Maize legume relay intercrops in Malawi - Microbial Ecology in Sustainable Agroecosystems - page 248

Table 9.7 Simple Regressions of Long-Term Soil Quality Indicator Variables and Maize Yields
for D1 and D2

D1 D2

Soil indicators Beta R 2 P value N Beta R 2 P value N

pH 0.150 0.000 NS 23 -0.261 0.068 NS 24
Extract P (ppm) -0.449 0.164 0.032* 23 0.097 0.009 NS 24
%C-WS 0.930 0.090 NS 38 -0.237 0.056 NS 34
%N-WS 0.085 0.007 NS 38 -0.180 0.032 NS 34
Note: Beta, slope of the regression line when predictor and independent variables are standardized (indi-
cates the direction of the relationship); R 2 , proportion of variability in the dependent variable attribut-
able to the regression equation; P value, significance level, with * P < 0.05, ** P < 0.01, and *** P < 0.001.
N , total number of cases; NS, not significant; WS, whole soil; LF, light fraction.

Bray-extractable soil P were observed; however, soil P was highly variable across sites and
time, with averages across treatment and year ranging between 2 and 51 ppm ( Table 9.6 ) .
The trend was for the dambo to have the lowest levels and the hillside the highest, with
the dambo margin intermediate ( Tables 9.5 and 9.6 ) . Soil P was significantly lower in the
dambo than the other landscapes in 2000 and lower in the dambo and dambo margin than
the hillside in 2004 ( Table 9.6 ) . Low extractable P levels in the dambo ( Table 9.3 ) may be
consistent with P fixation by iron and aluminum complexes, which occurs as seasonally
flooded soils dry (PPI, 2005). Soil P may have built up on the hillside due to low maize
productivity ( Figures 9.2 and 9.3 ), the practice of creating pockets of soil between rocks for
planting maize, or related to landscape geology, but more extensive sampling is needed
to determine how robust and widespread this pattern is. Harawa et al. (2006) performed
research in a nearby location and found P levels to be lowest in the dambo and hillside
and highest in the dambo margin. Snapp et al. (1998), however, found a very high level of
variability in soil P among smallholder farms in Malawi.

9.7.2 Distributional economic analysis
Yields varied depending on the wealth status of the farmers ( Table 9.1 ) , with a strong trend
for wealthier farmers to have the highest yields in most cropping systems, and signifi-
cantly higher yields in 1999, and in the fully fertilized treatments in 2003 ( t tests P < 0.05).
This difference likely reflected the preponderance of poor farmers cultivating on the mar-
ginal hillside soils, the frequency of poorer farmers selling their labor during critical peri-
ods in the maize cropping cycle (Sirrine, Shennan, Snapp, et al., 2010; Alwang and Siegel,
1999), and differences in prior field management practices, including history of fertilizer
use. Kamanga, Waddington, et al. (2010) also found that better-resourced farmers in cen-
tral Malawi experimenting with maize-legume intercrops had higher yields than poorly
resourced farmers, attributing differences to disparate field management practices prior to
the project's inception.
Given that the wealthier farmers could afford to wait for higher prices when selling
maize (see Sirrine, Shennan, Snapp, et al., 2010b), these differences in yield resulted in
even greater disparities in profitability for each system for wealthy versus poor farmers
( Table 9.1 ) . Design 1 legume-based systems were generally more profitable than the sole
maize systems receiving equivalent fertilizer quantities, with pigeon pea systems typi-
cally the most profitable. Under D2, the legume effect could not be isolated, but the dif-
ferent legume plus half fertilizer treatments were more profitable than the unfertilized