Soil Fertility as a Contingent Rather than Inherent Characteristic - Principles of Sustainable Soil Management in Agroecosystems - page 146

The second set of factorial trials at Anjomakely ( N = 240) held the rice variety
constant (all used a traditional cultivar, riz rouge ) and compared the effects of man-
agement practices on soils of contrasting quality; half were done on better (clay) soil
versus half on poorer (loam) soil. The climate was temperate, and the elevation was
1200 masl, so growing conditions were quite different from the first trials.
On these better and poorer soils, the respective yields with conventional practices
were 3.00 and 2.04 Mg ha -1 , while with SRI methods, paddy yields were more than
tripled on both soil types, 10.35 and 6.39 Mg ha -1 . Details on both sets of factorial
trials are given in Uphoff and Randriamiharisoa (2002) and Randriamiharisoa and
Uphoff (2002). In both sets of trials, the patterns of yield response as well as other
parameters measured (numbers of tillers and panicles, panicle length, root length,
and root density) were very similar, even though climatic and soil parameters at the
sites were quite different.
The advisor for these studies, Prof. Robert Randriamiharisoa, director of research
for the University of Antananarivo's Faculty of Agriculture, suspected from the first
trials that there could be some soil biology influences involved. Accordingly, the
research in the second trials examined the populations of the N 2 -ixing bacterium
Azospirillum associated with the roots of the rice plants grown under the different
treatments, as part of the evaluation of interaction among six factors using random
block design (Andriankaja 2001).
The data in Table 6.1 are averages for six replications of each of four treatments
on the same clay soil: (1) conventional practice: 20-day-old seedlings, 3 per hill,
continuous flooding without any soil amendments; and three versions of SRI: (2) SRI
practices with 8-day-old seedlings, one per hill, with intermittent irrigation, and
without any soil amendments; (3) SRI practices as in (2) but with fertilizer amend-
ments (NPK 11-11-16 at 300 kg ha -1 ); and (4) SRI practices as in (2) with compost
additions (5 Mg ha -1 ) rather than NPK. The compost was made from cow dung, rice
straw, and the legumes tephrosia and crotalaria, and had become stabilized before
use. These treatments resulted in very different numbers of tillers and in different
yield levels as seen in Table 6.1.
That grain yields correlated strongly with the number of grain-bearing tillers
per plant was expected. The almost sixfold production increase in yield associated
with SRI practices when these were combined with compost applications, compared
to conventional practice without soil amendments, was unanticipated. Such results
could have been dismissed as implausible if similar increases had not been seen on
some farmers' fields in the Ranomafana area.
The most interesting results from the analysis were the numbers of Azospirillum
associated with root tissue taken from plants in the respective plots and analyzed
at Institut Pasteur in Antananarivo. Because the roots were not surface-sterilized
before analysis, the organisms counted could have been endophytes or epiphytes
or residing in neighboring rhizosphere soil. The research was exploratory to see
whether SRI practices had an effect on microbial populations in, on, and around the
roots, not designed for hypothesis testing. Thus, the results were regarded as indica-
tive of SRI practice effects on the population dynamics of beneficial microorganisms
associated with rice, recognizing that more conclusive research on this subject would
have to be done.