Agriculture Reference
In-Depth Information
in the case study.* For brevity, we use only two land management treatments. The
first represents a land degrading practice commonly used in the countries included
in the case study, and the second represents fairly sustainable land management prac-
tices deemed feasible in both countries. Details of the two land management prac-
tices are given below:
(1) Land degrading practice . No chemical fertilizer or manure application and
50% harvesting of crop residues. This is a practice that reflects land man-
agement practices of most farmers in SSA—where it is estimated that only
10 kg/ha of nutrients [nitrogen, phosphorus, and potassium (NPK)] is used
(FAOSTAT 2009). Crop residue harvesting and in situ feeding by livestock
are also common practice in SSA.
(2) Integrated soil fertility management (ISFM) practices . The use of organic
inputs, improved crop varieties, and judicious amounts of chemical fer-
tilizers (Vanlauwe and Giller 2006; Tittonel et al. 2008). We use 40 kg
of nitrogen per hectare of chemical fertilizer, 1.7 tons of manure/ha, and
50% of crop residues as the treatment representing ISFM. The recom-
mended fertilizer rate for maize is about 80 kg N/ha for subhumid areas in
West Africa (Aduayi et al. 2002). We use the lower rate of fertilizer appli-
cation to reflect the benefit of ISFM, which typically reduces the chemical
fertilizer use requirement (Sharma and Biswas 2004; Vanlauwe and Giller
2006; Roy et al. 2006), and to reflect the low rates used by farmers in the
SSA region. Akinnifesi et al. (2010) also observed that agroforestry that
uses leguminous trees planted on maize plots in Eastern and Southern
Africa reduced the requirement of nitrogen from chemical fertilizers by
up to 75%.
Controlling for other drivers of crop yields and varying only the two soil fertil-
ity management practices listed above, it is possible to estimate the impact of land
degradation on crop yield. We show the trend of yields using a linear trend model to
give a rough estimate of the yield loss per year. Linear trend models are commonly
used in estimating crop yield trends (e.g., see Bruinsma 2009; Moss and Shonkwiler
1993). However, the model uses normality assumption. The Shapiro-Wilk normality
test (Shapiro and Wilk 1968, 1972) showed that maize and rice yields in each country
were normally distributed—something that was expected given that the data used
are simulated:
Y t = β 0 + β 1 t + ε t
(9.1)
where Y t = crop yield in year t , and β 1 = slope of yield trend depicting the annual
change of crop yield. A negative sign will show declining yield and an indicator of
impact of land degradation. It should be noted that this model only presents land
* Details of the calibration and other simulation procedures are available from the authors upon request.
Search WWH ::




Custom Search