West African cocoa farmer tree valuation data

The study was conducted at 15 sites across Côte d’Ivoire, covering a South-North climate and vegetation gradient from evergreen to semi-deciduous forests. These sites were selected to represent varying cocoa field characteristics, including structural complexity (ranging from monocultures to agroforests), cocoa field age (young to mature), and cocoa tree productivity (low to high). Each site had 10 cocoa plots, ranging from 0.3 to 5 hectares. The study also spanned different "cocoa loops," covering areas from the eastern to western parts of the country. Cocoa farmers were involved in a ranking exercise using the stone method to assign importance to tree species across 10 categories of use.

Methods
The adopted approach involved the use of the stone ranking method developed by Sheil et al. (2004) and applied by Jagoret et al. (2014) in cocoa fields in Cameroon. A comprehensive table with tree species listed in rows and 10 use categories in columns was presented to each cocoa farmer. Each farmer had a bag with an ample number of small stones, more than enough for the exercise. The farmer was then asked to place between 0 and 10 stones in each cell of the table, based on the importance they assigned to each combination of tree species and use. Due to the 10 categories of use, the farmer could distribute up to 100 stones per tree species. 

Three response variables were calculated to answer research questions:

Mp: Global motivations, representing total use values of trees per hectare.

MpU: Motivations by category of use, calculated per use category and per hectare.

Sp: Specialization of tree uses, the level of specialization of uses, which we have defined as the opposite of the observed diversity of uses. To obtain it, we calculated, for each field, the first-order diversity MpU transformed into a Hill number. Since the maximum possible value is 10 (all uses have exactly the same level of motivation, i.e. maximum diversity), we transformed the obtained value by subtracting it from 10. Thus, we obtain a measure of the specialization of uses.

Calculations of Sp were performed using R’s ‘entropart’ package.