4.1. Respondent selection 4.1.1. Grouping of respondents
Key-informant interviews and participatory mapping were conducted in 2008 in Kalimantan and Sumatra and in 2010 in Java in order to collect base-line information about potential study villages, and to identify – for the purpose of sampling – who the tree planters and non-planters were. The model of how the farmers planted trees was also determined; whether in woodlots, home gardens or on agricultural borders of their land. Tree planters and non-tree planters were defined for each study site according to the different species they were planting. In the three villages in Riau the farmers were divided into acacia planters and non-acacia planters; in Asem Jaya Village into kadam planters and non-kadam planters; in Ranggang Village into mahogany planters and non-mahogany planters; and in the two villages in Central Java teak planters and non-teak planters. Most of the farmers in all the villages planted several other tree species on their land (mainly fruit trees, and in South-Kalimantan and Riau also rubber; Hevea brasiliensis Muell. Arg), however only tree species that were planted primarily for timber or fiber production were included in the analyses. This means that when we refer to ‘tree-planters’, we are referring to farmers planting acacia, kadam, mahogany or teak only; whereas ‘non-tree planters’ are those farmers not planting these species.
Initial wealth ranking was conducted in order to have representative number of farmers from different wealth categories in the sample. The three different wealth categories were; poor (1), moderate (2), and wealthy (3), and the rankings were based on village statistics and the criteria determined by the key informants (Annex 1). The key informants included participants from the village, sub-village, and farmers’ group levels.
4.1.2. Sampling
The aim was to select the household respondents using stratified random sampling. First the province, district, sub-district, village, and sub-villages (in the case for the Central Java case study) were purposefully selected according to the criteria explained in section 3.1. Tree planting and non-tree planting respondents were randomly selected in each village, in proportions according to their wealth ranking (poor, medium, and rich). A minimum of 30 tree planters and non-tree planters per site were selected if available (the three villages in Riau were considered as one site while selecting the respondents and during the analyses). Modification of the sampling plan was required however, because random sampling would not have guaranteed a representative number of tree planters. This is why the farmers planting mahogany in Ranggang village, kadam in Asem Jaya village and teak in the sub-villages in Sendangijo were purposefully selected (all the households planting trees in even-aged woodlots were selected). In addition, in Ranggang village 24 households planting mahogany on the borders of their land were selected (however these households were not used for the study on the silvicultural practices and the quality of the stands).
A total of 412 farmers were interviewed; including 78 acacia planters and 39 non-acacia planters, 31 kadam planters and 36 non-kadam planters, 44 mahogany planters and 57 non-mahogany
planters, 67 teak planters and 60 non-teak planters. The number of the selected respondents varied between the villages because of the varying number of households, the varying number of tree planters and non-tree planters in the villages, and because of variation in the time, resources, and number of researchers available during each site visit.
4.2. Inventory methods
All of the selected tree planter’s stands that were planted in multiple rows with regular spacing were inventoried in the four villages in South Kalimantan and Central Java (excluding trees planted only in the borders of the land). Links were made between the household surveys and the forest inventories for all selected farmers except for the three villages in Riau, where the link could not be made because farmers did not know exactly where their two hectares of land was located (that was planted with acacia in the village land as part of the company-community partnership), and were not involved in the silvicultural management. This made it impossible to study the factors influencing farmers’ silvicultural management in Riau, thus only the factors affecting farmers’ decisions to join the partnership in this site were studied.
The unit of analysis for the inventory was a tree stand with an area of 0.1 ha or greater and containing a minimum of 50 trees that were planted by a farmer or their forbearers. The aim was to measure 20 trees per sample plot, and to establish one or more fixed area of sample plot (circular plots) in each stand (depending on stand size and planting density). Each stand was classified using visual observations in one of the following three stand quality groups: (1) well-maintained trees in generally good condition; (2) overall condition acceptable, some problems are likely to appear; and (3) poorly managed, and success or survival of the stand is doubtful.
DBH was measured for each tree. Total tree height and the height of the crown base were measured for 25-30% of the trees in each plot in the villages in South Kalimantan and for all the trees in the plot in the villages in Central Java. Each tree was classified into one of the following stem quality classes related to sawn wood value: (1) high, no major irregularities or defects;
(2) medium, in part usable as sawn wood, some defects; and (3) poor, unsuited for sawn wood, too many defects. In addition, trees were classified into three groups according to their overall condition with respect to viability and growth potential: (1) dominant, in the upper crown layer and superior to immediate neighbors; (2) sub-dominant, in the lower crown layer and receiving less light, smaller than immediate neighbors; and (3) others, all trees that neither had major nor medium irregularities and defects, that are not in groups 1 and 2.
4.3. Interview methods
Two structured questionnaires, a socio-economic and a silvicultural management-oriented questionnaire, were applied in the household interviews. The questionnaires were field-tested and then modified before implementation. Interviews were conducted at farmers’ houses by enumerators from the local Forestry Research Institutes and the Forestry Research and Development Agency (FORDA). The enumerators were trained, supervised and accompanied by the researcher. The socio-economic questionnaires collected data on household characteristic including the number
of the household members, age, gender, education, occupation, expenses, assets, income sources, market access, and participation in farmers’ groups or other social organizations and on farm characteristics including land area, land use and land ownership. The silvicultural management questionnaire included questions on: (1) farmers’ selection of species, silvicultural knowledge and practices; (2) farmers’ motivations and attitudes towards tree planting; (3) incentives received for tree planting; (4) disadvantages related to tree planting; (5) previous or planned use of income received from timber; and (6) farmers’ ideas for improving tree planting activity.
The socio-economic questionnaire was administered to all farmers, whilst the silvicultural management questionnaire was administered only to the tree planters. The questions related to farmers’ silvicultural practices were applied for each stand separately in South Kalimantan and Java.
4.4. Statistical analyses
Descriptive statistics, chi square, Mann-Whitney U test and t-tests were used for the analyses conducted in Study I in order to study the relationships between farmers’ socio-economic characteristics and their tree planting activity. To study the relationship between socio-economic factors and farmers’ silvicultural activity, data were analyzed using Spearman correlations, descriptive statistics, Mann–Whitney U tests and cross tabulations (Study II and III). The relationship between plantation quality and the silvicultural practices applied by the farmers was analyzed using descriptive statistics and by hypothesizing and statistically fitting a general linear model (Study II).