Agro-Morphological Characterization of Kenyan Slender Leaf (Crotalaria brevidens and C. ochroleuca) Accessions

Department of Agricultural Resource Management, University of Embu, P.O. Box 6–60100, Embu, Kenya Institute of Biotechnology Research, Jomo Kenyatta University of Agriculture and Technology, P.O. Box 62000–00200, Nairobi, Kenya Department of Life Sciences, South Eastern Kenya University, P.O. Box 170–90200, Kitui, Kenya Department of Medical Microbiology, Jomo Kenyatta University of Agriculture and Technology, P.O. Box 62000–00200, Nairobi, Kenya Department of Biological Sciences, University of Embu, P.O. Box 6–60100, Embu, Kenya


Introduction
Leafy vegetables contribute to additional taste and flavour as well as a source of protein, fibre, mineral, and vitamins, to human diet [1]. Slender leaf (Crotalaria spp) is one of the African indigenous leafy vegetables available in Kenya. e vegetable belongs to Fabaceae/Leguminosae family which comprise of annuals and short-lived perennials of agronomic importance [2]. eir largest centre of diversity is in Africa with approximately 500 species [3]. India is a secondary centre of origin, from where several species may have been dispersed to Australia and the East of the New World [4]. ere are two edible species of slender leaf, Crotolaria brevidens and C. orchroleuca, grown and consumed in Kenya, Tanzania, and Uganda among the East African countries [5]. e local name depends on the location. It is known as "marejea" in Tanzania, "mitoo" in Kenya, and "alaju" in Uganda [6]. Slender leaf grows prolifically in the wild as weeds in many parts of Kenya but its domestication and cultivation as a food crop are limited only to the western and coastal regions of the country.
Most of the African indigenous leafy vegetables have been proven to have the potential to contribute to the dietary improvement and food security for small households and less privileged families [7]. Slender leaf is not exceptional. e consumed portions of the slender leaf include young leaves and shoots and contribute 100% of the daily dietary requirement for vitamin A, vitamin C, iron, calcium, and proteins [8]. It provides 29.6 mg/g of zinc, 0.2 mg/g of calcium, 9.9 mg/g of vitamins, 11.1 mg/g of iron, and 1.9 mg/ g of proteins [9]. Slender leaf also has medicinal applications where it has been implicated in treating stomach-related ailments and malaria [10]. Furthermore, the crop has additional agronomic advantages like the ability to fix atmospheric nitrogen. It is also an important green manure legume due to high biomass production and with antagonistic effect on parasitic nematodes [11]. e slender leaf is, therefore, among the promising indigenous vegetables in Kenya whose potential is yet to be fully utilized.
Despite the high economic value of slender leaf, it has been for a long time neglected by researchers, farmers, and other stakeholders, who have concentrated on exotic and already improved popular vegetables [1]. To realize the agronomic potential and commercial viability of this highvalue vegetable, there is a need to develop its agronomic package, improve its leaf yields, and promote the production of quality seeds. is can only be achieved through intensive domestication, breeding, and selection. However, the success of every breeding programme relies on the level of genetic variability of the available germplasm [12,13] which forms the basis for selection. For the slender leaf, there is scanty information available on its genetic and agronomic attributes.
is knowledge needs to be harnessed to enable genetic and agronomic improvement of the crop. erefore, this study assessed the agro-morphological diversity of the two consumed species of slender leaf available in Kenya.

Description of Study Site.
Field experiments were set up at the University of Embu research farm for two seasons in October to December 2018 and March to May 2019. e University of Embu lies at 0°35′ 25″ S and 37°25′ 31″ E with an elevation of 1494 m above sea level. It lies within the upper midland 2 agro-ecological zone. e area receives an annual rainfall of up to 1230 mm. e rainfall pattern is bimodal with long rains falling between March and June and short rains between October and December [14]. e mean annual temperature is 19.5°C with a maximum temperature of 25°C and a minimum of 14.1°C [14]. e major soils are Humic Nitisols which are derived from basic volcanic rocks. ese soils are deep, highly weathered with friable clay texture and moderate-to-high inherent fertility [15].

Collection of the Test Accessions.
e seeds of thirteen accessions of C. brevidens and sixteen C. ochroleuca were collected from farmer's seed stock from nine Kenyan Counties, namely, Kakamega, Bungoma, Vihiga, Busia, Siaya, Homabay, Kisumu, Migori, and Kisii ( Figure 1). Geographical coordinates of the various collection points are presented in Table 1.

Experiment Layout and Design.
Field experiments were laid out in Randomized Complete Block Design (RCBD) with the 29 accessions as the treatments and replicated three times. Each treatment/accession was allocated a plot measuring 2 m × 2 m with a 0.5 m alley between plots. e seeds were directly sown through drilling in rows spaced 30 cm apart and then covered lightly with soil. inning was done four weeks after sowing to attain a spacing of 30 cm × 15 cm [16]. e experiments were purely rain-fed, and no fertilizer was applied. Manual weeding was carried out when necessary to maintain weed-free plots.

Data Collection.
Twenty quantitative and qualitative morphological traits were used to estimate the level of variation among the accessions based on Crotalaria descriptors adopted from Raj and Britto [17] with some slight modifications. e qualitative traits that were recorded include leaf type, leaf colour, stem colour, hairiness of stem, flower colour, dry pod colour, and seed colour. e quantitative data were collected on growth and agronomic traits including plant height, number of branches, petiole length, leaf length, leaf width, days to 50% flowering, days to first mature pod, pod length, pod width, number of pods per plant, number of seeds per pod, and weight of 1000 seeds. e leaf area was estimated at maximum leaf expansion using the formula A L � 0.73 (L L × W L ) adopted from Musyimi et al. [18], where A L is the leaf area, L L is the leaf width, and W L is the maximum the leaf width.

Data Analysis.
Qualitative and quantitative data were organized into a matrix and subjected to cluster analysis using XLSTAT Version 2019. Estimates of similarity among the morphotypes were calculated using Euclidean genetic distance, and dissimilarity indices were used to generate a dendrogram using the unweighted pair-group method with arithmetic average (UPGMA). Principal component analysis (PCA) was conducted to get information on the importance of the characters. A two-way analysis of variance (ANOVA) was used to analyze the quantitative data using XLSTAT Version 2019 to obtain the accessional variations, seasonal variations, and interactions between accessions and seasons. e means were then separated using Student's Newman Keuls (SNK) test at 95% level of confidence. Pearson's correlation coefficients were performed to compare the degree of association between quantitative traits.

Variation in Qualitative Characters.
A summary of the qualitative characters of slender leaf accessions that were studied is shown in Table 2. Out of the seven qualitative traits that were studied, only three were diverse among the accessions. e three variable qualitative traits were leaf type, flower colour, and seed colour. Accessions were found to be uniform for the rest of the traits, namely, stem colour, stem hairiness, leaf colour, and dry pod colour All the 29 accessions were glabrous (not hairy) with green-coloured stem and leaf. All the accessions also developed black dry pods at maturity but with variable seed colour ranging from pale yellow (45%), reddish (21%), and black (3%). e rest of the accessions had a mixture of two or three seed colours with 21% having a mixture of a pale yellow and reddish colour and 10% having a mixture of reddish, black, and pale yellow colours.
ree leaf types were identified among the   Most of the accessions (59%) had bright yellow flower corolla out of which 38% belonged to C. brevidens while 21% belonged to C. ochroleuca. e accessions with creamish yellow corolla comprised of 41% out of which 34% belonged to C. ochroleuca while 7% belonged to C. brevidens.

Variation in Quantitative and Agronomic Variables.
Significant variations (p < 0.05) between accessions were observed for leaf area index (p < 0.05), pod width (p < 0.01), and weight of 1000 seeds (p < 0.05). e rest of the quantitative variables were not significantly different (p > 0.05) ( Table 3). Leaf area index ranged from 12.54 in accession MKS221 from Kisii to 7.17 in accession BSY0113 from Siaya. Accession MMGR234 from Migori recorded the biggest pod width of 2.2 cm while BVG001 from Vihiga had the smallest pod width of 0.04 cm. e accessions were also found to vary significantly in weight of 1000 seeds with BKM203 from Kisumu recording the highest weight of 7.842 g while BKM197 from 5.447 Kisumu had the lowest seed weight.
ere were relatively higher variations in agronomic variables as compared to the quantitative genetic variables. e accessions recorded highly significant (p < 0.001) differences in days to 50% flowering which ranged from 64 to 74 days. ere were also highly significant (p < 0.001) variations between accessions for days to maturity which ranged from 95 to 117 days.

Pearson Correlations between Quantitative Variables.
ere were significant (p < 0.05) and positive correlations between several quantitative variables (Table 4). Plant height was found to be significantly correlated to leaf width (p < 0.040), pod width (p < 0.002), pod length (p < 0.0001), days to 50% flowering (p < 0.002), and days to maturity (p < 0.004). e leaf characteristics were found to have significant correlations among them with petiole length correlating highly with leaf length (p < 0.0001) which in turn correlated significantly with leaf width (p < 0.002) and leaf area (p < 0.0001). e leaf length and leaf area were also significantly correlated to days to 50% flowering (p < 0.03). Significant correlations were also observed between pod width and pod length (p < 0.0001), the number of seeds per pod (p < 0.002), days to flowering (p < 0.002), and days to maturity (p < 0.0001). e pod length was also positively correlated to the number of seeds per pod (p < 0.006), days to 50% flowering (p < 0.0001), and days to maturity (p < 0.003). e number of seeds per pod was also found to be significantly correlated with days to 50% flowering Mixture 1 � pale yellow and reddish; Mixture 2 � pale yellow, reddish, and black; the accessions whose code starts with letter B belongs to C. brevidens while those whose code starts with letter M belongs to C. ochroleuca.

4
International Journal of Agronomy (p < 0.029) and days to maturity (p < 0.004), both of which were also significantly correlated (p < 0.003).

Cluster Analysis.
Agglomerative hierarchical cluster analysis of both qualitative and quantitative traits grouped the 29 accessions into seven major classes (Figure 2). e diversity within classes was 24.87% while diversity between classes was 75.13%. Class 1 had only one accession of C. brevidens from Siaya county that was distinct with highest plant height, highest number of days to flowering (late flowering), largest pod length, and highest number of days to Means with the same letter within the column are not significantly different. * Significant at 5%; * * significant at 1%; * * * significant at 0.1%; NS: not significant; e accessions whose code starts with letter B belongs to C. brevidens while those whose code starts with letter M belongs to C. ochroleuca. International Journal of Agronomy 5 maturity (late maturity). Cluster 2 contained seven accessions with two C. brevidens accessions and five C. ochroleuca accessions from five different counties, three of which were from Homabay County. e seven accessions had 26.67 diversity between each other. Cluster 3 had ten accessions comprising of six C. ochroleuca and four C. brevidens accessions. e ten accessions had 35.53% diversity between each other. Clusters 4 and 5 had one accession each from Kisii and Migori counties, respectively, and both of them belonged to C. ochroleuca species. Cluster 6 contained only two accessions including one C. brevidens and one C. ochroleuca from Kisumu and Bungoma counties, respectively. e two accessions differed from each other with 22.92%. e highest within class diversity of 45.07% was observed in class 7 which contained seven accessions comprising of five C. brevidens accessions and two C. ochroleuca accessions from four different counties.

Principal Component Analysis.
e first three diversity factors accounted for 92.36% of the total agro-morphological variability among the accessions, out of which 84.30% variation was explained by the first two factors (Table 5). e number of seeds per pod, plant height, days to maturity, and days to 50% flowering were the principal components that made the highest contribution to factor 1 (F1) which explained 58.28% of the total variability. Factor 2 (F2) explained 26.03% of the total agro-morphological variation with plant height and number of seeds per pod making the highest contribution to this variability factor. Days to maturity, plant height, and number of seeds per pod were the principal components in F3 which explained to 8.06% of the total variation.
Correlations between variables and factors and between accessions and factors are illustrated in Figure 3. e accessions separated in a similar way as in hierarchical cluster dendrogram (Figure 2). e eight accessions in Clusters 1 and 2 ( Figure 2) were plotted relatively closer in the upper right quartile of the two-dimension plot (Figure 3(b)). ese accessions were characterized by high plant height, late flowering, long pods, and late maturity. e 10 accessions in Cluster 3 ( Figure 2) were also plotted together in the lower right quartile of the two-dimension plot (Figure 3(b)). ese accessions combined a high number of seeds per pod with long petioles and high seed weight. e accessions in the most diverse Cluster 7 were plotted in the upper left quartile of the two-dimension plot and were characterized by high leaf area and average branch number.

Discussion
e success of crop breeding programmes depends on the availability and identification of genetic diversity among the breeding materials [19]. Understanding the genetic diversity and agronomic performance of Crotalaria spp is the basis for enhanced adoption and genetic improvement of this highvalue vegetable crop. Insight on direct selection on leaf and seed yield can be provided by analysis of agronomic and related traits. In this study, agro-morphological diversity was observed among the different accessions of Crotalaria brevidens and C. ochroleuca in both qualitative and quantitative traits. However, this study did not identify any agro-morphological characters that could be uniquely linked to any of the two species as the characters were either similar in all the studied accessions or variables within and among the two species. Out of the seven qualitative characters that were studied, four of them, namely, leaf colour, stem colour, dry pod colour, and hairiness of the stem, were found to be similar among species. is observation concurred with Raj and Britto [17] and Schippers [20] who documented that C. brevidens and ochroleuca are closely related and their phenotypic information cannot be attributed to either of them.
Variation within and among species was observed in only three out of the seven qualitative characters. e three variable traits were leaf type, seed colour, and flower colour. e flower colour was observed to vary from pale yellow to creamish yellow with the former being more dominant in Crotalaria brevidens while the latter was more dominant in C. ochroleuca. is slightly differs from Schippers [20] who stated that C. ochroleuca can best be distinguished by the pale yellow or creamish colour of the flower, unlike C. brevidens whose flower is usually bright yellow. e leaf shape comprises of lanceolate, ovate, and linear types, with C. ochroleuca having more leaf type variations compared to C. brevidens in which the lanceolate type dominated. Soyewo and Omiyale [21] characterized the leaf morphology of the genus Crotalaria in Nigeria and reported that leaf features overlap within the genus with leaf shape varying among species from lanceolate to oblanceolate to ovate. e seed colour was found to be the most variable trait among and within the two species ranging from pale yellow to reddish to black and a mixture of either pale yellow and reddish colour or pale yellow, reddish, and black colours. A similar variation in seed colour was reported by Subramaniam et al. [22] among 38 Crotalaria spp in India including C. ochroleuca where the seed colour ranged from black, brown, steel grey, pale yellow, and orange red. ey concluded that seed colour is of little taxonomic significance in species characterization because there exists intraspecific variability in seed colour.
Similar to qualitative variables, most of the quantitative variables did not differ significantly among and within accessions. is could have been limited by the relatively fewer number of each species that were characterized in this study. Significant variation was demonstrated in only five out of the eleven quantitative variables studied. ese included three morphological variables, namely, leaf area index (LAI), pod International Journal of Agronomy width, and weight of 1000 seeds and two agronomic variables, namely, days to 50% flowering and days to maturity. According to Roux and Wyk [23], growth habit, leaf type, and quantitative morphometrics like petiole length, pod length, and width are phylogenetically important in the morphological characterization of Crotalaria. Variation in LAI was attributed to differences in leaf types. Leaf size reportedly contributes to the biomass yield and leaf yield and is, therefore, an important aspect in selecting candidates for leaf production [24]. e weight of 1000 seeds is an indicator of good quality seed and is an essential trait to consider during the selection of accessions to be used as parental stock in seed production during propagation or breeding projects. Variation in days to 50% flowering and days to maturity indicated that there were early and late flowering accessions and this can be attributed to the agro-ecological diversity of their areas of origin. However, early and prolific flowering is a limiting factor to leaf production and yield [25] and this should be considered during selection.
Significant and positive correlations were observed between several quantitative variables. is was an indication that some variables can be indirectly targeted in a selection criterion by using other related variables especially if the target trait does not provide better grounds for discriminative selection [26]. For example, although there was no significant variation between accessions for the variable plant height, the variable was found to be significantly correlated with leaf width, pod width, pod length, days to 50% flowering, and days to maturity. Plant height is considered one of the important traits that should be put into consideration in the agronomic improvement of many vegetables. A similar positive observation was made in soybean mutant accessions by Malek et al. [27]. ere was also a strong positive correlation between the leaf area and leaf width and leaf length. is was expected because the leaf area is a factor of leaf width and length. Similar findings were reported among amaranth genotypes by Gerrano et al. [28]. A significant positive correlation was also observed between days to 50% flowering and days to maturity. Besides, these two agronomic variables were also found to be positively correlated with the seed yield traits, namely, pod width, pod length, and the number of seeds per pod. Similar findings were reported in Corchorus accessions by Dube et al. [7]. ese traits have high levels of heritability and hence can be transferred to accessions of interest through breeding [29].
A cluster dendrogram is a good measure of diversity among and within species as it groups similar entries under one cluster [27]. Cluster analysis grouped the 29 slender leaf accessions into seven supported clusters based on the level of variation among the accessions. A similar strategy was applied by Zhang et al. [30] and Mekonnen et al. [31] on the agro-morphological characterization of Cucumis melo and lentils accessions, respectively. Diversity within supported clusters ranged from 0% (for singletons) to 45.07% as observed in Cluster 7. e high morphological diversity indicates a high potential of phenotypic selection which is a desirable aspect to plant breeders. Some accessions were placed separately in their clusters. ese were one Siaya accession of C. brevidens (Cluster 1), one Kisii accession of C. ochroleuca (Cluster 3), and one Migori accession of C. ochroleuca (Cluster 4). Such accessions are referred to as singletons and are given much attention in selection as they are suggested to be superior over other accessions [32]. Singletons were also observed in other genetic characterization studies [7,28,32].
Principal component analysis highlights and measures how each component contributes to total variation [33]. e factor loading through the eigenvectors indicates the level of contribution of each trait [7] and indicates which trait is the best target in genetic improvement of a certain crop. In this study, the first two eigenfactors accounted for 84.30% of the total phenotypic variability among the accessions. is variation was higher than what was reported by Akin-Idowu et al. [34] in grain amaranth and Ngomuo et al. [24] in jute mallow where the first two Eigen factors recorded a cumulative variance of 42% and 46%, respectively. Days to maturity, plant height, and number of seeds per pod made the most contribution to the morphological variation observed among and within the 29 slender leaf accessions that were studied. is indicates that these three traits can be utilized as target traits during hybridization. Comparing the correlations between variables, accessions, and factors offered a simple method of identifying the principal components in particular accessions and thus can be effective in aiding selection and hybridization.

Conclusions
is study revealed a moderate level of morphological diversity between and within Crotalaria brevidens and C. ochroleuca. However, there is potential of selection for some particular traits of agronomic importance including leaf area index, pod width, days to 50% flowering, and days to maturity. ese traits can be targeted for subsequent hybridization between and within the two Crotalaria species for crop improvement.
e study did not identify any phenotypic characters that can be uniquely associated with any of the two species. Uniformity in phenotypic traits such as stem hairiness, stem colour, leaf colour, and dry pod colour is of little taxonomic significance in the characterization of the two species since the said traits cannot be effectively used to discriminate between the two species. Similarly, seed colour cannot be effectively used in discriminating between the two Crotalaria species since it varies between species. is study sets the basis for genetic improvement of slender leaf in Kenya since the observed diversity can be exploited in selection and intraspecific and interspecific hybridization. e authors recommend further analysis of genetic diversity between and within the two species using molecular markers to provide deeper insights into the existing variability. Future studies should also consider a relatively higher number of accessions per species in order to increase the chances of identifying unique traits that can be used to clearly distinguish between the two species.

Data Availability
Some of the data used to support the findings of this study are included in the article. Additional data are available from the corresponding author upon request. 8 International Journal of Agronomy