Peer Review #1 of "Spatial analysis, local people’s perception and economic valuation of wetland ecosystem services in the Usumacinta floodplain, Southern Mexico (v0.2)"

1 Departamento de Conservación de la Biodiversidad, CONACYT-El Colegio de la Frontera Sur, San Cristóbal de las Casas, Chiapas, Mexico 2 Departamento de Conservación de la Biodiversidad, El Colegio de la Frontera Sur, San Cristóbal de las Casas, Chiapas, Mexico 3 Department of Natural Resources and Environmental Management, University of Haifa, Haifa, Haifa, Israel 4 Laboratorio de Análisis de Información Geográfica y Estadística, El Colegio de la Frontera Sur, San Cristóbal de las Casas, Chiapas, Mexico

Before the development of the oil industry in the 1980s, fishing was the main economic 145 activity in this region (Tudela 1989). The lagoon and its associated habitats serve as nursery and 146 feeding areas for important species of shrimp and about 214 species of fish in the southern Gulf 147 of Mexico (Lara-Dominguez et al. 1993). Previous studies have shown that fishery resources in 148 Campeche Sound and Terminos Lagoon depend strongly on the supply of nutrients, organic 149 matter, flood pulses and the movement of pre-adult fish and shrimp from the lagoon-estuarine 150 system to the sea caused by tidal action (Deegan 1986; Yañez-Arancibia et al. 1992). Fisheries 151 also include reef fish, coastal migratory pelagic fish, and large oceanic pelagics of great 152 importance at an international level (Yañez-Arancibia and Day 2004), which also depend on the 153 ecological integrity of the Usumacinta floodplain system, its waters and the quality of their 154 habitats. 155 Overall, 14 municipalities are located in the study region, all showing high degrees of 156 marginalization, with the exception of Carmen and Centro. Ciudad del Carmen is the most 157 important city, with a population of 221,000 inhabitants and infrastructure to support the oil 158 industry and other economic activities (e.g., tourism, fishing, harboring) (INEGI 2010). The 159 study area is mostly characterized by a warm, humid climate with 1,200-2,500 mm rainfall 160 during the summer and an average annual temperature of 26-27°C (INEGI 2008). 161 Materials & Methods 162 The methodological approach used in this research draws from various disciplines and includes: 163 (i) the classification and characterization of spatial distribution of land use and wetland 164 ecosystem types through remote sensing and GIS techniques; (ii) the valuation of local wetland 165 ecosystem services based both on the results of an international value transfer exercise and (iii) 166 the analysis of the perception of ecosystem services as elicited by local inhabitants through semi-167 structured interviews; and (iv) an analysis of the spatial overlap of the estimated ecosystem 168 service values with selected indicators of population marginalization and biodiversity richness in 169 order to provide policy-relevant information for decision-makers (Fig. 2). 170 Spatial analysis of wetland ecosystems distribution 171 Different types of floodplain wetlands, their distribution and extension were identified from the 172 classification of Landsat 8 OLI (path /row: 21/ 47; 22/47 and path /row: 21/ 48; 22/48) images 173 acquired in April, May and November 2014. IDRISI Selva and ArcGIS 10 were used for the 174 classification process and for the final coverage map. Before starting the classification procedure, 175 Landsat image was limited to the margins of the physiographic province of the Southern Gulf 176 Coastal Plain (INEGI 2008) using a masking process, and geographically projected to the UTM 177 zone 15 North (WGS84). Subsequently atmospheric correction was performed using the Idrisi 178 Selva AtmosC module. The atmospheric correction is an essential requirement to have an 179 accurate representation in the analysis of the surface properties from satellite images (Mahdavi et 180 al. 2018).

181
The classification process was carried out using a supervised method which required the 182 user to identify the wetland types, digitizing on-line from a color composite image a sample of 183 pixels (training sites) for each cover. Using the maximum likelihood algorithm individual pixels 184 were assigned to the cover class with the highest degree of similarity (Campbell 1996). For the 185 definition of the wetland types present in the study area, the Mexican wetland classification 186 proposed by Berlanga et al. (2008) was used.

187
The accuracy of the final map was assessed by an error matrix and the Kappa index (K'). 188 An error matrix was developed using 30 test points per class as reference data and compared by 189 cross-tabulation with pixels from the classification. Coincidences between both datasets (main 190 diagonal) were used to estimate the overall accuracy (%) and Kappa coefficient (K') to measure 191 the correspondence between the classification and the reference data (Congalton and Green 192 1999). The test points for the analysis were randomly selected from the resulting final map and 193 field validated with the assistance of a GPS. The accuracy for the classification was 86% and the 194 Kappa coefficient was 0.85, indicating that the classification results meet the accuracy 195 requirement of land cover classification. 236 a qualitative study of ecosystem service values was implemented using a participatory approach 237 in four typical villages distributed along the main type of wetlands of the Usumacinta floodplain 238 region (Fig. 1). The aim of this implementation was to compare the results obtained with both 239 methods, looking for similarities and differences, as well as better highlight the dependence of 240 local people from wetland ecosystem services. To achieve this, we collected relevant data 241 combining four different categories: 242 1) Ecosystem services: benefits people obtained from ecosystems, including provisioning 243 (e.g., support of commercial fishing), regulating (water quality improvement), habitat, 244 and cultural services (e.g., recreation) (TEEB 2011).

245
2) Capitals: benefits which are expressed in the form of capital accumulated through the use 246 of ecosystem services in human livelihoods (Scoones 1998), such as human capital (e.g., 247 health), social capital (e.g., associations) and man-made capital (e.g., income).

251
The methods applied to understand the importance of ecosystem services and its 252 associations to local people`s livelihoods were based on the FABE's approach (Hare and Pahl-253 Wostl 2002; Pahl-Wostl and Hare 2004; Hare 2011), which is generally used to build system 254 dynamics models on local people's dependence to natural capital. By incorporating the 255 sustainable livelihoods approach (Scoones 1998), the use of natural capital is understood on how 256 is transformed in different forms of capital such as human capital, social capital, economic 257 capital and infrastructure. Each of the interviewers made one casual loops model linking 258 ecosystems attributes, with the benefits and threats they perceive where affecting their 259 livelihoods as proposed by Fisher et al. (2013).

260
Ten semi-structured in-depth interviews were conducted between October 2014 and 261 March 2015 in each of the four communities selected (Fig 1.). They were organized following 262 the previously defined categories that were covered during the course of the interview, instead of 263 a sequenced script of standardized questions. Such format for the semi-structured interviews 264 allows for more focused, conversational, two-way communication (Meli et al. 2015). Interviews 265 were developed around initial questions such as "What does the river mean to you?", "What do 266 you eat normally?", "Where does your income come from?", "What kinds of problems are you 267 mostly concerned about?". Also, in the course of each semi-structured interviews, the 268 interviewers used the method of "cognitive mapping" (Hare 2011) to develop on paper, cognitive 269 maps of individual perspectives for each of the respondents. All interviews were digitally 270 recorded and professionally transcribed based on the "Grounded Theory" (Glaser and Strauss 271 1967) and the four previously defined categories. Subsequently, an individual qualitative 272 conceptual model for each participant was built on a computer (Fig 3.). 273 The selection of the communities was based on the following criteria : 1) location, 2) 274 population, 3) type of settlement, 4) antiquity, 5) level of transformation of the environment, 6) 275 economic activities, 7) ethnicity, 8) level of marginalization (education level), and 9) researchers' 276 safety (Table 1). With the support of the National Commission of Natural Protected Areas 277 (CONANP), communities in the region that met these criteria were selected. Once the four 278 communities were identified, the delegate of each community was contacted and with their help, 279 ten members were chosen in each of them. In the application, care was taken that the interviews 280 would reflect a representation of ages, gender, educational levels and social sectors in order to 281 have a wide representation of the uses of the ecosystems. Although we acknowledge that a larger 282 number of interviewees and communities would provide a more comprehensive perspective on 283 local ecosystem service perceptions, a more thorough investigation was not possible within the 284 constraints of the study. Yet, we believe the present results to be valuable in an area with severe 285 limitations in data availability. It is stressed that this research reports only results concerning the 286 ecosystem services identified by local communities in order to highlight the local dependence of 287 people from wetlands. The results of the other categories will be analyzed in depth in another 288 article that is in preparation.   Figure 4. The map indicates that urban areas, agriculture, wetlands 312 (coastal lagoon, mangrove, riverine, palustrine and lacustrine) and other inland cover areas are 313 the dominant cover types in the study area.
Considering the classification results, agriculture was the most widely represented land 315 use (1,018,186 ha), covering 21% of the entire area. Among wetlands, the largest ecosystems are 316 palustrine, covering 320,285 ha (12%) of the total surface in the study area. Palustrine and 317 coastal lagoon ecosystems were the most important wetlands in terms of coverage (548,486 ha), 318 representing 71% of the total area of wetlands. Mangroves, riverine and lacustrine ecosystems 319 occupy substantially smaller areas (Table 2). 320 Wetland ecosystem service values 321 Monetary values from benefit transfer 322 The results of the international value transfer analysis are presented in table 2. The number of 323 blank cells represents data gaps in the dataset. Overall, there is considerable variability in 324 ecosystem service values delivered by different wetland types. On a per-hectare basis, the 325 riverine ecosystem was the coverage with the highest annual median value of $12,833 326 USD/ha/year. The mangrove, palustrine and lacustrine ecosystems also contribute significantly 327 to the valuation analysis with a median value of $2,653, $9,689 and $6,366 USD/ha/year, 328 respectively. Coastal lagoons were the wetland type with the lowest median value of $1,926 329 USD/ha/year (Table 2). 330 The total economic values of ecosystem services show a considerable variability. Habitat 331 for species was the ecosystem service with the highest median value ($6,785 USD/ha/year), 332 followed by flood control and storm buffering and recreational activities ($6,677 and $5,093 333 USD/ha/year, respectively). In general, cultural services are the most valued (Table 2).

334
The value of the annual flow of ecosystem services delivered by each type of ecosystem 335 in the Usumacinta floodplain, calculated as the product of each ecosystem service median value 336 per area of the corresponding wetland type (Table 2), was over $4,000 million USD/year. Of all 337 wetlands included in the analysis, the palustrine ecosystem is one that contributed the highest 338 annual flow ($3,698 million USD/year), due to its substantial representation in the total area of 339 the case study (12%) combined with its relatively high value ($9,689 USD/ha/year). Estuarine 340 ecosystems (coastal lagoons and mangroves) together with the riverine ecosystem make an 341 important contribution to the total economic value of approximately $1,000 million USD/year, 342 while the lacustrine ecosystem made the smallest contribution ($397 million USD/year). 343 Importance of ecosystem services for local livelihoods 344 The analysis of semi-structured interviews allowed us to identify the dependence of local people 345 on wetland ecosystems through the provision of ecosystem services. Respondents to the survey 346 were all adults (>18 years) whose average age and time living in the community were 52 (range 347 between 28 and 82 years) and 32 years, respectively; 57.5% were male and 42.5% female.
348 Figure 5 shows that 15 types of ecosystem services are important to local people. Among 349 these ecosystem services, the greatest number of services mentioned in the interviews are those 350 corresponding to the provisioning (39%), followed by cultural (26%) and finally regulating 351 services (13%). Assessment of the provisioning services shows that the current wetlands provide 352 diverse ecosystem services for the local people, highlighting food (85%), support of commercial 353 fishing (80%), and fuel wood (65%). Other services frequently identified were water supply 354 (52.5%), harvesting of natural materials (50%), amenity and aesthetics (45%), medicinal 355 resources (40%), and recreational activities (32.5%). Flood control and storm buffering was the 356 service least often identified by respondents (5%).

357
Comparing these results with the estimates obtained through the benefit transfer exercise 358 reveals that the most valuable ecosystem services differ depending on the context. Cultural 359 services are highlighted as the most valuable services by global studies while provisioning 360 services are the most mentioned by local people. This differentiation may be due design of semi-361 structured interviews that were aimed at understanding the dependence of local communities on 362 the ecosystem services that are most immediately tangible to them, such as of provisioning, as 363 opposed to cultural and regulating services which are provided in an indirect way. Comparing these findings with the local perception analysis highlights that the 393 interviewed communities are distributed differently with respect to the spectrum of priority 394 areas. For instance, Quintin Arauz and Tembladeras are located in an area of high socio-395 economic marginalization and high ecosystem value (magenta).  The study also reveals interesting differences and suggests potential complementarities 428 between different valuation techniques. Although cultural and regulating services were highly 429 valued in the benefit transfer exercise, provisioning services, especially food and support of 430 commercial fishing, were identified as the most important services for local communities in the 431 semi-structured interviews. Although the latter is commonly expected in the context of . While we acknowledge that the proposed approach is a pragmatic rather 458 than theory-driven one, which partly relies on secondary sources and combines data originally 459 developed at different spatial scales, we believe it contributes to the development of the field 460 insofar as: (1) it creates a bridge at the science-policy interface whereby it shows how widely 461 used and readily recognized measures of ecosystem service values (i.e., monetary unit value 462 transfer estimates from international repositories) can be combined with socio-cultural 463 techniques and enriched with additional dimensions of analysis; and (2) it develops a rapid 464 assessment tool which can be useful in providing policy-makers in data-poor countries or regions 465 (such as, but not limited to, the investigated region in Mexico) with initial information on values 466 and priority areas, which can be used as a starting basis for more in-depth, targeted investigation.

467
Integrating the ecosystem service valuation analysis into a broader perspective allows to 468 infer further recommendations for the sustainable management of the local natural capital. The 469 implemented spatial analysis shows which of the important zones for biodiversity are faced with 470 deep marginalization and the ecosystem services that are most relevant in those areas, thus 471 helping to define a first prioritization of areas for intervention or, if the opportunity arises for 472 further and more in-depth investigation. This type of integration in a spatially explicit context, is 473 crucial for identifying and communicating the relative importance of conservation activities in 474 the context of limited availability of resources for ecosystem protection and could be applied to 475 similar environments at regional, country or global level. Also, it may be useful in contexts 476 where approximate value estimates and local perceptions are sufficient to greatly improve cost-477 benefit analyses for large scale developments, which can affect large areas of land, as in many 478 locations around the world, when they are projected in the short term.  The results show that wetlands in the Usumacinta floodplain generate a significant economic 502 estimated value; at least 4 billion USD was delivered to the inhabitants of this region in 2007, 503 highlighting the palustrine ecosystems as the most important wetland in terms of covered area 504 and value. Overall, cultural and provisioning services were the most relevant ecosystem services 505 in terms of their monetary value and local perceived importance. Decision makers in this region 506 will have to prioritize management policies in the municipality of Palizada, an area which 507 coincides with the sites of the analysis of local perception, as well as where the three indicators 508 in the spatial approach (ecosystem service values, marginalization and biodiversity) were highly 509 significant.

510
The approach used in this study can provide a useful starting point for a more 511 comprehensive analysis due to the ease with which the generated information and knowledge can 512 be integrated into a more democratic policy design, and the broad scope of analysis which 513 explicitly accounts for the perception of ecosystem services by local main users and allows for 514 the internalization of the most important wetland ecosystem services in value analyses, with 515 potential benefits in terms of the improvement of the livelihoods of the local communities. 516 Acknowledgements 517 We would like to thank Vladimir Erives, Samuel Aguilar and Héctor Trejo (LAIGE-Ecosur) for 518 their invaluable support with the spatial analysis. We also thank Eva M. Tello, Ariel Alarcón and 519 Roberto Muñoz for her valuable assistance in the analysis of local perception. We acknowledge 520 all respondents for kindly sharing their knowledge and time.