2.4. Soil sample collection, field measurements and analysisThe study was conducted on 36 plots in 6 farms. Six sampleplots in each farm were selected in the following range ofincreasing slope gradients: <20, 20–29, 30–39, 40–49, 50–59 and60%. Slope gradient was measured using a manual clinometer.Aerial cover and the height of the coffee plants and shade treestrata were measured using a laser distance meter. Samples weretaken from the first 10 cm of soil in the centre of each sample plot(point 0 in Fig. 3). Particle-size distribution was determined by thepipette method, after elimination of organic matter with H2O2 anddispersion with sodium hexametaphosphate (Loveland andWhalley, 1991). The calcium carbonate content was determinedby acid treatment with 1:1HCl in closed vessels and manometricreading of the pressure of CO2 released (Barahona, 1984). Totalnitrogen and carbon were analyzed by dry combustion in anautomatic LECO TruSpec-CN instrument, and organic carbon wasestimated by the differences between total carbon (from LECOanalysis) and inorganic carbon (from calcium carbonate analysis)(Sierra et al., 2013). The exchangeable bases were determined aftersaturation with ammonium acetate 1 N pH 7 and measurement ofthe elements in a VARIAN 220FS SpectrAA atomic absorptionspectrometer (USDA, 1972). The cation exchange capacity wasdetermined after saturation with sodium acetate 1 N pH 8.2 andextraction of the sodium retained with ammonium acetate 1 NpH7. The sodium representing the cation exchange capacity wasmeasured in a VARIAN 220FS SpectrAA atomic absorptionspectrometer (USDA, 1972). All atomic absorption spectroscopymeasurements were performed in triplicate and standard solutions(Merck quality) were used for calibration.The state of soil erosion and the erodibility factors weredetermined by bivariate correlation analysis (Spearman correlationcoefficient, Rho) and multiple linear regression analysis (stepwisemethod). The Bonferroni correction was applied to adjust thesignificance level (alpha) of the 24 correlations obtained betweenerosion and the independent variables included (alpha = 0.05/24 = 0.002). The erosion threshold, in relation to erodibility factors,was obtained by means of two types of variance analysis (Kruskal–Wallis test for several independent variables and Mann–Whitneytestfor twoindependentvariables). Thefirstwas a generalanalysis inwhich the results of erosion were distributed at different intervals,usingthe soilerodibilityfactors as the grouping variable. Twocriteriawere applied for this purpose: (a) the distribution should presentregular intervals, that is, representing a constant rate of growth (forexample, at intervals of 10: 10–20, 20–30; or of 20: 10–30, 30–50),because the variance is sensitive to the range if the results present ahigh degree of dispersion; (b) the distribution should present aminimum size (n) of 3 cases in each interval. The second type ofanalysis performed concerned the variance between the erosionand the erodibility factors found to be significant in the previousanalysis. This analysis consisted of examining pairs of consecutiveintervals. The erosion threshold was established at the boundarybetweenthe two intervals ofthepairs analysis for whichthehighestlevelof statistical significancewas obtained.Allthese analyzeswereconducted using SPSS 22.0 software.
การแปล กรุณารอสักครู่..