Estimations of amounts of soil organic carbon and fine root carbon in land use and land cover classes, and soil types of Chiapas highlands, Mexico

Amounts of organic carbon in the mineral soil (SOC) and fine-root (<5 mm) carbon (RC) were quantified, when possible down to I m depth, in 150 soil profiles from 39 sites in the highlands of Chiapas, Mexico. The sampling sites were classified and grouped according to the following soil type: Leptosols, weathered soils (Ferralsols, Acrisols, Lixisols and Nitosols), Luvisols and Cambisols/Phaeozems. Likewise, they were classified in the following groups with respect to land use/land cover (LU/LC): oak-evergreen cloud forest, fragmented forest, pine and pine-oak forest, and open land (cultivated land and pasture). No clear influence of soil-type grouping on amounts of SOC was revealed, between soil-type groups (soil type). The weathered soils had higher amounts of SOC and RC in the upper 50 cm than the other soil types and the lowest amounts of SOC and RC in the 50100 cm layers. The LU/LC groups showed marked differences in amounts of SOC. Open land had between 20 and 60% less soil carbon than the different types of forests. Oak-evergreen cloud forest had the largest amounts of SOC at all depths. The differences in amounts of carbon (SOC and RC) between oak-evergreen cloud forest and the other LU/LC classes were proportionally larger in the 50-100 cm layer. This suggests that the amounts of SOC in deep soil layers may be influenced by changes in LU/LC. The considerably smaller amounts of SOC and RC 174 and 38.5 Mg ha(-1), and 31 and 24.8 Mg ha(-1), exhibited in open land compared to oak-evergreen cloud forest and pine and pine-oak forests, respectively, suggest that conversion of forests to open land reduces SOC and RC. Despite the magnitude of the differences, evaluation with ANOVA (nested design) did not reveal a statistically significant influence of LU/LC. The large spatial variability in amounts of soil carbon prevented precise estimates. Changes induced by a gradient in altitude among the plots may have contributed to the large variability between the plots. (C) 2002 Elsevier Science B.V. All rights reserved.