Introduction

Cassava (Manihot esculenta Crantz) is a staple food for approximately 500–800 million people living in developing countries (Bull et al. 2011; Howeler et al. 2013) and worldwide it is second only to maize (Zea mays L) for production of starch (Howeler et al. 2013). In the developing world, cassava is amongst the top four most important crops in terms of production after rice (Oryza sativa L), maize and wheat (Triticum spp.). The potential yield of cassava is estimated at 90 t ha−1 of fresh roots under well managed conditions (El-Sharkawy 2004). Cassava plays a key role as a food security and income-generating crop for many smallholder farmers in developing countries (Ceballos et al. 2004; El-Sharkawy 2004; Tumuhimbise 2013). In East Africa, cassava is eaten after boiling and processing to flour to make porridge, local brew, ugali and bread, and sweet varieties lacking cyanogenic glycosides can be eaten raw (Kamau 2006; Mkumbira et al. 2003; Were 2011). In Rwanda, cassava is an important staple food and is currently being promoted as a cash crop through the establishment of cassava processing plants. In addition to its tuberous root, its leaves are treated as a vegetable called “Isombe”. Cassava is consumed in various forms (raw, paste/bread or ugali, boiled for breakfast, mixed with beans, vegetables, etc) and its cooking and preparation methods vary from one individual to another (mixed with beans, boiled, paste or ugali, etc.). It occupies the third place after banana and sweet potato for reducing hunger and poverty in the country (FAOSTAT 2011; Night et al. 2011). Cassava as a cash crop can be used in industries for production of animal feed and starch for use in pharmaceuticals, textiles and more (Ceballos et al. 2004; El-Sharkawy 2004).

Although cassava is a major food crop, its production is threatened by lack of good cultivars (early bulking, high yielding, and disease resistant), low soil fertility, poor agronomic practices and postharvest losses. Postharvest losses are mainly linked to the short shelf-life of cassava storage roots due to physiological postharvest deterioration (PPD), which presents major challenges to its production and utilization (Ceballos et al. 2004). The PPD begins within 24 h of harvest and rapidly renders the storage roots unpalatable, inedible and with a reduced market value (Sánchez et al. 2006). With poor road infrastructure and remote production sites, the short shelf-life severely limits marketing options and increases the likelihood of product losses and higher marketing costs. Physical methods such as underground storage, use of fungicides, pruning plants before harvest and cold storage (2-4 °C) can be used to limit PPD (Ravi et al. 1996). However, these techniques are ineffective and impractical because they are too expensive and complicated when handling large volumes of harvested roots (Sánchez et al. 2013). Subsistence farmers need to store food after harvest for home consumption and, due to limited land size, cassava cannot be stored underground (in the field) for long periods, thus there is a need for a technique that can extend shelf life of cassava for at least some weeks and preferably to several months. Conventional breeding and genetic engineering were suggested as long term strategies to delay the onset of PPD (Salcedo and Siritunga 2011). Recently, studies elsewhere have reported the development of nutritious carotenoid (pro-vitamin A with yellow/orange fleshed cassava) cassava genotypes; these genotypes due to the antioxidant property of carotenoids may retard or inhibit the onset of PPD (Morante et al. 2010; Sánchez et al. 2006; Zidenga et al. 2012). However, the hindrance factors for adoption of yellow fleshed cassava are still unclear in Rwanda.

Moreover, the high population pressure in Rwanda, resulting in small farm land size, can be a constraint in agricultural development by reducing the adoption of a long season or perennial crops such as cassava due to complications in cropping systems (rotation and intercropping practices) (Howeler et al. 2013). Farming systems and farmers preferences vary from country to country and from one culture to another. In Rwanda, cassava is grown as a subsistence crop in ten out of twelve agro-ecological zones and the main farming system is intercropping. The type of farming system and cassava variety preferences depend on the agro-socio environment such as farm size, climate and crop utilization (Were 2011). Thus, cassava is grown as a monoculture on large commercial farms and in intercropping systems on small land holdings for subsistence. Mbwika and Mayala (2001) reported that 46.9 % and 15.0 % of cassava was grown in intercropping and monocroping systems, respectively, while 38.1 % was grown in mixed cropping system in the country. Cassava was mainly intercropped with maize, beans, banana, and occasionally with groundnuts or sweet potato (Mbwika and Mayala 2001) and vegetable crops. Farmers preferred varieties based on traits such as yield, dry matter content, taste, and early maturing or early bulking. Cassava early bulking means shortened growth periods (Tumuhimbise et al. 2014) for accumulating starch and other yield components. Maximum cassava yields are obtained at 12–15 months after planting (Hillocks and Jennings 2003). However, the minimum growing cycle of cassava in Rwanda is around 10 months for the early bulking cultivars (personnel observation). The organoleptic qualities (taste and texture) and ability to cook quickly are important traits of cassava cultivars grown for food in most cassava growing countries. For instance, sweet cultivars are grown for raw consumption or after boiling, while bitter cultivars must undergo processing to reduce the cyanide content.

Despite the popularity and importance of cassava, there is lack of an operational breeding scheme in Rwanda. Thus, the cultivars grown are mostly non-adapted exotics (introduced from other countries) and a few landraces susceptible to the most devastating viral diseases (cassava brown streak disease (CBSD) and cassava mosaic disease (CMD)) in the region. Thus, understanding the adoption hindrance factors and appropriate varietal selection processes could improve cassava yield in the country. The national cassava programme invests much effort in evaluating the adaptability of introduced germplasm and focuses on yield and disease resistance as the main traits, which do not necessarily match all of the critical preferences of farmers. Although, there are some improved cultivars introduced from other countries, their adoption level is low and factors affecting their adoption are still unclear. In addition, relying on varietal introduction, participatory varietal selection is limited in Rwanda. On the other hand, the current research conducted in East Africa shows that the limited involvement of end-users in the formal breeding process affected the level of adoption of new cultivars negatively (Kamau et al. 2011; Tumuhimbise 2013; Were et al. 2014). Many breeding programmes in developing countries fail due to the lack of inclusion of participatory approaches which negatively affects the level of adoption of newly developed cultivars (Kamau et al. 2011; Were et al. 2012). Tumuhimbise (2013) reported that taste, cooking qualities and earliness are just a few of the dozens of crop traits of interest to smallholder farmers. Were et al. (2012) also reported that farmers have indigenous knowledge that could be of value to cassava improvement processes and improve the adoption of newly developed cultivars. In some countries, farmers are conservative and the understanding of information on factors affecting the adoption of new genotypes could be important in enhancing their adoption.

Participatory plant breeding (PPB) utilizes farmers’ skills in identification and selection of their preferred traits, breaks the barrier between farmers and breeders, reduces the gap between variety development and adoption and enhances availability of planting materials to farmers (Kamau et al. 2011; Kanbar and Shashidhar 2011; Smith et al. 2001). PPB is convenient for the adoption of new varieties because farmers participate in selection of parents and offsprings. When farmers and breeders select together the parents and new genotypes, the breeding programmes will be more efficient and effective (Ceccarelli 2006). Participatory plant breeding utilizes many approaches such as surveys and focus group discussions in so called “participatory rural appraisal” (PRA) approaches. The PRA relies heavily on participation by the communities. This method is designed to enable local people to be involved, not only as sources of information, but as partners in gathering and analyzing the information. These two approaches provide vital information on what is needed by farmers (Kamau et al. 2011; Were et al. 2012). The involvement of farmers at some breeding stages could change their conservative behaviour and promote the adoption of new genotypes containing preferred traits. Thus, the reorientation of cassava breeding and adoption of decentralised participatory breeding (Were et al. 2014) could enhance the adoption of new cultivars. Participatory rural appraisal (PRA) that is aimed at identifying gaps in cassava production could help to build a strong foundation for a cassava breeding scheme in Rwanda.

Materials and methods

Study sites

This study was conducted in three major cassava growing districts in Rwanda, namely Bugesera, located in East Province, Kamonyi, located in South Province, and Gakenke, located in Northern Province (Fig. 1). Geographically, Bugesera lies at 02°12′18″S 30°08′42″E, Kamonyi, at 2°0′0″S, 29°54′0″E and Gakenke at 1°42′0″S, 29°47′0″E (Fig. 1).

Fig. 1
figure 1

Map of Rwanda showing the study areas

Full size image

The selection of these districts was based on cassava production levels which could be influenced by different factors including altitude, temperature, rainfall and type of soil (Table 1).

Table 1 Description of study area
Full size table

Data collection

In order to assess cassava production constraints, preferred traits, factors affecting the adoption of new cassava cultivars, cassava market aspects and losses due to physiological postharvest deterioration (PPD), data were collected through focus group discussion (FGD) and farmers’ interviews. To facilitate data collection, FGD checklist and questionnaires were translated into the local language (Kinyarwanda). Three FGDs were conducted in each district, making nine for the whole study. Each FGD was composed of ten participants, namely one district agronomist, one district extension officer, one cassava extension specialist from the Rwanda Agriculture Board (RAB), and seven farmer representatives from different groups of farmer field schools (FFSs). Some participants, mainly district and RAB staff, participated in all FGDs within a district. Gender was balanced, with women being two thirds of the group, because they are much more involved in agriculture activities than men in the country.

Semi-structured questionnaires were developed to be administrated to cassava farmers, cassava traders and processors. Sampling was done within FFSs at district level where a total of 60 cassava farmers from each district were selected. Multistage sampling was performed, based on cassava growing areas and three districts were chosen. A random sampling was done within FFSs where ten FFSs were selected from each district. Another random sampling was performed at household level, whereby six household were selected from each FFS to participate in interviews. This makes a total sampling size of 60 participants per district and 180 participants in total.

Data analysis

Pair-wise ranking matrix and scoring matrix (Catley et al. 2007) were used to compute the data from PRA. Data for land size, land allocated to cassava, losses due to PPD, time to harvest and cassava yield were analysed using Post hoc test ANOVA (Hilton and Armstrong 2006). Other social data collected were screened and coded to be analysed using the Statistical Package for Social Sciences (SPSS), 16th version. Percentages, means and cross tabulations are presented in the following results section.

Results

Cassava farming system in Rwanda

Main food crops grown in Rwanda

Cassava, bean (Phaseolus vulgaris L), sweet potato (Ipomoea batatas L) and maize are the major crops grown in the study areas. In Bugesera and Kamonyi districts, cassava ranked first with 90.0 % and 75.0 % of respondents, respectively, while in Gakenke district, it occupies the fifth place after banana (Musa spp. L), bean, maize and sweet potato. Maize occupies the second place in Bugesera district, while bean occupies the same place in Kamonyi and Gakenke districts (Table 2). The places of other crops vary from district to district. Rice, coffee (Coffea arabica L), pineapple (Ananas comosus L) and sunflower (Helianthus annuus L) were ranked as minor crops in all districts.

Table 2 Rank of cassava and other main crops in three districts of Rwanda
Full size table

Land size and cassava production yield

The land allocated to cassava differed significantly from district to district (p < 0.001), and in general was greater than the land allocated to other crops. On average the total land was 0.69 ha per household while the average land allocated to cassava was 0.29 ha. The majority of farmers had a total land size less than one ha and hence practised intercropping systems (Table 3). Legumes (beans and soybeans) were the most common crops intercropped with cassava in the study area. However, some farmers mixed cassava with maize, pineapple, shrub crops and young trees such as eucalyptus. The farmers with land size larger than one ha, grow cassava in monoculture (Table 3).

Table 3 Land size and cassava cropping system in Rwanda
Full size table

The yield of cassava significantly varied significantly (p = 0.045) from district to district. Focus group discussions reported yields of 24.5, 24.2 and 16.7 t ha−1 in Bugesera, Kamonyi and Gakenke districts, respectively (Table 3).

Growing cycle, causes and effects of late bulking cultivars

Time to harvest varied significantly (p < 0.001) between districts and ranged from 6 to 24 months, but 16 months was the average bulking time for all districts. The early bulking cultivars (6–8 months) were reported in Bugesera and Kamonyi districts while in Gakenke district, the early bulking was at 12 months (Table 4).

Table 4 Time to harvest per district
Full size table

The first cause of late harvest was late bulking cultivars (the late bulking inherent in the cassava cultivars) as indicated by the majority of farmers in the study area. Agricultural practices and cold environment were commonly reported as the second causes of late bulking. Farmers from Bugesera district suspected that planting cassava at the wrong time (out of season) of the year is the cause of late maturity (Table 4) and affected yield. In Rwanda, the best planting time is September-October of each year.

Food crises (shortage of food, malnutrition and prolonged hunger), lack of cuttings, losses of other crops’ yields and delayed return on investments were commonly reported as the main effects of late bulking cassava cultivars. The other effects indicated by farmers in Bugesera and Kamonyi districts were poverty and crop rotation impediment (Table 5).

Table 5 Farmer perceptions on causes and effects of late bulking cultivars
Full size table

Clean cuttings availability

The majority of farmers (average of 66.6 % in all districts) confirmed that the availability of clean cuttings was a problem. The sources of clean cuttings were research institutes, farmers groups, cooperatives and NGOs (Fig. 2). Farmers lacked a source of clean cuttings from neighbouring farmers and their own preceding cassava crops (Fig. 2). However, farmers were unsure of the health status of the cuttings available.

Fig. 2
figure 2

Availability of clean cuttings

Full size image

Cassava production constraints in Rwanda

The majority of farmers (99.4 %) confirmed the presence of cassava production constraints. Lack of clean cuttings, late bulking cassava cultivars and diseases, especially Cassava Brown Streak Disease (CBSD) and Cassava Moaic Disease (CMD), were the main constraints of cassava production. Drought was the second most challenging constraint in Bugesera district while the lack of clean cuttings ranked the same in Kamonyi and Gakenke districts (Table 6). The other major challenging constraints per rank order were weathered soils, insufficient fertilizers, small land size, limited information (access to information), and lack of market, storage of fresh and dried cassava, theft and animal damage and the agricultural policy of crop regionalization. Several minor cassava production constraints such as cold environment, heavy rainfall and storms, limited knowledge on cassava cropping s were identified by farmers and varied from district to district (Table 6).

Table 6 Constraints to cassava production per district
Full size table

Cassava physiological postharvest deterioration

Fresh cassava storage constraints and storage techniques

A total of 96.7, 98.3 and 70.7 % of farmers in Bugesera, Kamonyi and Gakenke districts, respectively, confirmed the lack of effective storage technique (Fig. 3).

Fig. 3
figure 3

Effective cassava storage constraints

Full size image

The traditional technique of storage of cassava as flour (drying and processing cassava storage roots to flour) was the most widely used technique to deal with PPD, as confirmed by the majority of farmers (89.2 %) across the districts. Piecemeal harvesting (gradual harvest) can conserve cassava storage roots more than one year, as indicated by 71.9 % of farmers interviewed in the study area. However, once harvested, underground storage of cassava roots in the soil (interment) conserves the roots for only four days on average, as confirmed by 53.8 % of farmers (Table 7). Less common techniques highlighted by farmers in one district (Kamonyi) were storage of peeled cassava in water and precooking cassava. However, the latter cannot conserve cassava storage roots for more than three days.

Table 7 Traditional storage techniques of cassava fresh storage roots in Rwanda
Full size table

Local methods for detecting PPD in Rwanda

The colour change of cassava flesh was the most common method for measuring PPD as reported by 89.8, 80.7 and 86.4 % of farmers interviewed in Bugesera, Kamonyi and Gakenke districts, respectively. Rotting of cassava flesh and changing taste also were classified as the second and third most common methods to assess damage due to PPD (Table 8). However, FGDs indicated that difficulty of removing skin (peeling) of cassava storage roots and the increase of fibres in cassava flesh were among the local methods of detecting PPD.

Table 8 Local methods to detect PPD damage
Full size table

Losses due to PPD and value given to damaged cassava storage roots

There was no significant difference (p = 0.259) in cassava production losses due to PPD in the three study areas. Losses of 12.6, 10.3 and 13.3 % were reported in Bugesera, Kamonyi and Gakenke districts, respectively. PPD appeared at approximately three days after harvest (Table 9).

Table 9 PPD appearance after harvest and PPD losses in Rwanda 2014
Full size table

In cases of total deterioration due to PPD, the deteriorated cassava is processed into flour, given to poor families (low social class), or used to feed animals (especially pigs). In Kamonyi and Gakenke some farmers indicated that deteriorated cassava was processed into flour for constructing their houses (painting and mixed with cements). However, 44.1, 41.0 and 14.5 % of farmers in Bugesera, Gakenke and Kamonyi districts, respectively reported that damaged cassava was considered as garbage (Table 10).

Table 10 Place and value given to deteriorated cassava due to PPD
Full size table

Availability of PPD tolerant cassava cultivars in Rwanda

Based on the fact that PPD signs start appearing after 24 h, farmers indicated that some cultivars showed PPD signs only after three days, an indication of PPD tolerance. The common popular cultivars with delayed PPD were Rwizihiza, Mavoka, Cyizere, Seruruseke and Mbakungahaze (Table 11). Most of these cassava cultivars are introduced (assumed to have high yield, disease resistance and possibly improved dry matter, carotenoid content and other valuable traits). Some landraces (Gahene, Nyiramabuye, Rutanihisha, Yangwe, Rwicabana and Gapfutsi) were also reported to tolerate PPD; these landraces are bitter types with high cyanide content which could delay microbial attacks. The only sweet landraces tolerant to PPD were Gacyalicyali and Mushedile. However, none of the cultivars tolerate PPD beyond three days under normal conditions of storage (room temperature) as indicated by farmers. This shows that there is no cultivar in the country tolerant to PPD, compared to cultivars in other countries which have the ability to withstand PPD for one to several weeks.

Table 11 Some cassava cultivars in which symptoms of PPD were delayed
Full size table

Farmer preferred cassava traits in Rwanda

The adoption of newly introduced cassava cultivars must respond to farmers and consumers preferences. In order of importance, the preferred cassava traits were sweet taste, high yield, good quality ugali (viscosity and colour), resistance to pest and diseases, early bulking, multipurpose, good colour of flesh and flour, many clients, delayed PPD, dry matter content, good odour/smell, long storage in the field, many cuttings produced and good cookability (Table 12).

Table 12 Consumers and farmers traits preference
Full size table

Farmer perceptions on yellow fleshed cassava

All farmers confirmed the availability of two yellow cassava cultivars. However, the majority of them (95, 55 and 52.3 % in Bugesera, Kamonyi and Gakenke districts, respectively) dislike yellow fleshed cassava. They highlighted some reasons for the unpopularity of yellow fleshed cassava such as a drying problem (low dry matter), bad colour of flour, lack of taste, rapid rotting, carotene odour from volatile carotenoid compounds, low demand (few clients), poor storage in the field (spoiled when kept in the field) and not cooking well. However, 33 % of respondents liked yellow fleshed cassava cultivars for its early bulking, resistance to CMD, high yield, good ugali, multipurpose (eaten raw or processed into flour for ugali), vitamin A content and cookability (Table 13).

Table 13 Farmer perceptions on yellow fleshed cassava
Full size table

Factors influencing adoption of new cassava genotypes

Focus group discussions revealed some factors affecting the adoption of new cassava cultivars. Pair wise ranking listed the factors in descending order whereby farmer preferences, information and extension services, performance (yield, early bulking, disease resistant), quality (cooking aspect, taste, dry matter, viscosity of ugali, colour of ugali), market acceptability and stake production (Table 14) proved to be the main factors influencing farmers to adopt or reject the new introduced cassava genotypes in Rwanda.

Table 14 Pair wise ranking of factors affecting adoption of new cassava cultivar in three districts of Rwanda
Full size table

Discussion and conclusion

This study aimed at identifying the main constraints of cassava production, farmer preferred traits, effects of late bulking cultivars, losses due to PPD, and factors affecting the adoption of new genotypes. An understanding of these aspects in cassava farming system provides guidelines for breeding objectives for the cassava breeding programme in Rwanda.

The study found that cassava is among the main food crops in the country, but its place varies from district to district. This agrees with Stephen and Lecumberri (2011), who reported that cassava, beans, maize, banana and sweetpotato are the main foods crop grown in Rwanda. The findings showed that the average farm size was 0.69 ha with 0.29 ha allocated to cassava per household in average. However, many farmers possess less than the average farm size (0.6 ha). This is corroborated by Rurangwa (2013), who reported that the majority of households had less than 0.2 ha in Rwanda. Land allocated to cassava was greater than that for other crops and the predominance of intercropping was attributed to the small size of land available, a result of the dense population of the country (407 people/km2 according to Rwanda’s National Institute of Statistics in 2012), the greatest in SSA (Rurangwa 2013). Land in the study area was fragmented, necessitating subsistence farming. Subsistence farming was associated with the large spacing (1 × 1 m/10,000 plants per ha) required for cassava and the need for food diversification. Households with small plots tended to practise intercropping in contrast to those with larger plots who mostly practised monoculture. This is corroborated by Mbwika and Mayala (2001) who found that the majority of cassava farmers in Rwanda practised mixed cropping. Many scientists (El-Sharkawy 2004; Munga 2008; Were et al. 2012) report that cassava is grown by small scale farmers in intercropping and mixed cropping systems in developing countries.

The results revealed that the lack of clean cuttings, pests and diseases (especially CBSD and CMD) and late bulking cultivars were the main constraints of cassava production in the study area. These findings are in agreement with Tumuhimbise (2013), who showed that virus diseases such as CBSD and CMD are the most challenging constraint in Uganda followed by lack of early bulking cultivars. Kamau (2006) also reported that the lack of early bulking cultivars is a challenging constraint of cassava production in the East Africa region.

The majority of farmers confirmed the lack of effective storage techniques. This was linked to physiological postharvest deterioration (PPD) which starts a few hours after harvest, with most cassava production marketed as fresh roots for consumption as fresh boiled cassava. Fresh cassava roots for raw and boiled consumption, and for processing need to be free from any deterioration (PPD and bacterial rots). PPD differs from bacterial rots in that there is a change in colour of the flesh due to physiological activity. This begins within 24 h of harvest and starts as blue-black to black vascular discoloration (vascular streaking) and then spreads to the parenchyma, thereby rendering the storage root unpalatable (flavour, odour and colour) and unmarketable (Morante et al. 2010; Reilly et al. 2007; Sánchez et al. 2006). Ceballos et al. (2004) also reported that the short shelf-life of cassava roots, due to physiological postharvest deterioration, presents major challenges to increasing production and utilization in developing countries. Tackling the effects of PPD needs much effort; the private sector and government must invest in infrastructures and cassava processing plants in order to reduce the time between harvest and marketing and the initiation of processing activities.

The traditional measure to tackle losses due to PPD in the study area were to keep cassava storage roots in the field with gradual harvesting (piecemeal) and underground storage of harvested cassava roots, which conserves them for at least four days. The former can conserve cassava storage roots for approximately one year, but it can also compromise agriculture practices (rotation) because it occupies land for longer periods. Sayre (2011) reported that harvesting cassava can be delayed by months or even up to three years. On the other hand, cassava storage roots kept in the field for long periods can become woody and quality and flavour may be affected.

Traditional assessment of PPD by farmers included change in colour and taste, rotting, difficulty in removing skin (peeling) and increased fibre. The difficulty in skin peeling could be attributed to the increase of fibres which may be a defence mechanism of cassava after wounding against subsequent bacterial attack (Kpémoua et al. 1996; Luna et al. 2011; Uarrota et al. 2014).

The study found on average 11.9 % of cassava production losses due to PPD. FAO (2000) reported total postharvest losses of 29, 10 and 8 % of cassava in Africa, Latin America and Asia, respectively. In Rwanda, farmers considered cassava totally deteriorated approximately 3 days after harvest. This may be the result of bacterial attack which can cause total rotting of cassava flesh, leading to a total financial loss. For non-bacterial PPD farmers minimize financial loss by feeding the affected cassava to animals (especially pigs) (Ubalua 2007), selling it at low price as food for poor families (low social class) and processing the flour for use in cements and house paints. Okafor (2008) reported that cassava flour performs satisfactorily as a water reducing admixture in concrete. Cassava starch prevents floor cracks by stopping the formation of calcium silicate hydrates (CSH) in concrete mixtures (Abalaka 2012).

The significant differences in harvesting time observed in the districts was attributed to gene x environment interactions and ecological differences among districts as the majority of farmers reported that bulking time is an inherent characteristic of the cultivar and the growing environment. Food crises (lack of some important foods at specific times and locations leading to malnutrition) and lack of other crop yields were linked to the late bulking cultivars occupying land for long periods of time. These findings agree with Tumuhimbise (2013) and Kamau et al. (2011) who reported that the late bulking cultivars occupy land for extended periods and consequently the land cannot be effectively utilised for sequential cultivation of other crops. Additionally, Zidenga et al. (2012) reported that, in a semi-commercial setting, the land may need to be released for other uses. The lack of cuttings was also perceived as the result of late bulking cultivars.

Sweet taste, high yield, good ugali (good quality: taste, colour and ugali viscosity), resistance to diseases, early bulking and multipurpose uses were the most preferred traits of farmers. These traits, except the multipurpose trait, were reported by Tumuhimbise (2013) in Uganda, where farmers selected cultivars largely focused on high fresh root yield, early bulking, resistance to pests and diseases, and sweetness. The needs of farmers, the behaviour of consumers, the priorities of farmers, the production environment, the available transformation technologies and the farming systems should be some of the factors that dictate cassava selection.

The introduction and breeding of yellow fleshed cassava cultivar could be an effective way to reduce postharvest losses due to PPD. Many scientists (Bayoumi et al. 2010; Sánchez et al. 2006; Sánchez et al. 2013; Xu et al. 2013; Zidenga et al. 2012), reported carotenoids could delay the onset of PPD owing to their antioxidant properties (Morante et al. 2010; Zidenga et al. 2012). Despite the availability of some yellow fleshed cassava in the country, their adoption is hampered by the yellow/orange colour being linked to traits such as low dry matter and carotene odour when consumed fresh or boiled. However, the carotene odour disappears when improved carotenoid cultivars are processed into flour to produce cassava paste or ugali. This odour could originate from volatile compounds derived from boiling (heating) cassava storage roots with high carotenoid content. Rios et al. (2008) and Zepka et al. (2014) reported that oxidation and thermal treatment cause degradation of carotenoids which influences the aroma and flavour of products containing them. This explains the non-adoption of yellow cultivars with high β-carotene content. Salcedo and Siritunga (2011) reported that several agronomic traits such as starch and dry matter, are negatively correlated with PPD and Wenham (1995) also confirmed that delayed PPD was a associated with reduced dry matter.

In conclusion, this study revealed the main constraints of cassava production and losses due to PPD, farmers preferred traits and factors affecting adoption of new cultivars in Rwanda. Lack of clean cuttings, pest and diseases (CBSD and CMD) and late bulking cultivars, associated with 11.9 % losses due to PPD, handicap the cassava sector in the country. The newly introduced yellow cultivars, which are resistant to CMD, have early bulking (8 months) and high yield could be an option for reversing the main constraints reported by farmers. However, the lack of a local participatory breeding program has affected the adoption of introduced cultivars. Participatory plant breeding, involving farmers at some stages of the breeding process, can promote the ownership of the developed cultivars and consequently enhance their adoption. The development and adoption of improved carotene content for cassava paste (ugali), earliness, disease resistance and delayed PPD cultivars are expected to promote the cassava sector and improve the livelihoods of cassava growers in Rwanda.