Rapid Cycling Recurrent Selection for Increased...

crop science, vol. 53, november–december 2013 www.crops.org 1

ReseaRch

Cassava (Manihot esculenta Crantz) is an important food source for more than 70 million people in developing countries who

obtain more than 500 kilocalories per day from its roots (Kawano et al., 1998). Cassava is the second most important food staple (in terms of calories consumed) in Sub-Saharan Africa (Nweke, 2004; IFAD/FAO, 2005; Tarawali et al., 2012) and is called Africa’s food insur-ance because it gives stable yields even in the face of drought, low soil fertility, and low intensity management (Dixon et al., 2003). However, current cassava varieties produce roots with low levels of protein, fat, minerals, and micronutrients such as provitamin A carotenoids (PVAC) (Ceballos et al., 2007; Thakkar et al., 2009).

Vitamin A is an essential micronutrient for the normal func-tioning of the visual and immune systems, growth and devel-opment, maintenance of epithelial cellular integrity, and for

Rapid Cycling Recurrent Selection for Increased Carotenoids Content in Cassava Roots

H. Ceballos,* N. Morante, T. Sánchez, D. Ortiz, I. Aragón, A.L. Chávez, M. Pizarro, F. Calle, and D. Dufour

AbstrActImproving total carotenoids content (TCC) in cassava roots is an important strategy to reduce vitamin A deficiency in human populations that rely on cassava as a source of energy in their diets. The high heritability for TCC in the roots allowed the International Center for Tropical Agriculture to implement a rapid cycling recur-rent selection approach that reduced the stan-dard length of each cycle from the ordinary 8 yr to 3. Data from successive evaluation nurseries suggested that gains have been made through time. However, no comparison of different cycles of selection has been made when representa-tives of each cycle were grown together. This study compares 4 to 5 clones representative of cycles of selection from 2004 to 2009. Results demonstrated significant gains for TCC as well as for total b-carotene (TBC) expressed both in a fresh and dry weight basis. Although dry mat-ter content (DMC) was not a selection criterion during the selection process, it increased with the successive cycles of selection. This sug-gests that indeed, simultaneous gains for TCC, TBC, and DMC are feasible. This finding is rel-evant for the important ongoing efforts at the International Institute of Tropical Agriculture and African National Programs to release biofortified cassava clones in Africa with adequate levels of dry matter content.

H. Ceballos, N. Morante, T. Sánchez, D. Ortiz, I. Aragón, A.L. Chávez, M. Pizarro, F. Calle, and D. Dufour, International Center for Tropical Agriculture (CIAT). Apartado Aéreo 6713. Cali, Colombia; D. Dufour, Centre de Coopération Internationale en Recherche Agronomique pour le Développement. (CIRAD); UMR Qualisud, 34398 Montpellier Cedex, France. Received 26 Feb. 2013. *Corresponding author ([email protected]).

Abbreviations: CEUN, National University of Colombia in Candelaria; CIAT, International Center for Tropical Agriculture; DMC, dry matter content; HPLC, high-performance liquid chroma-tography; LSD, least significant difference; MAP, mo after planting; NIR, near-infrared spectroscopy; PVAC, provitamin A carotenoids; TCC, total carotenoids content; TBC, total b-carotene; UV, ultravio-let; VAD, vitamin A deficiency.

Published in Crop Sci. 53:1–10 (2013). doi: 10.2135/cropsci2013.02.0123 Freely available online through the author-supported open-access option. © Crop Science Society of America | 5585 Guilford Rd., Madison, WI 53711 USA

All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.

2 www.crops.org crop science, vol. 53, november–december 2013

reproduction (ACC/SCN, 2000; Combs, 1998). Improv-ing the vitamin A status of children reduces mortality rates by 23 to 30% (ACC/SCN, 1992; Beaton et al., 1993). It is estimated that 75 to 251 million children have sub-clinical symptoms (WHO 2009) of vitamin A deficiency (VAD). In addition to the direct effect of VAD, there is growing evidence of vitamin A having synergistic effects with iron and zinc bio-availability (Graham and Rosser, 2000). Carotenes from vegetables contribute two-thirds of dietary vitamin A, worldwide, and more than 80% in the developing world (Combs, 1998).

Three main strategies have been traditionally used to prevent VAD: dietary diversification, food fortification, and/or supplementation. These strategies are relatively cost-effec-tive, but have failed to completely eradicate the problem for a diversity of reasons (West, 2003). Recently, different pro-grams such as HarvestPlus (www.harvestplus.org), involving a global alliance of research institutions, initiated the imple-mentation of a fourth strategy (biofortification) to develop micronutrient-dense staple crops (Bouis et al., 2011; Mon-tagnac et al., 2009; Dwivedi et al., 2012). Among these ini-tiatives is the development of biofortified cassava clones with high PVAC in the roots. Biofortification can be achieved through conventional breeding techniques that take advan-tage of the genetic variability for micronutrients in different crops (Welch and Graham, 2005; Chávez et al., 2005), but also through genetic transformation (Failla et al., 2012; Wel-sch et al., 2010). It represents a sustainable strategy that aims at solving the root of the micronutrient problem: a deficient diet. Fortunately, the conversion of PVAC present in cassava roots into vitamin A in humans has proven to be highly effi-cient (Failla et al., 2008; 2012; Liu et al., 2010; Tanumihardjo et al., 2010; Thakkar et al., 2007; 2009).

Recent information has demonstrated that carotenoid content for a given clone in some cases can vary consider-ably as a result of sampling issues and the influence of dry matter content between and within the roots (Ceballos et al., 2012a; Ortiz et al., 2011). Studies on genotype ´ environment interaction for carotenoid content in cassava reported that the interaction was statistically significant but did not result in drastic changes of the relative ranking of the different genotypes (Ssemakula and Dixon, 2007). Narrow sense heritability of carotenoid content in cassava roots is relatively high (Morillo-C. et al., 2012). There is growing evidence that carotene content in cassava roots depends on few genes (Akinwale et al., 2010), with some of them having a recessive behavior (Morillo-C. et al., 2012). A recessive condition would be observed for genes, such as lycopene e-cyclase, carotenoid b-hydroxylase, or carotenoid e-hydroxylase, controlling the conversion of b-carotene into other molecules (Qin et al., 2012).

Progress in increasing carotenoid content in cassava roots has been significant during the last few years. Maximum lev-els of total carotenoids reported in 2005 were around 10 mg

g-1 of fresh root, whereas by 2010 the maximum level was almost 25 mg g-1 of fresh root (Ceballos et al., 2012b; Sán-chez et al., 2010). This indication of genetic progress has been obtained at the International Center for Tropical Agricul-ture (CIAT) by comparing maximum values observed in the successive evaluation nurseries grown in Palmira, Colom-bia. However, no comparison has been made, so far, among high-carotene clones from different cycles of selection grown together and simultaneously in the same experiment and under the same environmental conditions. The objectives of this study were to: (i) describe results from successive evalua-tion and selection nurseries during the first decade of breed-ing for biofortified cassava; and (ii) compare high-carotene clones developed since 2004 through the second semester of 2009 growing them together in two different locations and harvested at different ages.

MAtErIALs AND MEtHODsrapid cycling recurrent selectionThe conventional breeding method for cassava is a phenotypic recurrent selection and it takes about 8 yr to complete a cycle of selection (Ceballos et al., 2012b). However, for several years, it has been known that the heritability of carotenoids content in cassava roots is high (as recently demonstrated by Morillo-C et al., 2012). Therefore, a drastic modification on the breeding scheme was introduced in the work to improve the concentra-tion of PVAC in cassava roots by shortening the duration of each cycle of selection as described below (Fig. 1).

The rapid cycling recurrent selection for increased levels of PVAC in the roots relied on the selection of genotypes based solely on their high-carotene content (Fig. 1). Selection took place at the end of the growth of F1 (seedling) trials 10 to 12 mo after planting (MAP). Selected genotypes were then cloned and planted in a crossing block which was kept in the field for 18 mo. Crosses were made to produce full-sib families. Because of large variation in the flowering habit of cassava, crosses can be made starting at six through 14 to 15 MAP (Fig. 1). Fruits from early crosses could be harvested on time to be planted in the third year of the cycle (New cycle-A in Fig. 1). However, most of segregating progenies would be planted in a seedling trial in the fourth year (New cycle-B in Fig. 1).

The resulting botanical seed was germinated and seed-lings transplanted to the field (F1 stage in the selection process described by Ceballos et al., 2007, 2012b). Therefore, there was just one plant (from the germinated seed) per genotype and, depending on the year, from 1000 to 2000 plants to be evaluated. This rapid cycling recurrent selection shortened the normal breeding cycle from 8 to 2 to 3 yr (Fig. 1) as it takes advantage of the high-heritability reported for TCC in cassava roots. The method would not be suitable, however, for improv-ing low-heritability traits such as yield.

selection of High-carotenoids Genotypes in Each seedling NurserySeedling nurseries varied in size from 1000 to 2000 genotypes. The harvesting season at CIAT lasts for 3 mo. Extraction and quantification of carotenoids (protocols described below) is time


University of Colombia in Candelaria (CEUN) also in Valle del Cauca Department. From each cycle of selection the best five clones were selected and enough stem cuttings were planted for harvests to take place at 8, 9, 10, and 11 MAP. A report on the plant-to-plant variation (within genotype) and root-to-root varia-tion (within a plant) has been published (Ortiz et al., 2011). Results from that study suggested that appropriate sampling would be achieved by harvesting 2 to 3 roots per plant and, hopefully, two plants per genotype. There is not much environmental influence (e.g., specific location in the field) on carotenoids content. Based on that information, the material was planted so that two plants could be harvested at each age. From each plant, 3 to 5 commer-cial-size roots were taken and homogenized. Therefore, for each age, there were two samples (replications) analyzed independently. The five clones from each cycle were planted together to facilitate the harvesting process. There was, therefore, a confounding effect of cycle of selection with position in the field where the respective plants were grown. However, the entire material was planted in such a small area that this confounding effect was expected to have a negligible effect on the results.

Soils at CEUN have been described as fine-silty, epiaquert ustic isohiperthermic (Gómez Enríquez et al., 2010), while those from Palmira are fine-silty, mixed, isohyperthermic Aquic Hapludol (Duque-Vargas et al., 1994).

Harvest and root HandlingThe handling of the roots was done carefully to prevent physical damage. Roots were not stored. Harvest took place at sunrise and carotenoids extraction was done in the morn-ing hours. The quantification of TCC was done before noon and HPLC quantifications were performed in the afternoon hours. Root samples and extracts were protected from the light as much as possible. Roots from each replication were ground together with a food processor with stainless steel tools into a uniform mash (SKYMSEN Food Processor MODEL PA-7SE), from which subsamples were taken for DMC measurement, carotenoids extraction and quantification, near-infrared spec-troscopy analysis (NIR) (FOSS 6500, monochromator with autocup sampling module, Hilleroed, Denmark), and color intensity with a Minolta ChromaMeter CR-410 (D65 illlumi-nant and a 10° observer, Osaka, Japan). If the spectrophotomer and HPLC data did not agree, NIR and color intensity were used to identify the suspicious data point and correct it.

root sampling ApproachesIn early cycles of selection (up to year 2009), only one root per genotype was taken and split into four longitudinal quarters. Two

consuming. Up to 40 quantifications per day can be done by spectrophotomer. Early quantifications with high-performance liquid chromatography (HPLC) restricted the number of sam-ples analyzed to 8 per day (up to year 2009). Improvements of lab equipment and in the processing protocol allowed, in later cycles, to analyze up to 18 samples per day with the HPLC. Selection was through a tandem approach; first, a visual assessment of root color intensity was made in the field (genotypes with cream and light yellow roots were eliminated). This allowed for a drastic reduction of the number of genotypes whose roots were actually evaluated through spectrophotometer and HPLC. Selection was basically based on carotenoids content in the roots. As breeding evolved, the selection criterion became more stringent and min-imum acceptable levels of PVAC increased. Selected germplasm was immediately cloned and planted to be used as progenitors and generate a new cycle of selection.

Germplasm Used in this studyEarly stages of the biofortification process focused in screening accessions from the germplasm collection at CIAT known to produce yellow roots. This process was finalized by 2004. How-ever crosses were started by 2002 and therefore, by the year 2004, the first segregating progenies for enhanced carotenoids content were evaluated in the field. Data from each F1 trial from 2004 through 2012 was recovered and consolidated for this study.

In addition, the best five clones from successive cycles of selection (genotypes identified in the F1s planted in 2004, 2005, 2006, 2007, 2008, 2009A, and 2009B) were compared in this study. In 2008, there were a few severe flooding episodes in Colombia and some plantings had to be repeated. This is why in 2009 there were actually two cycles of selection, one planted in the first semester (2009A) and the other during the second semes-ter (2009B). It should be emphasized that most crosses among the best materials evaluated and selected, for example, in 2004, would emerge only in trials planted in 2006 and thereafter, and similarly for the successive cycles of selection. The best materials from later F1 nurseries were not included in this study as the lim-ited planting material available was used for new crossing blocks.

Locations and Experimental DesignThe evaluations for the rapid cycling recurrent selection process were conducted at CIAT Experimental Station in Palmira, Valle del Cauca Department, Colombia. In this station operates the laboratory for carotenoids extraction and quantification. The best five clones of successive cycles of selection were grown and evalu-ated simultaneously in two different locations: CIAT Experimen-tal Station in Palmira and the Experimental Station of National

Figure 1. Illustration of the chronology of rapid cycling recurrent selection in cassava for enhanced carotenoids content.


opposite quarters were used for dry matter content quantification and the remaining two quarters were homogenized and used for carotenoids content quantification. However, as reported by Ortiz et al. (2011), this sampling procedure lead to some concerns about root to root variation, and since 2009, three roots per gen-otype were harvested and homogenized in a stainless steel food processor. The analysis of the best five clones of different cycles of selection grown together at Palmira and CEUN experimental stations were made following the latest sampling approach.

Dry Matter contentA sample from the ground roots was taken for the quantification of DMC. To estimate it, about 100 g of ground root tissue were dried in an oven at 105°C for 24 h. Dry matter was expressed as the percentage of dry weight relative to fresh weight.

carotenoid Extraction from root samplesCarotenoids were extracted following the method suggested in the literature (Rodriguez-Amaya, 2001; Rodriguez-Amaya and Kimura 2004), except that separation of the solid and liquid phases was performed by centrifugation and not by filtration (Chávez et al., 2005). Carotenoids are sensitive to ultraviolet (UV) light, pro-oxidants or associated compounds, and high temperature. Thus, steps were taken to avoid any adverse changes in this pig-ment due to such effects protecting them from UV light and avoiding excessively high temperatures. Special care was taken to avoid direct exposure to sunlight, and the lights in the labora-tory were protected with UV filters. Five grams of root tissue were added to a vial with 10 mL acetone. After 10 min, 10 mL of petroleum ether were added and mixed using an ultra-turrax (IKA Janke and Kunkel) for 1 min. Samples were then centrifuged (Eppendorf 5804R, Hamburg, Germany), at 3000 RPM, for 10 min, at 10°C. The organic phase was collected and extraction was repeated on the residue with 5 mL of acetone and 5 mL of petroleum ether, followed by centrifugation.

Extractions were optimized until its residues turned color-less. Based on preliminary analysis, it was decided that three iterative extractions would be used. The organic phases were combined with 10 mL of 0.1 M NaCl solution and centrifuged (3000 RPM, for 7 min, at 10°C). This washing process was repeated two additional times. The aqueous phase was extracted with a pipette. Petroleum ether was added to the extracts to adjust volume to 15 mL.

carotenoid QuantificationWith the extracts obtained, TCC was determined by visible absorption spectrophotometry (Cecil CE2021, Cambridge, UK), at an absorbance at 450 nm and using the absorption coef-ficient of b-carotene in petroleum ether (2592) (Rodriguez-Amaya, 2001; Rodriguez-Amaya and Kimura, 2004). From the organic phase used for spectrophotometric quantification of TCC, aliquots (15 mL) were transferred to glass tubes and dried by a nitrogen evaporator (N-Evap 112, Organomation Associates, Berlin, MA, USA). Immediately before injection, in the same tube, the dry extract was totally dissolved in 1.5 mL of (1:1) methanol and methy tert-butyl ether MeOH:MTBE HPLC-grade after sonication (10 s) and agitation in a VWR multi-tube vortexer (2400 rpm; 60 s) and filtered through a

0.22 mm polytetrafluorethylene (PTFE) filter. Separation and quantification of carotenoids were achieved using a YMC Carotenoid S-5 C30 reversed-phase column (4.6 mm ´ 150 mm: particle size, 5 mm), with a YMC Carotenoid S-5 guard column (4.0 ´ 23 mm) in a HPLC system (Agilent Technolo-gies 1200 series, Waldbronn, Germany), using a DAD detector with wavelength set at 450 nm.

Peaks were identified by comparing retention time and spectral characteristics against a pure standard from Caro-teNature GmbH, Lupsingen, Switzerland: b-Cryptoxan-thin– N°0055 HPLC 97%; Lutein– N°0133 HPLC 97%; Lycopene– N°0031 HPLC 95%; a-Carotene– N°0007 HPLC 97%; b-Carotene– N°0003 HPLC 96%; (E/Z)-Phytoene– N°0044 HPLC 98%; Violaxanthin– N°0259 HPLC 95%; Zeaxanthin– N°0119 HPLC 97%; and from Sigma-Aldrich: Xanthophyll– N°X6250 minimum 70%.

The quantity of each carotenoid was determined by integrat-ing the peak area against the respective standard curve. Before making any determinations, the method was previously vali-dated according to the requirements of Thompson et al. (2002) and EURACHEM/CITAC (2000). For All-trans b-carotene, the linearity (correlation coefficient) was 0.99925 at 2, 6, 10, 14, 18, 25, 30 ppm levels, and based on three replicates. The relative standard deviation in repeatability conditions was 2.1%. Carot-enoids contents were estimated on a fresh and dry weight basis.

Data Analysis of representatives of Different cycles of selectionAnalysis of variance and regression analyses were made using the PROC GLM and PROC REG tools from SAS software (SAS, 2008), respectively. Least significant difference (LSD) values were estimated as suggested by Steel and Torrie (1960).

rEsULts AND DIscUssIONsummary of results from F1 seedling trialsRelevant results from the different nurseries grown over the years are presented in Table 1. The germplasm evaluated each year in these data grew under different environmental conditions. In addition, the protocols for carotenoids quan-tification as well as for root sampling evolved over the years to achieve improved accuracy. In spite of these limitations, the information presented in Table 1 provides an excellent summary of the work to develop biofortified cassava as well as some of the problems that had to be overcome.

The screening of accessions from the germplasm collection involved carotenoids quantification of 1315 genotypes which is consolidated as 2004 data in Table 1. Not all of these materials produced yellow roots as early research sought for the possibility of identifying high carotenoid levels in lightly colored roots (Iglesias et al., 1997). The average level of TCC from this large sample of germplasm was 2.4 mg g-1 of fresh root tissue with a maxi-mum value of 10.3 mg g-1. The latter can be considered the baseline at the start of the project. Unless otherwise stated, all measurements for carotenoids content are on a


associated with these measurements would tend to be con-siderably higher than those for average values. For instance, the high TBC value of 19.1 mg g-1 observed in 2010 is most likely an overestimation. In spite of these problems, aver-ages and maximum values provide consistent evidence of the progress achieved over the years.

Two important trends can be highlighted from Table 1. The first relates to the concern at the beginning of the project that as TCC increased, the proportion represented by TBC would gradually fall. This is an important issue as b-carotene is more efficiently converted into vitamin A compared with other carotenoids. Although there is some tendency for the TBC/TCC ratio to be somewhat lower at later cycles of selection, it can be said that increasing TCC drags TBC values up as well. Therefore, the nutritional goal of 15 mg of b-carotene per gram of fresh root was finally surpassed in later cycles of selection. The second trend is that a negative relationship between TCC and DMC observed in the early years of the project (regression coefficients pre-sented in the last column of Table 1) basically faded away in later cycles. This is an important finding as it suggests that the problems faced in Africa to combine high TCC or TBC with adequate DMC must soon be overcome.

Figure 2 illustrates the kind of variations observed in a full sib family evaluated in 2011. Within this family was the genotype with the highest level of TCC measured in F1 seedling nurseries (25.8 mg g-1), but also genotypes with low levels ( > 1.0 mg g-1). This type of segregation suggests that inheritance of TCC is controlled by more than one gene. However, the rapid gains attained demon-strate relatively high realized heritabilities.

Evaluation of representatives of Different cycles of selectionWeather during the growth of this experiment was unusual, with rainfall well above the average for this region of Colombia. As a matter of fact, the whole country

fresh weight basis. Because of limitations in the capacity to process large numbers of samples through HPLC in the early years, no data for TBC can be presented for early stages of selection. In 2008, the HPLC equipment used for this work broke down at the beginning of the harvest sea-son and, since samples cannot be preserved, only data for TCC from spectrophotometer analysis can be provided.

Table 1 also presents data on DMC and on its regres-sion on TCC. The relationship between these two vari-ables is important as developing biofortified cassava germ-plasm with adequate levels of DMC has proved to be dif-ficult in Africa (Chimaobi et al., 2012), as well as in genetic transformation efforts (Failla et al., 2012). The proportion of TCC related to TBC is also presented in Table 1. This proportion varied over the years as carotenoid pigments different from b-carotene (such as lutein, lycopene, and violaxanthin) contribute differentially to color inten-sity depending on the respective absorption coefficient. Because during each year, a different generation of crosses was evaluated, there was considerable variation in the rela-tive importance of these non b-carotene carotenoids with strong pigmenting capacity. It is important to emphasize that the first step in the selection process was based on a visual assessment of intensely pigmented root parenchyma.

It should be emphasized that each cycle of selection lasted 2 to 3 yr. Therefore, segregating progenies from materials selected a given year will make up most of the genotypes evaluated 3 yr later. Average values for TCC and TBC must be analyzed by taking into account that the sam-ples analyzed are not representative of each cycle of selec-tion as only the best materials were chosen for carotenoids extraction and quantification. In other words, a truncated approach was taken and only the most promising materials were further processed in the lab. As progress was made, the cutoff point gradually moved for stronger color intensity. Maximum values also face the problem that they represent single data points from a single plant per genotype. Errors

Table 1. Summary of the data generated in each seedling nursery over the years. The number of genotypes planted in the field and those selected for analysis at the laboratory is provided at the left of the table. Selection pressure increased as better preselection protocols were developed.†

Year

n DMC (%) TCC TBC

TBC/TCC

Regression coefficient

Field Lab Average Average Max. Average Max. DMC vs. TCC

% ——————————————— mg g-1 FW —————————————— 2004 2110 1315 37.0 2.4 10.3 n.a. n.a. n.a. -0.0012005 2458 930 38.1 2.8 11.2 n.a. n.a. n.a. -0.1112006 1568 288 34.9 3.1 12.7 2.3 9.9 74 -0.1182007 1456 173 29.9 6.8 19.1 5.5 12.8 81 -0.1232008 3062 178 33.4 7.3 15.0 n.a. n.a. n.a. 0.056

2009 2175 345 33.5 10.6 18.9 4.9 10.3 46 0.103

2010 6840 490 30.2 11.4 24.7 9.8 19.1 86 0.014

2011 3415 332 37.1 15.6 25.8 8.5 15.0 54 -0.0122012 4037 415 35.7 14.7 24.3 8.6 16.2 59 0.079† DMC = Dry matter content; TCC = total carotenoids content; TBC = total b-carotene; n.a. = not available.


suffered from severe flooding. This situation resulted in higher than normal root rots, particularly at CEUN where soils are heavier. For this reason, after harvesting the plants at 8 MAP, it was decided to harvest only at two plant ages (8 and 11 MAP) in this location. This decision allowed enough plants and roots for balanced data. Harvests for the four plant ages could be made at CIAT in Palmira. In most cases, three roots from two different plants (replications) were harvested. In some cases, only two commercial and healthy roots were available from a given plant. Data from five clones was obtained from 2004, 2005, and 2009A cycles. For 2006, 2007, 2008, and 2009B cycles, only four representative clones could be properly analyzed.

Data for TCC based on the spectrophotometer and HPLC had, as expected, a very close relationship with a regression coefficient of 1.086 and a R2 value of 0.95. Over the years it has always been observed that HPLC quantifi-cations are slightly higher than those from the spectropho-tometer. Data from this study was not an exception and a regression coefficient slightly above 1.0 reflects this. Also, as observed before (Ceballos et al., 2012a), most of carot-enoids were in the form of all-trans b-carotene (61.2%). Together, all isomers of b-carotene (all-trans, 9-cis, 13-cis and 15-cis) accounted for 82.2% of TCC. The rest of the carotenoids were accounted for mainly by violaxanthin,

antheraxanthin, lutein, b-cryptoxanthin (data not pre-sented). Phytoene and phytofluene (precursors in b-car-otene synthesis) were also quantified. The description of results will concentrate in TCC and all forms of b-car-otene (TBC), because of their nutritional relevance and relative prevalence.

Table 2 presents the results of the analyses of variance for key variables measured in this study. All sources of variation were statistically significant (P < 0.01) except for a nonsignificant location effect for DMC and significance at the P < 0.05 for age of plant at harvest time in HPLC-quantified TCC.

Taking into consideration the significance of most sources of variation, Table 3 is presented to illustrate trends related to cycles of selection and age of the plant. There were consistent gains throughout the different cycles of selection for every variable, in agreement with data presented in Table 1. The fact that dry matter content increased along with TCC and TBC is very relevant, even though it was not a major selection criterion. This obser-vation contrasts with the ongoing projects to develop high-carotene clones in Africa which have faced problems combining high carotene with high DMC levels.

Many of the sources for high-carotene used in the Afri-can programs were introduced through the collaboration

Figure 2. Illustration of the variation observed for root color intensity in genotypes from a full-sib family, whose total carotenoids content ranged from < 1.0 to 25.8 μg g-1 (fresh weight basis).


between CIAT, the International Institute of Tropical Agriculture (IITA), and Nigeria’s National Root Crops Research Institute (Njoku et al., 2011) as well as other National Agriculture Research Institutions. These materi-als lacked adaptation to the African conditions which could be a feasible explanation for the initial problems to com-bine high DMC and TCC in that continent (Chimaobi et

al., 2012). However, it is likely that this problem will soon be overcome as data from this study predicts. Moreover, the analysis of the relationship between DMC and TCC in the data from the earlier F1 seedling nurseries resulted in negative regression coefficients, but they gradually reduced in magnitude and by 2008, had turned into a positive rela-tionship (Sánchez et al., 2010, Table 1). However, R2 val-ues for these regression analyses were always very low and regression coefficients were not statistically different from zero (data not presented). Therefore, as further recombina-tion and additional variability is introduced into the Afri-can breeding populations, it is likely that the apparent link-age between high carotene and low DMC will be broken. In fact, there have already been promising results reported from the latest nurseries evaluated in that continent (Egesi et al., 2012; Okoro et al., 2012).

This is indeed a very relevant suggestion as DMC is, more often than not, a key trait that defines farmers’ adop-tion of new cassava germplasm (Dixon et al., 2008). Hon-gbété et al. (2011) reported that high DMC improves mea-liness. Cassava cultivars with high DMC are better suited for producing “attiéké” in the Ivory Coast (Bakayoko et al., 2009). Safo-Kantanka et al. (1997) reported that cooking quality was related to dry matter content in the root, and that both of these variables were related to the arrival of the rains. Dry matter content of the root has an impact on the cooking time (Lorenzi, 1994; Safo-Kantanka et al., 1997). As in the case of Africa, DMC is also a key trait to improve in cassava grown in SE Asia (Kawano et al., 1998) and in the Americas (Ceballos et al., 2012b; Lorenzi, 1994).

The regression for different variables on age of the plant resulted in the coefficients presented in Table 4. Regres-sions were made combining data across locations and indi-vidually for each location. In general, data from CEUN was less reliable as suggested by the magnitude of the standard errors of the regression coefficients, particularly for DMC (the only nonsignificant, but still positive, coefficient).

It has to be emphasized that most of the crosses from the best materials in the 2004 cycle, for example, would have been planted in 2007, and only few may have been

Table 2. Analyses of variance for dry matter content (DMC), total carotenoids content by spectrophotometry (TCCSPEC), total carotenoids content by high-performance liquid chromatography (TCCHPLC), total b-carotene (TBC), and total carotenoids content on a dry weight basis (TCCDW).

Sources of variation df DMC TCCSPEC TCCHPLC TBC TCCDW

Location 1 2.44 62.43** 75.67** 18.25** 396.88**

Cycle 6 63.86** 141.70** 196.94** 76.34** 1070.39**

Clone(Cycle) 24 74.17** 36.92** 46.19** 38.96** 412.81**

Age 3 197.85** 6.99** 7.12* 15.27** 295.10**

Location*Cycle 6 33.93** 5.06** 6.81** 5.97** 103.21**

Location*Clone(Cycle) 24 7.83** 4.83** 4.78** 2.75** 38.27**

Error 277 3.29 1.55 2.01 1.08 11.83

* Significant at the 0.05 probability level.

** Significant at the 0.01 probability level.

Table 3. Averages (year of original nursery and age of har-vested plants) for dry matter content (DMC), total carotenoids content by spectrophotometry (TCCSPEC), total carotenoids content by high-performance liquid chromatography (TCCH-PLC), total b-carotene (TBC), and total carotenoids content on a dry weight basis (TCCDW). Since sample size for har-vest at 8 and 11 mo after planting (MAP) was different than that for 9 and 10 MAP, least significant difference (LSD) val-ues are presented for different types of comparisons.

DMC TCCSPEC TCCHPLC TBC TCCDW

Year of original nursery

2004 34.89 7.78 8.17 5.26 23.71

2005 37.65 9.24 9.65 5.52 26.04

2006 35.67 7.50 7.33 4.67 20.61

2007 36.71 10.31 10.93 7.25 29.87

2008 37.39 10.37 10.93 6.21 29.57

2009A 38.38 11.76 12.45 7.48 32.47

2009B 38.57 12.96 13.77 8.61 35.96

LSD 0.05 0.72 0.49 0.56 0.41 1.36

LSD 0.01 0.94 0.65 0.74 0.54 1.79

Age of plants sampled (MAP)

8 37.97 9.48 10.01 6.53 26.42

9 37.96 10.39 11.09 7.11 29.21

10 38.04 10.61 10.75 6.68 28.22

11 35.03 10.06 10.54 5.77 30.17

For comparisons between 8 and 11 MAP

LSD 0.05 0.48 0.33 0.37 0.27 0.91

LSD 0.01 0.63 0.43 0.49 0.36 1.19

For comparisons between 9 and 10 MAP

LSD 0.05 0.65 0.45 0.51 0.37 1.23

LSD 0.01 0.85 0.59 0.67 0.49 1.62

For comparisons between 8 or 11 MAP and 9 or 10 MAP

LSD 0.05 0.47 0.32 0.37 0.27 0.89

LSD 0.01 0.62 0.43 0.48 0.35 1.17


included in the 2006 evaluations. Also, these genotypes continued to be included in the crossing blocks for one or two more years and then their use gradually faded away. A similar situation occurred with the successive cycles of selection. It is acknowledged that this situation is unusual for a study measuring cycles of selection, but this is the restriction imposed by the reproductive biology of cassava with some genotypes flowering early and profusely and others late and scarcely, the latter being the phenotypes preferred by farmers (Ceballos et al., 2012b). Therefore, rather than having distinctive cycles of selection that do not overlap with each other, as it happens in short season crops, the cycles of selection in this study overlap with each other as few genotypes from the best clones selected a given year may have been included in evaluation nurseries 2 yr later, with the peak of segregating progenies included in the nurseries 3 yr later, and then gradually (but not completely) reducing in frequency thereafter.

Impact of Age of the Plant and Dry Matter content on carotenoids contentsThe analyses of DMC in relation to age of the plant are also interesting (Table 3). In general, there is a trend for increases up to 10 MAP with a decrease at 11 MAP. As can be seen, DMC showed the most drastic reduction with values falling from 38.0 to 35.0%. Weather in Colombia has suffered drastic changes from normal patterns in the last few years. This has affected the typical bimodal rainfall pattern, which in turn has modified the planting and har-vesting times in the cassava program. Dry matter content at 11 MAP was reduced because the trials were planted a month later than usual and the rains arrived relatively early (before the last harvest had taken place). The reduction of DMC, therefore, is the result of the plants already reiniti-ating their growth, which requires energy from the roots. Rainfall patterns have a predictable response on DMC in cassava roots (Bakayoko et al., 2009; Ceballos et al., 2012b).

Dry matter content had a large impact on TCC and TBC. There was maximum TCC and TBC at 9 to 10 MAP and then a consistent reduction for the last harvest. However, when TCC values are expressed on a dry weight basis, the trend is more regular with consistent gains in older plants. It can be postulated that, as DMC varies with the physiological status of the plant, the relative content

of water will change accordingly, while carotenoids accu-mulated in the root remain more or less constant. The variation of water content in the root “dilutes” the con-centration of carotenoids. This is an effect that is not pres-ent when TCC is expressed in a dry weight basis. Similar observations have already been reported in retention stud-ies (Ceballos et al., 2012a). It is always important, there-fore, to provide DMC information in these kind of studies and to report TCC or TBC values, both on a fresh and dry weight basis for accurate comparisons.

These results suggest significant gains through conven-tional breeding. In fact, results suggest an advantage over results achieved through genetic transformation based on the most recent report (Failla et al., 2012). In that study, the average of seven transgenic lines was 4.8 mg g-1 (FW) and the maximum individual value was below 6 mg g-1. As expected, in that report, most of the carotenoids were also b-carotene, with an average of 4.5 mg g-1 (FW). A major issue that is still unresolved in the genetic transformation of cassava for increased carotenoids content is the unaccept-ably low levels of dry matter content, which, in the report of reference, was only 20.2% (no information was provided regarding the age of the plants at harvest). In spite of the clear disadvantages that genetic transformation may have, it still offers the advantage that farmers’ preferred varieties may be transformed (provided that DMC can be maintained).

cONcLUsIONsThis is the first report of a rapid cycling process to increase carotenoids in cassava on a large scale. Jos et al. (1990) reported on progress to increase carotenoids content in cassava roots after a few cycles of selection. Results indi-cated that cassava responded well to the methodology employed with consistent and significant gains not only for TCC and TBC, but surprisingly, also for DMC. The gains observed are comparable with those from similar studies in crops like maize. The major difference is that in cassava, rapid cycling can only be implemented for high heritability traits such as PVAC content. For low heritabil-ity traits, however, the low multiplication rate implies that several years are needed for producing the planting mate-rial that is required for multilocation trials.

The parallel gains in carotenoids content and DMC observed in this study are a clear demonstration that these

Table 4. Regression coefficients (and their respective standard error) for dry matter content (DMC), total carotenoids content by spectrophotometry (TCCSPEC), total carotenoids content by high-performance liquid chromatography (TCCHPLC), total b-carotene (TBC), and total carotenoids content on a dry weight basis (TCCDW) on year of selection. The independent variable is year of original nursery.

DMC TCCSPEC TCCHPLC TBC TCCDW

Across locations 0.492(0.092) 0.818(0.064) 0.904(0.071) 0.542(0.059) 2.030(0.197)

CEUN† 0.127(0.190) 0.725(0.102) 0.939(0.118) 0.639(0.106) 2.467(0.356)

CIAT‡ 0.659(0.096) 0.842(0.078) 0.862(0.087) 0.476(0.069) 1.765(0.232)† CEUN, National University of Colombia in Candelaria.‡ CIAT, International Center for Tropical Agriculture.


two traits are not necessarily linked (from the phenotypic point of view) and simultaneous improvement for them is feasible. This is an important finding for the ongoing efforts in Africa and provides hope that the current prob-lems to obtain higher DMC in yellow rooted cassava may soon be overcome.

Finally, the results of the study are further evidence of the high heritability for TCC and TBC and the cor-relation between these two variables. The results suggest a high magnitude of realized heritability which agrees with recent reports in this regard (Morillo-C. et al., 2012)

AcknowledgmentsThe financial support provided by the Bill and Melinda Gates Foundation, USAID, and World Bank through the Harvest-Plus initiative has been fundamental for the research to pro-duce high-carotene cassava germplasm. Carotene analyses were carried out at CIAT’s Nutrition Quality Laboratory which is funded by the Monsanto Fund and the AgroSalud Project (CIDA 7034161).

referencesACC/SCN. 1992. Second report on the world nutrition situa-

tion. Administrative Committee on Coordination, Subcom-mittee on Nutrition. Food Policy Research Institute, United Nations, Geneva, Switzerland.

ACC/SCN. 2000. Fourth report on the world nutrition situa-tion. Administrative Committee on Coordination, Subcom-mittee on Nutrition. Food Policy Research Institute, United Nations, Geneva, Switzerland.

Akinwale, M.G., R.D. Aladesanwa, B.O. Akinyele, A.G.O. Dixon, and A.C. Odiyi. 2010. Inheritance of b-carotene in cassava (Manihot esculenta Crantz). Int. J. Genet. Molec. Biol. 2:198–201.

Bakayoko, S., A. Tschannen, C. Nindjin, D. Dao, O. Girardin, and A. Assa. 2009. Impact of water stress on fresh tuber yield and dry matter content of cassava (Manihot esculenta Crantz) in Côte d’Ivoire. African J. Agric. Res. 4:21–27.

Beaton, G.H., R. Martorell, K.J. Aronson, B. Edmonston, G. McCabe, A.C. Ross, et al. 1993. Effectiveness of vitamin A supplementation in the control of young child morbidity and mortality in developing countries. ACC/SCN State of the Arts Series, Nutrition Policy Paper 13. World Health Orga-nization, Geneva, Switzerland.

Bouis, H.E., C. Hotz, B. McClafferty, J.V. Meenakshi, and W.H. Pfeiffer. 2011. Biofortification: A new tool to reduce micro-nutrient malnutrition. Food Nutr. Bull. 32:S31–S40.

Ceballos, H., M. Fregene, J.C. Pérez, N. Morante, and F. Calle. 2007. Cassava genetic improvement. In: M.S. Kang and P.M. Priyadarshan, editors, Breeding major food staples. Blackwell Publishing, Ames, IA. p. 365–391.

Ceballos, H., C. Hershey, and L.A. Becerra-López-Lavalle. 2012b. New approaches to cassava breeding. Plant Breed. Rev. 36:427–504.

Ceballos, H., J. Luna, A.F. Escobar, J.C. Pérez, D. Ortiz, T. Sán-chez, et al. 2012a. Spatial distribution of dry matter in yellow fleshed cassava roots and its influence on carotenoids reten-tion upon boiling. Food Res. Int. 45:52–59. doi:10.1016/j.foodres.2011.10.001

Chávez, A.L., T. Sánchez, G. Jaramillo, J.M.I. Bedoya, J. Ech-everry, E.A. Bolaños, et al. 2005. Variation of quality traits in cassava roots evaluated in landraces and improved clones. Euphytica 143:125–133. doi:10.1007/s10681-005-3057-2

Chimaobi, I., C. Egesi, E. Okogbenin, B. Olasanmi, H. Ogbuekiri, A. Udugwu, et al. 2012. Nutritional implication of the trend of genetic expression in yellow cassava roots. Paper presented at: Second Conference of the Global Cassava Partnership. Kampala, Uganda. 18–22 June.

Combs, G.F. 1998. The vitamins. Fundamental aspects in nutri-tion and health. Academic Press, London, UK.

Dixon, A.G.O., M.O. Akoroda, R.U. Okechukwu, F. Ogbe, P. Ilona, L.O. Sanni, et al. 2008. Fast track participa-tory approach to release of elite cassava genotypes for vari-ous uses in Nigeria’s cassava economy. Euphytica 160:1–13. doi:10.1007/s10681-007-9487-2

Dixon, A.G.O., R. Bandyopadhyay, D. Coyne, M. Ferguson, R.S.B. Ferris, R. Hanna, et al. 2003. Cassava: From poor farmers’ crop to pacesetter of African rural development. Chron. Horticult. 43(4):8–15.

Duque-Vargas, J., S. Pandey, G. Granados, H. Ceballos, and E.B. Knapp. 1994. Inheritance of tolerance to soil acidity in tropi-cal maize. Crop Sci. 34:50–54. doi:10.2135/cropsci1994.0011183X003400010009x

Dwivedi, S.L., K.L. Sahrawat, K.N. Rai, M.W. Blair, M.S. Ander-sson, and W. Pfeiffer. 2012. Nutritionally enhanced staple food crops. Plant Breed. Rev. 36:173–293.

Egesi, C.N., B. Olasanmi, T.J. Onyeka, E. Okogbenin, I. Chi-maobi, P. Ilona, et al. 2012. High provitamin A cassava vari-eties for Nigerian farmers: Current status and perspectives. Paper presented at: Second Conference of the Global Cassava Partnership. Kampala, Uganda. 18–22 June.

EURACHEM/CITAC. 2000. Quantifying uncertainty in ana-lytical measurements, 2nd ed. EURACHEM/CITAC. www.measurementuncertainty.org/mu/QUAM2000-1.pdf (accessed 10 Sept. 2010).

Failla, M.L., C. Chitchumroonchokchai, D. Siritunga, F.F. De Moura, M. Fregene, and R.T. Sayre. 2012. Retention during processing and bioaccessibility of b-carotene in high b-caro-tene transgenic cassava root. J. Agric. Food Chem. 60:3861–3866. doi:10.1021/jf204958w

Failla, M.L., T. Huo, and S.K. Thakkar. 2008. In vitro screening of relative bioaccessibility of carotenoids from foods. Asia Pac. J. Clin. Nutr. 17:200–203.

Gómez Enríquez, A., H. Rojas Palacios, F.A. Vallejo Cabrera, and E.I. Estrada Salazar. 2010. Water requirements of pep-per in Candelaria, Valle del Cauca, Colombia. Acta Agron. 59:442–448.

Graham, R.D., and J.M. Rosser. 2000. Carotenoids in staple foods: Their potential to improve human nutrition. Food Nutr. Bull. 21:404–409.

Hongbété, F., C. Mestres, N. Akissoé, B. Pons, J.D. Houn-houigan, D. Cornet, et al. 2011. Effects of cultivar and har-vesting conditions (age, season) on the texture and taste of boiled cassava roots. Food Chem. 126:127–133. doi:10.1016/j.foodchem.2010.10.088

IFAD/FAO 2005. Proceedings of the Validation Forum on the Global Cassava Development Strategy, Vol. 2. A review of cassava in Africa with country case studies on Nigeria, Ghana, the United Republic of Tanzania, Uganda and Benin. Rome, 26–28 Apr. 2000. International Fund for Agricultural Devel-opment (IFAD) and Food and Agriculture Organization of


the United Nations (FAO), Rome.Iglesias, C., J. Mayer, A.L. Chávez, and F. Calle. 1997. Genetic

potential and stability of carotene content in cassava roots. Euphytica 94:367–373. doi:10.1023/A:1002962108315

Jos, J.S., S.G. Nair, S.N. Moorthy, and R.B. Nair. 1990. Carotene enhancement in cassava. J. Root Crops 16:5–11.

Kawano, K., K. Narintaraporn, P. Narintaraporn, S. Sarakarn, A. Limsila, J. Limsila, et al. 1998. Yield improvement in a mul-tistage breeding program for cassava. Crop Sci. 38:325–332. doi:10.2135/cropsci1998.0011183X003800020007x

Liu, W., Y. Zhou, T. Sánchez, H. Ceballos, and W.S. White. 2010. The vitamin A equivalence of b-carotene in b-carotene-bio-fortified cassava ingested by women. FASEB J. 24:92.7.

Lorenzi, J.O. 1994. Variação na qualidade culinária das raízes de mandioca. Brangantia 53(2):237–245. doi:10.1590/S0006-87051994000200013

Montagnac, J.A., C.R. Davis, and S.A. Tanumihardjo. 2009. Nutritional value of cassava for use as a staple food and recent advances for improvement. Compr. Rev. Food Sci. Food Saf. 18:181–194. doi:10.1111/j.1541-4337.2009.00077.x

Morillo-C., Y., T. Sánchez, N. Morante, A.L. Chávez, A.C. Morillo-C., A. Bolaños, et al.. 2012. Estudio preliminar de herencia del contenido de carotenoides en raíces de pobla-ciones segregantes de yuca (Manihot esculenta Crantz). Acta Agron. 61:253–264.

Njoku, D., V. Gracen, C.N. Egesi, I. Asante, S.K. Offei, E. Okog-benin, et al.. 2011. Breeding for enhanced b-carotene content in cassava: Constraints and accomplishments. J. Crop Improv. 25:560–571. doi:10.1080/15427528.2011.594978

Nweke, F. 2004. New challenges in the cassava transformation in Nigeria and Ghana. PTD Disc. Pap. No.118. Environmen-tal and Production Technology Division, International Food Policy Research Institute, Washington DC.

Okoro, M.J., C.N. Egesi, B. Olasanmi, I. Chimobi, S. Njoku, U. Ikeogu, et al. 2012. Clonal evaluation trial of yellow root cas-sava genotypes in south eastern Nigeria. Paper presented at: Second Conference of the Global Cassava Partnership. Kam-pala, Uganda. 18 –22 June.

Ortiz, D., T. Sánchez, N. Morante, H. Ceballos, H. Pachón, M.C. Duque, et al. 2011. Sampling strategies for proper quantifica-tion of carotenoids content in cassava breeding. J. Plant Breed. Crop Sci. 3:14–23.

Qin, X., W. Zhang, and J. Dubcovsky. 2012. Cloning and com-parative analysis of carotenoid b-hydroxylase genes provides new insights into carotenoid metabolism in tetraploid (Triti-cum turgidum ssp. durum) and hexaploid (Triticum aestivum) wheat grains. Plant Mol. Biol. 10.1007/s11103–012–9972–4.

Rodriguez-Amaya, D.B. 2001. A guide to carotenoid analysis in foods. ILSI Press, Washington, DC.

Rodriguez-Amaya, D.B., and M. Kimura. 2004. HarvestPlus handbook for carotenoid analysis. HarvestPlus Tech. Monogr. 2. International Food Policy Research Institute (IFPRI) and International Center for Tropical Agriculture (CIAT). Wash-ington, DC and Cali, Colombia.

Safo-Kantanka, O., J. Owusu-Nipah, R. Acquistucci, M. Osei-Minta, A. Aubyn, and C. Quansah. 1997. Contribution to the understanding of the cooking quality of cassava (Manihot escu-lenta Crantz). In: Improvement of basic food crops in Africa through plant breeding, including the use of induces muta-tions. Proceeding of a final research co-ordination meeting. FAO/IAEA. Rome, Italy. p. 121–129.

Sánchez, T., D. Ortiz, N. Morante, H. Ceballos, H. Pachón, F. Calle, et al. 2010. New approaches for quantifying carotenoids content in cassava roots. Paper presented at: 11th Triennial Symposium. International Society for Tropical Root Crops, African Branch. Kinshasa, D.R. Congo. 4–8 Oct. 2010.

SAS. 2008. SAS/STAT 9.1 User’s guide. SAS Institute Inc., Cary, NC.Steel, R.G.D., and J.H. Torrie. 1960. Principles and procedures of

statistics. McGraw-Hill , New York. p. 106–114.Ssemakula, G., and A. Dixon. 2007. Genotype ´ environment

interaction, stability and agronomic performance of carot-enoid-rich cassava clones. Sci. Res. Essays 2:390–399.

Tanumihardjo, S.A., N. Palacios, and K.V. Pixley. 2010. Provitamin A carotenoid bioavailability: What really matters? Int. J. Vitam. Nutr. Res. 80:336–350. doi:10.1024/0300-9831/a000042

Tarawali, G., C. Iyangbe, U.E. Udensi, P. Ilona, T. Osun, C. Okater, et al. 2012. Commercial-scale adoption of improved cassava varieties: A baseline study to highlight constraints of large-scale cassava based agro-processing industries in South-ern Nigeria. J. Food Agric. Environ. 10:689–694.

Thakkar, S.K., T. Huo, B. Maziya-Dixon, and M.L. Failla. 2009. Impact of style of processing on retention and bioaccessibility of b-carotene in cassava (Manihot escultenta Crantz). J. Agric. Food Chem. 57:1344–1348. doi:10.1021/jf803053d

Thakkar, S.K., B. Maziya-Dixon, A.G.O. Dixon, and M.L. Failla. 2007. b-carotene micellarization during in vitro digestion and uptake by Caco-2 cells is directly proportional to b-carotene content in different genotypes of cassava. J. Nutr. 137:2229–2233.

Thompson, M.L., L.R. Stephen, S. Ellison, and R. Wood. 2002. Harmonized guidelines for single-laboratory valida-tion of methods of analysis. Pure Appl. Chem. 74:835–855. doi:10.1351/pac200274050835

Welch, R.M., and R.D. Graham. 2005. Agriculture: The real nexus for enhancing bioavailable micronutrients in food crops. J. Trace Elem. Med. Biol. 18:299–307. doi:10.1016/j.jtemb.2005.03.001

Welsch, R., J. Arango, C. Bär, B. Salazar, S. Al-Babili, J. Beltrán, et al. 2010. Provitamin A accumulation in cassava (Manihot esculenta) roots driven by a single nucleotide polymorphism in a phytoene synthase gene. Plant Cell 22:3348–3356. doi:10.1105/tpc.110.077560

West, K.P., Jr. 2003. Vitamin A deficiency disorders in children and women. Food Nutr. Bull. 24:S78–S90.

WHO. 2009. Global prevalence of vitamin A deficiency in popula-tions at risk 1995–2005. World Health Organization, Geneva, Switzerland.

Rapid Cycling Recurrent Selection for Increased...

Documents

Transcript of Rapid Cycling Recurrent Selection for Increased...