Mineral nutrition, soil factors and growth rates of Gmelina arborea plantations in the humid lowlands of northern Costa Rica

The purpose of this study was to determine the factors which are responsible for clearly visible growth irregularities in Gmelina arborea stands by relating growth rates to soil and site properties. All Gmelina plantations under study, established on degraded pasture soils in the Atlantic lowlands of Costa Rica, showed the same growth pattern: very poor growth and chlorotic foliage of trees in mid-slope positions, and fast growth and healthy leaves of trees on hilltops and hillbases. The variation in site and soil properties and tree growth rates was analysed, investigating 24 plots in eight different plantations. Leaf samples were taken from 120 trees (five trees per site). Foliar analysis revealed that the tree growth is highly dependent on the supply of N, P, K and S, indicating that poorly growing trees suffer from a multiple nutrient disorder. To identify the most restricting soil factor, simple correlations between growth rates and soil chemical and physical properties were applied to the entire data set. The best correlation was obtained with exchangeable soil K (r=0.78, P<0.001). Subdivision of the data set into plots on brown soils (eight) and plots on red soils (16) and subsequent correlation analysis resulted in much stronger relationships. Growth depressions in mid-slope positions had other causes on brown soils than on red soils. On the acid brown soils the combination of the variables Al saturation and bulk density could explain 82% of the variation of tree growth. Al saturation of up to 80% in brown soils inhibits nutrient uptake, particularly of N and P. The red soils dominate on Ca---Mg-enriched, alluvial terraces and were among the first soils to be cultivated in the region. During the agricultural use and at the establishment of the tree plantations, they received dolomitic lime to reduce Al toxicity. Here, very low K/Mg ratios (less than 0.03) may induce K deficiency. Therefore, the best multiple regression model for tree growth rates on red soils is obtained with K/Mg ratio and thickness of the humic A-layer (r²=0.75, P<0.001). Inclusion of the variable bulk density resulted in a clear improvement of the model, explaining 91% of the growth variability.