Agroforest Syst (2012) 86:493–504 DOI 10.1007/s10457-012-9513-9 Afforestation of savannah with cocoa agroforestry systems: a small-farmer innovation in central Cameroon Patrick Jagoret • Isabelle Michel-Dounias • Didier Snoeck • Herve´ Todem Ngnogue´ • Eric Male´zieux Received: 14 July 2011 / Accepted: 17 March 2012 / Published online: 4 April 2012 Ó Springer Science+Business Media B.V. 2012 Abstract Cocoa cultivation is generally considered grasslands and 182 cocoa plantations established in to foster deforestation. Contrary to this view, in the gallery forests. In a sub-sample of 47 grassland cocoa forest–savannah interface area in Cameroon, farmers plantations, we inventoried tree species associated have planted cocoa agroforestry systems on Imperata with cocoa trees and measured soil organic matter cylindrica grasslands, a soil-climate zone generally levels. Marketable cocoa yields were similar for the considered unsuitable for cocoa cultivation. We two types of cocoa plantations, regardless of their age: undertook a survey to understand the agricultural 321 kg ha-1 in cocoa plantations on grasslands and and ecological bases of this innovation. Age, cropping 354 kg ha-1 in cocoa plantations in gallery forests. history and marketable cocoa yield were assessed in a Two strategies were used by farmers to eliminate sample of 157 cocoa plantations established on I. cylindrica prior to the establishment of cocoa plantations, i.e., cropping oil palms in dense stands and planting annual crops. Farmers then planted cocoa trees and fruit tree species, while preserving specific forest trees. The fruit tree and forest tree densities -1 P. Jagoret (&) Á D. Snoeck respectively averaged 223 and 68 trees ha in CIRAD, UPR Syste`mes de Pe´rennes, 34398 Montpellier plantations under 10 years old, and 44 and 27 cedex 5, France trees ha-1 in plantations over 40 years old, whereas e-mail: [email protected] the cocoa tree density remained stable at 1,315 -1 P. Jagoret Á H. T. Ngnogue´ trees ha . The Shannon–Weaver index increased Perennial Crop Department, IRAD-Nkolbisson, BP 2067, from 1.97 to 2.26 over the same period although the Yaounde´, Cameroon difference was not statistically significant. The soil I. Michel-Dounias organic matter level was 3.13 % in old cocoa planta- UMR 951 Innovation Montpellier Supagro INRA CIRAD, tions, as compared to 1.7 % in grasslands. In conclu- Campus de la Gaillarde, 2 Place Viala, 34060 Montpellier sion, our results show that the occupation of grasslands cedex 1, France by cocoa agroforestry systems is both an important I. Michel-Dounias Á E. Male´zieux example of ecological intensification and a significant CIRAD, UPR HortSys, 34398 Montpellier cedex 5, farmer innovation in the history of cocoa growing. France Keywords Imperata cylindrica Á Sustainability I. Michel-Dounias Institut des Re´gions Chaudes, 1101 Avenue Agropolis, BP assessment Á Agrobiodiversity Á Ecological 5098, 34093 Montpellier Cedex 05, France intensification Á Forest–savannah interface 123 494 Agroforest Syst (2012) 86:493–504 Introduction matter levels. According to Burle (1961), cocoa trees require 1,500–2,500 mm of water a year, with a dry In Africa, cocoa (Theobroma cacao L.) cultivation is season that has less than three consecutive months of generally considered to be a major cause of defores- below 70 mm. Sandy-clayey soils with over 3 % tation in humid tropical areas since the trees are often organic matter in the topsoil are the most suitable soils planted on newly cleared forest land (Dixon et al. for cocoa growing. However, such conditions do not 2001). The expansion of cocoa plantations in Coˆte prevail in the forest–savannah interface area in central d’Ivoire, for instance, led to a reduction in the Cameroon where the annual average rainfall is around country’s forest area from 13 million ha in 1960 to 3 1,300 mm with sandy topsoil which contains less than million ha in 1990 (Hanak-Freud et al. 2000). The 2 % organic matter (Santoir and Bopda 1995). same pattern has been noted in Ghana, where it is Champaud (1966) also demonstrated that this area is estimated that 80 % of the forest area has disappeared far north of the boundary of the ideal cocoa growing since cocoa trees were introduced (Cleaver 1992). region in this country. Planting cocoa agroforestry In Cameroon, the role of cocoa cultivation in the systems on I. cylindrica grasslands in central Camer- deforestation process is not as clear since forests oon thus seems to contradict the opinion of cocoa are generally only partially cleared for planting cocoa, experts. many forest tree species being preserved in Vieira et al. (2009), in their review article on the resulting complex cocoa agroforestry systems the concepts of agroforestry systems and agro- (Duguma et al. 2001; Zapfack et al. 2002; Laird et al. successional restoration, highlighted the shortage of 2007; Sonwa et al. 2007). scientific literature on the description and analysis of In the present study, we show that cocoa growing agro-successional restoration examples in the world, contributes to the process of afforestation of grass- especially in tropical areas. These authors also stated lands dominated by Imperata cylindrica in the forest– that both scientists and farmers involved in agrofor- savannah transition area of central Cameroon. Pub- estry should think about how agroforestry and resto- lished reports about afforestation in forest–savannah ration might be integrated in different situations. In interface areas in Africa are relatively ancient. In most central Cameroon, grassland cocoa agroforestry sys- situations, this process was due to the natural dissem- tems are an as yet undocumented example of this. ination of forest tree species, as described by Boulvert The objective of the present study was to document (1990) in the Central African Republic, Clayton the gradual transformation of grasslands into cocoa (1958) in Nigeria, Swaine et al. (1976) in Ghana, agroforestry systems in the forest–savannah interface and Bonvallot et al. (1970) and Spichiger and Pamard area of central Cameroon. Firstly, we checked the age (1973)inCoˆte d’Ivoire. Only a few cases of voluntary of the cocoa plantations established on grasslands and afforestation have been described in Togo (Guelly we compared their cocoa yields to those of cocoa et al. 1993), Coˆte d’Ivoire (Blanc-Pamard and Peltre plantations in gallery forests. Secondly, we focused on 1984) and Guinea (Fairhead and Leach 1996), where the conditions required for establishing cocoa agro- Robusta coffee-based agroforestry systems were forestry systems on grasslands and farmers’ manage- planted, since this coffee cultivar is able to adapt to ment of the different associated stands of tree species. grassland soil and microclimatic conditions. Finally, we assessed the agrobiodiversity and soil To our knowledge, only one case of afforestation fertility in these systems using the Shannon–Weaver involving cocoa trees has been reported in Indonesia index and the total soil carbon content respectively. (Ruf and Yoddang 2004), where farmers in central Sulawesi planted cocoa trees on I. cylindrica grass- lands after conducting chemical weed control sprays Materials and methods and planting the leguminous tree Gliricidia sepium to eliminate the grass and temporarily shade the young Study site cocoa trees. According to Wood and Lass (1985), grassland areas are theoretically unable to fulfil the The study was carried out in the forest–savannah ecological requirements of cocoa trees because of the interface area in central Cameroon, around the villages insufficient and variable rainfall and low soil organic of Bakoa, Begni, Yorro and Kedia in Bokito district: 123 Agroforest Syst (2012) 86:493–504 495 4°30 latitude N and 11°10 longitude E; 450–500 m rainfall conditions would normally induce defoliation, elevation (Fig. 1). The farmers were mainly indige- which could be fatal to the trees (Burle 1961). Ferric nous ethnic Yambassa people. Bokito district Acrisols predominate in this area. The topsoil contains (7,125 km2) is a patchwork of gallery forests occu- over 60 % sand and less than 2 % organic matter pied by cocoa agroforestry systems and grasslands (Santoir and Bopda 1995), which are further con- dominated by I. cylindrica. Invasion of fields by straints for cocoa trees (Burle 1961). I. cylindrica is a major farming constraint because this heliophytic weed quickly reappears as soon as the crop Experimental design, stakeholder survey data fields are insufficiently maintained or when the crop and measured data cover is sparse (Deuse and Lavabre 1979). The forest–savannah interface area has a hot and In 2004, 339 cocoa plantations belonging to 282 humid climate with an average annual temperature of farmers were randomly selected from lists of farmers 25 °C. Mean annual rainfall is 1,300 mm, which belonging to local cocoa growers’ organizations. We represents a mean deficit of 200 mm relative to the interviewed each of these farmers to collect the minimal annual water needs of cocoa trees. The following data: year of establishment of each cocoa rainfall distribution is characterized by a dry season plantation (plantation age in years), vegetation cover lasting over 3 months (December–March) during before cocoa trees were planted (gallery forest or which the monthly rainfall is less than 70 mm (Santoir grassland), area and declared production for the three and Bopda 1995). In monocropped cocoa stands, these harvest seasons prior to the survey (2001, 2002, and 300 R (mm) 250 200 NORTH 150 100 Chad Nigeria 50 0 JFMAMJJASOND Central Bokito African Republic Yaoundé Atlantic Ocean Bakoa Equatorial Gabon Congo Kedia Guinea Begni Yorro State capital Regional borders Main road Study site Yaoundé Forest Savanna Study village Fig. 1 Map showing the location of the study sites in central Cameroon 123 496 Agroforest Syst (2012) 86:493–504 2003). The mean marketable cocoa yield of each cocoa variance (ANOVA) using a general linear model.