Rwanda has a digital land resource database including a medium scale - 1:50,000 - soil map (CPR: for
Carte Pédologique du Rwanda). The availability of this land resource information was expected to
improve significantly the way agricultural research and extension were conducted in this country.
Paradoxically, research and extension programs are still planned and implemented under a kind of trial
and error approach (Multi-Environment Trials) within large and heterogeneous Agro-Ecological Zones
(AEZs) without systematic consideration of different soil types within each AEZ. Overlooking the use
of the CPR in this process has detrimental consequences on the overall reasoning of agricultural
research and extension on the one hand, and the interpretation and extrapolation of obtained research
results on the other hand. Without a broader understanding of the national biophysical environment
and the systematic consideration of different soil types at watershed level it remains an illusion to
expect from scientists, the development of soil-specific and transposable technologies in the complex
soilscapes of Rwanda! A study was undertaken to understand how the CPR can be at the heart for
Rwandan agricultural research and extension thinking towards the Science-Policy-Practice Interface
(SPPI). Findings from this study show that in a country like Rwanda where the innovation model is
intended to be the Participatory Integrated Watershed Management (PIWM), and where agriculture is
practiced by small-scale farmers (0.5-1 ha) with an already functional Farmers’ Soil Knowledge
system (FSK), the usefulness and use of the CPR can be significantly improved by integration of the
scientific and FSK systems through communication bridges. The communication bridges allow
scientists to interpret the farmers’ soil-related practice rationality and to introduce new soil-related
technologies as compatible pieces of the FSK system.
In the Akavuguto watershed case study, the link between scientific and FSK consisted of the
equivalency between scientific and farmers’ land units and scientific and farmers’ soil types. It has
been observed that soils are distributed along the slope and that top soil properties and crop yields are
more influenced by soil type intrinsic properties than by the land use. In other words, in the low input
system of Rwanda, the response to the human management factor depends first of all on the fertility
potential of each soil type. The soils of Entisols order (Urubuye /Urusenyi) occupy the mountainous
and crests/interfluves. The soils of Ultisols order (Inombe) occupy the plateaus and shoulders. The
soils of Oxisols order (Umuyugu/Mugugu) occupy the hillsides/back slopes (the largest land unit);
while the soils of Histosols order (Nyiramugengeri) occupy the valleys. Except the mountainous which
are constrained by the steep slopes (gradient >55%), soils with good soil properties (slightly acid and
less leached) are located in the upper hill made up by hill summits, crests, plateaus and shoulders. In
these land units, soils can still produce relatively good crop yields under low farmers’ input system
(farmyard manure). The infertile soils (extremely acid and strongly leached) occupy hillsides. To
produce good staple crop yields, these soils imperatively need the combination of lime, organic
manure and fertilizers. Other infertile soils (extremely acid) are found in the valley bottom. The soils
vii
in this land unit need the combination of lime and fertilizers to be productive. Thus, the integration of
CPR soil resource information with the FSK system improves the accessibility of scientific soil
knowledge and constitutes an effective way of achieving soil-specific technologies and a practical way
of extrapolating results to analogous soil types. The most important practical implication is that the
representative regional soil reference systems integrating both scientific and FSK systems at watershed
level and circumscribed in the landscape context constitutes a key step towards an –ease-to-use Land
Information System (LandIS) for Rwanda. The user friendly LandIS is necessary for the sound
management of the Rwandan space and for more rational agronomic experimentation. The main policy
implication is that the Participatory Integrated Watershed Management should be institutionalized in
agricultural research and development organizations as a valid and valuable innovation model to
which policy, administration and finance institutions should adapt.