Phenotypic Characterization, Evaluation, and Classification of Cassava ( Manihot esculenta Crantz) Accessions in Ethiopia

,


Introduction
Cassava (Manihot esculenta Crantz) grows more frequently in subtropical and tropical regions despite being native to Latin America, and it is the fourth major food crop in the world, next to maize, rice, and wheat [1]. It is regarded as an important nutrient source for ensuring food security in developing countries and a key source of food calories for two out of every fve people in Africa [2]. A number of industrial products are made with cassava starch, including paper, cardboard, textiles, plywood, glue, alcohol, animal feed, nongrain starch, ethanol, and biofuel [3]. Cassava has been shown to adapt and grow well in diferent agroecologies in Ethiopia, with various levels of yield [4]. It is an essential food crop that provides a large portion of people's daily consumption as well as the principal supply of carbohydrates in southern Ethiopia [5].
Cassava is a major storage root crop with a lot of potentially useful genetic variation [6]. Jahufer and Gawler [7] stated that genetic variation is central to plant breeding, as good management of variation can produce permanent gains in the yield of the plant and can bufer seasonal fuctuations. Farmers continue to keep a diverse set of accessions (landraces) on their farms, despite the fact that they may be low yielding and susceptible to some biotic and abiotic stresses [8]. Tese landraces may have greater genetic variability, which may enhance gene fow via hybridization [8,9]. Tus, landraces are valuable genetic resources for breeding and other crop enhancement eforts [9,10]. Furthermore, maintaining genetic resources is crucial for achieving genetic progress in breeding programs throughout the recombination and selection cycles [11]. To make the best use of available genetic variability, the genetic material must be properly organized and analyzed, increasing the likelihood of selecting genotypes with superior performance on the traits of interest [12]. Both morphological and agronomic variables, as well as molecular analysis, can be used to evaluate genetic variability [12]. Te primary goal of characterization is to get rid of duplicate accessions, generate a genetically unique core sample for use in breeding programs, and produce progeny who have a higher level of heterotic [13].
Te morphological characteristics of cassava are prominently variable, signifying a high degree of interspecifc hybridization [14,15]. Information on the genetic relationship between accessions is an important component of plant breeding programs because it provides knowledge of the genetic diversity available for producing new allelic combinations in a segregated population [11]. In cassava, it is supposed that a broad range of genetic variability was produced through centuries of farmer management [10]. Morphological characterization and evaluation of locally available accessions are indispensable for making information available for cassava root yield improvement. According to Fukuda and Guevara [16], assessing existing genetic variability is necessary and should be based on appropriate and recognised descriptors. So far, studies using qualitative characteristics conducted in various countries around the world have revealed signifcant variability within farmer-owned cassava cultivars [17]. According to Asare et al. [18], morphological characters are commonly used in primary evaluation because they are a quick and easy way to determine the degree of variability. Tese morphological characters reveal the true variability as perceived by farmers, and morphological characterization has been used to identify genetic diversity among cassava varieties [19].
Cassava is an essential food crop in Ethiopia that provides food security and income as well as a signifcant percentage of the daily diet [5,20]. However, there are various research gaps in this genetic research on important morphological characteristics in cassava, which lagged far behind other root and tuber crops [5]. Such research gaps are attributed to major problems such as a lack of desired features for the cassava storage root and high-root-yielding cultivars. Tis highlighted the signifcance of extending research eforts to look at the morphological characterization and relationships among cassava accessions through the collection of available genetic resources. Terefore, the aim of this research was to characterize, evaluate, and classify cassava accessions obtained from diferent sources using qualitative characteristics to provide valuable knowledge for cassava improvement programs and conservation eforts.

Te Study Area's Description.
Te experiment was carried out during the 2020-2021 cropping season at the Tarcha research site in the Dawuro zone, Southwest Ethiopia People' Regional State, and Bonbe research site in the Wolaita zone, Southern Nations Nationalities and Peoples' Region State. Tarcha research site is situated at an altitude of 1250 meters above sea level at latitude 07°09′32″N and longitude 037°1 0′16″E [21]. Te average annual rainfall in the area is 1392 mm, with a mean maximum and minimum temperature of 30°C and 17°C, respectively [21]. Te soil in the study area is nitosol, which is weathered brown and has a pH of 5.6. Bonbe research site is located at an altitude of 1701 meters above sea level at latitude 07°08′15.5″N and longitude 037°34′54.1″E [21]. Te average annual rainfall in the area is 1450 mm, with a mean maximum and minimum temperature of 26°C and 15°C, respectively [21]. Te soil in the study area is nitosol, which is weathered red and has a pH of 4.25.

Plant Materials and Design.
Sixty-four cassava accessions were used for the study, ffteen were from the Nigeria and fortynine accessions were from the Jimma and Hawassa Agricultural Research Centers (Table 1). Te experiment was designed as an 8 × 8 simple lattice. Mature cassava cuttings measuring 25-30 cm long were planted in an experimental plot (7 m 2 ) at an inter-row spacing of 1 m and an intrarow spacing of 1 m on the top of the ridge at a slanting (an angle of 45°) position. Te evaluation was conducted on three plants for each accession per plot. All cultural practices were performed as recommended by Markos et al. [22] and farmers' practices in the area.

Data Collection.
Te morphological characters were observed at four periods, 3, 6, 9, and 18 months after planting (MAP), based on 30 descriptors of cassava as developed by Fukuda et al. [23]. Te most frequently observed variant was noted in three plants ( Table 2). Te presence or absence of seeds and fruit was recorded throughout the vegetative stage until harvest.

Data Analysis.
Morphological frequency distributions were estimated based on the characters observed in the accessions. To assess the phenotypic diversity of all accessions, the Shannon-Weaver diversity index (H′) was computed using the phenotypic frequencies. Te Shannon-Weaver was estimated using the following formula [24]: 2 International Journal of Agronomy Pi ln pi, where pi is the proportion of accessions in the i th class of the trait to the total number of accessions grouped in the trait, where "n" is the number of phenotypic classes in the trait. Multiple correspondent analyses (MCA) were performed using the method developed by Abdi and Valentin [25] to analyze a set of observations for a set of categorical variables using statistics [26]. Te MCA is a method helpful for understanding the similarities between the categories of variables and the associations between the variables [27]. It displays and allows the identifcation of factorial axes that reveal the most discriminant variables [28]. Clustering analysis was performed based on unweighted pair group methods with an arithmetic average (UPGMA). Te numbers of clusters were determined using the cubic clustering criteria (CCC) as described by Mohammadi and Prasanna [29]. Te appropriate numbers of clusters were determined from the values of pseudo-F and pseudo-T 2 using the SAS software [26] and R-software package [30].
In the present study, several morphological characteristics revealed sufcient variation within the tested accessions. Among the characters, the shape of the central leafet, leaf retention, petiole color, branching habit, the color of the stem epidermis, the color of the stem exterior, the external color of the storage root, and the color of root pulp were found to be the most variable factors in the evaluated cassava accessions towards frequency distribution. Te variability of characters as indicated by frequency distribution in cassava agrees with the fndings of Brice et al. [28], who found petiole color, branching habit, the color of root pulp, the color of the stem exterior, and plant shape to be the major discriminating characteristics. Te other study was conducted by João Afonso et al. [38], who observed high variation in the foliar scar prominence, the cortex stem color, the petiole color, and the shape of the lobe. Carine et al. [39] reported that petiole color, branching habit, the shape of the plant, fowering, shape of the central leafet, the orientation of the petiole, the color of end branches, and the color of the root cortex were observed to have high variability and ofered possibilities for selection. Furthermore, the study reported by Nadjiam et al. [40] stated three variants of the stem exterior color, the stem epidermis color, and branching habit. However, these authors also found that the cortex stem color, the measurement between leaf scars, and the foliar scar prominence did not show variability among the evaluated 59 cassava accessions. Tis may be due to genetic similarities among accessions.

Te Shannon-Weaver Diversity Index (H′).
Te Shannon-Weaver diversity index (H′) was used to calculate the diversity of cassava accessions based on the frequency distribution of 30 qualitative characters and phenotypic classes [41]. In this aspect, the H′ value for most observed phenotypic characters exhibited a normal level of diversity among tested cassava accessions, which ranged from 0.24 for root constriction to 1.47 for petiole color, with a mean value of 0.84 (Table 3). As adopted by Islam et al. [42], H′ is classifed as low (H′ < 0.50), medium (H′ � 0.50-0.75), and high (H′ ≥ 0.75). Based on this classifcation, four characters were categorized as low diversity, such as the color of the stem cortex, the presence or absence of fruit and seed, and root constrictions (Table 3). Additionally, eleven characters had medium genetic diversity, while the remaining ffteen characters had high genetic diversity.
A low level of diversity may indicate the narrow genetic base of the plant and be more frequent than others [43], while a high H′ value shows a relatively high level of diversity and an even distribution of the landraces [41,44]. Moreover, the overall mean H′ value of 0.84 defned the presence of high phenotypic diversity among the tested cassava accessions. Tus, hybridization among accessions could produce the best hybrids and desirable segregants. Tis result is in agreement with the work of João Afonso et al. [38], who found a high H′ value for petiole color, stem exterior color, and color of the root bark.

Multiple Correspondent Analysis.
Te multiple correspondent analyses (MCA) of the qualitative characters are presented in Table 4. Te frst three dimensions of MCA were found to be responsible for up to 39.39% of the overall variation among the accessions (Table 4). Te petiole color, leaf retention, the shape of the plant, pubescence on apical leaves, the external color of the storage root, and the color of root pulp were major contributors to the frst dimension in multiple correspondent analysis (Dim1), which explained 20.77% of the total variation (Tables 4 and 5). Te second dimension in multiple correspondent analysis (Dim 2) explained 9.98% of the total variation and was primarily related to the contrasting efects of root epidermis texture, lobe margins, root cortex color, leaf vein color, and petiole orientation (Tables 4 and 5). Tis fnding was supported by those of Brice et al. [28], who characterized 89 accessions using 19 morphological characters, where the color of the apical leaves, the petiole color, fowering, and the yellow pulp color were the major contributing variables. In another study, Agre et al. [45] assessed the degree of genetic diversity of 116 elite cassava collections in Benin using 41 qualitative traits, while the apical leaf color, the shape of the plant, leaf retention, and constriction of the root are the leading contributors to the total variability.

Cluster Analysis.
Te clustering analysis results are presented in Figure 1, and the four divergent clusters' formation was performed using the criteria with a pseudo-F and a pseudo-t-squared, which had values of 7.9 and 2.4, respectively; the maximum increase occurred at this point ( Figure S1). Te distribution of the 64 accessions (Table 6 and Figure 1) showed that 52 accessions were in cluster I, 6 in cluster II, 5 in cluster III, and 1 in cluster IV. Te accessions grouped under cluster I have mainly been identifed International Journal of Agronomy by the purplish green color of apical leaves, the absence of pubescence on apical leaves, outstanding leaf retention, greenish red petiole color, leaf lobe number of seven, smooth lobe margins, the horizontal orientation of the petiole, the light green stem cortex color, the white color of root pulp, and the pink color of root cortex (Tables S1, Table 6, and Figure 1). Additionally, the study found that three accessions, G23, G36, and G40, had roots with a yellow pulp color under this cluster (Table S1). It is interesting to note that one of the elements determining the root pulp's value is its color. In cassava commercialization, it is preferable to use the pulp's yellow color since it denotes a high content of carotenoids, which are precursors to vitamin A and are crucial for both human and animal diets [33,46]. Under cluster II, six accessions were grouped together by having the difcult cortex removed or peeled, the cream color of root pulp, and the yellow color of root cortex, while cluster III was represented by the light brown external color of storage root, the white color of root pulp, the white or cream color of root cortex, and the easy cortex removed or peeled (Tables S1, Table 6, and Figure 1). Finally, one accession was classifed as cluster IV due to the following characteristics: reddish green in less than half of the lobe, leaf vein color, red petiole color, inclined upwards orientation of the petiole, zigzag growth habit of the stem, the dark brown external color of the storage roots, the pink color of root pulp and cortex, difculty in cortex peeling, and rough texture of the root epidermis (Tables S1, Table 6, and Figure 1).   G12  G34  G24  G15  G44  G9  G25  G43  G7  G40  G22  G10  G14  G36  G28  G29  G59  G64  G48  G21  G38  G1  G50  G11  G41  G60  G33  G4  G6  G2  G3  G5  G23  G30  G17  G56  G52  G55  G49  G58  G46  G57  G47  G51  G26  G32  G20  G39  G35  G63  G13  G45  G31  G16  G42  G8  G27  G61  G19  G37  G62  G18 G53 G54 Figure 1: A dendrogram displaying the dissimilarity of 64 cassava accessions by an average distance method of hierarchical clustering patterns using thirty qualitative characters.
International Journal of Agronomy 9 Variations in morphological characters have been considered simple indicators of genetic variability in crop species and varieties. Cluster analysis is used to categorize accessions that are alike into one group and others into a diferent one [20]. Based on unweighted pair group methods, the 64 accessions were classifed into four clusters, and the accessions represented in each cluster varied from 1 in cluster IV to 52 in cluster I, indicating that the studied cassava accession has a good chance of improving a character's interest through efective selection. Te distribution of accessions (Tables S1, Table 6, and Figure 1) showed that there were 52 (81.25%) accessions in cluster I (39 from the Hawassa, 3 from the Jimma, and 10 from the Nigeria); 6 (9.38%) accessions in cluster II (3 from the Hawassa, 1 from the Jimma, and 2 from Nigeria); 5 (7.81%) accessions in cluster III (2 from the Hawassa and 3 from the Nigeria); and 1 (1.56%) accessions in cluster IV from the Jimma. Accessions from the same or diferent sources are grouped into diferent clusters, implying that the genetic make-up of the accessions difers.
Te clustering analysis result is consistent with a similar study in Côte d'Ivoire, where a total of 89 accessions of cassava were characterized using 19 qualitative characters and were classifed into three groups [31]. In Benin, a total of 116 accessions were classifed, by Agre et al. [45], into six clusters using 41 qualitative descriptions, and in Brazil, 45 cassava cultivars were classifed, by Nadjiam et al. [40], into 5 clusters based on 36 qualitative characteristics. All those studies reported the presence of wide variability among tested accessions.

Conclusion
Te aim of this study was to characterize and evaluate cassava accessions in order to provide useful information for breeding programs and conservation. In this regard, the evaluated accessions showed high variability for the shape of the central leafet, branching habit, leaf retention, petiole color, color of the stem epidermis, the color of the stem exterior, external color of the storage root, and color of root pulp towards frequency distribution analysis. Tus, the various organ colors had a great role in identifying accessions. Te Shannon-Weaver diversity index (H′) value for most observed phenotypic characters has exhibited an optimum level of diversity among accessions. Te multivariate statistical analysis allows for identifying and grouping the accessions into diferent categories for various characters individually. As a result, the frst two dimensions in the multiple correspondent analyses revealed 30.75% total variability, which was mainly related to 11 major characters. Tus, those characters are considered the most relevant for use in describing cassava accessions. In the cluster analysis, the 64 accessions are classifed into four groups, with the number of accessions shared by each cluster being not uniform and varying from 1 in cluster VI to 52 in cluster I. Tis study could also facilitate breeders' identifcation and classifcation of desired features for the cassava storage root from both a genetic improvement and an agronomic point of view. For instance, accessions G23, G36, and G40 are identifed for optimal nutritional content. Finally, this study confrmed the existence of sufcient genetic variability for the characteristics evaluated, which could be attributed to the dissimilar genetic backgrounds of the evaluated accessions. Tus, the observed variation could be useful for a breeding program to develop cassava cultivars with desired root characteristics and conservation.

Data Availability
Te datasets that support the fndings of this study are available from the corresponding author upon request.