Cassava Breeding: Current Status, Bottlenecks and the Potential of Biotechnology Tools
- First Online:
- Cite this article as:
- Ceballos, H., Kulakow, P. & Hershey, C. Tropical Plant Biol. (2012) 5: 73. doi:10.1007/s12042-012-9094-9
- 374 Views
Cassava is an important energy source in the diets of millions of people in tropical and subtropical regions of the world. It is a key subsistence crop, and its industrial uses are steadily growing. In spite of its economic and social relevance, relatively little investment has been made for research on cassava. However, conventional breeding resulted in more stable production through enhanced tolerance to biotic and abiotic stresses; increased productivity, both in fresh root production and increased dry matter content; and, more recently, improvements in qualitative traits such as starch quality and increased carotenoids content. The inbreeding of cassava has been identified as a key step for more efficient genetic improvement of the crop, therefore, research is underway to develop protocol(s) for the production of doubled haploids. Marker-assisted selection has been successfully applied to cassava, but in a more modest scale compared with other crops. More support and emphasis is needed on practical applications of molecular marker technology in cassava improvement. The availability of more efficient genotyping approaches and the cassava genome sequence promise to increase the impact of biotechnology tools on cassava improvement. Efficient and reliable phenotyping of cassava remains a challenging goal to achieve in the near future.
KeywordsPhenotypic recurrent selectionHeterosisGenetic variabilityBiotic and abiotic stressesIndustrial uses
Advanced yield trial
Cassava Brown Streak Disease
clonal evaluation trial
Cassava Mosaic Disease
dry matter content
Latin America and the Caribbean
post-harvest physiological deterioration
preliminary yield trial
quantitative trait loci
Targeted Induced Local Lesions in Genome
The center of origin of cultivated cassava is hypothesized to be in South America (Olsen and Schaal 2001; Allem 2002; Nassar and Ortiz 2008), although several questions about its domestication remain unanswered. Nassar (1978) proposed that domestication of cassava occurred from a natural hybrid between M. pilosa and other species. On the contrary, Olsen and Schaal (1999) and Léotard et al. (2009) suggested that cultivated cassava emerged from populations of M. esculenta ssp. flabellifolia (Pohl) Ciferri.
Although cassava is frequently considered a polyploid species (Westwood 1990), analyses conducted during diakinesis and metaphase I consistently indicated the presence of 18 small and similar pairs of associated homologous chromosomes, or bivalents (Hahn et al. 1990; Wang et al. 2011). In some cases occurrence of univalents/trivalents and late bivalent pairing has been observed. Cassava is therefore a functional diploid (2n = 2x = 36; Jennings 1963; Westwood 1990; De Carvalho and Guerra 2002; Nassar and Ortiz 2008). Magoon et al. (1969) suggested that certain portions of the genome are duplicated, indicating that cassava may be a segmental allotetraploid.
The culture and characteristics of cassava must be taken into consideration in order to develop an effective breeding program. Cassava is monoecious, with female flowers opening 10–14 days before the male ones on the same branch. Self-pollination occurs when male and female flowers on different branches or on different plants of the same genotype open simultaneously (Jennings and Iglesias 2002). Flowering time depends on the genotype and the environmental conditions. Branching occurs when an inflorescence is formed. Given that erect, non-branching cultivars are frequently preferred by farmers, crossing elite clones may become difficult due to the scarcity of flowers. Synchronization of flowering remains a difficult issue in cassava breeding. Some clones flower relatively early, 4 or 5 months after planting, whereas others flower 8–10 months after planting. Because of this, and the time required for the seed to mature, it takes at least a year to obtain seeds of a planned cross. For each pollination of a flower, one or two seeds usually develop, out of the three possible formed in the trilocular fruit. Several publications illustrate the procedures for controlled pollinations in cassava (Jennings and Iglesias 2002; Kawano 1980).
Cassava can be propagated either by stem cuttings or by seeds resulting from a sexual cross, with cuttings the most common practice used by farmers for multiplication and planting purposes. The root is not a reproductive organ. Propagation from true seed occurs occasionally in farmers’ fields. Seed is a starting point for the generation of useful genetic diversity (Alves 2002) and has occasionally been used in commercial propagation schemes (Iglesias and Hershey 1994; Rajendran et al. 2000).
At harvest, farmers cut the young branches and discard them. Before harvesting the roots, the main stems are cut and tied together in ~50-stem bunches. Stems will vary from 1–2 m in length, depending on the cultivar and growing conditions, and will be stored at full length to reduce dehydration during storage. Bunches are placed vertically on the ground, under the shade of trees or plastic screens and in an upright position. Before planting, farmers cut the stems into planting stakes (cuttings). Ideal stakes should have 5–7 nodes and a length of about 20 cm. Each stem yields an average of 5–7 stakes; however depending on the age and varietal characteristics, stems can yield from 3 to 12 stakes. This low multiplication rate strongly determines the way evaluation and selection (as described below) is implemented. There is no dormancy period, and stakes can be planted immediately after harvest. Even thin (green) stems can sprout and produce a vigorous plant.
Cassava is a cross-pollinated species. Farmers plant hybrids whose performance depends on heterosis (Cach et al. 2005, 2006; Calle et al. 2005; Jaramillo et al. 2005; Pérez et al. 2005a, b). If cassava shows heterosis, it also has inbreeding depression (Contreras Rojas et al. 2009). In this regard the genetic structure of cassava is similar to that of maize.
Cassava has seldom been self-pollinated. Successful cassava hybrids can be reproduced vegetatively. Therefore, there is no need to produce homozygous progenitors to be able to reproduce outstanding hybrids, contrary to the case of maize.
Limited knowledge exists on the inheritance of relevant traits. There is a remarkable vacuum in the current understanding of the genetic control of relevant traits. There has been limited work on inbreeding in cassava, very few reports on useful recessive traits, and a lack of knowledge on the inheritance of most relevant traits for this crop.
Cassava has a slow multiplication rate. On average, one cassava plant produces 7–8 stakes. It takes about 5 years to have enough planting material to conduct multi-location trials. Similarly it takes considerable time and effort to produce large-scale amounts of planting material of released varieties for the farmers.
Planting material can only be stored for a limited period of time. Logistics for breeding and production of planting materials for the farmers are cumbersome, expensive and oftentimes very complex. Stems cannot be stored for more than two months. In addition, planting material is bulky: a truckload of stems may be required to plant just 1–2 ha of cassava.
Planting material requires special phytosanitary management. As living tissue, the stems of cassava can carry a much wider array of pests and diseases much wider than botanical seeds. This considerably limits the flexibility of moving germplasm from one region to another.
Production of botanical seed from crosses is expensive. Flowering in cassava is not phenologically determined, as it is in cereal and legume crops. There is no “flowering season” in cassava. Workers walk large crossing nurseries daily in search of flowers that will be suitable for crosses. It takes two full weeks for a single worker to make the pollinations required to produce 1000 seeds of a given cross.
Planting material generates additional variation in field evaluations. Cuttings from green stems (slightly lignified) can sprout, but they are susceptible to attack by pathogens and insects and tend to dehydrate rapidly. Cuttings from stems older than 18 months are too lignified, contain small amounts of food reserves, and have reduced viability, delayed and slow sprouting, and poor vigor. In general the best sprouting occurs from cuttings coming from the mid-section of the stems.
A key objective in most cassava-breeding projects is high and stable production of fresh roots. The cassava crop´s reliability and resilience are among the characteristics most valued by farmers. Additional breeding objectives will depend heavily on the ultimate use of the crop. Productivity, plays a major role in industrial uses of cassava, whether for starch, animal feed or bio-ethanol, whereas stability of production is fundamental in the regions where cassava is the main subsistence crop. Industrial uses of cassava require not only high productivity of fresh roots, but also a minimum level of dry matter content (DMC) in these roots. This additional requirement of the starch industry arises from the fact that lower DMC leads to a larger amount of effluent liquids. Similarly, drying yards would require an additional day or two to complete the drying process. In areas where cassava is important for human consumption, cooking quality or starch characteristics may be more important than productivity.
Consumers frequently associate good cooking quality with other morphological traits, such as root peel color, as ‘markers’. Farmers frequently reject any change in such morphological traits, although they may have little or no correlation with actual cooking quality. Because of the specific needs and preferences of farmers and consumers, participatory approaches have been developed for cassava breeding (DeVires and Toenniessen 2001; Gonçalvez Fukuda et al. 2000; Gonçalvez Fukuda and Saad 2001). Other root quality traits relevant to cassava breeding programs throughout the world are the cyanogenic potential in the roots (Dixon et al. 1994), early bulking capacity, higher protein content in the roots, and reduced post-harvest physiological deterioration (PPD). The genetic variability for the latter two traits has been thought to be limited in M. esculenta. Inter-specific crosses with other Manihot species to introgress useful alleles have been attempted (Ceballos et al. 2006a, 2007a) especially for protein content and reduced PPD. More recently, the demand for cassava roots in the production of bio-ethanol has created new requirements linked to the costs of producing ethanol from the fresh or dry roots (Reddy et al. 2008).
Pests and Diseases
High and stable productivity relies heavily on adaptation to biotic and abiotic stresses specific to the growing environment. In Asia, cassava faces few diseases at serious levels, with the exception of cassava mosaic disease in India and Sri Lanka. In other continents, however, the disease pressures are higher. In Africa, Cassava Mosaic Disease (CMD) and Cassava Brown Streak Disease (CBSD) are important constraints (Calvert and Thresh 2002). A disease similar to CMD is also present in southern India. In certain regions of Latin America and the Caribbean (LAC), Frogskin Disease (FSD) causes roots to become “corky” and commercially unusable. It is not clear what organism is associated with FSD, but it is probably caused by a virus or a phytoplasm (Calvert et al. 2008; Alvarez et al. 2009). FSD is only found in LAC, and can be controlled effectively by avoiding the use of stems from diseased plants. Bacterial blight, caused by Xanthomonas axonopodis pv. manihotis (also known as X. campestris pv. manihotis; see review this issue), is found in Asia, Africa and LAC and can have devastating effects on yield and the availability of planting material, particularly in Africa and LAC (Hilloocks and Wydra 2002). Several fungal diseases also affect cassava productivity. Super-elongation disease, caused by Sphaceloma manihoticola (Teleomorph: Elsinoe brasiliensis) is widespread in the Americas, from Mexico to Southern Brazil. Phoma species cause leaf and stem lesions in the tropical highlands. Several species of Phytophthora, Sclerotium, Armillaea and Fusarium induce root rot. There are sources of genetic resistance to most of these diseases (CIAT 2001; Hilloocks and Wydra 2002).
Insect and mites decrease cassava productivity through its growing regions. In Asia pests cause more damage to cassava than diseases. Several arthropod pests feed on cassava and can reduce productivity (see Bellotti et al. review in this issue). Tetranychus spp. and other red mite species (from the genera Eutetranychus and Oligonychus) are the most conspicuous problem in Asia, whereas in other regions of the world it is the green mite (Mononychellus tanajoa) that can devastate cassava fields (Bellotti et al. 2002; Herrera-Campo et al. 2011; Nyiira 1975). The mealybugs Phenacoccus manihotis and P. herreri feed on cassava fields in Africa and LAC, respectively, and the former is becoming common in cassava fields in Thailand. Thrips (particularly Frankliniella williamsi and Scyrtotrips manihoti) considerably reduce yields of susceptible genotypes. Clones with pubescent leaves in their early stages of development offer excellent levels of resistance to these insects (Bellotti 2002), and this trait has been broadly incorporated into improved varieties. Whiteflies are among the most widespread pests in cassava. Aleurotrachelus socialis is the predominant species in northern South America, where it causes considerable crop damage through direct feeding. Bemisia tabaci is widely distributed in tropical Africa and several Asian countries. B. tabaci serves as the vector of the devastating CMD in Africa. However, high whitefly populations in Uganda and some regions of East Africa have been damaging through direct feeding and may be contributing to increase the spread of CBSD as well. Several other species of whitefly affect cassava in different regions, but Aleurodicus dispersus is probably the most common in Asia. Genetic resistance to whiteflies in cassava has been found, particularly for A. socialis, in several germplasm accessions from the CIAT collection (Bellotti 2002). Based on breeding work at CIAT, Colombia released the first whitefly-resistant variety of any crop.
There are several other arthropod pests affecting cassava roots, foliage and/or stems, particularly Lepidoptera, Diptera and Hemiptera genera. There is little or no known genetic resistance to those pests, and their management is commonly achieved through biological control measures. Attempts to produce transgenic cassava have succeeded, with the introduction of cry genes encoding insect-specific endotoxins (Bt toxins) from Bacillus thuringiensis (Fregene and Puonti-Kaerlas 2002; Taylor et al. 2004). Several attempts are currently underway to produce transgenic cassava with insect resistance.
There is a range of abiotic factors limiting cassava productivity. The crop is frequently grown in drought-prone regions and on low fertility soils. It can also be found in alkaline or acidic soils, most frequently the latter. Some traits associated with adaptation to these conditions have been suggested (Jennings and Iglesias 2002), such as leaf longevity (Lenis et al. 2006), optimum leaf area index, and ideal plant architecture (Hahn et al. 1979; Kawano et al. 1998; Kawano 2003). Research into transgenic cassava with higher leaf retention for increased tolerance to drought is ongoing (Peng et al. 2008) Photosynthesis is efficient in cassava and allows it to rapidly recover from stress (El-Sharkawy 2006; El-Sharkawy and Mejía de Tafur 2010). Cassava also has remarkable stomatal sensitivity, which allows a very rapid response to stress and results in reduced transpiration, high water use efficiency (WUE), and less soil water depletion (El-Sharkawy and Cock 1984; Cock et al. 1985; Alves and Setter 2000). Cassava´s advantage is in permitting photosynthesis only when environmental conditions allow it.
The capacity of the stems to withstand storage of up to 2 months from harvest to planting. Poor quality planting material can affect plant density and yields, and is especially important for areas with long dry spells or erratic rainfall. While there is known genetic variation for stem storability, it has not been a major breeding objective. Storability of planting material is likely to become a more relevant trait as predictions of climate change includes more erratic rainfall (see Jarvis et al. this issue).
A serious constraint to cassava production is the short shelf-life of its roots due to PPD. PPD begins within 24 h (Beeching et al. 1998; Rickard 1985) and rapidly renders the roots unpalatable and unmarketable. Consequently, cassava roots need to be consumed soon after harvesting (van Oirschot et al. 2000). Marketing options are limited due to high risk of loss, high marketing costs, and inability to access distant urban markets. The processes involved in PPD resemble changes associated with the plant’s response to wounding, which triggers a cascade of biochemical reactions that are frequently oxidative in nature (Beeching et al. 1998; Hirose et al. 1984; Uritani et al. 1984). Specific genes involved in PPD have been identified, characterized, and evaluated at the transcriptional level (Reilly et al. 2007).
Uses of Cassava
Animal Feed: Cassava can be a source of energy in animal diets but, due to its low levels of protein, requires supplementation with an additional source of protein (typically soybean derivatives). As a consequence, cassava sells at least 30% lower than maize, a crop with higher protein content (Tewe 2004). Higher levels of proteins in the roots is a relevant goal for the feed industry (Ceballos et al. 2006b). In addition, other nutritional traits, such as pro-vitamin A carotenoids, would be beneficial (Chávez et al. 2005; Posada et al. 2006).
Starch Industry: Cassava starch is particularly adapted to certain uses but completely unsuitable for others. This sector has always requested novel cassava starch types to diversify uses and to enhance the process efficiency for existing ones.
Ethanol and bioplastics: The biofuels and bioplastics industry is making a relatively new demand on cassava production and has been accentuated with recent increases in the price of oil. A “sugary” cassava (Carvalho et al. 2004) could decrease the economic and environmental costs of fermentation of cassava carbohydrates into ethanol or lactic acid (an alternative product in the pathway for the production of bioplastics). A key feature for production of ethanol is the ease of starch hydrolysis. A small-granule starch mutation (Ceballos et al. 2008; Rolland-Sabaté et al. 2012) yielded a product with a faster hydrolysis rate than any other source of commercial starch (unpublished data).
Processed food: Cassava roots without any trace of cyanogenic glucosides (acyanogenesis) is a highly desirable trait for the food industry. Additionally, properties of the starch, such as the relative proportion of amylose and amylopectin, the length of chains in the amylopectin molecule, the granule size and characteristics of its surface, strongly influence the use of cassava for processed food production (Charles et al. 2005; Hoover 2001; Peroni et al. 2006; Sánchez et al. 2005, 2010; Sriroth et al. 1999; Rolland-Sabaté et al. 2012). For example, syneresis (loss of water in refrigerated or frozen foods upon warming or defrosting) should be minimized. Expansion of products through extrusion depends heavily on proportion of amylose in the starch.
Cassava foliage may also be exploited for food and feed. Leaves are rich in minerals, proteins and vitamins, earning it the moniker “tropical alfalfa.” Many communities routinely consume cassava leaves, particularly in African countries such as Democratic Republic of Congo and Sierra Leone (Lancaster and Brooks 1983; Ngudi et al. 2003). High-density plantings for foliage harvests every 3–4 months over a 2-year cycle that is concluded with harvesting of the roots is under study and development in SE Asia.
Pre-breeding in Cassava
The genetic improvement of cassava starts with the assembly and evaluation of a broad germplasm base with selected lines used for the production of new recombinant genotypes. Since scientific breeding in cassava began only a few decades ago, the divergence between landraces and improved germplasm is not as great as in crops with a more extensive breeding history. As a result, landrace accessions play a more relevant role in cassava than in many other crops. For example, Nan Zhi 199, a very popular variety grown in the Guang Xi province of China, is actually a landrace from the germplasm collection at CIAT. Parental lines are selected based mainly on their performance, with little progress made to use general combining ability or breeding value as a criteria for parental selection (Hallauer and Miranda Fo 1988). Full-sib families can be produced by manual, controlled crossings, while half-sib families result from open pollinations in polycross nurseries.
The genetic variability available within Manihot has not been fully explored or screened. The limited evaluation of cassava genetic variability results from the difficulties and cost associated with collecting and maintaining cassava germplasm. Furthermore, detection of some of the economically important root traits is difficult. For instance, the many different kernel mutants in maize (popcorn, sweet, floury, opaque, waxy corn, etc.) are easily recognizable through their phenotype. No equivalent mutant had been reported for cassava until recently.
Nutritional quality factors studied to date show relatively low genetic variation, with the exception of the high carotene levels found in yellow cassava roots (Iglesias et al. 1997). However, an aggressive screening of cassava germplasm allowed Chávez et al. (2005) to report interesting variation in carotenoids and N content in the roots . Further analyses (Ceballos et al. 2006b) have confirmed the occurrence of cassava clones with 2–3 times higher levels of N, which gave hope than crude protein contents could be bred to be as high as 6–8% (compared with the typical levels of ~2% found in cassava roots). Recent (unpublished) data suggest that these high-N roots do not possess higher levels of proteins but rather a higher proportion of N-rich arginine. Accurate determination of the importance of non-protein nitrogen in cassava is critical for understanding if protein content can be significantly improved.
Another activity relevant to the proper screening of genetic variability is the introduction of inbreeding, which allows for the identification of useful recessive traits. CIAT started to systematically self-pollinate cassava germplasm consisting of both elite improved clones and materials from the germplasm collection in 2004. As a result, a waxy-type starch mutation (reduced amylose content) was discovered among self-pollinated progenies (Ceballos et al. 2007b, 2008). This discovery is important not only because of the economic value of such a trait, but also because it proves the usefulness of introducing inbreeding into cassava genetic improvement.
Pre-breeding activities also include wide crosses with wild relatives of cassava (Blair et al. 2007). Several traits of commercial importance have been found in these wild relatives and could be introgressed into the cassava gene pool. Among the most relevant ones are the tolerance to PPD in M. walkerae, increased protein content in M. tristis and M. Peruviana, resistance to the cassava green mite in M. esculenta sub spp. flavellifolia, and amylose-free starch in M. crassisepala and M. chlorosticta. M. glaziovii is suspected to be the origin of resistance to cassava mosaic disease and the hornworm in segregating progenies from crosses which involved this species as one of the progenitors (Blair et al. 2007). Improved nutritional quality has also been reported in wild relatives of cassava by Nassar and Ortiz (2008).
Different crossing schemes are used to produce botanical seed in cassava (Kawano 1980). For open pollinations, a field planting design developed by Wright (1965) is followed to maximize the frequency of crosses of all the parental lines incorporated in the nursery. At CIAT, seeds are germinated in greenhouse conditions, and the resulting seedlings transplanted to the field when they are 20–25 cm tall. At the International Institute of Tropical Agriculture (IITA), the same system is usually used, but in some nurseries seeds are directly planted in the field. Root systems in plants derived from botanical seed versus vegetative cuttings may differ considerably. The taproots from seedlings tend to store fewer starches than roots from cuttings (Rajendran et al. 2000). Because of this, it is difficult to correlate the root yield of clones at later stages in the evaluation/selection process with early results from the plants obtained from botanical seeds (Morante et al. 2005). However, when seeds are germinated in containers and later transplanted, the taproot often does not develop, and the seedling-derived plant may be more similar to subsequent stake-derived plants in terms of starchy root conformation.
The multiplication rate of cassava through vegetative cuttings is low. Under good environmental conditions, a cassava plant from a modern clone can easily yield up to 20 cuttings. However, when thousands of clones are handled in a range of environments, a realistic multiplication rate is in the range of 5–10 cuttings per plant. This imposes a critical limitation, because it takes several years until enough planting material is available for multi-location trials. One further complication is the number of factors that can affect quality of planting material. For example, the original positioning of the vegetative cutting along the stem considerably affects the performance of the plant it originates. Cuttings from the mid-section of the stems usually produce better performing plants than those from the top or the bottom. This variation in the performance of the plant due to the physiological status of the vegetative cutting results in larger experimental errors and undesirable variation in the evaluation process.
Description of the different evaluation and selection stages typically utilized in the CIAT cassava breeding program for a given target environment
Plants per plot
Clonal evaluation trial (CET)
Preliminary yield trial (PYT)
Advanced yield trial (AYT)
Regional trial (RT) - I
Regional trial (RT) - II
Description of the different evaluation and selection stages typically utilized in the IITA cassava breeding program for a given target environment
Plants per plot
Clonal evaluation trial (CET)
Preliminary yield trial (PYT)
Advanced yield trial (AYT)b
Regional trial (RT) - I
Regional trial (RT) - II
At CIAT, selection starts at nurseries planted with seedlings derived from botanical seeds (F1 in Table 1). Given the low correlation between the performance at seedling and clonal propagation stages, early selections are based on highly heritable traits such as plant type, branching habit and, particularly, reaction to diseases (Hahn et al. 1980a, b; Hershey 1984; Iglesias and Hershey 1994; Morante et al. 2005). The second stage of selection is called Clonal Evaluation Trial (CET). The few surviving genotypes from the single-plant selection conducted during the F1 or seedling stage produce the 6–10 vegetative cuttings required for this second step. The capacity to produce this number of cuttings is in fact another selection criteria utilized at the F1 stage. CETs usually range from 1500 to 3000 clones. Within a given trial, however, the same number of plants is used to avoid the confounding effects between number of plants and genotypic differences. Because the competition between neighboring genotypes in the CET may favor more vigorous plant architectures, selection at this stage relies heavily on heritable traits such as harvest index (Kawano et al. 1998; Kawano 2003; Morante et al. 2005). Plant type is an important selection criterion at early stages of selection: plants whose main stem does not branch until it reaches about 1 m are preferred (Kawano et al. 1978; Hahn et al. 1979). Other selection criteria at this stage include high dry matter and cyanogenic potential (Iglesias and Hershey 1994). At CIAT, between 100 and 300 clones survive the CET. During the first two stages of most programs selection is frequently visual with no data recording in order to manage a larger number of materials at lower costs.
One important trait that makes the harvest of large trials, such as the CET, expensive and time demanding is the measurement of dry matter content (DMC) in the roots. The productivity of cassava depends ultimately on the amount of fresh roots produced and the DMC of those roots. It is feasible to have excellent dry matter yields based on high production of fresh roots, even if they have below-average DMC. This situation, however, is generally not acceptable, because the transport and processing costs are too high. A sample of about 5 kg of roots is weighed in a hanging scale and the same sample of roots is weighed with the roots immersed in water. The relationship between the two weights provides an accurate estimate of DMC (Pérez et al. 2011). Morante et al. (2005) demonstrated that selection for high DMC during CETs had a good correlation with results at later stages of selection.
The following stage of selection at CIAT is the Preliminary Yield Trial (PYT), which consists of the evaluation of 10 plants in 3 replications. The ten plants in each replication are planted in two 5-plant rows. Rows are spaced only 0.8 m apart instead of the standard 1.0 m, and one empty row is left between plots to increase within-clone competition and to reduce between-clone competition. Large genetic variability occurs among clones, even within the same family. Although poor performing clones are mostly eliminated at the CET stage, there is still a considerable variation in the PYT trials. This highlights the needs for both a gradual process of selection and an avoidance of strong selection pressures. At IITA, the PYT trials typically have only two replications and no empty rows.
With the initiation of replicated trials, the emphasis of selection shifts from highly heritable traits to those with lower heritability, such as yield (Morante et al. 2005). Starting in PYT and increasingly during the Advanced Yield Trials (AYT) and the Regional Trials (RT), there will be a greater weight on yield and its stability across locations. Cooking quality, “gari” quality and “poundability” (IITA), and “farinha” quality (Brazil) testing will also began at these stages, when the number of genotypes evaluated has been reduced to a manageable size. AYT are typically grown in 1–2 locations. They consist of 3 replications per location and plots are 4 rows with 5 plants per row. Yield data are taken from the 6 central plants of the plot and the remaining 14 plants are used as a source of planting material for the next season. RTs are conducted for at least 2 years in 5–10 locations each year. Plots have 5 rows with 5 plants per row. Yield data is taken from the 9 central plants.
The process is indeed a mass phenotypic recurrent selection, because no family data are involved in the selection process.
Few data are taken in the early stages of selection, especially on genotypes that can be readily discarded by visual evaluation. Therefore, no data regarding general combining ability effects (i.e., breeding value) are available for a better selection of parental materials.
There is no proper separation between general (additive) and specific (heterotic) combining ability effects. The outstanding performance of selected materials is likely to depend substantially on positive heterotic effects, which cannot be transferred to the progenies sexually derived from them.
Inbreeding has been intentionally omitted in the breeding scheme. Therefore large genetic loads are likely to remain hidden in cassava populations and useful recessive traits are difficult to detect.
Two or more stages of selection may be based on non-replicated trials. A large proportion of genotypes are eliminated without a proper evaluation design.
Because of the above reasons, there are some clear opportunities to further improve the efficiency and effectiveness of cassava breeding. Kawano et al. (1998) mention that during a 14-year period, about 372,000 genotypes, derived from 4,130 crosses, had been evaluated at CIAT-Rayong Field Crop Research Center. Only three genotypes emerged from the selection process to be released as official varieties. Nonetheless, it should be mentioned that these varieties have achieved remarkable success in Asia, with more that one million hectares planted. Similar experiences have been observed at IITA, CIAT-Colombia and Brazil. The resulting increases in productivity account for a higher income (about one billion $US annually in SE Asia alone) to the poor farmers who grow the improved germplasm (Kawano 2003; Kawano and Cock 2005).
Major Breeding Achievements
Conventional plant breeding has one of the highest rates of return among the investments in agricultural research. The remarkable increase in the productivity of many crops during the twentieth century was to a large extent due to genetic gains achieved through crop breeding (Fehr 1987). Cassava has also benefited from technological inputs in the area of breeding (Kawano 2003). New varieties in Africa, Asia, and LAC have satisfied the needs of farmers, processors, and consumers and brought billions of dollars in additional income to small farmers. The application of tissue culture technologies (DeVires and Toenniessen 2001) as well as the definition of adequate cultural practices, particularly in relation to fertilization protocols (Howeler 2002), have also made positive contributions. Genetic transformation and molecular biology (Blair et al. 2007; Calderón-Urrea 1988; Fregene et al. 1997, 2000; Fregene and Puonti-Kaerlas 2002; Taylor et al. 2004) offer great potential but have not yet had any measurable commercial impact.
During the past 30–40 years, significant progress has been achieved in the initial phase of the science-based genetic improvement of cassava. With the turn of the millennium, the domestication of cassava was considered complete and, as a result, the crop shows greater adaptation to more intensive cultivation systems. This process involved assembling major traits such as improved yield (mainly through a higher harvest index), low cyanogenic content (when desirable), improved plant architecture and resistance/tolerance to the major diseases and pests. All these activities contributed to the general aim of increasing productivity and improving stability of production (Ceballos et al. 2010).
In addition to increases in productivity, Kawano (2003) reported a major improvement in DMC of cassava varieties released in SE Asia and demonstrated the importance of selection for adequate harvest index, particularly in early stages of the selection process. Jennings and Iglesias (2002) provided an assessment of the significant progress achieved towards developing cassava cultivars tolerant to the main viral diseases (CMD and CBSD), bacterial blight and super-elongation diseases. Resistance to CMD has been deployed and analyzed from the molecular point of view (Fregene et al. 2000, 2004; Egesi et al. 2007). Important progress in identifying and deploying tolerance/resistance to CBSD has also been achieved in recent years (McSween et al. 2006).
Improving the Efficiency of Cassava Genetic Improvement
Several approaches have been implemented in order to improve the phenotypic recurrent selection currently used to breed cassava (Ceballos et al. 2007a, 2010). These approaches ranged from the stratification of large experiments such as CETs to taking data from all genotypes (regardless if they are or not selected), which allowed the estimation of general combining ability of progenitors. These approaches will not be described further in this review. In spite of these innovations, several limitations in cassava breeding remain unsolved. This section will briefly describe limitations on cassava breeding and the approaches taken to overcome them.
Limited Knowledge of Cassava Genetics and the Bottleneck of Phenotyping
The first field study of Mendelian segregation of a commercially relevant trait in cassava took place as recently as in 2010. In order to develop an amylose-free starch (or “waxy”) commercial variety, a large, segregating population was screened using the reliable iodine test. The original source of the recessive mutation (wx wx) was crossed with eight elite non-waxy genotypes (Wx Wx) to produce about 800 F1 (Wx wx) genotypes. Non-related F1 genotypes were then crossed to generate a pseudo-F2 generation which was grown in the field. More than 12,000 genotypes were evaluated and, as expected, ¼ of the progeny showed the distinctive red staining (rather than the blue) of amylose-free starch (Ceballos et al. 2007b). This highlights the serious lack of knowledge regarding conventional genetics in cassava.
Gonçalves Fukuda and co-workers (2002) summarized Mendelian segregation for a few non-commercial traits. One such trait, narrow leaf lobe, is controlled by a single dominant gene. A second is that brown color of the root skin is dominant over white. Thirdly, a single dominant gene has been suggested for pale green in the stem collenchymas over dark green. Fourth, they deteremined that zigzagging stems is a recessive trait. Fifthly, red leaf nervures (veins) are dominant over the green coloration. Furthermore, at least two genes have been linked to the dominance of yellow parenchyma over white in the root (Iglesias et al. 1997).
Only a few articles regarding the inheritance of quantitative traits have been published (Easwari Amma and Sheela 1993, 1995; Easwari Amma et al. 1995; Easwari Amma and Sheela 1998; Losada-V 1990). These pioneering studies suggested important non-additive gene action for root yield, number of roots, root length, mean root weight, DMC, starch content, and reaction to bacterial blight. Although a molecular map has been already developed (Fregene et al. 1997; Mba et al. 2001), little knowledge based on traditional genetics has so far been produced. The reduction in costs and increase in efficiency in the genotyping process have advanced so much that the phenotyping that is required along with the molecular work is now the real bottleneck (Flavell 2008).
An aggressive and systematic screening of cassava germplasm should be made, starting with the cultivated species (Sánchez et al. 2009). There are many reports of desirable traits found in wild relatives of cassava (Nassar and Ortiz 2008), however, there is no evidence that these desirable traits cannot be found in M. esculenta itself. Key traits, such as sugary roots, waxy starch, tolerance to PPD, or high-protein (actually high-N), initially reported in wild relatives have now been found in M. esculenta (Carvalho et al. 2004; Ceballos et al. 2006b, 2007b; Morante et al. 2010; Rosero-Alpala et al. 2010). It is therefore reasonable to concentrate the search for useful traits in cultivated cassava. The identification of sources of useful traits should be accompanied by careful genetic analysis of its inheritance, as this knowledge is important for breeding purposes as well as to complement molecular maps and genome sequence information.
The Need for Homozygous Progenitors
The vegetative propagation of cassava is, without a doubt, a convenient and simple way to reproduce outstanding hybrids. However, it has had the negative impact of cassava breeding occurring without the need of inbred progenitors. The lack of inbred progenitors has several detrimental consequences for cassava breeding. Conventional backcross schemes cannot be implemented because there is no homozygous recurrent progenitor. As new sources for high-value traits are identified, the capacity to implement the back-cross scheme will become increasingly urgent. Genetic transformation work would also benefit considerably from the implementation of the back-cross approach. Heterosis, which is based on non-additive effects (dominance and epistasis), can be more efficiently exploited when progenitors are homozygous. Inbred progenitors facilitate the gradual and consistent assembly of favorable gene combinations, which in the current system occur just by chance. Inbreeding would also facilitate the reduction of the genetic load of this crop, which is expected to be relatively large at this stage of the evolution of the crop. Other advantages of the use of homozygous progenitors in cassava-breeding include facilitated germplasm exchange and conservation (as botanical seeds that breed true) and the replacement of diseased elite hybrid planting material by re-crossing the original progenitors (Contreras Rojas et al. 2009; Ceballos et al. 2010).
The possibility of a more dynamic interaction among the few cassava breeding projects of the world through the exchange of outstanding inbred progenitors should be emphasized. The synergies among the maize breeding programs from several state universities within the USA in the second half of the 1900s was critical for the remarkable achievements made in this crop. The chronic lack of resources for cassava research can be partially overcome by joint and coordinated efforts of different breeding programs.
At this point, inbreeding cassava through successive self-pollinations, requires 12–15 years and will, unavoidably, result in undesirable plant types (i.e., branching plants due to the early and profuse flowering phenotypes). Although self-pollinated seed may be obtained 6 months after planting in some genotypes, when it is attempted at a large scale, the synchronization of male and female flowers become a problem, particularly for erect plant types. Therefore, breeders need to clone the genotypes in order to have enough material to make self-pollinations. This extends the time required for each self-pollination cycle to at least 2 years. Several institutions are currently involved in a project to develop a protocol for the production of doubled haploids in cassava. This project, initially funded by the Rockefeller Foundation, is now supported by the Bill & Melinda Gates Foundation. Important progress has been achieved in understanding the development of cassava microspores and approaches to digest the early deposition of a thick exine wall (Wang et al. 2011). Ongoing work is attempting to produce doubled haploids through anther, microspore and/or ovule culture. These different approaches can now routinely induce cell divisions in gametes and produce multi-cellular structures (unpublished data). The research is now focusing on the regeneration of plants. Work for the production of doubled haploids through wide crosses with castor (Ricinus communis) is also underway. In this case, different treatments (e.g. application of 2–4 D or unviable cassava pollen) accompany the pollinations with castor pollen. Since this process usually results in weak embryos and/or dysfunctional or severely affected endosperm, early embryo rescue is likely to be necessary. Another approach for the development of homozygous cassava has recently been initiated, exploiting a centromere mutation first discovered in Arabidopsis thalliana (Ceballos et al. 2011b; Ravi and Chan 2010).
Induction of Flowering
As stated above, production of botanical seed from crosses among selected progenies is not a major problem in cassava, but it is still expensive, time-consuming, and relatively inefficient. An alternative would be to select early flowering genotypes, but this would result in a branching plant architecture that is generally disliked by farmers. There is a need, therefore, to induce early and profuse flowering temporarily in the crossing blocks. Flowering in cassava is under genetic control (highly heritable branching and non-branching phenotypes) and should be amenable to modulation through the exogenous application of phyto-hormones, as done for other crops (Botha-M et al. 1998; Dukovski et al. 2006; Trivellini et al. 2007). This approach would be very helpful for the implementation of rapid cycling, recurrent selection in cassava. An alternative would be grafting non-flowering types onto profuse and early flowering stocks
The capacity to induce flowering (in crossing blocks) would accelerate the breeding process, facilitate the development of mapping populations, increase the genetic variability in breeding populations, and allow for the production of homozygous germplasm.
Effective use of Molecular Markers
Molecular markers have not accelerated cassava breeding as it has in other crops. The first and only example in cassava for the successful use of marker-assisted selection (MAS) was the official release of CR 41-10 in Nigeria in 2010. This genotype had been originally selected using a set of molecular markers linked to resistance to the disease in Colombia (where CMD is not present) and evaluated in Africa for general adaptation and agronomic performance. In this way, the successful use of MAS in cassava has been similar to that in other crops: mostly constrained to simple, monogenic traits (Heffner et al. 2009). In another example, TILLING (McCallum et al. 2000; Till et al. 2003), a very promising reverse genetics molecular approach, has been applied to cassava. Rather than using TILLING to identify genotypes carrying a mutation in a specific gene, it was used to confirm that the amylose-free mutation was indeed located in the GBSS locus. However, results were disappointing (data not published), possibly because the homologous gene sequences used were from maize (Zea mays L.). Other applications of molecular markers include assessing genetic diversity (Morillo-C et al. 2011a; Kawuki et al. 2009), helping in the identification of germplasm grown by farmers (Alzate-G et al. 2010), understanding the inheritance of carotenoid content in the roots (Morillo-C et al. 2011b; Welsch et al. 2010) and assessing the haploid or doubled-haploid condition in the multi-cellular structures developed in the doubled-haploids project.
In spite of large investments and many articles describing molecular maps and the identification of QTLs, these technologies have yet to live up to original expectations. Xu and Crouch (2008) stated that “most of the [molecular markers] publications result from investments from donors with a strategic science quality or biotech advocacy mandate leading to insufficient emphasis on applied value in plant breeding. Converting promising publications into practical applications requires the resolution of many logistical and genetical constraints that are rarely addressed in journal publications. This results in a high proportion of published markers failing at one or more of the translation steps from research arena to application domain”. In this regard, cassava faces the same problems found in other crops.
Molecular technologies have evolved at astonishing speed. From 2000 to 2006, a private company reported a 6-fold decrease in costs per marker data point and increased the volume of their marker data by 40-fold (Eathington et al. 2007). Deficient field data and unreliable phenotypic information can explain the relatively limited examples of applied uses of molecular markers in cassava genetic enhancement. Other constraints common for most MAS projects are the problems related to the use of biparental mapping populations and deficient software. New alternatives for a more efficient use of molecular markers, such as genomic selection applied directly on actual breeding populations, are promising (Heffner et al. 2009). Additionally, the availability of homozygous progenitors would greatly enhance MAS for cassava´s genetic improvement. Perhaps one of the most important impacts that MAS has had is in backcrossing major genes, which requires homozygous recurrent genotypes. This highlights the importance of current ongoing efforts to develop a protocol for the production of doubled haploids in cassava.
New technologies (such as MAS or genetic transformation) can be effective only in the context of established and functional conventional breeding projects. Molecular and conventional breeding should not be competing capacities, but complementing ones. Molecular markers have many applications. When markers are developed for MAS, they should be sought after only when their use would be more efficient than conventional phenotypic screening. In this context, a marker that is not later used for selection purposes is not a marker, but an unacceptable waste of resources. Field breeders and molecular biologists should therefore carefully assess the circumstances where MAS or genomic selection offer a clear advantage. In addition, field breeders should be extremely careful providing reliable phenotypic data. They should also recognize that outcrosses in the development of mapping populations have devastating effects to the outcomes. Outcrosses in normal breeding do not have such negative implications. The quality and care required for crosses in molecular studies (as well as for conventional genetic studies that also remain to be made) need to be improved.
Currently CIAT is concentrating its efforts on developing molecular markers for tolerance to PPD and resistance to whiteflies. The former is a difficult trait to assess reliably and the current protocol requires relatively large number of roots (20–30), requiring cloning of the segregating progenies If a large number of genotypes is to be evaluated, the logistical problems for storing the respective number of roots are immense. The availability of molecular markers will facilitate the evaluation process and speed up recurrent selection projects to increase the tolerance to PPD. In the case of resistance to whiteflies, the main justification for the need of molecular markers is to circumvent the difficulty of achieving uniform infestation for reliable resistance screening, and to accelerate development of resistance.
Development of Genetic Stocks
Cassava diversity embodies alternative approaches to overcome most biotic stresses that it faces (Alvarez and Llano 2002; Bellotti 2002). Climate change, however, will most likely alter the dynamics of diseases and pests, which in some cases will lead to increases in economic losses. Predicting pest outbreaks and subsequent crop damage, in relation to environmental or climatic change, is a desirable goal, but difficult to achieve. It is already known that many arthropod pests, disease vectors, beneficial natural enemies, and diseases can be strongly influenced by climate (Ceballos et al. 2011a; Herrera-Campo et al. 2011). Clear identification of sources of resistance/tolerance to biotic stresses would therefore facilitate the rapid development of new cultivars that combine the required traits. In this regard, IITA and CIAT are advancing the idea of developing partially inbred genetic stocks. This concept can be expanded to include high-value traits such as enhanced nutritional quality, development of tolerant and susceptible checks for physiological studies on tolerance to abiotic stresses, and development of sources of resistance that allow for the identification of different races, isolates or biotypes of pests and diseases.
The availability of partially inbred genetic stocks would facilitate the maintenance and exchange of useful germplasm (as botanical seed), thus enhancing the collaboration among the cassava breeding projects, which currently work in a more or less isolated way. Partial inbreeding is desirable because such genetic stocks would be homozygous for the desirable trait, which implies doubling the breeding value compared with a heterozygous source. If the trait is quantitatively controlled by few or several genes (e.g. carotenoids content, dry matter content, cyanogenic potential), then the development of partially inbred genetic stocks would aim at concentrating favorable alleles, which enhances (but not doubles) their breeding value.
Cassava Genome Sequencing
The cassava whole genome sequence was recently developed by the Joint Genome Institute, Department of Energy (http://www.phytozome.net/cassava.php) and is composed of approximately 13,000 sequence scaffolds. This sequencing of cassava´s genome has been completed on a partially inbred line identified as AM 560-2. This achievement will strengthen and facilitate research for the genetic improvement of cassava. The discovery and genetic mapping of SNP markers is much needed to anchor and orient the scaffolds. The availability of gene sequences specific from cassava would facilitate the implementation of molecular tools such as TILLING and Eco-TILLING. Furthermore, the availability of a well-annotated, full-length cassava reference genome will be instrumental in facilitating valuable genomics, proteomics, and functional genomics applications. In the last decade, -omics-based research has expanded enormously from the model plant Arabidopsis into crop plant species (Mochida and Shinozaki 2010).
The sequence of the cassava genome will provide a better understanding of the genes that make cassava, as compared to other root crops, more drought-tolerant. The availability of genome sequence data should enhance genomics-assisted breeding in cassava for higher micronutrient availability or more resistant to pest and diseases. Moreover, genes identified in cassava could be useful for other crop species. This could be the case of the resistance to whiteflies (specifically against A. socialis) found in the landrace MECU72.
Cassava is a peculiar crop, facing unusual circumstances. It is a major staple crop that has received relatively little research attention in the past. Fortunately, this is changing, with both increased commitment by major granting agencies to support cassava research and interest on industrial applications by the processing sector (particularly for starch and bio-ethanol). Cassava enters the third millennium with promising prospects facilitated by new technologies, but with huge gaps in basic knowledge. It is critical therefore for policy-makers, the research community, and donors to understand that the challenges and opportunities of developing this crop for the future.