Thèse : "Weed management using a no-till system with Stylosanthes guianensis cover crop in upland rice-based cropping systems in the Mid-West of Madagascar"

Abstract: Food production in rich countries relies strongly on intensive use of external inputs at the expense of the environment and human health, whereas in poor countries, agriculture is characterized by low use of external inputs resulting in a meager production. To tackle these issues, agroecological practices based on crop diversification is considered as a promising approach to reduce inputs, and maintain or even improve crop production. In the Mid-West of Madagascar, the rain-fed and dry seeded upland rice system has become widely adopted by farmers because of limited irrigable lands. This cropping system is strongly infested by weeds, which are hardly under control because the only weeding intervention the poor farmers can afford is the time consuming hand pull method. An innovative upland rice cropping system using Stylosanthes guianensis cover crop, managed as a living mulch in a no-till system, was lately introduced in the region. This system was proven to supply nitrogen and suppress the parasitic weed Striga asiatica but its effect on weed community was not yet studied.

A three years research was conducted from 2016 to 2018 in the Mid-West of Madagascar in order to understand if this innovative system can be a useful tool of weed management. The methodology was based on weed floristic recording on farmers’ fields, experimental fields on real farms, experimental fields on station, and surveys on farmers’ perception.

Results indicated that the variables: weeding intervention timing, crop field age, the presence of maize intercropped with rice and soil management influenced the weed community. The no-till system with stylosanthes was observed to suppress dominant weed species of conventional tilled systems namely Digitaria spp., Richardia scabra and Eleusine indica, thus reducing the total weed biomass at harvest by 67%, and the total weed cover by 47% to 93% depending on the sampling dates. The innovative system also promoted other species namely Mitracarpus hirtus, Cyperus spp. and a few occasional species. The weed suppressive effect was attributed to the stylosanthes mulch which limited weed emergence and growth at early dates of the growing season. Mineral inputs combined with manure increased weed cover and weed biomass compared to manure alone. The dominant species Digitaria spp. increased in abundance with increasing mineral fertilization. The innovative system poses no threat to rice production, instead it was observed to increase the yield in a few cases. In weed-free conditions, higher yield in the no-till system with stylosanthes might be attributed to nitrogen supply by the legume cover crop and global soil fertility improvement such as soil texture or water regime reported in the literature. In weedy conditions, higher yield in the innovative system was interpreted as the reduction of yield loss due to competition with weeds. Results indicated that in a high weed infestations context, the yield loss was 81% - 99% in the conventional system, and reduced to 33% - 54% in the innovative system as a consequence of the weed suppression effect. In a moderate infestation context, yield losses were 39% and 26% in the conventional system and the innovative system respectively. The difference between the two systems is not significant if weed infestation is low. All this indicated that the benefits from the innovative system with respect to yield is more expressed in fields with severe weed problems. Mineral inputs increased rice yield in general, but decreased the yield when it promoted weed growth. The no-till system with stylosanthes affected also the critical period of weed interference by delaying the starting date of weed-crop competition by 11 days up to 18 days. This delay was also attributed to the stylosanthes mulch which limited weed emergence and growth at early dates of the growing season. In addition, results indicated that the presence of highly competitive weed species such as Richardia scabra may prompt the beginning of weed-crop competition at early dates. Farmers’ general perception was in agreement with the results from field experiments. They also perceived that the innovative system decreased the weed infestation and posed no threat to rice yield. Instead, they sometimes found that yield increased. The main factors that slow down the adoption rate of the innovative system at farmers’ level were the fact that a few years are required for its establishment and this time may be too long for them, insufficient fields to settle the new system, investment on necessary tools such as the roller-crimper and lack of technical skills for its management.

Information provided by this research should improve the knowledge about weeds in upland rice-based cropping systems, the relation between cover crops and fertilization, and the implication of cover crops on the critical period of weed interference. The potential benefits of stylosanthes highlighted in this study can be used to address weed issues in other countries where stylosanthes can be grown.

Keywords: weed community, weed management, cover crop, living mulch, Stylosanthes guianensis, upland rice, Madagascar

Jury:
Président  : Paolo Barberi (Scuola Superiore Sant'Anna)
Rapporteurs extérieurs: Bruno Chauvel (Inra), Thomas Lebourgeois (Cirad)
Examinateur: Marco Mazzoncini (Scuola Superiore Sant'Anna)
Directrice de thèse: Ana Camilla Moonen (Scuola Superiore Sant'Anna)
Co-encadrante : Aude Ripoche (Cirad)