| dc.description.abstract | Intensified agricultural practices lead to a reduction in ecosystems carbon stocks.
This is mainly due to removal of aboveground biomass as harvest with subsequent
burning and/or decomposition and loss of soil carbon as carbon dioxide and soil
through erosion. The effects of forest conversion and subsequent cultivation on
carbon stocks and soil properties were monitored in demarcated land use types
along transects in Mt. Marsabit ecosystem. The main objective of this study was to
develop a practical understanding of the impact of deforestation and the mitigation
measures being put in place on soil properties with emphasis on carbon stocks.
Near infrared reflectance spectroscopy (NIRS) and 12 chemical and physical
parameters were studied on soil samples collected from three different land use
systems namely forest, cropped and pasture land. Soil properties were calibrated to
spectral reflectance using partial least square (PLS) regression. Two hundred and
twenty two soil samples had been augured from the 0-20 em and 20-50 em depths .
Seventy four samples from the reflectance data were chosen for prediction. There
were four hundred and twenty seven soil samples obtained using core rings in soil
profiles laid in each land use type, to a depth of 150 em at 5 em intervals. The soil
samples were used to estimate the belowground carbon. The collection of litter was
done at 90 points using a 0.5xO.5 m quadrant. The litter was dried and weighed as
part of the aboveground carbon. The diameter at breast height (DBH) was measured
from 161 trees selected randomly in the forest for use in calculating the standing
aboveground carbon.
Total carbon ranging between 0.98 - 8.1 gkg', r2 was 0.97, bias was 0.021 gkg' with
root mean standard error of prediction (RMSE) of 0.78 gkg'. Predicted and measured
values for the other soil properties were: total nitrogen r2 = 0.95, bias was 0.001 gkg'
with RMSE of 0.14 gkg'; pH r2 = 0.95, bias was 0.0003 crnolkg' with RMSE of 0.01
cmolkg'; exchangeable magnesium r2= 0.94, bias was 0.0009 cmolkg' with RMSE of
0.18 cmolkg'; exchangeable calcium r2= 0.76 bias was 0.0007 cmolkg' with RMSE of
0.16 cmolkg' and CEC r2= 0.76 bias was 0.0008 cmolkg' with RMSE of 0.09 cmolkg'.
These soil properties were significantly different (P :S 0.001) (appendix 1) in cropped
and pasture compared to forest and their variation was successfully predicted (r"> 0.76)
using NIRS. The mean spectral reflectance was significantly different among the three
land use systems (P:S 0.001). The total (above and belowground) carbon stocks were
significantly different (P:S 0.001) in cropped and pasture compared to forest areas. The
carbon stocks declined by 47.4% and 45.6% in cropped and pasture land sites relative
to the forest site.
Belowground carbon tended to be almost constant at 100 em forest, 80 ern in pasture
land and 60 em in cropped land probably because soil carbon contents generally
decrease with depth, as organic inputs are primarily deposited on the soil surface
or occur in the topsoil where most of the turnover of fine roots occurs. In general,
however, decomposition processes were slower down the soil profile and the carbon
stocks that existed below the topsoil were better protected from physical disturbance
by vegetation roots and they were likely to change more slowly after land use change.
Also, the rooting system of the trees in the forest was dense compared to the bushes in
the pastureland and the crops in the cropped land use systems. | en |
