Open Collections

UBC Faculty Research and Publications

Agricultural land-use change in Kerala, India : Perspectives from above and below the canopy Fox, Thomas A.; Rhemtulla, Jeanine Marie; Ramankutty, Navin; Lesk, Corey; Kunhamu, T. K.; Coyle, Theraesa Jul 1, 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
52383-Fox_T_et_al_Agricultural_land-use.pdf [ 1.02MB ]
Metadata
JSON: 52383-1.0378871.json
JSON-LD: 52383-1.0378871-ld.json
RDF/XML (Pretty): 52383-1.0378871-rdf.xml
RDF/JSON: 52383-1.0378871-rdf.json
Turtle: 52383-1.0378871-turtle.txt
N-Triples: 52383-1.0378871-rdf-ntriples.txt
Original Record: 52383-1.0378871-source.json
Full Text
52383-1.0378871-fulltext.txt
Citation
52383-1.0378871.ris

Full Text

1  https://doi.org/10.1016/j.agee.2017.05.002 1 Title: Agricultural land-use change in Kerala, India: 2 Perspectives from above and below the canopy  3 Authors: Thomas A. Foxa*, Jeanine M. Rhemtullaa,1, Navin 4 Ramankuttya,2, Corey Leska,3, Theraesa Coylea,4, TK 5 Kunhamub 6 *Corresponding author, currently in the Department of 7 Geography, University of Calgary, 2500 University Drive 8 NW, Calgary, AB, Canada T2N 1N4, email: 9 thomas.fox@ucalgary.ca 10 a. Department of Geography, McGill University, 845 Rue 11 Sherbrooke O, Montréal, QC, Canada H3A 0G4 12 b. Department of Silviculture and Agroforestry, College of 13 Forestry, Kerala Agricultural University, KAU P.O. 14 Vellanikkara, National Highway 47, Thrissur, Kerala 680656, 15 India 16 1. Present address: Department of Forest and Conservation 17 Sciences, University of British Columbia, 2329 West Mall, 18 Vancouver, BC, Canada V6T 1Z4 19 2. Present address: Liu Institute for Global Issues and 20 Institute for Resources, Environment, and Sustainability, 21 Fox, T. A., Rhemtulla, J. M., Ramankutty, N., Lesk, C., Coyle, T., & Kunhamu, T. K. (2017). Agricultural land-use change in Kerala, India: Perspectives from above and below the canopy. Agriculture, Ecosystems & Environment, 245, 1-10.2  University of British Columbia, 2329 West Mall, Vancouver, 22 BC, Canada V6T 1Z4 23 3. Present address: Columbia University Center for Climate 24 Systems Research, 116th Street & Broadway, New York, NY 25 10027, United States 26 4. Present address: Fisheries and Oceans Canada, Centre 27 for Aquaculture and Environmental Research, 4160 Marine 28 Drive, West Vancouver, BC V7V 1N6 29 Abstract – Despite the availability of a wide range of tools, 30 measuring and explaining changes in land cover and land 31 use in tropical regions can be extremely challenging. Kerala, 32 India, is a biodiversity hotspot with a high population density 33 and a long history of complex agricultural land-use patterns. 34 Some reports suggest that agriculture in Kerala, which 35 historically is rice paddy-wetland and agroforestry-based, is 36 on the decline. However, the evidence is often anecdotal, 37 especially with regards to smallholding homegarden 38 agriculture. In this study we employ mixed methods, 39 including remote sensing, quantitative household surveys, 40 and semi-structured interviews, to unravel the complex land-41 cover and land-use changes occurring in Kerala.  42 3  Results indicate that, from a land-cover change perspective, 43 agroforests are in dynamic equilibrium with other land 44 covers, being cleared for roads and new buildings, but offset 45 by the expansion of younger, less diverse agroforests into 46 paddy wetlands. Yet beneath the canopy, agroforests are 47 undergoing rapid land-use change not discernible using 48 remote sensing. These changes include a reported decrease 49 in the cultivation of 80% of Kerala’s primary crop species 50 during 2003-2013, alongside a dramatic decline in chickens 51 (from 12.5 to 2.6 per homestead on average) and cows 52 (from 1.7 to 0.8). Over this period, no crop increased in 53 cultivation. According to farmers, the primary drivers of this 54 shift were declining profitability of agriculture in Kerala, 55 labour shortages, unreliable weather, unfamiliar pests and 56 diseases, and government policy.  57 Despite the undeniable move away from agricultural activity 58 in homegardens, we conclude that these ecologically and 59 culturally important systems are not disappearing, but rather 60 evolving to meet the needs of a less agricultural Kerala.  Our 61 research highlights the value of using mixed methods for 62 characterizing land-use and land-cover histories in tropical 63 regions. 64 4  Keywords: homegarden; land-use management; tropical 65 agriculture; mixed methods; agroforestry 66 1. Introduction 67 Changes in land use and land cover are an important 68 manifestation of human interactions with the environment, 69 with manifold consequences for ecosystems and human 70 livelihoods (DeFries et al., 2007; Foley et al., 2005). There 71 has been a rapid rise in scholarship over the last two 72 decades aiming to understand the ecosystem service 73 tradeoffs related to land-use practices (DeFries et al., 2004; 74 Nair et al., 2009; Tomscha et al., 2016). How best to 75 manage landscapes to balance human needs and 76 environmental conservation has become a key focus of 77 research and policy debate (Benton, 2007; DeFries and 78 Rosenzweig, 2010; Green et al., 2005). 79 Yet land use must be accurately measured before it can be 80 effectively managed. Various quantitative and qualitative 81 methods have been developed to identify and measure 82 changes in land use and cover (Lambin et al., 2003; 83 Luyssaert et al., 2011; Munsi et al., 2010; Veldkamp and 84 Verburg, 2004). These include, but are not limited to, 85 classification of remotely sensed imagery, physical field 86 5  measurements, consulting government records, and 87 interviewing land users or occupants.  88 Measuring land-use/cover change (LUCC) is complicated by 89 the dynamic nature of human-managed landscapes, which 90 experience changes at multiple scales, and not necessarily 91 at the same time. This is especially true of tropical 92 landscapes in developing countries, in which agricultural 93 land holdings tend to be both small and diverse in style of 94 agriculture. Agricultural landscapes in these regions range 95 from subsistence- to commercial-based and tend to exhibit 96 high spatiotemporal variability in crop selection, which can 97 be based on markets, available technologies, government 98 incentives, pest prevalence, investment potential, and so on 99 (Altieri, 2009; Wrigley, 1971).  100 Kerala, a tropical state in South India, is an example of a 101 region with a dynamic history of land-use change that has 102 not been particularly well-documented. Archaeological 103 evidence suggests that Kerala participated in global 104 agricultural markets for at least 2000 years, trading spices 105 first with the Romans, and later with Portuguese, Dutch, and 106 British merchants (Jeffrey, 2001). In addition to spices such 107 as black pepper (Piper nigrum) and cardamom (Elettaria 108 cardamomum), Kerala has been a major producer and 109 6  exporter of rice (Oryza sativa) and coconut (Cocos nucifera) 110 (Kumar, 2005). Traditionally, much of Kerala’s agricultural 111 activity has centered on homegardens. According to (Kumar 112 and Nair, 2004), homegardens are “intimate, multi-story 113 combinations of various trees and crops, sometimes in 114 association with domestic animals, around homesteads.” 115 Homegardens, which are the result of generations of 116 successive crop intensification, are renowned for their 117 species richness, multifunctionality and sustainability (Kumar 118 et al., 1994; Kumar and Nair, 2004). As such, it is important 119 to differentiate between homegardens, which are the places 120 – houses and farms – where people live, and agroforestry, 121 which is a land cover category. Agricultural land in a 122 homegarden is primarily agroforest, in which plantation crops 123 such as coconut, banana, or rubber (Hevea brasiliensis) are 124 either well integrated, or in which plantation-style cultivation 125 constitutes a limited proportion of homegarden area. 126 Agroforests, on the other hand, consist of not only 127 homegardens but also mixed agroforests not associated with 128 a homestead. The vast majority of Kerala’s rural homesteads 129 contain homegardens, yet these farms are quite small, and 130 other forms of agriculture such as plantations and paddy 131 land also exist. 132 7  Rapid agricultural land-use changes have occurred in Kerala 133 since the 1970s. In particular, local land-use scholars have 134 noted a shift towards monoculture and conventional cash-135 crop agroforestry, at the expense of traditional, species-rich 136 homegardens (Kumar, 2005; Peyre et al., 2006). While a 137 shift towards monoculture-style agriculture would be 138 consistent with shifts observed in other developing regions, it 139 would be at odds with the fundamental cultural importance of 140 tropical agroforestry to rural Keralites (Kumar and Nair, 141 2004). Furthermore, observations of this transition have 142 been mostly anecdotal, as land-cover data collected by the 143 state fail to account for the complexity of Kerala’s agricultural 144 landscapes (Kumar, 2005). In addition to the alleged shift 145 from traditional to monoculture-style agriculture, another 146 important land-use change has been the recent conversion 147 of paddy land into simple agroforests and other agricultural 148 crops (Guillerme et al., 2011). It is important to note that new 149 agroforests are often fundamentally different than traditional 150 homegarden agroforests, as the latter are, by definition, 151 intensively managed, more complex, and much older. 152 Understanding LUCC in complex landscapes requires a 153 multi-faceted approach (Lambin et al., 2003; Veldkamp and 154 Verburg, 2004). Using Kerala as a case study, we explored 155 8  the use of a mixed-methods approach to gain a more 156 complete understanding of LUCC at multiple scales. First, 157 using high-resolution satellite imagery, we estimated broad-158 scale land-cover changes in three of Kerala’s environmental 159 and agricultural zones. We then zoomed in to the scale of 160 the homegarden to conduct quantitative household surveys 161 and semi-structured interviews with farmers. While the 162 remote sensing analysis aimed to identify changes in the 163 areal extent of land cover, the farm-scale component of the 164 study aimed to identify the individual land-use changes that 165 were occurring, as well as the drivers of these changes. 166 Finally, we synthesized the disparate data sources to 167 develop a coherent explanation of agricultural LUCC 168 changes in Kerala over the last decade. 169 2. Methods 170 2.1 Study area 171 Despite its small size (38 863 km2), Kerala is topographically 172 and ecologically diverse, consisting of a mix of coastland, 173 wetlands, and plains to the west, and rolling hills and the 174 Western Ghats mountain range to the east. Crop choice 175 depends primarily on topography and elevation, but also on 176 crop profitability, soil type, availability of irrigation, and public 177 9  policy (Guillerme et al., 2011; Kannan and Pushpangadan, 178 1990; Narayanan, 2006). The most common crops grown in 179 Kerala are rice in the lowlands; tea (Citrus sinensis), coffee 180 (mostly Coffea arabica and C. canephora), and spices in the 181 uplands; and banana (various species), coconut and 182 arecanut palms (Areca catechu) nearly everywhere (Kumar, 183 2005). Common homegarden food crops include jackfruit 184 (Artocarpus heterophyllus), mango (Mangifera indica), curry 185 tree (Murraya koenigii), and banana, among many others. 186 Yet Kerala’s crop composition has experienced considerable 187 shifts since the 1950s, characterized by declines in rice and 188 increases in coconut and rubber (Kumar, 2005).  189 In addition to being biophysically, ecologically, and 190 agriculturally diverse, Kerala is socially, culturally, and 191 economically diverse, and is distinct from the rest of India. 192 Kerala has the highest Human Development Index (0.825 in 193 2015;(United Nations Development Program, 2015) and 194 highest literacy rate (93.91%; (Government of India, 2011) of 195 any state in India. Kerala’s universal social services have 196 resulted in a healthy, highly educated population that often 197 travels abroad to find gainful employment (Prakash, 1998). It 198 is estimated that one person works overseas for every five 199 people employed in Kerala, with foreign remittances 200 10  accounting for roughly 25% of the state’s economy 201 (Zachariah and Rajan, 2012). This mass exodus of skilled 202 and unskilled workers has come hand-in-hand with labour 203 shortages since the 1970s, which are generally assumed to 204 have contributed to the decline in paddy cultivation and a 205 rise in agroforestry in the 1980s and 90s (Kannan, 1998). 206 We conducted land-cover analyses in three panchayats (the 207 smallest political administrative unit in Kerala): Avinissery, 208 Kalikavu, and Poothrikka (Figure 1). Panchayat choice was 209 based on the availability of high-quality archival satellite 210 imagery as well as to best represent Kerala’s diverse natural 211 environments and varied population density. Avinissery is a 212 densely populated, low-elevation panchayat consisting 213 primarily of homegardens and paddy rice. Kalikavu is close 214 to the Western Ghats, has low population density, larger 215 farm size, and produces large amounts of tree crops such as 216 rubber, coconut and arecanut. Poothrikka, which produces 217 rice, rubber, and homegarden crops (e.g. mango, jackfruit, 218 and bananas), is between Avinissery and Kalikavu with 219 regards to elevation and population density. We selected five 220 additional panchayats for landholder surveys and interviews, 221 using the same environmental and demographic criteria as 222 described for the first three (Figure 1). 223 11   224 Fig. 1. Study area in Kerala, India. We conducted quantitative 225 surveys and semi-structured interviews in 8 panchayats (black), 226 representing 8 districts (dark gray) across the state. Land-cover 227 classification and change detection were conducted using satellite 228 imagery acquired from panchayats C, E, and F. The districts 229 (panchayats in parentheses) labeled are: A: Kozhikkode 230 (Thamarassery); B: Wayanad (Vengappally); C: Malappuram 231 (Kalikavu); D: Palakkad (Kadampazhipuram); E: Thrissur 232 (Avinissery); F: Ernakulam (Poothrikka); G: Idukki (Kattappana); 233 H: Alappuzha (Thiruvanvandoor). The inset map is of peninsular 234 India. 235  236 12  2.2 Remote sensing & land-cover change 237 For each of Avinissery, Poothrikka, and Kalikavu, we 238 acquired an IKONOS-2 image from early 2000 and a 239 GeoEye-1 image from 2012 (Table 1). We selected the 240 imagery using the following criteria: 1) high-quality images 241 with low cloud cover (<5%); 2) sufficient spatial overlap 242 between paired images to encompass the entire panchayat; 243 3) paired images as close as possible to 10 years apart; 4) 244 image pairs for each panchayat comprised a temporally 245 coincident 10-year period; and 5) minimal seasonal variation 246 between images in a pair. We collected ground control 247 points between June and November 2013 and used them to 248 georeference GeoEye-1 images. GeoEye-1 images were 249 then used for co-registration of IKONOS-2 images, and all 250 products were orthorectified using a 30 m ASTER DEM and 251 subsequently pan-sharpened. We used ENVI version 5.1 252 (ENVI, 2013) and ArcGIS version 10.3 (ESRI, 2011) for all 253 preprocessing. 254 Table 1. Satellites and imagery used to quantify land-cover 255 changes in selected Panchayats of Kerala state, India. 256 Panchayat Satellitea Bandsb Resolutionc (m) Coverage (km2) Date Cloud (%) Avinissery IKONOS-2 4 + Pan 3.2 (0.8) 27.06 2-Apr-2001 0 Avinissery GeoEye-1 4 + Pan 2.0 (0.5) 42.84 10-Dec-2012 5 13  Panchayat Satellitea Bandsb Resolutionc (m) Coverage (km2) Date Cloud (%) Kalikavu IKONOS-2 4 + Pan 3.2 (0.8) 34.81 22-Sep-2003 0 Kalikavu GeoEye-1 4 + Pan 2.0 (0.5) 34.81 13-Jan-2012 0 Poothrikka IKONOS-2 4 + Pan 3.2 (0.8) 44.88 13-Apr-2002 0 Poothrikka GeoEye-1 4 + Pan 2.0 (0.5) 44.88 1-Feb-2012 0 a 257 IKONOS-2: launched 29/09/1999, Global Average 258 Georeferenced Horizontal Accuracy: 15 m; GeoEye-1: 259 launched 06/09/2008, Global Average Georeferenced 260 Horizontal Accuracy: <4 m. 261 b 262 Spectral band wavelength range (in nm) for IKONOS-2: 263 Panchromatic − 526 to 929, Blue − 445 to 516, Green − 264 506 to 595, Red − 632 to 698, NIR − 757 to 853; GeoEye-265 1: Panchromatic − 450 to 800, Blue − 450 to 510, Green − 266 510 to 580, Red − 655 to 690, NIR − 780 to 920. 267 c 268 Presented: multispectral (panchromatic). 269 Preliminary pixel-based classification yielded insufficient 270 accuracy for the purpose of this study. While supervised 271 classification has been successfully used for land-cover 272 analysis in numerous contexts (e.g. Fretwell et al., 2012; 273 Goetz et al., 2003; Gutierrez and Johnson, 2012; Mumby 274 and Edwards, 2002), Kerala’s rural landscapes are a 275 complex mosaic of wetlands, tree plantations, and 276 agroforests with large spectral variability. Furthermore, built 277 surfaces such as houses and roads are often obscured by 278 14  overhanging trees or tree shadows, which led to an 279 overestimation of tree cover and underestimation of built 280 surfaces. We next attempted object-oriented classification, 281 but encountered the same issues with overhanging trees 282 and shadows. 283 We therefore opted for a manual classification approach 284 (although we used supervised classification to guide our 285 analysis as described in the next paragraph). Overhanging 286 trees and shadows were clearly visible to the naked eye, but 287 mischaracterized by both pixel-based and object-based 288 classification. Manually digitizing land-cover polygons by 289 hand is often more accurate, as shape, texture, and context 290 can be employed, in addition to spectral characteristics 291 (Lillesand et al., 2014; Lu and Weng, 2007). Manual 292 classification, or a combination of manual and object-based 293 image classification (OBIA), has been the preferred choice 294 for classifying complex tropical landscapes (Gibbs et al., 295 2010; Ramdani and Hino, 2013). 296 The major shortfall of manual classification is the necessary 297 time investment. Because classifying all 6 images in their 298 entirety was too time consuming (Achard et al., 2012; 299 Shimabukuro et al., 2014), we adopted a systematic 300 unaligned sampling approach (Bellhouse, 1977). For each of 301 15  the three panchayats, we divided image pairs into 8 equal-302 area sections, generated two random points for each 303 section, and used these 96 points to generate square 0.75 304 ha sample areas in ArcMap. We conducted maximum 305 likelihood supervised classifications using ENVI to ensure 306 that image samples were representative of the overall 307 image. Land-cover variability between samples and parent 308 images ranged from 0.2 to 7.9%. 309 We manually classified all 96 sample images into five land-310 cover classes: agroforests, bare ground, built environments, 311 water, and wetlands. Agroforests in Kerala are the dominant 312 land-cover class, and typically consist of mixed tree crop 313 species around homesteads, and in some areas 314 monoculture plantations (especially rubber). In this study we 315 did not differentiate between monoculture agroforestry 316 (a.k.a. silviculture) and polyculture agroforestry because: 1) 317 the two exist along a continuum with no clear point of 318 demarcation, and 2) our remote sensing approach could not 319 differentiate between these two sub-categories. Wetlands, 320 which are often the lowest-lying regions in Kerala, occupy 321 the remainder of the agricultural land, and are used primarily 322 for the cultivation of rice. However, fallowing paddy fields, 323 natural wetlands, and other non-tree wetland crops (e.g. 324 16  tapioca) are also considered in the wetland class. In this 325 study, all agriculture was classified into either wetland 326 (mostly rice paddy) or agroforestry, as these cover the 327 overwhelming majority of the landscape. It is important to 328 note that this land-cover classification is a broad 329 simplification of a highly complex landscape. Furthermore, 330 while non-treed dryland agriculture does exist in Kerala (e.g. 331 cassava), it is effectively absent from our selected study 332 regions.  333 Land-cover polygons were manually delineated and 334 classified if at least one of their axes was longer than 5 335 metres. Linear features were ignored if they were fewer than 336 3 pixels in width, because they could not otherwise be 337 reliably identified. In the rare case that an object could not be 338 identified by the user, images were opened in eCognition 339 Developer 8.8 so that the objects could be first isolated with 340 an appropriate segmentation, and then compared using 341 texture, shape, and spectral characteristics. 342 Land-cover change was assessed by subtracting one 343 classified image from the other (Singh, 1989), and first-order 344 Markov models were used to correct for variable image 345 acquisition dates (Urban and Wallin, 2002).  346 17  We could not validate the land-cover classification from this 347 study with ground truth points because image acquisition 348 dates ranged from one to twelve years prior to fieldwork. 349 Given the rapid change in Kerala’s landscape, any validation 350 conducted using recent field data would be highly 351 inaccurate. Furthermore, our GPS measurements would not 352 have provided a reliable validation, because we could not 353 make use of differential correction in most panchayats due to 354 their remoteness. But given the distinct textural, structural, 355 topological and spectral characteristics of Kerala’s 356 landscapes, the features digitized for this study were easily 357 discernable with the naked eye. In fact, scholars frequently 358 use both IKONOS-2 and GeoEye-1 imagery for validation of 359 lower resolution imagery (Huang et al., 2009; Potapov et al., 360 2014; Wickham et al., 2013), and here we used them directly 361 to classify land cover. 362 2.3 Surveys, interviews, & farm-scale change  363 2.3 Surveys, interviews, & farm-scale change  364 We conducted quantitative surveys and semi-structured 365 interviews at 115 homegardens between July and October of 366 2013. To maximize representation of Kerala’s geographical 367 diversity, we visited farmers in one panchayat in each of 368 18  eight contiguous districts in central Kerala (Figure 1). 369 Numerous criteria were considered for the selection of 370 panchayats, including the availability of satellite data (see 371 Section 2.2), population density, and elevation, with the aim 372 of achieving the broadest possible representation of Kerala’s 373 landscapes. In each panchayat, we randomly selected 15 374 farmers from household registries provided by local 375 governments, and included them in our sample if their 376 homegarden was at least 0.1 ha and contained at least 3 377 different cultivated tree species with a variety of understory 378 crops. In total, only one household failed to meet the 379 homegarden criteria, and two others chose not to participate. 380 In these cases, another homegarden was randomly selected 381 in order to keep samples sizes consistent between 382 panchayats. Due to logistical constraints, we were able to 383 visit only 10 homegardens in Wayanad.  384 We conducted both the quantitative surveys and semi-385 structured interviews at the homes of the respondents. For 386 each of the 115 homegardens visited, we first conducted the 387 quantitative survey, which lasted approximately 30 minutes, 388 followed by the semi-structured interview, which lasted 389 anywhere between 30 and 90 minutes. All surveys and 390 interviews were conducted with a head of household, but in 391 19  many cases the entire family contributed. A translator from 392 Kerala Agricultural University was hired to assist in 393 communicating with respondents not fluent in English. 394 Our quantitative surveys were designed to elucidate land-395 use histories by comparing the primary crops and livestock 396 produced on the homestead in 2013, and ten years prior, in 397 2003. Each survey consisted of a set of questions on 398 cultivation histories for 15 common crops (listed in Figure 4): 399 1) Do you grow this crop?; 2) Do you ever buy this crop?; 3) 400 Ten years ago, did you produce more, less, or the same 401 amount of this crop?; 4) Do you ever sell this crop?; 5) Ten 402 years ago, did you sell more, less, or the same amount of 403 this crop? We posed similar questions for livestock (cows 404 and chickens), asking farmers to estimate the number of 405 heads owned. In our quantitative surveys we also collected 406 location and size of homegardens using GPS (Trimble Juno 407 5 Handheld), and demographic data (e.g. size of family, 408 primary source of income). In analyzing crop and livestock 409 production, we controlled for changes in area by excluding 410 any homegardens that experienced changes in property size 411 between 2003 and 2013 (n = 23). 412 Our semi-structured interviews sought primarily to explore 413 the drivers of land-use change between 2003 and 2013. Our 414 20  leading question was: “Has agriculture on your land and/or in 415 this panchayat decreased over the past 10 years?” Following 416 this question, our interviews developed freely in various 417 directions, but were guided generally by questions such as 418 “What has caused agriculture to decrease on your 419 homegarden?” and “What has caused agriculture to 420 decrease in Kerala”? We recorded answers on paper and 421 digitized them using RQDA qualitative data analysis 422 software, developing 99 codes and 14 themes from 115 files 423 (Huang, 2012). 424 3. Results 425 3.1 Current homegarden characterstics 426 Average homegarden size ranged from 0.19 ha in 427 Thamarssery to 0.67 ha in Vengapally, with a mean across 428 all panchayats of 0.34 ha (Table 2). In general, 429 homegardens were smaller in densely populated 430 panchayats, which tend to be closer to the coast, and larger 431 in less populated areas, usually closer to the mountains. Of 432 the 115 farmers interviewed, 52% and 28% relied on farming 433 as their primary and secondary sources of income, 434 respectively, while 20% used their homegardens for only 435 personal use (Table 2). Many respondents (64%) owned 436 21  additional agricultural land nearby that averaged 0.87 ha and 437 was typically wetland (often in fallow) or plantation (Table 2). 438 Duration of homegarden ownership was highly variable, 439 ranging from 1-60 years with the current owner, and 1 year 440 to “time immemorial” with the family (Table 2). 441 Table 2. Descriptive statistics for selected Panchayats of Kerala 442 state, India. 443    Other Land Ownership (years) Agricultural Income Family Size Panchayat n Area (ha) Percent Farmersa Mean Area (ha) With Family Current Owner Primary Source (%) Secondary Source (%) 2003 2013 Avinissery 15 0.26 53 0.85 156 28 40 33 5.4 4.5 Kadampazhipuram 15 0.42 53 0.97 102 34 53 7 5.3 4.7 Kalikavu 15 0.23 60 0.81 43 26 60 27 7.6 6.7 Kattappana 15 0.37 53 1.85 39 21 80 13 5.4 5.4 Thiruvanvandoor 15 0.24 93 0.80 161 29 27 53 4.6 4.3 Poothrikka 15 0.33 40 0.19 118 35 40 27 6.1 4.9 Thamarassery 15 0.19 67 0.39 57 26 33 47 5.2 3.7 Vengapally 10 0.67 90 1.06 82 19 80 20 5.0 4.5 Total 115 – – – – – – – – – Mean – 0.34 63.63 0.87 94.75 27.25 51.63 28.38 5.58 4.84 St. Deviation – 0.16 18.82 0.49 47.97 5.60 20.37 15.77 0.92 0.90  444 a 445 Percent of respondents in a panchayat who report owning additional 446 land outside of the homegarden area. 447 Commercial crops such as areca and rubber were grown in 448 most homegardens, though not all farmers were actively 449 engaged in selling their crops (Figure 2a). In particular, 450 coffee, cardamom, and pepper each had market 451 engagement rates of just over 50%. Many commercial crops 452 were highly geographically concentrated. Cardamom, for 453 example, which was grown by only 20% of overall farmers in 454 22  our sample, was grown by 93% of farmers in Kattappana, a 455 hilly panchayat in Kerala’s uplands. The most extensively 456 planted food crops were coconut (present in 97% of 457 homegardens), banana (94%), jackfruit (88%), and mango 458 (87%; Figure 2b). Despite widespread on-farm production, 459 over 99% of households needed to purchase food to meet 460 domestic needs (Figure 2b). Fewer than 25% of farmers 461 were engaged in the cultivation of rice, and even those who 462 did grow rice explained that it was more sensible to sell 463 rough rice for processing and use the profits to purchase rice 464 from the store than it was for them to process the rice 465 themselves. Of the 115 homegardens surveyed, only one 466 claimed to be self-sufficient in the production of food. 467  468 Fig. 2. Major commercial (A) and food (B) crops grown, bought, 469 and sold by homegardeners. 470 23   471 3.2 Land-cover and land-use changes 472 LUCC results from the three methodological approaches 473 employed in this study were at times complementary and, at 474 other times, seemingly contradictory. Remote sensing 475 analysis, quantitative surveys, and semi-structured 476 interviews consistently showed a general decline in 477 agriculture across Kerala. Each of these methodological 478 approaches further suggested that the most prominent 479 change was a widespread increase in built surfaces in each 480 panchayat, accompanied by loss of wetland in all rice-481 producing regions (Figures 3,4).  482  483 Fig. 3. Land-cover changes (in percentage gain or loss between 484 images for each class) from remote sensing analysis for Avinissery 485 (2001–2012), Kalikavu (2003–2012), and Poothrikka (2002–2012). 486 24   487 Fig. 4. Percent of homegardeners producing more (white), less 488 (black) or the same (gray) amount of common crops in 2013 as 489 compared with 2003 (n = 92). We removed respondents who did 490 not grow a given crop in 2003 (when reporting declines) and 2013 491 (when reporting increases), as well as respondents whose property 492 area changed during the 10-year period of investigation. 493 Source: quantitative surveys. 494  495 Remote sensing results suggested that total net agroforest 496 area remained constant, while quantitative survey results 497 and interviews seemingly contradicted these results by 498 pointing to a decline in the cultivation of agroforestry crops 499 (Figures 3, 4). Of the 15 crops we surveyed, 12 decreased in 500 25  production over a period of 10 years, while the remaining 501 three remained constant, but did not increase (Figure 4). 502 Rice, pepper, and cashew were the most heavily affected, 503 with over 75% of respondents reducing production or 504 abandoning cultivation altogether. The least changed crops 505 were rubber, curry tree, and coffee, though none of these 506 exhibited overall increases in production. Livestock 507 ownership also declined, with the average number of 508 chickens per homegarden dropping from 12.5 in 2003 to 2.6 509 into 2013 and the average number of cows from 1.7 to 0.8 510 (Figure 5).  511  512 Fig. 5. Average number of cows (A) and chickens (B) per 513 homegarden for each panchayat in 2003 (black) and 2013 (white). 514 Paired sign tests indicate that overall trends are significant for both 515 chickens (s = 4, p < 0.0001) and cows (s = 8, p < 0.0001). 516 Panchayat names are abbreviated to the first three letters. 517  518 26  Net agroforestry land area did not decrease as agroforests 519 were in a state of dynamic equilibrium with wetlands (which 520 were decreasing) and built surfaces (which were increasing; 521 Table 3; Figure 6). Extensive net increases in built surfaces 522 across all study sites came primarily at the expense of 523 agroforests (70%) and bare ground (20%), but not from 524 wetlands. Built surfaces were the land-cover class that saw 525 the most growth; only half of those mapped in 2012 were 526 present a decade prior. A remarkable 31% of farmers 527 surveyed had built a new house on their property between 528 2003 and 2012, and in almost every case the new 529 construction came at the expense of agroforest. In 530 interviews, farmers from all panchayats supported this result 531 by making reference to increased rural construction 532 threatening agroforests. The absence of new construction on 533 wetlands in our land-cover analysis was also supported by 534 our interviews; farmers explained that it was not advisable to 535 build on wetlands due to state laws prohibiting new 536 development in these areas.  537 Table 3. Change detection analyses for Avinissery, Poothrikka, and 538 Kalikavu panchayats of Kerala state, India. For each land-cover 539 pair, the top numbers are transition probability matrices that list the 540 percent chance that a pixel of one land-cover class (rows) will 541 change to a pixel of another class (columns) between 2002 and 542 2012 (e.g. in Avinissery, 5.6% of the pixels on the landscape 543 changed from agroforest in 2002 to built in 2012). Note that the 544 27  top numbers represent the percent of all transitions that occurred 545 on the landscape (i.e. matrices sum to 100%). The numbers in 546 brackets list the percent contribution of 2002 land covers to new 547 land covers of 2012. For example, 66% of new built surfaces in 548 Avinissery came from agroforest. The bottom number is the total 549 area in square metres that transitioned from one land cover to 550 another (e.g. in Avinissery, 22 207 m2 of land transitioned from 551 agroforest to built). 552  Avinissery  2012   A  Agroforest Bare Built Water Wetland  Agroforest 40.6 3 (43) 5.6 (66) 0 (0) 0.2 (100)   143416 12063 22207 81 917  Bare 4.2 (55) 2.5 1.9 (22) 0.1 (100) 0 (0)   16748 6814 7709 323 0 2002 Built 1.9 (25) 0.2 (3) 5.4 0 (0) 0 (0)   7558 804 18848 1 0  Water 0.1 (1) 0 (0) 0 (0) 0.5 0 (0)   386 0 36 1842 0  Wetland 1.5 (19) 3.7 (54) 1 (12) 0 (0) 27.5   5977 14535 3979 0 97139  Poothrikka  2012   B  Agroforest Bare Built Water Wetland  Agroforest 64.6 2.6 (86) 2.1 (75) 0 (0) 0.3 (100)   229855 9414 7570 0 1150  Bare 9.5 (73) 0.9 0.7 (25) 0 (0) 0 (0)   33712 3320 2395 0 0 2002 Built 1.1 (8) 0.2 (7) 3.9 0 (0) 0 (0)   4068 542 13735 0 0  Water 0 (0) 0 (0) 0 (0) 0.2 0 (0)   0 0 0 632 2  Wetland 2.5 (19) 0.2 (7) 0 (0) 0 (0) 11.2   8872 808 0 0 39785  Kalikavu   2012   C  Agroforest Bare Built Water Wetland  Agroforest 83.4 3.3 (100) 2.7 (90) 0.3 (100) NA   300686 10527 8817 808 0  Bare 3 (91) 1.2 0.3 (10) 0 (0) NA   9762 5639 1047 0 0 2002 Built 0.3 (9) 0 (0) 4.5 0 (0) NA   906 146 16082 0 0  Water 0 (0) 0 (0) 0 (0) 0.9 NA   120 0 41 3071 0  Wetland NA NA NA NA NA 28    0 0 0 0 0  Mean   2012   D  Agroforest Bare Built Water Wetland  Agroforest 62.9 3 (59) 3.5 (70) 0.1 (100) 0.3 (100)   224652 10668 12865 296 689  Bare 5.6 (64) 1.5 1 (20) 0 (0) 0 (0)   20074 5258 3717 108 0 2002 Built 1.1 (13) 0.1 (2) 4.6 0 (0) 0 (0)   4177 497 16222 0 0  Water 0 (0) 0 (0) 0 (0) 0.5 0 (0)   169 0 26 1848 1  Wetland 2 (23) 2 (39) 0.5 (10) 0 (0) 19.4   4950 5114 1326 0 45641  553  554 Fig. 6. Primary land-cover changes based on aerial imagery as a 555 percentage of total sampled area for Avinissery (A; 2001–2012), 556 Poothrikka (B; 2002–2012), and Kalikavu (C; 2003–2012). Arrow 557 weight represents magnitude of flow, and only changes exceeding 558 0.5% are reported. 559  560 Of the panchayats we investigated with remote sensing, 561 Avinissery experienced the most dramatic net increase in 562 29  built area (94%), and Poothrikka the lowest at only 29% 563 (Figure 3). Despite such massive development, net 564 agroforest land-cover area in Avinissery and Kalikavu did not 565 change, and Poothrikka even saw growth of 11%, which 566 coincided with net losses of wetland and bare ground (which 567 we suspect was, at that time, recently cleared agroforest). 568 While newly constructed areas in Avinissery and Poothrikka 569 came primarily from gross losses in agroforest (66% and 570 75%, respectively), these losses were mitigated by 571 encroachment of agroforests on wetlands and bare ground 572 (Table 3). Kalikavu, however, which is located in the 573 Western Ghats and does not have wetlands, witnessed a 574 52% increase in built land cover, 90% of which was at the 575 expense of agroforest (Table 3). Yet agroforests remained 576 mostly unaffected, due to very low initial levels of built area 577 in Kalikavu, which is relatively remote, combined with 578 extensive agroforests that cover nearly 90% of the 579 panchayat (Figure 3). 580 Widespread losses of wetland area in rice-growing regions 581 were almost entirely replaced by local gross increases in 582 agroforests and bare ground (Table 3; Figure 6). Among 583 panchayats, wetlands saw the least growth, with 97% of 584 2012 wetlands unchanged from 2002 (Table 3). Despite 585 30  dramatic increases in built surfaces across study regions, 586 little construction occurred on wetlands (10%; Table 3). 587 Change detection findings were consistent with interview 588 outcomes, in which respondents commented on how farmers 589 would plant trees on the periphery of wetland areas, thereby 590 gradually converting wetlands to agroforests (this process is 591 described in Guillerme et al., 2011). 592 Semi-structured interviews provided some insight to land-593 use changes, and in doing so confirmed the quantitative 594 survey results. All but two farmers indicated that they had 595 observed a decline in agriculture on their property over the 596 same time period, as well as for the panchayat in general. 597 Most narratives of regional agriculture were consistent with 598 each other, pointing towards a widespread decline in both 599 garden-based and commercial farming, alongside an 600 increase in buildings and roads. 601 3.3 Drivers of change 602 Farmers provided numerous explanations for the decline of 603 agriculture in Kerala. The most common themes, from most 604 to least mentioned, were (percentage of respondents in 605 parentheses): 1) changing weather and climate (87%), 2) 606 decreased access to labour (80%), 3) declining profit 607 31  margins (60%), 4) poor access to pesticides and fertilizers 608 (43%), 5) increased problems with pests and disease (43%), 609 6) a stigma against agriculture (23%), and 7) construction 610 competing for land use (17%). Using these primary themes, 611 we present a conceptual organization of land use changes 612 and drivers in Figure 7. In considering the themes together, 613 one of the most widespread concerns was the increased 614 financial risk associated with a combination of high 615 investment costs alongside shrinking profit margins. Input 616 costs such as labour, pesticides and fertilizers, which were 617 deemed necessary by most farmers, have soared in recent 618 years (Thomas and Devi, 2016). 619  620 Fig. 7. Conceptual organization of observed land-use changes and 621 their drivers according to quantitative household surveys and semi-622 structured interviews. Arrows indicate direction of influence and 623 black lines suggest a causal relationship between lower and higher-624 order drivers. 625 32  Sociological factors and government policy were also highly 626 cited drivers of agricultural decline. Many of the farmers 627 interviewed expressed uncertainty over the fate of their farm 628 should they grow too old to work. In most cases their 629 children were educated with university degrees and were 630 pursuing work in other sectors, often only finding such work 631 abroad. Several farmers considered Kerala’s youth to be 632 over-educated, resulting in a shortage of white-collar jobs 633 and high levels of unemployment despite a serious shortage 634 in agricultural labour. The government was often criticized 635 for its inability to rectify Kerala’s agricultural decline by 636 helping farmers in meaningful ways. Specific criticisms 637 included: 1) the implementation of welfare strategies, such 638 as the Mahatma Gandhi National Rural Employment 639 Guarantee Program, which may act as a disincentive for 640 labourers to work on farms (Gulati et al., 2014; Harish et al., 641 2011), 2) not providing subsidies for expensive inputs and 642 technologies, and 3) not providing sufficient infrastructure to 643 connect farmers with regional and global markets. 644 Though it was mentioned less frequently than other drivers, 645 the increase in rural construction was identified by several 646 farmers as an important factor in the decline of agriculture. 647 Building new houses decreases agricultural production 648 33  locally because the new properties are typically erected on 649 what was agroforest. At the landscape scale, fragmentation 650 of properties as they are passed from one generation to the 651 next eventually results in agricultural holdings so small that 652 they cannot be farmed profitably. As farms become less 653 profitable, there is an increase in the relative value of real 654 estate as a land use compared to agroforest. The following 655 quote from a farmer in Kattappana highlights these 656 conflicting land uses (translated from Malayalam): 657 “The cost of labour is increasing, and the market price 658 for cardamom is decreasing. This year, the cost of 659 fertilizer and pesticide is double what it was last year. 660 Cardamom is like a child that requires too much care, 661 but no other crop is profitable. […] We are planning to 662 build a lot of houses on this land for renting instead of 663 agriculture. We want to build a new house every year, 664 as they are a stable source of income.” 665 4. Discussion 666 4.1 Land-use and cover changes 667 A combination of remote sensing analysis, quantitative 668 surveys, and semi-structured interviews provided a 669 complementary means to paint a clear picture of land-use 670 34  and land-cover changes in Kerala between early 2000 and 671 2013. At the landscape scale, the remote sensing analysis 672 showed a pronounced increase in built area that coincided 673 with a decrease in wetlands. However, these coincided with 674 only negligible net changes in agroforestry land cover. But 675 below the canopy, at the household scale, we discovered 676 using quantitative surveys that all forms of agriculture 677 (wetland, garden, and commercial) are experiencing decline. 678 Semi-structured interviews provided narratives supporting 679 and enriching land-use and land-cover changes measured 680 using quantitative methods. The interviews provided 681 additional insights into the environmental, economic, social, 682 and political forces driving these changes. 683 If agroforests aren’t decreasing in areal extent, why are 95% 684 of farmers reporting a decline in homegarden crops? First, 685 new agroforests have mainly been consigned to infilled 686 wetlands, where fewer tree crops are able to grow 687 successfully. While banana, coconut, and areca can grow 688 with limited success in wetter soils, most other dominant 689 agroforestry species have trouble establishing due to poor 690 drainage and increased prevalence of flooding. These 691 conditions typically result in highly simplified post-wetland 692 agroforests (e.g. a mix of coconut with arecanut palms and 693 35  banana in the understory). Second, and contributing to the 694 same problem, wetlands are often further from homesteads 695 making intensive management less convenient. Farmers are 696 therefore less likely to grow medicinal, ornamental, or even 697 food crops intended for household use due to ease of 698 access and additional labour requirements. Finally, and 699 perhaps most importantly, mature agroforests often have 700 better developed cultivated species composition than newly-701 planted agroforests, partly because slow-growing trees have 702 had time to establish, but also because older agroforests are 703 the product of generations of stewardship of culturally 704 important agricultural resources. 705 Are increases in building and road construction also driving 706 agricultural decline? At first glance, our remote sensing and 707 quantitative survey results appear to suggest that new 708 construction of built surfaces is driving a decline in 709 agroforestry, which is in turn expanding into, and thus driving 710 the decline of, less valuable wetlands (Raj and Azeez, 711 2009). While there is likely some validity to this 712 interpretation, semi-structured interviews and quantitative 713 surveys identified numerous other drivers of land-use 714 change (e.g. declining profitability, high levels of risk, etc.) 715 that convincingly supplant real estate and construction as 716 36  the primary causal agents of agricultural decline. 717 Furthermore, despite a surprising number of farmers (~30%) 718 reporting new construction on their property, far more (over 719 95%) reported a decline in agriculture. 720 We therefore propose that the high rates of building and 721 road construction reported in our study are not a driver of, 722 but rather an outcome of agricultural decline. Profitability, 723 and thus cultivation of paddy has dropped sufficiently and for 724 long enough in Kerala that farmers have allowed agroforests 725 to slowly encroach upon wetlands (Guillerme et al., 2011). 726 Agroforests, subject to the same economic and 727 environmental challenges as paddy (e.g. access to inputs, 728 climate, pests, etc.), albeit to a lesser extent, have become 729 attractive and affordable real estate for a population with 730 foreign money and an appetite for large, exurban houses. 731 Agroforests, while expanding into wetlands, are most 732 susceptible to new construction, given that houses cannot be 733 built on wetlands due to legal and logistical constraints. 734 According to this explanation, which is more in line with 735 farmers’ perceptions, increased rural construction is not the 736 driver of agricultural decline, but rather an unfortunate 737 byproduct of a decade or more of unprofitable and unreliable 738 agriculture. This interpretation is also in line with research by 739 37  (Lambin et al., 2001), which suggests that drivers of land-740 use and land-cover change are often primarily and 741 fundamentally linked to economic opportunities available at 742 the local level. 743 Some panchayat-specific results require brief consideration. 744 First, the unexpected 11% increase in agroforest measured 745 in Poothrikka is almost certainly related to the anomalous 746 62% decline in bare ground. We suspect that Poothrikka’s 747 agroforests experienced a surge in clearing just prior to the 748 acquisition of the IKONOS-2 images used for this study in 749 April 2002. This surge could have happened for various 750 reasons, such as felling trees for timber, re-planting old and 751 depleted rubber and/or coconut plantations, or clearing for 752 construction. Therefore, the measured increase in agroforest 753 is more likely a regrowth event rather than agroforest 754 expansion. 755 A second point of consideration is the state of land-use 756 change in Kalikavu, as well as other upland regions of 757 Kerala that have little to no wetland. The 52% increase in 758 built surfaces measured in Kalikavu did not have a 759 considerable effect on absolute agroforest area, despite 760 arising almost exclusively at their expense, because the 761 initial proportion of built surfaces was relatively low (only 5%, 762 38  compared to 15% in Avinissery). However, if Kalikavu’s 763 current rate of construction continues into the future, there 764 will be a disproportionate effect on agroforests. This is 765 because wetlands, which are absent in Kerala’s uplands, are 766 unable to mitigate agroforestry losses as they do in other 767 parts of the state. This consideration will be important for 768 land-use management in the future, as the effect of building 769 and road development on agriculture in rural Kerala is a 770 product of both the presence of wetland and the relative 771 areal extent of remaining agroforest. 772 In recent years, homegarden researchers have raised the 773 concern that plantation-style agriculture in Kerala is 774 replacing more species-rich and culturally important 775 homegarden-style agriculture (Kumar, 2005; Peyre et al., 776 2006). While our results cannot provide a conclusive or 777 detailed answer to this question, we can offer some insight. 778 While Kerala’s homegardens have modernized considerably 779 over past decades (Peyre et al., 2006), our results fail to 780 support the hypothesis that they are being replaced by 781 plantation agriculture. None of the commercial or plantation 782 crops investigated in our study increased in production over 783 the ten year study period, which would be expected if non-784 plantation crops were being replaced. Farmers were 785 39  undivided in their accounts of agricultural decline in general, 786 whether with regards to homegarden, plantation, or wetland 787 agriculture. As previously mentioned, the increase in 788 construction, and in turn the number of homegardens, 789 means a decrease in mean property size, and therefore a 790 larger number of smaller homegardens. Smaller properties 791 are less likely to have plantations, partly because new 792 residents are less likely to be farmers, and partly due to 793 spatial constraints (i.e. plantations benefit from economies of 794 scale and are not profitable when too small). Furthermore, 795 the growing number of households may even enhance 796 regional crop species diversity, given that smaller 797 homegardens in Kerala and Sri Lanka have been shown to 798 exhibit higher cultivated species richness (Kumar, 2011; 799 Mattsson et al., 2014).  800 Although it is unlikely that new construction between 2001 801 and 2012 was a cause of the concurrent agricultural decline, 802 increased built-area development may have other 803 implications for agriculture in Kerala. Along with such 804 development comes an increase in the rural population, 805 which could contribute in other ways to the agricultural 806 decline, most notably: 1) incoming landholders are less likely 807 to be farmers because they often come from cities or are 808 40  returning from working abroad, in which case they have 809 alternative sources of income; they are also less likely to 810 enter into agriculture as they are dissuaded by low returns 811 on investment; 2) newcomers inject capital into local 812 economies, bringing with them (both directly and indirectly) 813 employment opportunities for those no longer drawn to the 814 high risk and low returns of agriculture; 3) high post-815 secondary education attendance rates and narrowing social 816 boundaries between low- and middle-income classes have 817 produced a generation of young adults unwilling to take over 818 their parents’ farms, fostering a stigma against agriculture 819 and other labour-based livelihoods; 4) new buildings are 820 constructed on partitioned land, and after a certain number 821 of partitions it becomes impossible to take advantage of 822 economies of scale. In other words, a minimum amount of 823 land is required for an agricultural operation to be profitable, 824 and Kerala’s landscape is becoming increasingly 825 fragmented, consisting of smaller and smaller farms. 826 However, these potential impacts are speculative, and 827 further research would be required to monitor and identify 828 the effects of rural population growth and development on 829 Kerala’s agricultural systems. 830 41  This increasing number of rural holdings might seem like a 831 worrying trend for conservationists, especially considering 832 Kerala’s status as a biodiversity hotspot. But compared to 833 other developing hotspots, such as Southeast Asia and the 834 Amazon, where agricultural development of oil palm (in the 835 former), and pasture and soy (in the latter) have led to 836 massive deforestation (Barona et al., 2010; Koh and 837 Wilcove, 2008), Kerala’s agricultural model holds some 838 promise. Homegardens as a form of intensive, 839 environmentally friendly agriculture are a working example of 840 a system in which high population density, agriculture, and 841 conservation interests can coexist (Galluzzi et al., 2010). 842 However, more research is required to assess the yield and 843 biodiversity potential of this wildlife-friendly farming model 844 (see Green et al., 2005). 845 4.2 Mixed methods for LUCC research 846 Taken together, the mixed methods employed in this study 847 worked in a complementary fashion to illustrate that 848 homegardens, the most common and widespread 849 agroforestry system in Kerala, may be declining in 850 agricultural importance, though not necessarily in extent, 851 numbers, or cultural importance. Each of the three methods 852 that we employed demonstrated distinct advantages and 853 42  limitations that, when considered together, paint a more 854 complete picture of LUCC. Remote sensing, which provides 855 a valuable means of consistently and relatively objectively 856 inferring large-scale changes in land cover over the historical 857 record, fails to capture more nuanced land-use changes and 858 is unable to probe the intangible experiences and knowledge 859 of those inhabiting the landscape. Semi-structured 860 interviews, which lack the relative objectivity of remote 861 sensing or quantitative surveys, introduce the nuance and 862 complexity of human experience, and allow for a rich 863 understanding not only of the causes of LUCC, but also of 864 the implications. Quantitative surveys can help to bridge the 865 gap between remote sensing and semi-structured interviews 866 by providing further evidence that helps to validate the links 867 researchers draw between above-canopy images and 868 below-canopy narratives. While quantitative surveys focus 869 on the human perspective, they do so by collecting data that, 870 when generalized from sample to population, may reveal 871 trends not evident to the subjects. 872 In this study, any conclusions that would have been drawn 873 from any individual method would have been deeply flawed. 874 While remote sensing analysis suggested that agroforests 875 were relatively healthy, sub-canopy investigations revealed 876 43  declining agroforestry despite constant areal extent. Neither 877 remote sensing nor quantitative surveys could have provided 878 the insight relating to the multiple LUCC drivers revealed in 879 the semi-structured interviews. Therefore, based on our 880 research, and in support of the ideas proposed previously by 881 Turner et al. (1994) and Lambin et al. (2003), we conclude 882 that land-use and land-cover change quantification and 883 identification of drivers in tropical regions is highly complex, 884 and warrants the adoption of a mixed-methods approach. 885 Given the inherent complexity of LUCC research, failure to 886 employ mixed methods at multiples scales could conceivably 887 lead to incomplete information and therefore inappropriate or 888 counterproductive policy outcomes.  889 5. Conclusions 890 We examined agricultural land use changes in Kerala using 891 a combination of remote-sensing, quantitative surveys and 892 semi-structured interviews. We found little support for the 893 hypothesis that plantations are replacing homegardens. On 894 the contrary, we found that homegarden extent is remarkably 895 stable. However, we documented a general decline in both 896 the production and importance of homegarden crops for the 897 average Kerala household, driven by changing 898 socioeconomic circumstances. Land use policy in Kerala 899 44  should address the broader shifts in the socioeconomic 900 landscape, rather than the physical landscape. Our study 901 demonstrates the value of a mixed-methods approach for 902 developing a richer understanding of land use changes. 903 Acknowledgements: We would like to thank our field 904 assistants, Nijin BS, Niyas Palakkal, and Haseena Kadiri, the 905 farmers who engaged with us in our research, and the 906 welcoming faculty at Kerala Agricultural University in 907 Thrissur. We would also like to thank Carlo Soto for his 908 geospatial skills and all of the graduate students and staff 909 from the Rhemtulla and Ramankutty labs at McGill University 910 in 2013-14. We would also like to thank the McGill Ethics 911 Board Office - Research Ethics Board I for granting us a 912 Certificate of Ethical Acceptability of Research Involving 913 Humans.  914 Funding: This work was supported by the International 915 Development Research Centre of Canada through the John 916 G. Bene Fellowship entitled “Trees and People”, awarded in 917 2013 [IDRC Doctoral Research Award #107473-99907000-918 006], an NSERC Julie Payette Award to TAF, an NSERC 919 Andre Hamer award to TAF, and a James Lougheed Award 920 to TAF. 921 45   922  923 Works cited 924 Achard, F., Stibig, H.-J., Beuchle, R., Lindquist, E., D’Annunzio, R., 2012. 925 Use of a systematic statistical sample with moderate-resolution 926 imagery to assess forest cover changes at tropical to global 927 scale. Glob. For. Monit. Earth Obs. 125. 928 Altieri, M.A., 2009. Agroecology, small farms, and food sovereignty. 929 Mon. Rev. 61, 102. 930 Barona, E., Ramankutty, N., Hyman, G., Coomes, O.T., 2010. The role of 931 pasture and soybean in deforestation of the Brazilian Amazon. 932 Environ. Res. Lett. 5, 024002. doi:10.1088/1748-933 9326/5/2/024002 934 Bellhouse, D.R., 1977. Some optimal designs for sampling in two 935 dimensions. Biometrika 64, 605–611. 936 Benton, T.G., 2007. Managing farming’s footprint on biodiversity. 937 Science 315, 341. 938 DeFries, R., Hansen, A., Turner, B.L., Reid, R., Liu, J., 2007. Land use 939 change around protected areas: management to balance 940 human needs and ecological function. Ecol. Appl. 17, 1031–941 1038. 942 DeFries, R., Rosenzweig, C., 2010. Toward a whole-landscape approach 943 for sustainable land use in the tropics. Proc. Natl. Acad. Sci. 107, 944 19627–19632. 945 DeFries, R.S., Foley, J.A., Asner, G.P., 2004. Land-use choices: Balancing 946 human needs and ecosystem function. Front. Ecol. Environ. 2, 947 249–257. 948 ENVI, 2013. . Exelis Visual Information Solutions, Boulder, Colorado. 949 ESRI, R., 2011. ArcGIS desktop: release 10. Environ. Syst. Res. Inst. CA. 950 Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, 951 S.R., Chapin, F.S., Coe, M.T., Daily, G.C., Gibbs, H.K., others, 952 2005. Global consequences of land use. science 309, 570–574. 953 Fretwell, P.T., LaRue, M.A., Morin, P., Kooyman, G.L., Wienecke, B., 954 Ratcliffe, N., Fox, A.J., Fleming, A.H., Porter, C., Trathan, P.N., 955 2012. An emperor penguin population estimate: the first global, 956 synoptic survey of a species from space. PLoS One 7, e33751. 957 Galluzzi, G., Eyzaguirre, P., Negri, V., 2010. Home gardens: neglected 958 hotspots of agro-biodiversity and cultural diversity. Biodivers. 959 Conserv. 19, 3635–3654. doi:10.1007/s10531-010-9919-5 960 Gibbs, H.K., Ruesch, A.S., Achard, F., Clayton, M.K., Holmgren, P., 961 Ramankutty, N., Foley, J.A., 2010. Tropical forests were the 962 46  primary sources of new agricultural land in the 1980s and 963 1990s. Proc. Natl. Acad. Sci. 107, 16732–16737. 964 Goetz, S.J., Wright, R.K., Smith, A.J., Zinecker, E., Schaub, E., 2003. 965 IKONOS imagery for resource management: Tree cover, 966 impervious surfaces, and riparian buffer analyses in the mid-967 Atlantic region. Remote Sens. Environ. 88, 195–208. 968 Green, R.E., Cornell, S.J., Scharlemann, J.P., Balmford, A., 2005. Farming 969 and the fate of wild nature. Science 307, 550–555. 970 Guillerme, S., Kumar, B.M., Menon, A., Hinnewinkel, C., Maire, É., 971 Santhoshkumar, A.V., 2011. Impacts of public policies and 972 farmer preferences on agroforestry practices in Kerala, India. 973 Environ. Manage. 48, 351–364. 974 Gulati, A., Jain, S., Satija, N., 2014. Rising Farm Wages in India—The 975 “Pull”and “Push”Factors. J. Land Rural Stud. 2, 261–286. 976 Gutierrez, M., Johnson, E., 2012. Utility of GeoEye in assessing riparian 977 vegetation along the RioConchos, Mexico, in: Geological Society 978 of America Abstracts with Programs,. p. 15. 979 Harish, B.G., Nagaraj, N., Chandrakanth, M.G., Murthy, P.S., Chengappa, 980 P.G., Basavaraj, G., others, 2011. Impacts and implications of 981 MGNREGA on labour supply and income generation for 982 agriculture in central dry zone of Karnataka. Agric. Econ. Res. 983 Rev. 24, 485–494. 984 Huang, C., Kim, S., Song, K., Townshend, J.R., Davis, P., Altstatt, A., 985 Rodas, O., Yanosky, A., Clay, R., Tucker, C.J., others, 2009. 986 Assessment of Paraguay’s forest cover change using Landsat 987 observations. Glob. Planet. Change 67, 1–12. 988 Huang, R., 2012. RQDA: R-based qualitative data analysis. R Package 989 Version 02-3 2. 990 India, G. of, 2011. Census of India. 991 Jeffrey, R., 2001. Politics, women and well-being: How Kerala became “a 992 model.” Oxford University Press. 993 Kannan, K.P., 1998. Political economy of labour and development in 994 Kerala. Econ. Polit. Wkly. L61–L70. 995 Kannan, K.P., Pushpangadan, K., 1990. Dissecting agricultural stagnation 996 in Kerala: an analysis across crops, seasons and regions. Econ. 997 Polit. Wkly. 1991–2004. 998 Koh, L.P., Wilcove, D.S., 2008. Is oil palm agriculture really destroying 999 tropical biodiversity? Conserv. Lett. 1, 60–64. 1000 Kumar, B.M., 2011. Species richness and aboveground carbon stocks in 1001 the homegardens of central Kerala, India. Agric. Ecosyst. 1002 Environ. 140, 430–440. 1003 Kumar, B.M., 2005. Land use in Kerala: changing scenarios and shifting 1004 paradigms. J. Trop. Agric. 42, 1–12. 1005 Kumar, B.M., George, S.J., Chinnamani, S., 1994. Diversity, structure and 1006 standing stock of wood in the homegardens of Kerala in 1007 peninsular India. Agrofor. Syst. 25, 243–262. 1008 Kumar, B.M., Nair, P.R., 2004. The enigma of tropical homegardens, in: 1009 New Vistas in Agroforestry. Springer, pp. 135–152. 1010 47  Lambin, E.F., Geist, H.J., Lepers, E., 2003. Dynamics of land-use and 1011 land-cover change in tropical regions. Annu. Rev. Environ. 1012 Resour. 28, 205–241. 1013 Lambin, E.F., Turner, B.L., Geist, H.J., Agbola, S.B., Angelsen, A., Bruce, 1014 J.W., Coomes, O.T., Dirzo, R., Fischer, G., Folke, C., others, 2001. 1015 The causes of land-use and land-cover change: moving beyond 1016 the myths. Glob. Environ. Change 11, 261–269. 1017 Lillesand, T., Kiefer, R.W., Chipman, J., 2014. Remote sensing and image 1018 interpretation. John Wiley & Sons. 1019 Lu, D., Weng, Q., 2007. A survey of image classification methods and 1020 techniques for improving classification performance. Int. J. 1021 Remote Sens. 28, 823–870. 1022 Luyssaert, S., Hessenmöller, D., Von Lüpke, N., Kaiser, S., Schulze, E.D., 1023 2011. Quantifying land use and disturbance intensity in forestry, 1024 based on the self-thinning relationship. Ecol. Appl. 21, 3272–1025 3284. 1026 Mattsson, E., Ostwald, M., Nissanka, S.P., Pushpakumara, D.K.N.G., 1027 2014. Quantification of carbon stock and tree diversity of 1028 homegardens in a dry zone area of Moneragala district, Sri 1029 Lanka. Agrofor. Syst. 89, 435–445. doi:10.1007/s10457-014-1030 9780-8 1031 Mumby, P.J., Edwards, A.J., 2002. Mapping marine environments with 1032 IKONOS imagery: enhanced spatial resolution can deliver 1033 greater thematic accuracy. Remote Sens. Environ. 82, 248–257. 1034 Munsi, M., Malaviya, S., Oinam, G., Joshi, P.K., 2010. A landscape 1035 approach for quantifying land-use and land-cover change 1036 (1976–2006) in middle Himalaya. Reg. Environ. Change 10, 145–1037 155. 1038 Nair, P.K., Mohan Kumar, B., Nair, V.D., 2009. Agroforestry as a strategy 1039 for carbon sequestration. J. Plant Nutr. Soil Sci. 172, 10–23. 1040 Narayanan, N.C., 2006. For and Against Grain Land Use Politics of Rice in 1041 Kerala, India. Int. J. Rural Manag. 2, 123–144. 1042 Peyre, A., Guidal, A., Wiersum, K.F., Bongers, F., 2006. Dynamics of 1043 homegarden structure and function in Kerala, India. Agrofor. 1044 Syst. 66, 101–115. 1045 Potapov, P.V., Dempewolf, J., Talero, Y., Hansen, M.C., Stehman, S.V., 1046 Vargas, C., Rojas, E.J., Castillo, D., Mendoza, E., Calderón, A., 1047 others, 2014. National satellite-based humid tropical forest 1048 change assessment in Peru in support of REDD+ 1049 implementation. Environ. Res. Lett. 9, 124012. 1050 Prakash, B.A., 1998. Gulf migration and its economic impact: the Kerala 1051 experience. Econ. Polit. Wkly. 3209–3213. 1052 Raj, P.N., Azeez, P.A., 2009. Real estate and agricultural wetlands in 1053 Kerala. Econ. Polit. Wkly. 63–66. 1054 Ramdani, F., Hino, M., 2013. Land use changes and GHG emissions from 1055 tropical forest conversion by oil palm plantations in Riau 1056 Province, Indonesia. PloS One 8, e70323. 1057 48  Shimabukuro, Y.E., Beuchle, R., Grecchi, R.C., Simonetti, D., Achard, F., 1058 2014. Assessment of burned areas in Mato Grosso State, Brazil, 1059 from a systematic sample of medium resolution satellite 1060 imagery, in: Geoscience and Remote Sensing Symposium 1061 (IGARSS), 2014 IEEE International. IEEE, pp. 4257–4259. 1062 Singh, A., 1989. Review article digital change detection techniques using 1063 remotely-sensed data. Int. J. Remote Sens. 10, 989–1003. 1064 Thomas, J.M., Devi, P.I., 2016. Farm Size, Farm Income and Efficiency: 1065 The Case of Commercial Pineapple Farming in Kerala. Indian J. 1066 Agric. Econ. 71, 352. 1067 Tomscha, S.A., Sutherland, I.J., Renard, D., Gergel, S.E., Rhemtulla, J.M., 1068 Bennett, E.M., Daniels, L.D., Eddy, I.M., Clark, E.E., 2016. A guide 1069 to historical data sets for reconstructing ecosystem service 1070 change over time. BioScience 66, 747–762. 1071 Turner, B.L., Meyer, W.B., Skole, D.L., others, 1994. Global land-1072 use/land-cover change: towards an integrated study. Ambio 1073 Stockh. 23, 91–95. 1074 UNDP, 2015. Human Development Report. 1075 Urban, D.L., Wallin, D.O., 2002. Introduction to Markov models, in: 1076 Learning Landscape Ecology. Springer, pp. 35–48. 1077 Veldkamp, A., Verburg, P.H., 2004. Modelling land use change and 1078 environmental impact. J. Environ. Manage. 72, 1–3. 1079 Wickham, J.D., Stehman, S.V., Gass, L., Dewitz, J., Fry, J.A., Wade, T.G., 1080 2013. Accuracy assessment of NLCD 2006 land cover and 1081 impervious surface. Remote Sens. Environ. 130, 294–304. 1082 Wrigley, G., 1971. Tropical agriculture: The development of production. 1083 London: Faber and Faber. 1084 Zachariah, K.C., Rajan, S.I., 2012. Inflexion in Kerala’s Gulf Connection: 1085 Report on Kerala Migration Survey 2011. 1086  1087 Supplementary material Preliminary pixel-based classification yielded insufficient accuracy for the purpose of this study. While supervised classification has been successfully used for land-cover analysis in numerous contexts (e.g. Fretwell et al., 2012; Goetz et al., 2003; Gutierrez and Johnson, 2012; Mumby and Edwards, 2002), Kerala’s rural landscapes are a complex mosaic of wetlands, tree plantations, and agroforests with large spectral variability. Furthermore, built surfaces such as houses and roads are often obscured by overhanging trees or tree shadows, which led to an overestimation of tree cover and underestimation of built surfaces. We next attempted object-oriented classification, but encountered the same issues with overhanging trees and shadows. We therefore opted for a manual classification approach (although we used supervised classification to guide our analysis as described in the next paragraph). Overhanging trees and shadows were clearly visible to the naked eye, but mischaracterized by both pixel-based and object-based classification. Manually digitizing land-cover polygons by hand is often more accurate, as shape, texture, and context can be employed, in addition to spectral characteristics (Lillesand et al., 2014; Lu and Weng, 2007). Manual classification, or a combination of manual and object-based image classification (OBIA), has been the preferred choice for classifying complex tropical landscapes (Gibbs et al., 2010; Ramdani and Hino, 2013). The major shortfall of manual classification is the necessary time investment. Because classifying all 6 images in their entirety was too time consuming (Achard et al., 2012; Shimabukuro et al., 2014), we adopted a systematic unaligned sampling approach (Bellhouse, 1977). For each of the three panchayats, we divided image pairs into 8 equal-area sections, generated two random points for each section, and used these 96 points to generate square 0.75 ha sample areas in ArcMap. We conducted maximum likelihood supervised classifications using ENVI to ensure that image samples were representative of the overall image. Land-cover variability between samples and parent images ranged from 0.2 to 7.9%.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.52383.1-0378871/manifest

Comment

Related Items