Objectives: This study assessed the on-farm Musa germplasm diversity across different agro-ecologies of Rwanda and the socio-economic utilization options and selection practices that create/maintain this diversity on-farm. Methodology and results: A Musa germplasm diagnostic survey was carried out in 2007 in five Rwandan districts on a transect from Lake Kivu (West) to Kirehe district (East) bordering Tanzania. Across all sites, 118 farms, each having at least 50 mats were sampled for determining Musa diversity. Forty three Musa cultivars were recorded across the five districts. Higher diversity was observed in the east declining westwards to the Lake Kivu region as reflected by the number of cultivars and their relative abundance. Nearly half of the recorded cultivars had a low diversity index (Gini-Simpson 1-D < 0.2) and therefore prone to genetic erosion. Cooking cultivars only dominate in the district of Kirehe, while beer cultivars dominate the banana production landscape in the other districts. Taste/flavor, bunch size and market demand were the most important criteria for banana cultivar selection and thus greatly influenced cultivar conservation and distribution on-farm. Diseases such as Fusarium wilt and Xanthomonas wilt greatly contributed to genetic erosion. Conclusion and application of results: Musa cultivar diversity in Rwanda is under threat. Ex-situ conservation of the menaced cultivars is of crucial importance. Beer cultivars dominated the landscape. Cultivar diversity on-farm was influenced by the prevailing altitude; taste/flavor, bunch size, and market demand of the cultivars; and their susceptibility to diseases especially Fusarium and Xanthomonas wilt. Banana breeding or adaptation strategies therefore should take into account the farmer preferred traits. In addition, strategies for managing these diseases are critical for preventing the genetic erosion of the affected cultivars.

Banana Xanthomonas wilt was reported in Rwanda in 2005. The present study was conducted to determine the distribution and incidence of the disease and farmer knowledge of disease symptoms, modes of spread and control. A survey was conducted in Rwanda in 2009-2010 in 12 major banana-growing districts of the country. One hundred and eight banana growers were interviewed using a structured questionnaire. Farmers were asked about knowledge of disease symptoms, spread, control and use of control methods. They were also asked about their sources of information on Xanthomonas wilt. Direct field observations were made of the distribution and incidence of the disease as well. The proportion of fields with Xanthomonas wilt was highest in Rutsiro (89%) and lowest in Kayonza and Ruhango (11%). The disease was not found in Gakenke, Kicukiro or Ngoma. Within-farm incidence was highest in Rutsiro (average 36%) and lowest in Kayonza (1%). The awareness of disease symptoms ranged from 53% (discoloured fruit pulp) to 84% (wilting leaves). For modes of spread, the highest proportion of farmers (73%) was aware of the role of contaminated tools while the least known mode was spread via soil and water (24%). Some 72% of famers were aware of uprooting plants as a control measure. There were large differences between awareness and use of tool disinfection and destruction of infected plants as control measures. There is a need to develop user-friendly methods of disease control. The creation of awareness in newly affected and Xanthomonas wilt-free areas is advocated. Participatory approaches are encouraged as they may reduce the gap between knowledge and adoption of control measures.

Cultivation of non-host crops after uprooting Xanthomonas campestris pv. musacearum (Xcm)-infected banana plants has been advocated for breaking Xanthomonas wilt disease (XW) cycle in fields. Knowledge on the interaction of these crops with Xcm is limited. Maize, beans and sweet potato were planted after uprooting Xcm-infected banana plants in Rwanda and eastern Democratic Republic of Congo (DR Congo). A weed fallow (mixed species) served as the control. After one, two and/or four break-crop or fallow seasons, healthy plantlets were replanted and monitored for XW for 12-24 months. XW status in adjacent fields was monitored, and diseased stems within 100-300 m radius of the two-and four-season experiments were uprooted. In Rwanda, soil and plant parts from the one-season experiments were sampled for Xcm isolation and Xcm-like colonies confirmed with Xcm-specific primers using PCR. Pathogenicity tests were performed to confirm the ability of the PCR-positive isolates to infect healthy banana plantlets. XW was observed in all the one-season experiments, with higher cumulative incidences in maize and bean plots. However, no similar trends were observed in the two-season experiments, with a 6-8% incidence observed only in bean and potato plots in DR Congo. Lengthening time under break crops to two and four seasons, respectively, reduced the incidence to 3% and zero in Rwanda and 0-8% in the two-season experiments in DR Congo. Incidence in the first-season experiments highly correlated (R = 88) to that in the adjacent fields, suggesting possible re-infections from these fields. Two season with break crop plus collective XW control are recommended in these agro-ecosystems. PCR-positive Xcm-like colonies from break crops only induced localized cell death on banana, while PCR-positive isolates from symptomatic banana plants caused full XW symptoms. Cross-infection/inoculation studies under controlled conditions are still needed to conclusively elucidate Xcm interaction with these crops. J Phytopathol 164 (2016) 659-670 Ó 2016 Blackwell Verlag GmbH 659 J Phytopathol Breaking the cycle of Xcm in banana fields A. Rutikanga et al.

Xanthomonas wilt of banana, caused by Xanthomonas campestris pv. musacearum, is a devastating bacterial disease that can cause up to 100% yield loss when appropriate control measures are not implemented. Currently, cultural practices are the only recommended means for managing Xanthomonas wilt. No cultivar is reported as resistant, nor can the disease be controlled using chemicals. On-farm experiments were established at three sites (eastern and western Rwanda, and North Kivu in Democratic Republic of Congo (DR Congo)), with the main objective of determining the most effective fallow period for eliminating Xanthomonas wilt from a highly infected banana field. At each site, three banana fields with an initial disease incidence of at least 70% were selected. All banana mats and most debris were removed before the experiment began. Thereafter, suckers that were free of Xanthomonas wilt of the cultivars 'Kamaramasenge' (Musa AAB group) and 'Injagi' (AAA-EA) in Rwanda, and 'Kamaramasenge', the plantain 'Musilongo' (AAB) and 'Vulambya' (AAA-EA) in DR Congo, were planted in experimental plots at monthly intervals, following increasingly long fallow periods, over 10 months. In Rwanda, ten plants per variety were planted each month in parallel rows in each field. In DR Congo, ten plants of each variety were randomly planted across the three experiment plots in single rows of ten plants per plot. In both countries, disease incidence was monitored for 15 months (i.e. up to 15 months after the first planting). In Rwanda, 13-15 months after planting, Xanthomonas wilt incidence in 'Injagi' was 22% for the planting in month 1, 27% for month 2 and 9% for month 3, whereas in 'Kamaramasenge' it was below 2% for the first 3 months of replanting. In 'Injagi', disease incidence declined sharply from months 4 (2.4%) and 5 (1.7%), i.e. less than 11 months from planting. Healthy suckers of the two cultivars planted from month 6 onwards did not become infected. In DR Congo, a steady decrease in Xanthomonas wilt incidence from month 1 (up to 70%) to month 10 (10%) was observed in the experimental fields. The prolonged appearance of disease symptoms (i.e. beyond replanting month 5) could be linked to the extremely high (>80%) disease incidence of, and close proximity to, diseased fields. Possible transmission of the disease into the experiment by foraging small ruminants and larger birds could have occurred. In contrast, diseased mats were continuously uprooted in adjacent farmers' fields in Rwanda. Results suggest that under Rwandan conditions, the bacterium is likely to survive for up to 5 months in soil and/or remaining plant debris. Therefore, it is proposed that replanting of previously Xanthomonas wilt-infected fields should be carried out 6 months after thorough uprooting of bananas infected with X. c. pv. musacearum. The DR Congo study also indicates that there is need for rigorous application of preventive control measures in adjacent farms to avoid possible transmission by foraging animals, large birds or runoff water. Hence, in both Rwanda and eastern DR Congo, disease management efforts should be carried out by farmers in a concerted manner.

Xanthomonas campestris pv. musacearum is a bacterium that causes Xanthomonas wilt of banana. It mainly spreads through contaminated garden tools, infected planting materials and insect vectors. Following infection, the bacterium spreads systemically throughout the plant, including the lateral shoots. Cultural practices are the only available management option for controlling the disease. Researcher-led and farmerimplemented experiments were established in Rwanda and eastern Democratic Republic of Congo (DR Congo) to assess the effectiveness of selective uprooting of diseased mats combined with early male bud removal and the prohibition of the use of garden tools for routine field maintenance. Three farmers' fields, corresponding to initial Xanthomonas wilt incidence levels of 14-29%, 30-35% and 42-45% (proportion of mats with symptoms), were selected in each of four sectors in Rubavu District, western Rwanda, while in North Kivu, eastern DR Congo, two fields were selected with disease levels of 14-29% and 42-45%. Banana plants were monitored on a monthly basis for disease symptoms in both treated and adjacent control plots (with no management of Xanthomonas wilt) for 10 months. A separate experiment in South Kivu, DR Congo, compared the effect of removal of the whole mat with removal of single plants on the expression of the disease. In Rwanda, seven out of 12 farmers attempted to rigorously apply the recommended control options and this increased the proportion of visibly healthy plants fivefold (from 7% to 35%). In DR Congo, when the initial disease level was 14-29%, after 10 months of treatment 57% of plants were visibly healthy compared with only 15% in non-treated plots. At an initial disease level of 42-45%, 32% of plants were visibly healthy after 10 months of treatment compared with only 9% in non-treated plots A significantly higher number (57%) of visibly healthy plants was recorded under conditions of low initial disease incidence (14-29%) in DR Congo, while in Rwanda a higher number of asymptomatic plants occurred in plots where farmers attempted to adhere to the recommended management options for Xanthomonas wilt. Although selective uprooting under conditions of low initial infection, in combination with the rigorous application of management options that reduce disease transmission, resulted in a higher number of visibly healthy plants (35% in Rwanda and 57% in eastern DR Congo), farmers reported that new infections were still observed after the experimental period. We are thus inclined to recommend the complete removal of diseased fields and subsequent fallowing/establishment of break crops when disease incidence is above 14%. The experiment in South Kivu indicated that complete mat or single plant removal does not lead to the quick removal of the disease from a field in a small-scale farming context.