Agro-Morphological Characterization of Sweet Potato Genotypes Grown in Different Ecological Zones in Kenya

The main characteristic of sweet potato [Ipomoea batatas (L.) Lam.] is its high phenotypic and genotypic variability. There is no, or limited, information on the suitability of agromorphological characteristics for utilization in production and processing. Currrently, farmers are growing different sweet potato genotypes characterized with low yield. The aim of the study was to evaluate agro-morphological characteristics of 68 sweet potato genotypes in order to determine the best-performing ones. The sweet potato genotypes were grown at the Kenya Agricultural and Livestock Research Organization and the Embu and Miyare Agriculture Training College. The locations were chosen because they are the main sweet potato producing areas with different climatic and production conditions. Six storage root and five aerial characters were used in the characterization. The genotypes differed in storage root stalk and root length; vine internode length and vine internode growth rate; petiole length and leaf size, and yield at both sites. Genotype Nyautenge was the best performing in terms of storage root yield. There was poor correlation among agro-morphological attributes. The study demonstrates the potential of some sweet potato genotypes such as Nyautenge for high productivity.

informative descriptors were abaxial leaf vein pigmentation, the shape of roots and vine tip pubescence [4].
Morphological variation has been widely used to characterize sweet potato genotypes [9][10][11] and to eliminate duplicates among genetic accessions [12,13]. Additionally, [14] and [1], using agro-phenotypic characters, reported wide diversity among sweet potato genotypes. One of the main characteristics of sweet potato is its high phenotypic and genotypic variability [15] that confers adaptability to different climatic conditions. Through characterization, a diversity that exists in a germplasm population can be estimated and studied. Morphological characterization provides information on conserved germplasm, placing it in the most effective form for use, and the value of the germplasm increases as it becomes known and documented [9]. Agro-morphological characterization in sweet potato is done by assessing variations in the vine, leaf, flower and storage root characteristics. This method has been used for identifying sweet potato cultivars, duplicate accessions, detecting unique character traits, and correlation with characteristics of agronomic importance. Morphological and agronomic characters, such as the storage root to vine ratio, have been used to identify and select dual-purpose sweet potato varieties [1]. Currently, sweet potato yield in developing is far below compared with developed countries. It is essential to determine the best performing genotypes to be recommended to farmers for production [10]. This way, it will be possible to estimate the real variability maintained to make conserved germplasm available for effective use by researchers and farmers. Agro-morphological characterization provides information on conserved germplasm, placing it in the most effective form for use, and the value of the germplasm increases as it becomes known and documented. There is no, or limited, information on the suitability of agro-morphological characteristics for utilization in production and processing. The objective of the study was to characterize accessions of sweet potato genotypes based on their agro-morphological descriptors.

Materials and Methods
The experiment was done between October, 2013 and April, 2014 in season one and between March, 2014 to October, 2014, for season two at the Miyare Agriculture Training College farm situated in the Migori County (ACT-Miyare) and Kenya Agricultural and Livestock Research Organization in Embu County (KALRO-Embu). The latter is at an altitude of 1497 m a.s.l., annual rainfall of 1252 mm, and annual temperature of 19.5 °C with humic nitisols soils. The ATC-Miyare is at an altitude of 1460 m a.s.l., annual rainfall of 1700 mm, annual temperature of 16.5 °C and humic acrisols soils.
Sixty-eight sweet potato genotypes, collected as vine cuttings from sites in Kenya and Uganda were used ( Table 1). The genotypes included 57 local Kenyan landraces and 11 F1 hybrids from a polycross obtained from the National Crops Resources Research Institute, Uganda. The genotypes collected from Kenya were from the Western, Nyanza and Eastern regions. The criterion for genotype collection was based on genotypes commonly grown by farmers in the Western region of Kenya which have some resistance to weevils (Cylas spp.). The 68 sweet potato genotypes were multiplied at KALRO-Embu. The sweet potato genotypes were planted in a randomized complete block design replicated three times at ATC-Miyare and KALRO-Embu. Each plot was 1.5 × 3.75 m with a plant spacing of 30 × 75 cm having 25 plants per plot. Sweet potato cuttings measuring 30 cm long from each genotype were planted in 5 rows. Weeding was at both sites 6 weeks after planting. Experimental fields were rain fed. No fertilizer was applied as a common practice done by sweet potato farmers. Harvesting was 160 days after planting.
Agro-morphological characterization of above and below ground parts was with the International Potato Center (CIP) guide [16] at 100 and 160 days after planting, respectively. The evaluation was on 9 plants of each genotype excluding border plants of each plot. Key agromorphological characters for sweet potato genotypes such as vine growth rate, vine internode length, vine internode diameter, storage root cortex thickness, storage root stalk, mature leaf size, storage root length, storage root diameter, petiole length, the weight of largest tuber and yield were evaluated ( Table 2).
Analysis of variance of agro-morphological data was in SAS (ver. 10, SAS Institute Inc, Cary, NC). Data were classified according to genotypes, locations, blocks and replications. Variation between sites was in SAS. If interactions were significant they were used to explain the results. If interactions were not significant means were separated using LSD. Cluster analysis was done on standardized agro-morphological data based on Euclidian distance coefficient and the Un-weighted Pair Group Method with Arithmetic means (UPGMA) using NCSS-pc (ver. 11, Tarragona, Spain). The hierarchical program in Number cruncher statistical systems (NCSS-pc) was used to generate dendrograms. Data points with smaller distances between them were grouped together. The Pearson correlation matrix was done using DARwin, ver. 6. Table 2. List of agro-morphological descriptors for characterizing sweet potato genotypes.

Results
The ANOVA analysis indicated that the main effects of site and genotype affected all agromorphological variables except vine internode length (VIL), vine internode diameter (VID), storage root cortex thickness (SRCT), mature leaf size (MLS), petiole length and weight of the largest root (WLR) which not significant between sites (Table 3), and there was no significant interaction between sites and genotype. There were differences in vine growth rate (VGR) due to genotype and site ( Analysis of variance indicated differences in vine internode diameter (VID) of genotypes ( Table 4). The VID of all genotypes ranged from very thin to thin. There were no genotypes that exhibited intermediate, long or very long VID. Genotypes that had the least VID were Fundukhusia and SPK 031. Genotypes 36 Kalamb Nyerere, Mogesi Gikenja, Ejumula × New Kawogo 1 and Polo yiengo had a thin VID. There was no difference in VID between sites. Analysis of variance indicated differences on storage root cortex thickness (SRCT) for genotypes (     Volume 5 Analysis of variance indicated differences in storage root stalk (SRS) of genotypes and sites ( Table 4). The SRS of all genotypes ranged from short to very long. There were no genotypes that exhibited very short SRS. Genotypes that had the shortest SRS were Kibuonjo; genotypes Naspot 1 and Mogesi Gikenja had the longest SRS. The ACT-Miyare site had longer SRS than the KALRO-Embu site.
Analysis of variance indicated no difference in mature leaf size (MLS) of sweet potato genotypes ( Table 4). The MLS of all genotypes in ACT-Miyare were regarded as small. There was no difference in MLS between sites.
Analysis of variance indicated differences on storage root length (SRL) of sweet potato genotypes and sites ( Table 4). The SRL of genotypes ranged from short to long. Genotypes with the shortest SRL were Nyawo Nyathiodiewo. The genotype that recorded the longest SRL was Nyautenge. Genotypes at KALRO-Embu had the longest SRL compared to those at ACT-Miyare.
Analysis of variance indicated differences in storage root diameter (SRD) of sweet potato genotypes and sites (Table 4). Genotypes with the shortest SRD was Mugande × New Kawogo 2. Genotypes with the longest SRD were Kenspot 3 Kenspot 5 and Kibuonjo. Genotypes at KALRO-Embu had longer SRD than at ACT-Miyare.
Analysis of variance indicated significant differences on petiole length of sweet potato genotypes ( Table 6). All genotypes had very short petioles. Genotypes that had the shortest petioles were Mugande × New Kawogo 2 while genotypes Ejumula × New Kawogo 2 had the longest petiole length. The sites did not differ in petiole length.
Analysis of variance indicated differences in weight of the largest storage root (WLSR) of sweet potato genotypes ( Table 4). Genotypes that had the least weights were Ejumula × New Kawogo 2; the heaviest WLSR was from genotype Mbita. The sites did not differ.
Analysis of variance indicated differences in storage root yield of genotypes and sites ( Table  4). The genotype Kunykibuonjo had the lowest root yield and genotype Nyautenge the highest yield. KALRO-Embu had the highest yield compared with ACT-Miyare.
Quantitative characters used to generate the dendrograms were: vine growth rate, vine internode length, vine internode diameter, storage root cortex thickness, storage root stalk, mature leaf size, storage root length, storage root diameter, petiole length, weight of largest tuber and yield (Figs. 1, 2). From the hierarchical cluster analysis, quantitative characters indicated polymorphism of about 2.5 among the genotypes at ATC-Miyare (Fig. 1). The tree obtained separated genotypes into 2 major clusters (A and B) at about 2.5 Euclidean distance. Cluster A contained 36 genotypes and consisted of 2 sub-clusters. Cluster B contained 32 genotypes and formed 3 sub-clusters. Both cluster A and B did not exhibit any distinguishable relationship or pattern. From the hierarchical cluster analysis, quantitative characters showed a polymorphism of about 2.8 among the 68 sweet potato genotypes at KALRO-Embu (Fig. 2). The tree obtained separated genotypes into 2 major clusters (A and B) at about 2.7 Euclidean distance. Cluster A contained 22 genotypes and consisted of 2 sub-clusters. Cluster B contained 46 genotypes and formed 3 sub-clusters (Fig. 2). Clusters A and B did not show any distinguishable relationship or pattern.
Significant correlations occurred among quantitative agro-morphological characters of the genotypes in ATC-Miyare (Table 5). Positive significant correlations occurred between vine growth rate and vine internode length, vine growth rate and mature leaf size, storage root stalk and root yield, and root yield and largest storage root diameter. Similarly, significant correlations occurred among the quantitative agro-morphological characters of the sweet potato genotypes in KALRO-Embu (Table 6). Positive significant correlations occurred between vine growth rate and vine internode length, largest storage root diameter and weight of largest root, storage root length and weight of largest root. Root yield was significantly, and positively, correlated with the weight of largest root. Figure 1. The dendrogram (based on Euclidean distance coefficient) of sweet potato genotypes generated from quantitative data at ATC-Miyare. Genotypes connected by portions of the dendrogram are highly related. As the dendrogram couplets coalesce the genotypes are fit into groups that are related based on vine growth rate, vine internode length, vine internode diameter, storage root cortex thickness, storage root stalk, mature leaf size, storage root length, storage root diameter, petiole length, the weight of largest tuber and yield. Letters (A, B) and numbers (1,2,3) represent clusters and sub-clusters respectively.

8
Volume 5 Figure 2. The dendrogram (based on Euclidean distance coefficient) of sweet potato genotypes generated from quantitative data at KALRO-Embu. Genotypes connected by portions of the dendrogram are highly related. As the dendrogram couplets coalesce the genotypes are fit into groups that are related based on vine growth rate, vine internode length, vine internode diameter, storage root cortex thickness, storage root stalk, mature leaf size, storage root length, storage root diameter, petiole length, the weight of largest tuber and yield. Letters (A, B) and numbers (1, 2, 3) represent clusters and sub-clusters respectively.

Discussion
Assessment of agro-morphological diversity and relationships among sweet potato varieties is important for germplasm conservation, and for breeding, especially during selection of varieties having superior qualities [17]. Among genotypes, most agro-morphological characters were highly variable. Genotypes in KALRO, Embu had better performance on agro-morphological characters such as storage root length, largest storage root diameter, and root yield compared with ATC, Miyare. Genotypes in ATC, Miyare were superior in vine growth length and storage root stalk compared with KALRO, Embu. This could be attributed to changes in agro-morphological conditions in both locations. The high variability in sweet potato genotypes is caused by natural mutations [18]. No single genotype was superior in all agro-morphological traits, due to unique genetic constitutions. Genotypes exhibiting intermediate, or fast, growth rate can be suitable for animal feed since the vines of sweet potato usually form an excellent source of green fodder [19]. However, most genotypes were not ideal in terms of root yield stability except for genotype Nyautenge. High yield is a product of genetic make up of individual genotypes [5,17], increased weight of roots, or increased number of roots, per plant [20]. There was a potential of some genotypes to yield more if all roots harvested from each plot would be equal to the largest root. Some genotypes exhibited long storage root stalks at both sites. A long root stalk increases rooting depth. Deep rooting can act as an escape mechanism to weevil infestation. Deep rooting and early maturing genotypes are less susceptible to insect pest infestations than shallow rooting and late maturing genotypes [21].
The dendrograms trees could only indicate general germplasm relatedness and diversity. There was high polymorphism of 2.5 and 2.8 in ATC, Miyare and KALRO, Embu respectively. This there indicates high genetic variability among the studied sweet potato genotypes. The probable reason as to why clustering of genotypes was not uniform across the dendrograms is that expression of agro-morphological characters is environment dependency. Similar results were obtained by [22] when studying sweet potato genotypes in Tanzania. That genotypes sharing a common name did not express genetic similarities underlines that artificial naming of biological organisms has no bearing on genetic makeup. This was more pronounced in the F 1 clones. It is possible that the F 1 clones clustered in different groups because they are not genetically stable. The high variability of vegetative characters among varieties can be attributed to high polyploidy level in sweet potato [20].
In ATC, Miyare, there was a significant correlation between vine growth rate with vine internode length and mature leaf size, storage root stalk with root yield and largest root diameter with root yied. Inn KALRO, Embu there was a significant correlation between vine growth rate with vine internode length, storage root length with weight of the largest root, largest root diameter with weight of the largest root and weight of the largest root diameter with root yield. These correlations indicate the importance of sink-source relationships in plants. M.J. Mbithe et al. [20] observed a similar relationship on sweet potato genotypes in Uganda.

Conclusion
Agro-morphological characters were used to characterize the selected sweet potato germplasm. Findings of the present study reveal that sweet potato germplasm presented high diversity based on the agro-morphological assessment. There was also a significant correlation among the variables of the studied genotypes. The study also revealed that the agro-morphological characters used in this study could effectively discriminate the different genotypes as seen from the dendrograms. Genotype Nyautenge was the best performing in terms of storage root yield.