Development of End-User Preferred Sweetpotato Varieties

Sweetpotato (Ipomoea batatas (L.) Lam) is the fourth most important root and tuber crop in Ghana, in terms of production. Attainment of increased sweetpotato utilization has become an important breeding objective in Ghana recently. The major emphasis in breeding is on the development of farmer/consumer preferred varieties. This study aimed at developing farmer/consumer preferred sweetpotato cultivars for increased utilization in Ghana and beyond. One hundred and fifteen sweetpotato accessions were collected and evaluated at two ecozones in the major and minor cropping seasons in 2011 to identify low sugar parents for hybridization. Two released varieties (Histarch and Ogyefo) and eight breeding lines (AAT-03-025, CIP 442264, CRIWAC 25-10, CRIWAC 30-10, DOS 03-006, CRIWAC 11-10, CIP 440095 and CRIWAC 19-10) were selected and used as parents. Genetic variability was significant for all the traits studied. Sufficient useful genetic variation was present in the materials studied and was exploited to provide for substantial amount of improvement through selection of superior genotypes. Negative heterosis was observed for sugar content and this is very important for breeding because Ghanaians prefer non-sweet varieties. Fifteen percent of the F1 hybrids of Histarch and Ogyefo were non-sweet. These will meet the staple food needs of Ghanaians. Eight hybrids were identified as potential non-sweet varieties for further testing multilocation on-farm for release. These were Ogyefo × Histarch-11, Histarch × Ogyefo-13, Histarch × Ogyefo-52, Histarch × Ogyefo-37, Histarch × Ogyefo-65, Histarch × Ogyefo-88, Histarch × Ogyefo-39 and Histarch × Ogyefo-16.


Introduction
Sweetpotato (Ipomoea batatas (L.) Lam) is one of the most important root crops in the world with more than 133 million tonnes produced worldwide annually (Warammboi et al., 2011).It is the fourth most important root and tuber crop in Ghana in terms of production, after cassava, yam and cocoyam.Its annual production is estimated at 135, 000 tonnes, representing just under 0.6% of all root and tuber crops produced in Ghana (FAOSTAT, 2013).The attainment of improved crop yield is an important objective in most breeding programmes (Rausul et al., 2002), as is the development of end-user preferred improved varieties.Consumers in Ghana prefer non-sweet sweetpotatoes with high dry matter content (Baafi, 2014;Baafi et al., 2015;Sam & Dapaah, 2009).However, locally available clones have very sweet taste, which limits their consumption as a staple food (Missah & Kissiedu, 1994).Increased sweetpotato utilization has become an important objective in Ghana recently.The major emphasis in breeding is on the development of farmer/consumer preferred varieties.Dry matter content, starch content, sugar content and storage root yield are quantitatively inherited in sweetpotato (Jones, 1986).Heterosis for these traits is present in sweetpotato hybrids between certain varieties (Baafi, 2014;Grüneberg et al., 2009), and the identification and use of heterosis is important for breeding sweetpotato.The objective of this study was to develop farmer and consumer preferred sweetpotato varieties using a diallel mating scheme and to estimate the level of heterosis and heterobeltiosis among F 1 hybrids obtained from a diallel between low sugar sweetpotato genotypes.

Germplasm Collection and Evaluation
Germplasm were collected from farmers' field in the major sweetpotato growing areas in Ghana in 2010.These were the Northern, Upper East, Upper West, Volta, Eastern, Central and the Brong Ahafo Regions.Collections from the CSIR-Crops Research Institute, Kumasi and the CSIR-Plant Genetic Resources Research Institute (PGRRI), Bunsu, were also included.In addition, accessions were collected from the Crop Science Department, University of Ghana and the International Potato Centre (CIP) gene bank in Accra and Kumasi, respectively.A total of 115 sweetpotato accessions (Table 1) were collected.These represent four groups, namely local accessions (32), local improved accessions (13), exotic and local accessions in National Agricultural Research Systems (NARS) or Programmes (43), and exotic accessions from CIP,Kumasi germplasm (27).Evaluation of the sweetpotato germplasm was carried out under rain-fed conditions in two replications at Fumesua (Forest ecozone) and Pokuase (Coastal Savanna ecozone) in the major and minor cropping seasons from May to December, 2011 to identify low sugar parents for hybridization.Planting distance was 1 m between ridges and 0.3 m within row of ridge length 3.6 m and a total of 12 plants per ridge.

Data Collection
Harvesting was done at three and half months after planting.The 10 central plants were harvested and one large, one medium, and one small storage roots were randomly selected for determination of sugar content.Storage roots used for observations were those approximately over 3 cm in diameter and without cracks, insect damage or rotten parts (Ekanayake et al., 1990).The Workflow for Sample Preparation and near-infrared reflectance spectroscopy (NIRS) analysis of sweetpotato developed by the Quality and Nutrition Laboratory of the International Potato Centre (CIP), Lima, Peru was used.Fifty grams (50 g) fresh sample was used.

Data Analysis
Only data for 102 out of the 115 accessions were analyzed due to missing information.The analysis also excluded minor cropping season data for Pokuase because the experiment failed due to erratic rainfall.The data were subjected to Analysis of Variance (ANOVA) using Genstat statistical package (Genstat, 2007).The relative efficiency (RE) of an alpha lattice design over randomized complete block design (RCBD) was not significant.The RE was determined as shown below; Where, MSe = Error means square; RE is significant if the ratio is > 1 and vice versa.Hence, RCBD in two replications was used to analyze the data.

Hybridization and Genetic Material Used
The hybridization block was established at the research field of the CSIR-Crops Research Institute (CSIR-CRI), Fumesua in 2012.The list of parents used is shown in Table 2.The parents were selected from the germplasm collected and evaluated based on their low sugar content (Bottom 10% of accessions).They were ten genotypes, and were made up of two released varieties (Histarch and Ogyefo) and eight breeding lines (AAT-03-025, CIP 442264, CRIWAC 25-10, CRIWAC 30-10, DOS 03-006, CRIWAC 11-10, CIP 440095 and CRIWAC 19-10).The hybridization block was established during the minor cropping season in 2012 at planting distance of 0.3 × 1 m.The full diallel mating design was used.Flowers ready for pollination the next morning, were tied the previous afternoon using a piece of drinking straw to prevent out-crossing by insects.At the time of pollination, the corolla was carefully opened, pollinated and carefully tied again afterwards to avoid insect contamination after pollination.Although self-fertilization occurs only rarely in sweetpotato (Poole, 1955), emasculation was done on the female parents to eliminate such a possibility.

Field Layout
The F 1 progenies produced and their parents were evaluated at three locations across three ecozones of Ghana in the minor cropping season in 2013.The locations were the CSIR-CRI research station at Fumesua (Forest ecozone), and the National Agricultural Research Stations at Wenchi (Transition ecozone) and Pokuase (Coastal Savanna ecozone).Since sweetpotato is highly heterozygous, each cross between two different parent plants is genetically distinct and the variation in the F 1 families produced is equivalent to an F 2 generation in a crop like maize.Therefore, there was a need to evaluate variation between different F 1 families as well as the variation within crosses.For this reason, twelve full-sib families obtained from crosses among four parents (Histarch, Ogyefo, CIP 442264 and AAT 03-025) out of the ten parents selected for hybridization were evaluated.The families consisted of one hundred and eleven F 1 progenies but, due to poor vigour of some progenies, 92 were evaluated alongside their parents at the three locations in two replications.All entries were planted in a single row on ridges at five plants per progeny at planting distance of 0.3 m within row and 1m between rows.Four node vines from the middle portion to the tip were used for planting.Genotypes within family were randomised into adjacent plots.

Data Collection
Harvesting was done at three and half months after planting.The three middle plants for each progeny row were harvested, and one large, one medium, and one small, storage root were randomly selected for physico-chemical analysis after yield data were taken.Storage roots taken for the yield data were those approximately 3 cm or more in diameter and without cracks, insect damage or rotten parts (Ekanayake et al., 1990).The physico-chemical traits determined were dry matter, starch, and sugar contents.This was done at the Quality and Nutrition Laboratory of the International Potato Centre (CIP) at Fumesua, Ghana.The physico-chemical traits were determined using the Workflow for Sample Preparation and near-infrared reflectance spectroscopy analysis of sweetpotato developed by the Quality and Nutrition Laboratory of CIP Lima, Peru.Fifty grams fresh sample was used.The fresh sample was freeze-dried for 72 hours.The dry matter content was calculated as the ratio of the weight of the dry sample to that of the wet sample expressed as a percentage.

Data Analysis
Data were analysed using the approach of Buerstmayr et al. (2007).Analysis of Variance (ANOVA) was performed on data for all parents and their F 1 s to ascertain the performance among and between the F 1 s and the parents in 8 × 12 alpha lattice design.After this, data for the different F 1 families were analysed separately with their parents to estimate heterosis.All the analyses of the different F 1 families were carried out in a randomized complete block design (RCBD) except for the crosses between Histarch and Ogyefo which was done using an 8 × 11 alpha lattice design.This was because the relative efficiency (RE) of alpha lattice design over randomized complete block design (RCBD) was significant for the data involving crosses between Histarch and Ogyefo.The analyses were done using Genstat version 9.2.0.152 (Genstat, 2007).The percent increase or decrease of F 1 hybrids over mid-parent as well as better parent performance was calculated to estimate heterosis for the traits studied (Fonseca & Patterson, 1968), as shown below, Where, Ht = Heterosis; Hbt = Heterobeltiosis; MP = Mid-parent value; BP = Better parent value; F 1 = F 1 hybrid value.
The 't' test was used to determine whether F 1 hybrid means were significantly different from mid-parent and better parent means (Wynne et al., 1970), as follows; Where, F 1ij = Mean of the ijth F 1 cross; MP ij = Mid-parent value for the ijth cross; BP ij = Better parent value for ijth cross; EMS = Error mean square.

Results
Numerous ANOVA tables were obtained because numerous ANOVA were carried out for the different separate analysis involving genotypes within family (to estimate mid-parent and better parent heterosis).Only ANOVA table from the combined analysis (i.e.involving all the hybrids irrespective of family and the parents) is reported (Table 5).

Performance of the 102 Sweetpotato Accessions Based on Sugar Content
The phenotypic coefficient of variation (PCV) was higher than the genotypic coefficient of variation (GCV) (Table 3).The broad-sense heritability was 0.70.The expected gain from selection and genetic advance (as percent of grand mean) were 5.10 and 30.95%.There were significant differences in the sugar content of the sweetpotato accessions (Table 4).Sugar content ranged from 9.83% to 30.34%.These values were given by CRIWAC 19-10 and CIP 442850.

Performance of the Parents and F 1 Hybrids
There were challenges with most of the crosses due to cross incompatibilities which was either due to poor or lack of flowering and/or genetic incompatibilities.This occurs frequently with sweetpotato hybridization.The mean squares from the combined ANOVA showed highly significant (p < 0.01) differences among the genotypes for all the traits except sugar content which was significant at p < 0.05 (Table 5).Range of values for the traits were 36-48% for dry matter content, 9.01-17.53%for sugar content, and 68.27-76.25%for starch content (Table 6).Some genotypes did not produce storage roots but the highest root yield was 36.31 t/ha.The mean values for the traits were 43% (dry matter content), 13.45% (sugar content), 72.64% (starch content), and 16.04 t/ha (storage root yield).Their coefficients of variation were 6.3% (dry matter content), 20.7% (sugar content), 2.9% (starch content), and 46.9% (root yield).

Estimation of Heterosis and Family Performance for the Sweetpotato Genotypes
Significant variation (P < 0.05) was seen between the parents, the F 1 hybrids and between the parents and the F 1 hybrids for all the traits in the crosses between Histarch and Ogyefo (Table 7).Fifty-five per cent of the F 1 hybrids had sugar content lower than Ogyefo (13.58%).Crosses Histarch × CIP 442264 and Histarch × AAT-03-025 had significant differences (P < 0.05) for only dry matter content and root yield (Table 9 and 11).There were no significant estimates for heterosis between crosses Histarch × CIP 442264 (Table 10), but some F 1 progeny had significant heterosis for dry matter content, sugar content and storage root yield in crosses between Histarch and Ogyefo (Table 8).Heterosis was significant for dry matter content and storage root yield for crosses between Histarch and AAT-03-025 (Table 12).Both positive and negative heterosis was seen.For example, Histarch × Ogyefo-26 (Table 8) had significant negative mid-parent and better parent heterosis for dry matter content (-14% and -18%), and sugar content (-39.1% and -33.0%) while Histarch × Ogyefo-13 had positive heterosis for starch content (4.6% and 4.6%).

Discussion
Significant differences observed between the germplasm provided opportunity for identification and selection of superior genotypes as parents for hybridization.Significant differences demonstrate significant genetic diversity and indicates that meaningful selection and improvement of desired trait is possible (Mohammed et al., 2012;Nwangburuka & Denton, 2012).The observed difference between PVC and GVC could be attributed to environmental effects (Denton & Nwangburuka, 2011).GCV provides a measure to compare the genetic variability present in various quantitative traits.However, it is not possible to estimate heritable variation with GCV alone (Prasad et al., 1981).The use of GCV with heritability estimates give the best picture of the amount of advance to be expected from selection (Burton, 1952).Heritability indicates the effectiveness with which selection of genotypes can be based on phenotypic performance (Johnson et al., 1955).In this study, broad-sense heritability estimates for sugar content was high.Traits with medium to high heritability are influenced by additive gene effects (Denton & Nwangburuka, 2011).This suggests that selection based on phenotype will be effective.Two of the parents (CRIWAC 25-10 and CRIWAC 19-10) did not produce flowers, and for those that did, only Histarch, Ogyefo, CIP 442264 and AAT-03-025 gave successful crosses.Lack of seeds and fewer numbers of seedlings from some crosses may largely be attributed to poor flowering and genetic incompatibility.This is because, according to Martin (1967), Martin and Cabanillas (1968), the improvement of sweetpotato through conventional breeding is impeded by poor flowering and incompatibility.According to Fekadu et al. (2013), flowering prolificacy in sweetpotato is variety dependent.While some varieties do not flower at all, others produce very few flowers.In addition, many sweetpotato clones rarely flower under normal conditions as a result of differential response to seasonal variation.Most sweetpotato genotypes are day length sensitive.Thus while some genotypes flower readily any season, flowering in others occurs only during short day length (Martin, 1988).Among the parents that gave successful crosses, AAT-03-025 and CIP 442264 were very low in flower prolificacy that is why fewer crosses involving them were made and subsequently fewer seeds were obtained.
The range of values obtained for this study were comparable to those reported by Grüneberg et al. (2009).Values for dry matter content were also comparable to those reported by Shumbusha et al. (2014).Significant negative heterosis observed for sugar content is important for sweetpotato improvement in Ghana because the main trait preferred for increased sweetpotato utilization in Ghana is non-sweetness (bland taste) (Baafi, 2014;Baafi et al., 2015;Missah & Kissiedu, 1994;Sam & Dapaah, 2009).Breeding for non-sweetness is the most important breeding objective of the crop currently in Ghana.The sugar contents of the hybrids evaluated agrees with Kays et al. (2005).The authors classified sweetpotatoes based on sugar content as very high ≥ 38; high 29-37; moderate 21-28; low 12-20; and non-sweet ≤ 12. Based upon this classification, 15% of the F 1 hybrids of Histarch and Ogyefo were non-sweet.These will meet the staple food needs of Ghanaians.Based upon sugar content (mainly), dry matter content and root yield, eight hybrids were identified as potential non-sweet varieties for release.They were Ogyefo × Histarch-11, Histarch × Ogyefo-13, Histarch × Ogyefo-52, Histarch × Ogyefo-37, Histarch × Ogyefo-65, Histarch × Ogyefo-88, Histarch × Ogyefo-39, and Histarch × Ogyefo-16.These genotypes had high dry matter content of 42-46%, sugar content of 9-12%, and root yield of 12-31 t/ha as compared to Histarch (46% dry matter, 16.33% sugar content and 20.7 1t/ha root yield) and Ogyefo (41% dry matter, 13.62% sugar content and 13.61 t/ha root yield).Yields of the selected hybrids were low compared to Histarch.The first documented non-sweet, staple-type sweetpotato breeding line GA90-16 yields less than the most widely grown traditional North American cultivars Beauregard and Jewel (Kays et al., 2001).According to the authors, GA90-16 in Athens, total yields are generally ≈ 70% to 80% of 'Jewel', varying with year and location.Relatively lower yields for non-sweet varieties may be because higher yields are sacrificed for non-sweetness.In terms of yield some of these hybrids may be low but their non-sweetness will contribute so much for their increased acceptance and utilization in Ghana.In addition, 26 other hybrids were identified for other purpose such as source of sugar flour for sweetening porridge and aboolo.One of the oldest uses of sweetpotato is sweetening porridges and maize products, such as Aboolo (steamed or baked sweetened fermented maize dough) (Osei-Opare & Adjei-Poku, 1977).

Conclusion
Significant genetic diversity was found for sugar content.Sufficient useful genetic variation was present in the materials studied and was exploited to provide for substantial amount of improvement through selection of superior genotypes.Significant heterosis was found which is useful for the improvement of sweetpotato for increased utilization in Ghana and beyond.Negative heterosis observed for sugar content is very important because breeding for non-sweetness will raise sweetpotato to an increased staple food status in Ghana.The hybrids listed above will be further tested multi-locationally for potential release to farmers.These selected hybrids together with their parents used in this study will be used as the breeding population for sweetpotato improvement programme in Ghana.In addition to polyploidy and large chromosome number, this study suggested that flowering and self-or cross-incompatibility are major constraints to sweetpotato breeding in Ghana.

Table 2 .
List of sweetpotato parents used for the establishment of the crossing block and their attributes

Table 3 .
Genotypic and phenotypic coefficient of variation, heritability and expected genetic advance for sugar content of the 102 sweetpotato accessions

Table 4 .
Storage root sugar content for the 102 accessions evaluated across three locations

Table 5 .
Mean squares from combined ANOVA for the parents and their F 1 hybrids

Table 6 .
Performance of sweetpotato parents and F 1 hybrids