Experimental site and description area
The Barahona Province has an extension of 1,639 km2. It is located between latitudes 18°40′ and 18°12’N and longitudes 71° 17′ and 71°21′ W, southwest of the Dominican Republic. The area has an altitude between 700 and 1200 m.a.s.l, has a temperature annual range of 17 and 22 °C with an annual average of 26 °C1 The annual rainfall varies between 655 and 2296 mm (Pérez et al, 2002). The coffee cultivated area in the Barahona Province is 11,082 ha with about 3,565 participating farmers (CODOCAFE, 2002). Most of the area (98.6% or 30,710 ha) is in arabica cofee (C. arabica), whith the remainder (1790 ha) is planted to robusta coffee (C. canephora).
Farm survey and soil sampling
The fifteen coffee production areas of the Provice were identified from which 96 farms were selected. The farms were identified based on the 2001 National Coffee Census (CODOCAFE, 2001). The sample size was computed using the software InfoStat (2008), with an error of 10%, considering the total number of coffee farms (3,565 farms), in an area of 11,082 ha. Specific location of each farm was made using using topographic sheets for the area. All farm visits were made from April to July 2009.
The distribution of cases by location was based on proportion of farms and area occupied by location: Las Guazaras (7), Santa Helena (9), Bahoruco (6), Las Cienaga (12), Polo (3), Breton (10), Monteada Nueva (1), La Lanza (7), Los Charquitos (8), Platon (1), Leonardo (7), Los Patos (7), Maria Teresa (5), Chene (7) and El Pino (6). Each farm was georeferenced with GPS model Garmin GPS 76 (Garmin International Inc., Olathe, KS). A survey was prepared and personally supplied to each farmer in order to characterize current yields, agronomic practices and other related information, for example: weed control, pest management practices, soil conservation practices, extent of shade-tree pruning, coffee tree pruning and fertilization or manure management practices.
Soils were sampled to a depth of 30 cm with bore or cutting blade. A composite sample from each farm was gatherd by taking six subsamples from each of the three main geomorphologic positions (shulder, backslope, foot-toe slope) at the farm. Two subsamples were gathered from each geomorphologic position). All leaves and other surface debris were removed by hand. The six sub-samples were homogenized to obtain a composite sample of 2 kg. Samples were placed in previously identified plastic bags and sent to the laboratory for analisis within two weeks.
Soil characteristics
Soil samples were air-dried and sieved to pass a 2-mm mesh sieve. Soils were analyzed for soil fertility characteristics following methodology outlined in Page et al. (1982).
The pH was determined in a ratio 1:2 (soil: water) using a potentiometer. The organic matter (OM) was measured by the Walkey and Black method (potassium dichromate oxidation). The electrical conductivity was determined on the supernatant obtained from a 1:2 (soil: water) suspension using a conductivity bridge. The total nitrogen (TN) was measured using the Kjeldahl method. Available P was extracted using the Mehlich III extractant followed by quantification in a UV-Vis spectrophotometer. The exchangeable bases were measured by the extraction with NH4OAC followed by quantification using atomic absorption. Exchangeable acidity was extracted with 1M KC1, followed by the quantification of Al and H by titration. The micronutrients iron (Fe), copper (Cu), manganese (Mn) and zinc (Zn) were determined by the method of digestion with nitric-perchloric acid by followed by quantification by atomic absorption.
We calculated the effective cation exchange capacity (ECEC) based on the sum of the bases and exchangeable acidity. Aditional relations such as Ca/Mg, Mg/K, Ca+Mg/K, sodium saturation percentage (SSP) and aluminum saturation percentage (ASP) were calculated. The texture was measured in the laboratory using a hydrometer Bouyucos (Page et al, 1982).
All tests were performed according to the standard protocols of the CENTA-IDIAF Soil Test Laboratory. Each sample was tested the color (wet) with the Munsell color chart.
Statistical analysis
The data were analyzed using descriptive statistics, using the InfoStat software (version 2008). Multivariate analysis was applied to the data: principal component analysis (PCA) and cluster analysis, with clustering by Euclidean distance.