority (ANATI) indicated his interest in collaborating with members of Juntos para proteger nuestra Cuenca on land-use planning.

The main obstacle encountered during 3D modelling was campesinos’ poor participation during the conflict discussion phase. This can be explained by two reasons. First, campesinos might not benefit from talking about and agreeing on conflicts as campesinos who invaded land might profit from an unclear land tenure regime. As a matter of fact, in previous interviews conducted by St-Laurent et al. (2013) in the region, campesinos mentioned their fear of losing land during formal projects such as REDD+. Results showed that 100% of households without rights of possession and 52.9% of households with possession rights were afraid. That same study also noted a lack of local organizations and lack of motivation or initiative to participate in REDD+. Compared with indigenous communities, campesinos communities rarely combine efforts and do not have strong organizations or political institutions to represent them at a regional, national, or international level.

In vast landscapes inhabited by heterogeneous groups such as the Upper Bayano watershed, multiple actors are at the negotiation table when it comes time to reflect on land cover and land use. Stakeholders, or groups of stakeholders, likely differ in their concerns and preferred land-use options, which complicates the consensus-building task. Tan-Kim-Yong (1992) explained the importance of using simple methods and tools to create a common vision over land use and to facilitate consensus building.

Land-use planning processes often rely on maps (Evans et al. 2006) and the information presented on them provides a baseline to project future land-use scenarios. However, the understanding of spatial information presented on the map by stakeholders depends on their interpretation (Couclelis and Gottsegen 1997), with the same map possibly being interpreted differently by different actors. In a negotiation process, differential understanding of baseline representation can occasionally lead to confusion. Spatial information for land-use planning purposes must be clear and transparent, otherwise maps can lie and ignite conflicts (Monmonier 1996; Rambaldi et al. 2002).

In this project, the third dimension of a 3D model served as a catalyst to improve understanding of and interest in the spatial data (Fig. 6). As an example, during the painting process and official presentation of the 3D model, people of the region who came to see it systematically located the Bayano Bridge as a reference point. Once this fixed landscape element was located, inhabitants were able to “travel” their territory by following the spatial information presented on the model. The information of the 3D model was easily interpreted by all. These findings support previous research in which the third dimension was identified to give important cues to stimulate memory and establish spatial associations (Rambaldi et al. 2002; Rambaldi and Lanh 2003). Other studies also found that the information in a 3D model was easily understandable by all (Mascarenhas and Kumar 1991; Tan-Kim-Yong 1992; Corbett 2009, 16). We therefore argue that the third dimension allowed a common understanding of the spatial data and helped create a common baseline land cover representation (Fig. 6).

Compared with participatory mapping, the P3DM caught people’s attention and excitement, making people talk about this project. For example, young participants posted photos and updates of the project on social media. This created visibility through conversations and “likes” and fuelled participants’ excitement. It was striking that participants of the three sectors that participated in the P3DM did so voluntarily (i.e., without receiving salaries), which contrasts with standard practice in the region of receiving a compensatory salary for days spent working on a project. The Guna participants, when they worked on the traditional 2D mapping of their territories (Fig. 2), for example, were paid over the 2 d of the exercise, but, when it came to the 3D modelling sessions, they accepted to work as volunteers. The 3D model was something new to most people and clearly captured people’s interest. Compared with a two-dimensional map, its size and vertical dimension make it an innovative and intriguing object. Similar enthusiasm has been observed in previous studies where the use of a P3DM as a concrete tool increased people’s participation (Rambaldi and Lanh 2003; Maceda et al. 2009). Our experience with P3DM is therefore in line with past experience, such as that of Rambaldi et al. (2002) who used it to improve the indigenous negotiating position over a land resources dilemma with Vietnamese national park authorities. Likewise, De Vera (2005) presented the influence of a P3DM in the empowerment of indigenous peoples to advocate their demands and development priorities to the Philippine government.

From the beginning of the project, the interactive participation of all stakeholders created a sense of equity, trust, and group solidarity. We propose that it created what Berkes (2009) calls social capital among the stakeholders. At first, participants appeared shy, and participants from the three sectors did not mingle. But as the process went on, individuals of different sectors started to mix together and to work as a group. Rambaldi and Lanh (2003) mentioned that human interaction dynamics in P3DM are stepping stones towards improving stakeholders’ mutual trust and the understanding of other stakeholders’ positions. We consider that this increased understanding of each other’s positions is especially important in a conflict area. This common definition of the problem prior to formal meetings and negotiation (Amado et al. 2014) was well illustrated when mapping land delimitation of the Comarca Kuna de Madugandi. When considering the overlap between the Comarca and the Collective Land of Piriati, a discussion arose between participants, which was brought to the attention of the Guna authorities. The authorities confirmed the existence of the conflict between both sectors (Fig. 5), with the Emberas claiming that the area was part of their ancestral land although the Gunas were given the area in 1996 when the Comarca was created. The discussion evidenced the confusion because Piriati was also officially given the area in 2014 under Law 72 of 2008 (Panama 2008; Vergara-Asenjo and Potvin 2014). Authorities of both sectors are now considering ways to resolve the situation.

Social learning, as observed in this project, is a key element in co-management systems and a central process in decision-making and environmental management (Berkes 2009). It entails collective action, evaluation, and participation to improve human and environmental interrelation management (Keen et al. 2005, 4). We argue that the social capital created by interactive participation as well as the excitement and enthusiasm created by the third dimension led to “positive effects” (Fig. 6), creating an environment favorable to discussions and harmonious relations among stakeholders. The consensus among stakeholders was not only obtained by discussing and locating conflicts with the authorities but also facilitated by the positive effects (Fig. 6). Innes and Booher (1999) found that the consensus-building process is not only implemented to produce agreements and plans, it is mainly about experimentation, learning, changes, and building together a shared meaning. In fact, following the P3DM, a land-use planning project based on the aspirations, needs, and world views of the different actors that inhabit the region, was initiated to elucidate a variety of land-use scenarios for the Upper Bayano watershed and to model how this area could potentially change over the following two to three decades. Upon assessing the different possible scenarios in terms of land-use change of the Upper Bayano, this project aims to facilitate the selection among all actors of an optimal land-use trajectory that supports local livelihoods, curbs territorial conflicts, and ensures the protection and restoration of forests.