Ocean & Coastal Management 73 (2013) 1e12 Contents lists available at SciVerse ScienceDirect Ocean & Coastal Management journal homepage: www.elsevier.com/locate/ocecoaman The impacts of shrimp farming on land use, employment and migration in Tumbes, northern Peru F. Mialhe a, *, Y. Gunnell b, C. Mering a a b Department of Geography, Université Paris Diderot 7, CNRS UMR 8586, 2 rue Valette, 75005 Paris, France Department of Geography, Université de Lyon, CNRS UMR 5600 EVS, 86 rue Pasteur, 69365 Lyon Cedex 07, France a r t i c l e i n f o a b s t r a c t Article history: Available online 4 January 2013 Export-oriented shrimp aquaculture appeared in the coastal landscape of Peru in the late 1970s. Its rapid development has entailed both positive and negative environmental and socioeconomic impacts across the newly emerging shrimp farming territory of the Tumbes river delta. Here we combine ground surveys and interviews with remote sensing analysis of the transformed environment in order to characterize and quantify the nature, origin and scale of the environmental and social impacts caused by the intrusion of aquaculture. Results show that shrimp farming has encroached on several land cover categories and converted 17% of the Peruvian mangrove. We provide an overview of the history and nature of those impacts. Minimum full-time equivalent employment, which occurred during a white-spot epidemic, was attained in 2001 with 439 full-time jobs; employment peaked in 2006, with 2660 fulltime jobs. However, considering indirect benefits, remittances and the prevalence of part-time jobs, the population reliant on aquaculture probably exceeds 10,000. A geographical analysis of social networks reveals how shrimp farming draws a sustained influx of population from the Andes and how it modifies the interaction between population and the natural coastal environment. Recent sectorial improvements towards achieving a more sustainable management of the coastal resources are analysed, and additional recommendations predicated on a more integrated approach are provided. Ó 2013 Elsevier Ltd. All rights reserved. 1. Introduction Among the diverse productions of aquaculture, shrimp farming has received special scrutiny because of its negative environmental and social impacts. Worldwide production of farmed shrimps has increased from less than 100,000 metric tons in 1980 to more than 3.5 million metric tons in 2009 (FAO, 2012). Shrimps, as many other crustacean organisms, are high value food commodities. Crustacean production accounted of late for only 9.5% of the global production in mass but for 23.1% in terms of value (FAO, 2010). Growing demand among developed countries and emerging economies, technological advances conducive to the intensification of production, and the global decline in marine fish catches have been the main drivers behind this surge in shrimp farming (Biao and Kaijin, 2007; Paul and Vogl, 2011). Aquaculture has also been promoted to alleviate poverty through the supply of direct and * Corresponding author. Present address: Department of Geography, University of Namur, 61 rue de Bruxelles, 5000 Namur, Belgium. Tel.: þ32 485 36 34 24. E-mail addresses: francois.mialhe@fundp.ac.be, framialhe@hotmail.fr (F. Mialhe). 0964-5691/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ocecoaman.2012.12.014 indirect jobs, proteins and revenue (Hishamunda and Ridler, 2006; Asian Development Bank, 2004; Edwards, 2000; Heck et al., 2007). For example, several development projects conducted by multilateral agencies such as the World Bank and the Asian Development Bank focused on aquaculture during the 1980s (Goss et al., 2000; Hall, 2004) and shrimp farming, a non-traditional export activity, benefited substantially from them. Shrimp farming has since become a major source of export earnings for several Southeast Asian countries (Hall, 2004). The ensuing economic development was associated with both positive and negative impacts. In total, shrimp farming is estimated to occupy between 1 and 1.5 million hectares along the world’s coastlines (Berlanga-Robles et al., 2011). Mangrove degradation, for example, has largely been depicted as a major environmental consequence in American and Asian coastal areas. Almost half of the total mangrove land cover area has been depleted in the last 50 years (Curran, 2002). Although several factors have led to mangrove destruction, shrimp farming is considered a major culprit (Valiela et al., 2001). Until recently, development of shrimp farms in mangrove ecosystems was not prohibited and even sometimes encouraged by national and local governments, who assessed the traditional use of mangrove resources as inefficient (Stokstad, 2010; 2 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 Armitage, 2002). Other negative ecological (e.g. soil and water pollution) and social impacts have arisen with the development of the aquaculture industry, such as social inequalities in food security and resource allocation, and the marginalization of smaller scale farmers (Bailey, 1988; Stonich, 1995). A major unforeseen impact is that the development of aquaculture has failed to reduce the pressure on capture fisheries: instead of operating as a substitute it has expanded the demand for fish protein by creating new markets (Curran, 2002). A positive outcome of the activity is the job market generated by the farms, with opportunities for improving human livelihoods. Overall, global aquaculture production, including shrimp farms, is estimated to provide a direct source of income to 16.7 million people and indirectly to another 6.8 million (Valderrama et al., 2010). Even such a figure probably underestimates reality if we consider that just in Bangladesh 1.2 million people are directly involved in shrimp production and 4.8 million household members benefit from the industry (Islam, 2008). However, according to some case studies in Asia (India, Philippines, Indonesia), shrimp farming appears to have benefited mostly affluent and elite classes, and in some case has widened the gap between rich and poor (Adduci, 2009; Hall, 2004; Armitage, 2002; Stonich, 1995). Among the poorer sections of society, some individuals have benefited from aquaculture, mainly through job opportunities in the commodity chain (Edwards, 2000). Many, however, such as in Ecuador and Honduras, have seen a reduction of their livelihood options because of poor planning of aquacultural activities and contempt from the ruling class towards traditional resource management practices and local indigenous populations (Stonich, 1995; Hamilton, 2011). Privatization of the resource base that supports traditional livelihoods has caused some in- and out-migration of labour for economic reasons, with farm owners taking advantage of interregional or international wage inequalities and favouring non-local populations as an avoidance strategy against local collective protest (Resurrección and Sajor, 2010). In places such as the Philippines or Bangladesh, the development of shrimp farming has led to emigration because of environmental degradation (Primavera, 1997). However, through the remittances sent home by the migrants, aquaculture has also revealed a capacity to enhance the resilience of households in their original home territories by spreading economic risk and broadening economic opportunities. In Vietnam, remittances have also increased inequalities among households in the communities of origin (Adger et al., 2002). In Thailand, the development of inland shrimp farming has also promoted migration into the continental interior (Flaherty and Vandergeest, 1998). In Indonesia, the government-led transmigration of population from Java to Sumatra, Kalimantan and others islands has been critical in the development of shrimp farms outside Java, where shrimp diseases affects many production units (Armitage, 2002; Hall, 2004). Shrimp farming has also generated some international migration. In Thailand, where Thais tend to avoid unskilled work, the government has encouraged immigration policies advantageous to shrimp farms. These now employ low- and semi-skilled workers from Laos and Burma (Resurrección and Sajor, 2010). These impacts have been addressed in different ways by stakeholders of the shrimp farming sector. The wide spectrum of solutions reflects differences in diagnosis, in perception of the impacts, but also divergent interests. For example, Béné (2005) has highlighted a sharp opposition between a politicized discourse rooted in political ecology, which promotes a holistic approach involving extensive and mixed farming systems, and a technocratic and market-oriented discourse which advocates intensive, closed systems assorted with best management practices. Concepts and guidelines emanating from the prescriptions of Integrated Coastal Management (Chua et al., 2006; Primavera, 2006) and Ecosystem- Based Management, for example, strive to reconcile economic development and resource conservation objectives by reframing aquaculture through the lens of the ecosystem services narrative (Barbier et al., 2008). However, the successful conciliation of conservation goals and development objectives at sites where shrimp farms have become established still remains a largely unfulfilled ideal (Barbier et al., 2008). In South America, shrimp farming developed first in Ecuador in 1969 (Twilley et al., 1999). In Peru, it began in the 1970s, initially under the impulse of government-led experiments. The private sector was first allowed to invest in shrimp farming in 1978 following a shift in national policies towards liberal economic principles (National Institute of Natural Resources, or INRENA, 2007). That same year, the government allocated 6000 ha to aquaculture, mostly in saline areas on the inner fringes of mangrove swamps. Initially, the development of aquaculture was driven by a government-endorsed licence to exploit land resources, with the economic pattern broadly following an export-based system modelled on vent-for-surplus theory (Myint, 1958). Following the importsubstitution model of industrialization, export-led growth models of development through non traditional export were supposed to stabilize the balance of payments and revitalize economic growth in many developing countries (Barham et al., 1992). This model argues that export activities in the basic sector (export) create jobs in the non-basic sector through spillovers (Tiebout, 1956) and are particularly well suited to small regions or cities that need to import goods not produced locally (Malecki, 1997). In this perspective, shrimp farming was a suitable candidate activity since a number of advantageous conditions were locally available: unfarmed land, the proximity of Ecuadorian expertise, private capital, wild post-larvae in the local ecosystem, and an existing research unit in the local Fisheries department. The well-known collapse of Peruvian marine fisheries in 1972 due to anchovy overfishing (e.g. Thorpe et al., 2000) also provided a perfect conjuncture for developing aquaculture in the Tumbes delta. Based on a case study, this paper aims to assess some of the impacts generated by ex nihilo shrimp farming in the coastal area of Tumbes, northernmost Peru. In this so far poorly documented region, we first map the land cover and land use changes that have occurred through the development of aquaculture to unveil the environmental changes that have occurred at landscape level. Then we estimate the number of jobs created by the newly introduced shrimp farming system. Through an analysis of labour migration, we identify the determinants of population flow at the individual, household and regional levels and analyse the role of shrimp farming in driving immigration. We finally connect those three aspects to show how shrimp farming has shaped, and been shaped by, the interactions between society and the environment. 2. The study area The study area is part of Tumbes Department, in the northern coastal zone of Peru on the border with Ecuador (Fig. 1). Climate is influenced by the semi-arid conditions in the south, by the equatorial climate in the north, and by El Niño events. The 180 km-long Tumbes river is the main watercourse in the area (drainage area: 5656 km2) and the only navigable river of the Pacific coast of Peru. It rises in Ecuador (Zaruma Mountains) and forms a delta in the study area. The river discharge regime is unimodal and peaks in March. The coastal area comprises a mosaic of various ecosystems such as mangrove (association of Rhizophora mangle, Laguncularia racemosa, Conocarpus erectus and Avicennia germinans), dry forest, savanna and algarrobal (Prosopis pallida). Among the provinces forming the Tumbes Department (population: 191,713), Tumbes province was the most populated in 2005 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 3 Fig. 1. Location of the study area. 10000 12000 14000 occurrence of the seventh cholera pandemic in 1991; and (iii) an epidemic caused by an outbreak of white-spot syndrome (WSS) in August 1999. The farming systems involve semi-intensive and intensive monoculture of shrimps (P. vannamei). Post-larvae are produced in hatcheries mainly located in Ecuador (Evans and Tveteras, 2011) whereas formulated feed is produced nationally from forager fish. The stocking density of post-larvae varies from 15 m 2 for semiintensive to >50 m 2 for intensive systems. Farms are large scale, many of them with a production area exceeding 50 ha. Yields in semi-intensive systems approximate 1 to 1.5$103 kg per cycle with two to three cycles per year, while intensive systems yield between 5$103 and >10$103 kg per cycle. Farmers are now dramatically reducing intensive systems and have shifted towards semi-intensive systems following the recurrence of viral diseases. The absence of a local market likely to absorb the production unfit for export has provided a further impetus for the farmers to abandon the intensive system, in which production costs remain very high (E. Mialhe, pers. comm. October 2012). 3. Methods 8000 In order to assess the multidimensional nature of aquaculturerelated impacts both in space and time, a specific methodology was elaborated based on remote sensing and field work. 4000 6000 3.1. Mapping changes in land use 0 2000 Farmed shrimp production (Mt) (population: 139,073) (Instituto Nacional de Estadistica e Informativa, or INEI, 2009). Between 1981 and 2005, the Department population increased by 77% (83,649 inhabitants), the bulk of this increase (67%) affecting Tumbes alone. With a population density of 225 km 2, the coastal area, which includes the La Cruz, Corrales, Tumbes, Papayal, Aguas Verdes, and Zarumilla provinces, is much more thickly populated than the hinterland (10 km 2). Given also that demographic fertility is greater in the hinterland, migration has been an important driver of this demographic imbalance. Alongside services and international trade with Ecuador, agriculture is a major activity. Rice and banana are the two mains crops, both irrigated. Peruvian shrimp production statistics (Fig. 2) exhibit high variability over time. The major disturbances to a steadily growing production were (i) El Niño in 1983 and 1997e1998, which caused flooding, clogging of waterways by sediment, mangrove depletion, and a critical reduction of wild post-larvae of Penaeus vannamei, Penaeus stylirostris and Penaeus californiensis in 1984; (ii) the 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010 Years Fig. 2. Shrimp production in Tumbes between 1980 and 2006 Source: Peruvian Department of Fisheries. The assessment of environmental impacts at the landscape level was achieved through a multitemporal inventory of land resources obtained from a satellite image- and aerial photograph-based analysis of land use changes. Remote sensing is a convenient tool for obtaining synoptic and repetitive information about the state of land resources and then to assess the land use changes (Green et al., 1996; Rogan and Chen, 2004). A total of eight images were used in this study: three Landsat MSS, one Landsat TM, one Landsat ETMþ and three SPOT 5 (Table 1). Landsat images were acquired from the Global Land Cover Facility while SPOT images were acquired through the ISIS program of the CNES (Centre National d’Études Spatiales). Aerial photographs were taken in 1962 by the Fuerzas 4 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 Table 1 Imagery used in the study. Table 2 Image geometrical correction tests. Source Date (years) Aerial photographs Landsat MSS 1962 1977, 1982, 1983 1991 2000 2003, 2004, 2007 Landsat TM Landsat ETMþ SPOT 5 Ground resolution (m) 57 28.5 28.5 10 Armadas del Peru. In total, the image data set covers a period of 45 years. Because misregistration can decrease the accuracy of diachronic maps (Dai, 1998), geometric corrections were performed systematically (Fig. 3). Given field investigations and in situ GPS data collection carried out in 2007, a SPOT 5 satellite image from 2007 was selected as the reference image. All the other images were thus coregistered against the SPOT 5 using second-degree polynomial transformation and nearest neighbour resampling. The root mean square errors (RMSE) of the images are given in Table 2. Except for the Landsat MSS image acquired in 1983, for which the low quality did not allow the same level of accuracy to be attained, all the RMSE Image (with acquisition date) No. of ground control points RMSE Landsat MSS (1977) Landsat MSS (1982) Landsat MSS (1983) Landsat TM (1991) Landsat ETMþ (2000) Spot 5 (2003) Spot 5 (2004) 23 18 20 20 17 11 15 0.444064 0.471381 0.602474 0.499527 0.246883 0.102056 0.074806 values were suitable because below the recognized accuracy threshold of 0.5 (Hui et al., 2008). The image classification procedure followed the steps indicated in Fig. 3. Principal Component Analysis (PCA) is a multivariate statistical technique based on an orthogonal transformation designed to convert possibly correlated spectral data into a (usually smaller) set of linearly uncorrelated variables called principal components. PCA was first performed on each image in order to increase radiometric contrasts and reduce redundancy between strongly correlated channels (Chavez and Kwarteng, 1989; Eklundh and Singh, 1993; Lu and Weng, 2007; Campbell and Wynne, 2011). Components with the higher eigenvalues were then subjected to an unsupervised K-means classification, which has become a routine procedure of unsupervised pixel classification (Lillesand et al., 2007). The number of user-defined classes varied from 15 for the Landsat to 25 for the SPOT 5 images. These classes were first interpreted and were then merged on the basis of a land cover typology elaborated from field investigations (Mialhe, 2010). The typology included the following classes: wetlands, water bodies, bare soil, mangrove, dry forest, mixed vegetation (including savanna and riparian forest). The sea was masked by segmenting the water body classes. Because maps derived from remotely sensed images can contain errors, e.g. due to preprocessing or interpretative techniques, classification accuracy was assessed using the Kappa index, frequently used in remote sensing and derived from an error matrix (Story and Congalton, 1986; Foody, 2002, 2004). Kappa indices ranged here between 0.81 and 0.98. Aerial photographs were visually interpreted using the same class typology as for the satellite images. Finally, several change detection maps were computed using a post-classification method based on the individual land use maps (Singh, 1989). 3.2. Surveys Fig. 3. Flow chart of the remote sensing procedure for land use change mapping. Field work was conducted to collect information about employment in the aquaculture sector and labour migration. It was conducted between April and June 2007. First-hand data were collected through questionnaires and interviews with local stakeholders (government agents, farm operators, environmentalists) while second-hand data were compiled after visiting the appropriate institutions (referenced whenever relevant hereafter in the text). The unbiased assessment of population involvement in specific activities is not a simple task because employment can be informal, irregular or indirect. Here, quantitative estimates of employment levels relied on estimations of the labour force required at several stages of the commodity chain: farms, conditioning plants, and input salesmen. One of the three major conditioning plants was visited. Chief and section managers were questioned about the operations conducted within the plant and about the labour force required to perform each operation. Data was normalized by the mass of shrimps processed in order to allow for interannual computations. Information about the same operations in other plants 5 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 confirmed these numerical estimations. The assessment of jobs in the transformation chain is based on the following empirical equation: Ep ¼ ½ðp þ yÞ*x þ ðq þ sÞ*wŠ=z (1) where Ep is the number of annual full-time jobs per plant, p is the total mass of shrimps produced (in 106 g), q is the mass of whole shrimps destined for export (in 106 g, 10% of the production on average), s is the mass of unbeheaded shrimps (in 106 g), x is the average number of labourers involved in beheading operations (x ¼ 8$10 6 g), w the average number of people required for freezing and conditioning operations (w ¼ 10$10 6 g), y the proportion of total production to be beheaded (y ¼ 0.9), and z the number of days worked by each employee (z ¼ 260). A dozen of grow-out farms were visited, ranging from extensive to intensive and from medium to large-scale. Farm managers were questioned about the labour force required for permanent and temporary operations. External agents were also questioned about temporary work loads. The following information was sought: number of permanent workers for grow-out operations under different production intensities (ranging from extensive to intensive), proportion of skilled workers, number of temporary workers and number of annual production cycles. Because numerical data given by farm managers were suitably convergent, an average of reported figures was calculated. These numerical data were also consistent with estimations made by Berger et al. (2004). For onfarm operations, separate functions were computed for permanent and temporary workers. Employees include both skilled (technical, administrative, secretarial) and unskilled workers (for grow-out operations, security, monitoring), whereas the temporary workers are hired for harvesting and cleaning tasks. The empirical formula for on-farm employees is: Epf ¼ Q *F þ M*SI þ N*I (2) where Epf is the number of permanent workers on the farm, Q the average number of skilled jobs per farm, F the number of farms, M the number of permanent unskilled workers per hectare for semiintensive and extensive farms (M ¼ 0.2), N the number of permanent unskilled workers per hectare for intensive farms (N ¼ 2), SI the area of semi-intensive and extensive farms (in ha), and I the area of intensive farms (in ha). For temporary workers the function is: Et ¼ A*B*V þ I*B*W (3) where Et is the number of temporary workers, A is the total production area (in ha), B is the average number of cycles per year, V is the number of workers needed per harvest and per hectare, W the number of workers needed to clean the intensive ponds after harvesting operations, and I the area of intensive farms (in ha). The total number of jobs was calculated from equation (4): Ed ¼ Epf þ Ep þ Et þ T þ U (4) where Ed is the total number of jobs, T is the number of individuals working in local services dedicated to aquaculture (such as food and input supplies), and U is the estimated number of workers working in undeclared farms. Other variables are as given in equations (1)e(3). A questionnaire was submitted to unskilled workers (n ¼ 70) and served several purposes. Its aim was to estimate the proportion of migrants among the unskilled worker population and to identify the determinants of the migration at individual, household and contextual levels. Because most farm workers live most of the time on the farm, surveys were carried out on the farm during the rest of the time. Six farms presenting a range of production intensities were surveyed. Respondents were selected randomly among the workers, i.e. without indicating our intended focus on migrants. The questionnaire included a general section dedicated to all workers and a special section concerning immigrant workers. 4. Results and discussion 4.1. Land cover and land use change Land cover changes observed from the six diachronic land use maps produced (see Supplementary Material for five of these maps) have been interpreted and summarized in Table 3. The map sequence provides a detailed record of habitat depletion for each of the time slices indicated in Table 3. This table includes only nonseasonal changes, i.e. it does not consider phenological changes in the ecosystems or cropping systems. The successive main land use changes are related to the development of agriculture, the development of shrimp farming and the variability of shrimp production. The main transformative process occurring in the study area during the second half of the 20th century has been the destruction of natural habitat by agriculture and aquaculture. Agriculture first developed on fine-textured bare soil and mangrove in the delta and along the small river floodplains, and it was promoted in this lowland environment by the introduction of highyielding rice varieties and new irrigation schemes. Between 1977 and 1991, the ponds appeared primarily on bare soil (which includes beaches and saline soils with scattered halophytes, e.g. Distichlis spicata and Sesuvium portulacastrum), mainly on the landward side of the mangrove belt. Mangrove and dry forest Table 3 Land use changes detected by remote sensing methods. Period Time-sliced evolution 1962e1977 1977e1991 1991e2000 2000e2003 2003e2007 Summary 1962e2007 Qualitative land use changes Quantitative estimate of impact Expansion of aquaculture Development of ponds Coastline variation Interruption of aquaculture Recovery of aquaculture Continued recovery of aquaculture Average of 47 ha yr 1 gained at the expense of bare soil (526 ha) and mangrove (185 ha) Gained at the expense of mangrove (600 ha), dry forest (500 ha) and bare soil (2000 ha) Accretion associated with channel avulsion Development of bare soil at the expense of water bodies (w1800 ha) w1000 ha of recovered ponds, scattered along the coast w1000 ha of recovered ponds Expansion of agriculture Coastal accretion Gained at the expense of bare soil (1330 ha), dry forest (852 ha), savanna (853 ha) w375 ha of new coastal sediment (i.e. w3 km of seaward progradation) near present day sand spit Mangrove (w1000 ha), bare soil (1000 ha), savanna (1490 ha), dry forest (700 ha) Development of fish ponds 6 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 located on the mangrove fringes were first to be cleared. Some farms, however, were also implanted as enclaves in the mangrove heartland. Although the bulk of the mangrove was converted to ponds, some portions were also deforested for building the farm access roads and bridges (INRENA, 2007). El Niño (1997e1998) and the WSS outbreak episode explain the reduction of areas under productive aquaculture during the 1991e2000 decade. Estimating through remote sensing alone the shrimp farming areas that temporarily halt production is a difficult task because of the difficulty in distinguishing between naturally bare soil and disused (i.e. dried up) ponds based on spectral signatures alone. However, a visual inspection of the maps in the Supplementary Information shows that the total change (w1800 ha) corresponds mostly to a decline in pond use. During that same time interval, however, some new farms also developed on bare soil (w500 ha). The area corresponding to ponds that recovered during the following period (2000e2007) is w2000 ha. A more comprehensive account of the proportion of land cover converted to ponds is best provided by the map of land use change between 1962 (estimated from aerial photography) and 2007 (obtained from SPOT 5) (Fig. 4, see Supplementary Information for a colour version of Fig. 4). It reveals the contrast between the diversity and heterogeneity of initial land cover types and the comparative uniformity of present day land use as the delta became progressively converted to aquaculture. In total, the percentages of original land cover that have been converted to shrimp farms between the coastline and the pan-American highway are 16%, 18%, 14% and 28% for mangrove, bare soil, dry forest, and savanna, respectively. The absence of environmental and land use planning policy enforcement gave a free hand to indiscriminate aquaculture development, yielding a relatively anarchic spatial pattern of farm occurrence. Depletion of the TumbesePiura dry forest, recognized in this region as containing high rates of bird endemism (Best and Kessler, 1995), raises conservation issues of international proportion. Like the Guayaquil flooded grasslands in adjacent Ecuador, this tropical and subtropical broadleaf woodland is listed as a remarkable terrestrial ecoregion by the World Wide Fund for Nature and is really a mosaic of many habitats, with at least 14 endangered species and a catalogue of nearly 500 species of bird, mammal, batracian and reptile. The risk of desertification relating to the increase of grassland savanna over woodland is another concern. The encroachment of aquaculture on bare soil (whether dry or wet) has also impacted natural drainage because this type of land cover mostly corresponds to creeks in the intertidal zone. As a consequence, the mangrove ecosystems that depend on the ebb and flow of water in the creeks have been affected (INRENA, 2007). The construction of canals by farmers has also increased flood hazard in built-up areas. The impacts are thus not limited to the converted areas but also have off-site effects. Shrimp-farm effluents, which diffuse to the surrounding water bodies, usually damage both flora and fauna (Gräslund and Bengtsson, 2001; PàezOsuna, 2001). Evidence of such impacts on the local flora and fauna would require a purpose-designed ecological investigation, but previous studies have already highlighted a link between the Fig. 4. Changes in land use from 1962 to 2007. 7 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 development of shrimp farming and the reduction of fish catches and crab populations (e.g. Cardisoma crassum) within the Tumbes mangrove area (INRENA, 2007; McClennen, 2004). Deforested mangrove between 1962 and 2007 is mapped in Fig. 5 (see Supplementary Information for a colour version of Fig. 5). It reveals that 1228 ha of mangrove have been destroyed, a figure close to official data (1278e1294 ha) but less than claimed by environmentalists (3000e4500 ha) for that same time period. Depletion represents about one sixth of the total mangrove area. The creation in 1988 of a 2972 ha mangrove sanctuary, unique in Peru and further recognized as a Ramsar site (wetland of international importance) in 1997, was a direct consequence of aquaculture development. It has been effective in the limitation of encroachment by aquaculture within the limits of the sanctuary but much less in its buffer area, which comprises patches of dry forest and salt marsh and was designed to reduce extractive activities such as logging. Conversion of forested land to aquaculture had a direct impact on natural resources and on the level of ecological goods and services (Sathirathai and Barbier, 2001). The extension of aquaculture at the expense of mangrove habitat raises questions of sustainability because the mangrove habitat supplies natural resources and ecosystem services of direct utilitarian value to shrimp farmers (post-larvae, nutrient cycling). Because fish, molluscs and crustaceans that depend on the mangrove habitat are routinely harvested by smallscale fisherfolk and collectors (INRENA, 2007), the reduction in mangrove land cover has also negatively affected livelihoods. Avulsion of the main Tumbes channel causing migration of the river mouth has occurred several times as a consequence of Niñorelated floods. Fig. 5 maps the shoreline variations between 1962 and 2007 resulting from the beach erosion and accretion processes in response to Niño events. Because of the limited number of shrimp farms pumping Tumbes river water, the impact of flow extraction on river discharge is deemed limited and thus has not played a key role in the observed shoreline variation. Some farms, however, have suffered the consequences of shoreline variations, with several farms located near the beaches having ceased to N operate the more exposed ponds. Others farms have planted mangrove vegetation in the hope of stabilizing the shoreline. There is no evidence of newly developed farms since 2007 in the study area. However, some new farms have appeared to the back of the beach in the southern part of Tumbes Department. These new installations have caused detectable losses of dry forest. 4.2. Employment in the shrimp farming sector Peru lacks official statistics about employment in the shrimp farming sector (Evans and Tveteras, 2011). Based on our own data, Fig. 6 shows the evolution of total employment between 1980 and 2006. Employment after 2006 was not computed because information about surface areas under semi-intensive and intensive farming is scarce. A minimum was attained in 2001 (in the middle of the WSS epidemic) with 439 full-time jobs, and the peak was reached in 2006, with 2660 full-time jobs. This is consistent with the estimation of 1130e2260 permanent employees given by Evans and Tveteras (2011). Our estimate does not take into account the post-larvae collectors (larveros) who used to work along the beaches and tidal channels until this activity was definitively banned in 2000. Most of them were previously in farming and came from the Andean provinces during the shrimp boom (Peña-Ruiz, pers. comm., April 2007). According to the INEI (2009), there were 4000 larveros in 1997 and only 300 in 2002. Approximately 70% of them stayed in Tumbes after the ban (Peña-Ruiz, pers. comm., April 2007). On-farm permanent work is the most important source of jobs before temporary work on farms and in conditioning plants. The number of jobs has undergone important changes, with some booms in 1986e1988, 1993e1997 and 2002e2006 and some busts in 1990e1991 and 1998e2001. While busts are mostly related to environmental shocks and viral outbreaks, booms are related to economic incentives and technological innovation. A more accurate estimate of the people directly or indirectly dependant on aquaculture requires family members to be included because wages are locally shared with household members or sent Pacific Ocean Depleted mangrove Land mass Puerto Pizarro 0 N 2,5 5 10 kilometers 1962 / 1991 Pacific Ocean Recession Progradation Land mass Rivers Puerto Pizarro 0 2,5 5 10 kilometers N 1991 / 2007 Pacific Ocean Recession Progradation Land mass Rivers Puerto Pizarro 0 2,5 5 Fig. 5. Mangrove depleted by aquaculture between 1962 and 2007, with shoreline variations. 10 kilomete rs 8 4000 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 Farm − temporary Conditioning plant 3000 Total Farm − permanent Undeclared farms El Niño Intensification 2000 WSSV 0 1000 Full−time jobs Cholera PL shortage 1980 1985 1990 1995 2000 2005 Years Fig. 6. Full-time jobs in the local shrimp commodity chain. to the family by single migrants. Based on a conservative estimate of three beneficiaries per household, a minimum population of 8000 is thus estimated to depend on aquacultural wages. This figure is based on the hypothesis of full-time employment in the sector, but given that most of the workers are employed on a parttime basis, the estimated population reliant on aquaculture is likely to exceed 10,000. This figure is less, however, than the estimate of 20,000 by Berger (2002). The w10,000 jobs can be differentiated according to whether they are contract-based or not, to wage levels, working conditions, security levels, and variability in time. In Tumbes, the working population represents 54% of the total (INEI, 2009). On that basis, the proportion of the working population employed in the aquaculture commodity chain reached a maximum of 4% in the early 1990s. This estimation does not consider indirect employment created through the local expenditure of wages, e.g. on transportation, food, accommodation, retail, construction, or leisure. It also excludes activity losses (small-scale farming and livestock) in dry forest areas due to land conversion. 4.3. Shrimp farming and migration From the set of unskilled workers we interviewed, 81% of them are internal migrants, all male and mostly young. An overwhelming proportion of the migrants comes from Piura Department (87%), which is the adjacent province to the south, 7% from Lima, 2% from Cuzco, 2% from La Libertad and 2% from Cajamarca (Table 4 and Fig. 7). Geographic proximity is evidently a cause in immigrant provenance: Piura is the only Peruvian region contiguous with Tumbes, with the two respective provincial capitals w300 km apart by road. The bulk of migrants from the Piura region come from its rural and mountainous eastern provinces. These consist of deep valleys exposed to floods and soil erosion with small, lowproductivity farms (Rubio, 2007). Biophysical and agriculturerelated structural factors thus appear to be major constraints on economic prosperity. This is confirmed by poverty and extreme poverty indices, which are based on household level income and expenditure (INEI, 2009) and are nationally much higher in the mountainous provinces d respectively 72% and 39% d in the Andes compared to 45% and 5% in the coastal areas (INEI, 2009). In Tumbes Department the extreme poverty index is 7.4%. The advantages of Tumbes are linked to factors already mentioned such as international trade and the production of export commodities such as rice, banana and shrimp. The main migration stream, therefore, is clearly from rural and mountainous to urban and coastal areas. The three main motives given by the migrants for emigrating are economic: they expect to improve their standard of living, find a job, and support the household (Table 4). These all arise from push factors. Push factors, which can either be environmental, economic or social, are unfavourable conditions that prevail in the place of origin; conversely, pull factors are attractive conditions that prevail in the place of destination (Lee, 1966). In the last several years, agriculture in Piura has been adversely affected by El Niño anomalies and by an absence of effective agricultural policies. Together, these have further marginalized farmers who were already vulnerable to agrarian problems (Rubio, 2007). Pull factors are related to needs in the labour market and to the presence of relatives or acquaintances. The availability of jobs in shrimp farming ranks third among the list of pull factors, with 22% of migrants choosing Tumbes because of the aquaculture industry (Table 4). These factors are linked to the existence of social networks, i.e. links between individuals from the same community or kinship group, regardless of their current place of residence, thus including prior migrants (Massey, 1990). The existence of these social networks, where information about economic opportunities is intensively exchanged, explains why the migrants have chosen Tumbes and why almost half of them were aware of job opportunities in shrimp farming before leaving Piura. Social capital, i.e. the network together with the information or resources shared among individuals, thus contributes to the emergence of migration as a livelihood strategy by providing a direction to potential migrants. Social capital completes the pushepull paradigm, which successfully explained the difference between places of origin and destination but failed to explain why and how people move to a place with qualities similar to those of other places (Curran, 2002). The members of the social networks also support the migrants during the initial job seeking stages: earlier migrants succeed in securing jobs on their farm for new migrants by virtue of the mutual trust that has already had a chance to grow between them and the shrimp farm manager. Several data confirm that statement: our survey indicated that 97% of the workers had obtained their jobs through their social networks and that 80% of them had a relative already working in shrimp farming. As a result, every farm includes several clusters of workers coming from the same geographical area or from the same kinship group, thus confirming the more widely recognized importance of social capital in migration patterns (Portes, 1995; Gray, 2009). The presence of prior migrants within the social network fuels the influx of new migrants because the latter are informed of opportunity costs and are provided with appropriate material reassurances (Curran, 2002). A majority of migrants (68.5%) send remittances to their family. On average, farm workers send 40% of their wages to their families, i.e. 231 soles1 for an average monthly salary of 573 soles. These savings are declared to be mostly invested in the services sector in the province of origin, such as grocery retailing or automobile repair, or are used for basic household needs. Although this aspect would require a more detailed analysis, previous studies in Andean countries and in Asia have shown that remittances and return migration sometimes reduce social resilience and increase inequalities between individuals and households within and between home villages because wealth is not redistributed (Adger et al., 2002; Lipton, 1980). One of the consequences of these migrations is the evolution of the gender ratio in Tumbes Department, which has increased from 106 males for 100 females in 1993 to 112:100 in 2005 (INEI, 2009). In 1993, considering just the more recent migrants, the ratio even reached 135:100 (INEI, 2009). This 1 The currency rate exchange between US dollar and Peruvian sol in April 2007 was: 1 USD ¼ 3.06 soles. 9 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 Table 4 Characteristics of migrant population based on field surveys (n ¼ 57). Variables (various units) Results Former region of residence (%) Time already spent in Tumbes (years) Average Standard deviation Main reason for migration (%) Piura (87), Lima (7), Cuzco (2), Cajamarca (2), La Libertad (2) 6.6 6.3 Improve standard of living (53), Find a job (51), Help the household (45), Exerce the present position (11), Live with relatives (7) More working opportunities (66), Known resident person (35), Shrimp farming (22) No (54), Yes (46) Reason for choosing Tumbes (%) Awareness of shrimp farming jobs before leaving (%) Decision to migrate (%) Migration participants (%) First job after arrival (%) Personal (90), Household (10) Alone (64), With family members (26), Other (10) Shrimp farming (67), Agriculture (10), Manufacturing (2), Services (21) Through friend (54), Through relative (43) Access to present job in shrimp farming (%) Remittances sent home (%) Average monthly remittances (pesos) Average proportion of wage remitted (%) Estimated future duration of stay in Tumbes (%) Yes (68.5), No (31.5) 231 40 Forever (20), 5 more years (9), From 1 to 5 years (22), Less than 1 year (14), Do not know (35) imbalance has begun to pose serious local demographic challenges involving fertility and household renewal issues. Although the majority of the population is employed on a permanent basis, immigrants and locals do not occupy the same jobs on the farms (Table 5). Locals obtain the better paid jobs in maintenance and management. Furthermore, skilled jobs (not surveyed here) are dominated by locals with higher educational status. These advantages secure positions with higher income and fewer working hours. For immigrants, the relatively lower condition is expressed by a greater willingness to change jobs. Work in shrimp farming is a transitional activity for the immigrants, who spend much less average time in this occupation than the non-migrants (4.6 against 8.8 years, respectively). The future of immigrants after having left shrimp farming needs to be thoroughly investigated but local informants declared that they either chose to stay (then working in the services or fisheries sector, or as seafood collectors), returned to their homestead, or emigrated to Ecuador or to southern Peru. These are common situations but they were dramatically intensified during the El Niño and WSS events. The population involved in resource extraction from the mangrove N Ferrenafe Pacific Ocean Zarumilla B Tumbes C Pacific Ocean Huamanga Pacasmayo Contralmirante villar Tumbes department ECUADOR A ECUADOR Piura department A PERU B Ayacaba Pacific Ocean Lima Piura C Morropon Huacabamba Nber of shrimp farm workers per district of origin Sechura 0 25 50 75 100 kilometers 1 6-7 Department 2 9 Province 3-4 12 Fig. 7. Geographic origin and number of shrimp farm workers interviewed. 10 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 Table 5 Socio-economic profiles of immigrants and locals. Variable (various units) Immigrant (n ¼ 57) Local (n ¼ 13) Average age (years) Time spent in shrimp farming (years) Life cycle status (%) Single Married Cohabiting Average number of children Education level (%) Primary Secondary Further/Higher Previous job (%) Primary sector Manufacturing Services Other Contract type (%) Permanent Temporary Present work on farma (%) Production Security Maintenance Management Ancillaryb Weekly hours (h) Mean wage (pesos) Willingness to change jobs (%) 31 4.6 33 8.8 58 21 21 1.2 31 61 8 1.5 30 65 5 31 38 31 46 6 40 8 31 23 38 8 87 13 92 8 51 29 10 4 6 61 573 Yes (78) No (22) 38.5 15 38.5 8 0 53 682 Yes (42) No (58) a Production encompasses activities dedicated to the grow-out of farmed shrimps: feeding, fertilizing, pond cleaning, harvest. Security refers to guarding the farms against intruders. Maintenance includes the maintenance of the material used in production such as pumps or aerators. Management refers to decision making with regard to buying the inputs, to the daily operation of the farm, and to timing of the harvest. b On-farm ancillary work involves activities such as cooking and cleaning the buildings. sanctuary, for example, consists predominantly (65%) of migrants mainly originating from Piura department (Ayabaca, Huancabamba and Morropon provinces) (INRENA, 2007). Overall, immigrants are deemed to play an important role in the economic viability of Tumbes shrimp farms because of the lower labour costs compared to wage expectations among the local population. This is important because reducing the cost of labour is a major target among shrimp farm owners in order to remain competitive. The clustering of relatives and acquaintances at farm level also strengthens trust between shrimp farmers and workers. For the migrants, the advantages of a job on a shrimp farm are linked to the contractual engagement in which they find security even though the contract can be easily broken. Shrimp farms also provide a job soon upon arrival, thus reducing the costs of migration. Consequently, even though shrimp farming as a pull factor remains weaker than the push factors previously mentioned, the rise of aquaculture has evidently contributed to an influx of labour facilitated by the information provided by prior migrants about employment opportunities in Tumbes. 4.4. Shrimp farming, society and the environment in Tumbes 4.4.1. Existing environmental safeguards and outstanding issues The previous sections have provided some insights into land use change, employment in the shrimp farming sector, and internal migratory flows towards Tumbes. Together, those insights illuminate how shrimp farming in Peru has redefined the interactions between population and the environment. Through its dual process of extensification and intensification, and also its relatively volatile productivity determined by environmental and market factors, shrimp farming has shaped the natural and social environments through its unprecedented use and degradation of natural resources as well as by its requirements in terms of labour. Shrimp farming has also suffered under the impacts of its own externalities such as the pollution of canals by effluents. During the recent intensification of the industry, labour requirements grew and attractivity to migrants likewise increased. Further integration of these immigrants into the local economy poses an ecological threat from the moment they choose to exploit natural resources, but can also be beneficial to locals because immigrants are more willing to occupy the low-wage jobs. Due to increasing population in the coastal area, growing inequalities between skilled and unskilled workers and the competing uses of natural resources between various stakeholders have led on some farms to acts of sabotage, arson and theft. Conflicts have opposed shrimp farmers to inhabitants of Puerto Pizarro who live on the periphery of the shrimp farms. The principles of Integrated Coastal Management have never been implemented in the Tumbes region. Most of the policies that rule the use of natural resources are sectorial and concern isolated segments of the territory. Conservation-oriented management initiatives were nonetheless adopted to tackle the depletion of natural resources, the most important policies dedicated to the conservation of coastal ecosystems having been promulgated in 1978 (protection of channels, mangrove and dry forest) and 1988 (creation of the mangrove sanctuary and its buffer zone). The sanctuary was created because of the growth of shrimp ponds in the biodiversity-rich coastal habitats. The enforcement of rules dedicated to curbing shrimp farm encroachment on the sanctuary limits has proved to be more effective than in the sanctuary’s buffer zone. Shrimp farm implantation has continued in the buffer zone, which is an area where conservation practices and more sustainable activities should have prevailed. Slack enforcement encourages over-harvesting of aquatic organisms within the sanctuary boundaries by individual collectors, who also ignore best practice rules such as a restriction of harvesting activities during the spawning season. The ban in 2000 on wild post-larvae collection along the beaches and tidal channels, which was designed to prevent the intrusion of virus-bearing larvae into the production ponds, positively contributed to a reduction of industry-related environmental impacts not just because it helped to restore the importance of post-larvae in the food chain, but also because the post-larvae harvesting was carried out using finemeshed netting, a practice which entailed unsustainable by-catch of other aquatic organisms. The management of natural resources within the sanctuary was also devolved to the local community after 1999 through the creation of a Management Committee, which includes ordinary citizens, representatives of the shrimp industry, academic institutions and some small-scale farmers and fisherfolk. The purpose of the Committee is to ensure the sustainable use of natural resources within the sanctuary and in its buffer zone. On the ground, dialogue remains limited but some progress has occurred through the constitution of action groups dedicated to promoting tourism in the sanctuary. A recent collaborative project has been successfully developed by shrimp farmers, NGOs and the SERNANP (Servicio Nacional de Areas Nacionales Protegidas) for the production of spats of Anadara tuberculosis, a local and overexploited mollusc, with a particular focus on genetic diversity and disease prevention. Overall, the management policies still remain exceedingly sectorial, i.e. do not take sufficient account of the diversity of coastal activities and their interactions. Their scope is also limited in that they do not consider the entire coastal zone as an F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 integrated system. Finally, they remain largely ineffective because the existing regulatory framework lacks the appropriate resources for enforcement. Environmental regulations dedicated to controlling the effluents released by shrimp farms were introduced in 1994. All new farms were forced to carry out an environmental impact assessment (EIA) by a certified agency, and farms above 50 ha were required to produce an EIA declaration every six months. However, local officials openly admit that farms do not comply with the rules. Encroachments and the absence of EIA compliance point to a connivance between the shrimp industry, politicians and environmental agency employees, both at local and national level. In addition to giving the industry an opportunity to be competitive in the international marketplace, the numerous fiscal advantages granted to the industry are also interpreted as a consequence of cronyism. Among the more positive recent developments, the strong reduction of antibiotics and substitution by native probiotics isolated from the wild shrimp microbiota is a welcome noteworthy (E. Mialhe, pers. comm. October 2012). The methodology for shrimp virus diagnosis based on molecular tools also promises a decrease of viral outbreaks into the environment, subsequently decreasing impacts on wild shrimp populations. Local officials also recognize the existence of many unauthorized farms, particularly in the more remote areas. Some of them were visited during the course of this study. These illegal farms maintain a tacit agreement with authorized farms to sell their produce to the conditioning plants, which require the farmers to be declared. This informal system is well known but the willingness or resources to police it remain elusive. 4.4.2. Outlook for the future The findings brought into focus by this study provide a basis for formulating a few recommendations towards an improvement of coastal management in Tumbes. Although the perspective adopted here has been restricted to shrimp farming and does not, therefore, authorize prescriptions relevant to all aspects of Integrated Coastal Management, considerations regarding land use planning, social welfare and local governance would suggest the following priorities: (i) stricter urban planning in the vicinity of existing ponds in order to limit the total length of direct contact between residential land use and ponds; (ii) a moratorium on habitat fragmentation through stronger enforcement of existing resource harvesting policies; (iii) a regeneration of forest habitats and a restoration of bare soil areas (beaches, salt marshes); (iv) improved design of canals for both water intake and effluents. (v) Better monitoring of immigrant labour should also help to anticipate the increase in local demand for goods and services. Migrant workers would also gain from a stricter enforcement of labour laws, which would not only improve their economic status and working conditions but could also limit the urge to engage unsustainably in the harvesting of wild species. The setting up of the Management Committee in 1999 remains a landmark among the more innovative governance schemes in the region. Other frameworks conducive to dialogue between the different agencies involved in shrimp farming are called for, one being cross-border cooperation with Ecuador. This perspective would allow watershed-scale integrated planning, particularly given that gold mining activity in the Ecuadorian part of the Tumbes river catchment has been releasing inorganic mercury and sodium cyanide in the water, which raises concerns for the shrimp production in the coastal areas (Guimaraes et al., 2011). 5. Conclusion This case study has analysed some impacts triggered by the development of shrimp farming in northern Peru. Since the first 11 appearance of ponds in the late 1970s, shrimp farming has experienced severe upsets due to external environmental factors such as El Niño and world market instabilities, but also to internal mismanagement. The growth of aquaculture has also induced social and economic changes: shrimp farming has created a labour market draining migrants from poorer mountainous regions. In addition to the attraction exerted by shrimp farming, social capital was also found to be a major driver of migration. The production chain relies on a diversity of segmented labour tasks. Shrimp farming was estimated to support the livelihoods of 10,000 individuals contributing to the local economy, but a more thorough diagnosis would also require accounting for livelihoods that were destroyed or stifled by the development of commercial aquaculture. By remote sensing analysis we quantified and qualified the areas over which aquaculture-related land use has expanded. As in other contexts throughout the Tropics, aquaculture has expanded at the expense of mangrove but the intensity with which this has occurred must be put into perspective because aquaculture has also developed over land cover categories initially detected as bare soil, dry forest, and savanna. A moratorium on the destruction of relict mangroves has been in place since 1988 through enforcement of a habitat conservation policy. Recent history reveals several adaptations of the local socialeecological system such as the adoption of new management practices and an intensification of production systems. By providing a quantitative account of changes in land use, farmed shrimp production, labour markets and a qualitative assessment of migratory flows, this case study contributes to the ongoing global debate on the processes and policies best adapted to attaining sustainability in resource allocation and management. It promotes a holistic perspective on the impacts of shrimp farming on the socioeconomic and natural environments at the local and regional levels, with global ramifications and feedbacks relating to world markets and to epidemiological and climatic hazards. Acknowledgements The authors acknowledge the Isis program, Centre National d’Etudes Spatiales (CNES), for providing SPOT 5 images at a negotiated price. Three anonymous reviewers contributed helpful comments, resulting in a more balanced treatment of the many issues and specialisms that we have attempted to combine here in a cross-disciplinary synthesis. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.ocecoaman.2012.12.014. References Adduci, M., 2009. Neoliberal wave rocks Chilika lake, India: conflict over intensive aquaculture from a class perspective. J. Agrar Change 9 (4), 484e511. Adger, W.N., Kelly, P.M., Winkels, A., Huy, L.Q., Locke, C., 2002. Migration, remittances, livelihood trajectories, and social resilience. Ambio 31, 358e366. Armitage, D., 2002. Socio-institutional dynamics and the political ecology of mangrove forest conservation in Central Sulawesi, Indonesia. Glob. Environ. Change 12, 203e217. Asian Development Bank (ADB), 2004. Special Evaluation Study on Small-scale Freshwater Rural Aquaculture Development for Poverty Reduction. Operations Evaluation Department SST. REG 2004-07. Bailey, C., 1988. The social consequences of tropical shrimp mariculture development. Ocean Shore Manag. 11 (1), 31e44. Barbier, E.B., Koch, E.W., Silliman, B.R., Hacker, S.D., Wolanski, E., Primavera, J., Granek, E.K., Polasky, S., Aswani, S., Cramer, L.A., Stoms, D.M., Kennedy, C.J., Bael, D., Kappel, C.V., Perillo, G.M.E., Reed, D.J., 2008. Coastal ecosystem-based management with non-linear functions and values. Science 319, 321e323. Barham, B., Clark, M., Katz, E., Schurman, R., 1992. Nontraditional agricultural exports in Latin America. Lat. Am. Res. Rev. 27 (2), 43e82. 12 F. Mialhe et al. / Ocean & Coastal Management 73 (2013) 1e12 Béné, C., 2005. The good, the bad and the ugly: discourse, policy controversies and the role of science in the politics of shrimp farming development. Dev. Policy Rev. 23 (5), 585e614. Berger, C., 2002. Situacion del cultivo de langostinos en el Peru. In: Proceedings of the Primer congreso nacional de acuicultura, Lima. Berger, C., Quispe, M., Talavera, V., 2004. Programa Nacional para la competitividad de la acuicultura langostinera en el Peru 2005-2014. Asociación langostinera Peruana (ALPE). Berlanga-Robles, C.A., Ruiz-Luna, A., Bocco, G., Vekerdy, Z., 2011. Spatial analysis of the impact of shrimp culture on the coastal wetlands on the Northern coast of Sinaloa, Mexico. Ocean Coast Manag. 54, 535e543. Best, B., Kessler, M., 1995. Biodiversity and Conservation in Tumbesian Ecuador and Peru. BirdLife International, Cambridge. Biao, X., Kaijin, Y., 2007. Shrimp farming in China: operating characteristics, environmental impact and perspectives. Ocean Coast Manag. 50, 538e550. Campbell, J., Wynne, R., 2011. Introduction to Remote Sensing, fifth ed. The Guilford Press, New York. Chavez, P.S., Kwarteng, A.Y., 1989. Extraction spectral contrast in Landsat Thematic Mapper image data using selective principal component analysis. Photogramm. Eng. Rem. Sens. 55, 339e348. Chua, T.-.E., Bonga, D., Bermas-Atrigenio, N., 2006. Dynamics of integrated coastal management: PEMSEA’s experience. Coast Manag. 34 (3), 303e322. Curran, S., 2002. Migration, social capital and the environment: considering migrant selectivity and network in relation to coastal ecosystems. Popul. Dev. 28, 89e125. Dai, X., 1998. The effects of image misregistration on the accuracy of remotely sensed change detection. IEEE Trans. Geosci. Rem. Sci. 36 (5), 1566e1577. Edwards, P., 2000. Aquaculture poverty impacts and livelihoods. ODI Nat. Res. Perspect. 56, 1e4. Eklundh, L., Singh, A., 1993. A comparative analysis of standardised and unstandardised Principal Component Analysis in remote sensing. Int. J. Rem. Sens. 14 (7), 1359e1370. Evans, Y., Tveteras, S., 2011. Status of Fisheries and Aquaculture Development in Peru: Case Studies of Peruvian Anchovy Fishery, Shrimp Aquaculture, Trout Aquaculture and Scallop Aquaculture. FAO, Background Report. FAO, 2010. The State of World Fisheries and Aquaculture. FAO Fisheries and Aquaculture Department, Rome. FAO, 2012. Aquaculture Production 1950e2010. FishStat Plus. FAO Fisheries and Aquaculture Information and Statistics Service, Rome. Flaherty, M., Vandergeest, P., 1998. Low-salt shrimp aquaculture in Thailand: goodbye coastline, Hello Khon Kaen! Environ. Manag. 22 (6), 817e830. Foody, G.M., 2002. Status of land cover classification accuracy assessment. Rem. Sens. Environ. 80, 185e201. Foody, G.M., 2004. Thematic map comparison: evaluating the statistical significance of differences in classification accuracy. Photogramm. Eng. Rem. Sens. 70 (5), 627e633. Goss, J., Burch, D., Rickson, R.E., 2000. Agri-food restructuring and third world transnationals: Thailand, the CP group and the global shrimp industry. World Dev. 28 (3), 513e530. Gräslund, S., Bengtsson, B.-E., 2001. Chemicals and biological products used in south-east Asian shrimp farming, and their potential effect on the environment e a review. Sci. Total Environ. 280 (1e3), 93e131. Gray, C.L., 2009. Environment, Land, and rural out-migration in the southern Ecuadorian Andes. World Dev. 37 (2), 457e468. Green, E.P., Mumby, P.J., Edwards, A.J., Clark, C.D., 1996. A review of remote sensing for the assessment and management of tropical coastal resources. Coast Manag. 24 (1), 1e40. Guimaraes, J.R.D., Betancourt, O., Miranda, M.R., Barriga, R., Cueva, E., Betancourt, S., 2011. Long-range effect of cyanide on mercury methylation in a gold mining area of Southern Ecuador. Sci. Total Environ. 409, 5026e5033. Hall, D., 2004. Explaining the diversity of Southeast Asian shrimp farming aquaculture. J. Agrar Change 4 (3), 315e355. Hamilton, S.E., 2011. The Impact of Shrimp Farming on Mangrove Ecosystems and Local Livelihoods along the Pacific Coast of Ecuador. PhD thesis., the University of Southern Mississippi. Heck, S., Béné, C., Reyes-Gaskin, R., 2007. Investing in African fisheries: building links to the Millenium Development goals. Fish Fish 8 (3), 211e226. Hishamunda, N., Ridler, N.B., 2006. Farming fish for profits: a small step towards food security in sub-Saharan Africa. Food Pol. 31 (5), 401e414. Hui, F., Xu, B., Huang, H., Yu, Q., Gong, P., 2008. Modelling spatial-temporal change of Poyang Lake using multitemporal Landsat imagery. Int. J. Rem. Sens. 29 (20), 5767e5784. Instituto Nacional de Estadistica e Informatica (INEI), 2009. Perfil sociodemografico del departamento de Tumbes. Instituto Nacional de Estadistica e Informatica, Lima. Instituto Nacional de Recursos Naturales (INRENA), 2007. Plan maestro del santuario nacional los manglares de Tumbes 2007e2011. Instituto Nacional de Recursos Naturales, Lima. Islam, M.S., 2008. From pond to plate: towards a twin-driven commodity chain in Bangladesh shrimp aquaculture. Food Pol. 33 (3), 209e223. Lee, E.S., 1966. A theory of migration. Demography 3 (1), 47e57. Lillesand, T., Kiefer, R.W., Chipman, J., 2007. Remote Sensing and Image Interpretation, sixth ed. J. Wiley & Sons, New York. Lipton, M., 1980. Migration from rural areas of poor countries: the impact on rural productivity and income distribution. World Dev. 8 (1), 1e24. Lu, D., Weng, Q., 2007. A survey of image classification methods and techniques for improving classification performance. Int. J. Rem. Sens. 28 (5), 823e870. Malecki, E.J., 1997. Technology and Economic Development: the Dynamics of Local, Regional and National Competitiveness. Longman, London. Massey, D., 1990. Social structure, household strategies, and the cumulative causation of migration. Popul. Index 56 (1), 3e26. McClennen, C., 2004. White Spot Syndrome Virus. The Economic, Environmental and Technical Implications on the Development of Latin American Shrimp Farming. M.A. thesis., Tufts University, Medford. Mialhe, F., 2010. Le développement de l’aquaculture saumâtre dans l’Aire Pacifique. Evolution des paysages, dynamiques socio-économiques et impacts environnementaux dans deux territoires au Pérou et aux Philippines. PhD thesis., Université Paris 7 Diderot, Paris. Myint, H., 1958. The ‘Classical Theory’ of international trade and the underdeveloped countries. Econ. J. 68, 317e337. Pàez-Osuna, F., 2001. The environmental impact of shrimp aquaculture: causes, effects and mitigating alternatives. Environ. Manag. 28 (1), 131e140. Paul, B.G., Vogl, C.R., 2011. Impacts of shrimp farming in Bangladesh: challenges and alternatives. Ocean Coast Manag. 54 (3), 201e211. Portes, A., 1995. The Economic Sociology of Immigration: Essays on Networks, Ethnicity, and Entrepreneurship. Russell Sage Foundation, New York. Primavera, J.H., 1997. Socio-economic impacts of shrimp culture. Aquac. Res. 28, 815e827. Primavera, J.H., 2006. Overcoming the impacts of aquaculture on the coastal zone. Ocean Coast Manag. 49 (9e10), 531e545. Resurrección, B.P., Sajor, E.E., 2010. “Not a real worker”: gendering migrants in Thailand’s shrimp farms. Int. Migrat. 48 (6), 102e131. Rogan, J., Chen, D.M., 2004. Remote sensing technology for mapping and monitoring land-cover and land-use change. Prog. Plann. 61, 301e325. Rubio, E.T., 2007. Climate Change Impacts and Adaptation in Peru: the Case of Puno and Piura. Human Development Report, UNDP. Sathirathai, S., Barbier, E., 2001. Valuing mangrove conservation in southern Thailand. Contemp. Econ. Pol. 19, 109e122. Singh, A., 1989. Digital change detection techniques using remotely-sensed data. Int. J. Rem. Sens. 10 (6), 989e1003. Stokstad, E., 2010. Down on the shrimp. Science 328, 1504e1505. Stonich, S.C., 1995. The environmental quality and social justice implications of shrimp mariculture development in Honduras. Hum. Ecol. 23 (2), 143e168. Story, M., Congalton, R., 1986. Accuracy assessment: a user’s perspective. Photogramm. Eng. Rem. Sens. 52 (3), 397e399. Thorpe, A., Ibarra, A.A., Reid, C., 2000. The new economic model and marine fisheries development in Latin America. World Dev. 28 (9), 1689e1702. Tiebout, C., 1956. A pure theory of local expenditures. J. Polit. Econ. 64 (5), 416e424. Twilley, R.R., Montano, M., Valdivieso, J.M., Bodero, A., 1999. The environmental quality of coastal ecosystems in Ecuador: implications for the development of integrated mangrove and shrimp pond management. In: Yanez-Arancibia, A., Lara-Dominguez, A.L. (Eds.), Ecosistemas de manglar en America Tropical. IUCN/ORMA, NOAA/NMFS, pp. 199e230. Valderrama, D., Hishamunda, N., Zhou, X., 2010. Estimating employment in world aquaculture. FAO Aquacult. Newsl. 45, 24e25. Valiela, I., Bowen, J.L., York, J.K., 2001. Mangrove forests: one of the world’s threatened major tropical environments. BioScience 51 (10), 807e815.