1 Title: Diversification through agroforestry for small-scale farms on Malo Island, Vanuatu 2 Alternative title: How to take advantage of a new crop? The experience of Melanesian 3 smallholders 4 Authors: Laurène Feintrenie (1), Jean Ollivier (2), Frank Enjalric (2) 5 Authors’ affiliations: 6 (1) CIFOR, University of Montpellier III, IRD. Corresponding author. Address: CIFOR, 7 PO Box 0113 BOCBD, Bogor 16000, Indonesia. Phone: (62) 251 8622 622. Fax: (62) 8 251 8622 100. E-mail: [email protected] 9 (2) CIRAD, Montpellier, France. 10 1 10 Introduction 11 Malo Island (15° 36′ S, 167° 30′ E) is located at a cross point of trade exchanges between 12 the main islands of the Vanuatu archipelago in the Pacific. Coconut-based agroforestry 13 systems are one of the main components of small-scale farming on the island. These systems 14 associate coconuts (Cocos nucifera) with cocoa (Theobroma cacao) and several types of fruit 15 trees (Artocarpus altilis, Barringtonia procera and Citrus spp) or nut trees. The main food 16 crops are macabo (Xanthosoma sagittifolium), yam (the most common species is Dioscorea 17 nummularia), taro (Colocasia esculenta), island cabbage (Abelmoschus manihot) and sweet 18 potato (Ipomoea batatas). These agroforestry plantations combine home-garden traditions that 19 mix vegetables and root crops in a rotational system, including long forest fallows (Allen 20 2001), with smallholders’ cash crop plantations. Coconut and cocoa estate plantations were 21 introduced on the island at the beginning of the 19th century by European settlers 22 (Bonnemaison 1996). After Vanuatu gained independence in 1980, most of the estate 23 plantations returned to villagers’ ownership, and farmers began transforming them into mixed 24 tree systems. To increase labour and land productivity, people spontaneously associated 25 coconut and cocoa in the same plantations. As the results were convincing, the farmers began 26 planting coconut and cocoa in their new root crops swiddens every year. Coconut, which has 27 long been cultivated for home consumption, moved from staple food to cash crop in family 28 farming (Labouisse 2004). 29 In the 2000s, the choice in cash crops increased again, with the development of a local 30 market for vanilla and spices. How could smallholders best take advantage of this new 31 economic opportunity? Responses came quickly; local farmers adapted their production 32 systems without giving up their previous productions and while maintaining their food crop 33 cultivation. They tested different options for introducing vanilla to their farms, such as 34 cropping vanilla (Vanilla planifolia) under coconuts. 2 35 This study compares the economic results of the main associations of coconut, cocoa and 36 vanilla that were observed in Malo, and discusses the advantages of the different systems and 37 the choices made by smallholders. While in countries such as Indonesia or the Philippines, 38 agroforestry plantations are increasingly being replaced by monocultures, the people of 39 Vanuatu chose diversification of production rather than specialisation, at both the household 40 and the plot levels. Assuming that the economic results of the different types of plantation are 41 determinant in smallholders’ decisions, we used three economic indicators to compare the 42 plantations: return to land, return to labour and labour needs across a year. 43 44 Material and methods 45 A combination of surveys, field observations and measures was used to determine 46 farmers’ strategies, agricultural practices and decision-making processes, and to assess the 47 performances of agroforestry systems in terms of yields, labour needs and incomes. The data 48 collection was conducted in 2005 (Feintrenie 2006), during six months of field work, from 49 May to October). 50 Interviews were conducted of a sample of 30 households, to collect data on cropping 51 practices, inputs (such as fertilizers, pesticides and tools), labour needs, prices and yields. 52 This information was confirmed by a literature review of the most recent studies conducted on 53 Malo Island: Allen describes the local agrarian system (Allen 2001) and the traditional staple 54 food gardening system (Allen 2000); a detailed typology of the coconut-based agroforestry 55 systems was carried out by Lamanda et al. (2006). The spatial and temporal dynamics of a 56 family farm were also captured during these interviews; the history of the household and the 57 distribution of the plots used by the household across time were recorded, along with details 58 on the management of the plots and the tenure status. Farmers’ strategies and decisions 59 regarding the allocation of plots to one crop or another were discussed. 3 60 Coconut–cocoa and coconut–vanilla agroforestry systems were studied using the 61 synchronic approach, which permits the extrapolation of temporal dynamics from a set of 62 plots of different ages. This approach is relevant only if the spatial distribution of plots of 63 different ages in a relatively homogeneous environment at a given point in time can be 64 considered equivalent to the different stages of a plantation over time (Pickett 1991). Four 65 stages were selected to describe these agroforestry systems according to their spatial and 66 temporal components along the coconut life cycle. The definition of these stages was based on 67 the description of coconut-based agroforestry systems made by Lamanda (Lamanda et al. 68 2006). Measurements in the plots aimed to confirm the information available from previous 69 studies (Allen 2000; Lamanda 2006); therefore, a small sample of plots was considered 70 sufficient, as long as no contradictions with the literature or with farmers’ interviews were 71 observed. Three representative plots were chosen for each stage in the same morpho- 72 pedological unit, with constant agricultural practices over time. The sample was composed of 73 12 plots of coconut–cocoa plantations and 3 plots of coconut–vanilla. The floristic 74 composition of the plots was assessed to get a precise description of the cropping system 75 according to the age of the main perennial crop, the coconuts. Useful species were numbered 76 and their productions were measured. Interviews with farmers, using the same sample of 30 77 households, confirmed the data on plot floristic composition according to the age of coconuts, 78 and yields of the main species according to the age of the plants. 79 Modelling of economic results was done using the software Olympe (CIRAD et al. 2007; 80 Deheuvels and Penot 2007). Olympe is a decision-support software for agriculture. It 81 combines a database ‘ready to fill’ with economic information on prices, productions and 82 households with an accounting calculator which allows the automatic computation of 83 economic indicators. The software can also be used as a simulator to test a change in the 84 farming system or to evaluate a farm’s resilience to risks such as low harvest or price drop 4 85 (CIRAD et al. 2007). Yields, prices, costs of production and labour needs were processed in 86 the Olympe software and gross margins were calculated on the basis of a one-hectare plot, 87 with details of each perennial crop and food crop (Feintrenie 2006). Economic modelling 88 allowed the simulation of economic results over the coconut production cycle of 80 years. 89 First, a comparative analysis of coconut–cocoa and coconut–vanilla plantations using 90 economic indicators (return to land, return to labour and labour needs across a year), and 91 exposed over the coconut production cycle, revealed the advantages and constraints of the two 92 agroforestry systems. Second, the possibilities of combining the two cropping systems on a 93 family farm were simulated and evaluated using Olympe. 94 95 Results 96 Coconut-based agroforestry systems in family farms in Malo 97 Family farms in Malo have a median size of 5 to 10 ha of cultivated land and extensive 98 production systems. The main labour force is the family, with labour groups organised among 99 neighbours for some activities (such as the copra harvest). The farming system is based on a 100 combination of perennial cash crops and annual or pluri-annual food crops. Agroforestry 101 systems usually occupy half of the cultivated land area and are the first installed by a young 102 household (100% of respondents). Fig. 1 shows the dynamics of crops implementation in 103 space and time. Traditionally, farmers clear a plot of forest every year to install a garden, 104 which combines food crops and perennial cash crops; this garden is usually about 625 m² (or 105 1/16 ha). Perennial plantations thus increase at the same rate as forest clearance. A household 106 possesses a quite constant surface area of garden in its first, second and third years every year 107 until all the land is planted perennially. A minimal surface of forest reserve, often about one 108 hectare, is usually preserved for use for the home garden once intercropping of food crops in 109 plantations is no longer sufficient to meet the family’s needs. Allen (2001) describes the 5 110 traditional home-garden system as the rotation of root crop gardens and long-term bush 111 fallows (more than 15 years). One farmer used to cultivate a plot only once, and not come 112 back in the same place after several years but rather move to a new stand of forest for every 113 new home garden. Our interviews suggest that the increasing population density on the island 114 has led to more individual land appropriation. As a consequence, one hectare of land reserve 115 is not large enough to allow farmers to keep a long fallow rotation with their home gardens. 116 Ninety per cent of the households surveyed follow a complex rotation system, with a 117 combination of long fallows and short fallows. A garden is cultivated for four years. The fifth 118 year is the beginning of a short fallow period of two to three years. This short rotation of six 119 or seven years, including the food garden and short fallow, repeats three or four times. Then, 120 after about 20 years, the place is left to long fallow (20 years), and a new forest plot is cleared 121 for gardening. Most often this second site is far from the house, because all the nearest lands 122 have already been planted with perennial crops (80% of respondents). Within that scheme, a 123 farmer can exploit two to three different sites for gardens during his/her life. 124 The progressive establishment of crops allows a succession of productions. For a one- 125 hectare plot, the first 20 years are dominated by food crop production; perennial crops then 126 take the lead (100% of respondents follow this succession). Food crops are mainly destined 127 for home consumption but they can also be sold in the local market to get some cash income 128 (20% of respondents said they sell vegetables in the local market once or twice every two 129 months). The presence of food crops during the very first years of cultivation gives young 130 households an opportunity not to indebt themselves during the immature period of cash crops 131 (confirmed by 50% of households), which can be considered an investment period. After 132 these 20 years, corresponding to the progressive establishment of perennial crops and 4 years 133 of gardening, perennial crops generate a regular income that can last 80 years for coconuts. 134 Therefore, the progressive establishment of crops allows the family to spread out its income, 6 135 to adapt to its means during the period and especially to the family labour force, and to adjust 136 to the family’s needs for a cash income (Feintrenie et al. In press). 137 Shifting agriculture is usually practised in areas of low population density and where 138 arable land is not limited in either surface or access (Mazoyer and Roudart 1997). However, 139 this situation is currently changing on Malo Island. According to the inhabitants, land pressure 140 is rapidly increasing and access to land can be difficult. Therefore, perennial plantation is also 141 a way for farmers to appropriate and to claim land. The establishment of huge areas of 142 extensive plantations is a low-cost strategy to increase land property with no possible 143 contestation. 144 Coconut–cocoa agroforestry system 145 The values presented below are the results of the computation of the economic data 146 gathered in the field, using the Olympe software. We defined a typical coconut–cocoa 147 agroforestry system of one hectare, based on the data collected in the plots and during the 148 interviews. A one-hectare plot of coconut–cocoa agroforestry is established step by step 149 during 16 years, with the clearance of a small forest plot (1/16 ha) every year. The main work 150 during the establishment phase is the forest clearing and garden cropping. Little work is 151 needed for the young coconuts or cocoa trees. The busiest period is from August to October, 152 with the harvest of yams and taro, but labour demand still only reaches 92 hours/month, or 153 about 3 hours/day for one person (fig. 2c: Work table of a coconut-cocoa plantation during the 154 immature phase). If one person works 8 hours/day and 26 days/month, then available labour is 155 208 hours/month. Thus, one person can manage 208/92 = 2.25 ha of a coconut–cocoa 156 agroforestry system during the establishment period. Using the same calculation, it appears 157 that one person can manage 6.62 ha of a mature coconut–cocoa agroforestry system, when 158 production is at its highest level, which we call the ‘cruising period’ (fig. 2d: Work table 159 during the cruising phase of a one-hectare plot in the coconut–cocoa agroforestry system). 7 160 Return to land is at its maximum during the establishment of the plantation (fig. 2a: Return to 161 land of a one-hectare plot in the coconut–cocoa agroforestry system); it reaches more than 162 1 300 €/year after 5 years (for a cultivated surface of 3 125 m² at that time). This high 163 productivity of food crops maintains itself as long as the farmer enlarges the plot, that is, 164 during 20 years for a one-hectare plot. Then, the annual gross added value in the cruising 165 phase (for our example of one hectare of plantation) decreases to an average of 250 €/year 166 (fig. 2d). Thus, for 6.62 ha of plantation, the monthly gross added value would be 167 250/12 × 6.62 = 138 €/month. This amount is nearly equivalent to the net added value or to 168 the agricultural income, because of the very low costs of production (very cheap tools, no 169 expensive building or materials, no tax, no subvention, etc.). Therefore it can be compared to 170 the minimum wage in Vanuatu, which was 130 €/month in 2005 (Radio New Zealand 2005), 171 compared with 138 €/month produced by the plantation with less work. 172 Return to labour is also high during the establishment phase, with an average of 2 €/hour 173 (fig. 2b: Return to labour of a one-hectare plot in the coconut–cocoa agroforestry system). 174 Then, during the cruising phase, return to labour stabilises at around 1.10 €/hour for one 175 hectare of plantation. This too is higher than the national minimum wage hour income. 176 Coconut–vanilla agroforestry system 177 Vanilla has been cultivated on Malo Island since 2000; it is presently the fashionable 178 diversification crop for coconut–cocoa plantations. It is cultivated either in monospecific 179 plantations on Glyricidia sp. or Erythrina sp. live stakes, or under coconuts. As it was still a 180 new crop at the time of the field work (2005), cropping schemes were not yet strictly fixed by 181 farmers, who were trying different practices, among which was association with coconuts. In 182 Malo, vanilla was found associated with coconuts of every age, from 4- to 40-year-old 183 plantations. Others were planted at the same time as new coconut plantations, but this met 184 technical problems because of the spatial bulkiness of young coconuts; it is difficult to 8 185 intercrop vanilla between coconuts lines before the crown grows higher than 4 m. Because of 186 this inconvenience, we assumed that in the future farmers will plant vanilla only under 187 coconuts older than 8 years. Shading can be beneficial for vanilla at several stages, but 188 coconut shading may cause some difficulties because it cannot be controlled throughout the 189 year and palm or fruit falls can hurt vanilla ropes. On the other hand, as vanilla plants need 190 inert compost to grow, coconut is very complementary, as coconut husks and shells are 191 layered on the soil surface around each live stake. This mulch provides the vanilla’s aerial 192 roots with moisture, shelter and nutrition. 193 Economic performances of the coconut–vanilla agroforestry system were estimated using 194 the same modelling as for the coconut–cocoa system. The same dynamic of establishment was 195 modelled: one-hectare plot cleared by sectors of 625 m² every year, plantation of coconuts in 196 the first year, food crops intercropping during the first four years and then plantation of 197 vanilla in the fifth year. The results are presented along the whole length of the coconuts’ 198 production life, namely 96 years. We considered a plot of 1 220 vanilla ropes/ha, with a 199 production estimated through surveys at 24.4 kg/ha at 3 and 4 years, 48.8 kg/ha from 5 to 8 200 years, then 24.4 kg/ha at 9 and 10 years. Vanilla ropes are replaced in the 11th year, when 201 there is no production. 202 The analysis of the work table of a one-hectare plot in the coconut–vanilla agroforestry 203 system (fig. 3c: Work table during the cruising phase of a one-hectare plot in the coconut– 204 vanilla agroforestry system) shows that it has a high labour demand. A single person can 205 manage only 0.19 ha of this cropping system because of the huge amount of work needed 206 during the flowering and harvesting periods, from September to December (the same 207 calculation technique as above). Pollination must be done by hand as the natural agent of 208 fecundation, a small bee, is not in Vanuatu. This work and the harvest are carried out every 209 morning for 4 to 5 months. 9 210 Return to land is very high, at between 3 000 and 4 500 €/ha a year (fig. 3a: Return to land 211 of a one-hectare plot in the coconut–vanilla agroforestry system), with an average of 3 607 212 €/ha in a year, or 300 €/ha a month. This result is because of the high price of vanilla in 213 Vanuatu, especially in comparison with copra or cocoa. Nevertheless, this high return to land 214 is moderated by the low return to labour (fig. 3b: Return to labour of a one-hectare plot in the 215 coconut–vanilla agroforestry system). Indeed, return to labour is around 1 €/hour, which is 216 half the coconut return to labour or 4 to 5 times less than the food crops’ return to labour. 217 Thus vanilla can yield a lot by hectare, but only with a huge labour demand throughout the 218 year, which limits the possibility of relying on day workers. The valorisation of 0.19 ha by 219 one person is only at 685 €/year or 57 €/month. 220 Thus the establishment of a coconut–vanilla agroforestry system responds to different 221 farmers’ objectives and strategies to those of a coconut–cocoa agroforestry system. The 222 association of vanilla with coconuts is interesting for a farmer who has limited access to land 223 but available labour. On the other hand, a strategy of land occupation by extension of 224 plantations every year will be better served by a coconut–cocoa system, as it is less 225 demanding in terms of labour. 226 Comparison and combination in a coconut–cocoa–vanilla agroforestry system 227 These first analyses of the economic returns of vanilla and cocoa under coconut cropping 228 systems demonstrate that these two systems are promising and answer the complementary 229 needs and strategies of farmers. A second step in the search for the best way to integrate 230 vanilla into small family farms is to evaluate the feasibility and profitability of combining 231 vanilla with both the main crops. For this, a new economic model is proposed. The aim of this 232 exercise is to test the new cropping system observed in the field – vanilla under coconut – by 233 simulating its economic profitability over a whole production cycle. It is once again based on 234 a one-hectare plot, progressively planted by sections of 625 m². In this model, vanilla and 10